@prefix vivo: . @prefix edm: . @prefix ns0: . @prefix dcterms: . @prefix dc: . @prefix skos: . vivo:departmentOrSchool "Arts, Faculty of"@en, "Philosophy, Department of"@en ; edm:dataProvider "DSpace"@en ; ns0:degreeCampus "UBCV"@en ; dcterms:creator "Preinsperger, Kurt"@en ; dcterms:issued "2008-12-18T21:15:59Z"@en, "1992"@en ; vivo:relatedDegree "Doctor of Philosophy - PhD"@en ; ns0:degreeGrantor "University of British Columbia"@en ; dcterms:description """Numerous conflicting theories about the just distribution of goods in society have emerged in response to John Rawls’ 1971 treatise A Theory of Justice. Given that informed persons continue to disagee strongly about the demands of social justice, progress may come from better understanding the underlying reasons for disagreement about justice among philosophically informed people of moral goodwill. In this dissertation I explore the idea, suggested by Larmore, Pogge, Ackerman and others, that some of the disagreement about justice among informed people of goodwill is “rational” disagreement. I identify, in the literature on social justice, recurrent conceptual, normative and empirical issues which we have reason to consider currently (or conceivably forever) irresolvable, at the first-order level of moral reasoning, by any information, arguments or methods accessible to us. But claims concerning the possibility of rational disagreement about justice only differ non-trivially from skepticism about justice if plausible limits can be set to the scope of this disagreement. To characterize such limits, I seek to establish the following two wide-reflective-equilibrium-based presumptions. (1) A consequentialist metaethical framework is our most credible approach to moral justification (where the form of consequentialism defended is constructivist, non-foundational, value-pluralistic, and includes distribution sensitivity among its ultimate values). (2) In moderately well-off, pluralistic societies, only those conceptions of justice fall within the scope of rational disagreement which propose broadly egalitarian-liberal, “directly responsive” principles (i.e. principles applicable to individual or group shares and not merely to basic social structure). Some likely candidates for the status of rational disagreements about social justice are discussed: the criteria definition and inclusion problems; various balancing problems related to attempts to increase the comprehensiveness of principles (the priority problem, the aggregation-distribution problem and commensuration problems); the domain demarcation problem; and problems of imprecision associated with justifying claims about justice within a consequentialist framework. An improved understanding of major sources of rational disagreement about social justice, as presented in this dissertation, helps define the normative weight of appeals to justice. This in turn clarifies the need to resolve many issues of social distribution otherwise than by relying on invocations of justice."""@en ; edm:aggregatedCHO "https://circle.library.ubc.ca/rest/handle/2429/3128?expand=metadata"@en ; dcterms:extent "5926957 bytes"@en ; dc:format "application/pdf"@en ; skos:note "RATIONAL DISAGREEMENT ABOUT SOCIAL JUSTICEbyKURT PRENSPERGA THESIS SUBMIYI’ED IN PARTIAL FLTLFTLLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFDOCTOR OF PHILOSOPHYinTHE FACULTY OF GRADUATE STUDIESDepartment of PhilosophyWe accept this thesis as conformingTHE UNIVERSITY OF BRITISH COLUMBIAPre- &15fe-9 July 1992Signature(s) removed to protect privacyIn presenting this thesis in partial fulfilment of the requirements for an advanceddegree at the University of British Columbia, I agree. that the Library shall make itfreely available for reference and study. I further agree that permission for extensivecopying of this thesis for scholarly purposes may be granted by the head of mydepartment or by his or her representatives. It is understood that copying orpublication of this thesis for financial gain shall not be allowed without my writtenpermission.(Signature)Department ofThe University of British ColumbiaVancouver, CanadaDate______________£r (SDE-6 (2/88)Signature(s) removed to protect privacy11ABSTRACTNumerous conflicting theories about the just distribution of goods in societyhave emerged in response to John Rawis’ 1971 treatise A Theory of Justice.Given that informed persons continue to disagee strongly about the demandsof social justice, progress may come from better understanding the underlyingreasons for disagreement about justice among philosophically informedpeople of moral goodwill.In this dissertation I explore the idea, suggested by Larmore, Pogge, Ackermanand others, that some of the disagreement about justice among informedpeople of goodwill is “rational” disagreement. I identify, in the literature onsocial justice, recurrent conceptual, normative and empirical issues which wehave reason to consider currently (or conceivably forever) irresolvable, at thefirst-order level of moral reasoning, by any information, arguments ormethods accessible to us.But claims concerning the possibility of rational disagreement about justiceonly differ non-trivially from skepticism about justice if plausible limits canbe set to the scope of this disagreement. To characterize such limits, I seek toestablish the following two wide-reflective-equilibrium-based presumptions.(1) A consequentialist metaethical framework is our most credible approachto moral justification (where the form of consequentialism defended isconstructivist, non-foundational, value-pluralistic, and includes distribution-sensitivity among its ultimate values). (2) In moderately well-off, pluralisticsocieties, only those conceptions of justice fall within the scope of rationaldisagreement which propose broadly egalitarian-liberal, “directly responsive”principles (i.e. principles applicable to individual or group shares and notmerely to basic social structure).Some likely candidates for the status of rational disagreements about socialjustice are discussed: the criteria definition and inclusion problems; variousbalancing problems related to attempts to increase the comprehensiveness ofprinciples (the priority problem, the aggregation-distribution problem andcommensuration problems); the domain demarcation problem; and problemsof imprecision associated with justifying claims about justice within aconsequentialist framework.An improved understanding of major sources of rational disagreement aboutsocial justice, as presented in this dissertation, helps define the normativeweight of appeals to justice. This in turn clarifies the need to resolve manyissues of social distribution otherwise than by relying on invocations ofjustice.S 111CONTENTSAbstract iiList of tables ivAcknowledgments vINTRODUCTION 1Part 1: THE CONCEPT OF RATIONAL DISAGREEMENT1.1 The philosophical context 61.2 A conception of rational disagreement 141.3 Objections 21Part 2: THE CHOICE OF A CONSEQUENTIALIST MORAL FRAMEWORK2.1 Defining non-trivial limits to rational disagreement 282.2 Reflective equilibrium as a method of moral justification 352.3 Choosing consequentialism in reflective equilibrium 422.4. The structure of CND consequentialism 572.5 Rawis’ avoidance of moral theory 70Part 3: CONSEQUENTIALISM AND RATIONAL DISAGREEMENT ABOUT JUSTICE3.1 Disagreement about just-making criteria 813.1.1 The concept of social justice 813.1.2 The criteria definition problem 943.1.3 The criteria inclusion problem 1103.2 Disagreement about the comprehensiveness of principles 1323.2.1 Basic properties of principles of justice 1323.2.2 Sources of rational disagreement about principles 1443.2.3 Rawis master principle 1553.2.4 Limits of the master-principle approach to justice 1733.3. Disagreement about the justification of principles 1813.3.1 The basic nature of consequentialist justification 1813.3.2 Bona fide principles of justice and morally justified principles 1903.3.3 Defining injustice 1953.3.4 Rational disagreement about consequentialist justifications 2003.4 The conflict between liberals and non-liberals 2083.4.1 Liberal and non-liberal conceptions of justice 2083.4.2 On the nature of human well-being 2213.4.3 Limits to rational disagreement about liberal politics 2253.5 The conflict between libertarians and egalitarians 2393.5.1 The presuppositions of freedom 2393.5.2 The contractarian justification of libertarianism 2463.5.3 Limits to rational disagreement about libertarianism 265SUMMARY AND CONCLUSIONS 271BIBLIOGRAPHY 291LIST OF TABLESTable I: A taxonomy of views of justice Page 114Table II: Types of master principles of justice Page 139Table IlL Consequentialist formulations of traditional Page 201moral standards used as criteria for proposedprinciples of justiceivVACKNOWLEDGMENTSI feel deep gratitude toward my adopted country Canada for allowingmy lifetime dream of studying philosophy to come true. I owe the Universityof British Columbia many years of joy and growth in an atmosphere ofintellectual freedom.I owe thanks to my Committee. Earl Winkler took on my supervisionat a hectic stage in his own career and encouraged my fledgling thesis effortsup until his sabbatical year. Michael McDonald kindly agreed to take over asthesis supervisor, in spite of the crushing demands of his role as Director ofUBC’s new Centre for Applied Ethics. He gave me reams of valuablecomments which challenged me to think harder and dig deeper.Gary Wedeking and Peter Danielson invested hard work in my project,and I sincerely thank them for seeing it to completion. Thanks are also due toJim Dybikowski and Bob Modrow for serving as my University Examiners,and to Wayne Sumner of the University of Toronto for serving as myExternal Examiner.I also want to thank Bob Rowan, my M.A. thesis supervisor, whosepassionate teaching of Plato, Hobbes and Rousseau originally got me excitedabout political philosophy. In his unforgettable words, “Yes, Plato can be shotdown — but not with your guns!”And I wish to acknowledge some special friends for their affirmingsupport in sometimes difficult times: Dennis Bibby, Alice Gmuer, MarciaMonreal, Margaret Nicholson, Randy Reiffer, Antonia Rozario, MakikoSuzuki-Plimley.1INTRODUCTIONInescapable questions confront human communities about how to dividebenefits and burdens among their members. Justice is commonly invoked as a decisiveconstraint on social distributions of valued things. But there are fierce disagreementsabout justice which often provoke festering antagonisms and violent confrontations,even among philosophically informed people of moral goodwill.There has been an upsurge of theorizing about justice thanks to the publicationof John Rawl& A Theory of Justice in 1971. Rawis deserves praise for reopening thequestion of justice at a sophisticated level of argument. It is a striking fact, however,that Rawls’ work has not so much stimulated convergence toward philosophicalconsensus about the just distribution of goods in society, as a proliferation of rivaltheories. 1In response to this fact, my dissertation seeks to develop the concept of ‘rationaldisagreementH, in the hope of overcoming certain forms of conflict which arestandardly couched in terms of justice. The concept of rational disagreement, thoughnot yet prominent in justice theorizing, is not original with me. Some - for exampleLarmore- use it without, apparently, feeling much need to clarify jt.2 Others use moreI A very partial sampling of post-Rawlsian theories of social justice: Ackerman, Bruce A., Social Justice inthe Liberal State, 1980; Ewin, R.E., Liberty, Community, and Justice, 1987; Galston, William A., Justice andthe Human Good, 1980; Gauthier, David, Morals by Agreement, 1986; Heller, Agnes, Beyond Justice, 1987;Narveson, Jan, The Libertarian Idea, 1988; Nielsen, Kai, Equality and Liberty: A Defense of RadicalEgalitarianism, 1985; Nozick, Robert, Anarchy, State, and Lltopia,1974; Phillips, Derek L., Toward a JustSocial Order, 1986; Reiman, Jeffrey, Justice and Modern Moral Philosophy, 1990; Soltan, Karol Edward,The Causal Theory of Justice, 1987; Sterba, James P., The Demands of Justice, 1980; Walzer, Michael,Spheres of Justice: A Defense of Pluralism and Equality, 1983; Winfield, Richard Dien, Reason and Justice,1988.2 Larmore uses the concept in the following context: “If we have it in our power to satisfy the needs ofothers or to prevent their having physical pain, and if the good thereby effected is sufficiently great, wemay well feel obligated to set aside temporarily the pursuit of our own projects. And even if we do notconclude that the greater good overall should prove dedsive, we ought still to feel regret, or offer someexplanation, or perhaps make some amends. I am disinclined to believe that there is any illuminatinggeneral rule to decide such cases (How great should the greater good overall be? How much does it matterhow it is distributed among others?), Some cases may be decidable by judgment; others must be the object ofrational disagreement.”(14 1-2; emphasis added).2or less closely related concepts. Pogge, for example, speaks of “reasonabledisagreement”;3Ackerman speaks of “good-faith disagreement”;4Benjamin speaks of“the ineliminability of rationally irreconcilable moral conflict”;5 and Hardin speaks of“the limits of reason”.6 There is nothing very mysterious about the underlying idea ofsuch talk. Every remotely plausible approach to moral decision-making will have toadmit limits both to its precision and to its power to decide hard cases; and it is naturalto speak of “room for disagreement” in such contexts. But in spite of its centralimportance, the idea of rational disagreement seems to me to have been leftsurprisingly unclear in recent philosophical works on justice.I can attempt only part of this clarifying task here. My aims are essentiallythreefold: to give some sense of the complexities that arise from an attempt to apply theconcept of rational disagreement to disputes about social justice; to identify someplausible sources of rational disagreement; and to reach a provisional verdict aboutsome of the limits to rational disagreement.This inquiry is organized into three main parts. Part I seeks to motivate talkabout “rational disagreement” by sketching the need for such a concept in the context ofcontemporary justice theorizing and by offering a viable working conception of thisPogge’s use of this concept is especially illuminating: “What is needed is the recognition thatknowledgeable and intelligent persons of good will may reasonably disagree about the fundamental issuesdividing the world today. For example,should the means of production be controlled by nationalgovernments, or locally by workers or by private owners? Is the best forum for democratic discussion anddecision making afforded by a single-party, two-party, or multi-party system? Which is more important inthe appraisal and reform of social institutions, the protection of civil and political liberties or thesatisfaction of basic social and economic needs? If only we could understand our disagreements about suchmatters as reasonable disagreements, then we could jointly work toward a world in which alternativeanswers to these questions could coexist in a peaceful, friendly, and supportive internationalenvironment.”(232; emphasis added)Ackerman says this: “And so we come to a first difficulty of second-best government. What decision ruleshould be used when liberal statesmen disagree about the best way to compromise the liberal ideal ofundominated equality? Call this the problem of good-faith disagreement.”(274; emphasis added)5 Benjamin thinks that “An appreciation of the ineliminability of rationally irreconcilable moral conflictwill accord greater importance to the notion of compromise than one generally finds among ethicaltheories.”(2)V6 Limits of reason, according to Hardin, “are the usual cognitive and structural limits to good decision-making.... These include limits on mental ability, limits on time available for deciding, limits oninformation, and limits on relevant theory. Anyother major limit of moral reason that is not usuallydiscussed in this context is the inability of a single individual to determine an outcome independently ofthe actions of others. This is the problem of strategic interaction. ..“(xvii).3concept. Part II and III advance and defend the following theses respectively:1. Setting non-trivial limits to rational disagreement about justice will requiremetaethical commitments. (I will sketch a wide reflective equilibrium-based rationalefor choosing a form of consequentialism over rival theories— but obviously withoutbeing able to give anything approaching a full defense here).2. Once we accept a broadly consequentialist framework for moral thinking, thenthere arise a number of recurrent conceptual, normative and empirical issues whichconstitute plausible sources of rational disagreement about justice. (Many of theseissues will, however, arise mutatis mutandis within certain non-consequentialistframeworks as well.)I launch my search for plausible sources of rational disagreement about justice byexamining two central problems arising from the question of what sorts ofconsiderations are fundamentally relevant to determining justice. These problems arelabelled the criteria definition problem and the criteria inclusion problem (Ch 3.1). NextI explore two closely connected sources of disagreement about principles of justice:disagreement about the appropriate level of comprehensiveness of such principles(Ch 3.2), and about the structure of their justification within a consequentialistframework (Ch 3.3). Four recurrent dilemmas facing theories of justice are brieflycharacterized: the priority problem, the aggregation-distribution problem, thecommensuration problem, and the domain demarcation problem. Finally I try to assessthe possibility of rational disagreement with respect to two of the most prominentdisputes about justice: the dispute between liberal and non-liberal conceptions of justice(Ch 3.4), and the dispute between libertarian and egalitarian conceptions of liberalism(Ch 3.5).I conclude that there is a presumption that only “directly responsive” principlesof justice of a liberal-egalitarian type fall within the scope of rational disagreement(assuming a pluralist society in at least minimally favorable economic circumstances).77 Principles of justice will be labelled indirectly responsive to the existing pattern of sentiment and4Disagreement about competing principles of this kind will generally have to be settledin the light of fairly context-specific consequentialist considerations rather than at thelevel of fundamental moral theory. But even context-specific considerations willfrequently leave room for rational disagreement between mutually exclusive principlesor decisions. Such disagreement must then be resolved at the second-order level ofmoral reasoning, the level of political compromise-seeking and convention-setting —ideally, in clear recognition of the scope for rational first-order disagreement.Three cautionary remarks are in order. First, I cannot possibly defend theinterconnected strands of my account all at once, and must often defer discussions tolater chapters whose results are presupposed in earlier ones. Since a meaningful effortto think clearly about justice requires complex considerations to be kept in mindsimultaneously, there is no ideal order of presentation which can resemble the linearityof a simple argument. Only a view of the whole can illuminate each part.Second, I must presuppose acquaintance with contemporary theories of justice,most importantly John Rawis’ and David Gauthier’s work. There is simply no feasibleway for me to restate their work sufficiently fully here to prepare the ground for mycriticisms. I shall often refer to places in the literature where crucial arguments can belooked up whenever rehearsing them would lead me too far afield.Third, my interest here is not in giving a concise summary and critique of thevast contemporary literature on justice, or even a history of key disputes that seem togo on without resolution. Parts of this immense task have been ably accomplished by anumber of writers.8 I shall assume that the empirical fact of deep philosophicaldisagreement about justice is not in doubt, and focus on my chosen subject ofexamining the possibility that at least some of this disagreement may be rational in theexpectations about justice in a society if they specify a just system of basic socia’ institutions organized as awhole (e.g. Rawis’ approach). By contrast, principles of justice will be labelled directly responsive to thispattern if they specify particular characteristics which individuals (or groups) must have for justlyclaiming a particular share of social resources. (See Sec 3.1.3 §3 below.)8 See e.g. Philip Pettit, Judging Justice (1980); Will Kymlicka, Contemporary Political Philosophy (1990);or Tom Campbell, Justice (1988) for admirable restatements and criticisms of prominent theories of justice.5relevant sense.The dilemma which this sort of synoptic inquiry must face is that many issuesare both centrally relevant to it and yet far too complex to pursue to a desirable degreeof depth. I cannot follow many contemporary writers on justice into what I judge to beultimately blind alleys, but must be highly selective with respect to the issues I engage.All I can hope to do here, on many issues, is to establish basic bearings. My project ismotivated by a fundamental conviction, however, that finding our basic bearings in thewelter of conflicting ideologies is more important than to erect blinkers around somelittle corner of it. The scope of this inquiry is broad; but when it comes to contemporaryjustice theorizing, I believe that a broad view is required to see any light at all.6PART 1: THE CONCEPT OF RATIONAL DISAGREEMENTThis part seeks to motivate the concept of rational disagreement in thecontext of contemporary justice theorizing; to develop a viable working conceptionof this concept; and to confront several serious objections to the project ofidentifying sources of rational disagreement about justice.1.1 THE PHILOSOPHICAL CONTEXT§1 What sense can we make of the vast philosophical literature on justice?How much authoritative clarity about the demands of justice can decision makersgain from it? Are additions to the already crowded field of competing theories ofjustice likely to resolve conflicts about justice or more likely to exacerbate them?What sort of philosophical endeavor is now most likely to advance this field? Theseare my motivating concerns.Two kinds of endeavor seem obviously worthwhile. The first would be tooffer an adequate theory about the just distribution of goods in society. The second isto point out inadequacies in various proposed theories. What are the prospects forsuccess in either of these endeavors?It is a monumental challenge to offer an adequate theory of social justice: atheory, which no informed person of good will could reasonably reject, for a set ofusefully determinate principles to decide the justice of major social distributions.Taking on this challenge is easily confused with a far less difficult one - that ofproposing new principles of social distribution and defending them as no lesspermitted by reason than those we already have. My own initial desire was, in fact,to defend some principle as the following:Distribute economic goods in such a way as to ensure the most modest levelof material prosperity which is consistent with good health, and which hasthe psychological effect of minimizing preoccupation with materialacquisition while at the same time maximizing opportunities forexperiencing the beauty of unspoilt nature, the delights of art and scientificdiscovery and warm human relationships to the utmost compatible degree.7I came to realize, however, that there was little hope of justifying my favored socialvision as a principle of justice which any informed person of good will would haveto accept. Since a wide range of conflicting visions of the ideal society already exist, itseems objectionable to advance any new vision of this kind as a theory of justice -unless one is prepared to accept the burden of arguing that it is the best social vision.And, as I hope to make clear, any such argument faces immense problems.The second kind of endeavor, which makes up the bulk of current justicetheorizing, consists in detailed critiques of proposed theories of justice. Suchcritiques come in two grades of severity: let me call them “fatal” and “non-fatalt.Successful criticism of the non-fatal kind merely undermines a theory’s claim torational superiority and reduces its rational status to that of one competing optionamong others. Successful criticism of the fatal kind undermines even the modestclaim of a theory to being one competing option among others. Stated differently,non-fatal criticism merely aims to show that not all informed persons of good willmust accept the theory in question, while fatal criticism aims to show that everysuch person must reject the theory. The distinction between these two kinds ofcriticism, although rarely drawn explicitly, is crucial for my purposes, because nonfatal criticism places a theory squarely within the scope of rational disagreement.Many allegedly “devastating” criticisms of theories of justice are, in fact, charitablyinterpreted to be of the non-fatal rather than the fatal kind. They are devastatingonly to a theory’s monopoly claim to rational superiority. While all theories ofjustice qua theories of justice must, I think, make precisely this monopoly claim, itdoes not follow that undermining any such claim “refutes” the theory, in everyrelevant sense of this ambiguous term. Criticizing a theory for not compellinguniversal rational assent is not equivalent to having raised a fatal objection to it, aslong as the criteria for having justified a theory of justice as superior to all rivaltheories are themselves in dispute.8With practice one quickly gets proficient at spotting non-fatal defects in anytheory of justice yet proposed, because every such theory can plausibly be interpretedas sacrificing something of great moral importance in order to gain something else.But those who do not share each other’s fundamental normative perspective arerarely converted. Up to a point, the debate between defenders of various theoriesand their critics helps clarify the implications of each perspective, but I cannotdiscern much progress toward rational consensus in the predictable ritual whichnow unfolds every time someone proposes yet another theory of justice.Has justice theorizing reached a dead end? This is hardly an unreasonablesuspicion. Appeals to “justice” provoke flaming passions and festering conflictsthroughout the contemporary world- yet the current normative disarray in thephilosophy of justice must be disheartening for decision-makers seeking moralguidance. Still, if philosophers cannot illuminate the nature of justice, who can?In the face of this seeming impasse, a third endeavor in the field of justicetheorizing appears to me promising: to investigate the sources of normativeindeterminacy in this field in a more systematic way, with the goal ofconceptualizing the sources and limits of rational disagreement. How muchdisagreement about justice does reason permit? What are the conceptual andnormative factors on the one hand, and the intractable empirical factors on theother, which so often lock informed people of good will into bitter disagreementabout justice? Working toward an understanding of some important sources ofrationally irremediable disagreement about justice— this is the line of inquiry Ipropose to pursue here.§2 There is nothing strikingly new about a focus on the problem of sortingout the conceptual, normative and empirical factors that may allow a variety of9conflicting but equally rational beliefs about justice.1 Serious writers on justice, andmost impressively John Rawls, have explicitly acknowledged limits to rationaldecidability.2But efforts at determining the sources and limits of the rationallyundecidable tend to be selective and inadequate. The all-important distinctionbetween the rationally required and the merely permitted is not easy to draw, andoften not clearly acknowledged by writers who must surely be aware that the weightof reasons favoring their particular views about justice is far short of compelling.I want to group contemporary philosophical work on justice into three broadcategories. There are developed normative theories which generally includecritiques of major alternatives (e.g. Rawis, Nozick, Dworkin, Ackerman, Galston,Nielsen, Phillips, Walzer, Gauthier, Narveson, Reiman, Sterba and others). Thenthere are many detailed critiques and critical surveys of theories of justice, oftenstrongly informed by normative commitments without, however, presenting themas a developed alternative to existing theories (cf. Barry, Sandel, Wolff, Maclntyre,Larmore, Raz, Fishkin, Pettit, Campbell, Kymlicka and others). And in additionthere is a small but growing “counter-literatur& to the almost obsessive emphasison the construction and refutation of ambitious theories of justice in contemporarypolitical philosophy, a literature that tries to evaluate the limits of theory1 Hume gives the following as an example of what reason alone cannot determine in a dispute about justproperty claims (although his example is admittedly fairly minor and has to do with the need forconvention-setting):“Sometimes the interests of society may require a rule of justice in a particular case; but may notdetermine any particular rule, among several, which are all equally beneficial...Sometimesboth utility and analogy fail and leave the laws of justice in total uncertainty. Thus, it ishighly requisite, that prescription or long possession should convey property; but what numberof days or months or years should be sufficient for that purpose, it is impossible for reason aloneto determine.”(An Enquiry Concerning the Principles of Morals, 29-30).2 Rawis’ painstaking analyses of the “priority problem”(40f) and the “index problem”(93f) in ATheory of Justice remain guiding examples of the sort of project I think needs to be carried through moresystematically, in the light of the criticisms Rawis’ own theory has received. An article that isexceptionally sensitive to problems of achieving “political consensus in the face of fundamental moraldisagreement” is Amy Gutman and Dennis Thompson’s “Moral Conflict and Political Consensus” (1990).Such “second-order” agreement-seeking, however, presupposes some consensus on the limits to firstorder agreement.10construction by reassessing fundamentals of rational justification in moral andpolitical theory (I have in mind, for example, Mapel’s Justice Reconsidered, Shklar’sThe Faces of Injustice, as well as recent work by Sumner, Hardjn and Haslett).3How do I understand my own project in relation to this literature? Clearly, itfits the third category best. My hope is to reach at least some tentative conclusionsabout the limits of justified belief about justice, in the light of existing theories ofjustice and fundamentals of moral justification. The larger enterprise to which Ihope to contribute is to find ways of transcending the current impasse in justicetheorizing. This is best attempted, in my view, by first seeking a clearerunderstanding of rational disagreement about justice. It is hard to see how such anunderstanding could evolve without renewed efforts to rethink justificatoryfundamentals with a synoptic view of the shortcomings of existing theories ofjustice. But the issues involved in this enterprise, as I came to discover, are sorelentlessly complex that it is perhaps impossible for any single mind to encompassthem in their fullness.§3 I must immediately address three misconceptions of my project. I shallargue that there is room for rational disagreement about justice - which is not thesame as the skeptical claim that no usefully determinate principles of justice canpossibly be justified as rationally required or the dogmatic claim that I can give afinal verdict on what issues are rationally undecidable. It will be instructive toconsider these misconceptions briefly.First: I am not arguing the thesis that being “usefully determinate” and“rationally required” are mutually exclusive properties of principles of justice, butsomething much weaker: that on a plausible account of the nature of moralreasoning, far greater disagreement is rationally permitted about the demands ofL.W. Sumner, The Moral Foundations of Rights (1987); R. Hardin, Morality Within The Limits OfReason (1988); D.W. Haslett, Equal Consideration: A Theory of Moral Justification (1987).11justice than seems to have been contemplated by many contemporary theories ofjustice.Even if all theories so far proposed can be shown to be inadequate, that doesnot imply the impossibility of an adequate theory. Obviously, the idea of such atheory is not self-contradictory, and only self-contradictory ideas are demonstrablyunrealizable. My project does not imply that an adequate normative theory ofjustice is impossible. But by taking more seriously the possibility of rationaldisagreement about justice, we may clarify the limits of what anyone can expect toachieve in this field. It may be possible to develop concepts and arguments whichenable us to repudiate, with great confidence and clarity, exaggerated claims made bymany people about the rational status of their favorite conception of justice.Second: I cannot offer a final verdict on what disagreements about justice arerationally undecidable- I can only offer a list of what I judge to be plausiblecandidates for the status of rational disagreements. The concept of rationaldisagreement permits rational disagreement about what is to count as rationaldisagreement. But this fact does not, I believe, lead to a vicious regress; it ratherconfirms how indispensable the concept of rational disagreement is.Others will find shortcomings in my list of plausible candidates for rationaldisagreement which may lead me to revise this list. All value judgments arisenecessarily from someonetperspective. But judging an issue explicitly to be amatter of rational disagreement invites a sharing of perspectives about the questionof whether we sincerely believe this issue to be resolvable by information,arguments or methods currently at our disposal.§4 We must, however, immediately draw an important distinction between“rational disagreement” and “disagreement among rational people”. Disagreementamong rational people is disagreement which informed people with impeccablerational credentials carry on in good faith by the means of rational persuasion and12which they judge to be resolvable, given currently available information, argumentsand methods (otherwise they would find it pointless to carry on their dispute bysuch means). In contrast, rational disagreement - in a sense to be further clarifiedbelow - is disagreement which (although we may not be conscious of it) we havereason to judge to be irresolvable at the present, because we have exhausted allrelevant and epistemically accessible information, arguments and methods.Rational disagreement could be seen as a form of fruitless disagreement which isnot appreciated as such.That there exists disagreement among rational people about some issue is,therefore, neither a necessary nor a sufficient condition for this disagreement to berational disagreement, in the sense of Hrationap? which interests me here. It is notnecessary (i.e. not all rational disagreement is disagreement among rational people),because the fact that a disagreement is not carried on by informed people of goodwill through means of rational persuasion clearly does not affect the questionwhether this disagreement involves irresolvable conceptual, normative orempirical issues. And it is not sufficient (i.e. not all disagreement between rationalpeople is rational disagreement), for at least three possible reasons: (1) Most peoplewho freely invoke the concept of justice are surprisingly unsophisticated about theconceptual, normative and empirical sources of rational underdetermination in thisfield. In fact, many undoubtedly reasonable people seem not to have seriouslycontemplated the very possibility of rational disagreement about social justice, or atleast they often do not realize that a particular issue is a matter of rationaldisagreement. (2) Many issues of social justice are both enormously complex andemotionally absorbing. Progress is often possible in settling disputes about justice byappeal to rational argument. But there is a hard-to-define point at which manydebates cease to explore the limits of rational agreement in constructive ways andinstead lapse into a sterile repetition of inconclusive considerations. Even peoplewith impeccable rational credentials cannot always be expected to realize when this13point has been reached, especially given the primitive state of our vocabulary forarticulating varieties of rational underdetermination. (3) There may in turn beroom for rational disagreement about whether a particular issue is, in fact, a matterof rational disagreement. Whether all relevant and epistemically accessibleinformation, arguments and methods have been brought to bear on an issue is not aquestion whose answer is always clear-cut. Concepts like “relevance” and “epistemicaccessibility”, although indispensable in many contexts, are notoriously difficult toclarify. Clearly, how inaccessible a particular piece of information is may depend onthe effort we are prepared to make in getting it. If the cost of getting some particularinformation is prohibitive in relation to the value of goods whose just distributionis in dispute, then obviously such information is, for practical purposes, inaccessible.To the extent that the concepts of “relevance” and “epistemic accessibility” are notsharply bounded, the distinction between what is inside and outside the scope ofrational disagreement will also be somewhat hazy and shifting. But from the factthat a distinction is not sharp it never follows that drawing it is unimportant.Persistent disagreement among informed people of good will provides at bestevidence, not proof, that some issue is a matter of rational disagreement, in therelevant sense of “rational”. For someone to suggest, therefore, that a particularissue is outside the scope of rational disagreement implies neither that there is nolonger any dispute among informed people of good will about this issue, nor thatcarrying on such disagreement is necessarily irrational in every sense of thisambiguous term (see below).My claim is not that focussing on the question, “How much room is there forrational disagreement about social justice?” is a magic key to the ultimate truthabout justice or that it will transport us to a vantage point beyond ideology. Myclaim is simply that conceptualizing certain recurrent disputes about justice asrational disagreements defines a constructive line of effort for philosophicaldialectic - probably our best chance for progress. Those who are strongly committed14to their favorite normative perspective may well resist the possibility of rationaldisagreement. It is my goal to challenge the rationality of this kind of meta-positionabout rational alternatives without, however, embracing all-out skepticism.1.2 A CONCEPTION OF RATIONAL DISAGREEMENT§1 How can “rational disagreement” be defined? To understand what aconcept includes it is often helpful to ask what it is meant to exclude. No doubtdisputes about justice often result from the fact that the disputants are ignorant ofrelevant and accessible information, insincere in their professed concern for justice,overwhelmed by passion, misled by authority figures or otherwise clearly deficientin their reasoning powers. These and similar factors are common sources ofdisagreement about justice. But claiming that the intelligence, rationality orsincerity of some or all of the disputants is somehow blocked, that they must bestubborn, uninformed or misinformed, shortsighted, biased or otherwise mentallyblighted cannot plausibly explain the extent of good-faith disagreement about justiceamong informed people with impeccable rational credentials.The most satisfactory way to mark the distinction between “rational” and“irrational” forms of disagreement about justice would perhaps be to give anexplicit, uncontroversial definition of rationality which, when applied to problemsof justice, will neatly separate disagreements that are rational from those that arenot. But to my knowledge, no such definition of rationality exists. An extendedinvestigation of models of rationality, although relevant to my task, would quicklylead me too far afield.4Nevertheless, I have in mind what seems to me a tolerably clear and viableconcept of “rational disagreement”. I want to risk an explicit definition, realizingCf. e.g. H.I,Brown’s interesting revisions of the traditional model of rationality in Rationality,London: Routledge, 1988.15that both my definition and the clarifying remarks which follow give rise tonumerous questions:Rational disagreement about justice is either currently or fundamentallyirremediable disagreement between (proponents of) basically rationalconceptions of justice. It can be of a conceptual, normative or empirical kind.Rational disagreement can be regarded as fundamentally irremediable if wehave reason to think (whether we are conscious of it or not) that there arebasic conceptual or normative points at issue which no relevant information,clearer reasoning or conceivable method will ever settle, or if there areintractable empirical points at issue which would require epistemicallyinaccessible information. Rational disagreement can be regarded as currentlyirremediable if we have reason to think there are empirical or non-fundamental conceptual or normative points at issue which additionalinformation, clearer reasoning or conceivable methods will settle eventually.I shall assume that the distinction between the conceptual, the normative andthe empirical as marking off three contrasting categories of disagreement issufficiently settled in principle, without suggesting that it is always easy to draw inpractice. But what is a basically rational conception of justice? Is rationaldisagreement completely relative to whatever conception of justice one adopts?How can we know which disagreements are remediable or irremediable in thesenses defined? Why does my definition of rational disagreement includedisagreements which are likely to be remediable eventually by the usual methods ofrationally resolving such matters? I address each question in turn.§2 What is a basically rational conception of justice? I must answer byanticipating later results of this inquiry. A basically rational conception of justice is(i) a set of bona fide principles of justice which (ii) are consistent with a rationalmorality. A bona fide principle of justice is a principle which specifies a reasonablecompromise among competing justice-based claims of different individuals in somedistributive domain. Justice-based claims are claims to shares of societyt resourceswhich individuals (or groups), in pursuit of their well-being, press as their due onthe basis of just-making criteria such as desert, need, equality, rights or good-faith16expectations. A distributive domain is any distributive context (e.g. social roles,institutions, geographic areas, ethnic groupings) which, ideally, is sufficiently well-demarcated that outcomes or processes for the just distribution of valued things canbe specified within it. A principle of justice embodies a reasonable compromiseamong competing justice-based claims in some distributive domain if it is“appropriately responsive” (in a sense to be explained) to patterns of informedexpectations and sentiment about justice and can be approximated by some feasibledistributive mechanism.A basically rational conception of justice, I said, must both include bona fideprinciples of justice and be consistent with a rational morality. But talk of a “rationalmorality will immediately arouse the deepest suspicion. If we must first take sidesin intractable disputes in metaethics before applying the idea of rationaldisagreement to justice, have we not begged the most interesting questions?Informed people of good will are no more agreed on the ultimate nature of moralitythan they are on the demands of social justice. How then can we presume to judgewhat makes someonetunderstanding of the nature of morality rational orirrational?While deferring my full answer to this important objection to Part II, I mustissue at least a promissory note. We must provisionally introduce a moralframework to fix ideas and to define non-trivial limits to rational disagreement(otherwise this concept will simply become synoymous with skepticism aboutjustice). Many disputes about justice arise because disputants have fundamentallyconfused views about morality. Rational disagreement about justice, in the sense Ihave in mind, is possible only between disputants who hold basically rationalconceptions of justice- or whose dispute does not turn on the fact that they don ‘t.This last qualification is important for the feasibility of my project despiteintractable disputes in metaethics. While I can define limits to rationaldisagreement about justice only at the cost of making certain metaethical17commitments, the broadly consequentialist framework I shall adopt does not beg allthe interesting questions. For one, many theories of justice ruled out here asincompatible with this framework will contain plausible sources (i.e. plausiblecandidates for the status) of rational disagreements which are more or lessanalogous to those I shall discuss here.Therefore, what counts as a matter of rational disagreement, in the sense Ihave in mind, is not strictly relative to whichever particular conception of moralityor justice one adopts. If it were, then obviously no greater consensus could beexpected about the scope for rational disagreement than about those conceptionsthemselves. But many of the fundamental problems afflicting theories of justice —such as the justification problem, the priority problem, the aggregation-distributionproblem or various commensuration problems — can be appreciated as rich sourcesof rational disagreement from within many different conceptions of justice ormorality. The prospects for achieving at least partial consensus on what issues aresensibly to be regarded as matters of rational disagreement about justice are farbetter, I think, than the prospects for achieving consensus on the best conception ofjustice (although achieving greater consensus of the first kind may ultimatelypromote greater consensus of the second kind).It is important to realize that the idea of rational disagreement about justice,while presupposing commitment to a strong enough moral framework to definelimits to rational disagreement, does not presuppose similar commitment in orderto identify plausible sources of rational disagreement. Even theories of justice whichare incompatible with the broadly consequentialist moral framework I shall adoptwill, in most cases, contain sources of rational disagreement which are closelyanalogous to those I identify here.18§3 How can we know which disagreements are remediable or irremediable inthe senses defined? My project is to find reasons for considering certain issues to beplausible candidates for the status of rational disagreements, without claimingepistemic finality for my findings. If a particular debate about what justice requiresin some particular respect has divided informed people of good will for a long time,and if we can identify conceptual, normative or empirical issues about which, uponreflection, most of us are of divided minds within ourselves, then surely at somepoint we have reasonable support for the hypothesis that perhaps some of theseissues cannot be rationally resolved, at least not currently, or not at the level offundamental theory, or not by appeals to the concept of justice.5 This is not to saythat they must be irresolvable by any method whatever which we might broadlyconsider rational. Even if we came to realize that the demands of justice, or even ofmorality in general, are indeterminate in certain respects, it will generally still berational to solve our disagreements by compromise rather than force. But a rationalresolution to an issue, in this purely pragmatic sense of “rational”, does notpreclude continued rational disagreement about this issue, in the sense of “rational”of interest to me here.§4 Why does my definition include as “rational disagreements” not onlyirremediable disagreements, but also those which seem to us remediable eventuallyby the usual methods of clearer reasoning and better information? I have definedthe concept to reflect the purpose I have in mind for it. Surely we want to useTwo important (and easily confused) distinctions are that between fundamental moral theory andcontextual judging, and that between the first-order and second-order level of moral reasoning.Fundamental theory, unlike contextual judging, abstracts from empirical particulars of specific contextsand assumes only general information about the nature of reality, knowledge and human values. Thefirst order level of moral reasoning, unlike the second-order level, is the degree to which moralquestions can be settled by appeal to fundamental theory or contextual judging, but without resorting tobargaining or other forms of purely pragmatic consensus-seeking in the absence of principled guidance.So not all matters which cannot be resolved at the level of fundamental theory must necessarily beresolved at the second-order level of moral reasoning: often, principled ways of contextual judging arealso possible. And of course, second-order reasoning will in turn be constrained, in some ways, byfundamental theory.19“rational disagreement” in a sense in which saying, “Perhaps there is room here forrational disagreement” serves to signal to informed people of good will that wethink all pertinent considerations and currently accessible information have beenbrought to bear on an issue without settling it, and that the debate has reached astalemate of rehearsing inconclusive arguments over and over again. It is meant tosignal the need to shift the debate to another level of reasoning, the second-orderlevel of pragmatic consensus-seeking.6If I restricted the concept of rationaldisagreement so as to include only fundamentally irremediable disagreement, theresulting implications of normative finality would make its use far morecontroversial and would tend to shift debates simply to whether the disagreement atissue must, in fact, be forever irremediable. And this debate is likely to be asinconclusive as the earlier debate about what justice requires because, even if it istrue that no amount of relevant information or clearer reasoning will ever settlesome disagreements, I believe that we can rarely know this truth with certainty. Ourcapacity for empirical discovery, and perhaps to a lesser extent our conceptual andmoral frameworks, evolve in unpredictable ways and may change the terms inwhich we see our most perplexing dilemmas.But what about disagreements which seem irremediable at some abstruselevel of theorizing but lead to normative agreement in practice? Moral reasoning ismeant to solve practical conflicts, and without a practical conflict, there is no6 am assuming that civil war or other brute-force attempts at dealing with first-order rationaldisagreement will, in most contexts, be ruled out by a basically consequentialist moral framework (andgenerally even by the informed self-interest of all parties involved). My claim, therefore, is not thatappeals to justice must be altogether suspended at the second-order level of moral theory and give wayto the use of force or threat of force. My claim is only that questions about the justice of variousstrategies of negotiating the terms of mutual accommodation in recognition of the scope of rational first-order disagreement are different from questions about the fundamental constraints imposed by justice ondistributive outcomes and processes.But clearly, the first-order/second-order distinction between levels of moral theory dependsitself on a prior understanding of the concept of rational disagreement. Its normative content will,therefore, become clear only in the course of our inquiry. In what follows I shall be concerned purelywith sources and limits of rational disagreement at the first-order level of moral reasoning— althoughthe problem of rational disagreement may, of course, reoccur in regard to the justice of various possiblenegotiation strategies for second-order accommodation in the face of rationally irremediable first-orderdisagreement about justice.20problem for moral reasoning to solve. Why concern ourselves at all withdisagreements which dissolve at the applied level?7It is a fortunate experience indeed that many metaethical and high-levelnormative differences do not, for a number of reasons, preclude agreement at theapplied level. Different lines of moral reasoning often converge on the samedecision. People with clashing metaethical convictions often share similar moralintuitions about some particular case. Supporters of different ideologies experiencepressure to form coalitions with a common platform. And often a compromise onsome second-best outcome will have clear advantages, from everyone’s point ofview, over escalating the level of conflict. But the way I propose to use this concept,“rational disagreement” refers to currently or fundamentally irremediabledisagreement about conceptual, normative or empirical issues at whatever level ofmoral reasoning they occur. Moral differences at various abstract levels oftenconverge on agreement at the applied level, but in many cases they do not, and theyare deep sources of many festering conflicts dividing human communities. In anycase, agreement at the applied level despite strong disagreements at a more abstractlevel (e.g. the agreement of many conservatives and feminists about the desirabilityof outlawing pornography) is bound to create at best temporary moral bedfellows.But is perhaps the entire search for abstract principles of justicefundamentally misguided? Jonsen and Toulmin distinguishtwo very different accounts of ethics and morality: one that seeks eternal,invariable principles, the practical implications of which can be free ofexceptions or qualifications, and another, which pays closest attention to thespecific details of particular moral cases and circumstances.(2)8As will become clear, I fundamentally agree with Jonsen and Toulmin in endorsinga version of the second approach to moral decision-making. But I think we must be7 Cf. Jonsen and Toulmin, The Abuse of Casuistry (1988).8 Numbers in brackets following quotes give relevant page numbers in an author’s work, as listed intheBibliography. Titles or year of publication are given when more than one work by an author islisted. All references to Rawis’ work are to A Theory of Justice (1971), unless a year of publication isgiven.21careful not to draw a false dichotomy between case-based and principle-based ethics,as if our judgment in particular cases were dispensable in the search for principles oras if an appeal to principles were dispensable for our judgments in particular cases.The understanding of moral justification I shall defend, following Rawls, is thesearch for convergence among informed people of good will of their wide reflectiveequilibria between principles, judgments in particular cases, and backgroundtheories. On such a coherentist view of moral justification, both moral judgmentsand principles are rational according to the degree to which they mesh with amaximally extensive and coherent belief system (see Part II).1.3 OBJECTIONSJustice theorizing is a philosophical mine field. Almost every constructiveattempt confronts paralyzing objections.9The following objections have been urgedagainst my project, all of them serious enough that I can hardly proceed furtherwithout confronting them:Having decided that a disagreement looks irremediable, would that fact alonenot be sufficient to make continued disagreement irrational? If so, the conceptof rational disagreement would seem to be straightforwardly selfcontradictory Or if not self-contradictory, it would seem to lack sufficientunity for a meaningfully focussed project. In any case it does not seem to beinterestingly different from skepticism. And at best, a project of identifyingpoints of rational disagreement about social justice is a mere exercise intaxonomy or semantics.§1 Is the concept of rational disagreement self-contradictory? Having decidedthat a disagreement cannot be settled by drawing on information, arguments ormethods currently at our disposal, would that fact alone not suffice to makecontinued disagreement irrational? The very idea of Hirremediability?l would seemAt least two prominent contemporary justice theorists - Rawls and Nozick - have found it necessary torecant substantial parts of their original theories (cf. Rawis, 1985, 1988; and Nozick, 1989, 286-296).Gauthier is, I think, also forced to make major concessions (cf. Vallentyne, 1991, 25; 323-330).22to limit the forms of disagreement that could be rational. It is surely irrational topersist in arguing once one realizes that one’s own position is not backed up bybetter reasons than one’s opponent’s position. Suppose it could be shown ofconflicting positions about justice that none is ultimately rationally superior. Thisfact would imply, not that ongoing disagreement between proponents of opposingpositions is rational, but on the contrary, that persisting in such disagreement wouldbe utterly irrational. Therefore, the project of identifying points of rationaldisagreement about social justice, if successful, would make rational disagreementimpossible. Therefore, my project is self-defeating.This objection depends on the fact that the term “rationalt’is, of course,multiply ambiguous. If there are arguments in a dispute, understood by all parties,that can settle the issue in favor of one side, then we would surely want to callcontinued disagreement over this issue “irrational”, in some sense. But if there areno arguments in a dispute that can settle the issue in favor of one side, and allparties understand this fact, we would also want to call continued disagreementover this issue “irrational”, though clearly in another (incompatible) sense.1°Wemust, therefore, distinguish two levels of normative disagreement: “first-order”disagreement at the level of fundamental theory or contextual judging, and“second-order” disagreement at a meta-level which recognizes the scope and limitsof first-order agreement. Disagreement which is irrational if judged from this metalevel may not be irrational if judged from the level of fundamental theory. Therecognition of rational agreement at the level of fundamental theory should indeed10 In view of the multiple ambiguities of “rational’, would it perhaps be clearer to speak of “moralundecidability”, and to abandon talk of “rational disagreement” altogether? (I owe this suggestion toPeter Danielson.)I am not convinced that such a terminological switch would more clearly articulate my centralthesis: the claim that, at the root of many disagreements about justice among informed people withimpeccable rational credentials, there often are unrecognized or as yet poorly understood conceptual,normative or empirical sources of rational underdetermination. The undecidability of certain moralissues is merely the effect of rational underdetermination - which means that, in our search for theunderlying causes of certain disagreements about justice, there is ultimately no way around makingclaims about what rationality does and doesn’t permit. (I shall argue in Part 2 that my choice of a fixedmoral framework is itself constrained by rationally non-arbitrary considerations.)23produce agreement at the meta-level that an issue must be resolved by other meansthan purely an appeal to fundamental theory. It is precisely my hope that a vastlyclearer philosophical (and eventually public) recognition of the rationalundecidability of many issues of justice at the level of fundamental theory or evenat the level of contextual judging will motivate agreement on the need forconsensus-seeking at the second-order level, free of mutual accusations of injustice.If the search for consensus at the second-order level is not simply to degenerate intoa power struggle, however, it must be informed by an understanding of the reasonsfor first-order disagreement and itself be constrained by procedural principles ofjustice. Ultimately the hope is that all rational disagreements about social justicewill turn out to be “remediable” in at least this sense - that informed people of goodwill can recognize the ultimate reasons for a disagreement, recognize the impotenceof fundamental moral theory or contextual judging to resolve this disagreement,and settle at the level of negotiated compromise. This does not mean that it cannotbe rational for people to persist in their commitments to cherished personal valuesdespite the realization that these commitments are irremediably underdeterminedby reason.11 It does mean, however, that insofar as one’s cherished but rationallyunderdetermined commitments conflict with other people’s equally cherished andunderdetermined commitments, the recognition of this underdeterminationshould cause a readiness, in informed people of good will, to find a modus vivendiwhile constricting the legitimacy of appeals to justice.This claim does not imply, of course, that the recognition of rationaldisagreement in a given case necessarily makes acting on one’s preferred course ofaction irrational. On any plausible account, considerations of justice are generallyneither necessary nor sufficient for determining what makes actions rational orAs Larmore puts this point, “The fact that a conviction of mine about the meaning of life iscontroversial, rejected even by others whom I consider reasonable, may not offer me a sufficient reason tosuspend belief in it, if it continues to make sense of my experience.”(52)24irrational. I certainly do not make the sweeping claim, therefore, that recognizingthe existence of rational disagreement about one’s views about justice makes itgenerally irrational, from the perspective of some individual’s (or group’s) interestsor desires, to act on these views. I only make the vastly more modest claim that suchrecognition makes it irrational to think that justice requires one’s preferred courseof action.§2 Does the concept of rational disagreement lack sufficient internal unity fora meaningfully focussed project? I have proposed to include certain empirical aswell as conceptual and normative disputes within the extension of “rationaldisagreement”. But does such a concept not verbally unify what is, in substance,quite disparate? In the case of empirical disputes we disagree about cause and effectin complex social situations but agree on what truths would resolve them; whereas,in the case of fundamental conceptual and normative disputes, there is sometimesno truth to be found, or perhaps we disagree even on that.Let me make this objection as clear as I can. Is an inquiry with “rationaldisagreement” as its unifying concept a little like a book with “banks” as its unifyingconcept which includes chapters about lights or other objects arranged in neat rows,financial institutions, and the edge of water bodies? If I thought so, I could limit theextension of my central concept so as to exclude empirical issues. But I amconvinced that this would be a mistake because, when it comes to disagreementsabout justice, conceptual, normative and empirical questions are often inseparablyintertwined. For example, is it purely an empirical question what will in practicesatisfy the idea that each person can justly claim an equal right to the most extensivescheme of equal basic liberties compatible with a similar scheme for all? Is it purelyan empirical question whether mass starvation in the world should beconceptualized as a problem of overpopulation or of food distribution? Even wherethe empirical questions can be separated from the conceptual or normative ones, the25information required to settle some of the fiercest disagreements about justice isepistemically so inaccessible to us that we may have to classify them asfundamentally irremediable. For example, what exact mix of public and privateownership of the means of production will lead to whatever we may plausiblyspecify as the most desirable consequences overall? Or what total system of socialinstitutions will, over time, make the economically poorest ten percent of thepopulation as well off in absolute terms as possible? Might informed people of goodwill not be in (first-order) disagreement about the best answer to these questionsforever, simply because of the epistemic difficulties of assessing the differentconsequences of all the various options? Questions of social justice presentthemselves as a bundle of intertwined empirical, conceptual and normative threads,and trying to untangle the conceptual and normative threads in isolation from theempirical can’t work. Ideal theorizing that is not at least roughly guided by feasibilityconsiderations has led, I believe, to a great deal of remoteness from the real-worlddebates about justice and has made many theories of justice, in Mapel’s terms,“unworkably complex” or altogether “senseless”.12 My inquiry will, therefore, notexclude disagreement about intractable empirical matters from the rubric of rationaldisagreement.§3 Is rational disagreement about justice simply skepticism by another name?The possibility of rational disagreement would seem to imply that there can beconflicting principles of justice which are equally rational. Whether something isreasonably to be regarded as just would no longer seem to be the sort of question12 An illuminating critique of theories of justice whose proposed “metrics of equality” are either“senseless” or “unworkably complex” is developed by David Mapel in Social Justice Reconsidered(1989). According to Mapel, “A metric is senseless if it relies on imaginary calculations that areimpossible or if it fails to identify any common currency by which the relative value of different sortsof resources might be commensurated. A metric is unworkable if it requires information that is toocomplex and counterfactual to lead to practical pnnciples.”(6) He identifies crudal parts of Dworkin’s,Gaiston’s, Ackerman’s and Rawis’ theories of justice as senseless or unworkable.26that always has a disjunctive yes or no answer. How, then, is recognizing rationaldisagreement interestingly different from skepticism?Skepticism about justice can either mean that there are no true principles ofjustice to be discovered which exist in the nature of things independently ofpeople’s shifting understanding of what is likely to promote their well-being. Or itcan mean that there is no room between mind-independent truth and meresubjectivist preference for constructing a shared understanding of justice in a waythat will be motivationally potent for rational persons. I admit the first but deny thesecond. Recognizing rational disagreement implies only that some conflictingconceptions of justice are equally rational, but not that all possible conceptions are.Skepticism about justice is certainly an understandable reaction to the normativedisarray in current justice theorizing as well as to the rather indiscriminate use ofjustice-talk in everyday life. It is part of my aim in this inquiry to keep suchskepticism at bay by retrieving a relatively modest rational status for appeals tojustice.§4 Is the project of identifying sources and limits of rational disagreementabout justice a mere exercise in taxonomy or semantics? Just as we have madeprogress in the field of practical reasoning by labelling certain recurring fallacies, wecan perhaps make progress in the field of justice theorizing by clearly namingrecurring reasons for seemingly interminable disagreement among informed peopleof good will. There is power in naming; and by more clearly conceptualizing suchrecurrent sources of disagreement about justice - such as, for example, the criteriadefinition and inclusion problems, the justification problem, the priority problem,the aggregrega tion-dis tribution problem, various other commensura tion problems,or the description-relativity problem - we may be better able to discern the presenceof these problems in specific normative disputes. Although initial attempts atmapping plausible sources and limits of rational disagreement will seem to yield27only a rather unruly collection of points, clearer ordering principles may emergelater on, as our insights deepen about the ultimate nature of such disagreements.Attempting to sort out the ways in which informed people of good will canreasonably disagree about what justice requires is not, I believe, merely semanticallydifferent from those approaches to social justice which offer a closed set of principlesspelling out what justice requires. As Pogge puts it,What is needed is the recognition that knowledgeable and intelligent personsof good will may reasonably disagree about the fundamental issues dividingthe world today (232).Despite the diversity of views about justice, the possibility of rational disagreementis often not conceded, and neither the sources nor limits of such disagreement seemto me to have been investigated in enough depth and detail. Progress in justicetheorizing may need an expanded vocabulary - a vocabulary to characterizenormative indeterminacies in a way that allows people to express their first-orderdifferences about justice without seeming to draw each otherTsrationality or goodfaith into doubt. It would clarify the distinction between first-order and meta-leveldisagreement. It would not presuppose the adequacy of truth-valuationaldichotomies for judgments about justice.The concept of rational disagreement, I believe, deserves a central place indebates about justice. If patterns of rational underdetermination could beconceptualized as focal points in work on justice and widely taught - much like theroster of fallacies taught in practical reasoning courses - it might dislodge some ofthe dogmatism afflicting many disputes about justice. It might defuse conflicts byturning disagreements between people into uncertainty within each person. I see nomore promising way in which many disagreements about social justice could be, ifnot resolved, then at least transcended.28PART 2: THE CHOICE OF A CONSEQUENTIALIST MORALFRAMEWORKThis part gives the rationale for relativizing the concept of rational disagreement toa particular moral framework and sketches reasons for provisionally accepting aform of consequentialism.2.1 DEFINING NON-TRIVIAL LIMITS TO RATIONAL DISAGREEMENT§1 Making metaethical commitments to a particular moral theory, it mightseem, is both unnecessary in the context of a thesis on rational disagreement aboutjustice, and in any case bound to be unsuccessful. Why not define “rationaldisagreement’in a framework-invariant way and then show how the problem ofrational disagreement arises within some moral theories of current interest?To introduce a controversial commitment to consequentialism is not a step Ihave taken lightly. My original intent was to follow Rawls’ example of sidesteppingintractable disputes in moral theory altogether and to focus my inquiry oncharacterizing plausible candidates for the status of rational disagreements as theyarise across the spectrum of contemporary theories of justice. The problem turnedout to be twofold. First, I accumulated a large and chaotic array of different points ofdisagreement which, from some moral perspective or other, could plausibly becharacterized as ‘rational”. And second, the complete absence of any suggestion ofprincipled limits to rational disagreement made my project seem to collapse into adefense of skepticism about justice.There can hardly be interesting limits to rational disagreement if any disputeabout justice becomes immediately a matter of rational disagreement, provided onlythat a moral theory held by some people leaves this dispute irresolvable or providedonly that any of these theories conflict with any of the others about how to resolveit. If a theist appeals to the dictates of God, a relativist to beliefs cherished within her29particular reference group, a subjectivist to the strength of her personal convictions,a moral realist to his intuitions about some Platonic realm of moral entities orproperties, and a skeptic to the utter absurdity of moral claims, it is not difficult toimagine that they will possess little by way of a shared rational framework to settlespecific disputes about justice. If we could narrow the range of prima fade acceptablemoral theories to those fashionable among competent analytic philosophers at thepresent time - or even if we narrowed the range further by shortlisting onlydisagreements within and between various forms of consequentialism,contractarianism, or rights theories as potential candidates for the status of rationaldisagreements- several vexing problems would still remain. First, if all moraltheories held by minimally competent analytic philosophers were accorded equalstatus, the scope for rational disagreement would still come perilously close toskepticism. For example, a morally clear-cut issue for a consequentialist may not beat all so clear-cut for certain types of contractarians: I am thinking here of Gauthier’smoral views about the mass extermination of cultural groups whose vastly inferiortechnology renders the expected benefits of cooperation less attractive than theexpected benefits of outright extermination (1986: 231-2). Second, the task ofdeciding, with any precision, who to exclude from the ranks of minimallycompetent analytic philosophers would be fraught with difficulty, and there is inany case something embarassingly arbitrary about using, as one’s ultimate criterionfor a moral theory’s rational status, simply current fashionableness among one’sown fraternity.§2 But why not simply pick some moral theory or other and examine sourcesof rational disagreement within it, without making any claims about its rationalsuperiority compared to other moral theories? Would clarifying such conceptualinterrelationships not count as doing perfectly respectable philosophy? Why notabandon the ambitious idea that we might be able to say, about a given issue, that it30involves matters of rational disagreement - and instead contend ourselves withsaying that the issue involves matters of rational disagreement from the point ofview of such-and-such a moral theory, but of course, if you happen to subscribe tosome other equally acceptable moral theory, then of course it may not?Saying that will indeed be my failback option. But I don’t want to admit thedefeat of my original hopes for this project so easily. I would seem to face threeoptions for interpreting the concept of “rational disagreement”: (1) give up talkabout rational disagreement simpliciter and admit that whether an issue is a matterfor rational disagreement may well depend on which of several equally acceptablemoral theories one adopts - and that in the end there are perhaps no interestinglimits to rational disagreement about justice at all; (2) find plausible candidates forthe status of rational disagreements which arise whichever moral theory oneadopts; or (3) take the plunge into metaethics after all and defend my stand.Weighing the pros and cons of each option, I found myself attracted to the third onthe following grounds:(1) Talk of rational disagreement relative to a particular metaethicalframework is at best uninteresting and at worst misleading if it is admitted thatthere are other, equally acceptable frameworks relative to which the same issue maynot involve matters of rational disagreement. Such an admission would turn theconcept of “rational disagreement” into something of an oxymoron - unless , ofcourse, everything becomes a matter of rational disagreement simply because it isrationally irresolvable within at least one moral theory or it is resolved by differentequally acceptable moral theories in conflicting ways. But by relaxing the conditionsfor something to qualify as a rational disagreement in these ways, we have failed todistinguish rational disagreement from skepticism in any interesting sense.(2) Finding plausible sources of rational disagreement which arise relative toevery moral theory professed in good faith by sane people - even if there were suchdisagreements - seems quite obviously uninteresting. If many different moral31theories - for example, the range of metaethical beliefs defended by analyticphilosophers today (some of whom are moral skeptics) - are taken as equallyacceptable from a rational point of view, then there exist no principled, non-triviallimits to rational disagreement about social justice. The fact that some identifiablyanalogous types of disagreement (e.g. the description-relativity problem) occuracross different frameworks is simply an inconsequential curiosity.(3) Under the pressure of these considerations I wondered if I could perhapsprovisionally endorse some moral theory after all. But which theory? Obviously,the theory which would emerge as the theory least ravaged by devastating objectionsfrom the widest possible reflective equilibrium of available moral theories.§3 1 find myself forced, by the need to set non-trivial limits to rationaldisagreement, to relativize the concept of rational disagreement to a particularmetaethical framework. And, in an attempt to avoid turning this conceptimmediately into an oxymoron, I find myself forced to embrace the chosenframework for constraining rational disagreements not simply as one attractiveoption among many equally reasonable competitors, but as the framework which, Iclaim, our widest possible reflective equilibrium is likely to identify as ultimatelythe most reasonable way to think about the basic nature of morality.But my attempt to claim, however tentatively, some degree of rationalsuperiority on behalf of a certain form of consequentialism will immediatelyencounter a seemingly paralyzing objection:You have taken the fact of persistent disagreement about justice amonginformed people of good will as evidence for the possibility of rationaldisagreement. But just as informed people of good will have not convergedon one ‘correct’ conception of justice, so they have not converged on one‘correct’ metaethical theory. Debates about most or all of the moral theoriesyou dismiss as rationally unacceptable are still very much alive inphilosophy, and in fact many contemporary disputes about justice arebetween defenders of fundamentally different moral theories. The tensionbetween your insistence on rational disagreement about justice and your32dogmatic denial of equally rational disagreement about moral theory seemsirreconcilable.In philosophy we are quick to reject an enterprise when we detect an irreconcilabletension at its core. But while I admittedly face a dilemma, I think there may yet be away to salvage my project.What exactly is my dilemma? If I don’t commit myself to any moral theory, Icannot draw interesting limits to rational disagreement about justice, and myproject will be trivial. If I do commit myself to a moral theory, my defense of it willnecessarily contain important gaps, and my claims about the limits of rationaldisagreement about justice will seem weak.Is there any way out? Rawis’ initially appealing suggestion that we couldsidestep moral theory altogether in justifying a conception of justice turns out, inthe end, to commit him to a deeply problematic pragmatic-communitarianmetaethic (see below). Therefore, to set non-trivial limits to rational disagreementabout justice, I see no way around taking sides in metaethical disputes. And I believethere are four lines of defense against the objection of being dogmatically blind tothe possibility of rational disagreement about metaethics, four ways of weakeningthe force of this objection sufficiently to save the credibility of my project ofidentifying plausible candidates for rational disagreement about justice: (1) byoutlining key elements of my reflective equilibrium-based rationale for choosing aconsequentialist framework; (2) by conceding the tentativeness of my claims aboutthe rational superiority of consequentialism and allowing a framework-relativeinterpretation of my findings as a failback option; (3) by stressing that plausiblecandidates for rational disagreement which I identify within a consequentialistframework will probably emerge, in analogous form, in at least some prominentrival frameworks as well; and (4) by making a case that the consequentialistframework I defend offers greater promise for identifying “principles ofaccommodation between moralities” than most other moral theories, including33Rawis’ own attempt to identify principles of justice by avowedly sidesteppingmetaethics altogether (Rawls: 1975, 540; cf. 1985). Let me say a few words about eachof these lines of defense,(1) While I cannot fill in the details of a wide reflective equilibrium, I can atleast make clear what job would have to be done to make a rational choice amongalternative moral frameworks and to establish some presumption that it can indeedbe done.(2) My inquiry is concerned with the scope for rational disagreement withinconstructivist, non-foundational, distribution-sensitive consequentialism.’ I mayadmittedly be too optimistic about the prospect of an eventual convergence of thewide reflective equilibria of informed people of good will on the question of themost rational moral theory. Or perhaps some unexpected elaborations of thetheories I reject are rationally superior to the consequentialism I adopt, from theperspective of a wide enough reflective equilibrium. Moreover, the fact that knownmeta-ethical options are flawed cannot rule out the possibility that additionaloptions will be discovered. On the basis of the arguments I can offer I am certaintlynot entitled to claim, nor do I claim, that those who reject consequentialism mustinhabit a realm of superstition which places them beyond the reach of meaningfuldebate about justice. Those who endorse some alternative theory may interpret anyclaim p which I make about the sources and limits of rational disagreement asincluding an implicit proviso: “p if q”, where “if q” stands for “if CNDconsequentialism or something relevantly similar is indeed the most rational wayto think about morality.” The sort of consequentialism I sketch is indisputably amajor contender among moral theories.2 Therefore, even though this provisoAbbreviated CND consequentialism. This form of consequentialism will be clarified below (Sec 2.3and 2.4).2 A work that argues against contractarian and natural rights theories and clears the ground forconsequentialist constructivism is L.W. Sumner’s The Moral Foundation of Rights (1987). A work thatattempts part of the actual constructivist normative task is D.W. Haslett’s Equal Consideration: ATheory of Moral Justification (1987).34weakens the interest of my conclusions, it is still quite worthwhile to ask: “At whatpoints can informed people of good will legitimately disagree about social justice,assuming they share a broadly consequentialist framework?”(3) The limits to rational disagreement will strongly depend on which moralframework one adopts, but particular plausible candidates for the status of rationaldisagreements will not. For example, reconciling the tension between aggregativeand distributive values (which I label the aggregation-distribution problem) is a richsource of rational disagreement within many theories; and so are the criteriadefinition and inclusion problems, the priority problem and several others. Anumber of candidates for rational disagreement which I identify withinconsequentialism will also emerge, mu tatis mutandis, within rival moralframeworks. If so, my efforts at clarifying their nature will not have been in vain,even if some moral theory turns out to be rationally superior to the formconsequentialism I adopt here.(4) In assessing the charge that adopting a particular moral framework isdogmatic for purposes of identifying sources and limits of rational disagreementabout social justice, one must appreciate a special pragmatic advantage ofconsequentialism: this framework has a great deal of context-dependent flexibilitybuilt in. It leaves a great many matters up to contextual judging which othertheories would rule out at the level of fundamental theory. It can, therefore,accommodate politically (although of course not metaethically) a wide variety ofrival moralities in a way in which such rival moralities could not accommodateeach other. In complex, pluralist societies it can, I think, provide the “principles ofaccomodation between moralities” which Rawis speaks of by justifying a liberal typeof social order (see Ch 3.4).352.2 REFLECTIVE EQUILIBRIUM AS A METHOD OF MORAL JUSTIFICATION§1 Before sketching my reflective equilibrium-based rationale for CNDconsequentialism, I must address this question: What exactly is the method ofreflective equilibrium? Understanding this method is crucial for appreciating thelimits to the certainty and precision we should expect in moral matters. We mustsee, according to Rawis,...if the principles which would be chosen [in the original position] match ourconsidered convictions of justice But presumably there will bediscrepancies. In this case we have a choice. We can either modify the accountof the initial situation or we can revise our existing judgments, for even thejudgments we take provisionally as fixed points are liable to revision. Bygoing back and forth, sometimes altering the conditions of the contractualcircumstances, at others withdrawing our judgments and conforming themto principle, I assume that eventually we shall find a description of the initialsituation that both expresses reasonable conditions and yields principleswhich match our considered judgments duly pruned and adjusted. This stateof affairs I refer to as reflective equilibrium. It is an equilibrium because at lastour principles and judgments coincide; and it is reflective since we know towhat principles our judgments conform and the premises of their derivation.(19-20)On a superficial reading, Rawis’ idea of establishing a reflective equilibrium seemsvery simple - closely related, in fact, to the dialectic method of testing principlesagainst intuitively plausible counterexamples which moral theorists have employedall along. But Rawis is not arguing that simple coherence between principles andconsidered judgments is sufficient to justify acceptance of such principles. Themethod of reflective equilibrium differs from the traditional dialectic method in theway Rawls seeks to establish simultaneous coherence between our consideredjudgments about justice and a set of principles and between relevant moral as wellas non-moral background theories and a description of a fair contractarian choicesituation in which more or less the same principles of justice could plausibly beaccepted. Relevant background theories sketched by Rawls include alternative moraltheories, a theory of moral personhood, of procedural fairness, of the role of36morality in society as well as relevant findings of the social sciences, especiallymoral psychology. It might be thought that Rawis is trying to do even more thanestablish simultaneous coherence, namely simultaneous implication or derivability.He has sometimes been interpreted as arguing that his principles of justice arederivable from the contractarian choice situation independently of someone’sconsidered moral judgments about specific cases.3 But this interpretation of Rawis’argument, along with many criticisms based on it, are deeply misguided. AlthoughRawls misdescribes his theory repeatedly as “the contract doctrine”(e.g. 329), he doesnot, and could not, seriously claim that his original position has by itself the powerto yield very determinate principles. And it would add a further layer ofmisunderstanding to construe Rawis’ original position as a fundamentallycontractarian attempt to get moral conclusions from morally neutral premises,although here again Rawis originally misdescribes his theory as part of the theory ofrational choice (cf. 16, 583). But he also clearly admits, “We want to define theoriginal position so that we get the desired solution”(141).4Such statements have often earned Rawls the charge of circular reasoning (cf.e.g. Hare, 1973). But all Rawis claims, and can plausibly claim, is that our consideredjudgments about what constitutes a fair contractarian choice situation can be basedon various background theories and kept independent, to some extent at least, ofour considered judgments about what is just or unjust in particular cases; and that,therefore, trying to find a match or overlap between principles suggested by eitherset of considered judgements is not simply narrowly circular reasoning butestablishes a degree of coherence among our moral beliefs which can be construed as3 Ackerman, for example, seems unclear about the limited role of the original position in Rawis’argument: “When the Rawisian Zero confronts the Infinite Choice Set, it is impossible to choose anyprinciples of justice until he is endowed by his creator with some set of preferences to guide hisjudgment.”(339)As Rawls clarifies the role of the original position in his 1985 paper, “the original position is simplya device of representation: it describes the parties, each of whom are responsible for the essentialinterests of a free and equal person, as fairly situated and as reaching an agreement subject toappropriate restrictions on what are to count as good reasons.”(237)37the best sort of support which any such beliefs can ultimately have. AlthoughRawl& choice of words occasionally opens him up to misunderstandings, it is aserious mistake to criticize him as if he considered principles of justice to bederivable from the original position, independently of whatever principles areantecedently suggested to us by our considered judgements of what is just or unjustin particular cases.I think the way Rawis employs the method of reflective equilibrium must beunderstood roughly as follows. First he asks what could constitute a faircontractarian choice situation, based on our considered judgments about proceduralfairness. Second he asks what set of principles of justice would be supported by ourconsidered moral judgments about what is just or unjust in particular cases. Thirdhe asks what subset of such principles would also be supported by a faircontractarian choice situation. Fourth he gives strong arguments about why theprinciples so identified must apply, not to the question of a given individual’sdistributive share, but to the question of what overall distributive outcomes a justsociety’s basic structure must achieve. That most reflective people who followedRawl& reasoning would in fact also converge on much the same set of principleswas perhaps Rawis’ hope and, briefly, his promise. (I shall discuss the shortcomingsof Rawl& approach from the point of view of identifying sources and limits ofrational disagreement about justice in Ch 3.2.)§2 Daniels helpfully clarifies reflective equilibrium methodology as agenerally applicable method of moral justification. Following Rawis, hedistinguishes narrow reflective equilibrium (the traditional dialectic method oftesting principles against intuitively plausible judgments) from wide reflectiveequilibrium, which he characterizes as follows:The method of wide reflective equilibrium is an attempt to producecoherence in an ordered triple of sets of beliefs held by a particular person,38namely, (a) a set of considered moral judgments, (b) a set of moral principles,and (c) a set of relevant background theories... .We do not simply settle for thebest fit of principles with judgments, however, which would give us only anarrow equilibrium. Instead, we advance philosophical arguments intendedto bring out the relative strengths and weaknesses of the alternative sets ofprinciples (or competing moral conceptions). These arguments can beconstrued as inferences from some set of relevant background theories... .Wecan imagine the agent working back and forth, making adjustments to hisconsidered judgments, his moral principles, and his background theories. Inthis way he arrives at an equilibrium point that consists of the ordered triple(a),(b),(c).(258)The gist of the method of wide reflective equilibrium seems to me this: we mustestablish a first coherence between considered judgments about specific cases of acertain type and a moral principle which covers cases of this type, and a secondcoherence between these principles and relevant moral and non-moral backgroundtheories, with an important further proviso: the considered moral judgments whichconstrain our (moral and non-moral) background theories “must be to a significantextent disjoint” from the considered moral judgments that directly constrain ourprinciple (Daniels, 260). “Coherence”, however, is a vague concept that may denotelogical relationships of widely varying supportive strength, ranging from mereconsistency Of thematically unrelated beliefs, via increasing probability orderivability in some loose sense, all the way to deductive implication. Obviously,when we give coherence arguments in support of some controversial conclusion,we want in every case to make the connection of logical support between beliefs asstrong as we can possibly make it.As I understand it, applying wide reflective equilibrium methodology(WREM) to the problem of justifying principles of justice essentially meansconstructing, on the basis of one’s considered judgments about justice in specificcases, general principles which imply these judgments and which can then be usedto make new judgments, provided that (1) these principles are consistent with wellestablished background beliefs such as, for example, the scientific world view, thefindings of the various sciences, or well-founded beliefs about the hopelessness of39expecting a resolution to certain meta-ethical disputes; and provided that (2) thereasons that led one to espouse these background beliefs are, to some degree,independent of one’s considered judgments about justice in specific cases. Thismethod, as Rawls himself notes (49), is somewhat analogous to the way scientistsconstruct scientific hypotheses on the basis of empirical observations- hypotheseswhich imply these observations and which can then be used to predict newobservations (where such hypotheses are also independently constrained byrelevant background beliefs, such as the mass of already well-confirmed scientifictheories).5Just as a sufficient number of false predictions derived from somescientific hypothesis will eventually lead scientists to discard the hypothesis, so asufficient number of intuitively unacceptable judgments derived from a moralprinciple will eventually lead proponents of WREM to discard a principle. Just asscientific method aims to establish the greatest possible coherence betweenobservations, a particular hypothesis and all other relevant and already well-confirmed scientific theories, so WREM aims to establish the greatest possiblecoherence between considered moral judgments, a particular set of moral principlesand well-founded and relevant background theories. But that is as far as Rawlsallows the analogy to go, at least in the case of principles of justice (49). Whereas(most) scientists presumably regard hypotheses which are maximally consistentwith both observations and already well-confirmed scientific theories asapproximating some mind-independent physical reality, Rawls does not regardprinciples of justice which are maximally consistent with considered judgments andrelevant background theories as approximating any mind-independent moral5 Daniels rejects this analogy between “considered moral judgments and observation reports” becauseaccepting it seems to him to make reflective equilibrium methodology vulnerable to the “nocredibility” criticism (27-3). That criticism is that the causal stories we can tell about how we came byour observations gives them credibility, whereas the causal stories we can tell about how we came byour considered moral judgments does not give them similar credibility. Daniels succeeds in pointing outplausible disanalogies between observation reports and moral judgments; but I do not see how thatchanges the point of the analogy, namely to illuminate the basic functioning of the method of reflectiveequilibrium in a helpful way.40reality. He leaves no doubt about his uncompromising anti-realism about principlesof social justice. His two principles of justice, he says,are not regarded as a workable approximation to the moral facts: there are nomoral facts to which the principles adopted could approximate. (1980, 564).§3 What is the value of WREM as a method of justification in ethics, and inparticular for principles of social justice? Is it our best method, is it one plausiblemethod among several, is it completely worthless, or is it downright pernicious?There are arguments for all four views.There certainly is no guarantee, using WREM, that even thinkers who startout with roughly the same “considered moral judgments” will accept the sameprinciples in the end. As Rawls himself states (50), it is not at all a foregoneconclusion that different thinkers, seeking maximum coherence between theirconsidered judgments about justice, principles and relevant background theories,must converge on pretty much the same principles. There is an importantdifference between moral theorists working back and forth between moraljudgments, principles, and relevant background theories, on the one hand, andscientists working back and forth between observations, hypotheses and alreadywell-confirmed scientific theories. In this “working-back-and-forth” process,observations are constrained by empirical input in a way in which moral judgmentsare simply not, or at least seem not to be to many thoughtful people. For acontractarian, for example, a particular way of specifying the choice situation and theprinciples derived from it may be in reflective equilibrium with that person’sconsidered moral judgments and background beliefs. For a religious person, thestory of the divine origin of moral commandments, along with his particularreligion’s set of commandments, may be in reflective equilibrium with thatperson’s considered judgments and background beliefs. For an intuitionist realist,the belief in the mind-independent existence of moral properties and in a moral41faculty of intuition by which they can be known, along with these intuitionsthemselves, may be in reflective equilibrium with his considered judgments andbackground beliefs. At best, someone may have examined absolutely every possibleoption of thinking about morality, but the moral principles and the account of theirorigin which she eventually accepts will only constitute yet another very widereflective equilibrium with her considered judgment and background beliefs.A proposed method of justification for principles of justice would seem to beof dubious value, in fact downright pernicious, if it allows different thinkers, all ofthem using this method in good faith, to emerge with strongly conflicting ideologieswhile legitimizing the claim of each to have “justice” on his side. But there alsoseems to be a strong case for considering WREM to be trivial. It seems trivially truethat, if any method can supply consensus on principles of justice, wide reflectiveequilibrium method can, because in the final analysis there would seem to be noreasonable alternative to it. The idea of converging wide reflective equilibria issimply a description of the dialectic process at its most general, carried through to apoint of maximum coherence of all beliefs of all people about all things. If the ideaof converging wide reflective equilibria is construed in this way, as I think it mustbe, then it is of course true that it is our only hope of reaching rational consensus onmoral principles. The best we can ever do is to achieve one sort of reflectiveequilibrium or another, and the more aware a person is of all the coherentist prosand cons of the various alternative ways of thinking about the nature of morality,the more sophisticated his particular reflective equilibrium will be. In this sense,WREM subsumes all other methods of moral argument within its scope.We would seem, then, to have reached the paradoxical conclusion thatWREM is both pernicious and trivial. On the one hand, using WREM has leddifferent moral theorists to conflicting conceptions of justice. But on the other, therewould seem to be no genuine alternative to WREM as our ultimate method ofmoral justification. How can we resolve this paradox?42The lesson to draw from the recent history of justice theorizing is not, I think,that WREM should be discarded as either pernicious or trivial. Instead we mustrecognize that the intellectual demands which this method makes on a moraltheorist are so enormous and perhaps overwhelming that we can only have limitedconfidence in the claim of any one thinker to have established a conclusive widereflective equilibrium for a particular set of principles. A plausible explanation forwhy using WREM to produce convergence in philosopher’s beliefs about justice hasbeen strikingly less successful than using scientific method in producingconvergence in scientists’ beliefs about the structure of the natural world wouldseem to me to be the following. Justice theorizing must draw on backgroundtheories in meta-ethics, metaphysics, epistemology, history and the social sciences,all of them fields of inquiry which involve such unsettled and elusive conceptual,normative and empirical matters that formulating readily testable hypotheses israrely possible, and even informed thinkers acting in good faith are likely to emergewith substantially different conclusions. The unsettled state of knowledge in thesefields does not, of course, imply that progress is impossible and could not lead to aprogressive convergence of the reflective equilibria of different justice theorists. Butjustice theorists must, I believe, cultivate a fuller appreciation of the manifoldpossibilities for rational disagreement regarding the conclusions they have reachedemploying the method of wide reflective equilibrium.2.3 CHOOSING CONSEQUENTIALISM IN REFLECTIVE EQUILIBRIUM§1 I now want to sketch my understanding of how the method of widereflective equilibrium applies to the problem of choosing a theory about the basicnature of morality. Justifying a metaethical theory by using WREM requires us,ideally, to examine all options and choosing that option which coheres best withboth our considered moral judgments and all relevant background theories which43we have found reason to accept. I shall mention briefly the eight metaethicaloptions that struck me as prima facie plausible at some point, only to put aside thefirst five without further argument.The first five options, which I shall group together as tlnon_constructivistfltheories, include theistic moral theories, social relativism, subjectivism, all forms ofmoral realism, and radical skepticism. The next three options, which I shall callconstructivist’ theories, employ rights-based, contractarian, or consequentialistmethodologies, even though they all arise from the same basic understanding of themain job which morality as a social institution is to accomplish. The difference Iwant to mark by using the distinction between tTconstructivist” and ttnonconstructivist” is suggested by Rawis when he explains his own Kantianconstructivism:...the parties in the original position do not agree on what the moral facts are,as if there already were such facts. It is not that, being situated impartially,they have a clear and undistorted view of a prior and independent moralorder. Rather (for constructivism), there is no such order, and therefore nosuch facts apart from the procedure of construction as a whole; the facts areidentified by the principles that result. (1980: 568)On my understanding (which seems compatible with Rawlst), what makes a moraltheory‘tconstructivist” is that it regards morality fundamentally as a peculiarproduct of human reason, a sui generis type of rational social construct whosecentral function is to promote the satisfaction of human interests by suggestingsuitable ways of restraining and coordinating people?s behavior. Part of the idea ofmoral constructivism is the possibility that a variety of strongly conflicting systemsof constraints on peopl&s behavior will be equally serviceable for doing morality’sjob, although there will generally be a strong presumption against upsetting toomany of people’s ingrained customs and habits all at once. By contrast, what makesa moral theory “non-constructivist” is that this theory does not accord reason afundamentally creative role in constructing morality so as to promote mutual trust44and reconcile human interests, but (at most) a purely epistemic role in discoveringmorality’s pre-existing content.6An overview of five non-constructivist options in their simplest terms willbe helpful at this point to establish our bearings. Theists generally think that adivine being’s commands are either the direct source of what is morally right andgood, or our best epistemic criterion. Social relativists identify the moral right andthe good with the laws or customs of some preferred social reference group.Subjectivists, the limiting case of relativists, consider the moral right and good to bematters of personal opinion. Moral realists hold that the right and the good involvenatural or non-natural moral properties or entities existing independently of whatanyone thinks about them. It is common, although certainly not necessary, forrealists to be intuitionists, and to combine the idea that the right and the good aremind-independent realities with a distinctive doctrine about how we come to knowthis reality, called “intuitionism”. Radical skeptics, on the other hand, think thatmoral talk captures neither a mind-independent nor a mind-constructed rationalorder, but is altogether irrational in every sense of that word.What makes the above five theories non-constructivist? Theistic moralitiesare non-constructivist because they regard the content of morality as eithermetaphysically or epistemically given by God’s commands, rather than constructedby human reason in the light of an at least partial convergence of our evolvingunderstandings of what sorts of behaviors will best promote collective andindividual well-being. Social relativism (as I understand this term here) is nonconstructivist because it regards the content of morality as given by either the lawsor customs of a social reference group, no matter whether these laws or customspass reason’s test of being plausible means to well-being. Subjectivism (as I defined6 Note that there is no suggestion in my definition of a constructivist moral theory that it necessarilyinvolves the use of game theory or decision theory to derive moral principles from a contractarianchoice situation. This approach, although sometimes regarded as the paradigm of moral constructivism(cf. Gibbard, 265), is simply one of several prima facie plausible constructivist approaches to morality.45this term) is non-cons tructivist, because it simply regards the content of morality asdetermined by whatever happens to be the content of a person’s opinion aboutmorality. Moral realism is non-constructivist insofar as it regards morality assomehow part of the basic ontological fabric of the universe. And radical skepticismis non-constructivist in the (utterly trivial) sense that it denies that morality, even ifunderstood as an artificial construct of reason for the sake of promoting humangoals, has rational status. For reasons of space I shall reject all these theories aboutthe ultimate nature of morality out of hand here.But two questions arise: Could there be constructivist versions of the abovetheories? And do I mean to reject them simply because they are non-constructivist?The answer to the first question is no. There could, of course, be constructivistmoralities whose normative content is identical with the normative content of oneor the other of these theories, but the meta-ethical interpretation of this contentwould have shifted in such fundamental ways that it would only confuse matters toregard them as the same theories or to use the same labels for them.The answer to the second question is also no. I mean to reject the abovetheories, not because I claim to have an a priori argument to show that all nonconstructivist moral theories must be false, but because I think that each of thesetheories has its own characteristic shortcomings.What all non-constructivist options (aside from skepticism) have in commonis that they accord human reason at best an epistemic role in discovering morality’scontent, rather than a fundamentally creative role in constructing and improvingmorality to serve human purposes and interests. Before discussing prima facieplausible constructivist methodologies, I want to ask if I have perhaps left outimportant non-constructivist options. I might seem to have ignored Kantian ethicsor non-constructivist forms of consequentialism, such as Aristotelian or neoAristotelian teleology. But I think Kant’s moral principles - the idea of moralconsistency and the idea of respect for persons - are best given a purely constructivist46interpretation. Non-constructivist forms of consequentialism (let me call them“teleologies”) face the question of how we could ultimately determine the good,apart from reflecting on people’s diverse interests, values and purposes and thenusing WREM to set certain broad limits to the range of minimally plausible forms ofthe good life. But if a teleological theory is highly pluralistic in the goods it wants topromote, regards morality as a mind-constructed device for overcoming certainobstacles to social cooperation, and accepts WREM as the method for constructinglimits to what can plausibly be considered to constitute a good human life, then itdiffers at most in degree from the understanding of morality I am about to develop.7§2 Having set aside forms of non-constructivism, I will briefly consider threeconstructivist methologies which I shall call “rights-based”, “contractarian” and“consequentialist”. The methodology I want to adopt, in the end, is meant to beselected on the basis of a judgment, reached in the widest possible reflectiveequilibrium, about which of all prima facie plausible constructivist options is leastafflicted with serious shortcomings. My efforts to sketch such a reflectiveequilibrium here will, however, necessarily be very truncated.First, let me clarify further what I take all constructivist options to have incommon. For a moral constructivist, morality is a system of behavioral rules andattitudes, and social pressures to uphold them, which evolved in human history ina variety of ways and due to a variety of stimuli, but whose continued existence andnormative force and content can be rationally justified - and justified purely interms of its necessary function in the promotion of people’s interest in well-being.But why is morality useful or even necessary for the satisfaction of people’sinterests? Haslett summarizes lucidly what we ultimately need a morality for:It is important to be clear about the distinction between WREM and constructivism. Constructivism isa basic conception of the nature of morality. WREM is the ultimate method for deciding whether aconception of morality, or any particular moral belief, meshes sufficiently well with the rest of one’stotal belief system to be rationally acceptable.47Why should a society have a code of morality anyway? At least part of theanswer is that a decent code of morality enables there to be the degree ofsecurity, and of trust in others, that is necessary in order for there to bemutually beneficial interaction among people. It makes this security and trustpossible, in large part, by making possible a much greater degree ofuniformity, and thus predictability, in people’s behavior than there wouldotherwise be. In a society with a decent code of morality people will, becauseof this code, pay their debts, keep their promises, refrain from violence, tellthe truth, and so on. Not all of the time, but most of the time. Therefore, tothis extent, there will be a certain uniformity, and thus predictability, inpeople’s behavior, the kind of uniformity and predictability upon whichsecurity and trust depend. Making this uniformity, and thus predictability,possible is the job, or one of the main jobs, of a code of morality. (77)There can, of course, be rational disagreement about how sophisticated or well-adapted or farsighted any particular moral code is in doing its main job, but it seemsclear what this job is. It is to ensure enough uniformity and thus predictability inpeople’s behavior so as to create the necessary sense of security and trust in order topromote almost universally beneficial interaction.Another clear formulation that encapsulates the rational basis for morality isthe following by Peter Railton. Think of morality, he says,as a set of behavioral restraints and recognized permissions and obligationsthat function within a group to promote co-ordination in those circumstancesin which co-ordination would be mutually beneficial. Nothing here aboutrepresenting an independently-ordered reality. Still, we can see how the needto discourage special pleading and free riders, to promote consistentexpectations and co-ordination in the projection of existing norms into newcases, to have teachable principles and associated habits of conduct, to havepublicly-ascertainable procedures for application, and so on, would createquite practical pressure on behalf of consistency, simplicity, generality, theavoidance of singularities, and the appeal to abstraction to mitigate conflictand to accommodate approximation.. ..A practice of discouraging singularities[special pleading by individuals] within the group becomes a closerapproximation to a demand of universalizability as the group itself, or thereference group with respect to which it seeks to defend itself, becomes acloser approximation of all humankind. (187-8. Emphasis added.)48This description of morality’s function seems to me to provide the essentials forexplaining positive morality as well as for justifying changes to it.8 It suggests howmoralities have evolved under practical pressure as norms of co-ordination. And tothe extent that rational foresight can help us figure out which logically possiblenorms of co-ordination are practically superior to others in terms of what is to our“mutual benefit’, to that extent we will also be justified in choosing one particularcontent for these norms of co-ordination over others. It is true that Railtonconsiders this account of morality “rather starkly practical - and perhaps thereforeincomplete”(187); but it is not clear what more we are justified in believing there tobe to morality.§3 That rational morality does not describe a given order independent ofwhat we conceive to be to our benefit, but can only be constructed by fallible humanbeings out of their shifting understanding of what sorts of behavioral restraints willeffectively serve their mutual interests, is one of the fixed points of my own moralunderstanding, supported by my widest and most stable reflective equilibrium.How, then, might we go about restructuring positive morality in a more rationalway?1. The rights-based approach: Although rights-talk is notoriously associatedwith theistic or realist moral theories, there seems to be initially no reason whyrights could not, in principle, become fundamental building blocks of aconstructivist moral vocabulary. In fact, the core idea of this approach seemstemptingly simple. The core idea of a rights-based approach is to look to somesalient human capacities, vital needs, widespread interests or other commonIt is obviously important not to confuse the moral beliefs people actually have (the positivemorality) with the moral beliefs that would be rational for them to have (the rational morality).Sometimes the term “ideal morality” is used for what I call “rational morality”. This term, however,gives rise to confusion because any rational morality will have to be further subdivided into the moralbeliefs which we should rationally follow under ideal circumstances and the moral beliefs which weshould rationally follow under real-life circumstances.49features of human nature and simply to postulate rights of a certain content, scopeand strength for possessors of such features. Constructivist rights theorists (e.g.Mackie; the early Nozick) may claim, for example, that the fact that individuals arecapable of having their own ideas of the good should be construed as sufficient togive them a right of a certain strength and scope to seek their own good in their ownway.One problem with taking rights as morally basic by linking them up withhuman interests, needs or capacities is that there seems to be no clear rationale fordeciding which of potentially infinitely many interests, needs or capacities should beregarded as giving rise to rights of what sort of content, scope and strength. AsSumner asks, after reciting a familiar litany of currently popular, but stronglyconflicting rights-claims:• .which aspects of our nature are the relevant ones? We are beings capable ofchoice - do we therefore have liberty-rights? We are also beings capable ofbeing injured by others- do we therefore have claim-rights not to be harmed?We are also beings who need the support and assistance of others - do wetherefore have claim-rights to be given assistance? If we lack any of theserights, why do we lack them? If we have them all then how can our naturedetermine which is to take precedence when they conflict? How, in general,can we distinguish between the relevant and the irrelevant aspects of ournature without presupposing a particular outcome for theargument?... [N]ature, even our nature, underdetermines selection of a set ofbasic rights and thus provides no effective control over the proliferation ofbasic rights principles. (125-6).Linking features of human nature with rights-claims without any additional checks,it turns out, is far too simplistic a way to go about constructing a viable morality.Such a rights-postulating process might work in limited contexts if employed withrestraint, but it couldn?t accomplish morality’s main job in a large-scale pluralisticsociety, because it will immediately lead to an unprincipled proliferation ofconflicting rights-claims. As Sumner shows, there seem to be insuperable obstaclesto building ways of avoiding such conflicts into the strength and scope of each rightitself (1987; 92-126).50In addition to pure rights-based theories in which only rights are basic, therewould seem to be hybrid theories for which both rights and goals are basic (e.g.Nozick, Dworkin). In these approaches, rights place absolute or at least very strongside constraints on the pursuit of goals, or goals are invoked to settle conflictsamong rights. But hybrid theories face a dilemma. Either the rights postulated bysuch theories constrain the pursuit of consequentialist goals only in ways whichseem ultimately more effective in promoting consequentialist goals - in which casethese rights are justifiable in terms of goals and therefore non-basic after all; orrights are basic and therefore not justifiable in terms of goals - in which case theymust postulate rights of a kind which will effectively hamper the pursuit ofconsequentialist goals without any compensating beneficial consequences. The firsttype of hybrid theory obviously collapses into indirect consequentialism, while thesecond needs a very strong rationale for recognizing goals while yet postulating anticonsequentialist rights (and this rationale would then make rights non-basic afterall). And for a moral constructivist, there is only one prima facie promising placewhere a rationale for anti-consequentialist rights could come from: contractarianarguments.Since the practice of generating substantive rights directly by an appeal tofeatures of human nature does not seem to admit of any principled limit, sinceconflicts between such rights can only be sorted out by other considerations, sincehybrid theories would inevitably seem to collapse into other approaches and sinceconsequentialism can itself provide a clear rationale for rights-talk (see Ch 3.1), I cansee no good reason why moral constructivists would want to admit rights as basic.But doesn’t this rejection of rights as basic building blocks of a constructivistmorality overlook the way in which WREM is ultimately also rights-based— insofaras WREM seems to presuppose everyone’s right to be convinced by reason of moralprinciples? (Peter Danielson, pers. comm.)51In reply I want to deny that WREM presupposes a wright” to be convinced byreason, in the sense in which “rights” are at issue here: as morally basic constructswhich allow individuals or groups to make claims to certain means tomeet/satisfy/develop certain human human needs/interests/capacities, where themere existence of some need/interest/capacity is regarded as sufficient justificationfor the corresponding rights-claim. It is true, the philosophical demand forjustification of moral claims can itself be grammatically reformulated as a rights-claim, but this reformulation would, I believe, be misleading. The demand forjustification is so fundamentally tied up with the very enterprise of making sense ofthe social phenomenon called morality that I would regard “the right to beconvinced by reason” as of a clearly different order than all other rights. My claimabout rights is simply that the prospects seem poor that constructing a moralitypurely by spelling out a list of basic rights will be able to do the job for which weneed a morality in the first place. In fact, I would claim that no practical senseattaches to the idea of a “right to be convinced by reason”, because requiring a fullunderstanding of the complex WREM-based rationale behind many moralprinciples and judgments would surely overtax most people’s rational capacities. Ido not take the employment of WREM by moral philosophers to entail such arequirement.2. The contractarian approach: The core idea of the contractarian approach isto conceive of constructing the rules of morality as analogous to negotiating theclauses of a multi-lateral contract. This core idea can be elaborated in various ways(e.g. Hobbes, Locke, Rousseau, Rawls, Scanlon, Gauthier). I find it useful todistinguish simple contractarianism, contractualism and mutual-benefitindividualism.(1) Simple contractarianism wants to derive moral constraints on individualbehavior or social policy simply from the idea that these constraints must be such52that individuals could voluntarily agree to them for their mutual benefit. But canthis idea alone generate a morality? What sorts of principles would turn out to beaccepted by contractarian bargainers is highly sensitive to how the conditions arespecified under which the agreement is imagined to have been reached. That poses adilemma for the simple contractarian. Either the conditions under which agreementon principles is reached are specified to include moral constraints or they are not. Ifthe conditions under which principles are chosen are specified to include moralconstraints (as in the case of Rawls’ “original position’), then obviously thoseconstraints are not themselves based on contractarian reasoning and presupposesome other basis. If just any arbitrary specification of the contractarian choicesituation is picked, then the particular principles chosen under those conditions willhave no moral force, since different principles would have been chosen given adifferent arbitrary specification of the choice situation. Contractarian constructivistswho choose the original position which will give them their own favoriteprinciples obviously beg the question (unless they give consequentialistjustifications for their choices - but then they are basically consequentialists, notcontractarians). Therefore, it would seem, simple contractarian approaches toconstructing morality will fail (cf. Sumner, 1987a: 159-60; Ripstein, 115-137).Is this criticism, pressed strongly by Sumner, entirely fair? Is he not applying adouble standard - rejecting the contractarian approach because of theunderdetermination of the choice situation, while construing the majorindeterminacies of consequentialism benignly as leaving room for rationaldisagreement? It would seem, however, that the simple contractarian’s dilemmamust be understood, not as a problem of underdetermination, but rather as a blatantcase of question-begging. Contractarian bargainers cannot even begin to bring thelogic of bargaining to bear on the problem of getting morality off the ground withoutfirst making morally loaded assumptions about features of the choice situation.Lomasky makes this point eloquently when he says,53Contract may be a useful heuristic for fixing attention on the agent-relativityof value to which political prescriptions must conform themselves, butneither inside nor outside of an original position is it able in a non-question-begging way to establish generally binding principles of justice. (Th3-4)This is not to deny that the simple contractarian idea is methodologically useful, butit is best understood, in the manner of Rawls, as an expository device withinreflective-equilibrium-based consequentialism, after various plausible moralassumptions have been made about the fairness of the choice situation.(2) Con tractualism conceives of morality as those social arrangements whichcan be publicly defended as reasonable, in the sense that persons who are motivatedto seek informed, unforced general agreement on the terms of social cooperationwill eventually converge on these arrangements as creating the greatest possiblesocial space for their more or less widely differing ideas of the good. Contractualists(e.g. Scanlon, Larmore, Stout) use the idea of the social contract simply as a way tomake vivid the point that morality is socially constructed out of both aggregativeand distributive concerns of individuals. The idea of contractarian bargaining canmodel this construction in wide or narrow contexts in which the need for moralprinciples arises and, given certain moral presuppositions, can serve as a usefulcriterion for testing existing moral beliefs. These presuppositions - that theagreement must be informed, unforced, general, and essentially liberal - can beplausibly derived from consequentialist assumptions. In fact, the normativeimplications of contractualism and WREM-based consequentialism seem to meessentially indistinguishable.But if so, which is morally basic: consequentialism or contractualism? What,ultimately, matters morally? Is it the promotion of some interpersonal goal whichcan plausibly be regarded as maximizing well-being, or the achievement of unforcedgeneral agreement among informed, autonomous persons?I see no need for moral constructivists to be reductionist here. It would seemplausible that, in most contexts, the contractualist criterion can be given a54consequentialist rationale, and the consequentialist criterion a contractualistrationale. We could, therefore, plausibly aim to satisfy the consequentialist andcontractualist criterion simultaneously and construct interpersonal goals whichinformed persons who are motivated to seek unforced general agreement on thebasic terms of social cooperation would choose as affording the greatest possiblesocial space for realizing different but equally acceptable conceptions of personalwell-being.(3) Mutual-benefit individualism, however, is a different story. As Iunderstand it, the core idea of this form of contractarianism - associated mostprominently with David Gauthier- is to eschew any attempt at aggregating differentindividuals’ interests in order to define interpersonal goals (of either a global orcontextual nature), on the grounds that a rational person has absolutely no reason topursue interpersonal or collective goals, except insofar as they happen to coincidewith her personal goals. Rational persons are those who care only about maximalfulfillment of their own (“considered”) preferences, whatever these may happen tobe. The project of mutual-benefit individualists is to reconstruct the social order insuch a way that it embodies only those moral constraints which every person willalmost always find it rational (in his actual self-interest) to comply with. It is easyenough, of course, to lower one’s moral standards to the point where they coincidewith people’s perceived self-interest; but what is interesting about Gauthier’s theoryis the claim that redesigning the social order along libertarian lines would lead tomaximum informed preference satisfaction of all individuals (without, presumably,stunting their preferences in any objectionable way). Obviously, if mutual-benefitindividualists have a blueprint for social reform so as to ensure maximal informedpreference fulfillment - while avoiding any messy aggregation of the preferences ofdifferent individuals in the manner of standard cost-benefit analysis - then it wouldseem that Gauthier’s mutual-benefit individualism ought to be happily embraced bya consequentialist constructivist as well.55My main project in this inquiry - identifying sources and limits of rationaldisagreement about social justice - will obviously take a very different shape,depending on the success or failure of Gauthiefs contractarian justification oflibertarianism. I face a difficult decision. I clearly cannot proceed in my main projectwithout judging the merits of Gauthier’s theory. To discuss the merits of Gauthier’swork in a few paragraphs would simply seem to be attacking a strawman. But tomake space here for the extended discussion it deserves would threaten tooverwhelm other stage-setting concerns for my main project.It seems to me best, therefore, to anticipate my conclusion but postpone mydiscussion of Gauthier?s theory to a later chapter where I assess the possibility forrational disagreement between egalitarian and libertarian liberalism (Ch 3.5). Myconclusion will be that mutual-benefit individualism ultimately fails to justifylibertarianism and offers only a very partial substitute for moral decision-making interms of promoting interpersonal goals.3. The consequentialist approach: The core ideas of a consequentialistapproach, as I understand this term here, are concisely expressed by Sumner:I begin with some assumptions about the likely shape of a consequentialistgoal. The raw materials for that goal will be an inventory of ultimate goods.Whatever other goods this inventory might include, it seems reasonable tosuppose that in one way or another it will acknowledge the value of thosestates which are the standard sources or components of individual well-being:life, health, liberty, autonomy, sociality, the development and exercise ofpowers and abilities, and so on. These goods must then be collated into someglobal value. Whatever other considerations this global value might include,it seems reasonable to suppose that it will be aggregation-sensitive, thus thatit will acknowledge the force of increasing the overall extent to whichindividuals enjoy these central ingredients of their well-being... .It might wellconsist of more than this [maximizing the sum of individual welfare], if itadmits other particular goods as well or adopts a combinatory rule which isalso sensitive to factors other than their aggregation. (201-2)56Sumner’s account exemplifies the undogmatic, non-foundationalist way in whichconsequentialist constructivists must approach their task.11 Consequentialism, as Iunderstand it, is that form of moral constructivism which says that we should giveour morality whatever features are likely to lead to the most desirable combinationof satisfied interests of individuals in their well-being. I shall call this goal the “idealconsequentialist goal” and clarify below what “desirable” here can plausibly mean. Ishall also use the phrase “X promotes well-being” for the more cumbersome phrase“X promotes the ideal consequentialist goal’ (where X can be anything whatever-generally a principle, a right, an act, a judgment, a criterion, a decision, a policy, astate of affairs, a social institution, an attitude or something of that sort). Theconsequentialism I want to defend is both constructivist and non-foundational;constructivist, because it denies that moral properties would exist even if humanbeings had not constructed them to suit their purposes; and non-foundational,because it makes our understanding of what counts as “promoting well-being”dependent on a coherentist process of seeking a wide reflective equilibrium ofrelevant considerations (see below). Moreover, I want to defend a consequentialismwhich, in Sumner’s words, “adopts a combinatory rule which is also sensitive tofactors other than...aggregation”(202).Three major questions arise: Why, of all imaginable consequentialisms,should we choose this particular form? Is its goal ultimately coherent? And how canwe decide, in practice, what would count as having achieved or approximated thisgoal?11“Consequentialism” has, of course, been defined in many different ways. Traditional textbooks oftendefine it, unhelpfully, as the theory which evaluates actions purely by their consequences rather thantheir intrinsic nature. Others define it as doing whatever will produce the most good or the best state ofaffairs overall (cf. Scheffler). Some define it as the moral theory which uses agent-neutral as opposedto agent-relative reasons (cf. Nagel). Larmore defines it as that moral theory which demands that wemust hold ourselves responsible, not just for what we alone do, but also (to the extent that this is in ourpower) for what will follow because of what everyone else will do as a result of our acting this way (cf.147-8). Raz lists seven theses as “historically associated with consequentialism”(268). To clarify therelations between all these definitions of consequentialism, and to weigh the pros and cons of each,would be a book in itself.572.4 THE STRUCTURE OF CND CONSEQUENTIALISM§1 Why pick the standard of promoting the most desirable combination ofsatisfied interests of individuals in their own well-being as our ultimateconsequentialist goal? The answer is that given what we want a morality for in thefirst place, it makes sense to construct it out of individuals’ interests in their wellbeing. It is a necessary (though not sufficient) condition for acceptable moralarguments that they must be practical and engage motivations that people have orcould acquire through moral education. But why, of all the imaginable ways toconstruct a morality out of individuals’ interests and a minimally objectiveconception of well-being, should we pick exactly this way? Again my answer mustultimately appeal to a wide reflective equilibrium. This type of consequentialismseems to me best able to meet all serious objections to standard rule utilitarianismwithout giving rise to the objections afflicting perfectionist teleologies. In particular,it meets the objection that consequentialist aggregation fails to respect the“distinction between persons” or “individuals’ separate existences”, and thereforedoes not treat them with equal consideration.12The ideal consequentialist goal (tobe clarified below) seems to me to embody the equal consideration principle in thedeepest possible way. It is not easy to see what more it could possibly require torespect individuals’ separate existences than to use, as our guiding ideal, thisconsequentialist goal.In reply to the familiar objections to utilitarianism based on aggregation,Haslett’s lucid defense seems to me worth quoting at some length:12 These two expressions come from Rawis and Nozick, respectively. For the debate about the merits ofutilitarianism, cf. e.g. J.J.C. Smart and Bernard Williams, Utilitarianism: For and Against (1973) andAmartya Sen and Bernard Williams, eds., Utilitarianism and beyond (1982). I was persuaded of themerits of CND consequentialism by D.W. Haslett’s Equal Consideration: A Theory of MoralJustification (1987), which defends a version of it at length against anti-utilitarian objections.58...a social pressure system, including the moral norms inherent in it, must beevaluated in terms of people’s interests.... [But] in terms of whose interestsshould it be evaluated? The answer proposed here is: In terms of everyone’s,giving each individual’s interests equal consideration. If, indeed, a social-pressure system is to be evaluated in terms of people’s interests, it is hard tosee how, merely by insisting that everyone’s interests be considered equally,the utilitarian could thereby be accused of not taking the separation of personsseriously.But the critics might press matters still further. Why, they might ask, shouldwe consider everyone’s interests equally, especially in that sense of “equalconsideration” being defended here’ we should consider everyone’sinterests equally because doing so is in everyone’s best interests. It is ineveryone’s best interests in that, from a standpoint of self-interest, each of uswould ourselves choose that we do so if fully informed and not subject toirrelevant influences — if, in other words, our choice were made under idealconditions, the very conditions, I submit, that would have to be met in orderfor our choice to have been made altogether freely. This is why equalconsideration is justified; this is the connection between self-interest andmorality; and this is the most profound reply to any who would claim thatthe utilitarian theory proposed here does not take the separation of personsseriously. For if evaluating social-pressure systems by considering everyone’sinterests equally is in the best interests of each individual— each separate,distinct individual— then how could this utilitarian theory be said to treatanyone as a mere means? (221-2)Thoroughgoing individualists of the libertarian kind will, of course, remainunmoved by Haslett’s argument. They may see transaction-by-transactioncalculations of mutual advantage, rather than the quest for impartial justificationsof interpersonal goals, as the only promising motivation or rationale for socialcooperation. I shall defer the discussion of libertarianism to Ch 3.5.The question I must address now is whether the proposed consequentialistgoal is fundamentally intelligible, self-consistent or coherent. What exactly is “themost desirable combination of satisfied interests of individuals in their own wellbeing”?The following account of the ultimate consequentialist goal seems to me to bethe most plausible way to spell out the idea of achieving the most desirablecombination of satisfied interests of individuals in their own well-being. The ideabehind this goal is not very different from the idea underlying the Golden Rule orIdeal Observer theories of ethics, implemented from a perspective which is (in59principle) enriched by a full understanding of all other perspectives. I find a thoughtexperiment suggested by Haslett useful (29-39), although I want to clarify its detailsand limitations. You can judge for yourself, in principle, what would be the requiredcourse of action from the standpoint of achieving the most desirable combination ofsatisfied interests of individuals in their own well-being. Imagine you yourself hadeach person’s life to live, and so experienced the interest of each person in her ownwell-being- including, of course, the very strong interests of most people in notbeing seriously disadvantaged for the sake of other people’s greater well-being, aswell as their interest in having pleasant hedonic states brought about by actualexperiences rather than hallucinations. Suppose you had perfect knowledge of theextent to which each possible course of action would satisfy the interest of eachindividual in her own well-being. The interest I am talking about here is neithersome paternalistic “objective” interest nor a completely unreflective “subjective”interest, but an idea of interest that reflects each person’s subjective interestconstrained by the very wide limits of rationally permissible interpretations ofhuman well-being. Then, if you were guided by the consequentialist standard, youwould choose the course of action which you would choose if you yourself had eachseparate person’s life to live (and were not, of course, permitted any inconsistenciesacross person-stages). Then, given the assumptions about perfect knowledge and theabsence of other influences except concern for your own well-being at each personstage, you would choose a course of action within a certain range of equallyacceptable courses of action.Other perfect knowers, going through the same thought experiment, willprobably not settle on exactly the same course of action. The courses of action theywould choose would, however, differ only to the extent that they would modifytheir unreflective subjective interest at each person-stage differently from you in thelight of the very wide limits of rationally permissible interpretations of humanwell-being (cf. Sec 3.4.2). To the extent to which different people who strive for this60goal in an ideal way would settle on different courses of action, to that extent theideal consequentialist goal may simply be fundamentally indeterminate.There is no reason I can see why the ideal consequentialist goal should notcontain a degree of fundamental indeterminacy, in the sense that it permits a rangeof options. Because of this fundamental, as well as the vast epistemic, indeterminacyof this ideal goal, practical decision-making procedures will have to be adoptedwhich are plausible contextual approximations to it.§2 Let me explain the rationale behind three key components of myformulation of the ideal consequentialist goal. First: What all consequentialistsmust want, by definition, is to promote well-being in some plausible sense. Wellbeing in this sense must be somehow related to people’s interests, but this relationisn’t simple. It is not plausible that simply allowing everyone to pursue their rawsubjective interests will promote well-being, because such interests may be self-undermining or socially destructive, and can completely miss the objective contentof the concept of human well-being (see Sec. 3.4.2). But neither will forcing a personto pursue someone else’s idea of what his or her objective interests are (no matterhow far these diverge from subjective interests) plausibly promote well-beingbecause a person’s interests are, after all, constitutive of this person’s identity, andthe free pursuit of one’s interests is perhaps the deepest source of humanfulfillment. Any plausible empirical index of what makes people happy just doesnot bear out the claim that people’s happiness is a matter of everyone’s realizing thesame determinate set of values. If neither the (distribution-sensitive) aggregation ofraw subjective interests nor the satisfaction of rigidly objective interests plausiblycaptures what is involved in promoting well-being, then obviouslyconsequentialists need some sort of compromise between the two. That is why Ithink we must, in defining the ideal consequentialist goal, temper the distributionsensitive aggregation of subjective interests with a theory of human well-being61which limits the range of interests whose satisfaction can count toward promotingwell-being. The concept of well-being must, therefore, be regarded as having“objective” content which imposes broad limits on the range of interests whosesatisfaction CND consequentialists think morally important (see Sec 3.4.2).Second: The ideal consequentialist goal requires you, first, to put yourself intothe shoes or psyche of each person and to assess, from that person ‘s perspective,what is in his or her interest before trying to find a way that will best harmonize allpeople’s interests. Some such “perspective-switch” must obviously be incorporatedinto any consequentialist moral standard which is to do justice to the fact of ourseparate existences. You are to imagine that you yourself would actually be theperson suffering every single one of the consequences of whatever course of actionyou settle on, with no compensation in a future life. But having done yourimagining, you will actually have to decide on one course of action, because onlyone course of action can be realized in the actual world- and that means that somepeople’s interests will unfortunately have to be frustrated and their well-beingcompromised. If you yourself suffered every single consequence withoutcompensation, you would want to frustrate as few vital interests as possible, andcertainly not sacrifice all of some people’s well-being for the sake of marginalincreases in the well-being of the majority. So plausible second-best criteria toapproximate the ideal consequentialist goal, as well as extremely difficultjudgments, will be necessary. But no self-consistent standard - consequentialist orotherwise - can escape these judgments or avoid ever sacrificing some people’s vitalinterests or even their entire well-being. There are genuine tragic choices where, nomatter what we do, some people will have to suffer so that a greater number ofothers won’t have to suffer even more. So our proposed consequentialist standardpermits the inevitable, for example, sacrificing some so that others may survive. Butbuilt into this ideal standard is, it seems to me, the most plausible conceptualprocedure for capturing what giving equal consideration to everyone’s interests62would mean. How to make this conceptual procedure even remotely practical isanother question, the problem of second-best approximation, which depends on firstgetting as clear as we possibly can about our ideal aim.Third: It may seem inconsistent of this standard to require you to find aresolution among conflicting interests from the perspective of having to liveeveryone’s life. The deep problem here is not that it is actually impossible for you toknow what everyone’s interests are. There clearly is a fact of the matter here, andthere are in any case various practical ways to discover what people’s approximateinterests are. But when it comes to harmonizing all the conflicting interests ofindividuals, there exists no neutral perspective, in addition to all the separateperspectives of individuals. To be sure, each individual can in principle harmonizeconflicting interests from his perspective, but if all perspectives of all individualsbecome yours, which of the resulting perspectives should you choose to settle on thecourse of action that necessarily sacrifices some people’s interests for the sake ofsatisfying those of others? Whose perspective is the privileged one in whose interestthe conflict will get resolved?I think this objection, while troubling, is not fatal to distribution-sensitiveconsequentialism. Take first a two-person universe in which the interests of A andB are in conflict. What would it mean to resolve the conflict by choosing the courseof action which each would choose if each had both lives to live? If the conflict isresolved from A’s perspective, A’s interests would seem likely to fare better thanB’s; if from B’s perspective, B’s interests would seem likely to fare better than A’s.A’s and B’s perspectives are the only ones there are; there is no third, neutralperspective from which conflicts could be resolved. But this is an oversimplifieddescription of the situation. It is true that the conflict must be resolved from eitherone perspective or the other; it is not true that those two perspectives, after beingsubjected to the process described in our thought experiment, are necessarily very farapart. The very point of the thought experiment is that you imagine being the63subject of other persons’ experience. If each of two people came to know what theother person’s experience is like, then a psychological merging of perspectives couldtake place which would not be identical to the original perspective of either. It is thisenriched perspective- which, in the real world, we can of course only approximatein the crudest way - from which the consequentialist standard asks you to resolveconflicts between different people’s interests. If, in the two-person case, yourjudgment is informed by the impartial sympathy for both perspectives required bythe Golden Rule, you would choose that course of action which ensures orpromotes the well-being of both individuals together, wherever that is possible. Youwould not routinely maximize (total or average) interest satisfaction between A andB, because in that case, B might forever end up making sacrifices, and that is not thechoice you would make if you actually had each life to live and experienced B’sinterests, including his interest in not being seriously disadvantaged for the sake ofmaximizing interest satisfaction overall. In situations where sacrifices on the part ofeither A or B become inevitable, and the current level of well-being of both isroughly equal, both will be called upon to sacrifice some amount. In some tragicsituations where, for example, one person’s life has to be sacrificed for the other’ssurvival, there simply may not be a morally preferred answer to the question ofwhose life should be sacrificed (although some procedural ways of settling thisquestion may be morally preferable to others).It may still be objected, however, that it is far from clear how a definitedecision results. Will at some point a commitment to maximization take over, orare there some distributive constraints built into the imaginative projection ofmyself as having everyone’s life to live? As McDonald has put this objection,What you picture [in the above thought experiment] is that when A looks atboth his own and B’s life and B does the same their judgment will converge.The process of convergence seems to be causal. There is no perspectivalcriterion advanced, such as maximization. But then it may well matter how64A and B feel about the separateness of individuals. What if A is in his guts amaximizer and B a separator (valuing the separateness of persons more)?Moreover, what if A and B are not equally prudent in their enrichedjudgments- what if A’s enriched sense of their interests leads him to approveof risking admittedly vital interests for trivial ones, just to add zest to life? Tobe sure A and B each picture what is best on the basis of their enrichedperspectives. Neither leaves out of account something the other has included.But each values or weighs matters differently. Or to use another analogy, eachcolours the same picture with a different hue. (Personal comm.)In principle we could, of course, handle the sort of problems which McDonald raiseshere by requiring iterations of the ‘perspective-switch’ thought experiment untilconvergence is achieved. We can imagine the first attempt at achieving an enrichedperspective to remain unenriched by such factors as people’s differing in riskaversion or being “maximizers” as opposed to being “separators” at heart. Duringthe second run of our thought experiment, once the results of the first run are in,we can factor into our enriched perspectives a mutual appreciation of the extent towhich risk aversion and maximizer-separator preferences are differentiallydistributed. Even given such mutual appreciation of differences, however, risk-friendly maximizers or risk-friendly separators may still weigh outcomes differentlyfrom risk-averse maximizers or risk-averse separators. Perhaps we need to imagineadditional iterations of our ‘enriched perspective’ thought experiment to achievesufficient convergence in our judgments about what constitutes the most desirablecombination of satisfied interests of individuals in their well-being.But this suggestion strikes me as rather gimmicky. There is no denying thatboth problems McDonald raises— the problem of differences in risk aversion andthe problem of reconciling maximizing and distributive considerations — constituterich sources of rational disagreement about justice. There will be many occasions,then, when consequentialists will, at the first-order level of moral reasoning,converge on similar decisions only in a very limited sense. While theconsequentialist well-being test will rule out some distributive options asunacceptable, the choice among a wide range of remaining options will have to be65made at the second-order level of moral reasoning, by seeking conventions ofmutual accommodation through political bargaining.I cannot discover any way to spell out, in the abstract, just at what point wecould allow maximization to take over without opening ourselves up to the chargeof failing to respect the separateness of persons. We must admit, I believe, that atlevel of fundamental theory divorced from specific contexts, imagining myself tohave everyone’s interests can generally yield only very vague answers to theproblem of resolving different people’s conflicting interests. We can say, forexample, that it is generally better to satisfy a greater aggregate of interests of a largernumber of people than a smaller aggregate of a smaller number, provided only thatthe vital interests of the few are not sacrificed for the less vital interests of the many.At the applied level, however, once we know what the actual interests of peopleaffected by a decision are, we will usually be able to make more or less confidentqualitative judgments which strike a credible balance between aggregative anddistributive considerations.I conclude that the proposed consequentialist goal is self-consistent, albeitafflicted with both fundamental and epistemic indeterminacies. How we are tojudge, in practice, what best approximates this goal will be discussed in greater detailbelow (Ch 3.3).§3 My project is to identify plausible sources and limits of rationaldisagreement about justice. I approach this project by (1) tentatively accepting aconsequentialist moral framework as our most promising metaethical option; and(2) examining how this framework constrains principles of justice. But my approachfaces the following objection:You insist that morality is constructed by ourselves as an instrument for thepromotion of human well-being and that it constrains ideas about justice. Buthow can ideas about justice be constrained by your artificially constructedmoral standards unless you yourself have built these constraints into them?66And how can you build these constraints into them, unless you already knowwhat the right ideas about justice are? You reject the contractarian approachas question-begging. But isn’t your attempt to find rational constraints onpermissible ideas about justice by deriving such constraints from yourconsequentialis t goal hopelessly question-begging as well?This seems like a devastating objection. But it rests on a misunderstanding. Itinterprets the constructivist project of contractarians and that of consequentialistsin analogous ways: as a foundationalist enterprise. But while moralfoundationalism is indeed characteristic of contractarians (at least of the mutual-benefit individualists among them), the form of consequentialism I endorse iscoherentist rather than foundationalist. We cannot avoid talking metaphoricallyabout our most central epistemic assumptions. What does the justifying of moralclaims, on my account, is a web of cohering considerations whose basic geometry isgoverned by a clear understanding of what we want a morality for, and whosepoints of contact with the empirical world are both universal and contextualinterests, a fairly open-ended conception of human well-being, social meanings andconventions, widespread and deeply ingrained patterns of expectations andsentiments, and the best available information about the state of the world and thecausal workings of things. The ideal consequentialist goal, and plausible derivativesof it, are simply nodes in the web of our examined moral and non-moral beliefs,“Archimedean points’ within a reflective equilibrium upon which many strands ofreasoning converge and from which many strands of reasoning radiate outward (cf.Rawls, 260). What I have called the ideal consequentialist goal is meant to sum upthe general spirit and guiding direction of the consequentialist enterprise, and notsomehow to stand alone as a self-evident, deductive foundation for moral decisionmaking which is known prior to and independently of everything implied by it.My project of rationally reconstructing morality is, therefore, not linear in thesense of starting from independently justified foundations which then allow us todeduce principles of justice without circularity. We must construct all parts of ourmorality together in a coherent way, and cannot first produce an account of moral67foundations, untainted by any assumptions about justice, and then somehow derivestandards to constrain ideas about justice. If our ideas about justice had to beconsistent with moral standards which are untainted by assumptions about justice,then obviously justice-based considerations would be routinely overridden by thenon-justice considerations contained in these moral standards. And that, as manycritics of utilitarianism point out, would fundamentally conflict with ourconsidered moral judgments. It would also conflict with our background theory ofmorality’s main job and would, therefore, be doubly ruled out by the method ofwide reflective equilibrium. It is clear that morality’s job can only be accomplishedby taking people’s justice-based claims seriously as an important source ofinformation about their well-being and by avoiding as far as possible the destructiveunleashing of their sense of injustice.Let me consider the feature of the basic consequentialist standard whichseems perhaps most obviously question-begging. Why should consequentialistconstructivism embody a general commitment, not simply to maximizing interestsatisfaction, but also to considering different people’s interests equally? Therationale for including, at the core of the basic consequentialist standard, acommitment to equal consideration has to do with elementary facts about humanpsychology which Brian Barry, among others, has eloquently spelled out:Someone who was engaged in impartially choosing principles to govern hislife with others would not endorse a principle on the basis of its favoringhimself, his friends and relations, or those with whom he felt some kind ofaffinity. But in just the same way someone seriously engaged in the search forprinciples that could not reasonably be rejected by others engaged in the samesearch would surely recognize that it would be a waste of time (as well as abreach of good faith) to put forward a principle whose only merit was that itwould be favorable to himself, his friends and relations, or those with whomhe felt some kind of affinity. For it is obvious that those who were put at arelative disadvantage by such a principle would wish to reject it, and it couldnot possibly be said that they were being unreasonable in so doing. (1989: 290)In constructing the details of our basic moral standards, we had better incorporatefrom the start, as integral building blocks, some explicit reference to a sort of68impartiality which plausibly captures, as traditional utilitarian theories did not, thedistributive concerns of individuals making claims qua individuals. The reasons aretwo-fold. First, the greatest well-being overall varies in direct proportion to the wellbeing of individuals, and an individual’s justice-based claims to social resources area useful, though far from infallible, source of information about what is likely topromote that individual’s well-being. Second, if an individual’s expectations to theshares of social resources which he sincerely regards as his due get disappointed in away that cannot be explained to that individual in terms of the necessity ofbalancing competing claims of comparable urgency, there is a great likelihood ofanger, humiliation, resentment and similar sentiments. These are, from aconsequentialist point of view, undesirable in themselves and also have potentialfor being unleashed in socially destructive ways. That is why any plausible form ofconsequentialism must take the difficult job of balancing the competing claims ofindividuals to a share of social resources in a principled, publicly defensible wayextremely seriously.§4 It may be doubted whether distributive constraints on the aggregation ofgoods can themselves, consistently, become one of the goods to be factored into thisaggregation. But “goods”, in the relevant sense, are simply whatever human beingscare about within objectively plausible limits of what can constitute personal wellbeing. “Constraints” on the aggregation of goods are whatever measures arenecessary or useful to prevent individuals from sabotaging each other’s pursuit ofpersonal well-being. Where do we get such constraints from? Where else, but fromasking ourselves what basic types of goods we care about sufficiently to want to buildthem into the goals spelled out by our ultimate consequentialist goal. Wesufficiently care about having the distinction between persons recognized indistributional matters to make the recognition of this distinction a justifiableconsequentialist constraint. So by the definition of what makes something a good69and a constraint, respect for the separateness of persons is both a good and aconstraint. In fact, whether one says that respect for the separateness of persons andhappiness are both goods to be weighed against each other in pursuit of anacceptable mix, or that the one functions as a constraint on the pursuit of the other,is a purely semantic matter for a moral constructivist.Including the good of equal consideration, along with the good of happinessand possibly other values, into our ultimate consequentialist goal makes thisstandard more complicated to apply as well as fundamentally indeterminate inmany cases. But no form of consequentialism could avoid this complicationwithout becoming immediately vulnerable to persuasive counterexamptes. Onecannot consistently criticize a fundamental moral standard both for not takingseriously the separateness of persons and for becoming more complex and lessdeterminate as a result of doing so. Of course, the ideal consequentialist goal willrequire plausible second-best approximations to become normatively useful inspecific contexts of application (see Ch 3.3).As we construct, or more accurately reconstruct and modify, the socialpressure system called morality in a rationally transparent way, we can’t all at oncetoss everything out that exists either in the philosophical tradition or in the culturalworld around us. Although nothing is beyond question, and every component ofour morality must ultimately find its functional rationale in the whole, we mustimmediately distinguish two intertwined components of any practicable moralitywe are likely to end up with: components for which a universal rationale can begiven in terms of the reasonable preconditions for any plausible type of social living;and a much larger proportion of components whose rationale is culturally orotherwise contextually relative to the way of life of specific groups and their selfunderstanding as members of that group. To expect a non-contextual rationale forevery component of our rationally reconstructed morality, or even to ask that thetwo basic types of components be neatly separable, is not a reasonable expectation of70the enterprise of constructivis t, non-foundational, distribution-sensitiveconsequentialism.2.5 RAWLS’ AVOIDANCE OF MORAL THEORY§1 If one of the greatest moral philosophers of our time thinks he can bypassmost metaethical issues in justifying principles of justice, we cannot simply dismissthis idea. The case for explicit metaethical commitments (and for a broadlyconsequentialist commitment in particular) will be strengthened, I believe, by brieflyconsidering Rawis’ rationale for wanting to avoid moral theory at almost any cost.In his later writings, Rawls still maintains that, ultimately, the wide reflectiveequilibria of different people will converge essentially on his two principles ofjustice. But he now gives these principles an explicitly pragmatic-communitarianinterpretation (cf. 985, 1988). In ‘The Priority of Right and Ideas of the Good”, forexample, he saysthe are several ways in which I would now revise the presentation ofgoodness as rationality; perhaps the most important would be to make surethat it is understood as part of a political conception of justice viewed as aform of political liberalism, and not as part of a comprehensive moraldoctrine. The distinction between a comprehensive doctrine and a politicalconception is absent from Theory, and while I believe nearly all the structureand substantive content of justice as fairness (including goodness asrationality) is unchanged when it is seen as a political conception, theunderstanding of the view as a whole is very significantly shifted. CharlesLarmore in his Patterns of Moral Complexity.. .is quite correct in vigorouslycriticizing the ambiguity of Theory on this fundamental matter. (1988: 254footnote)If WREM applied to the problem of social justice is followed through beyond theconflict of ideologies - Rawls now seems to think - the ultimate equilibrium to bereached does not yield a rationally superior moral theory, but merely “fundamentalintuitive ideas viewed as latent in the public political cultur&’(1988, 252). Theseideas alone can serve as principles of political accommodation among groups with71irreconcilably diverse moral beliefs. A theory of justice has a very limited role:defining whatever terms of political accommodation are most likely to promotepeace and cooperation in a particular society. It would be a mistake to base one’stheory of justice on any controversial beliefs about the nature of morality, becauseonly by not taking sides in the intractable debates of moral theory can a theory ofjustice discharge its social role. The terms of mutual accommodation among groupswith diverse ideas of the good may be very different for different societies. Inadvanced pluralist democracies, Rawis claims, his own principles of justice definesuch a society’s best chance for achieving a reasonably peaceful and stable socialorder.’3The hope that WREM allows the political philosopher to altogether avoid thecomplex debates in moral theory- a hope which Rawls perplexingly labels “themethod of avoidance”(1988, 240) - certainly has its initial attraction. But is this hopejustified? If wide reflective equilibrium methodology subsumes all other moralarguments within it - as I suggested above - then it seems doubtful that bypassing allquestions about the nature of morality is compatible with WREM. On the otherhand, if no agreement on the nature of morality is realistically to be hoped for, thenthis verdict itself will, of course, become one of the well-established backgroundbeliefs to be factored into our reflective equilibrium. But it is far from clear that thefact of irresolvable disagreement about morality among informed people of goodwill - rather than implying moral skepticism - would imply that the task of justicetheorizing consists purely in making explicit whatever fundamental intuitive ideaslatent in a society’s public political culture may best lend themselves to the purposeof mutual political accommodation among different groups.13 The idea of justifying liberal principles of justice purely in terms of their ability to provide a modusvivendi is eloquently defended by Charles E. Larmore, Patterns of Moral Complexity (1987), p. 74-6 and123-7.72A number of troubling questions arise in the face of Rawis’ pragmaticcommunitarian shift. Is the fact that some set of principles fulfills the social role ofjustice (i.e. defining terms of political accommodation) sufficient to accept them asprinciples of justice- rather than, say, as a concession to power politics? Canprinciples actually fulfill the social role of justice unless they are first recognized orjustified on independent grounds- that is, on the basis of an understanding ofjustice as more than simply those widespread intuitions which will best promotepolitical accommodation? Can reflective equilibrium methodology actually justify asuitable (i.e. sufficiently determinate) set of principles convincingly as principles ofjustice to allow these principles to fulfill the social role of justice without involvingany controversial metaethical commitments whatever? These issues are obviouslyimportant for an inquiry into the sources and limits of rational disagreement aboutjustice. I want to approach them in two steps: First, why exactly did Rawis find itnecessary to shift to such a pragmatic-communitarian justification of his theory?Second, is Rawl& new interpretation of his theory be convincingly derivable byapplying WREM to the problem of social justice?§2 What motivated Rawls’ pragmatic-communitarian shift? In “Justice asFairness: Political not Metaphysical”, Rawis tells us that understanding his theory asa political conception of justice is intended to counter the criticism thatthis conception depends on philosophical claims I should like to avoid, forexample, claims to universal truth, or claims about the essential nature andidentity of persons. (1985: 223).The claims to universal truth Rawls wants to avoid are claims about which generalconception of the nature of morality is rationally superior. “The essential point,” hesays,is this: as a practical political matter no general moral conception can providea publicly recognized basis for a conception of justice in a modern democraticstate.... [Sluch a conception must allow for a diversity of doctrines and the73plurality of conflicting, and indeed incommensurable, conceptions of thegood affirmed by the members of existing democratic societies. (1985: 225).The first and obvious reason for Rawlst shift in the interpretation of his theory ofjustice is his desire to find principles of accommodation between differentmoralities without taking sides in the intractable disputes of moral theory. But thereis another important reason for this shift. By interpreting his two principles ofjustice, not as spelling out the demands of universal justice sub specie aeternitatis,but far more modestly as a basis for an overlapping consensus among citizens ofadvanced pluralist democracies, Rawls hopes to meet the communitarian objectionabout the nature of the self presupposed by his theory. Rawls’ description of theidentity of the bargainers in the original position, stripped of most of theirindividuating characteristics, seemed to postulate a fictitious disencumbered” selfwhich can make choices in splendid isolation from the particulars of its socialenvironment. Rawls had even claimed (although in a different context) that ttheself is prior to the ends which are affirmed by it(56O).Communitarians such as Maclntyre, Sandel and Taylor were quick to arguethat this Kantian idea of the self as choosing its goals according the universaldictates of reason is untenable: the self is embedded in a particular community andculture, it is constituted by its choices, and these choices are fundamentallydependent on its social roles and relationships.’4Rawls answers this criticism in two steps. First, he stresses that the descriptionof the bargainers in the original position was always intended, not to capture theessential nature and identity of persons for whom his theory defines justarrangements, but purely as an expository device designed to make vivid a14 Cf. Maclntyre, After Virtue (1981) and Whose Justice? Which Rationality? (1988); Sandel,Liberalism and the Limits of Justice (1982); and Taylor, Hegel and Modern Society (1979). The Kantianidea of the self attacked by communitarians assumes, as Kymlicka puts it,“that the self is prior to its socially given roles and relationships and is free only if it iscapable of holding these features of its social situation at a distance and judging them accordingto the dictates of reason.”(1990)74conception of what constitutes reasonable conditions of impartiality for choosingprinciples of justice (1985, 237). That Rawl& reply to his communitarian criticsrepresents more than a minor clarification can be gleaned from his admission that• . .it was an error in Theory (and a very misleading one) to describe a theory ofjustice as part of the theory of rational choice, as on pp. 16 and 583. What Ishould have said is that the conception of justice as fairness uses an accountof rational choice subject to reasonable conditions to characterize thedeliberations of the parties as representives [sic] of free and equal persons; andall of this within a political conception of justice, which is, of course, a moralconception. There is no thought of trying to derive the content of justicewithin a framework that uses an idea of the rational as the sole normativeidea. (237, footnote)Second, Rawls now tries to make his theory of justice compatible with thecommunitarian idea of the self as embedded in communal practice simply byadopting a radically communitarian interpretation of his theory. He makes clearthat his theory is not a universal theory of justice for human beings in a widevariety of cultural settings, but on the contrary a highly restricted theory applyingonly to persons whose identity has already been shaped by conditions prevailing in aparticular type of community - namely advanced industrialized pluralistdemocracies. Rawls stresses that his theory “starts from within a certain politicaltradition”, and it“tries to draw solely upon basic intuitive ideas that are embedded in thepolitical institutions of a constitutional democratic regime and the publictraditions of their interpretation’(1985, 225).People whose self has been shaped by such a cultural environment, Rawls seems toargue, will generally have the sort of self which values roughly his list of primarygoods as all-purpose means to the satisfaction of their goals.Rawls’ pragmatic shift attempts to undercut the case of his communitariancritics by reinterpreting his theory as a version of communitarianism. But Rawls’bid to join the communitarian camp seems to have made him few friends amongcommunitarians. The debate seems simply to have shifted from the question as to75whether Rawis’ morally justified liberalism depends on an implausible theory ofthe self as detachable from its ends to the question as to whether Rawis’pragmatically justified liberalism can take seriously the value of community as afundamental ingredient of the good life for human beings.§3 Is Rawis’ pragmatic-communitarian interpretation of his theory of justiceconsistent with regarding reflective equilibrium methodology as the ultimatemethod of justification? If the two were inconsistent, it is clear that Rawls wouldhave to give up his pragmatic justification rather than WREM. Obviously, the ideathat a pragmatically justified principle (i.e. one which promises to promote politicalaccommodation) could override a principle which is justified by a wide reflectiveequilibrium (the greatest possible coherence of a moral theorist’s beliefs) isincoherent. Therefore, Rawls’ pragmatism is sustainable only if it is consistent withWREM.Given the central importance of showing the pragmatic justification ofprinciples of justice to be fully consistent with and, in fact, implied by WREM, it isstrange that, in Rawls’ 1985 paper, the only reference to his earlier much-acclaimedmethod of moral justification occurs as a parenthetical remark in the followingcontext:We look, then, to our public political culture itself, including its maininstitutions and the historical traditions of their interpretation, as the sharedfund of implicitly recognized basic ideas and principles. The hope is that theseideas and principles can be formulated clearly enough to be combined into aconception of political justice congenial to our most firmly held convictions.We express this by saying that a political conception of justice, to be acceptable,must be in accordance with our considered convictions, at all levels ofgenerality, on due reflection (or in what I have called “reflectiveequilibrium”). (1985: 228)I admit to being puzzled by how this remark, or any of the other things Rawls saysin his 1985 paper, could possibly be construed as sufficient to make the logicalcompatibility between WREM and the pragmatic thesis clear and convincing.76Although Rawis has weighty reasons for his pragmatic-communitarian shift, thereare even weightier considerations for thinking that Rawls’ “method of avoidance”is inadequate for justice theorizing.(1) The epistemic worry: Rawis’ argument for bypassing moral theory reliesheavily on the claim that no sufficient measure of agreement is to be hoped for inthis area. But how does someone know this to be true, and with what degree ofcertainty? From the fact that universal agreement has not happened in the past, itdoes not follow that expecting a sufficiently widespread measure of agreement inthe future is completely unrealistic. Can we categorically say that none of ourmetaethical options shows any promise whatever, if articulated with clarity, to winthe adherence of at least a clear majority of rational people? In any case, such arather pessimistic conclusion can obviously not be accepted by all justice theoristssimply on the strength of our faith in Rawis’ judgment, but must be reached in widereflective equilibrium, as the outcome of examining the pros and cons of ourvarious metaethical options in the light of sociological realities. The pessimisticconclusion that the disputes of metaethics are intractable presupposes intimateacquaintance with these disputes and a deep understanding of the potential of eachpossible moral theory. In what sense, then, is it correct to say that justice theorizingcan bypass moral theory? And how plausible is it for Rawls to imply that all justicetheorists employing WREM must emerge with the pessimistic conclusion that nomoral theory shows sufficient promise to become widely accepted as our mostrational understanding of morality? Certainly not all justice theorists, applying themethod of reflective equilibrium the way Rawls’ instructed them to do in A Theoryof Justice, have ended up with Rawls’ skepticism about the prospects of findingsome measure of agreement about the best understanding of morality by ruling outsome currently popular metaethical positions as rationally unacceptable.(2) The pragmatic paradox: But suppose Rawls is right, and WREM forces thepessimistic conclusion on us that we cannot hope to resolve the conflict between77rival metaethical position or to win public acceptance for any such resolution.Would it follow that we must then adopt a purely pragmatic justification ofprinciples of justice and accept whatever principles promise to work best, relative toa given society, in producing an “overlapping consensus”? It seems to me that theattempt to give a purely pragmatic justification for principles of justice confronts asomewhat analogous paradox to that which haunts William James’ attempt to givea pragmatic justification of religious belief. J. L. Christian explains the pragmaticparadox as follows:A human condition in which an idea must be believed to be true in terms ofcorrespondence- that is, one must be convinced that some object/event existsas a real entity- before the idea can be considered to be true on the pragmatictest. For example, one must believe that immortality exists as a realevent.. .before the belief in immortality can produce positive results in his life.(596)Let me make clear the analogy I see between the paradox faced by James and Rawls.Just as James’ pragmatic justification of religious belief will have little or no successin instilling such belief in the absence of support from antecedently held religiousconvictions, so Rawls’ pragmatic justification of principles of justice will have littleor no success in motivating agreement on these particular principles in the presenceof conflicting moral convictions. A pragmatic justification of principles of justice, ifunderstood as merely pragmatic, will not have sufficient normative force in theminds of people to override their antecedently held beliefs about justice. People’sbeliefs about justice are generally nourished by deep moral and often religiouscommitments. It would seem fairly obvious that even the most pragmaticconception of social justice must claim moral authority to overrule the dictates ofthose moral and religious commitments which conflict with any such conception;and to that extent (which may be considerable) it cannot avoid taking sides inmetaethical disputes. In ethics, what counts as a sufficient justification alwaysdepends on what is the best justification someone thinks he can get; and if someone78thinks that he can directly intuit the truth about social justice with absolutecertainty, or simply look it up in some Holy Book, he will hardly take great interest(unless he has no choice and has already become convinced of the virtue of liberaltolerance) in Rawis’ pragmatic justification for a set of principles reached by acomplicated process of wide reflective equilibrium, let alone accord them overridingmoral authority.’5Only insofar as Rawls’ reflective equilibrium arguments areacknowledged to have genuine moral force, therefore, will they persuade competinginterest groups in society to strike a bargain that does not simply reflect powerdifferentials and threat advantages. So when Daniels claims, “I think it is a virtue ofreflective equilibrium that it leaves open metaethical considerations”(282), he seemsto neglect seeing the issue from the perspective of those with strong metaethicalcommitments.(3) The indeterminacy of pragmatic justifications: Is it really plausible to thinkWREM necessitates regarding all and only those principles of justice as rationallyacceptable which are justifiable in a purely pragmatic way? Are there any suchprinciples at all? And if there are, do they resemble anything recognizable to peopleas giving each of them justice? And how convincing is Rawls’ claim that his twoprinciples would be singled out from among a host of possibilities by a pragmaticjustification?Two decades of criticism of Rawis’ theory have made it clear that the searchfor a purely pragmatic mode of political accommodation is highly unlikely toconverge on his particular principles of justice. As Arneson stresses, Rawis’ hope ofsome core consensus in democratic society seems unduly optimistic:15 Rawis only touches on this objection to his approach: “One might say.. ..that to develop a politicalconception of justice without presupposing, or explicitly using, a metaphysical doctrine, for example,some particular metaphysical conception of the person, is already to presuppose a metaphysical thesis:namely, that no particular metaphysical doctrine is required for this purpose.”(1985, 240 footnote) Hisreply: “Following the method of avoidance, I should not want to deny these claims.”(240). But canRawis legitimately invoke the method of avoidance in answer to the charge that the claim of beingable to avoid controversial meta-ethical assumptions is itself a controversial meta-ethicalassumption?79• . the Kantian ideal [of persons who give priority to preserving the conditionsfor the exercise of their moral powers to choose and to cooperate with otherson fair termsj is present in democratic culture- but so are several other firmlyrooted conflicting ideas, such as (a) competitive individualism, (b) patriarchalconservatism, and (c) Judeo-Christian benevolence... .Their adherentswould... .politely decline the invitation to join in an overlapping consensuson Rawis’s egalitarian liberal principles of justice or anything remotely closeto them. (705).The argument at this point in the debate becomes admittedly somewhat subtle andelusive, and it is not clear who is begging the question against whom. I take Rawis tobe arguing that a pragmatic accommodation is in everyone’s interest in preciselythose situations where any particular group’s attempt to get the better of all otherswould have disastrous consequences for all groups, and that his two principles ofjustice spell out the most rational terms of this accommodation. Arneson can retortthat there are many possible modes of accommodation that are pragmaticallyrational, some based purely on existing power differentials and threat advantages,others - such as Rawis own proposal - laden with more or less controversial moralassumptions; and that Rawls has failed to give the various interest groups acompelling rationale to converge on his two principles as the terms of their mutualaccommodation. Certainly, if history is the relevant test here - and for athoroughgoing pragmatist it would seem to be ultimately the only relevant test - itseems clear that Rawis’ theory has made few converts in the two decades of itsexistence, even among serious students of his work.§4 It seems to be Rawis’ view that, since we are not going to get agreement ona moral justification for a particular theory of justice, the task of the politicalphilosopher at the present time is restricted largely to a sort of damage control in theface of this fact (cf. 1985, 227). But Rawis’ radical pessimism about the prospects ofsome modest measure of moral agreement among informed people of good willseems to be at least as controversial as Rawis’ optimism that, in the absence of moral80agreement, we can still get pragmatic agreement on his two principles of justice (oron any other usefully determinate set of principles). Therefore, Rawls pragmaticjustification for his particular theory of justice can be considered to have failed on itsown pragmatic terms. But even if Rawls is right that no agreement on a moraljustification for a theory of justice is to be expected at the present time, it would notfollow that the political philosopher is simply restricted to the role of mediatorbetween interest groups, rubberstamping whatever forms of mutualaccommodation may be feasible in a particular society at a particular time. If a strongreflective equilibrium-based case can be made for accepting a broadlyconsequentialist moral framework, and if this framework turns out to support aliberal social order, not simply as providing pragmatic terms of mutualaccommodation, but as embodying morally justified principles of justice, then aliberal social order is quite likely to win the allegiance of many thoughtful people.Such allegiance, while falling far short of universal agreement, may contributesubstantially to keeping a liberal regime stable.The consequentialism I defend stands in sharp and obvious contrast to thesort of modus vivendi pragmatism about a just social order which tries to beagnostic about the possibility of a rational moral theory and wants to leave theclutter of clashing moral dogmatisms in place. A pragmatism of this kind seemslikely to collapse into mere power politics. Even though accepting a consequentialistframework for moral reasoning will not abstractly fix the demands of justice in avery precise general way, it does make the reasons for the uncertainties inherent inprinciples of justice at the fundamental level clearly intelligible. If appreciated asperhaps our most credible alternative to moral skepticism and Realpolitik, a broadlyconsequentialist understanding of morality may motivate people to settle manydisagreements about justice at the level of negotiated compromise instead ofallowing them to degenerate into raging confrontations.81PART 3: CONSEQUENTIALISM AND RATIONAL DISAGREEMENTABOUT JUSTICE3.1 DISAGREEMENT ABOUT JUST-MAKING CRITERIAAfter clarifying the concept of social justice and other key concepts, this chapterconsiders two central disagreements associated with the cluster of considerationsrelevant to determining justice: the criteria definition problem and thecriteria inclusion problem.3.1.1 The concept of social justice§1 The concept of “justice is almost invariably invoked wheneverindividuals or groups advance claims to society’s resources. As a first step towardclarifying this concept, we must obviously be able to establish some parameters ofrelevance, some basic agreement about what we take people to disagree about whenthey disagree about justice, some widely assumed core meaning of the concept ofjustice as opposed to competing interpretations of this concept.’The common assumption most people seem to make when they argue aboutwhat is just or unjust is that certain valued things are certain people’s due or areowed to them for a variety of possible reasons, and that social distributions of thosevalued things should try to give individuals their due. To say that the core meaningof “justice” is “to give everyone their due” is unhelpful, however, unless the notionof something being someone’s due is clarified. For what reasons, or on what basis,do people typically claim that something is their due?1 It is not a foregone conclusion that there must be a common idea unifying all uses of this concept toenable people to argue meaningfully about what justice requires. It is quite plausible that the concept of“justice” can be understood in the manner of Wittgensteinian family resemblance. Such an analysis ofthe concept allows room for disagreement about some of the exact items which make up the cluster ofjust-making criteria, even if these criteria are regarded as constitutive of the core meaning of “justice”.There is also the more radical possibility that “justice” is used equivocally - that there are reallyseveral senses of this term linked not even by family resemblance. But the sense of justice that I aminterested in, that defines my subject matter here, is only that invoked (i) by individuals claimingsomething as their due, and (ii) by attempts to work out theories about how best to respond to suchclaims.82This question is perhaps best approached negatively. People typically thinkthat they have not been given their due (and speak of injustice) when it seems tothem that, in the distribution of valued things, individual merits or achievementshave been overlooked, needs disregarded, arbitrary inequalities introduced, rightsviolated, or expectations of reciprocity disappointed.In general we could say that considerations of justice arise in contexts wherethere are benefits or burdens to distribute, and where there are too many claimants(for the benefits) or not enough volunteers (for the burdens). Typically, argumentsto decide the justice of an individual’s bundle of benefits and burdens appeal to arather vaguely demarcated cluster of evaluative considerations, such as merit ordesert, needs, equality, rights or expectations held in good faith. I shall call claimsbased on such considerations “justice-based claims” (without thereby implying thatjustice requires every such claim to be satisfied). In a preliminary way, then, we candefine the core meaning of “social justice” as “appropriate responsiveness to theclaims which members of a politically organized society make - as individuals or asgroups - to a share of social resources in pursuit of their well-being, by appealing todesert, need, equality, rights or good-faith expectations.”2I now want to briefly clarify a set of important ideas: the distinction between atheory of justice and a mere conception; the “problem of social justice”; and the twoseparate levels at which I think the problem of social justice must be conceptualized.By a theory of justice I mean a systematic conception of justice which includes2 We can usefully distinguish justice at the personal, the social and the global level. Personal justiceapplies ideas about justice to the distribution of goods in small-scale, usually face-to-face contexts. (Ifollow the practice of using the term “goods for both benefits and burdens.) Social justice applies ideasabout justice to the context of goods whose distribution is thought to be a feasible and proper concern ofpolitical regulation. Global justice would go even further and apply ideas about justice to thedistribution of goods among all people or even all sentient beings in the world. Although a personsinitial understanding of the concept of justice is likely to arise out of experiences in the realm ofpersonal justice, the transfer of our understanding between personal, social and global justice is clearlynot all one way. The concerns of justice at these three levels grade into each other, but as will becomeclear, it would be a mistake to think that the distinction between them simply marks an expandingcontinuum involving only a quantitative increase in complexity (cf. especially Sec 5.1).83standards to assess the justice of important social distributions in a principled way,along with proposed justifications for these standards, and more or less detailedinstructions for how to implement or credibly approximate them in practice. By aconception of justice I mean any interpretation of what justice requires, no matterhow confused, that goes beyond the core meaning of justice as defined above. Aconception of justice, at its simplest, may be embodied in judgments about specificcases, or in principles at varying levels of generality and determinacy, or - at its mostcomplex - as a well-elaborated, comprehensive theory. The problem of social justice,which theories of justice try to solve, is how to resolve disagreements about socialdistributions in a way that has potential to become acceptable to rational people asan appropriate response to their justice-based claims and to lead to stable terms ofmutual accommodation.It is important to distinguish two fundamental levels of the problem of socialjustice, because many of the disputes about justice stem from confusion betweenthese two levels and failure to appreciate the difficulties of achieving a fit betweenthem. Let me call these the level of “feeling injustice’ and the level of doingjustice??, and explain briefly what happens at each level. (The asymmetry of termshere is intended, because I want to make the point that doing justice does notnecessarily imply avoiding injustice, in every plausible sense of that term.) At thelevel of feeling injustice, individuals have strong expectations about what they canclaim as their share of social resources in particular contexts, and they tend to feelanger, resentment or indignation (a “sense of injustice??) when such expectations aredisappointed. By the familiar process of Humean sympathy, personally unaffectedbystanders may, of course, experience much the same sense of injustice as the actualvictims of perceived injustice.The expectations and sentiment about justice of all individuals collectivelyform a pattern of often conflicting justice-based claims to a society?s resources. At thelevel of doing justice, there usually exist a range of individual, social and political84alternatives about how to respond to the pattern of justice-based claims in such away as to achieve a balance among these conflicting claims in a particular domain.No matter which alternative is chosen, some individuals’ justice-based expectationsare almost always bound to be disappointed. Often the expectations which getdisappointed are every bit as legitimate as the expectations which get met. One wayto put this point would be to say that some injustices will usually be unavoidable inthe process of doing justice. We could, of course, insist that the term “injustice”should not be used for what can plausibly be called ‘injustice” from the perspectiveof an individual, but only for what can plausibly be called “injustice” from someperspective that takes all justice-based claims of all individuals into account. One ofthe major challenges faced by a theory of justice is, in fact, to give usefullydeterminate conditions for diagnosing an injustice, and I shall raise objections bothto making these conditions too permissive and to making them too restricted (Sec3.3.2).§2 As defined above, “doing justice” consists in an appropriate response tothe claims to a share of society’s resources which individuals make in pursuit oftheir personal well-being, by appealing to such criteria as desert, need, equality,rights or good-faith expectations. But any substantive claim about which criteria arejust-making considerations is bound to be controversial.Several distinctions are necessary to clarify the level of analysis in thischapter. The first and fairly obvious distinction is between those types ofconsiderations that are held to be sufficient to judging justice in a specific domainand those types of considerations that are held to be relevant to making suchjudgments in general. I want to call a “just-making criterion” a consideration whichis generally relevant to making judgments about justice. In contrast, I want to call a“principle of justice” a statement that spells out sufficient conditions fordetermining justice in some domain.85A note about the relationship between criteria and principles, as I define themhere. A principle may obviously contain more than one criterion. The role of just-making criteria in a principle of justice is either to specify conditions a person mustmeet to justly claim certain goods or to specify conditions which a distributiveoutcome or process must meet to be just. What someone regards as the “demands ofjustice” results from the set of all judgments that this person is willing to makeabout what things are just or unjust on the basis of applying his interpretation of alljust-making criteria to all distributive domains. The task of justice theorizing is toreduce, as far as possible, the degree of rationally permissible intersubjectivevariability in identifying the demands of justice.We also need to distinguish commonly invoked just-making criteria fromideal criteria. Commonly invoked just-making criteria encompass a fairly standardlist: (1) desert; also referred to as ‘merit’ or “equity”, and variously interpreted asability, effort, contribution or achievement; (2) needs; (3) equality, in the sense ofnon-discrimination on the basis of specified traits; (4) rights; and (5) good-faithexpectations. Fairly clear examples of what are commonly regarded as non-justiceconsiderations are such political values as national security or prestige, economicproductivity, or environmental conservation. It will be convenient to call demandsbased on just-making criteria “justice-based claims”. But even if ordinary peopledraw the distinction between justice-based and nonjustice-based claims in aparticular way, it obviously does not follow that a moral theorist cannot propose theinclusion of additional considerations in the list of just-making criteria or theexclusion of commonly invoked considerations from this list. From the fact that allindividuals in society, or members of certain groups, invoke a particular list ofcriteria when they make claims to society’s resources, it does not follow that a theoryof justice will include all and only these criteria in formulating its proposal for whatshould count as an “appropriate response” to the overall pattern of expectations andsentiment about justice.86We can distinguish three connected sources of disagreement about just-making criteria. There is controversy over how to define the exact nature of suchcommonly invoked criteria as desert, need, equality, rights or good-faithexpectations; controversy over exactly which items should be included in the list ofjust-making criteria; and controversy over what to do when different criteria eitherpresuppose unavailable resources, specify conflicting uses for available resources, orconflict in other ways. I want to call the first issue the criteria definition problem,the second the criteria inclusion problem and the third, following Rawis, the(criteria) priority problem (cf. 40-45).Each commonly invoked just-making criterion would warrant a book, andnumerous books have, in fact, been written about each. My own brief analysis inthis chapter aims only to clarify the basic nature of the criteria definition andinclusion problems to set the stage for the vastly more complex analysis ofprinciples of justice in the next chapter. But two major objections to my way ofproceeding need to be addressed right away: the objection that analyzing justice interms of a complex of just-making criteria is circular; and the objection that rightsare not plausibly regarded as one of several criteria or grounds of justice, but forman independent though perhaps overlapping moral category.§3 First, the circularity objection. As mentioned, each just-making criterionraises the question of its own criteria for being justifiably invoked or properlyapplied. Answering this question takes us back to justice. So analyzing justice interms of a list of just-making criteria is circular.My claim is not, however, that we only need to enumerate a list of commonjust-making criteria, and we will know what justice requires. My claim is that theimmensely complex idea of justice has normative constituents which aremeaningfully distinguishable in terms of their separate functions in our moralvocabulary, and that exploring disagreements about them is a helpful way of87organizing our initial inventory of fundamental sources of disagreement aboutjustice. Of course, a definitive normative account of what can legitimately beclaimed on the basis of desert, need, equality, rights or good-faith expectations mustawait an adequate theory of justice. But in the absence of such a theory, and as a wayof working towards it, how else could we get our bearings than from widespreadunderstandings and deeply ingrained sentiments about the meaning of various just-making criteria? These understandings and sentiments, like the concepts that haveevolved to express them, are of course riddled with inconsistency, vagueness, andother complexities. What could a (consequentialist) account of justice be, if notessentially a proposal for sorting out those complexities and inconsistencies in sucha way that the point of morality is preserved? Morality’s point - creating enoughpredictability and uniformity in people’s behavior so as to generate the necessarysense of security and trust for beneficial mutual interaction - can best and in factoniy be preserved, if the claims of individuals in pursuit of their personal wellbeing are taken very seriously. Taking them seriously, that is, achieving a reasonablecompromise among the often conflicting demands of just-making criteria, will be alarge part of the substance of any consequentialist morality.§4 Second, the rights objection. Is it really plausible to consider rights to bemerely one just-making criterion among several, or are rights perhaps bestconsidered to form a separate moral category? What exactly is the relation betweenjustice and moral rights? Which category is wider in scope - are injustices onepossible form of rights-violations, or are rights-violations one possible form ofinjustice? Do these categories perhaps have overlapping scopes? Or yet anotherpossibility: are these categories, though conceptually and functionally distinct,perhaps extensionally equivalent?In our attempt to sort out conceptual (and associated normative) confusions,we must be very clear about what our analysis is supposed to accomplish. Obviously,88the task of conceptual analysis may be approached very differently by those whothink that distinctions, such as that between justice and the rest of morality, orbetween justice and rights, are somehow carved into the nature of things or come tous from some Platonic realm of essences, and those who think that the ordinaryusage of moral concepts and distinctions reflects a confused mix of irreduciblyconventional, prudential, and historically or otherwise contingent elements. If onesubscribes to the latter view, as I do, the question then arises whether one shouldapproach conceptual analysis with reporting” or Hreforming intentions. Is theanalysis meant to be faithful to ordinary usage and trace out its inconsistencies, or isit meant to stay only as faithful to this usage as is compatible with the goal ofintroducing greater coherence into our moral vocabulary? My objective here isagain the latter: to differentiate between moral concepts with a consequentialistrationale in mind, by trying to assign to each that rationale that seems to me tocohere best with ordinary usage. Specifically, I want to assign rights-talk a place thatmakes sense in a coherent consequentialist conceptual scheme. If we take thisapproach, it seems certainly implausible to subsume all justice-based claims underthe category of rights. As Sumner remarks, ‘Though rights will surely be part of thestory in any adequate theory of justice they are unlikely to be the whole story”(1987a,136).The concept of justice and that of rights can be functionally differentiated interms of their consequentialist rationale. Rights-claims (from a consequentialistperspective) are fundamentally based in the recognition that exempting certain typesof choices or interests of individuals or groups from the well-being test on everyoccasion will promote the greatest well-being in the long run.3 Not all justice-basedclaims are, however, reducible to demands for the protection of types of interests orAlthough I shall develop my account of rights in terms of individual rights, there is nothing in thefundamental consequentialist rationale for rights that would seem to preclude the possibility ofcollective rights. Which collective rights particular social subgroups should have, or whether weshould accept irreducibly collective rights at all, will of course again be subject to the consequentialistwell-being test.89choices which, if recognized and socially protected as rights, would promote thegreatest well-being overall. It is an open question if any particular claim based ondesert, need, equality or good-faith expectations should be accorded the normativelypotent (though not absolutely inviolable) status of a right. It is also an open questionwhether a particular right has anything to do with claims based on desert, need,equality or good-faith expectations (see below).Campbell comments helpfully on the relation between moral rights andsocial justice:There are in life many minor injustices whose rectification may be verymuch an optional extra(20)....it can be argued that rights have a narrowerscope than justice. Rights seem most at home in limited areas whereindividual interests are protected by definitive rules....This may be importantfor the rectification of injustices done to individuals, but it does not seem tohave the same foothold where matters of wider collective or ‘social’ justiceare concerned. Unjustified economic inequalities, the absence of educationalopportunity and discrimination in employment are all grave social injustices,but they are not objectives whose attainment can always readily be effectivelypursued by means of legally enforceable entitlements.... (37-9)Although Campbell speaks of “legally enforceable entitlements” here, I think heclearly means to refer to moral rights protected by legal means rather than to purelylegal rights in the conventionalist sense. Moral rights call for whatever socialprotection promises to be most effective (usually legal enforcement).Of course, rights-claims will pick out many of the same types of actions forprotection as claims based on desert, need, equality or good-faith expectations,although the degree of protection may be quite different. A consequentialistunderstanding of moral rights arises, we said, from the recognition that protectingcertain types of actions from being subject to the well-being test on each occasionwill, in the long run, most effectively promote well-being. A consequentialistunderstanding of social justice arises from the recognition that promoting wellbeing requires responsiveness to those claims to social resources which individualsmake in pursuit of their personal well-being on the basis of criteria which they90allow others to use as a basis for similar claims and which they are capable ofbacking up with some of the fiercest sentiments known. Some of the justice-basedclaims people make will clearly be such that giving them rights-status will promotewell-being in the long run. But clearly, some claim might satisfy the rationale forjustice but not for rights.Could the reverse also be true- so that the scope of rights-claims wouldoverlap the scope of justice-based claims? In this case, some but not all rights-claimscould be considered just-making criteria. I think the answer is no: consequentialistscan safely subsume all rights-claims under the scope of justice-based claims. Rights-violations can generally be regarded as one possible form of injustice.On the other hand, I think the temptation of regarding all injustices asnecessarily rights-violations should be resisted. Given the function of rights in ourmoral vocabulary, a rights-based claim is meant to be normatively more potent, andcapable of overriding, other moral considerations. Claims based on desert, need,equality and good-faith expectations will become rights if, and only if, giving themrights-status is likely to promote the greatest well-being overall. Certain needs-basedclaims may qualify for consequentialist rights-status in any society that can afford tomeet these needs, but most justice-based claims will not qualify for rights-status,because that would give rise to far too many conflicting rights. This is not to deny, ofcourse, that practically all justice-based claims and, in fact, all sorts of completelyfrivolous claims are popularly advanced in the normatively debased language ofrights (cf. Sumner, 1987a, 2-4). But as consequentialists we have a fairly stringentcriterion for generating rights which will disqualify many popular rights-claims.It is plausible to think that all rights-violations constitute injustices, butclearly less plausible to construe every failure to satisfy other justice-based claims asan injustice. In spite of widespread readiness to pronounce all sorts of things unjust,the problem of finding plausible necessary and sufficient conditions for whatconstitutes an injustice turns out to be extremely complex (cf. Sec 3.3.2). If91consequentialists do not want to deprive the concept of “injusticet’of its normativeforce by legitimizing its trivialization, then they cannot permit pronouncing failureto satisfy every claim based on a just-making criterion unjust, but only failure tosatisfy those which survive an appropriate process of balancing the competingjustice-based claims within some appropriately demarcated domain. But thefunction of according rights-status to certain justice-based claims is precisely toshort-circuit the unpredictable vagaries of this balancing process to a large extent,and doing so may indeed be defensible on consequentialist grounds.So by assigning rights a place among my proposed ordering scheme for theunruly family of commonly invoked just-making criteria I do not mean to denythat my list (or any such list) contains important asymmetries, considerable mutualoverlap and a sort of normative pecking order. Claims grounded in basic needs, forexample, will generally have far greater moral weight than mere good-faithexpectations; and different needs and expectations will in turn have varying degreesof moral weight. But while I am prepared to admit that the normative weight ofrights-claims usually allows them to override other justice-based claims, and thatrights-violations should usually count as injustices, as a consequentialist I cannot, ofcourse, go so far as to accord rights-claims infinite moral weight. A person’s havinga right may become compatible, morally, with her being prevented from exercisingit, if the consequences of exercising this right would be sufficiently disastrous.4It is plausible, we said, that all rights-violations should count as injustices(and since rights can conflict, that obviously means injustices can sometimesbecome morally permitted). But should consequentialists affirm the converse aswell: should all injustices count as rights-violations? Should we perhaps regardNarveson seems to hold a wildly implausible absolutism about rights when he says, “If my right todrive a car is compatible morally with your forcibly preventing me from exercising it, then there is nopoint whatever in speaking of ‘my right”(1991, 331).92rights as the cut-off criterion to decide when disregard for some justice-based claimbecomes an injustice?5It is, of course possible, to use the term “injustice” in a sense such that afailure to satisfy a justice-based claim is called an injustice only if it constitutes arights-violation. I want to reject this usage for three reasons: (1) Since there may besituations in which none of the competing justice-based claims have thenormatively potent status of moral rights, this usage would mean that, in such asituation, it would be conceptually impossible for an injustice to be committed. (2)The cumulative weight of such justice-based claims of many individuals may, onoccasion, even outweigh some individual’s recognized moral right at some point,although rights usually will trump other justice-based claims. So it seemsimplausible to regard a violation of someone’s rights as a necessarily sufficientcondition for the occurrence of an injustice, as a “cut-off criterion” which wouldsomehow specify when failure to satisfy a claim based on another just-makingcriterion becomes an injustice. (3) We could, of course, postulate a particular sort ofright, the right never to be unjustly treated, and indeed many people want to claimsuch a right - but I have three consequentialist objections against recognizing such aright. First, postulating such a right would further debase the normative currency ofrights, for two reasons: (i) it would make the actual existence of this right asindeterminate as the diagnosis of an injustice (and there are good consequentialistreasons for wanting to assign the normatively potent status of rights only to fairlydeterminate types of claims); and (ii) rights-violations would become as numerousas injustices (and there is a fairly clear consequentialist rationale for using the term“injustice” in a sense such that many injustices are morally defensible, eitherbecause they are unavoidable or because they are on balance morally preferable (seebelow). Second, postulating a general right of everyone not to be unjustly treatedI owe this question to Michael McDonald, along with numerous other questions which greatly helpedme clarify my position.93would be useless as a cut-off criterion, because invoking it would already presupposethat we know how best to balance the competing justice-based claims in somedomain. Our consequentialist rationale for generating rights would be far tooindeterminate to be up to the job of making rights a useful cut-off criterion for whento consider a failure to satisfy a justice-based claim an injustice. Certainly thereseems nothing conceptually necessary about rights taking on this job. Third, makingevery injustice at the same time a rights-violation- a violation of the alleged “rightnot to be unjustly treated’- (after we already granted that every rights-violation canplausibly be regarded as an injustice) will further contribute to the conflation ofthese two functionally distinct concepts and to the resulting conceptual andnormative confusions. It is an understandable attempt at securing double protectionagainst being victimized by injustice; but raising every injustice to the status of arights-violation is simply bound to debase the normative currency of rights-claims.It is true that this analysis creates a situation in which someone can be done aninjustice, and yet I do not want to say that there is a failure to respect a right. But thatis precisely what, from a consequentialist perspective, seems to me desirable, giventhe relatively subordinate moral weight which I think we must assign to many ofthe things individuals want to call, and should legitimately be able to call,“injustices”.From a consequentialist point of view, I see the division of conceptual laborbetween morality, justice and rights somewhat analogous to the division ofacademic labor between a university as a whole, the faculty of humanities and thedepartment of philosophy, but with lots of cross-appointments of faculty membersin different departments within the humanities. I see nothing inconsistent aboutsaying that, regrettably, someone’s rights have to be violated to avoid a majorinjustice. It might seem inconsistent for me to say, however, that even a minorinjustice has to be committed in order to respect someone’s rights because, on myaccount, rights are part of the input into the process of balancing competing justice-94based claims, and if other justice-based claims lose out over a particular right in aparticular case where the appropriate process has been followed, then no injusticewould seem to have been committed. But this would be a hasty conclusion. Thereare reasons for wanting to use the term “injustice” in a sense in which it islegitimate for us to admit that we may have had to commit an injustice againstcertain individuals, even if overall we have achieved the morally best resultpossible under given circumstances (see below).In summary, then, rights seem to have all the requisite properties of a just-making criterion, but clearly not every claim based on a just-making criterion givesrise to rights. And considering rights to be a criterion among others on a list ofoverlapping, complex just-making criteria is quite compatible with admitting thespecial normative force of rights.3.1.2 The criteria definition problemNext we must wrestle with the criteria definition problem. It is tempting topursue in detail the intricacies and interconnections of such complex concepts as“desert, “needs”, “equality”, “rights” and “good-faith expectations”, but doing sowould make us lose the main thread of our inquiry. I shall, therefore, restrict myselfto a few clarifying remarks about each criterion from a consequentialist perspectiveand mention some plausible points of rational disagreement.§1 DesertClaims based on desert are central to real-life disputes about justice. What aredesert-claims based on? The bases of desert, at least as understood in Westernsocieties, are well analyzed by Sher, who differentiates six classes of desert-claimsaccording to their justificatory basis (1987, 150-1). The first two classes include desertclaims based on traits of a person. They can be non-moral traits. It makes perfectsense to say, for example, that “The best-qualified applicant deserves the job”. Or95they can be moral traits, as when we say, “The virtuous deserve to be happy.” Sher’sother four classes include desert-claims based on features of a person ‘s actions.People’s sense of desert is affected by such factors as whether the action involveseffort, involves competition, is freely willed, or is morally blameworthy. Again, itmakes perfect sense to say that “People deserve the fruits of their efforts”, “Peopledeserve prizes won in competitive performances”, “People deserve the foreseeableconsequences of their freely-willed actions”, or “People deserve punishment formorally blameworthy actions” (cf. Sher, 1987, 150).A consequentialist understanding of desert arises from the recognition thatrecognizing individuals’ claims to social resources on the basis of conscientiousefforts which serve a socially beneficial purpose will effectively encourage suchefforts. For a consequentialist, then, there is no such thing as “desert” independentlyof a conception of what sorts of results count as socially beneficial. In merit-basedhiring of job applicants, for example, what counts as merit is not fixed once and forall, but will change with our changing understanding of the social purpose that wewant a particular profession or institution to serve. There will often be heateddisputes, both because there can obviously be disagreement about the proper socialpurpose of an institution or profession, and also because job applicants who haveworked hard to measure up to established criteria of merit have good-faithexpectations that these criteria will not suddenly be changed.One fundamental dispute about desert that seems to go on withoutresolution, at least in the philosophical literature, is disagreement about the verycoherence of desert-claims in a world of causes. Sidgewick brought this issue intosharp relief. Asking whether rewards for a person’s services should be apportionedto the effort made or to its utility for others, he argues that• . .the actual utility of any service must depend much upon favourablecircumstances and fortunate accidents, not due to any desert of the agent: oragain, may be due to powers and skills which were connate or have beendeveloped by favourable conditions of life, or by good education, and why- should we reward him for these2 And certainly it is only in so far as moral96excellences are exhibited in human achievements that they are commonlythought to be such as God will reward. But by drawing this line we do not yetget rid of the difficulty. For it may still be said that good actions are dueentirely, or to a great extent, to good dispositions and habits, and that these arepartly inherited and partly due to the care of parents and teachers; so that inrewarding these we are rewarding the results of natural and accidentaladvantages, and it is unreasonable to distinguish these from others, such asskill and knowledge, and to say that it is even ideally just to reward the oneand not the other. Shall we say, then, that the reward should be proportionateto the amount of voluntary effort for a good end? But Determinists will saythat even this is ultimately the effect of causes extraneous to the man’s self.(283-4)Exactly this line of argument is invoked by Rawis to exclude desert altogether as arelevant criterion for determining the justice of society’s basic structure. Rawlsdefends his skepticism about desert in trenchant words:It seems to be one of the fixed points of our considered judgments that no onedeserves his place in the distribution of native endowments, any more thanone deserves one’s initial starting place in society. The assertion that a mandeserves the superior character that enables him to make the effort tocultivate his abilities is equally problematic; for his character depends in largepart upon fortunate family and social circumstances for which he can claimno credit. (104)6In more explicit form, Sidgewick’s and Rawls’ argument against including desertamong just-making criteria could be summarized as follows:1. A person’s achievements are the result of ability, effort or circumstances.2. A person’s circumstances are undeserved.3. A person’s ability and effort are wholly the result of his endowments ormotivations.4. A person’s endowment and motivations are wholly the result of heredityand circumstances.5. A person’s heredity and circumstances are undeserved.6. If a person’s achievements are wholly the result of undeserved factors, thenthose achievements are not deserved.Therefore, a person’s achievements are not deserved.6 But Rawis sometimes vacillates in his deterministic assumptions. In “Social Unity and PrimaryGoods”, for example, Rawis says: “As moral persons citizens have some part in forming and cultivatingtheir final ends and preferences...But to argue this [that it is unreasonable to hold people responsiblefor unduly expensive tastes] seems to presuppose that citizens’ preferences are beyond their control aspropensities or cravings which simply happen. Citizens seem to be regarded as passive carriers ofdesires. The use of primary goods. ..relies on a capacity to assume responsibility for our ends.”(168-9) Ido not see how these statements could be made consistent with Rawis’s radical skepticism about desert.97The fourth premise of this argument can be made extremely plausible with the helpof empirical support from biology, psychology or sociology. It represents, of course,the familiar determinist thesis that a person’s actions are merely links in anunbroken causal chain which started long before that person was born. From thethesis that no one can ever help doing what he actually does, it is tempting toconclude that no one ultimately deserves anything. But once the conclusion thatultimately no one deserves anything is combined with another alleged demand ofjustice which many find intuitively plausible, namely that people should becompensated for undeserved inequalities, a strongly persuasive chain of reasoningresults:1. Justice requires that people should be compensated for undeservedinequalities.2. All inequalities are ultimately undeserved.3. So justice requires that all inequalities should be compensated.4. But justice also requires that the position of the worst-off group in societyshould be improved.5. Some economic inequalities may be necessary to improve the position ofthe worst-off group in society.Therefore, justice requires that all inequalities should be compensated, exceptwhen economic inequalities are necessary to improve the position of the worst-offgroup in society.These arguments (together with arguments about the “priority of liberty”) lieat the heart of Rawis’ theory, and many deep rifts in contemporary justicetheorizing can be traced back to disagreement about some premise or other of thesearguments. Let me only give the briefest synopsis here. Nozick, for example,forcefully challenges Rawls’ assumption that whatever results from somethingundeserved is itself undeserved. Nozick argues that unequal wealth may well bedeserved, in the sense that no one is entitled to take it away from a person, if itresults from something which this person simply has (such as natural endowments)98and which no one else is entitled to have (cf. 224-7)) Dworkin disagrees with Rawlsthat what people achieve as a result of effort and ambition is just as undeserved aswhat they achieve as a result of endowments, and therefore proposes socialdistributions which are, as far as possible, “endowment-insensitive but ambitionsensitive”(1981, 311). Most of Rawls’ critics have argued that improving the positionof the worst-off group in society is plausible neither as a necessary nor as a sufficientcondition for identifying morally acceptable economic inequalities (see Ch 3.2).Some have denied that it is reasonable to prohibit tradeoffs between political libertyand economic gains when the loss of liberty would be small and the economic gainslarge (e.g. Barry, 1973.). Some have regarded the idea that people should be free toexercise their talents in the market as a much more plausible basic intuition aboutjustice than that people should be compensated for undeserved inequalities (e.g.Nozick, Narveson), but many possible intermediate positions have also beendefended.There is near universal consensus in the literature on social justice thatreason neither requires nor perhaps even permits us to accept Rawls’ principles ofjustice. My aim here is not to trace the details of this protracted debate, which hasbeen ably chronicled in a number of works, but only to locate a basic disagreementabout desert.8We seem left with essentially three alternative interpretations of “desert”about which rational people may plausibly disagree (even though perhaps withvarying degrees of rationality): (1) No one ever really deserves anything because thecapacities underlying a desert claim are themselves wholly undeserved. (2) Peopledeserve things in proportion to their justified desert claims, even though theseIt is sometimes claimed that adopting an original position will necessarily blunt claims of desert. Butthe contrary seems to me true. Bargainers in the original position who don’t know their individualendowments, but know enough about human nature to appreciate how demoralizing and resentmentprovoking neglect of desert-claims is, will make sure to include a provision to reward desert in theirprinciples.8 E.g. Kymlicka (1990); Mapel (1989); Campbell (1988); Pettit (1980).99claims are based on wholly undeserved capacities. A person deserves rewards on thebasis of ability and effort, not because these are the result of something which isitself deserved, but because they are manifestations of who a person is and of valueswe want to socially uphold. A person’s endowments and motivations are withinthe inner boundaries of his self, as well as expressing socially valuable qualities. (3)People deserve things in proportion to the extent that their effort and achievementsresult from the exercise of free will, properly construed. A person’s abilities, effort ormotivations are more than simply products of heredity and circumstances; theyescape complete causal determination in the relevant “desert-undermining” sense.The disagreement between those who hold (1) or (2) and those who hold (3)strikes me as remediable only if one of the knottiest philosophical conundrums, thefree will problem, could be satisfactorily resolved. From a consequentialist point ofview we want to say that, whatever the deep truth about human free will may be,recognizing claims of desert, in the sense of effort with socially beneficial results,promotes such effort, and therefore promotes well-being and is rationally justified.§2 NeedNeed has such a central place among commonly invoked just-makingcriteria, and such strong support from people’s sense of justice, that theories ofjustice which exclude need (such as libertarian theories) provoke passionateopposition for that reason.If “need” is to be a credible just-making criterion, however, two senses of thisconcept have to be kept apart. In the relational sense, need is simply a relationbetween means and ends: “Person A needs good B to achieve goal C.” Whatsomeone needs, in this relational sense of “need”, is purely a matter of finding theproper means to satisfy a person’s preferences, no matter how frivolous they maybe. Clearly, the relational sense of “need” is useless as a just-making criterion.100Braybrooke has tried to rehabilitate the concept of need in a sense in which itcan be said that a person A needs goods of type B, not to achieve just any goal, but tofacilitate that person’s basic physical and social functioning. Braybrooke calls suchneeds “course-of-life needs’t.Course-life-needs are needs which, unlike preferences,human beings have in common. In distinguishing such needs from preferences,Braybrooke says,.while it must be acknowledged that in preferences other people may be sodifferent as to be hardly understood, much less sympathized with, needsunite human beings in their conception of themselves.”(1987, 237)Braybrooke lists the following “course-of-life” needs:Needs based on physical functioning: the need to have a life-supportingrelation to the environment; the need for food and water; the need to excrete; theneed for exercise; the need for periodic rest, including sleep; the need for whateverelse may be indispensable to preserving the body intact.Needs based on social functioning: the need for companionship; the need foreducation; the need for social acceptance and recognition; the need for sexualactivity; the need to be free from fright and harassment; the need for recreation. (Cf.1987,36)Braybrooke is also careful to distinguish these “matters of need” (which everyperson has to some extent) from “minimum standards of provision” (which reflectthe actual extent of each need, and which will vary from person to person and forthe same person at different times). Braybrooke allows that his list of core needs canbe cautiously expanded - technology, resources and generosity permitting- toaccommodate other plausible candidates for course-of-life needs, such as the needfor satisfying work or needs arising from the requirements of personalitydevelopment. Whenever “fruitless” disputes occur about such expansions,Braybrooke recommends falling back on core needs. Against the obvious objectionthat satisfying some people’s needs comes at the expense of other people’s liberty,Braybrooke argues that an expansion of this list may itself be an exercise of liberty,provided the expansion happens as a result of social consensus (231ff). Braybrooke’sclarification of all these issues is illuminating, even though there remains room for101rational disagreement about a number of questions. What exactly should beincluded in the list of course-of-life needs; what minimum standards of provisionshould be contemplated; and how exactly can we regulate the expansion of this list?9Nevertheless, Braybrooke has made sense of the way needs can legitimatelyfigure, and do in fact figure prominently, in disputes about justice. Course-of-lifeneeds underlie many rights-claims and many expectations that are strongly backedby our sense of justice. It may well be an individuaPs strongest right and her mostlegitimate expectation that society’s decision-makers will not enact policies whichwould sabotage her basic physical or social functioning, policies which would makeit harder or even impossible for her to meet her course-of-life needs. Often thecausal chains are complex, and deprivations may be unforeseen side effects of well-intentioned policies, but there are many blatant cases all over the world wheregovernments cater to the preferences of the privileged while leaving even the mostbasic subsistence needs of the poor unmet. It is hard to see any reason why thosewhose course-of-life needs go unmet through no commensurate fault of their own,when there are potential resources to meet them, should feel any loyalty or respectfor the laws and institutions of the country whose citizens they are.§3 EqualityEquality is a notoriously vague and contentious concept. I want to begin bydistinguishing two senses: the inclusive sense, in which “equality” simply denotesan ideal of just individual shares, so that any injustice can therefore beconceptualized as a deviation from equality in this sense; and a non-inclusive sense,in which equality is one just-making criterion among others, and not everyinjustice can be conceptualized as a deviation from equality in this sense.Although Braybrooke says, in reply to a critic, that “the book allows for uses of the concept of needsexpanded as it were with unanimous consent”, surely a strict condition like unanimous consent willeffectively squelch any expansion. (Cf. 1988, 520)102Equality in the inclusive sense can be used either as a purely formal criterion(i.e. one that is compatible with any possible distribution whatever if it is suitablydescribed) or as a more or less substantive criterion (i.e. one which rules out somedistributions as unacceptable). The prime example of equality in the inclusive sense,used as a purely formal criterion, is Aristotle’s principle of justice, according towhich justice consists in treating equals equally and unequals unequally inproportion to their relevant differences (Cf. Bk V, Nichomachean Ethics). A goodexample of equality in the inclusive sense used as a substantive criterion isHonderich’s conception of justice:A society should seek to secure, as far as is practicable, lives of equalsatisfaction for all its members. It should do this by in general seeking tosecure, as far as practicable, equality of income and wealth, equality of respect(where that is other than....the mere recognition of the relevance of allpersons), equal political and legal freedoms, the full development of thedifferent potentials of individuals by means of education and in work,equality in housing and environment, equal medical care and provision forold age. (183)This is clearly an ideal of just individual shares, a master principle which defineswhat society owes each individual as a matter of justice. Like all such directlyresponsive, internally complex principles, it faces a multitude of problems. I shallconsider these matters in the next chapter.The sense of “equality” of interest to me here is the non-inclusive sense,equality as a just-making criterion which leaves room for other just-making criteria,although it may overlap them. Equality in this sense is best regarded as synonymouswith non-discrimination on some specified basis, such as race, sex, physicaldisability, sexual orientation, age or other features which are considered irrelevantto the just distribution of some good at issue. “Equality” is used in the non-inclusivesense if a certain good is specified along with criteria on the basis of whichdiscrimination in the distribution of this good is forbidden.103Types of equality which have figured prominently in social justice theorizingare: equality of freedom; equality of opportunity and equality of outcome; equality-in-meeting-needs; equality of life chances; equality of resources and equality ofwelfare (or well-being or satisfaction); and equality of consideration, respect,concern, attention or treatment. Some uses of “equality” clearly involve theinclusive sense and some the non-inclusive sense. Equality of life-chances, equalityof resources or well-being, or equality of treatment seem synonymous with an idealof just individual shares and therefore involve “inclusive” equality. Equality-claimsto more specific goods, such as equality of access to medical care or education,commonly specify criteria, in the form of features of persons, on the basis of whichdiscrimination in the distribution of such freedoms or opportunities is forbidden,and therefore commonly involve ‘non-inclusive” equality.There are a variety of plausible general ways of characterizing what it meansto treat people unequally or to discriminate against them. All are fraught withproblems of interpretation, but they nevertheless provide some useful conceptualcommon ground. We could say that for something to constitute inequality ordiscrimination, two conditions must be met: there has to be (unjustified) harminflicted on people; and some people must be harmed more than others inrelevantly similar positions. Another roughly equivalent way of characterizingdiscrimination would be to say that discrimination exists whenever people are beingplaced at a serious disadvantage relative to some appropriate reference group,through no (commensurate) fault of their own, in the distribution of importantsocial benefits. But perhaps the currently most widespread way of defining“discrimination” is to say that discrimination takes place whenever people aredenied equal opportunity in acquiring X - where X stands for a variable domains ofwidely valued goods whose distribution can be socially regulated.But what is equal opportunity in acquiring X? And how, apart from equalityof outcome, can we tell if it has been achieved? I can only touch briefly on these104immensely complex issues here. For the sake of simplicity I want to restrict mydiscussion to one representative domain, desirable jobs. In this domain the sharedcore meaning of “equality of opportunityt’seems to be the idea of distributing thisgood on the basis of performance-related traits rather than on the basis of race, sex,age, political or religious beliefs, sexual orientation, disabilities or such traits, unlessthey are demonstrably performance-related. “Performance-related” here must mean“furthering the social purpose of the job in question”. Beyond this widely sharedmeaning of equal opportunity there are, however, three sharply differentinterpretations of this concept.(1) Equal opportunity in the pure performance sense could be held to exist ifindividuals are selected purely on the basis of performance-related characteristics,no matter whether they acquired them under highly privileged or strongly adverseconditions. (2) Equal opportunity in a performance/motivation sense could be heldto exist if people are selected purely on the basis of performance-relatedcharacteristics and if all people with sufficient talent and motivation are assured ofat least minimally favorable conditions for acquiring performance-relatedcharacteristics. (3) Equal opportunity in a performance/motivation/endowmentsense could be held to exist if people are selected purely on the basis of performance-related characteristics, and if people with sufficient natural endowment andmotivation are assured of at least minimally favorable conditions for acquiringperformance-related characteristics, and if people with similar natural endowmentsare assured of (roughly) similar motivating conditions to make them want toacquire those characteristics.Cauthen describes this third sense of equal opportunity in these terms:.every individual with similar gifts who makes the same effort should haveequivalent life chances for success. Hence efforts must be made to overcomethe hindrances of family and social background that inevitably cripple somein comparison with others. (77)105Rational disagreement is probably possible about how much effort on the part ofsociety is required to equalize the life chances of people whose motivation todevelop their gifts may have been crippled by their family and social backgrounds.Some defenders of equal opportunity in the performance/motivation/endowmentsense argue that justice (and other considerations) require us to relax the standardperformance-related qualifying criteria for members of groups in whose casediscrimination may have denied opportunities or undermined motivation toacquire equal qualifications. Others argue that no relaxation of standards is involvedif we can plausibly reinterpret the social purpose of the job, or our social goals ingeneral, in such a way that membership in a particular racial, sexual or other groupbecomes itself a performance-related characteristic.Proponents of this view often propose to measure equality of opportunity bywhether members of different racial, sexual and other groups are represented ineach category of desirable jobs in proportion to their percentage of the totalpopulation. Given what we know about motivational and genetic factors, there is apoorly defined point at which such attempts at strict proportionality interfere withpeople’s authentic preferences as well as the autonomy of the family (cf. Fishkin,1983). There is scope for rational disagreement about what equality of opportunityrequires, but laying out the consequentialist pros and cons of each contemplatedpolicy clearly can go a long way toward narrowing the scope for such disagreement.§4 RightsWhat do we mean be calling something a “right”? Why shouldconsequentialists recognize rights at all? And which rights should we recognize?(1) What are rights? Rights are meant to protect certain types of choices orinterests of individuals or groups. Sumner’s lucid account makes clear just howcomplex a task it is to define a right in all its dimensions:106The content of a right is what it is a right to do or to have done; this is givenby the content of its core liberty or claim. The scope of a right has twoingredients: the subjects of the right and its objects. The subjects of a right arethose who hold it; this is given by the holders of the various Hohfeldian10advantages (liberties, claims, powers, immunities) which are ingredients ofthe right. The objects of a right are those against whom it is held; this is givenby the bearers of the various Hohfeldian disadvantages (chiefly duties anddisabilities) which are correlated with ingredients of the right. It follows thatthe content and scope of a right have been completely specified only when thecontent and scope of its several ingredients, both core and peripheral, havebeen completely specified. Finally, the strength of a right is its ability tooverride, or susceptibility to being overridden by, competing moralconsiderations. The strength of a right has been completely specified when itsweight has been given relative to every sort of consideration with which itmight compete. (1987a: 124)(2) Why should rights be recognized at all? From the point of view ofpromoting well-being, it is clearly best if certain types of actions are always protectedthan if we engage in complicated case-by-case calculations about whether interferingwith such an action would be more likely to promote well-being in this particularcase. As Campbell explains this consequentialist rationale,As long as individuals have desires (however altruistic) on which they wishto act and as long as such acts are facilitated in a socially beneficial way byallowing the individual a range of legal powers with which to pursue hisinterests, then there is reason to have a rights-based system of rules. (188-9)For a consequentialist, something counts as a right if giving this extraordinaryprotection to the particular type of action so protected constitutes a goodconsequentialist gamble. Rights ultimately protect certain actions (which often, butnot necessarily, constitute vital interests of the individual agent) from routineinterference, even interference for the sake of whatever honestly seems, at themoment, to government officials or other decision-makers to best promote wellbeing. Of course, such protection cannot be absolute; in extreme circumstances and10 Sumner clarifies and elaborates Wesley Hohfeld’s pioneering (1919) analysis of the fundamentalconceptual building blocks of legal rights for purposes of analyzing moral rights. (Cf. Sumner, 1987a:1 8f)107after a special process of deliberation, interference with someone’s rights may bejustified after all to avert a major disaster.(3) What types of actions are justifiably protected as rights? Some candidatesfor the status of rights are the right to life, to security of the person, to bodilyautonomy, to certain forms of property, to the pursuit of happiness, to a ratheropen-ended list of civil and political liberties like freedom of speech and assembly,to a reasonably substantive equality of treatment in the allocation of sundry socialbenefits like jobs or educational opportunities, and a roster of welfare rights that areclosely tied to elementary human needs. The welfare rights a society shouldrecognize are of course relative to what the society can afford. There is room forrational disagreement, I think, both about what kinds of actions deserve protectionas rights and how strong the protection should actually be. These questions are fartoo complex to be explored in the necessary detail here.Combining such general rights with specific circumstances, one can derive aplurality of more specific rights, though rarely in an uncontroversially deductivefashion. Derivations of this kind are fraught with potential for rationaldisagreement.The following reconstruction of the central argument in Shue’s book on BasicRights provides an example of an attempt to derive welfare rights from liberty rightsand, in effect, to challenge the moral status of this distinction:L If a society is not committed to protecting a small number of basic rightswhich are necessary for the effective exercise of any other right, a society does notreally protect any rights at all.2. All societies should protect some rights like everyone’s right to life,freedom or the pursuit of happiness.3. Basic rights like the right to physical security (a right not to be murdered,tortured or assaulted) and the right to subsistence (a right to minimally adequatefood, clothing, shelter, air, water and health care) are necessary for the effectiveexercise of any other right.Therefore, any society should be committed to protecting people’s right tophysical security and subsistence.108Shue’s argument links up with. the disagreement about how best to interpretthe concept of equality of opportunity. Shue’s argument makes a strong case, I think,that the pure performance sense of equality of opportunity must be rejected. It doesnot show, however, that equality in the performance/motivation/endowmentsense can be demanded as a moral right, as is sometimes claimed by those who callfor proportional representation of various groups in all desirable social positions.Desert-claims, needs and more or less urgently felt desires, demands forvarious forms of equality and expectations created in various ways can all too easilybe expressed as, or translated into, rights-claims. Many disagreements about rightswould be remediable if all the disputants shared a consequentialist moralframework and a rational conception of how rights must ultimately be justifiedwithin such a framework. But sometimes there are such intractably complexempirical issues involved that there is every reason to think that rationaldisagreement about the merits of conflicting rights-claims will persist indefinitely.§5 Good-faith expectationsBy “good-faith expectations” I mean existing expectations of individuals, heldin good faith, about what is their due and which, if ignored, arouse a sense ofinjustice. I speak of “good-faith” rather than “legitimate” expectations because Iwant to avoid the implication that claims based on such expectations are necessarilyjustified demands of justice rather than merely one type of justice-based claims. Thiscategory obviously overlaps all the other just-making criteria. Many, but surely notall, expectations which people hold in good faith about what is their due are need-,desert-, equality-, or rights-based. Good-faith expectations may also be based simplyon established practice, law, reasonably hoped-for reciprocity or some otherconsiderations. As I shall use this term, “good-faith expectations” is meant to referto a residual category of considerations that are relevant to justice. Such a residualcategory is needed because sometimes it is not clear which, if any, of the other just-109making criteria has been disregarded, and yet a person may feel a sense of injusticeand may indeed have a plausible case for complaint on grounds of social justice.’1An example might be a sudden raising of university admission standards whichdashes a person’s long-nurtured hopes for admission, or a drastic reduction in socialinsurance allocations. Or take the person who has spent years training to be aresearch scientist, encouraged in his belief that as a scientist he would be able tocontribute greatly to society. Now he is ready to set up his research project. He hashad good reason to believe that public funding would be forthcoming (the researchhas no sufficiently immediate commercial promise to interest private investors),but the government, in an effort to reduce the national debt, suddenly excludes therelevant category o research funding.Many good-faith expectations, as well as many rights-, desert- and equality-based claims, appeal to reciprocity, to a sense that we should justly get something ofvalue in return for giving something of roughly equal value. Other important typesof expectations are based, not on reciprocity, but on established practice. It isexpectations in this sense which, though underpinned by strong emotions, are oftendifficult to integrate with other just-making criteria.Rescher aptly called the dilemma that expectations pose from the point ofview of justice the “Reformer’s Paradox”:the actual, existing distribution.. .is a very important one, from thestandpoint of distributive justice. ..And this existing situation carries within itan existing body of claims, claims which must, in the interests of justice, betaken into account. Justice limits utility at exactly the point of the “Reformer’sParadox”: Given an imperfect existing initial distribution, any redistributionin the interests of arriving, from the standpoint of justice, at a superiordistribution runs headlong into the pattern of existing claims that cannot - inBut are all members of this residual class of just-making considerations necessarily “expectations”?If someone who lacks any expectation of getting x could nevertheless have a just claim to it, thisresidual class would obviously be misnamed. But is this a real possibility? Here, I think, we must againdistinguish consequentialist from more metaphysical conceptions of justice. On a consequentialistaccount, if someone neither deserves x, nor needs x, nor has a right to x, nor expects to get x, nor is treateddiscriminatorily by not getting x, then I see no consequentialist rationale for thinking that it couldpossibly be a concern of justice that this person does not get x.110the interests of the very justice that provides the rationale for the entireenterprise- be brushed aside as an irrelevant obstacle. (121)It is the stranglehold of long-standing expectation, among other things, whichmakes reforms of society’s “basic structure” in the name of justice extremelydifficult. Such reform tends to fail, in large part because an abrupt break with good-faith expectations mobilizes a tremendous sense of injustice in many people.When people have fashioned life plans based on the way things always havebeen, disappointing their expectations tends to arouse their sense of injustice. Anexample of a social practice which is in conflict with most plausible conceptions ofequality, desert, need and moral rights and which, from these perspectives, seemsstrikingly unjust, but is arguably legitimated by good-faith expectations (and a fewother consequentialist considerations), is the practice of inheritance. But there istremendous room for rational disagreement about what makes an expectation alegitimate just-making criterion, and for how much it should count, especially if itis not based on desert, need, equality or rights.3.1.3 The criteria inclusion problem§1 If we survey the range of things which human beings value, we find anumber of values which could plausibly compete with commonly invoked justice-based considerations in constraining social distributions. In other words, we need todistinguish the set of considerations that are held to make some distribution right,from the proper subset of considerations that are held to make it just. What is ofvalue to human beings, and which values can plausibly take on the status of apolitical rather than a mere personal value is, of course, itself contested.But the range of plausible political values competing with justice may include- depending on what is already included in someone’s conception of justice -freedom, happiness, peace and order, national security and prestige, economicproductivity, artistic achievement, scientific progress, a sense of community andcultural integrity, cultural diversity, the preservation of valued traditions, the111conservation of the environment, or the needs of international development. Itcertainly seems possible that competing values could, on occasion, justifiablyoverride considerations of justice. But some philosophers have regarded the valueof justice as “indefeasible”,12How could that be?To answer this question, we must clarify the basic options available forhandling the criteria inclusion problem. One can simply take stock of theconsiderations commonly invoked as relevant to distinguishing just from unjustdistributions, draw up a rough list of standard “just-making criteria” and then busyoneself with clarifying the complexities of each item on this list. But one can alsoexclude commonly invoked considerations from one’s list of just-making criteria orinclude some considerations that are not commonly invoked as relevant todistinguishing just from unjust distributions. Which option is chosen is not merelya semantic matter, but has potentially significant consequences, because by includingwithin the scope of justice-based considerations many considerations commonlyregarded as irrelevant to distinguishing just from unjust distributions (or byexcluding from their scope many considerations commonly regarded as relevant),one in effect proposes to realign the unusually strong sentiments which normallyunderpin justice-based claims with a new set of criteria.Such a proposal may or may not work. By diluting the commonly invokedlist of just-making criteria with considerations normally thought to be completelyirrelevant to giving individuals their due qua individuals, the concept of justicemay simply lose some of its distinctive normative force. But because the cognitiveand emotive aspects of peoples sense of justice are to some degree separable, there isclearly some room for such realignment by rational or non-rational means ofpersuasion. Non-rational persuasion by means of propaganda, “nationalist” appealsto common ethnic identities and various brainwashing techniques can probably12 Cf. J.P. Day, ‘The Indefeasibility of Justice”, in Liberty and Justice (1987). Day mentions Socrates,Kant and Mill as holding a Principle of the Indefeasibility of Justice.112achieve much greater realignments, however, than someone can hope to achieve bymeans of a theory of justice which tries to offer rational justifications. A theory ofjustice cannot stray arbitrarily far from the prevailing pattern of people’sexpectations and sentiments about justice if it is to be acceptable, or evenrecognizable, to them as a theory of justice, and therefore to have sufficient potentialto harness their sense of justice. Still, within hard-to-define limits, reasoned appealscan substantially change some people’s minds and sentiments about justice. There isboth hope and danger in the fact that some theories of justice have had largeinfluence on political leaders who then set the machinery of public indoctrinationin motion.§2 Depending on what one includes in one’s list of just-making criteria, andon how much weight one attaches to these criteria relative to other political values,one can construe the relation between justice and other constraints on socialdistributions in several ways.Before I examine this relation, I must clear up a potential confusion aboutwhat exactly is meant by “constraints on social distributions”. Are there politicalvalues which are not constraints on social distributions? John Rawls seems to thinkso. Since his master principle of justice defines the distributive outcome whichsociety’s basic structure as a whole must achieve to be just, and since he argues thatthe demands of justice are “uncompromising”(4), all constraints on socialdistributions must already be incorporated and somehow harmonized within hismaster principle. But he also says,A complete conception defining principles for all the virtues of the basicstructure, together with their respective weights when they conflict, is morethan a conception of justice; it is a social ideal. The principles of justice are buta part, although perhaps the most important part, of such a conception. (9;emphasis added)Unless Rawls were considered to commit an inconsistency here, the only way I seeto interpret his remarks is that he thinks that not all “virtues of the basic structure”113function as intended constraints on social distributions. But what would be anexample of such a virtue? I cannot think of any political value that might not,under some conceivable circumstances, become an intended constraint on a socialdistribution. It is possible, however, that there are political values which do notordinarily function as such constraints, but are simply part of the cultural givenswhich no one questions. Examples might be cultural diversity or territorialsovereignty. To say that such values might not consciously or directly serve tojustify distributions in a particular society cannot mean that they could not emergeas extremely important constraints on distributions under some conceivablecircumstances. The question, “Are there political values which are not constraintson social distributions?”, is therefore best answered as follows. There are politicalvalues which do not ordinarily function as intended constraints on distributions;but all political values constrain distributions in indirect ways and all could becomeintended constraints under conceivable circumstances.§3 What, then, is the relation between the demands of justice and otherordinarily intended constraints on social distributions? Thinking about this relationin a systematic way, in terms of mutually exclusive and jointly exhaustivecategories, will require us to distinguish four basic types of views about the list ofjust-making criteria and two types of views about the combined normative weightof these criteria relative to other constraints on social distributions. I shall try toassign reasonably descriptive labels to each type and mention their advantages anddrawbacks, and important representatives of each.I want to call the four views about the list of just-making criteria “thestandard view” of justice, “the constricted view”, “the expanded view” and the“mixed view”, depending on whether a view accepts our standard list of commonlyinvoked just-making criteria as constitutive of justice-based claims or modifies thislist in one of three ways: by excluding certain criteria, including additional criteria,114or both excluding and including criteria. Each basic view has two subtypes, which Iwant to call “the indefeasible view” of justice and ‘the defeasible view”, dependingon whether a view denies or allows that the demands of justice must be balancedagainst, and can therefore be defeated by, other political values which constrainsocial distributions. The following table summarizes the distinctions I wish to drawbetween different views about justice:Accepts Excludes certain Includes certain Both excludescommonly commonly additional just- certain commoninvoked just- invoked just- making criteria criteria tLLLmaking criteria making criteria not commonly includes newas constitutive of invoked criteriajusticeDemands of The standard The constricted The expanded The mixedjustice need UQL indefeasible indefeasible indefeasible indefeasiblebe balanced view of justice view of justice view of justice view of justiceagainst (Kant) (libertarians) a) the partly (Rawis)competing expanded viewvalues b) the catch-allview jRescher?)Demands of The standard The constricted The expanded The mixedjustice must be defeasible view defeasible view defeasible view defeasible viewbalanced against of justice of justice of justice of justicecoinpeting values (comrnonsense (Marx?)pluralism; eg.Berlin)Table I: A taxonomy of views of justice, according to(i) whether the “standard list” of just-making criteria is accepted asconstitutive of the concept of justice or modified; and(ii) whether other political values are accorded sufficient weight to requirebalancing with the demands of justiceI. The standard view of justice1. The standard indefeasible view: One can accept some list of commonlyinvoked just-making criteria as constitutive of justice-based claims and deny othervalues sufficient weight ever to override the demands of just-making criteria. Thisposition is well captured by the phrase “Let justice be done even though the earthmay perish”, and it can be clearly ruled out on consequentialist grounds. Althoughsuch a view of justice is a major problem in politics, I know of no contemporary115justice theorist who defends it. Kant, of course, seems to have thought that not onlythe principles of justice, but all moral rules flowing from his Categorical Imperative,are absolutely inviolable. About retributive justice, for example, he says,The law concerning punishment is a categorical imperative, and woe to himwho rummages around in the winding paths of a theory of happiness lookingfor some advantage to be gained by releasing the criminal from punishmentor by reducing the amount of it - in keeping with the Parisai motto: It isbetter that one man should die than that the whole people should perish.” Iflegal justice perishes, then it is no longer worth while for men to remainalive on earth. (The Metaphysical Elements of Justice, 100)Even if a civil society were to dissolve itself by common agreement ofall its members (for example, if the people inhabiting an island decided toseparate and disperse themselves around the world), the last murdererremaining in prison must first be executed, so that everyone will duly receivewhat his actions are worth and so that the bloodguilt thereof will not be fixedon the people because they failed to insist on carrying out the punishment;for it they fail to do so, they may be regarded as accomplices in this publicviolation of legal justice.” (The Metaphysical Elements of Justice, 102)Despite Kant’s many striking pronouncements about the uncompromising natureof justice and other moral imperatives, his derivations of the demands of justicefrom the two general formulations of the Categorical Imperative seem often obscureand controversial, and he does not seem to have come to terms with the crucialproblem of how to handle conflicts between allegedly uncompromising moralimperatives.Minor and even major injustices seem inevitable in the life of any society,and it is hard to see any rationale for accepting with equanimity whateverconsequences may result from prohibiting trade-offs between justice and othervalues, such as national security or great artistic and scientific achievements.Consequentialists will tend to accept many minor injustices and focus onpreventing major ones; but how to draw a principled distinction here between“major” and “minor” is obviously a knotty problem.2. The standard defeasible viezv:This is the commonsense view of politicalpluralists who see no reason why ultimately all virtues of the good society must becapable of principled integration which avoids the need for intuitive, context-116dependent balancing of competing virtues. Viewing justice as an important butdefeasible constraint on social distributions, and including essentially those generaltypes of considerations in the list of just-making criteria which individualscommonly invoke in pursuit of their personal well-being and back up with theirsense of justice would seem to have the advantage of defining the concept in a waythat is relatively consistent with ordinary usage and therefore avoids both thefanaticism of the exclusive view and the many pitfalls of the catch-all view (seebelow). Such a definition would clearly not preclude piecemeal attempts to getpeople to change their perspective about the precise implications of particular just-making criteria in the light of a wide variety of consequentialist considerations. Butit would make the demands of justice defeasible in the final weighing of distributiveconstraints imposed by all political values.Admittedly, ordinary usage is ambiguous on the question as to whether it canever be right to commit an injustice, perhaps because admitting that values of majorimportance can sometimes be mutually exclusive is extremely painful. But eitherwe admit that the properly balanced demands of commonly invoked just-makingcriteria must be further balanced against the demands of other important constraintson social distributions; in which case committing injustices must sometimesbecome justifiable. Or we deny that committing injustices could ever be justifiable;in which case we can never allow trade-offs between the properly balanced demandsof commonly invoked just-making criteria and other important constraints onsocial distributions. We cannot have it both ways, and often people are simplyconfused on this point.We could, of course, stipulate that whenever the properly balanced demandsof commonly invoked just-making criteria must reasonably be overridden by thedemands of other ordinarily relevant constraints on social distributions, we willmake a semantic adjustment in the ordinary meaning of justice and say that aninjustice has not been committed after all. But from a consequentialist point of117view, such a semantic adjustment would seem to be a bad idea. First, such ameaning-change is likely to create conceptual confusion because it divorces theconcept of justice from what individuals claim in pursuit of their personal wellbeing and back up with their sense of justice. Second, the legitimacy of such ameaning-change in any given case would obviously be contingent on someonesjudgment of what constitutes the best trade-off between the properly balanced claimsof just-making criteria and other political values. Such judgments are likely to beextremely controversial. Allowing a semantic adjustment in the ordinary meaningof justice, depending on the judgment of political decision-makers as to whether aparticular trade-off was necessary, would seem to invite attempts to conceal thepainfulness of such trade-offs and to dismiss justice-based grievances by gliblydenying that any injustice has been committed. Allowing the suggested meaning-change would deprive us of the vocabulary to carry on this debate.II. The constricted view of justice:One can exclude from one’s list of just-making criteria some or even allconsiderations which are commonly regarded as relevant to distinguishing justfrom unjust distributions. “From each according to ability; to each according toneed” has been a hugely influential principle of this kind. Many theories of justiceexist which single out need, or desert, or rights, or equality, as the only just-makingcriterion. Many libertarians constrict just-making criteria to a single basic right,namely the right to use one’s own mind and body to make oneself better off as longas one does not thereby make others worse off. The constricted view of justice ismotivated by a desire to make the seemingly chaotic diversity of justice-based claimsmanageable, but the price for severely constricting the list of commonly invokedjust-making criteria is high. It will make one’s view of justice seem lopsided andalmost certainly deprive it of any sufficiently close relation to individuals’ claims tosocial resources which are backed by their sense of justice. It will also have several118other drawbacks of the expanded view of justice, to be discussed below. (Ch 3.5examines whether there could nevertheless be a sufficient rationale to adopt thelibertarian view of justice.)One can, of course, hold either a constricted indefeasible view or a constricteddefeasible view, depending on whether one either denies or allows other politicalvalues sufficient weight to override the demands of just-making criteria on someoccasions. The first alternative will appear to be a peculiarly lopsided form of justicefanaticism, while he second would seem to give up most of the gains achieved bysimplifying the list of just-making criteria when it comes to balancing the demandsof justice with competing constraints on social distributions.III. The expanded view of justice1. The expanded indefeasible viezv: This view includes in the list of just-making criteria some or all constraints on social distributions, many of which arenot commonly regarded as relevant to distinguishing just from unjust distributions,and it regards justice as a value which can never override other ordinarily relevantconstraints on social distributions. Here we must distinguish two varieties, which Iwant to call “the partly expanded view” and “the catch-all view”.(a) The partly expanded indefeasible view: It is possible to include in one’s listof just-making criteria some considerations not commonly invoked as relevant todistinguishing just from unjust distributions without, however, including allconsiderations ordinarily regarded as relevant constraints on social distributions.The indefeasible variant of this view would seem to share the drawbacks of thestandard indefeasible view as well as the drawbacks of the catch-all view to a lesserextent.(b) The catch-all view of justice: This view simply expands the meaning of theterm “justice” to incorporate all values that are ordinarily regarded as relevant toconstraining social distributions. Rescher could be interpreted as making this move119when he includes “social utility” and “supply and demand?? among his eight??canons of distributive justice (cf. Rescher, 73-81). It can be argued that ??socialutility” or ??supply and demand?? involve many paradigmatic non-justiceconsiderations, and regarding these values as constitutive of justice-based claimscollapses a useful distinction that marks an important difference. Although it isultimately a matter of semantics whether one uses ?justice?? as an umbrella termreferring to an uncoordinated plurality of considerations many of which are notbased on the claims individuals make to social resources and which they back upwith their sense of justice, such a promiscuous use of this term would seem to meto have hardly any advantage and several drawbacks. It will lack any sufficientlyclose relation to individuals? claims to social resources which are backed by thesense of justice and will, therefore, necessitate another term for justice in thisnarrow sense. It will empty appeals to ??justice?? of their normative force. And it willalmost certainly give rise to conceptual confusion and equivocation. As Campbellpoints out,If ?justice? is defined as the overall standard of social rightness, then logicallyno other value can stand prior to justice since all relevant values aresubsumed under its umbrella.... It is possible arbitrarily to define justice as theprime social value and then to go on and fill in its content with whatsoever isthought to be morally most important in social distribution, and perhaps alsoin the aggregation of benefits and burdens. But this dogmatic approach hasthe effect of undermining our efforts at conceptual clarification by removingthe constraints imposed by the informal logic of the language of justice inactual political debate, thereby rendering dangerously misleading anysubsequent appeals to our ?intuitions? about what we think is ?just? or?unjust?, for such intuitions are rooted in our operative rather than in ourstipulative normative conceptions. (8)We can give the term ??justice?? any semantic content we like. But what we clearlycannot do is to save ourselves painful trade-offs between political values by simplyconceptualizing all of them as considerations of ??justice?? If someone like Reschersucceeded in broadening the ordinary usage of the term ??justice?I in the way hesuggests, we will still be left with the problem of how to conceptualize the type of120claims which individuals press as their due with the sort of passionate convictioncharacteristically associated with claims of desert, need, equality, rights and good-faith expectations.2. The expanded defeasible view: This view expands the list of just-makingcriteria without, however, going so far as to include all relevant constraints onsocial distributions, and it does not regard “justic&’ as an indefeasible value. Itshares to a lesser extent the drawbacks of the catch-all view. The expanded view ofjustice, whether in its indefeasible or defeasible form, seems to me to lack anyconvincing rationale to outweigh these serious disadvantages. The possibleexception is an expanded view which explicitly includes productive incentives as ajust-making criterion. I shall discuss the consequentialist pros and cons of thisversion of the expanded view of justice below.IV. The mixed view of justiceThis view both excludes from the list of just-making criteria considerationswhich are commonly regarded as relevant to distinguishing just from unjustdistributions, and includes considerations which are not commonly regarded asjust-making criteria.1. The mixed indefeasible view: This mixed view regards the value of justiceas indefeasible, that is, the demands of justice need not be balanced againstcompeting constraints on social distributions. This, of course, is the view held byJohn Rawis (and also the view conveyed by various “ideals of the good society”proposed as theories of justice in the wake of Rawis’ theory). The mixed indefeasibleview, as well as the indefeasible versions of the standard, constricted and expandedviews of justice, all share the characteristic that they could, without inconsistency, beproclaimed in exactly the same ringing words with which Rawis begins his book:Justice is the first virtue of social institutions, as truth is of systems ofthought. A theory however elegant and economical must be rejected orrevised it if is untrue; likewise laws and institutions no matter how efficient121and well-arranged must be reformed or abolished if they are unjust. Eachperson possesses an inviolability founded on justice that even the welfare ofsociety as a whole cannot override. For this reason justice denies that the lossof freedom for some is made right by a greater good shared by others. It doesnot allow that the sacrifices imposed on a few are outweighed by the largersum of advantages enjoyed by many. Therefore in a just society the liberties ofequal citizenship are taken as settled; the rights secured by justice are notsubject to political bargaining or to the calculus of social interests. The onlything that permits us to acquiesce in an erroneous theory is the lack of a betterone; analogously, an injustice is tolerable only when it is necessary to avoidan even greater injustice. Being first virtues of human activities, truth andjustice are uncompromising. (3)Why does this paragraph seem so startling? The reason is that most readers willnaturally assume the standard meaning of justice in terms of commonly invokedjust-making criteria and jump to the conclusion that the writer proposes to defendwhat I have called the standard indefeasible view of justice. Obviously, if someoneintroduced beliefs about justice prior to telling his audience that he has switched toan unorthodox view of the meaning of this term, he might be accused of creatingmisleading expectations. And Rawls has indeed switched to an unorthodox viewwhich assumes not only a radical departure from any ordinary list of just-makingcriteria, but also an even more radical departure from what is ordinarily regarded tobe the immediate subject of justice-based considerations. For him, the “primaryt’(immediate or direct) subject of justice is not the distributive share of an individualor group, but certain fundamental aspects of the entire set of distributive outcomesin society. Rawis can say that, for him, the demands of social justice are inviolableoniy because he identifies these demands with all those considerations which heconsiders relevant to constraining social distributions in ordinary circumstances. So,of course, do proponents of the catch-all view. But the difference between any twoviews of justice could hardly be greater than that between Rawis’ mixed indefeasibleview and Rescher’s catch-all view.122Suppose it were possible to find a set of constraints on social distributionwhich, under ordinary circumstances,’3coordinates the demands of all politicalvalues in a way that can be justified to individuals as giving each what he canreasonably demand in the name of justice and therefore promises to harness mostpeople’s sense of justice in support of it. It would be natural and clearly defensible toidentify this set of constraints as “principles of justice”. This, of course, is Rawis’ useof the term when he thinks of his theory not just as an alternative to the utilitarianaccount of justice (as outlined, for example, in Mill’s Utilitarianism, Ch. V orSidgwick’s Methods of Ethics, Bk III), but, at least in the realm of political decision-making, as an alternative to utilitarianism as a whole.Although both the catch-all view and the mixed indefeasible view include allordinarily relevant constraints on social distribution in their respective lists of just-making criteria, the difference between these two conceptions of justice is now clear.Whereas the catch-all view considers just-making criteria to form anuncoordinated, perhaps incommensurable plurality of criteria whose demandsmust be balanced according to the intuitions of political decision-makers in specificcontexts, the mixed indefeasible view welds some commonly invoked just-makingcriteria together with other constraints into a sort of social algorithm or at leastdraws the outlines of a social blueprint. By a “social blueprint” I mean distributiveconstraints on those social arrangements reasonably regarded as being offundamental importance, constraints whose acceptance is required for any rationalperson, and which avoid any need for unguided intuition at this fundamental level.If we use “justice” in this sense, then, on the face of it, we might seem to needyet another term to describe the criteria which individuals invoke when they claimsomething as their due in pursuit of their personal well-being and back it up withtheir sense of justice. But of course, Rawls’ radical idea was that principles sufficient13 Of course, national emergencies, such as the threat of military invasion, will justify a suspension ofthese constraints.123for constraining the distributive outcome of the entire system of basic socialinstitutions could be found and justified as plausibly giving each individual his orher due qua individual (given that there is tremendous dispute about what thisamounts to in any case). If Rawls’ project had succeeded, all would be well indeed.We would now have a social blueprint which would finally have laid majordisputes about justice to rest. But that has not happened. Instead, Rawls’ proposalfor a social blueprint at the level of society’s basic structure has triggered somethingof an avalanche of alternative visions of the good society, now standardly offered astheories of “justice”. The net result is perhaps more confusion than ever about thedemands, and indeed the core meaning, of justice.It will be convenient to have additional terminology on hand to clearlydistinguish the type of mixed defeasible view of justice which Rawls proposed fromother views of justice that may be superficially similar. Theories of justice differ in afundamental way, according to whether they attempt to be directly or indirectlyresponsive to individuals’ justice-based claims. Directly responsive theories, whichtake justice to be concerned with individual distributive shares, defend a standard ofideally just individual holdings of socially distributable goods, and they interpretjustice as requiring redistributions of such goods from those who, according to thisstandard, have too much to those who have too little. Indirectly responsive theorieslike Rawls’, which take justice to be concerned with society’s basic structure, defenda standard of ideally just distributive outcome patterns or processes for certainimportant social goods, and they interpret justice as requiring a system of socialinstitutions which will enforce these distributive outcomes or processes.2. The mixed defeasible view: This view both includes and excludes criteriafrom the standard list, but requires that the demands of justice be ultimatelybalanced against other relevant constraints on social distributions. Since this viewwould combine the serious drawbacks of both the constricted and expanded views,and it would seem in the end to require intuitive balancing against other constraints124on social distributions, it would need a very strong rationale. I cannot think of anyexamples of this theoretically possible view.§4 Contemporary justice theorizing is a conceptual swamp. Among manyconceptual confusions, perhaps the most fundamental confusion has to do with themeaning of the term “justice” itself. Which of two very different meanings of theterm “justice”, if any, should justice theorists adopt? Should we allow the legitimacyof equating “principles of justice” with the goal of finding a social blueprint forsociety’s basic structure organized as a whole? Or should we equate “principles ofjustice” with the vastly more modest aim of constructing principles which strike aconsequentialist balance among people’s justice-based claims in fairly context-dependent ways? This conceptual decision will, of course, depend on the mostreasonable answer to the question of whether a social blueprint for society’s basicstructure is really possible. Before answering, let us get very clear about what such ablueprint would involve.(1) It would have to single out certain features from the total set of a society’sdistributive outcomes as uniquely relevant to assessing the moral acceptability ofthese outcomes.(2) These features would have to be sufficiently straightforward that itbecomes actually verifiable whether the total set of distributive outcomes prevailingin a given society at any given time has or doesn’t have these features.(3) The blueprint would have to suggest, at least in rough outlines, somefeasible procedure for deciding which existing social institutions should be reformedin what ways so as to give the total set of distributive outcomes those features whichwould make it morally acceptable.(4) One particular proposal for how to single out these features and how toreform society in ways so that the total set of distributive outcomes would actuallyhave these features must be justifiable to all rational people as the best practicable125response to the claims individuals make in pursuit of their well-being quaindividuals, in order to have credible potential to enlist people’s sense of justice insupport of this proposal.If there is realistic hope of justifying principles which meet these conditions -which both provide determinate constraints for the overall distribution of thebenefits and burdens of social co-operation and can somehow be justified as bestgiving individuals what they can reasonably demand in the name of justice - thenthe rationale for adopting the mixed indefeasible view of justice seems compelling.But if there are fairly compelling reasons for thinking that principles at such a highlevel of generality are destined to be normatively empty abstractions, or to involveunmanageable complexities, or to fit poorly with people’s justice-based demands, orto be hopelessly controversial for other reasons, then the rationale for sticking withsomething like the standard defeasible view of justice seems equally clear. (I shallexamine this issue in the next chapter.)§5 But even if the search for a uniquely rational social blueprint turned outto be hopeless, another strongly contested part of the criteria inclusion problem stillposes a major obstacle to accepting any particular basic view of justice: the questionof productive incentives. Should concessions to people’s self-interest, for the sake ofproviding incentives to economic productivity, be considered to fall inside oroutside the list of relevant considerations that help make a distribution just? Thereseem to be strong arguments in support of both alternatives. It is not even entirelyclear whether we should say that a view that includes our standard list plusproductive incentives as an additional criterion is best regarded as an example of theexpanded view of justice, or that the standard view of justice is somewhatindeterminate on the question of productive incentives.The basic case for including productive incentives among just-making criteriais very simple. Suppose we lived in a society of economic equals. And make the126realistic assumption that everyone is fairly poor in this society and that almosteveryone could become considerably better off economically if the talented andambitious were given special economic incentives to exercise their talents andambition in productive ways. If doing justice consists in both satisfying individuals’justice-based claims in an impartial way and to the greatest possible extent, and if theexercise of talent and ambition will generate the resources to satisfy these claims to agreater extent, then giving talented and ambitious people a larger-than-averageshare of society’s wealth as an incentive to exercise their talents and ambitioncannot reasonably be considered unjust. Suppose we adopt a Rawisian scheme andtolerate only those economic inequalities which make the least advantaged fivepercent of society better off. Nevertheless, a certain number of individuals who, asindividuals, might have done better under the old egalitarian order or under somealternative social order suffer hardship under this one, hardship which cannot beplausibly rationalized as ultimately self-inflicted and deserved. Are theseindividuals justified in claiming to be victims of social injustice? The sensibleanswer, for a sophisticated moral theorist, would seem to be no. Clearly we don’twant to call a social order unjust which satisfies the justice-based claims ofindividuals to the greatest possible degree, even though some individuals whomight have done better under a different order will suffer special hardship underthis one. After all, it is pointless to call a social order unjust if it is impossible for anyalternative order to be more just.Yet, upon reflection, the case for excluding productive incentives fromamong just-making criteria seems extraordinarily resilient. When we accept aparticular form of capitalism as, on balance, preferable to its alternatives, are wereally committed to accepting all the possibly resulting inequalities as perfectly just?Wouldn’t that invite rationalizations of paradigmatic injustices? Is it really just thatthe rich should get tax exemptions in the perhaps justified hope that they mightthen have an incentive to invest locally and create jobs, while at the same time the127less well off must bear a heavy tax burden? Should we not say that such inequality isperhaps morally justifiable although unjust, rather than that it is justifiable andtherefore not unjust? It is not at all obvious why it should satisfy the leastadvantaged in the slums of capitalist countries to be told that their destitution in thepresence of extravagant and often unearned wealth is just, simply because somesophisticated moral theorist may be sincerely convinced that ours is the besteconomic system overall and that the percentage of the destitute under anyalternative system would be higher. We would not regard it as respecting theseparateness of persons to tell an individual who claims to be victimized by socialinjustice: “Yes, you suffer undeserved hardship, and under some other system youas an individual would probably be better off, but our system produces the mostwealth on average- so you aren’t really done an injustice.” Should we regard it asfully respecting the separateness of persons if we now tell this same individual thatthe least advantaged five percent of the population would probably be even worseoff under any alternative economic system, and that she should therefore desistfrom claiming to be a victim of injustice? From the point of view of this individual,haven’t we only trivially changed our tune and left her just as badly off as before?Suppose some hypothetical economic system makes the major ingredients toa good life dependent on paid employment and assigns some percentage of thepopulation to a state of involuntary unemployment, low self-esteem, and suicidalunhappiness through no fault of their own, so as to create a reserve pool of laborwhich will exert downward pressure on inflationary wage demands. Now it seemshardly good enough simply to tell such individuals that they should considerthemselves fortunate to live in a just society and assure them that they have noreason for complaint on grounds of justice. They will, in any case, continue to usethe language of justice, or evolve some substitute for it, in an attempt to callattention to their fate and to motivate whatever assistance to them may be feasible.But suppose we have convinced the least advantaged in a Rawlsian type of society128that they cannot possibly have any legitimate grievance on grounds of justice. Whyshould other individuals higher up the socio-economic ladder agree that justice hasbeen done to them, if social arrangements are systematically designed for themaximum benefit of the bottom five percent? Is it unjust to invest social resourcesin space exploration unless, as a result, the condition of slum dwellers promises tobe ameliorated more effectively than through any alternative use of theseresources? Could it be justifiable, on grounds of justice, that the relatively leastadvantaged, who may already be quite well off in absolute terms, should hold theother ninety-five percent back from realizing possibly important life goals inscientific or artistic endeavors of no foreseeable benefit to the least advantaged?§6 These are complex questions, and obviously relevant to settling thequestion we are now addressing: whether we should build productive incentivesinto the heart of the concept of justice, or whether we should sharply distinguishbetween considerations of justice and productive incentives and openly considerbalancing such considerations to be a matter of painful tradeoffs. It is clear that thosewho consider a uniquely rational social blueprint for society’s basic structurepossible must go against the grain of many people’s firmest intuitions and includeproductive incentives among just-making criteria, even at the risk of possiblytrivializing the undeserved destitution of many individuals as simply theinevitable byproduct of doing justice. Therefore, supporters of indirectly responsiveconceptions of justice have no choice.But supporters of directly responsive conceptions of justice, who conceive ofprinciples of justice as spelling out an ideal of individual shares of either societywide or domain-restricted scope, do have a choice: they can hold fast to the standarddefeasible view of justice and exclude productive incentives from the list of justmaking criteria (except insofar as those incentives can be construed as rewardingdesert). Or they can adopt an expanded defeasible view of justice and include129productive incentives into the list of just-making criteria, even where suchincentives lead to windfall profits for people who clearly do not deserve them.Suppose we find reason to adopt a directly responsive conception of justice.Which view of justice is then superior, the expanded view (expanded to includeproductive incentives) or the standard view (which excludes productiveincentives)? Or is there perhaps room for rational disagreement here? I think thisissue can be settled in favor of the standard view. Importing productive incentivesas a just-making criterion into the heart of the concept of justice would complicatethe proper use of this concept to such an extent as to radically reduce its usefulnessfor ordinary individuals who want to give voice to what they regard as their due.But. that productive incentives are best excluded from the list of just-making criteriadoes not mean that they can also be excluded from the list of all morally relevantconsiderations. Committing an injustice for the sake of the morally desirableconsequences of allowing productive incentives can sometimes be morallyjustifiable. (I will argue for a distinction between bona fide principles of justice andmorally justified principles of justice in Ch 3.3).But, it may still be objected, must we not include productive incentives in ourlist of just-making criteria on grounds of logic alone? Productive incentives createresources which allow us to satisfy justice-based claims to a greater extent thanwould otherwise have been possible. If we exclude them from our list of just-making criteria, we are left with the paradox that we may have made society as justas we can humanly make it, and yet we grant some individuals that they may havelegitimate grievances on grounds of justice.A consequentialist response to this argument is as follows. It is not at all aforegone conclusion that we should try to rid the language of justice of this paradox.What purpose is served by the language of justice? Justice, we said, is fundamentallya two-level or dual-purpose concept. Its basic purpose is both to give voice to whatindividuals regard as their due qua individuals and to articulate principled130constraints for rearranging aspects of the social order in greater conformity to whatindividuals regard as their due. Any such concept, whether it includes productiveincentives as a just-making criterion or not, must contain the deep paradox thateven a social order which satisfies justice-based demands to the greatest humanlypossible extent will not be able to silence charges of injustice. A principled responseto individuals’ justice-based claims requires balancing the justice-based claims ofdifferent individuals, and balancing means aggregating, and aggregating implies thatthe claims of some individuals, through no special merit of theirs, will win outover equally legitimate claims of others. Therefore we will in any case have aparadox at the heart of the language of justice: we are forced to acknowledgelegitimate individual grievances even in cases in which we think that social affairsare arranged as justly as we can possibly make them. But allowing a dual-purposeconcept into our vocabulary that contains precisely this paradox has importantconsequentialist pros and cons. On the one hand, it may discourage complacencythat everything already works for the best in the best possible social order and mayhelp motivate the perennial search for better social arrangements. By hypothesis wemay imagine to have reached some Rawlsian optimum, but we can probably neverbe sure what this optimum is or whether we have reached it. On the other hand,allowing the concept of justice to retain this paradox may also promote social strifeand encourage expectations among some groups which simply cannot be met.Whether it would be better, on balance, to include productive incentives along with,or instead of, desert in our list of just-making criteria may be a matter for rationaldisagreement. My point here is that whether moral theorists should follow Rawl&lead and try to purify the concept of justice of its paradoxical implications is not amere logical question which clearheaded conceptual analysis can resolve, but aquestion of what sort of social dynamics we want to encourage for purposes ofmotivating reforms toward greater well-being.131Our brief survey of commonly invoked just-making criteria as well as ofdifferent views about the criteria which are held to be constitutive of the meaning ofjustice makes evident the formidable obstacles faced by any theory of justice whichdoes not simply appeal to some group’s partisan intuitions about what justicerequires, but actually tries to provide a compelling rationale for combiningcommonly invoked just-making criteria (and other considerations) in somenormatively determinate way which any rational person must accept.In view of the varied and imprecise nature of just-making criteria and all thedifferent options of combining just-making criteria into a conception of justice, howcan we ever convincingly argue for or against the justice of a distribution? We mustexamine what level of comprehensiveness we can reasonably demand fromusefully determinate principles of justice, and what sorts of justification can begiven for such principles.1323.2 DISAGREEMENT ABOUT THE COMPREHENSIVENESS OF PRINCIPLESThis chapter explores the level of comprehensiveness which is feasible forprinciples of justice.3.2.1 Basic properties of principles of justice§1 The last chapter dealt with just-making criteria, as I called thoseconsiderations which are considered relevant to making judgments about justice ingeneral. In this chapter I want to examine the role of such criteria in principles ofjustice - by which I mean propositions which specify sufficient conditions for whatjustice requires in a particular domain. What important types of principles of socialjustice can we distinguish and what fundamental problems do they face? AlthoughRawls has done a great deal to give us the analytic vocabulary to deal with thesequestions, I shall have to coin a number of new terms as we go along. But first, letme clarify the important difference between a principle of justice and a mere justice-based claim.We defined a‘tjustice-based claim’T as a claim to a share of social resourcesmade by an individual in pursuit of personal well-being, based on a cluster ofcriteria which that individual would allow others to base analogous claims on,which typically includes such criteria as desert, needs, equality, rights or good-faithexpectations. Principles of justice spell out what justice requires with respect to somedomain by proposing a reasonable compromise between the conflicting justice-basedclaims of different individuals in that domain. The largest possible domain for aprinciple of social justice is of society-wide scope and includes all individuals and allgoods whose distribution is thought to raise considerations of social justice. Morerestricted domains are demarcated by types of goods, institutions and practices, socialroles, cultural subgroups, geographic locations and other individuatingcharacteristics. Since one domain may be included in the scope of another, aprinciple which proposes sufficient conditions for determining justice within a133particular domain may only provide relevant conditions for determining justice insome larger domain. Moreover, since domains may overlap in complex ways, thereis obviously ample room for conflict between domain-restricted principles.This situation seems unsatisfatory. The ultimate goal of justice theorizinghas, therefore, generally been regarded as finding principles of society-wide scopewhich jointly provide sufficient conditions for deciding whether anything whateverthat might be claimed to be socially unjust is unjust. Rawis, for example, says:Thus justice as fairness seeks to identify the kernel of an overlappingconsensus, that is, the shared intuitive ideas which...turn out to be sufficientto underwrite a just constitutional regime. (1985: 246-7; emphasis added)Principles of justice, whether of society-wide or domain-restricted scope, stateabstractly what conditions are sufficient for achieving a proper balance or reasonablecompromise between competing justice-based claims to social resources (cf. Rawls,10). But what makes a balance proper or a compromise reasonable? This chapteraims to illuminate the complexities associated with this question.Principles are often divided into two sets of categories: “formal” versus“substantive”; and “procedural” versus “material”. The distinction between formaland substantive principles refers to the degree of normative guidance a principleprovides, and raises important questions (see §2 below). The distinction betweenprocedural and material principles can be briefly explained as follows. Principles ofjustice are purely procedural if they define constraints on a distribution process andspecify any resulting distributive outcome as just, as long as it results from thisprocess. Principles are purely material if they define a just distributive outcome,without specifying by what process this outcome must or should be achieved. Manyprinciples will, of course, combine procedural and material constraints.But there is a confusion lurking here that needs clearing up. Rawis’ claim isfamiliar that social justice must be conceptualized as a matter of just procedures or“pure procedural justice”, because “there is no independent criterion for the right134result”(86). Rawis claims that trying to achieve justice in a society by defining justindividual shares and then keeping track of whether each individual has more orless than their just share at any given time gets us into hopeless complexities. Butgiven this claim, it seems initially confusing when, on the face of it, Rawl& ownprinciples turn out to be obviously material. Is it perhaps tempting to think that hecalls them ‘procedural” because they are (partly) derived from a contractarian choiceprocedure; but that would add yet another layer of confusion. The solution to theseeming paradox of Rawls’ obviously material principles being called “pureprocedural” is that they specify a distributive outcome at the level of society’s basicstructure and not at the level of individual distributive shares. Hence Rawls’principles are material when it comes to the design of the basic structure, but purelyprocedural when it comes to distributing particular goods to particular individuals.Rawls argues that we cannot find workable criteria for the share of society’sresources which a given individual could justly claim and which a just society couldtherefore be said to owe this individual (cf. e.g. 87-8). But he thinks we may still beable to find workable criteria for designing a scheme of basic social institutionswhich does a reasonably good job overall of responding to concerns about justice,and which has the potential to enlist people’s sense of justice in support of it. SoRawls’ principles deserve to be called “pure procedural”, not because they arederived from a hypothetical contractarian procedure, but because they are meant torespond to the justice-based claims of individuals purely by setting up society’s basicstructure in a certain way, and not by regulating individual distributive sharesdirectly. Individuals can use Rawls’ principles only to argue that the total system ofsocial institutions should be reformed to more nearly achieve this society-widepattern of distributive outcomes. But once this pattern has been achieved, Rawis’principles can no longer be used to justify complaints on grounds of justice, nomatter how destitute a given individual’s position may be.135§2 What sorts of principled responses are possible to the justice-based claimswhich individuals advance to a share of society’s resources? An analysis ofprinciples of justice must focus on two questions: how comprehensive the proposedprinciples are; and how convincingly the principles are justified. Ideally, we wouldlike a theory of justice to offer us maximally comprehensive principles (see below)and to show their acceptance to be necessary for any rational person. We might alsowish them to be universally valid, valid for all societies throughout history, at leastwhere reasonably favorable circumstances obtain (Hume’s famous “circumstancesof justice”).’ But contemporary justice theorists have rarely been so ambitious.Rawis, for example, regards his theory as applying to the basic institutionalarrangements of a “more or less self-sufficient” society and assumes “strictcompliance” with his principles (8-9; 457). In recent writings he has further limitedthe scope of his theory as applying only to reasonably affluent pluralist democraciesand left largely unclear, even in the case of these societies, what parts of theirexisting basic structure, if any, he finds objectionable on grounds of justice and whathe would put in their place (cf. 1985, 223-51; 1988, 252).What would it mean for a principle to be “maximally comprehensive”? By aprinciple’s level of comprehensiveness (or what I shall also call the size of itsnormative payload) I mean how much substantive normative guidance it gives,which is a function of two quite distinct but frequently confused variables: generalityand determinacy. Since the distinction between a principle’s generality anddeterminacy is absolutely crucial, but rarely made explicit, let me make it explicithere. A principle which applies only to a narrow range of cases (low generality) andrules out few alternative courses of action (low determinacy) has a small normativepayload. A principle which applies to a wide range of cases and rules out mostalternatives obviously has a large normative payload. Of two principles which are1 Cf. Hume, An Enquiry Concerning the Principles of Morals, 15-25.136identical with respect to their determinacy, the principle which has greatergenerality has a greater normative payload. Of two principles which are identicalwith respect to their generality, the principle which has greater determinacy has agreater normative payload. The important exceptions are principles which havezero determinacy, e.g: 11Always do what is right and avoid what is wrong.?? Suchprinciples have no normative payload whatever, no matter what their generality.We can define he level of comprehensiveness or the size of the normativepayload of a principle in terms of the number of situations in which the principlecould conceivably be applied in an action-guiding way. (Philosophers of scienceoften define the empirical content of a hypothesis analogously, in terms of thenumber of situations in which the hypothesis could be tested.) It is the failure todistinguish between generality and determinacy which encourages confusions, evenamong philosophers, between principles with high (even universal) generality - e.g.??Treat equals equally??- and principles which actually give us determinatenormative guidance.We can now define our topic more clearly. In asking, ??What level ofcomprehensiveness is achievable for principles of justice?, we are asking what sortof problems arise when we try to make principles both as general as possible and asdeterminate as possible. Maximally comprehensive principles of social justicewould be principles which have both the greatest conceivable generality (their scopeis society-wide) and the greatest conceivable determinacy (they tell us exactly what todo, down to the most specific level of practical decision-making, by always rulingout all but one alternative). What is the relationship between a principle?s level ofcomprehensiveness and the difficulty of justifying it? Are less comprehensiveprinciples generally less difficult to ??justify?? (whatever that comes to) than morecomprehensive principles? Can justified principles of justice achieve a high level ofcomprehensiveness, or do they often merely combine high generality with lowdeterminacy, and therefore deliver a very small normative payload?137As a step toward getting clear about these questions, we must distinguish two types ofprinciples of justice: master principles and what Rawls calls “commonsense precepts”(e.g. 36). Commonsense precepts are imprecise, often highly context-dependentconsiderations which people invoke in everyday life, both when they make justice-based claims and when, as decision-makers, they balance competing justice-basedclaims in an intuitively satisfying way. Such precepts are unsatisfactory, because theyoften conflict with other precepts and leave immense scope for self-serving bias. It willdo little good to confront “the chaos of competing claims” with a chaos of competingprecepts.Master principles of justice are proposals to infuse this chaos of intuitive preceptswith at least some degree of rational order. A master principle consists of an internallyoften complex set of principles that (1) individuate a fixed and usually quite short list ofdistributive goods along with criteria according to which these goods are to bedistributed; (2) are offered as jointly sufficient conditions to determine what is sociallyjust or unjust; (3) have, or are intended to have, the property of internal consistency,that is, they do not mandate mutually exclusive distributions (although they may leavedifficult balancing problems unresolved); and (4) are defended as justifiable in wayswhich any reasonable person (i.e. any person who is interested in finding publiclydefensible principles of social cooperation) must accept. Each of these objectives posesserious difficulties.It will make our discussion more concrete to have on hand several prominentexamples of what I call “master principles”, excerpted from the contemporary justiceliterature:a. Rawis’ master principle of social justice:“(1) Each person has an equal right to a fully adequate scheme of equal basic libertieswhich is compatible with a similar scheme of liberties for all.” [This scheme indudes: “freedom ofthought and liberty of conscience; the political liberties and freedom of association, as well as thefreedoms specified by the liberty and integrity of the person; and finally, the rights and libertiescovered by the rule of law.”](2) Social and economic inequalities are to satisfy two conditions. First, they must be attached tooffices and positions open to all under conditions of fair equality of opportunity; and second, theymust be to the greatest benefit of the least advantaged members of society.”(1987a: 3; cf. 1977: 160-1)138b. Nielsen’s master principle of social justice:“(1) Each person is to have an equal right to the most extensive total system of equal basicliberties and opportunities (including equal opportunities for meaningful work, for self-determination and political and economic participation) compatible with a similar treatment ofall...(2) After provisions are made for common social (community) values, for capitaloverhead to preserve the society’s productive capacity, allowances made for differingunmanipulated needs and preferences, and due weight is given to the just entitlements ofindividuals, the income and wealth (the common stock of means) is to be so divided that eachperson will have a right to an equal share. The necessary burdens requisite to enhance humanwell-being are also to be equally shared, subject, of course, to limitations by differing abilities anddiffering situations.” (289)c. Narveson’s master principle of social justice:..our sole basic duty is to refrain from utilizing the fundamental resources of otherswithout their consent; and those resources include, at a minimum, the bodies and minds of thoseothers. The implications I will suppose we can draw from this are these, at least:(1) We have no fundamental general duty to provide others with such goods as thenecessities of life, let alone some particular proportion of all the socially distributable goods thereare;(2) we have in general a duty not to interfere with the operations the ‘market’, so long asit is truly a free market and not something else masquerading as same;(3) we should always in general prefer voluntary social arrangements to involuntary oneswhenever this is definable and feasible.(4) Governments, in particular, are severely restricted in what they may properly do...”(1988, 166)d. Phillips’ master principle of social justice:“In a just social order, there must be(a) welfare rights to the provision of such goods and services as free and comprehensivemedical care and education, public housing, and guaranteed employment; plus(b) a redistribution of wealth and income to assure that everyone is enabled to obtain anadequate level of these and other necessary goods and services; and(c) the implementation of rectification policies to rectify earlier systematic injustices.After these stipulations have been met,(d) any inequalities that remain in the distribution of income and wealth are permissibleif and only if they do not lead to concentrations of wealth that are demonstrably detrimental topeople’s exercise of their generic rights” (427)e. Sterba’s master principle of social justice:Principle of Need: Each person is guaranteed the primary social goods that are necessaryto meet the normal costs of satisfying his basic needs in the society in which he lives. (Theprinciple of need applies under the assumption that there are sufficient social and economicresources in the society to maintain the guaranteed minimum, but not necessarily withoutrequiring a minimal contribution to society.)Principle of Appropriation and Exchange: Additional primary social goods are to bedistributed on he basis of private appropriation and voluntary agreement and exchange.Principle of Minimal Contribution: A minimal contribution to society is required of thosewho are capable of contributing, when social and economic resources are insufficient to providethe guaranteed minimum to everyone in society without requiring that contribution or when theincentive to contribute to society would otherwise be adversely affected, so that persons wouldnot maximize their contribution to society.Principle of Saving: The rate of saving for each generation should represent its faircontribution toward realizing and maintaining a society in which the members can fully enjoy thebenefit of its just institutions. (61)139f. Nozick’s master principle of social justice:“Ignoring acquisition and rectification, we might say: From each according to what hechooses to do, to each according to what he makes for himself (perhaps with the contracted aid ofothers) and what others choose to do for him and choose to give him of what they’ve been givenpreviously (under this maxim) and haven’t yet expended or transferred.” (160)§3 One caveat before we proceed. My analysis of principles in this chapter mayseem to pile distinctions upon distinctions. I remain convinced, however, that thisanalysis proceeds at the right level of abstraction- the only level that promises toadvance our understanding of social justice beyond the ideologically partisan. Despitethis chapter’s complexities, I am conscious of having unravelled only part of theinherent complexity of the subject matter.I now want to distinguish four types of master principles, using two cross-cuttingdistinctions. The first distinction, already introduced, is between directly responsive andindirectly responsive principles; the second between intuition istic and non-in tuitionisticprinciples. A table will be useful to provide an overview.Principles which spell out apolitical response to individuals’justice-based claims by specifyingwhat is to count as a just processor outcome of the total scheme ofsocial institutions:Principles which spell out apolitical response to individuals’justice-based claims by specifyingwhat is to count as individuals’just distributive share:Principles which do not spell out Directly responsive, Indirectly responsive,a way of balancing conflicting intuitionistic master intuitionistic masterjust-making criteria, but leave principles of justice principles of justicesuch balancing up to largely (e.g. Nielsen, Phillips, most (An empty category, because aunguided intuition: egalitarian theories, most rule principle cannot crediblyconsequentialist theories) claim to specify a just schemeof social institutions whileleaving the balancing ofconflicting criteria tointuition).Principles which (claim to) avoid Directly responsive, non- Indirectly responsive, non-the need for intuitive balancing of intuitionistic master intuitionistic mastercriteria by specifying a single principles of justice principles of justiceultimate maximizing criterion or (An empty category, because (Rawis, Sterba (?), libertarianlexical ordering: no directly responsive theory theories)can credibly claim to be ableto avoid the need for intuitivebalancing of conflictingcriteria)Table II: Types of master principles of social justice140All principles of justice must, by definition, formulate an “appropriate”,“reasonable” or “credible’ response to what individuals claim as their due inpursuit of their personal well-being, based on their sense of justice. Indirectlyresponsive principles respond to the justice-based claims of individuals byspecifying what is to count as a just scheme of social institutions and simply identifyan individual’s just share as whatever outcome such institutions happened toproduce in that individual’s case. Directly responsive principles respond to thejustice-based claims of individuals by specifying what is to count as individuals’ justdistributive share and worry only later, if they worry about it at all, about theinstitutional means by which a society is to secure each individual’s just share atany given time.Indirectly responsive principles, which can either be material or procedural,will have some such general form as the following:“Justice requires that the total scheme of social institutions be designed insuch a zvay that the distribution of good A follows a process or achieves anoutcome that meets criterion x, and the distribution of goods B, C, D, and Efollows a process or achieves an outcome that meets criteria x, y in lexicalorder.”The clearest examples of indirectly responsive procedural master principlesare libertarian principles, which specify the free market as the fundamental socialinstitution which all other institutions are designed to uphold, and which simplyspecify that all goods should be distributed according to the basic right of individualsto use their own minds and bodies to make themselves better off without therebymaking others worse off - which is to say, in practice, according to supply anddemand in the market. The paradigm examples of indirectly responsive materialmaster principles are, of course, Rawis’ master principle and various traditionalforms of utilitarianism. Although it may not be immediately obvious from141comparing Rawl& master principles with other non-libertarian principles, hisprinciple is not only the paradigm example of a non-utilitarian indirectlyresponsive material master principle, but also the only remotely credible candidateof any such principle. Other justice theorists proposing more complex materialprinciples may also claim to fix on the basic structure rather than on individualdistributive shares, but such claims seem to me to involve a seriousmisunderstanding of what it means for a principle to fix on the basic structure ofsociety organized as a whole, and of what extreme simplification would be requiredto get any sort of credible normative determinacy out of such a principle (see §5-8below). Although Rawls’ theory is surely not the only possible indirectly responsiveliberal-egalitarian theory of justice, all other non-libertarian theories yet proposedseem to me most charitably reinterpreted as directly responsive.Traditional forms of utilitarianism are distribution-insensitive, althoughthey claim to be able to account for most of people’s considered judgments aboutjustice by appealing to such ideas as the principle of diminishing returns or thefrequently bad side effects of what are intuitively considered injustices. Traditionalutilitarians only specify one basic just-making criterion: that the total set of goods ina society should be distributed according to whatever will maximize utility- a termthat can be cashed out as average or total happiness or pleasure, preference-satisfaction, or intrinsic value of whatever sort. Some utilitarians postulate equalityas a subsidiary criterion which kicks in whenever several alternative distributionsare equally compatible with maximizing utility. I shall not rehearse the well-knowncriticisms of traditional utilitarianism’s inability to respect the separateness ofpersons.§4 Directly responsive principles, which can be procedural or material, willhave some such general form as the following (which can obviously be furthercomplicated in many ways):142“Justice requires that good A be given to individuals who meet criterion x,goods B, C and D to individuals who meet criterion y, and good £ to individualswho meet criteria z and w.”Given such a principle, it should theoretically be deducible, for any givenindividual who may meet some criteria but not others, what his just share consistsin. How difficult this calculation is would seem to depend primarily on the scope ofa principle’s application, the determinacy of the particular just-making criteriainvoked and the extent to which balancing problems arise in particular contexts ofapplication (see below).As noted in the previous chapter, in the case of directly responsive principles,the term “justice” does not generally encompass all political values that make up avision of the good society or even only those that ordinarily constrain socialdistributions. Rather, the demands of justice will generally have to be balancedagainst the demands of those other values. In the case of indirectly responsiveprinciples, however, there would seem to be no room for such balancing. The wholepoint of taking the total scheme of social institutions as the subject of justice is toavoid the complexities of having to keep track of whether each individualhas more or less than their just share at any given time with respect to some good,while yet doing a sufficiently credible job overall of responding to concerns aboutjustice so that there is a realistic hope of enlisting enough people’s sense of justicein support of this scheme to achieve social stability. But having developed ablueprint for such a scheme, could someone like Rawls argue that we can regularlyadjust this scheme to accommodate competing social objectives? The answer seemsto be clearly no, except in national emergencies or other extraordinarycircumstances.2Someone who takes individual shares as the subject of justice can2 I have already remarked on Rawls’ ambiguity on this point. His introductory paragraph on the “roleof justice”(l) affirms the thesis of the indefeasibility of justice; but a few pages later he says that “Acomplete conception defining principles for all the virtues of the basic structure, together with theirrespective weights when they conflict, is more than a conception of justice; it is a social ideal.”(9,emphasis added)143easily argue that we may regrettably have to commit an injustice against particularindividuals to avoid even greater injustices against others or to achieve moreimportant social goals. But someone who takes the basic structure as subject cannotargue that we can regularly modify the just scheme of basic social institutions toaccommodate competing considerations, both because once established such ascheme is not easily modified, but more importantly because permittingunprincipled ad hoc modifications would seem to undermine the ever-precariousability of such a scheme to do a sufficiently credible job overall of responding to thepattern of justice-based expectations and sentiments so that there is a realistic hopeof enlisting enough people!s sense of justice in support of this scheme. It wouldtherefore seem that, unlike a directly responsive principle, an indirectly responsivemaster principle of justice cannot simply leave aside traditional non-justiceconsiderations, but must somehow amalgamate them all in one coherent principlewhich can feasibly constrain the design of society’s basic institutional arrangements.§5 The second distinction between different types of master principles, whichcross-cuts the first, is the distinction between intuitionistic and non-i ntuitionisticprinciples. Intuitionistic principles contain several different just-making criteria andleave the balancing of at least some of these criteria up to unguided intuition (cf.Rawls, 34-45). Clear examples of strongly intuitionistic principles are Nielsen’s andPhillips master principles, despite some superficial attempts at a priority ordering.Non-intuitionistic principles can specify either one or several criteria.Examples of single-criterion principles, or as Rawls calls them, “single-principleconceptions with one ultimate standard”(41), are necessarily non-intuitionisticbecause they propose to determine the justice of all distributions ultimately on thebasis of a single criterion. They may, of course, specify any number of subsidiarycriteria, but they are ultimately single-criterion principles as long as they specify acommon currency, such as happiness-maximization, in terms of which the relative144importance of each of these sub-criteria may at least in principle be assessed whenthey conflict. Prominent examples are the simplest utilitarian principles whichdefine the morally right (including the just) as promoting the greatest total oraverage happiness or preference satisfaction; and libertarian principles, such asNozick’s or Narveson’s, which define the socially just as whatever is distributed onthe basis of uncoerced individual decisions or free market transactions. But nonintuitionistic principles can also specify several ultimate criteria and avoid conflictbetween them by specifying a “lexical ordering”. At least in principle, a lexicalordering of principles obviates the need for intuitive balancing.3Examples of nonintuitionistic principles which make a serious attempt at lexical ordering are Rawl∧ far less credibly, Sterba’s, Nielsen’s and Phillips’ master principles as well.3.2.2 Sources of rational disagreement about principles§1 Now that we have analytically informative distinctions, we can confrontthe question of what problems different types of master principle face. What type ofmaster principle offers the best hope for providing a rational basis to settle conflictsabout justice? What types of master principles can be confidently eliminated asbeing beyond the scope of the rationally permissible? Or must we perhaps fall backaltogether on common-sense precepts in our arguments about justice?These are urgent questions; and considerably easier to ask than to answer.One way of justifying an answer to these questions, and probably the mostsatisfactory way, would be to examine the whole range of contemporary theories ofjustice in detail, to achieve the widest possible reflective equilibrium of’ relevantconsiderations. No such monumental effort is feasible here. An interesting projectmight be structured around a comparative study of the difficulties associated with avariety of theories. In fact, several critical surveys of contemporary theories of3 See Rawls, 42-3, for an account of lexical ordering.145justice already exist (cf. e.g. Kymlicka, Campbell, Pettit and others). My aim here isnot to simply follow in their footsteps, but to move beyond these critiques towardsome general conclusions about important sources and limits of rationaldisagreement about justice. In an effort to assess the prospects of various master-principle approaches to social justice, I shall first try to define the main problems ofeach type of master principle in the abstract and then assess the way John Rawistried to surmount these problems. I focus on Rawis’ theory, not because its faults areparticularly glaring, but because, on the contrary, it would be hard to surpass manyof its virtues and especially Rawls? awareness of balancing problems. Rawl& workstill has a lot to teach us about the limits of reason in deciding on the rationallyrequired content for master principles of social justice. If Rawls’ master principle,despite the ingenious simplifications it incorporates, proves unworkable, othermaster-principle approaches of society-wide scope which do not incorporate thesesimplifications are also likely to fail. In fact, many master principles proposed in thewake of Rawls’ work seem to me to stumble over precisely the sort of difficultieswhich Rawl& theory was designed to circumvent.Rawis’ motivation for seeking an indirectly responsive, non-intuitionisticmaster principle (which incorporates massive simplifications even at the risk ofputting it fundamentally at odds with many people’s sense of justice) is bestappreciated in the light of the difficulties faced by alternatives. Four major problemsfaced by master principles of social justice are the justification problem, the priorityproblem, the aggregation-distribution problem and various remainingcommensuration problems. Let me clarify the nature of these problems.§2 The justification problem consists in giving a compelling rationale -compelling, that is, for any person with an interest in publicly defensible rules ofsocial cooperation- for accepting a particular master principle over many competingprinciples of justice at various possible levels of comprehensiveness. Different146theories seek to resolve this problem by appealing to a range of mutuallyincompatible meta-ethical positions. Although explicit appeals to God’s will ordesign have gone out of fashion in Western analytic philosophy, neo-Aristotelianappeals to the human good, realist appeals to natural rights or categoricalimperatives, contractarian and rights-based constructivism, unabashedintuitionism, and especially invocations of intuitively appealing forms of equality,as well as a pragmatic liberalism based on a mixture of relativism and meta-ethicaldespair are all very much in evidence. The best one can do in matters of ultimatemoral justification, I have argued, is to seek a wide reflective equilibrium whichtries to work toward a principled understanding of the basic nature of morality byassessing all the prima facie plausible meta-ethical alternatives in the light ofconsidered moral judgments and relevant background theories, including thescientific world view. I have tried to establish some presumption that a widereflective equilibrium supports a form of consequentialism, although much detailremains to be filled in. According to this moral framework, a theory of justice mustbe judged, relative to alternatives, by the extent to which it shows promise to helpaccomplish morality’s main job: to create enough uniformity and thus predictabilityin people’s behavior so as to generate a sense of security and trust to allow mutuallybeneficial interaction. More precisely, the ideal consequentialist goal is to helpachieve that combination of satisfied interests of individuals in their own wellbeing which takes seriously the distinction between persons. We have found reasonto think that this goal can only be achieved if we take very seriously the claimsindividuals make to social resources in pursuit of their personal well-being, andwhich they back up with their sense of justice. That means that nothing can count asa justified master principle of justice if it does not respond to the pattern ofexpectations and sentiment about justice in a way so as to have credible potential toharness enough people’s sense of justice in support of it to make this principle thebasis for a stable social consensus about justice. Another necessary condition for a147rationally acceptable principle of justice is that it must not involve unworkablebalancing problems which would completely vitiate its normative determinacy.How well a principle meets these conditions is a matter of degree, at least up to ahard-to-define point at which a principle may asymptotically approach zerodeterminacy. Of those principles which meet these necessary conditions equallywell, taking the principle with the greatest comprehensiveness would seem to makethe most rational sense, because the point of a theory of justice is, after all, to pullpeople’s conflicting judgments about justice as far up to the level of principles aspossible. What can count as having “justified” a principle of justice is obviously nota subject that admits of very great precision.While our consequentialist moral framework rules out some proposedmaster principles as complete non-starters, it requires us to assess the comparativemerits of many others in the laborious way in which I will try to assess the merits ofRawis’ principle. Despite Rawls’ self-proclaimed anti-utilitarianism, there isnothing inconsistent about assessing the rationale for Rawls’ two principles fromwithin a consequentialist framework. Since I defend indirect consequentialism, it isan open question what sort of principles of justice will most plausibly promote theideal consequentialist goal. Rawls’ two principles are certainly among the primafacie contenders. If I ultimately reject Rawls’ principles, it is not simply because thejustification Rawls gives for them conflicts with a consequentialist framework, butbecause they fail Rawls’ own necessary conditions - which are also plausibleconditions from a consequentialist perspective- for any adequate theory of justice.§3 The priority problem is how to weigh the relative importance of thevarious just-making criteria invoked by an internally complex principle wheneverthese criteria conflict (cf. Rawls, 40-45). Such conflicts can take several possibleforms: two criteria may demand the same scarce resource; or one criterion mayqualify a person for obtaining a certain good and another may disqualify him from148obtaining it. Solving the priority problem exacts a steep price. It means that onemust either accept a single-criterion conception of justice or else specify a lexicalordering. Single-criterion conceptions of justice end up with a criterion which eithergives merely semantic unity to an underlying pluralism of values or else has noplausible potential to enlist many peopl&s sense of justice. Lexical ordering tends tobe vulnerable to obvious counterexamples, because it specifies an absolute order ofpriority for principles and prohibits any trade-offs between a minimally lessersatisfaction of principles of higher priority even for the sake of a hugely improvedsatisfaction of principles of lower priority. Indirectly responsive principles mustsolve the priority problem (and other balancing problems) through radicalsimplifications, because principles which specify constraints for the design ofsociety’s basic structure while leaving open the question of how to balanceconflicting just-making criteria would have no normative determinacy whatever,and de facto amount to having no principle at all (see below).Solving the priority problem is far less urgent for directly responsiveprinciples. I have called those directly responsive principles which solve the priorityproblem (and pay the price for solving it) “non-intuitionistic”, and those whichdon’t solve it “intuitionistic”. It is, of course, a matter of degree how intuitionistic adirectly responsive principle is - that is, to what extent it leaves the balancing ofcriteria up to the moral intuitions of relevant decision-makers. It is important to seewhy it is less crucial for directly responsive principles than for indirectly responsiveprinciples to solve the priority problem and other balancing problems. Allowingsome degree of balancing of conflicting just-making criteria by unguided intuition isless damaging for principles which specify the justice of individual distributiveshares rather than of society’s basic structure for several reasons. The local context inwhich an individual’s share will have to be determined will generally imposeadditional constraints on the intuitions of decision-makers; grievances can be dealtwith locally, often in face-to-face contact between the parties most affected;149procedures can often be more easily agreed upon within such contexts because thepeople involved have a direct personal stake in resolving disputes about justice; andthe injustices committed will be less momentous because confined to more or lessrestricted domains. Material master principles which specify a plurality of criteria fordetermining just individual shares in a plurality of goods will have to besupplemented with purely procedural principles and allow for a great deal ofcontext-dependent interpretation. But in the end, procedural principles(supplemented by a guaranteed social minimum) may often be the only type ofmaster principle we can get in many domains. Nevertheless, leaving theinterpretation of a directly responsive, intuitionistic master principle up tocontextual balancing of conflicting criteria is far from ideal and, as Rawls stressed,may often be little better than having no principle at all (cf. 34-40).§4 The aggregation-distribution problem: All principled ways of respondingto individuals’ competing justice-based claims in a credible way will necessarilyinvolve both distributive and aggregative considerations. The problem thenbecomes how to balance these two types of considerations. The problem (which thethought experiment of imagining oneself living all lives in sequence was meant toresolve in theory) gets more and more intractable in practice as a principle’s domainof application expands over a heterogeneous range of goods and a heterogeneousrange of beneficiaries of those goods. Even if we do not include additional valuesinto our list of just-making criteria except those considerations which arecommonly invoked as relevant to determining justice, we still have two objectives:satisfying individuals’ competing justice-based claims to the greatest possible extent,and satisfying them in a way that respects the separateness of persons. And thoseobjectives may easily conflict. Satisfying the important claims of even the poorestfraction of the population may mean forgoing the satisfaction of a greater amount ofless important claims of the rest of the population. The larger and more150heterogeneous the domain to which a principle applies, the more intractable thisbalancing problem would seem to become, because we have to make a multitude ofqualitative judgments of the relative importance of different claims of differentindividuals. Only by factoring this problem into small parts, that is, by eitherbreaking the problem of social justice up into a plurality of particular domains or(possibly) by using Rawis’ device of judging the justice of the total scheme of socialinstitutions from a representative social position, can we hope to be sufficientlysensitive to the disparate justice-based claims of enough individuals to harness theirsense of justice for social cooperation.Rawls tried to simplify the aggregation-distribution problem in a bold andhighly original way. Instead of giving up on the idea of a master principle of society-wide scope, he introduced one of the most controversial simplifications of histheory, the idea of a representative social position.4In trying to balance satisfactionof the greatest aggregate of justice-based claims against the satisfaction of the claimsof greater importance, he asks us to assume the position of a representative worst-off person and and to arrange the total scheme of social institutions in such a waythat the justice-based claims of this person are satisfied to the highest possible extent(cf. Rawls, 64f). There is obviously a lot to be said for judging the demands of justicefrom the perspective of those whose needs are greatest, given that any attempt tojudge them from everyone’s perspective simultaneously gets us into thecomplexities of the thought experiment of imagining oneself living all lives insequence. Still, for persons whose sense of identity is highly tied up with their socialclass, Rawis’ strategy of maximizing the expectations of one particular socialposition (that of the representative worst-off person) hardly seems to guaranteeRawls, of course, describes two types representative social position: the position of equal citizenship,from which we are to judge whether the liberty principle is satisfied; and the position of the worst offrepresentative person, from which we are to judge whether the difference principle is satisfied (Cf.64f). I shall only concern myself with the social position of the worst off representative person in whatfollows.151an acceptable level of respect for the separateness of persons without assuming anextremely dubious chain connection of benefits rippling upward through thedifferent social strata as a result of maximizing the lot of the worst off.From the perspective of distribution-sensitive consequentialism, in contrast,we can generally pronounce outcomes acceptable which would both produce moretotal satisfaction of justice-based claims more equally distributed than any otherfeasible outcome (assuming a safety net for the very worst-off). We can also say withconfidence that an outcome which increases some plausible index of aggregatesatisfaction at the expense of unequal distribution will eventually become just asintolerable as an outcome which increases the equal distribution of satisfaction atthe expense of every plausible index of aggregate satisfaction. But within broadlimits there seems to be no general way to rank, in the abstract, alternativecombinations of aggregate satisfaction and degrees of equal distribution. Hereperhaps any theory of justice must admit limits to rational decidability in theabstract and defer to more context-dependent reflective equilibria, which is to say,defer to the wisdom of decision-making bodies faced with concrete distributiveproblems.§5 The commensuration problem is the set of balancing problems left afterthe priority and aggregation-distribution problems are resolved. Such additionalbalancing problems arise when we try to compare the value of the different holdingsof all the different individuals in society in such a way that we can tell, at any giventime, what redistributions are required from those whose holdings exceed their justshare to those whose holdings fall short of it. The problem of tracking “the endlessvariety of circumstances and the changing relative positions of particular persons”,as Rawls so clearly realized, makes it seem like a forgone conclusion that anydirectly responsive principle of society-wide scope will be unworkable in practice.Rawis emphasizes the virtue of his approach in the face of the commensuration152problem:5Now the great practical advantage of pure procedural justice [i.e. letting a justbasic structure take care of individual shares, without individuals being ableto claim redress in the name of justice] is that it is no longer necessary inmeeting the demands of justice to keep track of the endless variety ofcircumstances and the changing relative positions of particular persons. Oneavoids the problem of defining principles to cope with the enormouscomplexities which would arise if such details were relevant. It is a mistake tofocus attention on the varying relative positions of individuals and to requirethat every change, considered as a single transaction viewed in isolation, be initself just. It is the arrangement of the basic structure which is to be judged,and judged from a general point of view. Unless we are prepared to criticize itfrom the standpoint of a relevant representative man in some particularposition, we have no complaint against it. Thus the acceptance of the twoprinciples [governing the basic structure] constitutes an understanding todiscard as irrelevant as a matter of social justice much of the information andmany of the complications of everyday life. (87-8)We can divide the commensuration problem into three basic parts, and differentmaster principles are mainly distinguished according to how they cope with eachdifferent part of it.(1) Take a principle of society-wide scope which defines just individualshares. Determining whether someone has a just grievance requires us to determinewhether a particular individual’s share at a given moment is actually just. First, wewould have to find ways to establish the total value of an individual’s holdings ofrelevant distributive goods relative to the value of the holdings of all otherindividuals in society in such a way that we can tell whether he exceeds his justshare or falls short of it. The holdings of different individuals will, of course, becomposed of different types of goods and different amounts of each type, and thesame good will have different value for different persons. If we are to compare thevalue of one person’s holdings relative to another person’s, we need to have aI cannot discover that he gave it any particular name in A Theory of Justice. In “The Priority of Rightand Ideas of the Good”, however, Rawis calls something closely related to it “the problem ofinterpersonal comparisons of citizens’ well-being”: “given the conflicting comprehensive conceptions ofthe good [individuals have], how is it possible to reach a political understanding of what is to count asappropriate claims?”(1988, 255-6).153common measure to be able to add up the value of different goods a person mayhave.6We could, of course, reduce the index problem to the extent that we specify aseparate criterion for different goods; but that has obvious drawbacks. It will escalatethe priority problem, insofar as different goods ultimately depend on the sameresources to produce them. Moreover, it would obviously be unjust not to take intoaccount, for example, that some people may have more money, but other peoplemay choose to compensate for that by having more leisure. The fact is that there isno good index to compare the value of diverse goods, from the point of view ofjustice. The market value of goods is notoriously problematic, partly because themarket value of some important goods for which there now exists no regularmarket is hard to determine, and partly because markets are distorted by thepreferences of the already affluent. We could use the amount of happiness theperson derives from each good to index the total value of each person’s holdingsand then compare his holdings to everyone else’s, to judge whether his distributiveshare was above or below average in value. But interpersonal comparisons ofmental states are also extremely problematic. One popular way of reducing the indexproblem without escalating the priority problem or giving rise to obvious injusticeis, of course, simply to specify standards of minimum provision for certain strategicgoods, regardless of what other goods a person may have, and often regardless ofwhether someone finds himself short of such goods through his own fault.(2) Having indexed the total value of a person’s holdings according to somecommon measure, one would then in principle have to do the same with everyother person’s holdings, because redistributing to those who have less than theirjust share may require taking from those who have more, and to do that we have toknow who is in which relative position at any given time (or perhaps over a wholelifetime). Either case requires keeping track of the endless variety of individuals’6 This part of the commensuration problem was called the “index problem” by Rawis (cf 93f).154circumstances and changing relative positions which Rawis mentions. Althoughcomputer technology has allowed great progress in keeping updated records ofindividuals’ changing circumstances, a centralized information bank on allindividuals of this sort obviously gives rise to other troubling moral issues.(3) Having determined who has too much and who too little, we now have todevise effective, cheap, peaceful and fairly continuous ways of redistribution, toreestablish or maintain a just balance of benefits and burdens for each individual.There are many additional problems which will arise in considering variousoptions for accomplishing redistributions on a massive, continuous scale.The magnitude of the commensuration problem a directly responsive masterprinciple faces will vary, in complex ways, with its scope of application, the numberand concreteness of distributive goods it individuates, as well as with otherparticulars of a principle. As Mapel sums up the difficulties of directly responsivemaster principles of ambitious scope that try to “provide standards for redressing theoverall situation of each member of society”,[They] must find some coherent and practicable way of distinguishing thenatural, social and personal factors that together presumably determine whatindividuals are capable of claiming. Of the many obvious difficulties... .1 willnote only the most salient: (1) the impossibility of distinguishing between theholdings that have been acquired by actual members of society and theholdings that might have been acquired in another society or a ‘state ofnature’; (2) the impossibility of establishing the counterfactual claimsnecessary to justify policies of direct compensation to individuals fordiscrimination or unfairness occurring any time in a more or less remotepast; (3) the necessity of nevertheless rectifying the cumulative effects ofindividual transactions, both free and fraudulent; (4) the impossibility ofregarding market demand as an objective measure of the worth of individualcontributions, together with the impossibility of commensurating the valueof most goods and services at all except through the operation of markets; (5)the conceptual and practical difficulty of assessing welfare, developmental orotherwise; and (6) the requirement that principles of justice neverthelessallow reasonable economic efficiency and political stability. (147)Directly responsive master principles of justice of ambitious scope may enunciatenoble-sounding manifestos of an individual’s just share. But if all the above155difficulties stand in the way of their implementation, they may simply serve to armmany people with yet another mischief-causing rhetorical device for pressingunsatisfiable claims in the name of justice. It clearly took genius not only toconceive the possibility of a radically different approach, but to develop it againstseemingly hopeless odds to the degree of sophistication Rawls did. Our questionnow is whether Rawls’ indirectly responsive master principle can, in fact, escapebalancing problems of unworkable proportions without, at the same time,simplifying beyond the bounds of what is acceptable if his principle is to havecredible potential to enlist rational people’s sense of justice in support of it.3.2.3 Rawis’ master principle§1 From the perspective of conceptual intelligibility alone, the number ofpossible principles of justice is obviously huge, because of the immense diversity ofpossible goods to which some combination of just-making criteria mightconceivably be considered relevant. A master-principle of social justice mustindividuate those types of goods it considers strategic enough to become the concernof social justice, and must align these goods with a particular combination of just-making criteria whose demands should be met.The fundamental normative idea embodied in Rawls’ master principle issimply that inequalities are unjust unless they work for everyone’s advantage:“Injustice, then, is simply inequalities that are not to the benefit of all.”(62) Hedistinguishes between political and economic “primary goods” and argues that,given a certain minimum level of material prosperity, any inequalities in people’s“right to the most extensive total system of equal basic liberties compatible with asimilar system of liberty for all” could never work to everyone’s advantage (250;541f). But when it comes to social and economic inequalities, he thinks that a justsystem of social institutions must allow even large inequalities, because certainincentive structures motivate individuals to exert effort and develop their talents to156create economic wealth which, despite producing relative disparities, will make allindividuals better off in absolute terms.Rawls supports these conclusions essentially with two arguments linked in awide reflective equilibrium: a narrow-reflective-equilibrium argument that claimsthat these conclusions cohere with our considered moral judgments, “duly prunedand adjusted”, about what is just or unjust in particular cases; and a social contractargument which derives these conclusions from an original position whosedescription is in turn supported by considered moral judgments and relevantbackground theories of various sorts. Since Rawis’ arguments are now well knownyet difficult to state concisely, I shall not rehearse them in detail here.7 Thefollowing summary of the essentials of two decades of criticism of Rawls’ theory ismeant to be intelligible only to those already reasonably conversant with theliterature.What are the main ways in which Rawis’ theory of justice is widely thoughtto fall short in justifying his proposed master principle as a principle of justicewhich any rational person must accept? The social contract argument seems tomany to fall short on two main counts. It is argued that Rawls fails to show that hisdescription of the original position is in reflective equilibrium with everyone’sconsidered moral judgments about what constitutes necessary and sufficientconditions for a fair choice situation. Moreover, even if his description of the choicesituation were in reflective equilibrium with everyone’s considered moraljudgments, Rawls fails to show convincingly that in this position rational chooserswould necessarily pick out his favored principles.Rawls’ narrow-reflective equilibrium argument (which tries to work fromour considered judgment in particular cases to general principles) is also widely seento fall seriously short. Rawis’ favored principles cannot be shown any moreSummaries are to be found in Barry (1973, 1989); Kymlicka (1990) or Campbell (1988), and in almostany other text on contemporary political philosophy. But there is no substitute for Rawis’ own work.157convincingly to be in reflective equilibrium with different people’s consideredjudgments than a multitude of competing principles within a very broad range. Infact, most people’s considered judgment remains stubbornly opposed to Rawis’rejection of desert as a just-making criterion. Finally, it seems now commonlyaccepted that Rawis wide reflective equilibrium argument is ultimatelyinconclusive. He fails to show convincingly that there is a sufficiently determinateset of principles reached via the original position which overlaps anothersufficiently determinate set of principles reached in narrow reflective equilibriumwith most people’s considered judgments in exactly such a way that this overlapcontains the following two principles:(1) Each person has an equal right to the most extensive scheme of equal basicliberties compatible with a similar scheme of liberties for all.(2) Social and economic inequalities are permissible provided that (a) they areto the greatest expected benefit of the least advantaged; and (b) attached topositions and offices open to all under conditions of fair equality ofopportunity. (1977: 160-1)I will not offer a detailed rehearsal of all the seemingly devastating and easilyaccessible arguments of Rawls’ critics, but I do want to make a general point aboutcriticizing Rawls. Although I think it is true that, even considered in combination,Rawis’ arguments do not come anywhere near justifying his favored principles asuniquely required by reason, much of the criticism he has received seems to me tomiss the point of his theory. Much of this criticism depends on taking Rawis’derivation of the two principles from the original position in isolation and also onthe fact that we simply do not have sufficiently precise criteria by which to decidewhat counts as having shown a master principle of justice to be required by reason.But an extremely plausible necessary condition for any rationally acceptable theoryof justice is that it should capture the pattern of expectations and sentiment aboutjustice in a way that has credible potential to enlist enough people’s sense of justiceto make it the basis for a stable social consensus about justice without, however,158involving balancing problems that would completely empty it of any normativedeterminacy in practice. If the degree of a theory’s rational acceptability is judged bythis criterion, then it won’t do for critics to argue that Rawis’ theory hasn’t capturedtheir considered judgments about justice and then propose in its place yet anothernoble-sounding ideal of just individual shares which cannot be justified overdozens of similar ideals and which, in addition, involves utterly unworkablebalancing problems.§2 Rawis’ objective, as I understand it, was to construct a theory whichcaptures a sufficient part of the pattern of expectations and sentiment about justiceso as to have credible potential to make enough people, upon reflection, accept thistheory as sufficiently responsive to their justice-based claims to allow it to serve as abasis for an overlapping social consensus about justice, at least in large, reasonablyprosperous pluralist societies. “However attractive a conception of justice might beon other grounds”, Rawis says,it is seriously defective if the principles of moral psychology are such that itfails to engender in human beings the requisite desire to act upon it. (455)However philosophically elegant, ingenious or provocative a theory of justice mayotherwise be, it must have credible potential to allow justice to play the social rolefor which a theory of justice is needed in the first place:to enable all members of society to make mutually acceptable to one anothertheir shared institutions and basic arrangements, by citing what are publiclyrecognized as sufficient reasons. (1980, 517)And it is hard to see what could give a theory of justice such credible potential to dothat, if not a high degree of responsiveness to the existing pattern of expectationsand sentiment about justice along with persuasive justifications for disappointingsome of these expectations.At the same time, an acceptable theory of justice must incorporate as much159simplification as is necessary to keep the priority, aggregation-distribution andcommensuration problems within manageable bounds. As Mapel and others haveconvincingly argued, the alternative master principles offered by Rawls’ non-libertarian critics such as Dworkin, Galston or Ackerman pose insurmountablebalancing problems and require “second-best approximations which areunrecognizable travesties of the original principle.8And having an unworkabletheory of justice may well be worse than having no theory at all, because theunsatisfiable expectations raised by such a theory may only serve to stir up adestabilizing sense of injustice. There is only one class of theories of equallyambitious scope which escape these problems, namely libertarian theories whoseonly just-making criterion is a particular interpretation of free choice and whichleave it completely up to the market to commensurate the value of different goodsfor different individuals. But avoiding unworkable balancing problems is only oneplausible necessary condition for an acceptable theory of justice; capturing asufficient part of people’s expectation and sentiment about justice to have crediblepotential to become acceptable to them as an account of justice is another. In thisrespect, I think libertarian theories fall short (see Ch 3.5).So Rawls seems easy to criticize but, given his extremely plausible objective,he is certainly not refutable simply by looking at each stage of his argumentseparately and showing that neither his description of the original position, nor hisfavored principles, are uniquely picked out by these arguments. Rawls’ project wasto achieve a wide enough reflective equilibrium which simultaneously takes everyconceivably relevant consideration into account to decide which master principle ofsociety-wide scope captures people’s expectations and sentiment about justice to thegreatest extent possible while requiring a minimum of controversial simplifications8 Cf. e.g. Campbell (1988); Mapel (1989) and Kymlicka (1990) for detailed critiques of these and othertheories. As Kymlicka remarks, “How can we fairly implement [Dworkin’s] insurance scheme? We taxthe rich, even though some got there purely by effort with no natural advantage, and support the poor,even though some...are there by choice without any natural disadvantage.”(82).160to keep the balancing problems manageable. He claims to have achieved such a widereflective equilibrium and has presented a awe-inspiring architectonic ofphilosophical insight to back up this claim.§3 I see only two ways to criticize Rawis without completely begging thequestion about what counts as a necessary condition for a rationally justifiabletheory of justice. First, one can attempt to develop an even more sophisticatedarchitectonic than Rawls does, within an even wider reflective equilibrium, whichwill make apparent that a very different master principle than Rawls’ best balancesthe potential to capture people’s expectations and sentiment about justice with asufficient minimum of controversial simplifications to avoid unmanageablebalancing problems. Second, one can attempt to show that his theory fails on its ownterms and cannot plausibly meet the necessary conditions for a rationally acceptablemaster principle. The first attempt is likely to achieve at best inconclusive resultsbecause there are fundamental indeterminacies involved, both in clearly definingthe various terms of the objective and in deciding what background beliefs areultimately relevant. Although I can see no better ultimate method of theoryacceptance than wide reflective equilibrium, it will by now seem obvious why thismethod, applied to the problem of social justice, is likely to allow for a great deal ofconceptual, normative and empirical disagreement among informed people actingin good faith. Once a master principle plausibly meets the two necessary conditionsfor being rationally acceptable, there would seem to be no conclusive way ofshowing which master principle would be uniquely picked out by a wide reflectiveequilibrium which draws on all the relevant arguments and best availableinformation.But can any master principle of society-wide scope plausibly meet the twonecessary conditions? Here is the second way of criticizing Rawls: asking whether hehas succeeded in his own terms. Does his theory solve the priority, aggregation-161distribution and commensuration problems in a practically workable way while, atthe same time, having the potential to capture enough people’s expectations andsentiment about justice to have credible potential to become acceptable to them asdoing justice and thus to serve as the basis for a stable social order? This is myguiding question in the rest of this chapter.§4 It will be well, first, to remind ourselves of the essential components ofRawis’ master principle. The four essential components of any master principle ofjustice are its primary subject; its scope or domain of application; the particulargoods it individuates as the concern of justice; and the just-making criteria itinvokes. Obviously, the potential number of intelligible combinations of thesecomponents is staggering. Every selection one makes from the field of theintelligible must have a rationale.Rawls chooses a very high level of abstraction for all four components. Hischoice of the basic structure of society as the primary or direct subject of justice isfamiliar. His master principle is meant to have society-wide scope; that is, there isnothing left for social justice to be concerned about as long as this principle issatisfied. The basic structure he fixes on is not individuated at the level of everydaydecision-making in society’s existing institutions, but at the vastly more abstractlevel of a four-stage design of basic institutions in an ideally just society. The goodswhose distribution society’s basic structure is designed to constrain are individuated,not at the level of a plurality of very concrete goods, but again at an abstract level ofwhat Rawls calls primary goods”. Rawls’ five types of primary goods are basic rightsand liberties, powers and opportunities, and the social bases of self-respect. Primarygoods are supposed to be neutral all-purpose means for carrying out any rationallife-plan (1988, 257). And finally, Rawis’ invokes three types of (lexically ordered)just-making criteria. In the domain of political goods, he invokes the criterion of(formal) equality of rights to the greatest system of equal basic liberties compatible162with similar liberties for all; in the domain of social and economic goods, thecriterion of “maximin” inequality (inequality regarded as justified only if itmaximizes the social and economic positions of the representative worst-off person)combined with the criterion of “fair” equality of opportunity in competing foroffices and positions.Rawls has an impressive array of rationales for the particular levels ofabstraction he chooses. I want to review them in the briefest of terms. Rawis’rationale for making society’s basic structure the focal concern of justice seems to beessentially threefold: (1) People have hopelessly conflicting intuitions about what istheir just share in particular situations, which gives rise to “a chaos of competingclaims”(99); and attempts to determine just individual shares by measuring eachindividual’s bundle of goods by some ideal standard and then redistributingresources in a way to conform to such a standard gives rise to hopeless complexities.(2) It is the total scheme of basic social institutions which pervasively conditions theprospects of individuals and often causes conflict at the level of individual shares;and only by knowing what sorts of basic institutions are just can we make thenecessary changes. (3) Only against the background of a system of social institutionswhich create certain opportunity structures and expectations does it even makesense to conceptualize an individual’s desert, rights, or legitimate expectations (cf.304-5; cf. also “The Basic Structure as Subject”, 1977).Next I want to briefly consider Rawis’ basic rationale for fixing on theparticular bundle of strategic goods he calls primary goods. Any master principle ofjustice faces an obvious dilemma when it comes to the individuation of goodswhose distribution it regards as important enough to be the concern of justice. If itsingles out only one type of good, say freedom or happiness, there will be theobvious objections that making only this good the concern of social justice isunacceptably lopsided, or unintelligibly abstract, or a purely verbal unification of anunruly plurality of goods. If a master principle of justice allows many different types163of goods to be the distributive concern of social justice, at the extreme anything thatat least some people can plausibly value, there will be an exponentially growingindex problem, unless a separate just-making criterion is specified for each good (butthat gives rise to an unmanageable priority problem). Rawis’ choice of a short list ofgoods is an obvious compromise between the vexing problems of the single-goodand the many-goods approach. Rawis notes that a theory of justice can capitalize on“a partial similarity in the structure of citizens’ permissible conceptions of thegood”:...comprehensive conceptions of the good, however distinct their content andtheir related religious and philosophical doctrines, require for theiradvancement roughly the same primary goods, that is, the same basic rights,liberties, and opportunities, as well as the same all-purpose means such asincome and wealth, all of which are secured by the same social bases of self-respect. (1988: 256-7; cf. Theory, 90-95)The list of primary goods is meant to include all and only those things which almostanyone wants, and wants to the greatest possible extent, whatever else he may want,because they are necessary or useful means to pursuing almost any plausibleconception of personal well-being (92).Rawis’ rationale for a lexical ordering of his principles parallels, in a way, hisrationale for primary goods. Just as, in the individuation of social goods, the choicebetween a single-good approach and a many-goods approach presents a dilemma, soin the individuation of just-making criteria, the choice between a single-criterionapproach and a many-criteria approach presents a dilemma. Single-criterionapproaches (which may, for example, propose to distribute whatever goods theyindividuate purely on the basis of desert, or purely on the basis of need, or purely onthe basis of particular rights, or purely on the basis of market success) face theobjection of being unintelligibly abstract, unacceptably lopsided or unified only in averbal sense. Many-criteria approaches will, of course, face a priority problem (forexample, what to do when considerations of need conflict with considerations ofdesert), a problem which increases exponentially with the number of criteria. Rawis164well-known proposal to deal with this dilemma is again to seek a compromisebetween too many and too few criteria, and to solve the priority problem amongthose few criteria he thinks relevant to deciding the justice of the basic structure bytrying to justify an ingenious lexical ordering for their application (cf. 42f; 61f; 82f;151 f).Last I turn to Rawis’ rationale for judging the justice of a distribution in termsof what will maximize the “expectations of a representative member of the leastadvantaged group in society. Why should we judge the justice of the basic structureby what is desirable from the perspective of one particular social position, whenstrictly speaking there are as many perspectives as there are individuals? Privilegingthe perspecitve of one of countless social positions in this way- is that not itself aninjustice? We have already touched on Rawls’ dual rationale for this radical stepfor the sake of necessary simplification. It is natural for a person to think of socialjustice from his particular perspective - that is, to interpret just-making criteria inthe light of goals and aspirations which reflect the social context in which thatperson has constructed his self. Justice-based claims are projections of suchembedded selves. Social class is an especially strong determinant of what claims willbe backed by a person’s sense of justice, because people are naturally more likely tocompare themselves with their approximate socio-economic equals than theirinferiors, and to adjust their justice-based expectations accordingly. But strictlyspeaking, there are as many different positions in society from which justice can bejudged as there are people, and depending on whose perspective we imagineourselves to be taking, we would intuitively assess the relative merits of theconflicting justice-based claims of different individuals very differently. Rawisargues that thinking of social justice instead in terms of what will maximize theprospects of a “representative social position” can alone make the commensurationproblem manageable by reducing it to the index problem (44; 64f; 95-100). Moreover,it simplifies the aggregation-distribution problem by giving us a single, coherent165perspective for assigning weights to conflicting aggregative and distributiveconsiderations. Mapel lucidly explicates Rawis’ rationale:By focussing on abstract social positions, rather than individuals, it is nolonger necessary to follow and compare the claims of each and everyindividual within society. Instead, we can leave the matter of individualshares largely to pure procedural justice, as long as the least-advantaged aretaken care of. The device of representative positions also enables us to dealwith the problem of balancing philosophical first principles, by ‘factoring theproblem of social justice into small enough parts.’ Since it is no longernecessary to try to maximize expectations with respect to more than one pointof view...., the problem of balancing general aggregative and distributiveconsiderations has been simplified considerably. These practical adjustmentsmay well give us pause, but [they] seem necessary if we are to avoid thedifficulties involved in trying to be fair to each and every individual and eachand every social group that might possibly raise a claim on resources. Thedevice of representative social positions seems the only solution to theproblems. (Mapel, 98-9; cI. Rawis, 44; 95-1 00)Given, then, that an indirectly responsive master principle can solve balancingproblems oniy by judging the justice of distributive outcomes from the perspectiveof a particular social position, which position shall a defender of such a principlechoose? The social position which Rawls picked to assess the justice of social andeconomic institutions is, of course, the position of the representative worst offperson: both because the justifiably felt injustice seems most acute for this group andbecause, Rawis thinks, judging distributive outcomes by whether they maximize theprospects of the worst off segment of society allows for a “chain connection” ofbenefits rippling upward through society.This completes my brief overview of Rawis’ rationale for thinking of socialjustice in terms of designing social institutions around the goal of distributing a setof primary goods according to lexically ordered principles which stipulate the equaldistribution of political rights and liberties, equality of opportunity in thedistribution of access to offices and positions, and the maximization of an index ofsocial and economic goods with respect to the worst off representative person.Although no such overview can possibly capture the subtleties of Rawis’ own166argumentation, it will do for our purposes here.§5 Is Rawis’ attempt to meet the two necessary conditions for a masterprinciple of justice by resorting to such simplifying devices as a focus on society’sbasic structure, primary goods, lexical ordering and representative social positions atleast a qualified success? Does Rawis succeed on his own terms- providing a theoryof justice of society-wide scope which has plausible potential to capture enough ofthe pattern of most people’s expectations and sentiment about justice to becomeacceptable to them as a theory of justice and to enlist their sense of justice in supportof these principles, while at the same time avoiding unmanageable complexitieswhich would prevent his theory from giving concrete practical guidance?Concerns about the normative determinacy of Rawis’ theoryA. The basic structure: Rawls’ seemingly compelling rationale for choosingthe basic structure as subject raises troubling questions of its own. First, it is far fromclear whether a sufficiently principled distinction between the basic structure andthe rest of the social order can be drawn. Rawls defines “the basic structure” as “theway in which the major social institutions distribute fundamental rights and dutiesand determine the division of advantages from social cooperation.”(7) What, then,are major social institutions?By major social institutions I understand the political constitution and theprincipal economic and social arrangements. Thus the legal protection offreedom of thought and liberty of conscience, competitive markets, privateproperty in the means of production, and the monogamous family areexamples of major social institutions. (7)But Rawls himself admits, “Now admittedly the concept of the basic structure issomewhat vague. It is not always clear which institutions or features [of an ideallyjust society] should be included.”(9) Brian Barry clarifies Rawls’ concept of society’s“basic structure” as follows:167[The basic structure] is made up of the institutions that together determinethe access (or chances of access) of the members of a society to resources thatare the means to the satisfaction of a wide variety of desires. These resourcescan be grouped under three headings: power, status, and money. Examples -this is not an exhaustive list - of institutions that fall under the basic structureof the society are the following: the rules that allocate fundamental legalrights and privileges (equal or unequal) to members of the society; the rulesthat specify how access is gained to political decision-making power; the rulesthat permit concentrations of private decision-making power, such as thatexercised by business corporations, real-estate developers, and so on, inmarket societies; the institutions (including the educational system and rulesof apprenticeship or licensure) that determine access to professions, crafts, orother sought-after occupational positions; and the whole complex ofinstitutions (including some already mentioned but also including rules forthe acquisition and inheritance of property and the system of taxation andtransfers) that determine the distribution of income and wealth in a society.(1989:146)Now indeed we have a clearer picture of what the basic structure” involves. But ifthe basic structure breaks down into many separate domains, surely the questionarises whether Rawls’ focus on the basic structure is really a simplification. Is itlikely that we can tell whether such a large variety of different “institutions”actually produce the pattern of distributive outcomes which Rawis’ principlesrequire? And how is the division of distributive labor among these manyinstitutions to be decided?If Rawis’ master principle were meant to give guidance at the level of allthese specific institutions which make up society’s basic structure, then on the faceof it they would seem to leave decision-makers largely at a loss. How can aconnection be made between maximizing the prospects of the least advantagedrepresentative person and particular policy decisions within institutions? In reply tosuch worries Rawls introduces “an elaboration of the original position”, a “fourstage sequence that clarifies how the principles for institutions are to beapplied”(194-201). Rawls’ basic idea is “a division of labor between stages in whicheach deals with different questions of social justice”(199), accompanied by a complexprocess of gradual lifting of the veil of ignorance. Campbell gives a helpful168summary of Rawis’ own rather difficult exposition of the four-stage sequence:.in the first [stage] the two principles of justice are chosen; in the secondthere is a constitutional convention to set up a system of government; in thethird ordinary laws are legislated and in the fourth these laws are applied byjudges. During this process the veil of ignorance is progressively liftedsomewhat so that at the second stage the constituent assembly knows thenature of the society in question, at the third stage the legislators know thebasic economic facts about their society and at the fourth stage the veil ofignorance is totally removed so that for the first time citizens know their owncircumstances and characteristics. (81)But Rawls instructions are puzzling and counterfactual - the four-stage sequence isof course meant to be a thought experiment (200). Even if decision-makers seekingnormative guidance from Rawl& two principles were able to place themselvesbehind a partial veil of ignorance, what determines how high the veil is to be liftedat each stage? Rawis says thatThe flow of information is determined at each stage by what is required inorder to apply these principles intelligently to the kind of question of justiceat hand, while at the same time any knowledge that is likely to give rise tobias and distortion and to set men against one another is ruled out. (200)We will agree that applying any principles of justice always requires absence of selfserving biases. It is not clear to me, however, how stipulating intelligence and theabsence of such bias will help someone test the justice of contemplated and existingeconomic institutions and the justice of decisions made within them by appealing toRawls difference principle. According to this principle, inequalities are justifiedonly as long as they are to the long-run advantage of the least advantaged (and donot interfere with fair equality of opportunity in qualifying for offices and positions).But how is Rawls’ test fOr the justice of any contested distribution in a givensociety supposed to work? We must ask if the thought-experiment of the four-stagesequence shows the basic structure of a given society to be in essential conformitywith Rawl& difference principle. Now there would seem to be two possibilities:either a contested distribution is itself part of the basic structure, or it is not. If it isnot, and merely flows from a basic structure that is itself just, then - according to169Rawis’ theory - no charges of social injustice are legitimate, as long as the basicstructure is just. But what if the contested distribution is itself part of the basicstructure (which, given a sufficiently broad understanding of‘tbasic structure,isprobably the usual situation in the case of important distributions)? Now when itcomes to deciding whether a contemplated policy conforms to the differenceprinciple, who bears the difficult burden of proving that a contemplated or existinginequality among people will or wont bring about socioeconomic changes that canreasonably be expected to make the worst off group in society better off? And, tofocus on an easily overlooked aspect of this question, what time frame should wecontemplate? How long can benefits be delayed in the name of producing thegreatest long-run’ advantage? Obviously, some policies will make some particularoccupants of the worst-off social position even worse off in the short run but maypromise to make future occupant of this position vastly better off in the end (e.g. bymaking clear that they cant rely on government handouts). By doing what willmake some representative worst-off group in society better off at any given time, wemay well have blocked opportunities for subsequent cohorts of the relatively worst-off to achieve equal or greater benefits they might otherwise have achieved.8Andthere may be all sorts of alternatives with different pay-off schedules, often requiringextremely difficult judgments of probability, which obviously leave tremendousscope for rational disagreement. As Mapel remarks,Even if the parties possessed a social science with perfect predictivepowers.... the problem of balancing the relative benefits and burdens ofvarious reform strategies over time would remain.”U20)Our question is how anyone, at whatever level of decision-making withinsocietytmost important institutions, can get normative guidance from Rawis about8 In the present Canadian context, for example, recognizing the rights of Native peoples to control overlarge territorries, political sovereignty and large cash payments, we may well make one of the mostdisadvantaged groups in Canada better off by changing the face of the country in ways which the mostdisadvantaged among future generations of Canadians will live to regret.170how to reform specific social institutions so as to make the overall system of suchinstitutions more just. Perhaps all that can be said in the end about this question, orall I can say, is that we must largely trust the impartiality and good sense of decision-makers to balance conflicting claims within particular contexts, within theparameters set by a democratically elected government and legally protected rights.But given the apparent ambitions of Rawis’ theory, one might have hoped forclearer normative guidance than Rawis seems to offer us in the end, when he asksus to decide the justice of economic arrangements by going through the thought-experiment of the four-stage sequence.9The normative indeterminacy of Rawis’theory seems almost palpable when he says, in ‘The Basic Structure as Subject”,the two principles of justice do not insist that the actual distribution reflectany observable pattern...What is enjoined is that (permissible) inequalitiesmake a certain functional contribution over time to the expectations of theless favored.. .Institutions must organize social cooperation so that theyencourage constructive efforts. We have a right to our natural abilities and aright to whatever we become entitled to by taking part in a fair social process.The two principles of justice define the relevant fair process and so whateverdistributive shares result are fair. (1977: 164)B. Primary goodsRawls restricts his focus to the just distribution of so-called “primary goods”,which are meant to be all-purpose means to pursuing any rational conception of thegood, and again his rationale seems impeccable. His list of such goods contains sixstandard liberal political freedoms (including the right to own personal property, butnot necessarily ownership of the means of production) and five social and economicIn fact, Rawls himself sometimes seems to disregard the complexities of the four-stage sequence, forexample when he says,“It is fairly straightfoward to ascertain what things will advance the interests of the leastfavored. This group can be identified by its index of primary goods, and policy questions can besettled by asking how the relevant representative man suitably situated would choose”(320;emphasis added).Surely this passage cannot be construed as meaning that each specific policy question within society’sbasic institutions can simply be settled by asking what will favor the least advantaged? But how elsecan it be interpreted? I do not see how this passage could be reconciled with Rawis’ insistence thatsociety’s basic structure, rather than particular distributions, are the subject of justice.171goods: wealth, income, power, authority and the bases of self-respect (cf. 62; 90-95).Several troubling questions arise about this list of goods. Is it really possible tofind a common index of aggregation for these different goods to determine when, infact, the difference principle has been satisfied? Are they really all-purpose means toany rational life plan, or do they reduce a heterogeneous variety of goods verbally toa few simple categories? Do these primary goods cover all the means necessary forwell-being? Let me only address the first of these worries here.Although Rawls’ acute awareness of balancing problems is unsurpassed inthe literature on social justice, it is far from clear to me whether his proposedsolution allows us to identify who the least advantaged in society are from whoseperspective we are supposed to judge if the difference principle is satisfied. What weare supposed to do, if I understand Rawls correctly, is to compare individuals whohold varying bundles of social and economic primary goods and identify therepresentative least advantaged person. But social and economic primary goodsencompass, as Mapel stresses,the powers and prerogatives of offices and positions of responsibility in all ofthe main economic and political institutions of the basic structure, income andwealth, and the social bases of self-respect. Leaving aside the numerousproblems in identifying and measuring each of these goods, there remains theproblem of assigning a relative weight to each part of the index. (Mapel, 105)But if we are not clear about how to index primary goods - if anyone can attach toeach (compenent of a) primary good whatever relative value seems to that personintuitively plausible- how are we going to reach any sort of social consensus aboutwhat will maximize the share of the worst off? Rawis’ ingenious simplicationsreduce the commensuration problem (having to rank the relative value of allpossible bundles of goods for all individuals) to a mere Hindex problem” (having torank only the relative value of different bundles of goods possessed by plausiblecandidates for the social position of worst-off representative group). But the scenarioof many different groups vying for the status of being especially disadvantaged is all172too familiar in our contemporary society. The two related problems here are thatRawis’ index problem is solvable for anyone only in the crudest intuitive way andthat, therefore, there will be nothing approaching a unique solution for everyonewhich could serve as the basis for a social consensus.In fact, if even Rawis’ parsimonious list of primary goods gives rise totroubling problem of intersubjective indeterminacy when we try to comparevarying bundles of such goods to determine who is actually worst off, one hesitatesto take Rawis to task for not including other important goods. In a later paper, Rawlshimself suggests expanding his list of primary goods by including such things as“leisure time, and even certain mental states such as the absence of physical pain,provided we “respect the constraints of simplicityH and keep the goal of a politicalconception of justice in mind: “to serve as the focus of an overlappingconsensus”(1988: 256-7). There are obvious pressures to expand the list of primarygoods; but if Rawls thinks he must give in to them, I am puzzled how he can thinkthat the index problem for primary goods is still sufficiently solvable to give usnormative guidance.The normative indeterminacies introduced by the highly subjectivecalculations required by the index problem, together with the indeterminaciesassociated with the thought-experiment of the four-stage sequence of deciding onthe basic structure of society, would seem to make the normative content of Rawls’theory of justice highly interpretation-sensitive. Rawls leaves me at a loss as to howto generate convincingly determinate answers to contested questions of socialredistribution. Kymlicka is even more blunt in assessing the normative payload ofRawls’ theory:Other than a rather modest proposal to limit inheritances, Rawls gives us noidea how to confront the injustices in our society, or indeed whether hethinks there are substantial injustices to be confronted”(1990: 89).173§6 Concerns about Rawis’ ability to enlist peoplets sense of justiceEven if Rawis’ theory carries no determinate normative payload once we tryto calculate our way through the complexities of the index problem and the fourstage sequence, can at least the spirit of it inspire and harness people’s sense ofjustice? Our answer here can be brief. From the point of view of many thoughtfulpeople’s considered beliefs about justice, the very idea of the difference principle -arranging the economic system for the greatest benefit of the worst-off group - seemsfundamentally misconceived. Even in Rawis’ original position, behind the veil ofignorance, the contracting parties would hardly agree to principles which, at least inspirit, seem to require correcting certain inequalities which strike most people asclearly just (for example, someone’s failure to acquire, through laziness, what thediligent person acquired through effort); and which seem to sanction inequalitiesthat strike most people as clearly unjust (for example, a vast increase in inequalitywhich would marginally improve the lot of the worst off in a society where therelatively worst off are already quite affluent). Even from the point of view of theoriginal position it would not seem irrational to agree to a principle that does notpermit the relatively well-off to enjoy a huge gain whenever the relatively less well-off, who may be quite affluent in absolute terms, suffer a marginal loss as a result. Infact, it has been a persistent criticism of Rawis’ derivation of this two principlesfrom the original position that he does not take the risk-taking side of humannature sufficiently into account.We must conclude, I think, that Rawis’ master principle, even if it wereusefully determinate, would have dubious promise to enlist the sense of justice ofenough people to serve as the basis for a stable social consensus about justice.3.2.4 Limits of the master-principle approach§1 What general lessons can we draw from the shortcomings of Rawis’work? Can other theories succeed where his seems to have failed? What can be174achieved in the field of social justice theorizing?First, another caveat. I face a dilemma. On the one hand, I wish to claim thatno other theory has succeeded in the seemingly very plausible terms which Rawisset for a theory of justice: finding a master principle of justice of society-wide scopewhich can give normative guidance by managing to avoid hopelessly complexbalancing problems while at the same time having credible potential to harnesspeople’s sense of justice sufficiently to make it a basis for an informed socialconsensus. On the other hand, any such sweeping claim should clearly be backed upby solid evidence in the form of a critical study of every theory yet proposed.Although anyone familiar with the critical literature on contemporary justicetheorizing will probably find it safe to say that no one yet succeeded where Rawisfailed, I am clearly not justified in dismissing alternatives to Rawis’ theory withoutdiscussing them. But one cannot fairly criticize a theory of justice without going intoconsiderable detail, and my project here is to engage in synoptic reflections about thepossibilities for rational disagreement about justice - not simply to write yet anotherbook focussed on showing the inadequacy of existing theories. My caveat is, then,that the following tentative and somewhat disjointed speculations should only beaccepted with that degree of confidence which they seem to warrant in the light ofthe reader’s own background knowledge of contemporary theories of justice.§2 What scope seems possible for a master principle of justice? Preferably wewould like to have a distributive social algorithm which any rational person mustaccept for assessing the justice of all important distributions with some reasonableprecision. Rawis was right, I think, that such an algorithm would have to beindirectly responsive, that is, focus on fundamental social background featureswhich condition people’s individual situation in ways that do not requireindividualized distributive assessments and corrective measures. Such analgorithm would also have to be either non-intuitionistic or at least extremely175parsimonious in the just-making criteria and primary goods it individuates, toavoid giving rise to unmanageable priority and commensuration problems. I canonly see two workable suggestions of how, in principle, such a distributivealgorithm could be constructed. Rawis’ suggestion was, of course, to start with apicture of the distributive pattern which a system of just institutions shouldcollectively aim at, institutions which then set up a procedural hierarchy withinwhich distributive decisions at lower levels inherit just constraints from the levelsabove them. I cannot see a way to avoid the problems which effectively emptiedRawl& master principle of any useful normative payload. The other suggestion, thatof libertarians, is to rely almost exclusively on markets to coordinate allocations toindividuals. But it seems extremely hard to justify the libertarian principle in a waythat has the potential to harness people’s sense of justice sufficiently to preventthem from wanting to invoke the concept of justice to press additional claims basedon desert, special needs, rights and other expectations. And there would seem to belittle point in claiming that justice requires us to accept market outcomes todetermine individual distributive shares (supplemented perhaps by a guaranteedsubsistence income), and at the same time allowing all sorts of additionalredistributions in the name of justice. This would simply open the floodgates yetagain to the chaos of competing claims.It is hard to see how any master principle of society-wide scope, whetherfocussed on individual distributive shares or society’s basic structure, could escapewhat we might call Razvls’ dilemma. Either a master principle embodies thenecessary simplifications to keep the priority, aggregation-distribution andcommensuration problems manageable; but then it would seem unable to captureenough of people’s expectations and sentiment about justice to make this principle apromising basis for social consensus. Or a master principle captures enough ofpeople’s expectations and sentiment about justice to make it a promising basis forsocial consensus; but then it would seem to face unmanageable balancing problems176which make the principle normatively empty in practice. It is fair to say thatlibertarian theories typically get impaled on the first horn of Rawis’ dilemma,egalitarian theories on the second. Rawis own theory, in an attempt to avoid eitherhorn, seems to get impaled on both.But if we cannot have master principles of justice of society-wide scope, whatcan we have? It is clear that commonsense percepts which people intuitively applyare always our fall-back option. People are not about to give up justice talk. The needto express, in some vocabulary, claims to a share of social resources whichindividuals perceive as their due is too deep a part of human nature. But surelythere is room to move between a master principle of society-wide scope andcommonsense percepts. These percepts can be given some degree of order whichwould at least mitigate “the chaos of conflicting claims” and which rational peoplewould therefore have to accept, even if these ordering principles do not allow us toassess either the justice of individual shares on a society-wide basis or the justice ofsociety’s basic structure organized as a whole. But perhaps we shouldn’t think ofjustice as having either society’s basic structure or individual shares on a society-wide basis as its proper subject.What exactly is the relationship between the basic social arrangements andwhat can intelligibly be called unjust? Although Rawls was undoubtedly right thatmany disputes about justice arise out of pervasive features of basic socialarrangements, perhaps it is misleading to equate these background features withbackground “injustice”. Although an indirectly responsive master principle aims tostand above our warring intuitions about justice and wants to arbitrate betweenthem, it seems such a principle must fail to capture the complexity of causes whichunderlie people’s sense of justice and instead gives rise to intractable, fruitlessdisputes among rival master principles. Perhaps our understanding of justice flowsfrom context-dependent considerations in a way that makes us lose our grip onwhat we mean when we talk about the injustice of the basic structure of society as a177whole. Mapel’s remarks seem helpful here:The alternative to dropping the idea of the basic structure, understood as asingle, global unit of analysis, is not one of ignoring background injusticealtogether. Rather, what we must give up is the idea of redressing the effectsof background injustice all at once, from some comprehensive perspective.(124)Master principles of justice geared toward society’s basic structure run the danger ofrejecting evolved and deeply entrenched features of society which are nourished bypeople’s basic value orientations and are usually impossible to change all at oncewithout destructive consequences. The historically well-substantiated risks ofrevolutions and radical reforms create a strong consequentialist presumptionagainst them, but these risks do not, of course, rule them out completely under allcircumstances. In some cases, as probably in the case of the communist SovietUnion, extremes of cultural and economic decay do justify a wholesale revampingof basic principles of social and economic organization. But usually there is a bitterprice to pay for any such abrupt break with the past. Society’s basic structures rarelyget revamped to conform to someone’s master principles of justice, and when theydo, the result often turns out to be very different from what the reformersenvisaged. To insist that we cannot sensibly hope to make headway on questions ofsocial justice unless we first agree about which basic structure it would be just forsociety as a whole to adopt may only discourage piecemeal change while holding outa futile hope for magic consensus.This is not to say, of course, that one is not entitled to have a particularvision of the best distributive pattern within society as a whole, but only that it isnot clear how implementing any such vision could be justified, to every rationalperson, as a demand of justice. If a master principle is interpreted, in a vastly moremodest sense, as rationally permitted, it will articulate only one of many possiblevisions of the good society, and will be useless for settling disputes about justiceamong defenders of competing visions.178Obviously, if we cannot even justify some basic moral framework as requiredby reason, the question of what justice requires becomes moot; but I have tried toargue that a form of consequentialism is indeed required by reason. Aconsequentialist moral framework does not seem to pick out one very definitevision of the good society as required for any rational person to accept, although Ithink it allows us, given minimally favorable circumstances, to confidently embraceliberal democracy. But given the epistemic uncertainties of consequentialism, wemust be extremely careful about condemning whole economic systems, socialstructures, and civilizations on the grounds that they do not conform to someone’sexpectations and sentiment about Hjusticefl.But beyond that, there exists, at least in Western pluralist democracies, anuneasy sort of evolved consensus on certain institutional fundamentals:representative government, a list of constitutionally guaranteed rights andfreedoms, due process in legal matters, a market economy supplemented by aneconomic safety net, publicly funded basic education and publicly subsidized highereducation, and a whole lot more. The basic structure of such societies can be alreadysaid to embody an open-ended, strongly pluralistic “master principle” in the waythey regulate the distribution of a multiplicity of goods among their members. Andyet, when it comes to the distribution of many economic goods in particular, socialstrife is rampant and in certain cases escalating, as members of various groupsbecome more militant in demanding “justic& for their cause. Liberal societiesrespond to such matters largely in an ad hoc way, allowing decision-makers atvarious levels of responsibility to balance conflicting claims of diverse pressuregroups in whichever way seems best designed to appease them. This situation is farfrom satisfactory.§3 If we are not simply to resign ourselves to “the chaos of competingclaims”, at what level of generality and determinacy can we pitch principles of179justice? We want principles which are firmly rooted in peopl&s context-boundsense of justice and at the same time clearly justifiable on the basis ofconsequentialist considerations. It seems clear that any justifiable master principlewill have to be quite strongly pluralistic in both its criteria and its goods, and willtherefore seem to face insuperable balancing problems if its domain of application istoo heterogenous. In any case our principles will have to rely heavily on unguidedintuitive balancing of context-dependent considerations in the light of whateverwisdom, impartiality and good will political decision-makers can muster. Thereseems to be only one basic way to reduce such balancing problems, and that is bythinking of principles of justice as applying to restricted domains. Although thiswill reduce the three balancing problems only by introducing a domain demarcationproblem and a domain balancing problem, these problems may be moremanageable, to the extent that there exists a fairly settled understanding of differentdomains of goods. An important type of goods are social roles, and part ofpromoting a settled understanding of different distributive domains will be topromote a stable system of reasonably clear role differentiation among members of asociety and to attach expectations to each role which can be clearly justified in termsof its function. Walzer, for example, individuates eleven distributive domains or“spheres of justice”, each with its own set of roles and goods which are constitutiveof that domain, and each with its own principles of justice which must be applied byintelligent, responsible, caring persons who are both familiar with the “socialmeanings” of the goods at issue and occupy an impartial viewpoint.’0Walzer’seleven domains are community membership; security and welfare; money andcommodities; office; work; leisure; education; love and kinship; religion; andrecognition. Walzer’s approach, which relies on settled social meanings, faces manyobvious limitations and needs to be supplemented with a consequentialist moral10 Cf. Michael Waizer, Spheres of Justice (1983). Waizer also has an interesting principle of theautonomy of different domains which raises many problems. I cannot pursue them here.180framework if it is not to lapse into social relativism. But it correctly describes, Ithink, how societies must try to channel individualstjustice-based claims intomanageable patterns of expectations to which a satisfactory political response ispossible.1813.3 DISAGREEMENT ABOUT THE JUSTIFICATION OF PRINCIPLESThis chapter explores the structure of justification for principles of justice within aconsequentialist framework.3.3.1 The basic nature of consequentialist justification§1 Given the many competing principles which have been proposed bydifferent theories of justice, what sorts of considerations can ever make compliancewith any such principle morally required? What exactly is meant by “justifying’ aprinciple of justice within a consequentialist framework, and how much room doconsequentialist justifications leave for rational disagreement? When is failure tosatisfy someone’s justice-based claims an injustice, and are all injustices morallywrong? Answers to these questions are urgent to inject clarification into manyconfused debates about justice.I will again focus on concerns at the metaethical level, on conceptualstructure rather than on judging the justice of specific distributions. To recapitulatebriefly, I tried to establish a presumption that a wide reflective equilibrium is likelyto support a constructivist, non-foundational, distribution-sensitiveconsequentialism, according to which morality’s main job (promoting enoughuniformity and predictability in peoples behavior to create the necessary sense ofsecurity for a climate of mutual trust and cooperation) is best accomplished bythinking of moral constraints as aiming at an interpersonal goal which combinesboth aggregative and distributive values. A broadly consequentialist approach tomorality requires us to evaluate a system of behavioral constraints by whether itscomponents are individually necessary and jointly sufficient for plausiblypromoting the most desirable combination of satisfied interests of individuals intheir own well-being. In determining which combination of satisfied interests ofindividuals is most desirable, each of us is to be in principle guided by the idea ofgiving to every individuals interest in his or her well-being the same consideration182which each of us would give it if each of us had this individual’s life to live andjudged from the perspective of that individual’s values. In judging a person’sinterests from that person’s perspective, however, the pursuit of purely subjectiveinterests is to be constrained within the bounds of what can plausibly count as “wellbeing” (see Sec 3.4.2). Although many proposed forms of consequentialism arerightly rejected as fundamentally irreconcilable with our considered judgmentsabout the wrongness of routinely sacrificing the vital interests of the few for themarginally greater happiness of the many, this particular form of consequentialismcan arguably achieve a self-consistent synthesis of fundamental distributive andaggregative concerns, although necessarily at the expense of a massive aggregation-distribution problem at the level of fundamental theory. But the idealconsequentialist goal is not to be understood as a deductive moral foundation,justified independently of our considered judgments in particular contexts andbestowing justification on these judgments uni-directionally. It is better understoodas a central node in the web of contextual principles selected in a wide reflectiveequilibrium between considered judgments and background theories. The gapbetween normative indeterminacy at the abstract level and useful determinacy atthe practical level can be bridged only by thinking of moral justification as seekingmaximum consistency among a hierarchy of principles at different levels ofgenerality, each of which is a plausible approximation to the ideal consequentialstgoal in its particular domain.§2 But can even maximum consistency be sufficient? Or must moraljustification be cashed out in terms of some stronger relation between principles,such as strict deducibility, or perhaps derivability or coherence in some less intimatesense, or perhaps only some loose degree of mutual support? It is not easy tocharacterize the relations that hold between moral principles of different content,generality and determinacy. Moral principles do not arrange themselves as a183deductive hierarchy whose implications somehow reach, like tentacles, into everycorner of our lives. The model that doesn’t fit morality at all well, yet stubbornlypersists in some minds, is that of an interpreted axiomatic system which requires usto know only logic and a small number of basic moral formulas to derive, in a moreor less mechanical way, all required moral theorems. These theorems would thensimply require us to have only factual information to generate the relevant moralconstraints in each real-life situation.I will not analyze the ways here in which any minimally sensible account ofmoral reasoning must conflict with this sort of model. For one, the modeloverlooks a major source of rational disagreement we have not considered: theproblem of description-relativity. Obviously, we can only make principled moraljudgments if a state of affairs is understood first to be of a type to which a particularprinciple applies. Whether a particular principle is applicable depends on how astate of affairs is described, but there are often various ways, at varying levels ofgenerality and determinacy, to describe the same state of affairs. Any principledapproach to moral issues runs, therefore, into the fundamental problem that thesame principle will apply to a state of affairs under one description but not apply toit under another. Applying moral principles to real-life situations - connecting anabstraction in the mind up with our experience of the world - always requires acreative leap of the imagination, tremendous selectivity in one’s focus on relevantfeature of experience, and a judgment in reflective equilibrium that will often leaveroom for rational disagreement.1While I have not worked out any alternative model for the “ecology” of ourmoral constructs, something like the following seems to me worth stressing. Ourcollective understanding of moral concepts evolved from a largely subconsciousinductive process which made human beings coordinate their behaviors in certain1 A helpful discussion of the problem of description-relativity applied to political philosophy can befound in Shapiro, The Evolution of Rights in Liberal Theory (1986), 290f.184ways to satisfy desires they naturally had. This coordination can be greatly helped bythe ability of human beings to group situations and behaviors into types withcertain common features, and to communicate with one another about suchmatters. Moral principles can thus be expected to contain, as their chief constituents,commonsense characterizations of types of behaviors along with commonsensedescriptions of types of situations in which these behaviors should be displayed.What relationships is it reasonable to expect between different principles ofthis sort? There are many different ways to carve up the experienced worldconceptually, depending partly on one’s interests and purposes. Since humaninterests and purposes vary, the resulting moral concepts will not neatly carve upthe world into mutually exclusive and jointly exhaustive domains of application,but into a patchwork of overlapping domains and subdomains. Moral concepts orprinciples can only be marked off from one another— for example, justice frombeneficence or gratitude — to the extent that the functional roles they play aredistinguishable. And the functional roles of different moral concepts, as we saw inthe case of rights and justice, often overlap greatly with respect to the sort ofbehaviors they actually require of people. The functional role of the concept ofjustice in a consequentialist moral vocabulary, we said, is twofold: to give voice tothe claims to a share of social resources which individuals or groups make inpursuit of their well-being, backed by their sense of justice; and to articulate apolitical response to the pattern of such claims within some domain. Why such asense of justice might have evolved in human beings is an interestingsociobiological question. At least in general outline, it is not difficult to imaginehow survival pressures might select in favor of types of community-forminganimals whose interest in their own well-being causes them to make demands oncommunal resources, but who are at the same time willing to recognize othercommunity members’ analogous interests and to constrain the pursuit of their owndemands accordingly.185As long as behavior-coordinating principles, arising from various differentpurposes we have, do not contradict by prescribing two or more behaviorssimultaneously that cannot be carried out together, we are justified by our purposesin doing as each prescribes. If “justifying’ a moral principle means exhibiting itsshare in carrying out morality’s job of channelling human behaviors in ways thatpromote well-being , then rather than thinking of moral principles in terms of amore or less well-ordered hierarchy, we should think of them in terms of eachprinciple regulating sub-autonomous domains of goals and purposes. If morality isto accomplish its job, principles need to be more or less tightly coordinated so as notto lead to frequent conflicts between prescribed behaviors. And that means thatlower-level principles must be at least consistent, or loosely cohering, with higher-level principles, not that they must be strictly implied by them. If some such accountof the underlying nature of moral principles is correct, then it is hard to see whatelse it could mean to “justify” a principle of justice than to make plausible that itboth responds appropriately (in a sense to be clarified) to the pattern of expectationsand sentiment about justice in some domain and also happens to fit in reasonablywell with an overall system of principles whose wellbeing-promoting tendencies arealready inductively well-established.§3 But what is this system of principles which has already met theconsequentialist well-being test? What is the fundamental content of aconsequentialist system of personal and political morality? My aim here is not todevelop such a morality in great domain-specific detail, but simply to sketch inoutline how it can be launched. I shall assume that we are designing a morality for acomplex, pluralist society and move quickly through some very complicated issues(some of which will be taken up in the next chapter).What system of principles, standards, norms or constraints on the behavior ofindividuals, groups, and governments can by justified in a wide reflective186equilibrium, given our understanding of such a system’s main job, of the idealconsequentialist goal, of the nature of typical human interests and other features ofhuman psychology, of prevailing moral traditions, of the causal structure of theworld, and other relevant background knowledge? What system of principles willhave the most desirable consequences?In justifying a particular principle as having more desirable consequencesthan its alternatives, we must ask not merely how desirable its total consequenceswould be if everyone complied with this principle. Assuming perfect compliancemay, at best, be a very rough initial approximation to eliminate obviouslyunacceptable principles from the rest. Consequentialists, for whom only theconsequences a principle will produce in the real world matter, must also ask threefurther questions: (:1) Will the proposed principle still be the most desirablealternative, given the likely extent to which people will, for many complex reasons,fall far short of perfect compliance? (2) Will the proposed principle still be the mostdesirable alternative, given some often overlooked costs associated with its effectivepropagation, such as disrupting peoples established habits, creating anxieties, andrequiring teaching and enforcement efforts? (3) Will the proposed principle still bethe most desirable alternative, given that it should lead to as few conflicts as possiblewith any of the other principles which are to make up society’s total system of moralnorms ?2So strictly speaking, we cannot justify a principle of justice simply by deciding,in reflective equilibrium, whether perfect compliance with it would plausiblypromote well-being better than its alternatives. The project of consequentialistconstructivism requires us to ask a far more complicated question: HWill the totalsystem of social norms of which this principle is to become a part produce the most2 Cf. Richard Brandt, A Theory of the Right and the Good G979), for the distinction between idealcriteria of right and wrong and a practical moral decision procedure, in which those criteria arewatered down so as not to put the morally conscientious person at too unfair a disadvantage in a societyin which these criteria are not generally complied with.187desirable consequences, given the extent to which people will probably fail tocomply perfectly with this system, and given also that there are costs of maintainingthis system?” Let me call this the ultimate conquentialist test, as opposed to the idealconsequentialist goal.Although the information and computations required to assess theconsequences of a proposed principle and all its alternatives along all thesedimensions would seem to be staggering at first, in practice the approximate range ofprinciples which are plausible candidates for passing the ultimate consequentialisttest is not at all hopeless. We know, for example, that even otherwise idealprinciples won’t do us any good if enough people cannot be motivated to complywith them at a reasonable enforcement cost. To achieve sufficient compliance withmoral principles requires most people’s active participation in maintainingcompliance pressure on people around them. To achieve such compliance pressure,given that human beings are largely creatures of habit, will require us not to taxtheir reflective capacity concerning fundamental moral matters too heavily, but torespect the inertia of people’s childhood conditioning as one of the facts ofcommonsense moral psychology. So any acceptable system of moral norms willhave at least three crucial pragmatic characteristics. It will contain only principlessimple enough so that most people inhabiting the domain of application of theseprinciples will understand them. And it will contain only principles that do notgenerally require people to make the sort of self-sacrifices which would conflict withtheir (reflectively rational) self-interest to such an extent as to motivate widespreadnon-compliance.3Andit will contain a strong presumption in favor of requiring3A rational morality must ultimately be reconcilable with self-interest, not in the implausible Socraticsense that doing the morally wrong thing could never make anyone better off, but in the sense that, first,the system of moral norms must plausibly aim at promoting the most desirable combination of satisfiedinterests of individuals in their own well-being, and second, the system of moral norms must be suchthat, with a not too costly system of rewards and punishments, ultimately most people, most of thetime, regard it as their best bet to comply with these rules and cultivate, out of enlightened self-interest, the disposition to be morally responsible. Of course, no moral system can ensure that doing themorally wrong thing will not make someone better off, given the status quo as the baseline forcomparison. All that is required for an effective morality is an appropriate system of rewards and188moral innovations to be made piecemeally and to be clearly defensible asconsequentialist improvements, rather than making a clean sweep of existingpractices and beginning de novo.Given these pragmatic conditions on acceptable moral principles, plausiblecandidates for the core of a consequentialist morality are not hard to identify. Wecan make a fairly confident judgment that the system of moral principles shouldinclude the following: “Tell the truth”, “Don’t steal”, “Keep your promises”, “Obeythe law”, “Don’t engage in unprovoked physical attacks”, “Take care of yourchildren” or “Don’t discriminate against people on the basis of race or sex”. Let usfollow Haslett in calling these the “primary norms of personal morality”(cf. 88ff).One conspicuously absent norm, which clearly fails the ultimate consequentialisttest for primary norms, is the act-consequentialist principle, “Always do, on everyoccasion, what you think will have the most desirable consequences.” This principlefails the test, and is therefore not among the primary norms of a rationallyjustifiable personal moral code, because it is far too epistemically demanding andwould lead to the sorts of disagreements, failure of coordination, andrationalizations for freeloaders which could not accomplish morality’s main job.4Primary norms of political morality will specify the central task ofgovernment, namely to solve coordination and free rider problems by enforcing asystem of legal sanctions. Primary norms of political morality will further consist ofnorms specifying acceptable conduct for government officials and restrictinggovernments and individuals in what they can permissibly do to other individualsor groups. In the contemporary context, many of these norms will be stated in thelanguage of rights. Nothing about consequentialism necessitates rights-talk (exceptpunishments which makes it a good bet for most individuals to comply with the important moral ruleseven in the face of most temptations.4Harsanyi, among others, has given an excellent account (which I shall not try to repeat here) of theadvantages of rule over act consequentialism in terms of coordination, expectation and incentive effects(cf. 56-60).189perhaps a right to life and certain basic freedoms). The point of having a morality isclear; but whatever conceptual division of labor we evolve to carry out morality’sbasic job - whether we use talk about rights, goals, virtues, or other moral concepts -is relatively immaterial and partly a matter of historical contingencies.Norms of acceptable governmental conduct, in addition to rights, would verylikely have to include the requirement that governments must themselves obey thelaw, keep their agreements, give aid to other countries, or take care to some extent ofthe interests of animals and future generations. About the details of these normsthere can obviously be considerable rational disagreement, but hardly about thequestion as to whether we should have norms governing these matters at all. Andthe consequentialist rationale for enforcing norms about a certain matter is, ofcourse, that allowing individuals to act against such norms in order to promotewhat seems to a given individual the greatest overall well-being on a particularoccasion is, given our epistemic limitations, our propensity for self-interestedrationalizations and other people’s expectations of us, almost always a badconsequentialist gamble.5Controversy will frequently erupt about the area of governmental actionwhich is left over after governments have observed all constitutionally or otherwisepredefined norms - an area which is necessarily governed by some version of theact-consequentialist norm after all. Within limits, usually spelled out in a country’sconstitution, consequentialists want governments to promote the greatest wellbeing according to their best judgment in the context of current opportunities. Infact, precisely because there can be, within limits, many conflicting interpretations ofwhat the specifics of this goal amount to, governments are necessary as arbiters inany society. Because a government-of-the-day’s power of discretion is so great andmistakes are potentially fateful, however, another plausible norm ofconsequentialist political morality is that the individuals forming each particular5See Haslett (135) for detailed arguments to this effect.190government should be subject to a system of checks and balances which periodicallyincludes free elections.In addition to primary norms, a complete system of personal and politicalmorality must, of course, also include secondary norms which tell us what to do infour special types of cases: (1) emergency cases (and golden opportunity cases) whereit seems virtually certain that we could prevent immense evils (or achievetremendous good) by ignoring ordinary norms; (2) conflict cases or borderline caseswhere two or more norms seem to sanction incompatible courses of action; (3)oddball cases which are so infrequent that it is not worth complicating our system ofexplicit norms for their sake; and (4) cases without precedent, which are types ofcases which we had no occasion to contemplate in the past and for which ourexisting norm system is not prepared. Although I cannot pursue this matter here, itis possible, I believe, to come up with plausible consequentialist norms to resolvethese types of hard’ cases in a principled way most of the time (cf. Haslett, 100ff; alsoDworkin, 1985). I now turn to the question of how principles of justice areembedded within a general system of consequentialist moral principles.3.3.2 Bona fide principles of justice and morally justified principles§1 We must distinguish two types of principles of justice, according to thestage they have reached in the process of consequentialist justification: bona fideprinciples and morally justified principles. Bona fide principles of justice constitutean “appropriate” response to conflicting justice-based claims in a given domain bystriking a reasonable compromise between them. Morally justified principles ofjustice are principles that require a morally responsible person’s compliance. A bonafide principle of justice is not generally the same as a morally justified principle,because there can be many competing bona fide principles in a given domain, andonly after all bona fide principles have been subjected to the trade-off considerationsnecessary to determine the morally best can one of them be regarded as morally191justified.A bona fide principle of justice is a proposal for achieving a reasonablecompromise among different individuals’ justice-based claims to social resourceswithin some sufficiently well-bounded distributive domain. Bona fide principlescan be more or less comprehensive. The most comprehensive principle conceivablewould be a master principle of global scope which encompasses all distributivedomains in existence and would balance conflicting justice-based claims to globalresources both within and between its many subdomains to any desired degree ofprecision. Given the intractable problems arising with a principle’s increasing levelof comprehensiveness, however, no such “ultimate master principle” seemsfeasible. Most bona fide principles of justice will simply be proposals for achieving acompromise among conflicting justice-based claims in a restricted domain.I do not think there can be anything very exact or systematic aboutindividuating distributive domains. Domains are simply those contexts ofcooperative human activity which shape the identity of individuals inhabitingthem. Many large, inhomogeneous domains have large numbers of importantsubdomains; for example, the domain of paid employment. Domains aredemarcated primarily according to evolved understandings about the nature ofdifferent goods, institutions and practices, the functional differentiation betweensocial roles, ethnic groupings, or geographic boundaries. The social understanding ofa domain’s boundaries, even though not of just distributions within it, must bereasonably well settled for domain-specific principles of justice to become viable.Although the consequentialist outlook is fundamentally cosmopolitan, the choiceof a politically organized “society” or “nation state” as marking off an especiallysalient domain is justifiable for many purposes.66 Peter Danielson suggests (personal comm.) that the question of international redistribution is anespecially important test for consequentialism. Contractariansm, it would seem, can keep theobligations of affluent nations toward the Third World within reasonable bounds, whereasconsequentialism would require a tremendous transfer of wealth from rich to poor countries. This mayseem to some a reductio ad absurdurn of consequentialism. I cannot digress into this complex issue here.192The major balancing problems afflicting principles of justice can be alleviatedby dividing the problem of justice into problems within different domains, even iftheir exact nature and boundaries are often themselves in dispute. Where socialunderstandings are unsettled, what matters is that individuals have opportunitiesto voice their dissent and change prevailing conditions within a domain or, ifnecessary, to exit one particular domain for another that suits them better (cf.Skhlar, 166f). Different domains may compete for a common stock of scarceresources, but admitting this obvious fact does not undermine the claim that we canonly make the problem of justice manageable by dividing it into problems withinseparate domains. For example, problems of justice arising in the domain ofallocating scholarship funds among graduate students can generally be settled byclearly articulated domain-specific procedural and material principles whichstructure the expectations of individuals inhabiting this domain. Problems of justicearising in the wider domain of deciding a university’s budget can also be settled,usually after lively debate, by a domain-specific combination of principles. Andproblems of justice arising in the even wider domain of deciding a governmenttspending priorities can in turn be settled by principles appropriate to that domain,such as, for example, some established variant of a fundamentally democraticprocess cons trained by constitutionally protected consequentialist liberty and welfarerights. But the process of creating constraints which are acceptable to people’s senseof justice works both top-down and bottom-up. Subdomains must adjust theirprinciples as a result of pressures from wider domains, wider domains must adjustBut it is not obvious to me that consequentialists in affluent countries must rush in and help the starvingwith all the resources they have. Arguing about principles of global redistribution within aconsequentialist framework will require (1) a strong probability that a proposed set of measures willindeed restructure the complex causal web of the world’s environmental, demographic and economicproblems in an desirable way; and (2) a factoring into this probability a realistic estimate of themotivational barriers resulting from substantial sacrifices required of individuals in the name of moralidealism. Proposals to expand the scope of theories of social justice beyond nation-states - such as, forexample, Pogge’s proposal to apply Rawis’ maximin principle globally - are likely to impose suchstrains on most people’s moral commitments in affluent parts of the world as to predictably fail toharness their sense of justice in support of such proposals, and thus also to fail the ultimateconsequentialist test.193their principles as a result of pressures from subdomains. At every stage in this first-order contextual process it is important to be sensitive to sources of limited rationaldisagreement which require pragmatic compromise-seeking informed andconstrained by an awareness of the sources and limits of first-order disagreement.§2 We defined a bona fide principle of justice as a proposal for achieving areasonable compromise among those conflicting claims to social resources made bydifferent individuals in pursuit of their personal well-being which are backed bytheir sense of justice. Let me now explore the most reasonable spelling-out of theidea of a “reasonable compromise” within a consequentialist moral framework. Aproposal for achieving a compromise among conflicting claims to social resources ina given domain is reasonable if and only if (1) it can be implemented, or crediblyapproximated, by some feasible distributive mechanism (which implies amongother things that it does not allocate resources to conflicting uses); and (2) it isresponsive to existing patterns of expectations and sentiment about justice withinthe relevant domain. A third condition must be added, for obvious consequentialistreasons: (3) The existing pattern of expectations and sentiment about justice are theresult of adequate information available to people and not the result of effectivebrainwashing techniques which have made segments of the population willinglyembrace slavery or its equivalent.These seem to me plausible consequentialist conditions for makingsomething a bona fide principle of justice, though they contain elements ofvagueness, and judging when they are met will often leave some room for rationaldisagreement. How, most importantly, are we going to explicate the exceedinglyvague responsiveness condition? It is clear that a bonn fide principle of justice mustbe, to some hard-to-define degree, responsive to existing patterns of expectation andsentiment about justice in order to be recognizable as a principle of justice at all.7 AtThe charge is often levelled against Gauthier’s “mutual advantage” contractarianism that it is not a194one extreme, a principle would be strongly responsive to existing patterns ofexpectations and sentiment about justice if it could somehow satisfy everyexpectation and conform to everyone’s sentiments. At the other extreme, a principlewould be weakly responsive to existing patterns of expectations and sentiment aboutjustice if the balance it proposes among conflicting justice-based claims has someminimally credible potential that enough people’s sense of justice could realignitself with this balance eventually to produce tolerably favorable conditions forsocial cooperation. What is the most reasonable compromise between thesealternatives?A CND consequentialist (unlike certain contractarians) will insist that theidea of a “reasonable compromise” be spelled out so as to embody a plausibleconception of impartiality. We might say that a balance between the demands ofdifferent just-making criteria is adequately responsive to existing patterns ofexpectations and sentiment about justice if it corresponds to the decision that couldplausibly be reached by a tribunal whose members have an in-depth understandingof the domain at issue, have personally nothing to gain or to lose from theirdecision, and are guided by a desire to satisfy different individuals’ competingjustice-based claims to the greatest extent compatible with respect for theseparateness of persons.8This idea is lucidly expressed by Brian Barry:What, after all, is the virtue of impartiality, as we seek it in, paradigmatically,judges and arbitrators other other public officials? An impartialdecisionmaker is one who acts ‘without fear or favor’. This includes, ofcourse, not taking bribes, anticipating rewards, or succumbing to threats. Butit also means the absence of more subtle departures from objectivity;impartiality excludes favoritism based on friendship, similarity of race orclass, and so on.. .The reasons for deciding should be general, publicly statable,theory of justice at all, but merely a description of realpolitik. See Kymlicka, 1990, P. 131, who citesRawis, Barry and Sumner as making this point. This charge is not, of course, a refutation of Gauthierwho claims that his theory is the best we can have by way of rational agreement on principles of socialcooperation.8 I address below the complicating factors for any such criterion of “adequate responsiveness” arisingfrom the fact that different representative committees might well reach different distributionaldecisions.195and publicly defensible.... (1989: 290)This idea of procedural impartiality can be readily (though usually far fromperfectly) operationalized in a multitude of interesting ways which I cannot discusshere. But what makes something a bona fide principle of justice, rather than merelyan expression of personal opinion or sentiment, seems now fairly clear. It must be aproposal for achieving a reasonable compromise, relative to some sufficiently well-bounded domain, among different individuals? conflicting justice-based claims tosocial resources, where ??reasonable compromis& means that (1) the proposal can besatisfactorily approximated by feasible distributive mechanisms; (2) it corresponds toa decision that could plausibly be reached, after due deliberation, by a group of(intelligent) decision-makers of good will who are guided by a thoroughunderstanding of the nature of justice-based claims in the domain at issue (and ofneighboring domains), and guided also by the desire to satisfy these claims to thegreatest possible extent compatible with respect for the separateness of persons; and(3) the pattern of people?s expectations and sentiment has been shaped by adequateinformation and not by effective brainwashing techniques which have madesegments of the population willingly embrace slavery or its equivalent. As isevident from this definition, there is no avoiding the aggregation-distributionproblem at any level of moral decision-making, no calculus that would makequalitative judgments unnecessary, although the difficulty of making thesejudgments to the satisfaction of all or most individuals affected will become quitemanageable for experienced decision-makers in suitably restricted domains.3.3.3 Defining injustice§1 If something like this is indeed the most plausible set of conditions forbona fide principles of justice, then there will, almost inevitably, be a range of suchprinciples relative to most viable domains. So there can be rational (first-order)disagreement about which principle should be implemented, given that only one196can. Now given that justice permits, but does not require, the implementation ofany particular bone fide principle from within this range, what does it mean forsomething to be an “injustice’? For something to be justifiably called an injustice, itis clearly not sufficient that some justice-based claim has to be disregarded or thatsomeone’s sense of injustice has been aroused. Something that we must demandfrom any minimally plausible account of justice is that it spell out sufficientlyrestrictive conditions for calling something an “injustice” so as not to debase thenormative currency of appeals to justice. At the same time, the conditions must notbe so restrictive that non-justice considerations routinely overrule any reasonablecompromise between conflicting justice-based claims. Here consequentialists face adifficult conceptual choice.Should an injustice be diagnosed whenever a distribution violates some bonafide principle or other within the permissible range of such principles in a givendomain? Or only if it violates every principle within this only approximatelydelimitable range? Or only if, in addition to violating every relevant bona fideprinciple, it meets further conditions, such as passing the ultimate consequentialisttest which tries to harmonize justice-based claims within a domain with all-things-considered morality?I shall label these three options for the use of the concept of injustice “theleast restrictive”, “the moderately restrictive” and “the most restrictive” account ofinjustice. Which should a consequentialist accept? There would seem to be strongpros and cons on all sides of this conceptual issue. It would seem plausible to adoptthe least restrictive account (allowing the violation of whatever bona fide principlesome individual might have preferred to be called an injustice- at least to thatindividual), because all rival principles within this permissible range are equallybona fide principles of justice, and choosing one does violate all the others, and itmay make all the difference to an individual which principle is actually acted on.But it would seem equally plausible to adopt the moderately restrictive account197(allowing only distributions that violate every bona fide principle within a givendomain to be called injustices), because a particular distribution can only satisfy onepermissible rival principle, and it is clearly a matter of rational disagreement whichprinciple within this range is preferable, and we do not want to debase thenormative currency of injustice-claims. And, finally, it would also seem plausible toadopt the most restrictive account (allowing only distributions which violate a bonafide principle that has met the test of all-things-considered morality to be calledinjustices), because in the popular mind the term “injustice’t strongly connotes amoral wrong, and violating a bona fide principle is morally wrong only if thisprinciple does not fail the ultimate consequentialist test, and this account wouldseem least likely to debase the normative currency of injustice-claims.How do we want the concept of injustice to function in our moralvocabulary? Which use of the term “injustic& should consequentialists legitimizeand therefore encourage people to adopt? There is no ideal solution. On balance Iincline to the moderately restrictive account, despite some very serious drawbacks.The moderately restrictive account of injustice makes it both less likely than theleast restrictive account that the unjust and the morally wrong will be very far apart,and less likely than the most restrictive account that legitimate uses of the conceptof injustice will be radically divorced from what individuals claim as their due inpursuit of their well-being. This account implies that, in order to make an injustice-claim legitimate, more is required than simply the violation of some plausible bonafide principle or other in some domain, but less is required than being able to defendthis injustice-claims against the demands of all-things-considered morality. On theone hand, this account requires someone who is displeased with a distributivedecision to look at a whole domain of application and ask whether the violation ofhis preferred bona fide principle was not perhaps required to avoid violatinganother such principle. It requires an individual who feels unjustly treated toconsider whether perhaps a distributive decision did not go his way because others198had as good a claim as he did to having the decision go their way. This may putunrealistic demands on some individuals, but making such demands, trying tohabituate individuals to judging justice from a wider perspective, seems eminentlydesirable if appeals to justice are to have a harmonizing rather than divisive effect.On the other hand, the moderately restrictive account of injustice does notdivorce uses of this concept so radically from the particular contexts of individuals’justice-base claims that it would lose its connection with them, would have nopotential to harness most people’s sense of justice, would require them to wade intothe complexities of all-things-considered morality and would get us into all thedifficulties of ambitious master principles of justice of society-wide scope.§2 Given that Rawls’ project of tying the concept of injustice fundamentallyto an ideal of society’s basic structure has not proved feasible, I see no alternative fora consequentialist account of justice than to follow Walzer in tying the necessaryand sufficient conditions for justifiably claiming an injustice fundamentally,although not indefeasibly, to those spheres of evolved social understandings inwhose context the identity of individuals is constructed and embedded. Justice-basedclaims are fundamentally projections of such embedded selves. The pattern ofpeoples expectations and sentiments is structured by their understanding of what isproper within particular domains of goods and roles. It is true that individuatingdomains poses problems, including problems that can themselves be intelligiblyconceptualized as problems of justice, and that different domains often draw on acommon pool of scarce resources and interact with each other in complex ways. Butfollowing Rawls in going beyond individual domains in our diagnosis of socialinjustice and trying to assess the totality of distributive domains all at once wouldonly make us lose our grip on questions of who can reasonably demand what in thename of justice, or of what we mean by “justice” at all.- But isn’t my balancing view of justice simply a form of social relativism,199taking as just whatever evolved social meanings and patterns of expectations andsentiment prevail in a particular social context? There is admittedly a strongelement of conservatism in the account outlined above, and it has this much incommon with relativism that the specific content of morally justified principles ofjustice may vary from society to society, and from social context to social context, ifpatterns of expectations and sentiment vary. But all accounts of justice will havethis feature to some extent, unless they propose some ideal of what justice shouldmean, no matter how strongly this conflicts with people?s evolved understanding ofwhat is their due in pursuit of their well-being as individuals. And if the acceptanceof any particular alternative among many such ideals cannot be shown to berequired by reason, proposing such an ideal as an account of justice represents meredogmatism.But my account differs from both unfettered social relativism and inflexibleconservatism in two main ways. First, although I consider bona fide principles ofjustice to be indeed a complex function of existing patterns of expectations andsentiment about justice, the three conditions for what is to count as a “reasonablecompromise” ensure that such principles do not necessarily reflect the expectationsand sentiment embodied in existing laws or held by some politically dominantgroup. Second, even though bona fide principles of justice may embody a highdegree of cultural and domain-dependent relativity, they are morally justified onlyif they cannot be shown to conflict with the ultimate consequentialist test.Injustices can thus be morally justified at times. People must be educated torecognize more clearly why having determined an injustice relative to some moreor less well-defined domain is not necessarily the end of all moral debate. It onlycreates a very strong presumption that a moral wrong has been committed andcreates a heavy burden of moral justification. If we are using the term “injusticet’inthe way that seems to be advisable from a consequentialist perspective, then somethings that are quite properly termed injustices can only be rectified by sacrificing200values of equal or greater moral importance or by committing similar or evengreater injustices in other domains.But admittedly, this account makes legitimate charges of injusticeconceptually fuzzier than many people’s facile and confident use of this termreflects. What makes a distribution unjust, on this account, is that it violates allprinciples which propose a reasonable compromise between justice-based claims insome viable domain. But what seems like an injustice if viewed within somenarrowly restricted domain may, of course, no longer be so clearly an injustice ifviewed from a more expansive domain. Moreover, overall moral justification hasto look to more than one domain; it has to look to the best consequences fordistributing resources among all distributive domains. Therefore, on the moderatelyrestrictive (domain-relative) account of injustice, morally justified injustices mayarguably occur in just those cases when every single one of the bona fide principlesof justice can be be plausibly argued to fail the ultimate consequentialist test.3.3.4 Rational disagreement about consequentialist justifications§1 But how do we apply this test in practice? Ideally, of course, we wouldbring our entire system of already existing personal and political norms to bear on aproposed principle of justice and reach a decision in reflective equilibrium aboutwhere any necessary adjustments should be made in case of conflicts. But for manypurposes we clearly need a more manageable decision-making process. Are thereperhaps plausible approximations to our ideal consequentialist goal which we coulduse as criteria to decide which particular bona fide principle of justice among allthose principles striking a reasonable compromise among conflicting justice-basedclaims in a given domain should actually be followed?I can think of no better criteria for choosing between conflicting bona fideprinciples than the idea of using certain traditional moral standards in combination.These standards accord with the considered judgments of many thoughtful people201and direct our attention to plausible aspects of the ideal consequentialist goal, suchas the promotion of happiness, moral consistency, respect for persons, or voluntaryagreement. The following table states what seem to me plausible interpretations ofsuch moral standards, applied to the problem at hand:The greatest “Will including this principle as a principle of justice in ourhappiness test system of moral norms foreseeably create more unhappinessthan some viable alternative?”The moral “Will recognizing this principle as a principle of justiceconsistency test foreseeably set an example which could not be adopted inother domains without disastrous consequences?The respect-for- “Will recognizing this principle as a principle of justicepersons test foreseeably (and avoidably) treat some people merely as ameans to other people’s ends?”The voluntary “Could informed people of good will voluntarily agree toagreement test recognize this principle as a principle of justice for ourmutual benefit (at least if we judge from a reasonablyimpartial standpoint)?”Table III: Consequentialist formulations of traditional moral standardsused as criteria for proposed principles of justiceThese standards cannot, of course, serve in the fashion of mechanical decision-making devices, but only point in the general direction for looking beyond arestricted domain in our attempt to pick the most defensible option amongconflicting bona fide principles.Giving confident answers to the tests of our basic moral standards will oftenbe difficult. They may therefore leave ample, though not unlimited, scope forrational disagreement. But many (in some cases perhaps all) bona fide principles ofjustice will clearly fail to meet the test of these standards. In this way, thesestandards constrain what can permissibly be done in the name of justice. Insofar asdisagreement about what can justifiably be done in the name of justice arises fromone party’s unwillingness to accept these constraints, I do not consider it to be aform of rational disagreement.Of course, there may often be several bona fide principles of justice relative to202a given distributive domain which pass the test of these moral standards. So therewill, at this final stage in the process of (first-order) moral justification, often beconsiderable room for rational disagreement about which principles of justice amorally responsible person is obliged to comply with. This result is no surprise.Several rival principles of justice will often be equally morally justified at the first-order level by any plausible consequentialist account of moral justification (and theparticular principle which actually requires compliance can only be singled out by asecond-order legitimation process).What creates so much room for rational disagreement about principles ofjustice? These principles must be morally justified by applying the test of our basicmoral standards to various alternative ways of achieving a reasonable compromisebetween the conflicting justice-based claims of different individuals. But our basicmoral standards often leave a wide range of strongly conflicting ways of achievingsuch a compromise equally permissible, not only because of our epistemiclimitations, but also because different types of social distributions will realizedifferent but equally plausible forms of well-being. The best we can reasonablyexpect, in most important distributive domains, is that our rational standards,together with our conditions for what counts as a bona fide principle of justice, willselect a relatively narrow range of competing principles from the immense range oflogically possible principles.§2 Although this consequentialist account of justice seems to me to haveinteresting normative implications for many contemporary moral issues, I want tosteer clear of applied ethics here. Doing even rough justice to all the consequentialistpros and cons of any interesting real-life moral issue would consume far too muchspace. Nevertheless, some illustration of my consequentialist account of justice is inorder. Suppose we want to find a principle which provides a sufficient condition forjudging justice within some distributive domain, say distributing opportunities for203higher education. Applying first the greatest happiness standard, we might ask:“Which principle specifying a reasonable compromise among justice-based claimsin this domain is least likely to create unhappiness?”Fortunately, bona fide principles of justice will generally have some inbuilttendency to be compatible with promoting happiness, because each just-makingcriterion evolved in the human understanding as a response to individuals’ deepaversion to types of distributions causing misery. Plausible answers to the question,“Which policy both achieves a reasonable compromise between justice-based claimsin the domain of opportunities for higher education and is not obviouslyincompatible with promoting happiness (and will also satisfy our other moralstandards)?” will yield our justified principles of justice in this particular domain.Since there will probably remain a range of such principles, the question, “Whichparticular principles of justice must an informed person of good will comply withwhen it comes to distributing opportunities for higher education?” has nocompletely definite answer at the first-order level of moral reasoning, and the finalanswer must be determined by purely procedural principles at the level of second-order accommodation.What justice-based claims are there in the domain of opportunities forhigher education? Many claims are advanced within this domain on the basis ofdesert, need, equality, rights and good-faith expectations, in a way strongly backed bypeople’s sense of justice. For example, those who worked hard to meet highacademic standards argue that they deserve to be admitted to university. Those whocome from socially disadvantaged ethnic groups which need doctors or lawyers oracademics as role models argue that they especially need a university education andshould perhaps enjoy relaxed admissions standards. Those with relevantly equalabilities argue that they should be treated equally, regardless of socioeconomicbackground or other characteristics, when it comes to access to university spaces.Each of these claims might be couched in the language of rights and some may even204be justifiable as rights. Certainly, a lot of people have good-faith expectations ofbeing admitted and often feel unjustly treated whenever admissions standards aresuddenly raised to keep them out.Now, finding a principle or policy which specifies a reasonable compromisebetween the conflicting justice-based claims in a way that will not conflict with theultimate consequentialist test or its approximations obviously requires all thejudgment, wisdom, experience and intelligence which we can mobilize for the task.It may be impossible ever to be absolutely sure that one has hit upon the bestsolution, but the line of effort required by a rational conception of morality is fairlyclear. Most possible principles or policies fall clearly outside the range of what isplausibly justifiable as morally permitted. One such possible principle that is outsidethis range is: “Distribute university spaces preferentially to those with lowintellectual capacity.” This principle is outside the range of justifiable principles ofjustice, both because it does not specify a reasonable compromise between people’sactual justice-based claims and because distributing university spaces preferentiallyto those with low intellectual capacity will clearly not promote happiness, not eventhe happiness of those who get to occupy these spaces (without fundamentallychanging the character of the institutions we now call “universities”). Much morehappiness will be created by giving spaces preferentially to those most able andmotivated to benefit themselves and society from a university education.One principle of justice which might fall inside the range of the rationallypermitted is: “Make university spaces available to all those who achieved at least an80% grade point average in the final year of high school and applied before June11th, except for applicants from certain disadvantaged backgrounds.” Of course,literally thousands of different variations of principles of justice can be imagined inthis situation which are all rationally permitted by first-order moral reasoning.We can try to narrow down the number of competing principles somewhat byapplying, in combination, the greatest happiness test, the moral consistency test, the205respect-for-persons test and the voluntary agreement test. The remaining principlesmust then be submitted to the judgment of legitimate decision-makers elected orappointed for this purpose. Only after one such principle has been selected by thissecond-order process can a bona fide principle of justice be regarded as morallyjustified, in the sense that compliance with it would generally be required.§2 Does our consequentialist account of first-order moral justification actuallysucceed in meaningfully narrowing, in various domains, the range of principles ofjustice that any informed person of good will must regard as morally justified (atleast subject to second-order legitimation)? Or can everyone fit their particularnormative prejudices into this framework? The general answer I will defend in thenext chapter is this. In complex, pluralist societies with at least moderately favorableeconomic prospects, a consequentialist framework requires us to embrace directlyresponsive principles of justice which are broadly liberal-egalitarian in character.Within these limits, however, lawmakers (and indirectly the people who electthem) have morally considerable room for context-dependent judging in theirchoice of principles of justice as well as for first-order rational disagreement. But aconsequentialist framework could usually, I believe, ensure enough commonground for fruitful debate. If someone is dissatisfied with the degree of normativeindeterminacy which a consequentialist framework entails, in the abstract, for theproject of spelling out the demands of justice, I can only say: “Give me a theory ofjustice which is more determinate at the abstract level, and I will give you reasonswhy informed people of good will need not find its acceptance rationally required(or even permitted).”There are some things of which a theory can be had, and others of whichperhaps only an account can be given. A theory of justice could plausibly be thoughtto consist in showing how rational agreement on a highly comprehensive set ofnormative principles for the distribution of the benefits and burdens of social206cooperation can be reached which is sufficiently responsive to patterns ofexpectations and sentiment about justice to enlist enough people’s sense of justicein support of these principles to make them the basis for a stable social consensusabout justice. An account of justice consists in explaining why no such agreementhas been reached or is to be expected in pluralist liberal democracies, except withinvery broad limits, and stating how we can nevertheless implement measures tominimize the destructive force of people’s sense of injustice while, at the sametime, trying to foster gradual convergence on such principles over time by building astable structure of expectations. I have only attempted to accomplish a fraction of thefirst part of such an account here and hardly touched the second part, namely thequestion of what kind of second-order accommodation constrained by whichprocedural rules, is most promising in the context of a particular society and withinvarious domains.We cannot predict, with great precision, which principles of justice will turnout to be preferable in various domains, independent of information about existingpatterns of people’s expectations and sentiment about which just-making criteria areappropriate in this domain. On this account, finding a principled response topeople’s justice-based claims is no longer a matter of erecting a grand structure oftimeless abstractions on the bedrock of pure rationality, but will involve the appliedethicist in sociological field studies or at least require an awareness of thecontributions of social scientists. What people already feel to be just, and hownegatively they would react to a principle that ignored such sentiments, willstrongly influence (but not by itself determine) which principles promote happiness,which principles minimize the need to treat individuals as a means to others’ goals,and which principles it is to our mutual benefit to voluntarily adopt.The recognition of this fact may, of course, itself motivate the pressing ofextravagant claims in the name of justice. The proliferation of blatantly self-servingclaims in the language of justice is a common practical problem for consequentialist207constructivism or any other theory of justice whose principles emerge partly fromcontextual and pragmatic considerations (i.e. all remotely plausible theories). Butthe illegitimacy of such claims is often evident by their lack of having crediblebacking from just-making criteria. Satisfying brazenly phony claims put forward inthe name of justice would generally have strongly negative consequences and will,therefore, be ruled out by the ultimate consequentialist test. In a politicalcommunity, extravagant claims and sentiments cloaked in the language of justicewill be held in check by counter-claims and counter-sentiments, although at a cost ofrising tensions. It is precisely in such situations that a clearer appreciation of thepossibility of rational disagreement about justice may be helpful to motivate as wellas constrain compromise-seeking.2083.4 THE CONFLICT BETWEEN LIBERALS AND NON-LIBERALSThis chapter offers a brief assessment of the scope for rational disagreement aboutthe conceptions of justice offered by liberals and their opponents.3.4.1 Liberal and non-liberal conceptions of justice§1 Two value tensions at the heart of the concept of justice have created suchprofound divisions between substantive conceptions of justice that no credibleattempt to assess the scope for rational disagreement about justice can pass them by.I mean the divisions between liberal and non-liberal theories of justice, and betweenthe theories of libertarian and egalitarian liberals. The literature generated by thesevalue tensions is so vast that I cannot possibly hope to do more than wrestle withsome fundamental issues and to create some presumption in favor of myconclusions.Liberals share the core belief that grown-up, mentally competent individualsare best left to judge for themselves how to live their lives, and that they shouldenjoy as much freedom to live according to their own preferences as is compatiblewith a similar freedom for all. This belief is not the same as believing that somelives are not lived better than others, or that individuals cannot make majormistakes in their quest for the good life. The liberal core belief has been given anumber of formulations, and they are not obviously equivalent in every respect.One can say, with Mill, that people or governments can justifiably interfere with theactions of any person only to prevent harm to others:...the sole end for which mankind are warranted...in interfering...is selfprotection. That the only purpose for which power can rightfully be exercisedover any member...is to prevent harm to others. His own good, eitherphysical or moral, is not a sufficient warrant. He cannot rightfully becompelled.. .because it will be better for him to do so, because it will make himhappier...These are good reasons for remonstrating with him. ..but not forcompelling him. (13)209Or one can say that there is a private sphere of action for each individual into whichthe state must not step, and which the state must protect against interference byothers.1 Or, to use the formulation that has served as the starting point for muchcontemporary debate, one can say that the state should be neutral between diverseideas of the good.2In a helpful attempt to clarify the complexity of the standard liberal creed,McDonald has taken care to distinguish three levels of liberal political thinking: thejustificatory level, the political-thesis level and the contextual level (1991, 220f).These levels of political thinking are roughly analogous to the meta-ethical,normative and applied levels of moral thinking. According to McDonald, the two-part core of the standard liberal creed, at the political-thesis level, is (i) therecognition of rights which “establish normative limits on the use of both state andprivate power for the sake of members of the society in question”(225), and (ii) “theattribution of rights to individuals” (226).While granting that liberals may, in principle, attribute rights to groups aswell as to individuals, McDonald stresses the basic ways in which liberal politicstends to be strongly individualistic. For liberals, the primary role of the state is toprotect the civil and political rights of individuals; the state is to protect these rightswithout enforcing some ideal of what makes a life worthy or a person virtuous; thechoice of how to live is vested with individuals as the locus of intrinsic value andimportance; and individuals must be given a significant range of choices (226).All capsule characterizations of the standard liberal creed are afflicted withmuch the same vexing problems of interpretation. Which freedoms are compatiblewith the same freedoms for all? What constitutes sufficient harm to others to justifyinterference with an individual’s actions? What are the boundaries of the “privatesphere of action” or “the normative limits on the use of both state and private1 Cf. e.g. Narveson (1988).2 Cf. Dworkin, “Liberalism” (1985)210power”? What exactly does state neutrality among diverse ideas of the good amountto? How are conflicts between the rights of different individuals to be mediated?Does our consequentialist moral framework permit us, within the usual range ofindeterminacy, to answer such questions in principled ways, or must we fall back onpurely ad hoc pragmatism? To ask an even more fundamental question: mustconsequentialists necessarily be liberals in pluralist societies? Or are there perhapsforms of non-liberalism with greater promise for helping us progress toward thegoal of achieving human well-being?This chapter can accomplish only a small part of the task of defending whatseem to me reasonable answers to a few of these immensely complex questions.§2 I want to begin by clarifying the relation between liberal politics and viewsabout what constitutes a good or worthwhile life. Liberals are frequently contrastedwith perfectionists. Kymlicka seems to draw such a contrast when he says:Such theories [as Marxist perfectionism] are ‘perfectionist’ because they claimthat certain ways of life constitute human ‘perfection’ (or ‘excellence’), andthat such ways of life should be promoted, while less worthy ways of lifeshould be penalized. This is unlike liberal or libertarian theories, which donot try to encourage any particular way of life, but rather leave individualsfree to use their resources in whatever ways they themselves find mostvaluable. (1990: 187)But contrasting liberals with perfectionists, as if “perfectionist” were equivalent to“non-liberal”, seems to conflate two cross-cutting distinctions. A term is clearlyneeded for a certain view within ethics, the view that some ways of life are knownto be better or worthier than others (I shall use the terms “good” and “worthy”interchangeably here). And another term is needed for a certain view of politics, theview that the state should protect a private sphere of individual action withinwhich individuals can pursue whatever ways of life they prefer, whether worthy inanyone’s opinion or not. If we adopt “perfectionism” as the term to refer to the firstview, and “liberalism” as the term to refer to the second view, then the category of211perfectionists and that of liberals are not mutually exclusive. It is an open questionwhether or not someone who believes she knows which ways of life are worthierthan others will espouse the politics of protecting a private sphere of action.There is, however, a problem with using the term “perfectionists’ simply forall those who believe that we can judge some forms of life to be better or worthierthan others while reserving the term “non-perfectionists” only for those who denythat any such assessment can in principle be made. Used in this way, the distinctionbetween “non-perfectionist” and “perfectionist” would simply duplicate thedistinction between “nihilist” and “non-nihilist”. Even if there are people who canremain sincerely nihilistic or even agnostic about whether some lives go better thanothers, they are of no interest to me here. So if we do not want to define“perfectionism” in such a way that all non-perfectionists become essentiallynihilists, I think we must clarify the distinction between perfectionists and non-perfectionists roughly as follows:Perfectionists are those who think that there are objective standards of thegood human life - i.e. standards which are completely indifferent to facts about whatways of living people actually prefer. As I shall use this term, perfectionists alsobelieve that these objective standards usually allow them to determine or predictthe goodness of a person’s life with sufficient confidence and precision so thatcoercing people into leading good lives becomes viable. Perfectionists who thinkthat there exist objective standards of the good human life, but that they are sovague, epistemically elusive and difficult to apply that we can’t generally tellwhether a person is or isn’t leading a good life are, for practical purposes, hardlydistinguishable from non-perfectionists. Different types of perfectionists will, ofcourse, be more or less pluralistic in what sorts of values they include amongconstituents of a good human life.Non-perfectionists, as I shall use the term, are those who think that there areno objective standards of the good human life (in the sense defined above), and that212at least one necessary condition for what sorts of values make a mature human lifegood always is that the person living it likes or chooses to live according to thesevalues. Although non-perfectionists don’t deny that the goodness of a person’s lifecan generally be determined by observation, they do not believe that currently (inthe absence of a highly accurate predictive psychology) we can tell in advance withgreat confidence and precision whether a particular way of living will constitute thegood life for a given person.Again in the interest of avoiding empty categories, I must further distinguishradical from moderate non-perfectionists: those who think that knowing a person’spreferences are a necessary and sufficient condition for determining what makesthat person’s life good, and those who think that such knowledge is only a necessarycondition. For radical non-perfectionists, the question of what is good or worthy fora person reduces exclusively to an empirical question about what this person likesor prefers. For moderate non-perfectionists, some general values whose presencegenerally tends to make a human life better can in fact be known, and persons wholike or prefer the opposite of these values can, in fact, be mistaken about their owngood. I feel compelled to make this somewhat awkward distinction because, again, Idoubt that there are in fact radical non-perfectionists among sane people who havethought the issue through. A radical non-perfectionist would have to deny thatanyone’s life could ever go better than anyone else’s, no matter how uninformed orself-destructive the choice of someone’s values, because they are after all the valueshe prefers. Radical non-perfectionism would again seem to collapse into a sort ofnihilism or at least a value subjectivism which, in the face of our knowledge of atleast some of the factors which determine people’s preferences, strikes me as highlyimplausible. In what follows I shall, therefore, concern myself only with moderatenon-perfectionists: those who think that the goodness of a person’s life dependscrucially on what is liked or preferred by that person and cannot be predicted withgreat confidence and precision for any given person, even though some very213general values whose presence tends to make human life better can in fact beknown.What is the relationship between liberals and perfectionists, in the sensesdefined? The most that can be said, I think, is (1) that some who espouse liberalpolitics are perfectionists and some are not, (2) that non-perfectionists need notnecessarily espouse liberal politics, and (3) that perfectionists who espouse liberalpolitics can do so for very different reasons.(1) A non-perfectionist who espouses liberal politics may do so for one ofthree basic types of reasons: because it provides the most pragmatic response to thebrute fact of pluralism about the human good; because it is the most neutral”ground in the face of a perfectly rational diversity of ideas about the good; or becauseit is empirically most likely to lead to social conditions which enable individuals todiscover their own good. (I shall examine these reasons below.)(2) But non-perfectionists need not necessarily espouse liberal politics. Thereis nothing strictly inconsistent about holding that what gives a life its value ishighly individuated and at the same time believing that the state need not protect aprivate sphere of action within which individuals can do whatever they prefer. Thenon-perfectionist thesis is only one of many compatible theses within ethics aperson may hold. A non-perfectionist might, quite consistently, also hold the thesisthat there is nothing more to morality than verbal posturing or venting ofemotions, and that anything whatever is permitted. And such a view clearly neednot lead to liberal politics.(3) A perfectionist who espouses liberal politics may do so for at least tworeasons. All perfectionists must by definition think that there are objectivestandards of the good life by which they can determine the worthiness of someoneTlife with considerable precision and confidence. What I shall call type-Aperfectionists espouse liberal politics because they think that only the selfdetermined life is worthy, and that a liberal social order is most likely to permit self-214determined ways of life. Any perfectionist who affirms self-determination as thecentral ingredient of a worthy life must also affirm a whole web of political valuesthat can plausibly be considered conducive to creating favorable conditions for self-determination such as, for example, a liberal education, access to economic meansfor the realization of one’s choices, or the creation of valuable opportunities for theexercise of self-determination. Type-A perfectionists, by the very nature of theirparticular idea of what makes a human life worthy, will necessarily recognize amuch broader diversity of equally worthy ways of living than other types ofperfectionists. But self-determination is clearly a value that presupposes the use ofstate power in order to create favorable conditions both for the cultivation of self-determining character types and for the effective exercise of self-determination.(Examples of contemporary type-A perfectionists are Joseph Raz or Thomas Hurka.)Type-B perfectionists do not regard self-determination as necessary or evenimportant for a worthy life, but nevertheless espouse liberal politics (in pluralisticsocieties) because they think either that any attempt to use state power would simplyunleash a destructive struggle for control of the state among opposing groups or thatusing state power to enforce the worthiest way of life would be downrightcounterproductive to converting people to such a way of life.3Both type-A and type-B perfectionists could intelligibly be called liberalperfectionists or perfectionist liberals. Type-C perfectionists are a residual category ofnon-liberal perfectionists which includes all and only those who think (i) that theyare in possession of objective standards of the good life by which they can determinethe worthiness of almost any persons life with great precision and confidence;(ii) that self-determination is not necessary for a worthy life; and (iii) that statepower should be used to enforce worthy ways of living for everyone. It may beupsetting to those who want to see a simple contrast between liberals and3 These two arguments are forcefully advanced by Locke in defense of freedom of religion (see A LetterConcerning Toleration). See Susan Mendus for an assessment of the comparative merits of Locke’s andMill’s arguments for liberal politics (Toleration and the Limits of Liberalism, 1989).215perfectionists, and who define liberalism in terms of state neutrality betweendiverse ideas of the good, that by no means all perfectionists advocate the use ofstate power to enforce favorable conditions for their idea of the good; while thoseliberals who justify liberal politics in terms of the value of self-determination clearlydo advocate the use of state power to enforce favorable conditions for their centralidea of the good.But even most hardcore type-C perfectionists who want the state to imposetheir particular idea of the good on everyone would probably agree that there shouldbe some a priori restrictions on what the state can do, such as a prohibition onconcentration camps for the large-scale extermination of dissenters. Moreover,many type-B perfectionists may also value self-determination to a degree, but onlyas one value among others, so that it may at times become a difficult judgment forthem what it is more important- to respect someone’s right to self-determination orto prevent her from making a major mistake about the good life.The contrast between those who espouse liberal politics and those who claimto be able to determine objective standards of a good life for each person is furthercomplicated by the fact that even the most non-perfectionist liberal would certainlywant to keep certain antisocial tendencies in check, while even the most ferventtype-C perfectionist would hardly want to control every trivial detail of individuals’lives. Both those who espouse liberal politics and and those who don’t will have tomake judgments about which behaviors are antisocial enough to be made illegal ortoo trivial to be bothered with at all. Kymlicka seems again surprisingly sweepingwhen he asks,Should we therefore be perfectionists, supporting state policies whichdiscourage trivial activities to which people are mistakenly attached?”U990: 203)It may well turn out that many behaviors which liberals tolerate are trivial enoughin the eyes of many non-liberal perfectionists to be tolerated by them as well, whilebehaviors which many non-liberal perfectionists want to make illegal are judged to216be antisocial enough by most liberals to be declared illegal by them as well. Neitherliberals nor perfectionists are united among themselves about what behaviors canbe tolerated without causing unacceptable harm to identifiable others or to society’svital long-term interests in civil peace, security, prosperity, cultural flourishing anda livable natural environment.§3 Contrary to what is sometimes implied, then, the two categories of“liberalism” and “perfectionism” are best regarded as intersecting in complex waysrather than as sharply disjoint. Only because of an unfortunate conflation of crosscutting distinctions has it been possible for some perfectionists to accuse liberals ingeneral of being skeptics about the good and for some liberals to accuse perfectionistsin general of being totalitarians by definition. Drawing clear distinctions about thedifferent reasons for espousing liberal politics, including perfectionist reasons, canavoid a great deal of sterile disagreement. This is not to deny that manyperfectionists who see a realistic chance for achieving control of state power will, asa rule, want to use state power to get the obstinate to lead worthy lives. But liberals,whether they be type-A perfectionists or non-perfectionists, are certainly noexception to the rule: they, too, propose to use state power to enforce their ideal ofthe best social order. The obvious question is whether liberals can, withoutinconsistency, consider the choice of their favored value — i.e. self-determination orletting people live by their own preferences— as sufficiently special to entitle themto use state power while denying type-C perfectionists the same right. I shall addressthis question below. The issue I want to address next is how ideas about the good lifeand about the best social order are connected with conceptions of justice.§4 Having distinguished perfectionist and non-perfectionist views about thegood life, and liberal and non-liberal views about the best social order, we now face a217choice: on which of these two distinctions should we base our fundamentaldistinction between different substantive conceptions of justice? This choice is nontrivial because differences marked by our fundamental distinctions will channel ourthoughts and disagreements in profound ways. The distinctions we draw betweensubstantive conceptions of justice can either mark the difference between varioustypes of views about the human good that can ultimately be advanced to justifydifferent principles of justice. Or they can mark the difference between various typesof views about the best social order which can also be advanced to justify differentprinciples of justice.4If we choose the first alternative, our distinctions between different basicconceptions of justice will mark such differences as, for example, that between non-perfectionists, type-A perfectionists and type-B perfectionists- all of whom may inpractice, at least in a pluralist democracy, converge on a broadly liberal social orderand on similar substantive ideas about justice. But not only will very different basictypes of views about the good life sometimes be consistent with the samesubstantive principle of justice, it will also be the case that subtypes of the same basictype of view about the good life will often justify very different substantiveprinciples of justice. The relation between views about the good life and principlesof justice would seem to be far more intractable than the relation between viewsabout the best social order and such principles. We could, for example, distinguishbetween an essentially liberal conception of justice and various types of non-liberalconceptions supported by those perfectionists who want to use state power toadvance some ideal of the good incompatible with self-determination. Thistaxonomy of different conceptions of justice assumes, of course, that there is‘ Or they can mark some other interesting difference between proposed principles of justice, such aswhether they are directly or indirectly responsive, intuitionistic or non-intuitionistic. Whichdistinctions we draw depends, of course, on our purposes. But sometimes we seem none too clear about ourpurposes. It is probably an understatement to say that confusion reigns in political philosophy abouthow to group different principles of justice into conceptions of essentially the same type.218something sufficiently special and cohesive about liberal politics to justify ourdistinguishing liberal conceptions of justice from the immensely wide and diverserange of all the others. That there is something that makes liberalism special in thisway is vigorously disputed by some non-liberals like Alistair Maclntyre (cf. 345).I will argue below that, indeed, liberal politics is special in a way that justifiesus in drawing a fundamental distinction between substantively liberal conceptionsof justice and the rest. But first I want to draw my overview of basic distinctionsbetween different conceptions of justice, and draw them rather starkly.Understanding distinctions in their unqualified starkness is often necessary, Ibelieve, before one can intelligibly blur them - especially as we enter the conceptualmaze of liberalism, its alternatives, and its internal divisions. My categories are not,of course, meant to be exhaustive but simply useful for entering contemporarydebates.I. Liberal conceptions of justice: “Social goods should be distributed in a way thatallows individuals the freedom to live as they prefer, while promoting thewidest range of valuable opportunities for self-determination.”A. Libertarian liberalismB. Egalitarian liberalism (often called “welfare” liberalism)II. Non-liberal conceptions of justice: “Social goods should be distributed in away that eliminates unworthy ways of living and promotes the greatestpossible flourishing of worthy ways, regardless of the extent to which doing soconflicts with people’s preferences.”A. Communitarian non-liberalismB. Utopian non-liberalismC. Religious non-liberalismEach of these generic principles characterizing liberal and non-liberalconceptions of justice does, admittedly, gloss over a multitude of substantive219differences within each camp.5 Liberals can be usefully grouped into a number ofopposing factions (I shall briefly consider the conflict between libertarian andegalitarian liberals). Non-liberals can also be grouped into many different types suchas, for example, into communitarian, utopian and religious non-liberals (which willfurther subdivide into countless antagonistic factions).Of course, communitarians, utopians or religious believers may consistentlysupport or oppose liberal politics. Broadly speaking, communitarians believe thatcommunity is an essential constitutent of the good life, that a person’s self isthrough and through the product of a particular cultural community and can realizeitself only through participation in the life of that community, and that our socialobligations, commitments and attachments are are not simply a matter of voluntarychoice, but deeply part of who we are. Maclntrye, Sandel and Charles Taylor areusually cited as leading contemporary communitarians, even though a great deal ofwhat they say seems to me to have been well articulated in F.H.Bradley’s EthicalStudies. Those communitarians who oppose liberal politics seem to be concernedwith strengthening individuals’ identification with their social roles and withmaintaining whatever received traditions are most likely to sustain a close-knitnetwork of warm human attachments and a sense of belonging within acommunity, even at the expense of denying individuals such core liberal rights asthe right to freedom of religion, expression, or association.6A prominent sub-type5 Examples of egalitarian liberals are John Rawls, Kai Nielsen, Derek Phillips, Brian Barry, DavidBraybrooke, Ronald Dworkin, James Sterba, as well as R.H. Tawney and L.T. Hobhouse. Examples oflibertarian liberals are Robert Nozick, David Gauthier, Jan Narveson, and Friedrich Hajek. A possibleexample of a communitarian perfectionists is Alistair Maclntyre. Michael Walzer, William Gaiston,Michael Sandel or Will Kymlicka are best seen as liberals who try to make room for communitarianvalues within liberalism. Examples of utopian perfectionists are Plato, Marx, Thomas Moore or B.F.Skinner.6 There can be communitarian liberals, or liberal communitarians. Cf. Will Kymlicka, Liberalism,Community, and Culture (1989), for an outstanding attempt to accommodate communitarian concerns in aprincipled way within a basically liberal vision of the best social order. Allen Buchanan has alsoargued that liberalism can make room for the good elements of communitarianism while avoiding itstotalitarian elements (1989). An excellent discussion of the complexities associated with the“collective rights” solution to accommodating the value of community within liberalism is to be foundThe Canadian Journal of Law and Jurisprudence, Vol IV, No 2, July 1991, which is devoted entirely tothe topic of collective rights.220of non-liberal communitarians are classical conservatives, such as Roger Scruton.For non-liberal communitarians, principles of justice will have this generalcharacter: “Distribute social goods according to whatever distributions willstrengthen community bonds and make individuals identify themselves asmembers of a shared culture with a common good- no matter whether suchpolicies interfere with most individuals’ preferred life choices.”Utopian non-liberals believe that promoting human excellence consists inrevamping society according to some blueprint of ideal social arrangements, ifnecessary without the informed consent of the majority of the population. Statedbluntly, utopian principles of justice will have this character: “Distribute socialgoods according to whatever my particular blueprint of the ideal society specifies.”Religious non-liberals believe that promoting human excellence consists infollowing the moral code associated with what each considers the true religion.Their principles of justice will have this character: “Distribute social goods accordingto whatever principles are laid down in the Holy Scripture of our particularreligion.” Whereas, for communitarian non-liberals, religion may play animportant role as one means of preserving community bonds, for religious non-liberals, spreading the true religion is the ultimate value, even where it destroyscommunity bonds. It needs stressing, however, that many religious believers, justlike many communitarians and utopians, consistently can, and in fact do, supportliberal politics rather than the use of state power to enforce their particular ideal ofthe good.This, then, is a starkly drawn matrix of how I see the basic divisions betweenvarious substantive conceptions of justice. The variety of ideas about justiceappealed to in real-world debates is bewildering; the number and complexity ofarguments offered in support of conflicting ideas is often overwhelming. We nowhave enough of a synoptic perspective, I think, to sketch the outlines of the strong221consequentialist case that can be made for a substantively liberal conception of socialjustice.3.4.2 On the nature of human well-being§1 If those perfectionists are correct who hold that we can know withconfidence both what is good for each person and that this good has little or nothingto do with self-determination, then obviously the case for liberalism is weak. In thatcase the only two reasons I can see for espousing liberal politics are that attempts toenforce the worthiest way of living will unleash a destructive struggle for control ofthe state among opposing groups in pluralistic societies or will often be downrightself-defeating as attempts to convert people to such a way of living.But do perfectionists who hold that we can know with confidence what isgood for each person and that this good has little or nothing to do with self-determination have a plausible case at all? Is there perhaps a possibility of lookingobjectively at some basic psychological facts about human nature and the humancondition and drawing some conclusions about what makes a human life good orworthy? I think there is.What could possibly count as making a human life good or worthy? At themost fundamental level, only two interestingly different prima facie plausibleanswers seem to me possible. Either we accept one of many competing religious ormystical accounts - the view, for example, that we are God’s creatures, here on earthto earn our place in heaven through obedience to God’s will; or the view that welive in a morally ordered universe whose moral ordering principles can bediscovered in much the same way as its natural ordering principles. Or we are in amorally indifferent, uncaring universe governed by blind material forces, throwninto existence by the mechanistic process of evolution, and all of us must reflect onour own experience and try to discover values that seem most likely to promotewhatever well-being the constraints of our situation permit. I think, though I shall222not try to argue this here, that a sober look at the evidence, as opposed to wishfulthinking, shows belief in the second alternative far more plausible than belief in thefirst- at least insofar as belief in the first alternative involves specific claims aboutthe nature of God or the nature of this independent moral order.A number of questions arise. Is at least some degree of rational consensus onthe range of plausible conceptions of personal well-being possible? Assuming theanswer is yes, what can we say about this range of plausible conceptions of wellbeing for different individuals? What defines the limits of this range? Are theresome properties which all plausible forms of personal well-being have in common?Such questions, I believe, must at least make sense to any grown-up, mentallycompetent individuals who approaches them with an open mind.After due reflection on the answers offered by others, each of us will have tointrospect his or her own deepest understanding to answer these questions. Whatultimately matters for a person’s well-being? Having abundant resources of a certainkind, such as Rawis’ primary goods? Or being in that kind of hedonic state whichindividuals in this state identify with well-being? I think the most plausibleconception of well-being is not captured by either alternative. Well-being cannotconsist in having abundant resources, because no matter how abundant one’sresources, they are not sufficient for well-being unless a happy state of mind isactually achieved through them. Beyond a certain minimum, additional resourcesare not necessary for well-being; an abundance of possessions may, in fact, detractfrom a happy state of mind. Nor can well-being simply be a pleasant hedonic state.A happy state of mind caused by drug-induced delusions or by the life-longstimulation of one’s cerebral pleasure spots, while one lies strapped down on a bed,doesn’t plausibly capture what ultimately matters to a person’s well-being.77 Nozick’s example of the “experience machine” vividly makes this point: cf. Anarchy, State, andUtopia, 4245. It is, of course, conceivable that some thoroughgoing hedonists will opt for a life hookedup to the Experience Machine, and most of us would probably like to plug into it intermittently.223What, ultimately, matters to a person? Let me sketch my own rough idea.Others, in the light of their own introspective understanding of what theyfundamentally care about, may well find reasons to disagree with it: I am a friend ofrational disagreements. Apart from satisfying on&s biologically-determined needs,what constitutes well-being, I think, is working toward and achieving the goals onehas chosen from a range of meaningful options, and experiencing a happy state ofmind as a result of doing so. I believe that almost always, for a happy state to beexperienced as a result of doing what one values, one must expend some degree ofreal effort, even though the happiness achieved is not proportional to effortexpended. Not everything a person values doing and actually does will result in herexperiencing a happy state of mind. A happy state of mind is certainly not the onlything a person can intelligibly value; in fact, a happy state of mind is generally abyproduct of sincerely valuing something else for its own sake and making progressin promoting this value. It is quite possible for people not to value their ownhappiness, or to sacrifice it for the sake of other people or other values, or toknowingly act contrary to their well-being out of weakness of will, and all toofamiliar for them to value their long-term well-being less highly than intenseimmediate enjoyments.The most important point is this. What is conducive to a persontswell-beingis something which that person can be mistaken about, but at the same timesomething which may differ considerably from one individual to the next. The basictypes of things that human beings, such as they are, can actually value in this world,such as it is, with the result of experiencing a happy state of mind which is notoutweighed by subsequent unhappiness, add up to a lengthy, varied, but quite finitelist. Although this list will get less universal and more controversial the morespecific it gets, it will almost certainly, for most people, include items like securingsuch things as food, shelter, and other requirements for basic physical and socialfunctioning, instinctual enjoyment, affectionate human relationships, self-esteem, a224meaningful social role or other ways of getting a sense of social belonging, as well aspersonal freedoms. Out of varying amounts of these and other values, eachindividual fashions his own mix of values to live by, although there will of coursebe a heavy cultural ingredient in any such mix.What gets something onto this list of values which are constitutive ofhuman well-being? Nothing more mysterious, than that we can recognize it asfrequently or universally necessary or conducive to people’s experiencing their livesas good. Some degree of convergence is possible, I believe, between the reflectiveequilibria of informed people of good will on what should go onto this list. Such alist of constitutents of human well-being, I think, can furnish the basis for anappreciation of the diversity of possible forms of personal well-being as well as anunderstanding of plausible limits to such forms. In cashing out the importantnotion of “plausible limits on the range of values which can conduce to personalwell-being”, a broad consensus is possible that certain values will clearly fall insidethe range (e.g. friendship) and certain values will clearly fall outside (e.g. heroinaddiction), while others may well be matters for rational disagreement.But I take it to be a fundamental fact about the human condition, obvious toany grown-up, intellectually competent person with an open mind (and perhaps acertain minimum of education), that there exists a diversity of values for humanbeings which no one life can encompass and which can be permutated in almostinfinitely many ways. And this fact, I believe, furnishes a powerful consequentialistargument for liberal politics - an argument which is neither identical to the purepragmatist’s or political realist’s argument based on the avantage of mutualaccommodation nor to the perfectionist argument based on self-determination (seebelow).2253.4.3 On the nature and justification of liberal politics§1 Let me try to characterize the nature of liberal politics more clearly beforeconsiderating justifications for it. I follow the current practice of taking Dworkin’sdefinition of liberalism in terms of the principle of government neutrality amongpeople’s diverse ideas of the good life as my point of departure:...government must be neutral on what might be called the question of thegood life. ..The first theory of equality supposes that political decisions mustbe, so far as is possible, independent of any particular conception of the goodlife, or of what gives value to life. (1985: 191).In the years since Dworkin’s 1978 paper “Liberalism” tried to rally liberals aroundthis ‘liberal neutrality” principle, competing interpretations of liberal neutralityhave emerged. What does it mean for liberals to claim that they are neutral betweendifferent ideals of the good - and how can liberal neutrality be justified?There seem to be at least three basic ways in which liberal neutrality might beunderstood. The terminology here is far from standardized, and I want to adoptreasonably descriptive labels that make sense to me. I want to distinguish“neutrality of aim” from “neutrality of effect”. Neutrality of effect can be furthersubdivided into what I shall call “non-differential” and “differential” neutrality.Which kinds of neutrality can we have?Neutrality of aim requires that the official reasons for justifying the state’spolicies should never include the promotion of some group’s ideas of the good lifeover that of other (legitimate) groups, even though most policies will, of course,impact quite differently on different groups. Non-differential neutrality of effectrequires that the policies of the state should be neutral in their impact on differentlegitimate groups. All legitimate groups, that is, should be helped or hinderedequally, in some plausible sense, in allowing each individual member of such agroup to pursue her idea of the good life or in allowing the group as a whole to gainadherents to its values. (These twin objectives may well diverge.) Differentialneutrality of effect requires that no legitimate groups be effectively prevented from226pursuing their ideas of a good life, although non-coercive means of persuasion andaffirmative action programs may be used to correct specific undesirable inequalitiesby giving differential encouragement to disadvantaged or underrepresented groupsover others. Difficult judgment calls will obviously be required to balanceaffirmative action for the disadvantaged or underrepresented without denying orseriously impairing other groups’ effective pursuit of their ideas of a good life.Obviously, all liberals must define some standard of legitimacy for groups:groups which value murder, rape, theft and the like are ruled out. And all liberalsmust draw a workable distinction between purely personal values (between whichthe state should be neutral), such as someone’s religion, career, sexual preferences,and a multitude of lifestyle choices; and political values (between which the statecannot be neutral), such as liberal neutrality itself, peace, civility, tolerance, politicalfreedoms and civil liberties, a range of social institutions like the market anddistributive goods which serve as flexible means to diverse ends, and someprinciples of justice according to which these things may be distributed.8Assuming that the distinctions between legitimate and illegitimate groupsand between personal and political values can be drawn without begging crucialquestions, which of the above three basic interpretation of liberal neutrality shouldliberals be interested in justifying? As a useful criterion for government action,neutrality of aim would seem to be too weak and too strong at the same time, atleast for consequentialist liberals. Too weak, because it should often count decisivelyagainst a policy that it is foreseeably disastrous for some group, even if the stat&sofficial justification make no reference whatever to its effect on any particulargroup. Too strong, because it is hard to see how most policies could avoid havingthe promotion of the good of some group (disadvantaged or otherwise) as their goal.Endorsing neutrality of aim would clearly open liberals up to the charge ofhypocrisy.8 See Rawis (1988) for a lucid summary of political values liberals must accept.227Non-differential neutrality of effect would also seem to be too strong, becauseit is impossible to live up to and because it would effectively paralyze thegovernment at every step. Rawis has this to say about non-differential neutrality ofeffect, which he calls ‘neutrality of effect or influence” and defines as the position“that the state is not to do anything that makes it more likely that individuals willaccept any particular conception rather than another unless steps are taken to cancel,or to compensate for, the effects of policies which do this” (262):..it is surely impossible for the basic structure of a just constitutional regimenot to have important effects and influences on which comprehensivedoctrines endure and gain adherents over time, and it is futile to try tocounteract these effects and influences, or even to ascertain for politicalpurposes how deep and pervasive they are. We must accept the facts ofcommon-sense political sociology. (1988: 262)These and similar considerations seem to me decisive against non-differentialneutrality of effect. Of our three prima facie options for interpreting liberalneutrality, differential neutrality of effect seems to come closest to a viable politicalvalue. Rawls defines this position as meaningthat the state is to ensure for all citizens equal opportunity to advance any[permissible] conception of the good they freely affirm. (1988: 262)But this definition, too, is in obvious need of both clarification and qualification.And how a liberal clarifies and qualifies it will depend very much on the reasonsthat made him espouse liberal politics in the first place. Raz, for example, who baseshis liberalism on the perfectionist value of self-determination, is quite explicit inadvocating policies with substantially differential effects on different groups:...a government whose responsibility is to promote the autonomy of itscitizens is entitled to redistribute resources, to provide public goods and toengage in the provision of other services on a compulsory basis, provided itslaws merely reflect and make concrete autonomy-based duties of itscitizens... .Autonomous life is valuable only if it is spent in the pursuit ofacceptable and valuable projects and relationships. The autonomy principlepermits and even requires governments to create morally valuableopportunities, and to eliminate repugnant ones....Perfectionist goals need notbe pursued by the use of coercion. A government which subsidizes certain228activities, rewards their pursuit, and advertises their availability encouragesthose activities without use of coercion. (417)A host of questions now confront liberals who espouse differential neutrality ofeffect, Is the idea of “equal opportunity’ to be given a libertarian or some form ofegalitarian interpretation? How is the line between persuasion and coercion to bedrawn? When does affirmative action for some group become unjustdiscrimination against another? Which policies should liberals prefer concerningcultural minorities, melting-pot policies or multi-cultural policies? How can liberalsdraw a principled line between legitimate and non-legitimate groups without beingaccused of excluding socially benign ways of life and including socially destructiveones?9 I cannot pursue these complex questions here, but I consider it fairly obviousthat informed people of good will, even if they share a broadly consequentialistmoral framework, can be in immense disagreement about how exactly to answerthem.§2 If one is committed to some plausible form of liberal neutrality, how canone justify such a commitment? The consequentialist framework I have outlinedwould seem to allow at least four connected lines of justification to converge onliberal neutrality. Liberal neutrality can be defended (1) as providing the politicalpragmatist with the most promising way to avert civil strife in the face of pluralismabout the good; (2) as being the most coherent position or most “neutral ground” inthe face of a perfectly rational diversity of ideas of the good; (3) as being empiricallymost likely to lead to the sort of social conditions which enable the greatest numberof individuals to achieve their interest in personal well-being; and (4) as beingnecessary for individuals to achieve one particularly important good, namely self-9 Discussions of the harm principle are numerous in the literature on liberalism. For recent examples,see Susan Mendus, Toleration and the Limits of Liberalism (1989); Joseph Raz, The Morality of Freedom(1986); and Lloyd Weinreb, Natural Laze and Justice (1987). For a discussion of different effects of stateintervention on the affairs of groups - and a defense of permitting certain “self-constituting” groups alarge measure of autonomy even at the expense of the autonomy of individuals within such groups - seeMcDonald, “Should Communities Have Rights? Reflections on Liberal Individualism”(1991).229determination (although appealing to the personal value of self-determination tojustify the political value of liberal neutrality seems, on the face of it, blatantlyinconsistent). Let me briefly consider these lines of justification.(1) Since some degree of political accommodation of diverse ideas of the goodis essential to securing peace in strongly pluralist societies, since social peace isimportant to the pursuit of almost any idea of the good, and since liberal neutralitycan be plausibly construed as an attractive form of political accommodation, liberalpolitics would clearly seem to be a pragmatically rational response to the fact ofpluralism. But it has at least three limitations. First, this pragmatic defense ofliberalism in strongly pluralistic societies will, of course, turn into an equallypragmatic defense of non-liberalism in less pluralistic societies and possibly justifyrepression of minorities. Second, this defense fails to give non-liberals a deeprationale for why they should not try to topple liberal institutions when they see anopportunity to do so. And third, it is hard to see what should make hardnosedpolitical pragmatists endorse liberal principles rather than naked power politics.Raz illustrates the limitations of any attempt to defend liberalism from thepoint of view of the pure pragmatist. He describes an interesting liberal-sounding“perfectionist” approach to establishing basic social arrangements that promises tocircumvent the objection that non-liberal perfectionism must necessarily sufferfrom excessive intolerance. Such a perfectionist “doctrine of justice”, he says,will fulfill the social role of justice, but it will be agreed upon by a process ofreasoning quite different from [Rawls’]. First, different ideals of the good farfrom being excluded will form the starting points of the argument about adoctrine of justice. Second, and as a result, supporters of different conceptionsof the good will follow different routes in arguing for the doctrine of justice.There will be unanimity in the conclusion but (given the different startingpoints) no unanimity on the route to it. Third, the common feature of mostroutes will be the reliance on a rational reconstruction of a process ofbargaining by which the common overriding goal of reaching an agreementleads the parties to compromise by accepting a less than perfect doctrine as theoptimally realizable second best. (129)230Raz imagines supporters of different forms of perfectionism to come together in acontractarian bargaining session, and to negotiate agreement on a set of principleswhich are the most favorable “second-best principles possible from the perspectiveof each group. Whereas Rawls imagines his bargainers to be free and equal personsbehind a thick veil of ignorance, Raz imposes no constraint at all on hisperfectionist bargainers except the “overriding goal of reaching an agreement”. (Heassumes that the bargainers know that there is sufficiently widespread disagreementabout ideas of the good in their society that no one such idea can be implemented. Itis not clear whether he also assumes equal bargaining power or threat advantage,but presumably not.)The outcome of such negotiations would then seem to be a complex functionof the particular perfectionist starting points, the range of diversity of perfectionistbeliefs, and relative threat advantages. If the social role of a doctrine of justice isconceived purely in the political pragmatist’s terms, Raz concludes that one doesnot require Rawls’ “veil-of-ignorance”. Individuals will reach agreement on a set ofdistributive principles, even if they know how they differ in their ideas of the good,“provided they are united in ranking the need for a public conception of justice veryhighly in their order or priorities.”(128) Nevertheless, mere agreement on principlesthat fulfill the social role of justice cannot be enough to make them morallyacceptable. “For reasons which are fairly obvious”, Raz says, “this procedure ofbargaining may in some societies lead to the endorsement of highly wickedprinciples as that society’s doctrine of justice”(129). What he must have in mindhere is that justice cannot simply be equated with whatever agreement on principlesof social coexistence can be reached, presumably because of unequal bargainingpower may allow some fairly pernicious types of perfectionism to impose their ideasof the good on the rest of society. He is right, I think, that we cannot hope to supportliberal politics very strongly in terms of a purely pragmatic modus vivendi.231(2) But the crucial point is not simply that there exists pluralism about thegood, but that there is no one correct way to live for everyone. Even for any givenperson there is always a broad range of possible ways of life that would have beenequally good. As we concluded above, it seems to be a fundamental part of thehuman condition that there is a diversity of values for human beings which no onelife can encompass and which can be permutated in almost infinitely many ways.Although we cannot create favorable social conditions for all valuable ideas of thegood to flourish simultaneously, our social arrangements can at least enableindividuals to explore a wide diversity of values and to seek out their own versionsof the good life. If this value diversity is widely appreciated as arising from the factthat the constituents of human well-being are naturally diverse and is not equatedwith confusion about the good, and if social institutions can be so designed to makeroom for a wide diversity of values to flourish, then those institutions should enjoywidespread support from people with many diverse ideas of the good.This, I think, is essentially the defense of liberalism adopted in Rawls’ morerecent writings, as well as in those of Stout and Larmore, although they callthemselves (misleadingly) pragmatistsJ° Stout writes,We need not agree on all matters of moral importance to agree on many, andwhere our judgments happen to coincide, we need not reach them for thesame reasons. The judgments we share about the achievable good are ampleenough to justify our practice of ascribing rights. They allow us to offer apragmatic defense of certain liberal arrangements and concepts as justified,here and now, given that we can not presuppose, nor bring about byacceptable means, a more nearly perfect meeting of minds. (226)Larmore has vigorously argued that liberal neutrality can and must be justified in away that is itself Hneutra1 in some sense, rather than based on any suchperfectionist value as autonomy, although liberals may quite consistently hold suchvalues in their private lives (50f, 73f).10 Cf. Jeffry Stout, Ethics after Babel (1988); Charies Larmore, Patterns of Moral Complexity(1987);and Rawls,”The Priority of Right and Ideas of the Good”(1988).232In order to take up a neutral position by abstracting from our substantialconception of the good life, we are not. ..repudiating that conception orlessening our attachment to it. ...But we are declining to appeal to it for aparticular purpose- that of devising common political principles. Neutralityis simply a means of accommodation. It is a stance that we adopt in order tosolve a specific problem to which our various commitments give rise, and soit is not a stance that expresses our full understanding of our purposes. Itestablishes a modus vivendi between persons whose ultimate ideals do notcoincide.A number of different considerations move us to take up such a stance.There is, first of all, a norm of rational conversation....that requires that if wewant to solve some problem but encounter disagreement about how to do so,we should retreat to neutral ground. Then, there are the reasons we have forwanting to converse with others about what political principles to adopt.Besides a desire for civil peace and sympathy for those whose ideals aresimilar to ours, the norm of equal respect demands that we explain ourproposals to those whom they will affect. To these reasons may be added thelikelihood that our pursuit of some substantial ideal of the good life will beprotected if others, with different ideals, also agree that common politicalprinciples must be sought from neutral ground. (75)Dispute has centered on whether appeals to norms of rational conversation and“equal respect” are, in fact, relevantly value-neutral; whether a retreat to sharedvalues can, by itself, provide sufficient ground for workable social policies; whetherseeking such shared values will only lead to widespread but not unanimousagreement and therefore leave certain groups out; and whether policies based onwhatever shared values there are will differentially affect people who attachdifferent degrees of importance to these values and will, therefore, in a sense benon-neutral.11 Assessing these objections raises complex matters. In spite of allthese complications, I think the fact remains that a political system that strives toaccommodate a wide diversity of personal ideals can reasonably claim toapproximate neutrality, in the relevant sense, more closely than a political systemthat does not.(3) The third cluster of justifications for liberal neutrality are justificationswhich consider liberal neutrality as most likely to further the interest of individuals11 Cf. Richard Arneson, “Neutrality and Utility”(1990); Will Kymlicka, “Liberal Individualism andLiberal Neutrality”(1989); and Robert E. Goodin and Andrew Reeve, eds, Liberal Neutrality (1989).233in their own well-being. To be sure, individuals make mistakes, but history showsthat entrusting corruptible governments with the authority to coerce people fortheir own good harbors far greater dangers. Allowing individuals to engage in“experiments in livingt’ is our best promise for creating a rich cultural life,economic prosperity and other social conditions conducive to reasonably good livesfor most.On the other hand, it has been claimed that liberalism sacrifices many of themost important forms of human excellence for a chaotic diversity of inferiorforms.12 A marketplace of ideas of the good will no doubt allow some inherentlyoffensive or oppressive values to win a large following, which necessarily makesthe social climate inhospitable for others, destroys community bonds, and promotesthe kind of mindless greed, self-absorbed individualism, alienating workingconditions, anomie, exploitation, and cultural decadence which are allegedly soevident in contemporary liberal societies.There is some empirical evidence for both sides in this debate. I shall notdiscuss it here, but simply state that, on balance, the evidence in the contemporarywold seems to me convincing that societies which strive for liberal neutrality tendto be no less successful in maintaining a rich cultural life and relatively prosperouseconomic conditions than societies that do not.(4) The fourth defense for liberal policies stresses the special importance of thevalue of individual self-determination or autonomy. What exactly “autonomy” is,and what forms of it consequentialists should value, cannot easily be spelled outclearly, but a degree of liberal neutrality seems necessary to promote autonomy.Many individuals, if allowed to make their own life choices, make major mistakes.But the value of letting people make such mistakes, from a consequentialist12 Cf. e.g. Macintyre’s chapter “Liberalism Transformed into a Tradition” in Whose Justice? WhichRationality? or Sandel’s Liberalism and the Limits of Justice.234perspective, is that experiencing the consequences of such mistakes constitutes partof controlling one’s own destiny in an authentic way and also constitutes a valuableopportunity to learn life’s lessons and to reach a more conscious, enlightened state.Life, it is sometimes said, can only be lived from the inside; without freedom todiscover our own truth about what the good life involves, and to live by this truth,we cannot realize our selves, but become little more than robots controlled byothers. In Ackerman’s words,What could be better for a person than his own development of a plan of lifethat seems to him good? (368)We are beings who, by nature, care about social living, but we also care about a senseof determining our own course of life, especially after we have been educated to asufficient level to discover exciting opportunities for weaving the strands of our lifeinto a creative pattern. Moreover, since the constituents of human well-being are sodiverse, even the most intelligent attempts at large-scale government paternalismare likely to be far less successful in promoting individuals’ well-being thaneducating our children to a level of responsible autonomy.It seems to me that there is a great deal of truth in all the above claims, andthat an appeal to the widely (but perhaps not universally) shared value of autonomygoes a long way to providing a rational basis for affirming a basically liberal socialorder.§3 But now we confront again the question raised earlier: What is so specialabout self-determination to justify liberals in using state power to enforce favorableconditions for this value without thereby also justifying attempts to use state powerto enforce non-liberal values? Defending liberal politics both in terms of neutralityamong diverse ideas of the good life and in terms of promoting the non-neutralvalue of self-determination seems obviously inconsistent. Maclntyre, for example,charges liberalism with this inconsistency:Liberalism provides a distinctive conception of a just order.... [Tihe theory andpractice of justice within such a polity are not neutral with respect to rival235and conflicting theories of the human good. Where they are in force theyimpose a particular conception of the good life, of practical reasoning, and ofjustice upon those who willingly or unwillingly accept the liberal proceduresand the liberal terms of debate. The overriding good of liberalism is no moreand no less than the continued sustenance of the liberal social and politicalorder. Thus liberalism, while initially rejecting the claims of any overridingtheory of the good, does in fact come to embody just such a theory. (345)Maclntrye’s argument seems to be that the liberal belief that society should allow thegreatest possible variety of differing conceptions of the good contradicts the otherliberal belief that society should not promote one particular conception of the good,because the former belief itself necessarily embodies one particular conception of thegood. How can liberals respond? They would seem to face the following dilemma.Either they claim that liberal politics is designed to achieve substantive neutralityamong the personal values that may plausibly make up individuals’ diverse ideasof the good life or they freely admit that liberal politics is indeed designed to createthe most favorable conditions for personal autonomy. In the former case, liberalswould have to show convincingly that their politics is not biased in favor of theparticular non-neutral value of autonomy. In the latter case, it would seem, liberalscould no longer claim to be neutral among the personal values that make upindividuals’ diverse ideas of the good life, and cannot consistently deny the use ofstate power to non-liberals to attempt to enforce their values.Larmore choses the first horn of this dilemma and defends the peculiar“neutrality” of liberal politics, in the sense that such politics constitute a retreat tothe most “neutral ground” possible and necessary in the face of pluralism about thegood. Publicly, Larmore maintains, liberals must refrain from appealing to the valueof self-determination in their defense of liberal politics and stick with “neutrality” asthe peculiar virtue which justifies a liberal social order. Privately they may of courseindulge in justifications for liberal politics which appeal to self-determination (50f;73-76).But this response seems to me to involve precisely the sort of hypocrisyagainst which Maclntrye inveighs. Unlike Larmore, I think the only credible236response to the dilemma posed for liberals by Maclntyre is to choose the secondhorn and to admit freely that liberal politics is indeed designed to create the mostfavorable conditions for personal autonomy. But then, how could liberalsconsistently deny that the use of state power by liberals to enforce favorableconditions for autonomy justifies, by parity of reasoning, the use of state power bynon-liberals to enforce their idea of the good? Here is what seems to me to be thehonest and sensible answer. Liberal politics (at its best) tries to be as unbiasedbetween people’s personal values as any political system can possibly be. The liberalconception of justice, as Rawis puts it,does indeed, presuppose a theory of the good, but within wide limits this doesnot prejudge the choice of the sort of person that men want to be (1988:260).. ..a just liberal society may have far more space than other social worlds,but it can never be without loss (1988: 266).It is not, of course, possible for such a political order to be neutral between thevalues of autonomy and non-autonomy. Nevertheless, there is somethingsufficiently special about autonomy to entitle those who value autonomy to usestate power to enforce favorable conditions for it, without thereby justifying the useof state power by those whose idea of the good does not include personal autonomy.Allowing non-liberals to enforce perfectionist values other than autonomy woulddeprive those who value autonomy of the opportunity to lead autonomous lives,whereas liberal politics does not deprive people who do not value autonomy of theopportunity to practise a non-autonomous way of life by joining, for example,whatever religious sect they please. So liberalism is indeed special or “neutral” in away no other political creed can be.There can be no question, however, that the dynamics of a liberal society willprofoundly shape or reshape a person’s self. Brian Barry makes this point well:...a liberal must take his stand on the proposition that some ways of life, sometypes of character, are more admirable than others.. .Liberalism rests on avision of life: a Faustian vision. It exalts self-expression, self-mastery andcontrol over the environment, natural and social; the active pursuit ofknowledge and the clash of ideas; the acceptance of personal responsibility for237the decisions that shape one’s life. For those who cannot take the freedom itprovides alcohol, tranquilizers, wrestling on the television, astrology,psychoanalysis, and so on, endlessly, but it cannot by its nature providecertain kinds of psychological security. (1973: 126-7).The liberal self is very different from a communitarian or tribal self. If this isthe main point of Maclntrye’s criticism, it can be readily granted withoutundermining the above argument that liberalism is special in a sense which isrelevant to justifying its monopoly on the use of state power.’3It is not liberals, butnon-liberals, who would seem to face the more serious dilemma. They claim thatthere are standards of the good life by which they can determine worthy ways of lifewith great confidence and precision. But which are those? Either non-liberals arewilling to offer a fairly specific list of goals and values which, they claim, all humanbeings should be educated to pursue; or they offer a broad range of vaguegeneralities. If they do the first, they will immediately encounter rival proposalsfrom other non-liberals, with no compelling rationale for choosing among them.14If they do the second, liberals will argue that these are personal values which aliberal society readily allows or encourages, or political values which it constrains itsmembers to embrace as well. For these reasons, liberal politics is sufficiently specialto justify the use of state power to create a favorable climate for self-determinationwithout, by parity of reasoning, legitimating the use of state power by non-liberals.13 The above argument is, however, sufficient to justify liberalism’s monopoly on the use of state poweronly under reasonably favorable economic and social conditions. This is so for two reasons. Only underminimally favorable economic and social conditions can the means to the effective exercise of personalautonomy be made widely available, and only under such conditions of relative economic security and ahigh level of education is it reasonable to have confidence in the constructive use of this autonomy onthe part of most individuals.14 Non-liberal communitarians may not be persuaded by this dilemma. Rather than claiming thatsome fairly specific list of goals and values constitutes the good of persons under all conditions, theywill argue that the good of persons consists in living by the traditions of their community, whateverthey happen to be, as long as they provide a sense of belonging. The communitarian claim in thisgeneral form seems implausible to me, because members of communities with strong ethnic identitiesoften purchase the good of ethnic belonging at a tremendous cost in opportunities for human betterment,and because there is often a great deal of ethnic chauvinism or fanaticism about the way in which manysuch communities have shaped the identity of their members. This is not to say that the well-beingargument in favor of liberalism necessarily justifies the sort of radical uprooting of indigenous groupsthat took place, for example, in the case of native Indian communities.238Few political distinctions go deeper, in theory and practice, than thedistinction between a non-liberal and a basically liberal state- between a state whichenforces someonetpreferred idea of what makes a human life worthy, as opposedto a state which strives, in good faith, to give individuals wide scope to live as theyprefer, in an atmosphere of free expression, meaningful opportunity and mutualtolerance. This is not to say that liberal practice always works out as it should, farfrom it. But the same can be said about any conception of the best social order.2393.5 THE CONFLICT BETWEEN LIBERTARIANS AND EGALITARIANSThis chapter considers the possibility of rational disagreement about the stronglyconflicting conceptions of justice held by libertarian and egalitarian liberals.3.5.1 The presuppositions of freedom§1 All liberals prize the freedom of individuals to seek their own good intheir own way. In commonsense terms, individual freedom is perhaps best spelledout as “not being prevented by others from doing what, given adequate informationand deliberation, one desires to do and could actually do, unless prevented byothers.” A liberal conception of social freedom, by contrast, can be understood as asystem of constraints on the exercise of individual freedom so as to avert conflict,while maintaining individual freedom to the greatest possible extent.But what should those constraints be? There are bitter divisions amongliberals about the answer to this question. Perhaps the deepest rift exists between twofamilies of principles which can be usefully labelled “libertarian” and “egalitarian”.In an attempt to make vivid the difference between them, I want to begin with twoarguments which lead from the same liberal core idea via plausible premises tostrongly incompatible conclusions. First, a basic libertarian argument:1. The state should be as impartial as possible between diverse ideas of thegood.2. If the state should be as impartial as possible between diverse ideas of thegood, then it may interfere with any individual’s activities only when they causeharm to others, and not to impose someone else’s idea of the good.3. If an individual’s activities make that individual better off without therebymaking others worse off, then any interference with these activities would imposesomeone else’s idea of the good rather than prevent harm to others.Therefore, the state must not interfere with any of an individual’s activitieswhich make that individual better off without thereby making others worse off.240Compare that to a basic argument for what I shall call “egalitarianliberalism”:1. The state should be as impartial as possible between diverse ideas of thegood.2. If the state should be as impartial as possible between diverse ideas of thegood, then it must strive to create at least minimally favorable conditions foreveryone to pursue his or her own idea of the good (from within a wide range ofmeaningful options), regardless of whether or not a person was born wealthy, smart,healthy or otherwise fortunate.3. If the state does not ensure at least a minimally adequate degree ofuniversal access to a stock of goods which serve as important means to a successfulpursuit of a wide range of ideas of the good life, then those who happen to be bornless fortunate in their endowments will be able to pursue their idea of the good farless effectively as those born rich, smart, healthy or otherwise lucky.Therefore, the state should ensure at least a minimally adequate degree ofuniversal access to a stock of goods which serve as important means to a successfulpursuit of a wide range of ideas of the good life (such as, for example, access toproper nutrition, education, medical care, child care, employment, housing, culturalevents, physical and economic security, and others).Both arguments can be suitably elaborated to acquire great rational appeal, and yettheir conclusions seem to remain irreconcilable. I defined social freedom as a systemof constraints on individual freedom which seeks to avert conflict while at the sametime maintaining individual freedom to the greatest possible extent. The abovearguments make clear why libertarians and egalitarians will be sharply divided overthe nature of this system of constraints.For libertarians, social freedom requires the state to enforce principles whichguarantee non-interference with an especially important and fairly sharplydelimitable subset of individual actions (namely those actions which make an agentbetter off without thereby making others worse off). For egalitarians, social freedomrequires the state to enforce principles which promote desirable new freedoms orperhaps even the most freedom overall, even though that will mean frequent241interference with that subset of actions of each individual which the libertarianconsiders especially important. The egalitarian’s idea of social freedom requiresgovernments to create a complex system of duties of mutual assistance to provide aframework of substantive equality of opportunity to enable all individuals to pursuetheir goals effectively— an idea which the libertarian opposes because forcing someto render assistance onto others would interfere with some individuals? freedom touse their own minds and bodies as they choose while not thereby making othersworse off.Let me label the idea of freedom cherished by libertarians basic freedom?? anddefine it, in very preliminary way, as “any person’s right to non-interference withany actions of that person which do not make others worse off.” It would completelymiss the point of the libertarian position to justify some constraint on anindividual’s basic freedom by arguing that some plausible index of aggregatefreedoms as a result of interference will sometimes clearly exceed the index ofaggregate freedoms that would have existed without such interference. Libertariansdo not deny that (although they may well deny that any such index is possible evenin principle). They deny that trade-offs between an individual’s basic freedom andother values, including maximum freedom overall, are ever permitted.Exactly what sorts of actions are those which make an individual better offwithout thereby making others worse off? And why should such actions be regardedas entailing an absolute or near-absolute right to non-interference? I explore theseconcerns below. But first I must address a pressing question. It may seem an all tooobvious empirical fact that the libertarian right to non-interference cannot pass thewell-being test required for rights by a consequentialist moral perspective. Iflibertarian ideas about social justice are grounded on little more than a bareassertion of natural rights, can we not exclude them offhand from the realm ofrational disagreement about justice?242The answer is no. First, some libertarians, for example Hayek (and apparentlyNozick, Narveson and Gauthier as well), deny that libertarian rights-claims wouldin fact fail every plausible version of the consequentialist’s well-being test. Theyargue that the libertarian’s favored social institution, the free market, has eitherfortuitously or necessarily the socially most beneficial effects of all knownalternatives.1Second, there has been a recent, prominent and sustained attempt byNarveson to defend a con tractarian form of libertarian liberalism whose practicalimplications (though not its justification) are claimed to be largely compatible with autilitarian value perspective. If this striking claim is really defensible — if moralpractices grounded in Gauthiers mutual-benefit individualism can plausiblypromise greater social benefits than more egalitarian-liberal practices built onconsequentialist justifications—, then of course such contractarianism could bereinterpreted as a form of indirect consequentialism. Consequentialists would thenhave to regard mutual-benefit individualist methods of moral reasoning asjustifiable on consequentialist grounds. They would have to support thedismantling of most redistributive social institutions along with the demise ofpractically any other institutions, values or relationships which cannot survive theharsh justice” of free markets.§2 Differences between libertarian and egalitarian conceptions of freedomproduce vastly different conceptions of social justice. For liberals generally, the verypoint of social justice is to ensure the greatest equal liberty for all. For egalitariansliberals, the greatest equal liberty for all cashes out as guaranteeing everyone anadequate basic scheme of equal political rights and freedoms compatible with thesame rights and freedoms for all, together with a more or less extensive list of socialand economic rights and opportunities. For libertarian liberals, however, social1 Cf. Friedrich Hayek, The Mirage of Social Justice (1976); Robert Nozick, Anarchy, State, and Utopia(1974); Jan Narveson, The Libertarian Idea (1988); David Gauthier, Morals by Agreement (1986).243justice is a matter of respecting the right to non-interference with each other’s basicfreedom to use our individual minds and bodies, and the products of our mindsand bodies, as each of us sees fit as long as he or she does not thereby make othersworse off. This cashes out as, essentially, being able to exercise one’s talents in amore or less laissez faire market and letting others do the same. This particular typeof freedom is what Narveson has in mind with is striking claim thatconstraining the liberty of some in order that others may prosper (or live atall) is contrary to justice. (:101)2Libertarians typically regard social distributions as just if and only if there is afree market in goods which allows individuals to acquire, through voluntarytransactions, whatever private property they desire, and to dispose of this propertythrough whatever voluntary transactions they choose, without any regard whateverto such commonly invoked just-making considerations as need or equality.Egalitarians, on the other hand, consider social distributions to be just only ifindividuals enjoy substantive equality of opportunity in their ability to developtheir natural endowments, to make effective use of political rights, and to acquirethe most important means to well-being. The spectrum of egalitarianism is,however, very broad, ranging from those who want to ensure for every individual aminimally favorable starting point in life’s competition to all those who want notonly to eradicate all socially caused disadvantages which some now suffer, but as faras possible compensate for natural disadvantages as well. (Some may even want toregard all disadvantages as ultimately socially caused.)To further clarify the depth of the division between the egalitarian andlibertarian outlook, I want to distinguish between two types of social distributions:intended and unintended. Distributions may be intended, as when governmentenacts a social welfare policy, or unintended, as when distributions result2 Bare page numbers in brackets following quotes from Narveson and Gauthier refer to The LibertarianIdea (1988) and Morals by Agreement respectively (1986).244inadvertently from many people’s uncoordinated decisions. Many distributiveoutcomes which were not intended by anyone’s conscious decisions can, however,be modified by deliberate political choice. According to egalitarians, distributiveoutcomes resulting from people’s uncoordinated market transactions are unjust ifthey put some groups at a systematic disadvantage on the basis of irrelevantcharacteristics, such as race, sex, ethnic origin, and others. But according tolibertarians, if distributive outcomes resulting from people’s uncoordinated markettransactions put some groups at a systematic disadvantage, this may be bad luck forthem, but is not unjust. Libertarians often reason that anything that is notconsciously intended by someone merely happens, and what merely happens to aperson cannot be unjust.3The core intuition of egalitarians about justice is perhaps best expressed bysaying that there is a presumption of social injustice if some people, or members ofsome group, end up significantly worse off than others through no fault of theirown. The corresponding core intuition of libertarians about justice could beexpressed by saying that there is no presumption of social injustice if some people,or members of some group, end up considerably worse off than others through nofault of those others. Egalitarians will say that even if the better-off are not directlyor intentionally causing others to be worse off, such undeserved inequalities arenevertheless often correctable faults of the system. Libertarians will reply that thesealleged faults of the system could only be corrected at the expense of ongoing,intolerable interferences with the outcome of people’s creative use of their ownbody and mind and with their voluntary transactions.4In any case, egalitarianscannot agree among themselves on a “correct” standard of equality. They have far3 Cf. e.g. J.P.Day, Liberty and Justice (1987). For him, the fact that the offspring of the poor may haveconsiderably less access to quality secondary and tertiary education, for example, and have therefore nochance at all to qualify for many of society’s most desirable positions, is simply “bad luck, like beingfemale. And, as we have seen, there is no right to compensation for misfortune.”(94)Cf. Nozick’s Wilt Chamberlain example, now the classic argument of this kind (1974: 161-163>.245greater trouble than libertarians when it comes to agreeing, in concrete terms, abouthow much inequality social justice permits. No doubt, the promise of escapingchronic infighting among egalitarians on this issue is, for many, part of the allure oflibertarianism.§3 The libertarian idea, according to Narveson,is to maximize individual freedom by accounting each person’s person as thatperson’s own property, that is, by giving each person the maximal level ofrights to the disposition of that particular bit of ‘property’ that is compatiblewith everyone else’s having such rights. (:175)Libertarians will typically identify a wide range of activities as coming under thisproperty right to one’s own person, such as, for example, engaging in any “capitalistacts among consenting adults”, bequeathing an inheritance to one’s offspring, orrefusing to hire particular individuals as employees of private firms for whateverreasons one pleases. Libertarians also identify a whole range of political practices asinterference with the property right to one’s own person, such as, for example, landuse restrictions or building codes, compulsory universal health or unemploymentinsurance and pension schemes, and most forms of taxation, most notably theprogressive income tax. I shall call the disputed right to non-interference in thisdistinctive sense “libertarian property rights??.When reasonable, informed people of good will are locked into bitterdisagreement, as they certainly are over the extent of property rights, each side islikely to have a strong defense.5 Although the dispute between libertarians andegalitarians is an example of a disagreement between informed people withimpeccable rational credentials, we can nevertheless not simply conclude, I think,that it must therefore be a rational disagreement which currently available5 Of the master principles listed in Sec 3.2.1, Narveson and Nozick’s principles exemplify thelibertarian approach to social justice. Nielsen and Derek Phillips’ principles exemplify the stronglyegalitarian approach while Rawis and Sterba’s principles exemplify a more moderate egalitarianapproach.246arguments, information or methods have no promise of resolving. I obviouslycannot discuss all the constructivist justifications which libertarians have offered indefense of libertarian property rights, but to establish at least a very tentativepresumption about the scope for rational disagreement, I want to briefly considertwo of the most prominent contemporary justifications: Narvesons mutual-benefitindividualist defense and the more straightforwardly utilitarian defense.3.5.2 The contractarian justification of libertarianism§1 In The Libertarian Idea, Narveson proceeds from seemingly innocuousassumptions to libertarian property rights. I shall summarize what I take to be thestructure of his argument, relying heavily (as does Narveson) on concepts pioneeredby Gauthier.Narveson begins his “central argument” with a carefully worded question:The question is whether I am motivated, by virtue of my possession of almostany set of values you can imagine, to acknowledge, when I deal with others(other more or less autonomous beings, that is) any principles restricting myfreedom of action in any way, and if so, which. (175)“The general form of the answer,” Narveson continues, isthat we shall all adopt principles restricting the performance of actions byothers that would make ourselves worse off than we would be if we didn’thave that restriction. We must, by definition, favor the nonperformance ofsuch actions, since by hypothesis we are assessing worse-offness in terms ofour own values, whatever they may be. (175)Narveson does not specify any hypothetical position for contractarian bargaining.Doing so would, of course, immediately invite the charge of question-begging. Butcan he plausibly claim that he is referring to any conceivable situation whatever inwhich rational persons may find themselves - such as, most importantly, the hereand now in the actual word? If “rational person” is understood to mean “a personwho cares only about the maximal fulfillment of his own preferences”, then itwould certainly seem that, starting from our situation in the actual world, he couldnever make out a case that it must be rational for each of us, regardless of our247differences in power and values, to adopt identical principles restricting theperformance of actions “by others” in certain ways. (After all, “we” areinterchangeable, for the purposes of Narveson’s argument, with those “others.)Is there a hypothetical bargaining position which we- assuming we arerational persons (individual utility maximizers caring only about maximalfulfillment of their own preferences)- would accept? If it is not rational for each ofus, given our present situation, to agree to a certain principle restricting ourbehavior in the actual word, how could it be rational for each of us, given ourpresent situation, to accept a certain hypothetical bargaining position in which eachof us would have found it rational to agree to such a principle? From the point ofview of a rational person in the actual world, is there any interesting choice to bemade between agreeing to a certain principle and accepting a hypothetical positionin which he would have agreed to this principle? If someone wouldn’t rationally dothe first, why would she do the second?The answer, of course, is that she would not. Each of us would find it rationalto accept some hypothetical position only if the principles she agrees to in thisposition would also be rational for her to agree to in the actual world. Nevertheless,postulating a hypothetical position may be a useful expository device. It may modelour rationale for adopting principles in a vivid enough way to actually suggestpromising principles to us. As long as we can make out a case that a certain principlewhich each of uswould agree to in a hypothetical position is such that it is alsorational for each of us to agree to it in the actual world, we have begged no question.§2 I shall now address four questions in the following order: (1) What sort ofhypothetical position should we postulate for expository purposes? (2) Whatprinciple would rational persons agree to in this position? (3) Is it indeed rationalfor us in the actual world to agree to this principle as well? And (4) Is it rational forus in the actual world to comply with this principle - or if not, is it perhaps still248rational for us to do something about the fact that it is not rational in the actualworld to comply with this principle?(1) Which hypothetical position should zve assume for expository purposes?The answer to that question is strongly suggested by Narvesons “general form ofthe answer’ to the carefully worded question which he wants rational persons to askthemselves (see quotes above). It seems to be an obvious analytic truth that todetermine whether I have become worse off as a result of actions by some specificother, I must determine how well off I would have been had that particular personnever acted in such a way as to affect me at all. I am supposed to imagine a stateprior to being affected, positively or negatively, by any actions of Y. And similarly, todetermine whether I have become worse off as a result of actions by some person orother or other people generally, it would seem that I must determine how well off Iwould have been had other people never acted in such a way as to affect me at all.This situation in which I imagine myself to be when I imagine that other peoplenever acted in such a way as to affect me at all will be called “the presocial position”(cf. Gauthier, 1991:25).(2) What principle would rational persons agree to in the presocial position?Let us imagine rational persons in a presocial position preparing to negotiate thefundamental principle of a social contract: would it be rational for each of them toaccept some principle restricting the behavior of each? The answer to that isobviously yes. Although ex hypothesis we imagine them not to have affected eachother in any way yet, the following developments can be plausibly foreseen in theabsence of any restrictions whatever on each other’s behavior. Certain general factsabout the human condition - scarcity of goods relative to people’s desires for them,sufficient equality of power so that usually even the relatively weak can inflict harmon the relatively strong, and greater concern among most individuals for satisfyingtheir own desires than those of others - will predictably tempt some individuals totake goods away from others. This will lead rapidly lead to high levels of mutual249resentment and general mistrust along with preemptive defense measures on thepart of many individuals who want to protect their lives and possessions. Thissituation can plausibly be imagined to escalate into Hobbes’ famous “war of everyman against every man” which would make practically everyone’s life “solitary,poor, nasty, brutish and short.”Rational people clearly would not want to lapse into a Hobbesian state ofnature. But neither would rational people in the presocial position simply acceptany principle restricting each other’s behavior that will merely stave off theHobbesian state of nature — perhaps making their lives a little bit less poor, nasty,brutish and short. Obviously, for that purpose a great many principles would do.Rational people want to do better than that. I take my clue from Narveson when hesays,“Each rational agents wants, after all, to get the best possible deal, and not justto beat the state of nature.”(155).Although there is a sense in which it is clearly rational for each of us to accept anyprinciple averting the state of nature to any principle which would not avert it, weare looking for something better: one particular principle which can be shown to beuniquely in the rational self-interest of both the relatively weak and the strongamong us to agree to in the presocial position.Which principle is that? Is there perhaps a principle such that a rationalperson would not voluntarily agree to anything less favorable to himself — while atthe same time realizing that other rational persons would not voluntarily agree toanything more favorable to him? Well, take the principle permitting everyone tobetter his situation in life while not permitting others to worsen it at any timewhatever. Let us call this principle “constraint X”. Let us consider now whether, inplausible circumstances, both the relatively powerful and the relatively powerlesscould be expected to voluntarily agree on any constraint that is either stronger orweaker than constraint X. (By “plausible circumstances” we do not mean to importcontroversial assumptions like equality of power, which obviously do not hold in250the actual world. We are only assuming three things which do very often hold: thatcircumstances obtain in which a cooperative surplus is possible which can then besplit to the mutual advantage of both the weak and the strong; that each person is“rational”, in the sense of caring only about the fulfillment of his own preferences;and that no rational person would willingly accept a principle restricting eachperson’s behavior that is clearly less advantageous to him than to the others.)In these circumstances, then, would the relatively powerful and the relativelypowerless agree on any constraint either stronger or weaker (from the point of viewof any given individual) than constraint X? Stronger constraints than constraint X(i.e. more restrictive from the point of view of any given individual) would be, forexample, a constraint permitting a given individual to better his situation only if hethereby also betters the situation of others; or a constraint permitting a givenindividual to better his situation while permitting others to worsen it. It seems quiteplausible that stronger constraints than constraint X will be unattractive, both to thepowerful and the powerless. After all, we are beings who want to be free to betterourselves. A principle that allows a rational person to better his situation while notallowing others to worsen it would seem to be clearly preferred over principles thatallow him to better his situation while also allowing others to worsen it, or to betterhis situation only if he also betters the situation of others.Very well, but what about weaker constraints than constraint X? A weakerconstraint (i.e. less restrictive from the point of view of each individual) would be aconstraint that permits that individual to better his situation and does not permitothers to worsen it, but permits him to worsen the situation of others. Wouldn’tthis constraint seem extremely attractive from the point of view of a givenindividual?Not so, as a little reflection shows. Every rational person who knows thatothers are also rational already knows beforehand that such a constraint would notbe acceptable to them, and would therefore give rise to “costly interaction”.251As Gauthier puts this point,Rational persons will voluntarily accept an agreement only insofar as theyperceive it to be equally advantageous to each. To be sure, each would behappy to accept an agreement more advantageous to herself than to herfellows, but no one will accept an agreement perceived to be lessadvantageous. Agents whose rationality is a matter of common knowledgewill recognize the futility of aiming at or holding out for more, and minimizetheir bargaining costs by coordinating at the point of equal advantage. (1.991:28; cf. also 1986: 227)This crucial- and perhaps initially a bit counterintuitive- step in theargument is worth belaboring for a moment. It is clear why the relatively powerlesswould not demand a weaker constraint than constraint X on their own behavior:they know beforehand that it will not be acceptable to the relatively powerful andwould give rise to conflict — in which the powerless would by definition lose out.But why would the relatively powerful hesitate to demand a weaker constraint thanconstraint X on their behavior vis-a-vis the relatively powerless? Isn’t itcommonsense that it would be rational for the powerful — fulfill more of theirpreferences — if they reserved the right to use the powerless any way they feel like,much the same way in which many people currently use animals? Why should thepowerful settle for a constraint that doesn’t permit them to better their ownsituation at the expense of the powerless, when they could clearly get away with it?As Narveson explains,Would the ones doing better voluntarily move to a situation in which theiradvantages were negated? In fact, the suggested answer is that they will do so,provided their situation after the agreement is better than it was before theagreement, even if the agreement negates their relative advantage over theweak. (177)So there are highly persuasive reasons why, in plausible circumstances, thepowerful might benefit far more from voluntary cooperation rather than from theoppression, enslavement or outright extermination of the powerless. But if nostronger nor weaker constraint than constraint X will be voluntarily accepted byrational persons, we would seem to have achieved a significant result: that rational252persons would converge on constraint X in plausible circumstances in a presocialposition.(3) But is it rational for us to agree to constraint X in the actual world as well?We said that it is rational for us to accept the presocial position if and only if theprinciples agreed to in this situation would also be agreed to by rational persons inthe actual world. But why would it be rational for both relatively powerful andpowerless persons, in the actual world, to agree to a principle constraining theirbehavior in such a way that they can better their own situation, but may neverworsen the situation of others?It would clearly be rational for them to do so if each could be sure ofeveryone’s general compliance with this principle, and if general compliance withthis principle would make greater cooperative surpluses possible in the actual wordthan are being realized at the present, and if each could be sure that he would not beleft out when it comes to splitting this cooperative surplus. But these would seem tobe three incredibly tough, counterfactual conditions to meet.First, how could each rational person be sure of everyone’s generalcompliance with constraint X? Making the plausible assumption that a rationalperson Y knows that other people are also rational, Y also knows that everyonewould generally comply if each thought that doing so would make a cooperativesurplus possible which would be split so as to benefit him relative to his presentsituation. For reasons to be spelled out below, it is extremely plausible thateveryone’s compliance with this principle would, in fact, make a cooperativesurplus possible which could then be split to everyone’s benefit. Now if everyonecould be made to understand the reasons why it is extremely plausible thateveryone’s compliance with this principle would make a cooperative surpluspossible which could be split to everyone’s benefit, wouldn’t that give each rationalperson in the actual world sufficient incentive to comply?253Unfortunately, no. In the actual world, a rational person cannot expectgeneral compliance with constraint X, but at most feel more or less confident that, atleast in many cases, other people will comply with whatever principles reflect theactual moral beliefs prevailing in a given society. If so, it is generally rational tocomply not with the principle that everyone may better his own situation and noone may worsen it, but with whatever principles one is expected to follow in thatsociety, on pain of moral condemnation, legal sanctions and various other forms ofsocial ostracism. But this prevailing state of affairs is deeply unsatisfactory from arational point of view, for at least five main reasons.(a) First of all, it is unsatisfactory because different people in a given societytend to have conflicting moral beliefs which lead to social conflict and costs in termsof forgone cooperative surpluses.(b) Second, it is unsatisfactory because the diversity of moral beliefs and theoften transparently superstitious justifications given for them can only promotewidespread moral skepticism among intelligent people. Rational persons will bemotivated to violate moral principles in surreptitious ways, whenever they judgedoing so to their personal benefit.(c) Third, it is unsatisfactory because the system of sanctions necessary toenforce compliance with whatever moral beliefs prevail in a given society can beextremely costly to maintain (costly in terms of foregone cooperative surpluses).(d) Fourth, it is unsatisfactory because even if all people agreed on the samemoral beliefs, and each of them voluntarily displayed perfect compliance, theintrinsic nature of these principles would still be highly unlikely to lead anywherenear to an optimal allocation of social resources to their most productive uses.Universal compliance with the moral code of most societies would still be highlyunlikely to produce the types and division of cooperative surpluses which wouldmaximally fulfill the preferences of each or even of most individuals.254(e) Fifth, it is unsatisfactory, because to the extent that rational persons realizethis last point, their realization will tend to reinforce their unwillingness to supportthe existing social order and will thus promote further social instability andforegone cooperative surpluses, over and beyond the instability produced by factorsoutlined under (a) and (b) above.Nevertheless, unsatisfactory as this state of affairs is, the conclusion stands: inthe present situation, it is generally not rational for a given individual (restrictedeconomic contexts now governed by free markets are an exception) to comply withconstraint X, rather than to make complex judgments, in the light of a host ofconsiderations, to what extent compliance with prevailing moral beliefs may be tohis benefit. But does that fact not deal a fatal blow to Narveson’s and Gauthier’sproject? The answer, surprisingly, is no. From the fact that it is not generallyrational to comply with constraint X in the actual world, it does not follow that itmay not be rational, for many individuals, to do something about this very fact.(4) Is it rational for us to do something about the fact that it is not rational formost of us in the actual world to consistently comply with constraint X? And to thisquestion the answer would seem to be clearly yes for most rational people (if notnecessarily for multimillionaires). If, as argued above, the present situation in theactual world indeed involves tremendous forgone opportunities for producingcooperative surpluses, then it would clearly be rational for us to move to a socialorder that is fundamentally structured so as to reliably produce cooperativesurpluses as a result of practically every social interaction — one in which observingconstraint X wouldn’t only be actually (but unbeknownst to most people) rational,but would also be known to be rational by confirming its own rationality intransparent ways in almost every social interaction. But that would require thatsocial interaction would have to consist almost entirely of transactions voluntarilyengaged in for mutual benefit; in short, free market transactions. There are twoimportant points to note about a social order permeated at every level by free255markets. (1) In such a market order, constraint X would be almost universally self-enforcing, because people wouldn’t voluntarily choose to keep engaging intransactions with others (at least not more than once or twice) who worsen theirsituation. Because constraint X would be observed almost universally, this socialorder would be stable in way in which no order can be which routinely allows somepeople to make others worse off. And (2), in a market order, decisions aboutallocating resources are made in a decentralized fashion, by each individual usinghis or her special knowledge of opportunities for mutually beneficial cooperationwith particular others. This sort of special knowledge is necessarily the property ofindividual persons. As Hayek never tires to remind us, no single mind can possiblyencompass all this special knowledge of all individuals and thus plan the allocationof resources in a centralized way. Therefore, in a market order, individuals haveincentive to use their special knowledge of opportunities for mutually beneficialcooperation so that resources will gravitate to their most productive uses — that is, tomaximum fulfillment of each individual’s preferences.But if a market order embodies these two features — the greatest possiblestability of any conceivable social order and a strong tendency for resources togravitate toward their most productive uses — then it becomes extremely plausible tothink that this order makes far larger cooperative surpluses possible than any othersocial order. Now as soon as rational people realize this fact, the existing social orderwill become even more unstable than it already is. In most situations, it will becomerational for individuals to provisionally comply with prevailing moral principleswhile at the same time working toward a radical change in the existing social order.This will be so for a given individual when the expected costs to him of workingtoward such change are outweighed by the expected benefits. And, at least indemocratic states which honor freedom of expression and association, the expectedcosts may be quite small. In dictatorships (such as the former Soviet Union), therisks of working toward a radical change in the social order are, of course, vastly256greater, but since the anticipated payoffs may also be vastly greater, it may wellbecome rational for many individuals, even in dictatorial states, to work towardchange by intelligent risk-minimizing means.This, in brief, is what I understand to be the basic structure of Gauthier’s andNarveson’s con tractarian argument for a libertarian social order.§3 The force, coherence and sheer intellectual beauty of the mutual-benefitindividualist’s arguments have certainly not eluded me. Their tight coherence as aself-contained conceptual system makes it tempting, however, to overlook someserious limitations.Before I express my worries about these limitations, I must stress threefundamental respects in which I take contractarians like Gauthier to be fellowtravellers, and in fact intellectual leaders, toward goals shared by consequentialistconstructivists. Both consequentialist constructivists and mutual-benefitindividualists regard “morality” as a humanly contrived system of constraints onbehavior which is worth having because such a system, at least if intelligentlydesigned and sufficiently internalized by most people, will vastly improve almosteveryone’s chances for a good life by overcoming coordination and freeloadingproblems (and by offering intrinsic emotional satisfactions to many). Both considerthe ultimate goal of a constructivist morality to be essentially the maximalsatisfaction of people’s informed preferences, although they disagree on the meansto promote this goal most effectively. And both will want to employ contractarianreasoning or game-theoretic modelling of moral decisions in an attemptconceptualize moral issues and to identify “invisible-hand” processes which mayachieve far better outcomes than other methods of promoting consequentialist goals(although here again they disagree on the extent of their reliance on such processes).But eventually consequentialists and contractarians like Gauthier must partcompany. While Gauthier’s work has inspired a number of promising research257projects (c.f. e.g. Danielson’s work), I now want to state some fundamental worriesabout his mutual-benefit individualism as well as the radical libertarianismNarveson wants to base on it.(1) Even if fully rational persons would agree to a principle which allowsthem to better their situation while not permitting others to worsen it, it is far fromclear that this abstract principle can simply be mapped onto real life by the idea ofall-pervasive markets replacing political decision-making or morally cementedcommunity practices. In theory it might seem that a social world permeated bymarkets at every level of cooperative interaction— instead of such things asredistributive social programs— is a world where there is optimum resourceallocation, no coercion, no need for a clumsy and corrupt bureaucracy, utter bliss, asclose as we can humanly come to creating heaven on earth. But our empiricalknowledge of markets must surely give us mixed feelings. Markets have their place.But they also have their own coercive dynamics, their horrendous externalities,their monopolistic momentum, their distinctive failures. So when Narvesonclaims that Gauthier’s arguments imply the libertarian principle of justice, whichsays (among other normatively potent things), thatWe [i.e. via our government] have no fundamental general duty to provideothers with such goods as the necessities of life, let alone some particularproportion of all the socially distributable goods there are;.... (166)I cannot help wondering how many of us would find ourselves falling by thewayside if this principle were implemented. (It is only fair to point out thatGauthier himself is far more cautious about the concrete normative implications ofhis work.)Nozick once saw markets as a panacea too and, in fact, saw utopia as averitable orgy of markets, including markets in types of communities whichindividuals would freely join or leave according to their preferences (1974: 312f). Ifound it most sobering to reflect on what Nozick has to say fifteen years later:258“The libertarian position I once propounded now seems to me seriouslyinadequate, in part because it did not fully knit the humane considerationsand joint cooperative activities it left room for more closely into its fabric. Itneglected the symbolic importance of an official political concern with issuesor problems, as a way of marking their importance or urgency, and hence ofexpressing, intensifying, channeling, encouraging, and validating our privateactions and concerns toward them. Joint goals that the government ignorescompletely- it is different with private or family goals- tend to appearunworthy of our joint attention and hence to receive little. There are somethings we choose to do together through government in solemn marking ofour human solidarity, served by the fact that we do them together in thisofficial fashion and often also by the content of the action itself... .A concernfor the expression and symbolization of values that can best and mostpointedly, not to mention most efficiently, be expressed jointly and officially -that is, politically- is continuous with a concern for individual self-expression. There are many sides of ourselves that seek symbolic self-expression, and even if the personal side were to be given priority, there is noreason to grant it sole sway. If symbolically expressing something is a way ofintensifying its reality, we will not want to truncate the political realm so as totruncate the reality of our social solidarity and humane concern for others..The libertarian view looked solely at the purpose of government, not at itsmeaning; hence, it took an unduly narrow view of purpose too.. ..If ademocratic majority desires to jointly and symbolically express its mostsolenm ties of concern and solidarity, the minority who prefer differently willhave to participate sufficiently to be spoken for.” (1989: 286-290; emphasisadded)Obviously, the scope and limits for markets in the best social order is an immenselycomplex topic. I shall briefly consider some of the pros and cons of markets below,when I discuss the utilitarian defense of libertarianism.(2) But if the connection between Narveson’s extreme libertarianism andGauthier’s contractarianism is severed, it becomes deeply problematic what exactlymutual-benefit individualist arguments prove for disputes about justice in the realworld. Gauthier certainly seems to have made a case that rational people inplausible circumstances might converge on the principle that each should bepermitted to better his own situation while no one else is permitted to worsen it.But how precisely can this be operationalized in the context of contemporary realworld applications?Gauthier maintains that the intuitive and ambiguous principle must bespelled out in terms of what he calls “the Lockean proviso”:259...just as I better your situation in so far as you prefer the outcome ofinteraction with me to what you would have expected otherwise, in myabsence, so I better my own in so far as I prefer the outcome of interactionwith you to what I should have expected otherwise, in your absence orunavailability for interaction. And I worsen my own situation in interactionwith you in so far as I prefer what I should have expected in your absence tothe outcome that I actually bring about...We interpret the Lockean proviso sothat it prohibits worsening the situation of another person, except to avoidworsening one’s own through interaction with that person. (205; emphasisadded)What exactly counts as “interaction” in this context? Let me unpack Gauthier’scrucial idea. We interact with someone only if the other person’s presence causallyaffects the process of bettering ourselves in any way. If, in the other person’s absence,this process would have unfolded in exactly the same way, no interaction in therelevant sense actually took place (209). But this account will cause alarm, and themore so after reading Gauthier’s own provocative example:Suppose that we live as fisherfolk along the banks of a river... .But if you,living upstream from me, merely use the river for the disposal of yourwastes, then even though you thereby kill many of the fish in my part of thestream, you do not violate the proviso. For although you worsen mysituation in relation to what I should expect in your absence, you do not betteryour own situation through interaction with me. You are no better off thanyou would be were no one to live downstream from you. (211-2)So if I own a factory upstream from you and better myself by polluting the river andkilling all the fish on which you depend for your livelihood, I have not interactedwith you because even if you had never existed, this would not in any way havecausally affected the process by which I better myself. Since no interaction took placebetween you and me, I have not, in this case, made myself better off by making youworse off through interaction. Therefore, Gauthier concludes, the Lockean provisohas been fully met.Any acceptable formulation of the Lockean proviso must contain the ideathat in bettering oneself one must not make others worse off or take advantage ofthem. But I see no good reason why we should accept Gauthier’s particular accountof “interaction” as a necessary feature of the commonsense idea of not taking260advantage or not making others worse off. In fact, Gauthier’s own example is aparadigm case of one person’s taking advantage of her position to exploit resourcesin a way which makes someone else worse off, even though the causal interactiongoes all one way. It would simply seem to beg the question in favor of libertarianismto spell out “taking advantage” in the particular causal way Gauthier proposes. If theLockean proviso is interpreted this way, then it would not have been the principlethat every individual utility maximizer who wants to form a society would alreadybe prepared to accept as her or she approaches the bargaining table.If the Lockean proviso is interpreted so that it clearly is to everyone’sadvantage to accept, it will only allow a person to better himself in ways that do nothave negative effects, however indirect, on others, in which case its scope ofapplication will be rather narrowly restricted. But, Gauthier may reply, surely theLockean proviso must mean something more. How should it be interpreted, if notthe way he proposes?If the Lockean proviso were interpreted as allowing a person to better himselfonly in ways that do not have any (non-trivial) negative effects on others, then itwould become unrealistically restrictive because, in practice, individuals’ attemptsat bettering themselves almost always causally impact on the prospects of others inmany subtle and not-so-subtle ways. Any relative or absolute increase in the wealthof some will, for example, worsen the situation of others with respect to inherentlyscarce goods in ways that are certainly not trivial. No one lives in a vacuum; and sothe Lockean proviso is almost never uncontroversially satisfied. Within hard-todefine limits, bettering oneself, even if it is sometimes at the expense of indirectly ormarginally worsening the situation of others, must under some conditions beperfectly compatible with social freedom and justice. In practice, a clean causalseparation between some people’s wealth-acquiring activities and other people’spoverty-stricken situation is rarely possible, even though the causal connections areoften far too complex for anyone to trace exactly. But this opens up all the questions261surrounding Mill’s harm principle, a principle of which the Lockean proviso isclearly only a variant. The question arises what, if anything, we are actually free todo if we are only free to do what will not harm others. The best we can do inapproaching this question, it seems to me, is to weigh the consequentialist pros andcons of different types of actions or policies with as much foresight as we canhumanly muster, making intelligent judgments about which action-types shouldqualify for consequentialist rights-status while recognizing that what does and doesnot harm, in a way serious enough to be a concern of justice, will change with ever-changing circumstances. Certainly, given the complex causal web whichinterconnects the actions of individuals in a society, Mill’s harm principle is far toosimplistic by itself to adjudicate between permissible and impermissible types ofindividual freedom or permissible and impermissible types of egalitarianism. Icannot address these complexities here. What I have argued is simply that theLockean proviso does nothing whatever to settle the dispute between egalitariansand libertarians in favor of libertarianism. If the Lockean proviso is interpreted so asto be in everyone’s self-interest to accept prior to bargaining, it does not havelibertarian implications. If it is interpreted so as to have libertarian implications, it isnot in everyone’s self-interest to accept prior to bargaining.(3) Gauthier goes on to give a mathematically precise interpretation of what itmeans for co-operators be be constrained by his version of the Lockean proviso.According to Gauthier, a division of the cooperative surplus among fully rationalcooperators requires them to determine, for each cooperative interaction, what sortof utility outcomes would have resulted in the complete absence of both cooperativeand non-cooperative interaction and then to minimize the “maximum relativeconcession”. But the counterfactual information required to establish the nonagreement point in typical real-life situations in which disputes about justice ariseseems elusive. Gauthier seems to me to abstract from the particulars of real-lifesituations to such an extent, and would require such enormous informational262input, as to leave completely unclear how the relative concessions required of thedifferent parties should be calculated. In fact, many of those conversant withbargaining theory find it unclear why Gauthier’s minimax relative concessionprinciple should be regarded as authoritatively defining the agreement whichrational persons must reach when they bargain over the division of a cooperativesurplus. This point would, of course, require detailed technical discussion.6But it ishard to see what follows from the mathematical precision of Gauthier’s approachfor real-life moral decisions. The fact is that most real-life situations in whichdisputes about justice arise seem to involve far too complex variables for ex antegeneralizations about the most rational outcome of such bargaining among self-interested utility maximizers.(4) When criticizing alternative moral theories, Gauthier seems to promiseus a robust mutual-benefit individualism but in the end, I think, he delivers onlyan abstract one (while equivocating between the two). Let me make this importantdistinction clear. Either mutual-benefit individualists conceive of rational self-interests abstractly, in terms of the maximum gain to be realized by each member ofa society of mutually disinterested utility maximizers who are highly successful inavoiding costly lapses into non-cooperation due Prisoner’s Dilemmas (freeloading)and Assurance (coordination) Problems. Or mutual-benefit individualists conceiveof individual interests robustly, in terms of what sorts of interests mutuallydisinterested utility maximizers, given the epistemic limitations of ordinary humanbeings, will plausibly find motivationally potent for themselves.In the case of “abstract” mutual-benefit individualism, I cannot see what, inspecific distributive contexts, it actually amounts to, because it would seem to6 This objection to Gauthier’s contractarian libertarianism is most lucidily elaborated by Brian Barry(1989). Gauthier’s theory presupposes the possibility of a general solution to real-life bargainingsituations to give principled guidance to decision-makers. Barry argues persuasively that there aremany possible types of “splitting-the-difference” solutions, none of them having any rationaladvantage over others (cf. 388-92). Others also make a forceful case that Gauthier’s proposed solutionis, at any rate, faulty (cf. e.g. Danielson, in Vallentyne, 1991; or Gibbard, 1991).263require an uncanny form of prescience on the part of individuals. While relativelysimple and highly artificial situations are conceivable where Gauthier’s strategy mayindeed work to maximize everyone’s preference satisfaction, how can this strategybe applied to the dynamic rough-and-tumble of interaction in a complex society? Ican make no sense of the idea of keeping track of counterfactual Prisoner’sDilemmas which have been avoided, and keeping track of the actual payoffs whichhave been realized as a result, thanks to compliance with moral constraints on thepart of a multitude of identifiable individuals. But if the contribution of specificindividuals to the avoidance of (counterfactual) Prisoner’s dilemmas cannot becredited, then how is the cooperative surplus gained from such avoidance to becalculated and its division arranged? And why should those who are relativelypoorly off at an earlier point in time expect that giving the already well-off adisproportionate share of any cooperative surplus will serve as a special incentivefor the well-off to cooperate in ways which will make the relatively poor better off inthe long run than they would have been in the absence of such a skewed division?It would seem clear that the “abstract” mutual-benefit individualist can offer a self-interested utility-maximizer neither specific normative guidance about how tomaximize his rational self-interest nor, a fortiori, motivationally potent reasons fornot maximizing his self-interest by worsening the situation of others.“Robust” mutual-benefit individualism, on the other hand, would seem tosanction the sort of predatory actions which will vitiate it as a plausible candidate forcarrying out morality’s main job. According to this theory, only when the relativelystrong and well-off judge, according to their own epistemic lights, that they havemore to gain from cooperating with the weak and poorly-off rather than excluding,oppressing or even exterminating them are they rationally (and thus morally)required to cooperate. Robust mutual-benefit individualism would seem torationally require individual utility maximizers to cultivate dispositionscommitting them to a mixed strategy of intelligently combining constrained (but not264too constrained) maximization with straightforward maximization whenever amajor scoop can be made with impunity. Since morality, for Gauthier, arises onlybetween parties whose cooperation will lead to a cooperative surplus from whichboth benefit, he cannot consistently say that there is anything whatever morallywrong whenever those with vastly superior technology have nothing to gain fromcooperation with their technological inferiors and brutally exterminate them: infact, Gauthier explicitly affirms the “rationality” (and therefore morality) of doing so(231-2). The h?robustH mutual-benefit individualist has a morality which simply isn’tconducive to creating the sort of mutual trust which would seem to provide theright climate for optimum social cooperation.Abstract” mutual-benefit individualism may seem to save individualutility-maximizers the price of straightforward maximization due to nearlyubiquitous Prisoner’s Dilemmas in social interaction. But it presupposes acompletely unrealistic prescience on the part of self-interested individuals of howspecific Prisoner’s Dilemmas will arise in the rough-and-tumble of social interactionand therefore fails to be motivating for actual individuals. “Robust” mutual-benefitindividualism, on the other hand, succeeds in motivating actual individuals to actaccording to their best judgment of their own self-interest, but only at the price ofallowing collective outcomes to reflect the price of straightforward maximization inmany Prisoner’s Dilemma situations.While morality is indeed justifiable only as a cooperative scheme for mutualadvantage, I do not see how the mutual-benefit individualist can, in the end, launcha scheme of social cooperation without also having to engage in cost-benefitcalculations which separate individuals’ (non-measurable) contributions to thesuccess of any such scheme from the share of benefits each individual gets out of thetotal cooperative surplus resulting from this scheme. There is no credible way inwhich the avoidance of non-cooperative behavior in potential Prisoner’s andAssurance Dilemma situations could get systematically credited to specific265individuals, and the cooperative payoff as a result of such avoidance could getapportioned among particular cooperators according to how well off each one ofthem had been if neither cooperative nor non-cooperative interaction had takenplace. But this means two things. It means giving up on the idea thatcontractarianism could somehow ensure moral compliance by harnessingindividual utility-maximizers’ self-interest and could thus make up for deficits inindividuals’ moral motivations zoithout external sanctions. And it also means thatthere is really no way around ranking the desirability of alternative systems ofdistributing the payoffs from social cooperation according to whether they maximizesome plausible aggregate satisfaction of individuals’ interests.3.5.3 Limits to rational disagreement about libertarianism§1 In the debate between egalitarian and libertarian liberals, the key issue - forconsequentialists- is whether almost exclusive reliance on the market for social andpolitical decision-making will, ultimately, meet the consequentialist well-being testbetter than an attempt to justify such decisions by an appeal to a hierarchy of moralprinciples. The core thesis defended by such utilitarian libertarians as, for example,Hayek, Friedman or Hospers, is captured well by N.P. Barry:[T]he co-ordinating mechanisms of the market and the incentives structureprovided by the institutions of the private property system always produce,accidentally, more beneficial outcomes than all known alternatives. (20)In other words, utilitarian libertarians defend capitalism, the widely dispersedownership of the means of economic production under competitive marketconditions. Contractarian libertarians do so as well, but they seek to first justify theright of individuals to use their own minds and bodies as they choose as long asthey do not thereby make others worse off, and then defend capitalist economicarrangements as flowing from this right. Utilitarian libertarians, in contrast, defendcapitalist economic arrangements by a direct appeal to their beneficial consequencesas compared to any alternative economic arrangements. They argue that capitalism,266because it tends to reward work with socially beneficial effects, provides maximumincentives for such work and therefore maximizes socially beneficial effects. Itmakes sense to equate justice with market outcomes, they argue, both becauseultimately only markets can solve large-scale commensuration problems, and alsobecause rewarding creative or conscientious work rewards desert in some plausibleform. When such libertarians argue that the harsh justice of the market” shouldbe left undisturbed, they imply, in effect, that capitalism, or at least some essentiallycapitalist form of economic arrangements, has the potential to capture enough ofthe pattern of peopl&s expectations and sentiment about justice to becomeacceptable to them as just arrangements and to enlist their sense of justice insupport of these arrangements.How plausible is this thesis? I cannot delve into the details of the utilitarianarguments offered by Narveson, Hayek and other libertarians. Nevertheless, Icannot neglect their claims entirely, because if they were true, then aconsequentialist moral framework would require us to equate social justice largelyor even exclusively with free market outcomes and whatever politicalarrangements are required to uphold the freedom of the market. To cast some doubton the claims of utilitarian libertarians, I can only suggest here the complex natureof the pros and cons that would have to be weighed to decide which preciseeconomic arrangement will best promote well-being. But it is fairly clear thatleaving all distributive matters up to real-life (as opposed to highly idealized)markets will not achieve this goal.Capitalism has many consequentialist advantages. Countless economicdecisions are coordinated without any need for an expensive, corruptible centralbureaucracy. It saves us from having to balance the demands of conflicting justmaking criteria or to commensurate the value of different goods for differentpeople. When all goes well, it provides incentives to creativity, risk-taking andsocially beneficial effort of many kinds, and generally there is at least a very rough,267partial correlation between people’s contributions and rewards. Capitalismharnesses some of the the strongest human motives, notably greed and fear, in theservice of economic goals and thereby liberates staggering productive energies. Inmany countries, capitalism (accompanied by social programs) has provencompatible with a high degree of political freedom combined with relativelywidespread prosperity.But capitalism seems clearly guilty of major evils as well. It is true thatcapitalist markets spare us complex balancing problems, but only at the expense ofmaking effective preference the sole criterion of who gets what - which means anypreference backed up by money. Capitalism tends to overproduce luxury items forthe wealthy, often without providing some of life’s essentials for others. It createschronic tensions as extreme wealth and poverty exist side by side. It has led to thepredation of third-world countries by multinational corporations. There is theoveremphasis on often rather joyless material consumption together with wasteand discontent resulting from artificially created desires. There is the concentrationof economic power which frequently translates into political power and privilege.There are the alienating conditions of work for many, the reinforcement of some ofthe worst traits of human personality and the erosion of community bonds. There isthe emotional upheaval and pervasive climate of insecurity associated with therelentless pace of change under capitalism. There is reason to fear that the sameproductive energies whose liberation makes capitalism seem so beneficial may,before long, overwhelm the earth’s ecosystem.Obviously, all these points would need extensive empirical support in orderto allow us to decide on the best economic system from a consequentialist point ofview. But it is surely clear that the choice of the best economic system is far morecomplex than many utilitarian libertarians would have us believe. Economicexperience has not shown, despite Hayek’s slippery-slope “road to serfdom”arguments, that judicious redistributive taxation undermines people’s incentive to268risk their capital in pursuit of wealth. Even if, at some point, taxation underminesthe incentive to invest and stifles the initiative of the enterprising to work hard, itcan also quite plausibly be argued, purely on grounds of economic efficiency, that agreat deal of inequality gives a massive headstart to the offspring of the well-to-dowhile demoralizing the downtrodden and depriving them of both motivation andopportunity to contribute their best to society. And there are other complexproblems faced by the utilitarian libertarian?s defense of capitalism. Keepingcompetitive markets intact over time is not simply a matter of allowing capitalistincentives to work their ?invisible hand?? magic, but requires a powerful enoughstate apparatus to prevent price collusions, monopolies and environmentalspillover effects, to promote wide dissemination of accurate product information,and to maintain an infrastructure to keep transaction costs low. Large corporationsmay amass such wealth and power that it becomes often difficult for governmentofficials to preserve their integrity, rather than promoting the interests of capitalistsin exchange for various sorts of direct and indirect forms of support. I will notpursue these difficult issues further here. It is a fixed point among my consideredjudgments, however, that something far short of libertarian property rights willprovide the optimum consequentialist mix of private and public control of themeans to production and the profits flowing from them.§2 Although the case for libertarian liberalism strikes me us ultimately notpersuasive, this is not to deny the importance of the libertarian?s cherished freedomto claim ownership over one?s own body and mind and to exercise whatever talentsor effort one can marshal in a market. There seems to be no convincingly principledway for liberals to delimit a private sphere of action for individuals once thelibertarian?s way of delineating this sphere is rejected. When asked what degree ofinterference with a person?s actions is justified for the sake of maximizing someindex of aggregate freedom or of some other important value, egalitarian liberals269will propose a variety of domain-restricted principles in an effort to strike areasonable balance in the light of weighing consequentialist pros and cons in thewidest possible reflective equilibrium. This method of generating political principlesis, of course, afflicted with enormous imprecision, presupposes large amounts ofinformation and goodwill for its proper application, and usually permits a widerange of equally rational alternatives. At the second-order level of moral reasoning,actual decision-makers in legitimate positions of power will have to select the mostpolitically opportune principle from among all those which pass the first-order tests.Apart from giving us the standard liberal core of political and civil rights,employing this method will often have an air of sheer arbitrariness which is likelyto give rise to controversy, especially among those who understand politicalmorality in much simpler terms.Nevertheless, I think the libertarian way of delineating a private sphere ofaction for individuals is rationally untenable. It seems clear that Narveson’scontractarian libertarianism has no credible potential to capture enough of thepattern of people’s expectations and sentiment about justice to become acceptable tothem as defining a just social arrangement and to enlist their sense of justice insupport of it. Utilitarian defense of libertarianism consists essentially in extollingthe consequentialist virtues of capitalist economic arrangements over allalternatives, but the consequentialist case for putting some restraints on theunfettered workings of capitalist markets seems to me overwhelming, especially ina world threatened by ecological disasters. The unfettered workings of free marketscould lead to a situation where the effective exercise of the freedom whichlibertarians so treasure, the freedom of individuals to use their own minds andbodies as they choose as long as they do not thereby make others worse off, couldbecome restricted to a relative minority of wealthy individuals.Freedom to use our own minds and bodies as we choose as long as we do notthereby make others worse off presupposes conditions which allow us to develop270our minds and bodies in the first place, and a social framework within which thisfreedom can be effectively exercised. Providing these conditions and this frameworkin a substantive way for the greatest number of individuals will require restrictionson the very freedom for whose sake they are to be provided. This paradox, perhapsappropriately called “the libertarian’s paradox”, seems to me to be the crux of thewhole dispute between libertarians and egalitarians. Since both egalitarians andlibertarians agree that freedom to exercise one’s talents is valuable, and sinceexercising one’s talents presupposes conditions which permit them to be developedin the first place, some moderate egalitarianism combining an economic safety netwith competitive markets would seem to me rationally superior, from aconsequentialist perspective, to any pure libertarianism or a radical egalitarianextreme. But there are many conflicting compromise principles of this sort. Whereexactly the proper balance lies must be judged for a given society at a given time andwill surely leave immense room for rational disagreement.27 1SUMMARY AND CONCLUSIONSWhat accounts for the immense disagreement about social justice amonginformed people of moral goodwill - and how could at least some of it be resolved?To approach these questions intelligently, in the face of the recentproliferation of conflicting theories of justice, I proposed to explore the possibility of“rational” disagreement and identified some plausible sources and limits of rationaldisagreement about justice. Let me now sum up what I think has beenaccomplished.I defined rational disagreement as disagreement about conceptual, normativeor empirical issues which are either currently irresolvable (there are empirical ornon-basic conceptual or normative questions at issue which additional information,clearer reasoning or conceivable methods may settle eventually) or fundamentallyirresolvable (there are basic conceptual, basic normative or intractable empiricalquestions at issue which no amount of accessible information, clearer reasoning orinnovative methods will ever settle).Before this idea of rational disagreement could be applied to the problem ofsocial justice, parameters of relevance had to be established. I analyzed “justice” asfundamentally a dual-purpose concept. The first purpose of this concept, at the levelof feeling injustice, is to give voice to the claims which individuals or groups,backed by their sense of justice, want to make to a share of social resources. Thesecond purpose, at the level of doing justice, is to articulate an appropriate politicalresponse to the pattern of expectations and sentiment about justice in a society.The sense of justice has an emotive and a cognitive component. Thecognitive component of the sense of justice allows individuals to assume aperspective beyond mere self-interest in pressing their claims and to recognize a(culturally variable) set of criteria such as desert, need, equality, rights and good-faithexpectations as justifying other people’s analogous claims to a share of social272resources. There is generally a vast range of logically possible options for designingsocial institutions to respond to the pattern of expectations and sentiment aboutjustice in a society. An important task of justice theorizing consists inconceptualizing and rationally ranking these options in a way that, ultimately, canguide political decision-makers.I insisted on a somewhat neglected distinction between theories of justiceand social ideals. Just as there can be many equally rational conceptions of the goodlife, so there can be many equally rational visions of the ideal society. Claims aboutwhat justice requires are not mere claims about what this or that persons vision ofthe ideal society requires. Trying to pass off one of many rationally permittedvisions of the ideal society as a theory of justice may well contribute to mischiefcausing expectations and exacerbate the clash of dogmatisms.What is a theory of social justice? I argued that a proposal for the distributionof valued things in society must satisfy two criteria to count as a theory of justice:(1) it must propose a politically feasible response to the conflicting claims to socialresources which different people (as individuals or groups) make in pursuit of theirwell-being and which they back up with their sense of justice; and (2) this proposedresponse to the pattern of justice-based claims must be explicitly and credibly -though not, of course, necessarily with success - argued as having a sufficient degreeof rational superiority over rival proposals so that informed people of good will(insofar as they are interested in finding publicly defensible rules of socialcooperation) cannot reasonably reject it. Without meeting this second condition, aproposal for the distribution of valued things can immediately be consigned to thescrap heap of social ideals.I then proposed two necessary, though admittedly quite vague, conditions forthe adequacy of a theory of justice: (1) the theory must have sufficient normativedeterminacy to provide guidance to political decision-makers, at least in the form ofrecognizable second-best approximations; and (2) it must respond to the pattern of273expectations and sentiment about justice in such a way as to have credible potentialto enlist enough people’s sense of justice in support of this theory to make it theeventual basis for a reasonably stable social consensus about the just distribution ofgoods. But offering necessary and sufficient conditions for an adequate theory ofjustice requires yet another condition: (3) An adequate theory of justice must, at leastin the minds of informed people of good will, uncontroversially meet the first twonecessary conditions to a greater degree than rival theories, or else it must haveadditional innovative features which will make it superior to its alternatives.We currently have many theories of justice, but none, I think, whichplausibly satisfies the above criteria of adequacy. Political decision-makers who seeknormative guidance from contemporary theories of social justice will discover aperplexing array of competing ideas, a quagmire of abstractions, a continuation ofreal-world conflicts about justice carried on by other means.It is clear that a theory of justice must meet stricter conditions thansomeone’s pipe dream of the ideal society. But why the above criteria? Aren’t webegging the question against those who want to identify the ‘correct’ theory ofjustice in radically different ways - for example, by intuiting a realm in whicheveryone’s just distributive share exists in its perfect Platonic form?Obviously, claims about the possibility of rational disagreement about justiceare non-trivial only if reason can define limits to it. Unless there can be somerationally compelling way to rule out many widespread ideas about the ultimatenature of morality, the judgments of philosophers about justice are really no morecompelling than those of people shouting from the barricades.On the one hand, I felt strong pressure to endorse a moral framework capableof setting non-trivial limits to rational disagreement about justice; but on the otherhand, I was sensitive to the objection that making strong enough moralcommitments to set such limits, in the face of intractable metaethical disputesamong people with impeccable rational credentials, would seem dogmatic in the274context of my basic project of identifying plausible sources of rational disagreementabout justice. In the face of this dilemma I decided to accept a form ofconsequentialism and to use four lines of defense against the above objection: (1) bymaking an (admittedly very truncated) effort to suggest how the widest possiblereflective equilibrium of relevant considerations is ultimately likely to converge onsome broadly consequentialist framework among all prima facie plausible metaethical options; (2) by admitting the tentativeness of my claims about the likelihoodof such eventual convergence and allowing a reading of my findings about thelimits of rational disagreement as presupposing a broadly consequentialist moralframework; (3) by pointing out that a number of plausible sources of rationaldisagreement within consequentialism are likely to emerge in analogous form inmany other moral theories as well; and (4) by suggesting that the context-dependentflexibility of the form of consequentialism I adopt offers greater prospects forrealizing Rawls’ hope of devising “principles of accommodation betweenmoralities” than plausible alternatives.Canvassing prima fade plausible meta-ethical options in an attempt toestablish the widest possible reflective equilibrium of relevant considerations, I wasled to adopt a constructivist, non-foundational, distribution-sensitive form ofconsequentialism. This moral framework is constructivist because it conceives ofmorality as a social pressure system constructed by human beings for the purpose ofcreating sufficient uniformity and thus predictability in people’s behavior toestablish the trust and security required for mutually beneficial interaction.1It isnon-foundational because it conceives of the justification of moral principles asshowing them to be supported by a web of cohering considerations rather thandeduced in a non-circular fashion from independently given foundations. It is aform of (indirect) consequentialism because it regards moral reasoning as involving1 Cf. Haslett (1987), p. 77. I paraphrase his felicitous formulation here of the basic rationale forhaving a morality at all.275the task of making it plausible that following certain principles will actually havethe consequence of promoting well-being. And it is distribution-sensitive, because itregards the goal of promoting well-being to involve both aggregative anddistributive values. For a consequentialism of this kind,2 promoting the goal ofwell-being means ideally aiming at that combination of satisfied interests ofindividuals in their own well-being that all of us would converge on from anenriched perspective which involves giving every individual’s interest in his or herwell-being exactly the same consideration which each of us would give it if each ofus had this individual’s life to live and judged from this individual’s perspective.A consequentialist understanding of justice arises from the recognition thattaking seriously those claims to social resources which individuals or groups makein pursuit of their well-being and back up with their sense of justice is likely topromote the goal of well-being both directly and indirectly: directly, because people’sjustice-based claims are a good though fallible source of information about what willpromote their well-being; and indirectly, because ignoring these claims is likely tounleash destructive sentiments. But at least in complex pluralist societies, noparticularly smooth fit will generally be possible between, the function of the conceptof justice at the level of feeling injustice and its function at the level of doing justice.At the level of doing justice, then, what is inside the scope of rationaldisagreement about justice and what is outside? I found reason to be hopeful thatmany issues dividing different conceptions of justice can indeed be settled at thelevel of fundamental theory with the rational means currently at our disposal forresolving the conceptual, normative and empirical points at issue. Such issues are,therefore, outside the scope of rational disagreement, as I defined this concept. Itneeds to be remembered, however, that by excluding some issue from the scope ofrational disagreement I do not imply that there is no longer any dispute about itamong reasonable people of good will. All I mean to imply is that, from the2 To be labelled simply “consequentialism”in what follows.276perspective of the widest possible reflective equilibrium I have been able to achieve,I tentatively judge the issue to be resolvable by close attention to currently availableinformation, arguments and methods. I cannot claim finality for my judgments. Anissue which I optimistically classify as decidable by rational means currently at ourdisposal may, from the perspective of an even wider reflective equilibrium, turn outto be rationally undecidable after all, or decidable in a very different way than mylimited understanding allowed me to see.On the basis of my inquiry, which issues promise to be rationally decidable,and which do not, when it comes to social justice? In fact, this question raises twodistinguishable concerns: first, how to define non-trivial limits to rationaldisagreement to keep my position from collapsing into skepticism; and second, howto identify plausible sources of rational disagreements about justice.Plausible limits to rational disagreement about justice (within consequentialism)There are several major issues about which I tried to establish at least a somepresumption of rational decidability so as to justifiably exclude them from the scopeof rational disagreement. The following summary is meant to be at best suggestiveof the general lines of reasoning that seem to me cogent.1. The issue between basically liberal and non-liberal conceptions of justice:There are two fundamental views about the best social order. The liberal view isthat social institutions do not exist to enforce worthy ways of living, but primarily topromote the freedom of individuals to live as they prefer, within a system ofconstraints designed to avert conflict while maintaining individual freedom to thegreatest possible extent. The non-liberal view is that social institutions shouldconstrain individuals to pursue a fairly narrow spectrum of ways of living known tobe worthy in advance, regardless of the preferences of those individuals. Theproblem of justice will take different forms for liberals and non-liberals. If the stateexists to enforce a limited spectrum of ways of living which can be known, with277sufficient confidence and precision, to be worthy, then we have available a more orless well-defined blueprint for determining just distributions. If, however, the stateshould not coerce individuals to pursue worthy ways of living, then the problem ofjustice will involve the complexities discussed below.I tried to establish some presumption that, within at least moderatelyprosperous pluralist societies, the dispute between basically liberal and non-liberalviews about the best social order is not a disagreement between equally rationalconceptions of the ideal society. Broadly egalitarian forms of liberalism can bejustified as the best social order on consequentialist grounds, provided that we arelooking for a basic social arrangement which can be publicly defended as reasonableto individuals with different ideas of the good. This justification depends on the factthat a liberal order leaves greater social space for a wide variety of ideas of the goodthan any other alternative. It does not, however, depend on denying that non-liberal orders can embody cultural virtues which are likely to succumb tocompetitive pressures in an egalitarian-liberal order, nor on denying that, from thenarrow perspective of the self- or group interest of a member of a non-liberalculture, egalitarian liberalism may well not be the best social order.The defense of a liberal order I have in mind proceeds as follows: (1) We canknow that there is a diversity of possible values for human beings which no one lifecan encompass, and therefore that there are many equally good ways of living ratherthan one particular way which is best for everyone. (2) Every non-liberal proposalfor the best way of living which the state should enforce encounters rival proposalsfrom other non-liberals, and there exists no non-partisan rationale for choosingamong an immense array of non-liberal options. (3) In the face of perfectly rationaldiversity about what is of value in life, one basic political scheme stands out in thechaos of competing schemes as offering a way of mutual accommodation whileminimizing mutual coercion. This is the proposal to set up social institutionsaround the idea of permitting individuals to seek their own good in their own way278(and equipping them through education to do so wisely), within a system ofconstraints designed to avert conflict while maintaining individual freedom to thegreatest possible extent. There are many different systems of constraints thatarguably satisfy this admittedly very general idea equally well. Although a basicallyliberal social order can take on a variety of specific shapes, a commitment to such anorder takes us a long way from skepticism about justice and makes talk about“rational disagreementTT meaningful.Within a consequentialist framework, this justification for a basically liberalsocial order will, of course, be most persuasive given already fairly pluralist societiesenjoying at least moderately favorable economic prospects.3Some additionalconsequentialist reasons for a basically liberal social order are: the desire to avoid acivil war between competing non-liberal forms of perfectionism - liberalism offeringperfectionists of many different stripes probably the most viable second-best terms ofaccommodation; the notorious fallibility of government paternalism; the character-building effects of making one’s own choices and mistakes; and the innovationsarising out of allowing experiments in living. Therefore, there is a sufficientlystrong presumption against non-liberal conceptions of justice to exclude them fromthe scope of rational disagreement about justice, at least in already fairly pluralistsocieties enjoying moderately favorable economic prospects.2. The issue between directly and indirectly responsive conceptions of justicein liberal societies: Directly responsive conceptions respond to people’s justice-basedclaims by interpreting justice as requiring redistributions of certain important goods3 For societies facing massive starvation and illiteracy problems, I would want to raise very differentconsiderations, given the quite likely case that liberal-democratic institutions would further add tothe political and economic chaos. For culturally homogenous societies which embody a non-liberal but“benign” form of perfectionism that has deeply shaped the self-understanding of individuals in thatsociety (e.g. certain indigenous tribal communities), the consequentialist case against promoting radicalsocial change is also very strong, and the argument I make for liberalism would have be qualifiedconsiderably. Moreover there is a real possibility that, within a few decades, population pressureswill not leave space for liberal democracies at all, as ten billion people lead a cramped existence on anecologically ravaged planet. Liberal justice theorizing may become a lot simpler in this situation, oraltogether irrelevant. But I cannot pursue these matters here.279from those who have too much to those who have too little, according to a standardof ideally just individual holdings of these goods. Indirectly responsive conceptionsrespond to people’s justice-based claims by interpreting justice as requiring a schemeof basic social institutions organized in a particular way, according to a standard ofan ideally just pattern of overall distributive outcomes or procedures.The question of which of these two basic conceptions of justice is superiorcan, I think, be rationally decided in favor of directly responsive conceptions. Ilooked at two prominent examples of indirectly responsive conceptions: JohnRawis’ theory of justice and the libertarian theory developed by Narveson onGauthier’s contractarian foundations. These two conceptions, although clearly notthe only possible indirectly responsive conceptions, are probably the initially mostplausible current examples of their kind. Rawis’ theory specifies a distributiveoutcome, libertarian theories a distributive process, which society’s basic structuremust be designed to uphold. I argued that Rawis’ theory, and probably any otheroutcome-specific indirectly responsive theories of justice that might be proposed,can be ruled out for three main reasons:(1) The method of wide reflective equilibrium, if directly applied to thequestion of which overall distributive outcome society’s basic structure shouldachieve, is too indeterminate to support more than a broad range of stronglyconflicting principles. (2) A distributive blueprint for society’s basic structure mustincorporate massive simplifications of commonly invoked just-making criteria -most notably the idea of judging justice by maximizing the expectations of oneparticular representative social position- so as to end up having little or nopotential to harness enough people’s sense of justice in its support to make thisblueprint the basis for a stable social consensus about justice. (3) Despite massivesimplifications such a distributive blueprint is not likely to carry much normativepayload in the end, when it comes to giving useful guidance to would-be reformersof society’s basic structure because it cannot specify, with any precision, a process to280decide the division of responsibilities among particular institutions for achievingthe required outcome.Libertarian theories can also be ruled out, for reasons to be outlined below.Therefore, only directly responsive conceptions of justice would seem to fall withinthe scope for rational disagreement.3. The issue between libertarian and egalitarian conceptions of justice:Libertarian conceptions of justice require the state to uphold a free market and toenforce a right to non-interference with the property of individuals which theyacquire by using their own minds and bodies as they choose without thereby makingothers worse off. In contrast, egalitarian conceptions of justice require the state toprovide a social framework of universal access to a more or less extensive stock ofgoods which constitute essential means to a successful pursuit of almost any idea ofthe good life, such as access to proper nutrition, education, medical care,employment, housing, legal aid, child care and many other possible goods.Currently available information and arguments are probably sufficient tosettle the issue between libertarian or egalitarian conceptions of justice in favor ofbroadly egalitarian conceptions. Ruling out non-constructivist moral appeals tonatural rights, we looked at two prima facie credible constructivist attempts toground libertarian property rights: Gauthier’s contractarian argument and theconsequentialist arguments of utilitarian libertarians like Hayek. Although theseseem to me the initially most plausible versions of libertarianism, I do not deny, ofcourse, that other possible versions would have to be examined as well.Narveson’s attempt to ground the libertarian right to non-interference in aprinciple of rational choice proposed by Gauthier (a principle that everyone issupposed to be equally ready to accept, whatever the relative strength - within widelimits- of his or her bargaining position), succeeds in evading the contractarianTsdilemma only at the expense of becoming indefensible on at least three relatedgrounds:281(1) There is reason to believe that no general principle exists which would beequally acceptable to strong and weak bargainers prior to bargaining. Even if thenormatively rather toothless idea behind the Lockean proviso (that no rationalperson would object to others making themselves better off if doing so hadabsolutely no detrimental effect on them) is a principle of rational choice, it is clearthat Gauthier’s particular spelling-out of this proviso cannot be regarded as asufficiently uncontroversial interpretation of it. Narveson’s attempt to connectGauthier’s interpretation of the Lockean proviso up with strong libertarian noninterference rights failed precisely because there is every reason to think that neitherthe strong nor the weak would find it in their self-interest to agree to this principleregardless of the specifics of a particular bargaining situation.(2) Despite the seeming precision of Gauthier’s approach at the abstract level,he abstracts from the particulars of real life situations to such an extent that histheory gives no useful normative guidance. Even if justice is equated to theoutcome of bargaining among individual utility-maximizers, most real-life disputesabout justice involve far too complex variables for ex ante generalizations about themost rational agreements from the point of view of self-interested bargainers. (Onthis question of normative guidance, Gauthier seems to equivocate between what Ihave called “abstract and “robust” mutual-benefit individualism.)(3) Even if Gauthier’s contractarian theory of “justice” had a usefulnormative payload, it would have no credible potential to enlist enough people’ssense of justice in support of this theory to make it the eventual basis for awidespread social consensus about justice. It fails to respond to the pattern ofexpectation and sentiment about justice in three ways: it systematically splits thecooperative surplus arising from each cooperative interaction in such a way thatexisting inequalities will be maintained or even exacerbated; it allows the strong tobetter themselves by undermining the livelihood of the weak whenever thebenefits of doing so are greater than cooperative interaction; and it explicitly affirms282the rationality for those with “superior technology” to exterminate the weak whentheir doing so outweighs the potential gains of cooperation (231-2). Because the self-interested calculations of GauthierTsindividual utility-maximizers may well makesuch actions rational and therefore morally required, Gauthier’s theory seemsspectacularly unsuited for encouraging a real sense of security and trust amongpeople and thus holds no promise for accomplishing morality’s main job.In contrast to contractarian libertarians, utilitarian libertarians rely on adefense of the libertarian right to non-interference which tries to show that thisright passes the consequentialist well-being test required for rights after all. Theyequate the exercise of this right with exercising one’s talents and ambition in a freemarket, and extol the ability of a fairly unbridled capitalist economic system topromote human well-being more effectively than any alternative. I have brieflysketched my reasons for thinking that their arguments cannot justify the libertarianright to non-interference, but can establish at best a strong presumption for a marketeconomy supplemented by redistributive social programs. But of course, suchcorrective redistributions tailored to the situation of particular individuals willopen the floodgates to competing claims once more (which can only be resolved indomain-restricted ways). Although consequentialists must acknowledge that thefreedom of individuals to use their own minds and bodies to make themselvesbetter off is indeed an important value, exercising this freedom presupposes meansand conditions which allow individuals to develop their minds and bodies in thefirst place, and therefore requires a social framework of substantive equality, or atleast limits to inequality, of access to these means and conditions. Hence protectingthe universal exercise of the very freedom that utilitarian libertarians appeal to (aswell as realizing other important values) requires restrictions on this freedom. Itherefore conclude very provisionally - not having examined all libertarianarguments in sufficient detail - that the scope of rational disagreement can probably283be effectively narrowed to liberal-egalitarian conceptions of social justice whichpropose directly responsive principles.4. The issue between directly responsive conceptions of justice that are ofsociety-wide scope and those that are of domain-restricted scope: Conceptions ofsociety-wide scope propose a master principle of justice which (1) is intended toprovide necessary and sufficient conditions for determining every individual’s justshare of every type of good at any given moment in time throughout society; and (2)individuates a fairly static list of goods whose distribution it regards as the concernof justice, along with a fairly static list of criteria a person must meet to qualify forcertain amounts of certain goods. In contrast, conceptions of domain-restricted scopepropose a plurality of different (and more or less easily modifiable) master principlesof justice of widely varying characteristics for a more or less large plurality ofrestricted and often changing domains. Conceptions of justice whose principles areof domain-restricted scope may vary greatly in the degree of their ambition to pullup people’s conflicting intuitions about justice in particular cases to the level ofprinciples. A complete collection of all such domain-restricted master principles, ifthey could ever actually be collected, would of course add up to a sort of masterprinciple of society-wide scope, even though its internal complexity and lack ofpriority ordering would not allow us to determine every individual’s just share atany given time without resorting to extremes of intuitive balancing which wouldrender it effectively indeterminate.I think that the issue between directly responsive conceptions of justice ofsociety-wide and of domain-restricted scope is decidable, by currently availableinformation and arguments, in favor of domain-restricted conceptions. Tounderstand the reasons why, however, we need to draw upon a further layer ofdistinctions. I will briefly recapitulate these distinctions, since the reasoning for myconclusion depends crucially on keeping them clear.284Conceptions of justice, we said, are either intuitionistjc or non-jntuitionistic.Intuitionistic conceptions of justice propose a plurality of relevant criteria foridentifying a just distribution of a particular bundle of goods, without specifying analgorithm for assessing the relative importance of the various criteria when theygive rise to conflicting claims. Non-intuitionistic conceptions propose either a singlecriterion as ultimately relevant for identifying a just distribution of a particulargood, or else they specify a lexical ordering as a method for avoiding the need forintuitive balancing of criteria in cases of conflict.Single-criterion (directly responsive) principles of justice of society-wide scopespecify only one criterion which individuals must meet to qualify for a particulardistributive share. But ultimately there are no plausible criteria of this kind. Criteriawhich might seem to meet this condition, like needs or (non-libertarian) rights, areinternally complex and require intuitive balancing after all, as well as beingunacceptably lopsided as a credible response to individuals’ justice-based claims to avariety of different goods in different social contexts. Nor can we go very far inaligning a plurality of different just-making criteria with this variety of goods andthen finding a plausible lexical ordering for different criteria, because such a lexicalordering will appear unacceptably rigid in the face of trade-off opportunities whichare clearly required on consequentialist grounds. So all prima fade plausible masterprinciples of society-wide scope must either be directly responsive and intuitionistic(or if non-intuitionistic, then highly domain-restricted). Now there exists anoverabundance of conflicting proposals for society-wide principles of theintuitionistic kind, and they are easy to make up, according to one’s favorite socialutopia. Each would seem to face a standard set of next-to-insuperable obstacles: thejustification problem, the priority problem, the aggregation-distribution problem,the index problem as well as additional commensuration problems. Such obstaclescan be overcome, if at all, only in very small, non-pluralist societies.285There is a strong presumption, therefore, that only domain-restricted liberal-egalitarian principles of justice (which in wide domains will be intuitionistic and infairly narrow domains may be relatively non-intuitionistic) fall within the scope ofrational disagreement in complex pluralist societies. Only domain-restrictedprinciples can factor the balancing problems into small enough parts to keep themmanageable while having a hope of responding to the pattern of domain-relativeexpectations and sentiments about justice in a way to enlist enough people’s senseof justice in support of such principles.Under conditions currently prevailing in many societies, then, ourconsequentialist framework narrows the scope for rational disagreement aboutjustice to a basically liberal-egalitarian social order. It is clear that such a social orderwill pass the consequentialist well-being test only if it strikes a balance between twoimportant values which are irreconcilably in tension. It must respect the value ofproviding secure opportunities for everyone to gain reasonable access to thenecessary means for realizing almost any legitimate idea of the good (including,most importantly, sufficient means to meet course-of-life needs by instituting aneconomic safety net). And it must also respect in large measure the freedom ofindividuals to use their own minds and bodies as they see fit and to pursue theirown intentional tracks through life. But within such admittedly broad parameters,attempts to find principled responses to people’s justice-based claims must consist inseeking support for domain-restricted principles by promoting stable expectationswithin restricted domains - rather than in multiplying ideals of everyone’s justindividual share of total social resources or in articulating ambitious ideals of theoverall distributive outcome to which society’s basic structure must conform.Our consequentialist framework leaves considerable room, at the level ofabstract theory, for context-dependent justice theorizing and for rationaldisagreement which must be resolved at the second-order level of politicalnegotiation and compromise.286Plausible sources of rational disagreement about justice (within consequentialism)I shall now recapitulate, in the briefest of terms, some plausible candidates forthe status of rational disagreements about social justice. Deeper insights into what is,admittedly, a still rather unruly collection of problems will emerge only after a greatdeal of clarifying dialectic.4Although my account of these problems was developedagainst the conceptual background of a form of consequentialism, they may ofcourse also arise, in recognizably analogous forms, for theories of justice embeddedin alternative moral frameworks.1. The criteria definition and inclusion problems: At the level of feelinginjustice, there is an unruly family of very imprecise evaluative considerationswhich people invoke in making claims in pursuit of their well-being while beingwilling to allow others to make claims based on the same considerations. Theseconsiderations include many (culturally variable and even intraculturally diverse)interpretations of desert, need, equality, rights and good-faith expectations. At thelevel of doing justice, difficult problems will arise — solved by different conceptionsof justice in different ways — about what considerations to include as just-makingcriteria in a principled response to the pattern of expectations and sentiment aboutjustice.2. The problem of feasible comprehensiveness: On the one hand, we want topull judgments about justice up to the level of principles; on the other, we cannotpitch the level of generality of these principles very high without emptying themboth of normative determinacy and of credible potential to enlist people’s sense ofjustice. Individuals construct their selves in the context of restricteddomains, and their justice-based claims are projections of such domain-embeddedselves. Further work is clearly needed to link up my very general remarks aboutdomain-restricted principles of justice with the viable domains in a particularRawis named the priority problem and the index problem.287society. Doing so would obviously be a highly complex, society-specific inquiry intoevolved and contrived institutions, practices, social roles, cultural groupings,economic development, geographic particulars and other contingent circumstances.3. Balancing problems: These problems include the priority problem (whatweights to assign to the various just-making criteria invoked by a theory wheneverresources are insufficient to completely satisfy them all); the aggregation-distribution problem (how to decide between satisfying a greater amount of justice-based claims and satisfying at least some of the most urgent justice-based claims ofall people); and various other commensuration problems (how to assess in practicaldetail, for all individuals in society, what redistributions are required from thosewhose holdings exceed their just share to those whose holdings fall short of it).Rawls’ method of judging the justice of society’s overall pattern of distribution fromthe perspective of a particular social position vastly simplifies commensurationproblems. It reduces many of them essentially to the index problem (finding anindex for different bundles of primary goods held by different people to comparetheir values). But even if this index problem could somehow be kept manageablefor indirectly responsive theories, Rawls would still face difficult commensurationproblems in deciding which elements of the basic structure to reform in which ways.While breaking the problem of social justice into domain-restricted principlesis, I argued, probably the only way to make the priority, aggregation-distribution andcommensuration problems manageable while yet maintaining credible potential toharness enough people’s sense of justice in support of these principles, such adomain-restricted, directly responsive approach to social justice will still face allthese problems to some extent, in addition to two new problems: the domaindemarcation problem (how to select viable domains which are sufficientlyindependent from neighboring domains for regulation by their own domainspecific principles); and the domain balancing problem (what to do when differentdomains compete for the same resources). These problems require contextual288perspective-seeking in reflective equilibrium, and will be rich sources of rationaldisagreement, at the first-order level of moral reasoning, about the demands ofsocial justice and those of morality in general.4. Various problems of first-order justification (within consequentialism):There can be rational disagreement about which bona fide principle of justice aninformed person of good will is actually obligated to comply with. The concept of abona fide principle of justice is central to my analysis. Bona fide principles of justiceare, in essence, all those proposal for balancing justice-based claims in a domain thatcould be reached by decision-makers who are guided by a thorough understandingof the distributive domain at issue and the desire to satisfy people’s justice-basedclaims in that domain to the greatest extent compatible with respect for theseparateness of persons.Bona fide principles are not equivalent to morally justified principles ofjustice - principles which are such that compliance with them is morally required. Abona fide principle can be regarded as morally justified only if it passes the test ofthose moral standards which are plausibly derivable from the form ofconsequentialism we have adopted. But usually a range of competing principles willpass this test relative to any given domain, and this range defines the scope for first-order rational disagreement about what fundamental moral theory requires in thisdomain. To identify a particular principle within this range, such that actualcompliance with it is morally required for a person, will generally take us beyondthe first-order level of moral reasoning, into the sphere of legitimation of aparticular principle by appropriate political decision-makers.Rational disagreements arise at several other points in the process of moraljustification. I adduced consequentialist reasons for thinking that the most plausibleway to interpret the demands of justice is as follows: “Justice requires that at leastone among the bona fide principles in some viable domain be implemented.” Ifjustice requires only that at least one among the bona fide principles in a domain be289implemented, and if a range of different bona fide principles is equally permissiblein most domains, it follows that first-order moral reasoning can identify thedemands of justice in a domain only with that degree of precision whichcorresponds to this range of equally permissible bona fide principles. It also followsthat the important concept of “injustice” (i.e. a violation of what justice requires)will be defined in a largely domain-relative way: as referring to a distribution that isincompatible with all bona fide principles within a given domain. To the extent thatdomains are not securely demarcated, therefore, something may be seen as aninjustice if domains are demarcated one way but not if they are demarcated another.Moreover, what justice requires in one domain may conflict with what justicerequires in another. And what justice requires may conflict with what moralityrequires, all things considered. Although the consequentialist rationale for thisstructure of moral justification is clear enough, in practice there will often be roomfor rational disagreement about whether an injustice has actually been committed,and if so, whether committing it is morally justifiable, all things considered.In developing my account I mentioned additional sources of rationaldisagreement and no doubt passed over other potential sources. I am conscious ofhaving unravelled only a modest part of the enormous complexities involved inany effort to overcome the current impasse in justice theorizing.The upsurge of justice theorizing since the publication of A Theory of Justicetwo decades ago has given us interesting social ideals. None of them is sufficientlycompelling, it appears, to serve as a promising basis for a social consensus about thedemands of justice within large, complex pluralist democracies. These theorieshave, however, taught us important lessons. A particularly important lesson seemsto me to be this: curbing conflicts about justice demands a vastly clearerunderstanding of the nature of rational disagreement than is currently available.Only a systematic focus on the sources of rational disagreement about justice defines290a line of effort which, I believe, may yet fulfill the sort of hope eloquently expressedby John Rawls:The hope is that.. .existing differences between contending political views canat least be moderated, even if not entirely removed, so that social cooperationon the basis of mutual respect can be maintained. Or if this is expecting toomuch, this method may enable us to conceive how....a public understandingcould arise consistent with the historical conditions and constraints of oursocial world. Until we bring ourselves to conceive how this could happen, itcan?t happen. (1985: 231)Conflicts about justice incite flaming passions throughout the contemporaryworld. To make sense of justice-talk in our time, to avoid both dogmatism andskepticism as well as the pragmatism of Machiavellian power politics, and to invokethe concept of justice in ways that will help mediate rather than escalate conflict, allthis requires a fuller recognition of the conceptual, normative and empirical sourcesof rational disagreement about justice. Such recognition will perhaps move humanminds from a state of moral certainty to that state of doubt which is the beginning ofwisdom. Such recognition can also fill the gap between opposing beliefs, producingan enriched perspective from which the need to negotiate compromises and to setnew conventions is better appreciated. A philosophy of justice which promotes anunderstanding of the deepest sources of rational disagreement and thereby producessuch an enriched perspective may well offer our best prospect for transcendingmany raging conflicts about justice, both within human societies and between them.291BIBLIOGRAPHYAckerman, Bruce A., Social Justice in the Liberal State, New Haven: YaleUniversity Press, 1980.Aristotle, Nichomachean Ethics, in: The Complete Works of Aristotle, Vol.Two, ed. Johnathan Barnes, Princeton: Princeton University Press,1985.Arneson, Richard J., “Introduction”, Symposium on Rawlsian Theory ofJustice, Ethics, Vol 99 No 4, Julyu 1989, pp. 695-711.Attig, Thomas, Donald Callen and R.G. Frey, ed., Social Policy and ConflictResolution, Bowling Green Studies in Applied Philosophy, Vol VI -1984.Baier, Kurt, “Justice and the Aims of Political Philosophy”, Ethics, Vol 99 No4, July 1989, pp. 771-790.Baldwin, R.W., Social Justice, New York: Pergamon, 1968.Barrow, Robin, Injustice, Inequality, and Ethics: A Philosophical Introductionto Moral Problems, Totowa, N.J.: Barnes & Noble, 1982.Barry, Brian, Political Argument, London: Routledge & Kegan Paul, 1980.Barry, Brian, The Liberal Theory of Justice, London: Oxford University Press,1973.Barry, Brian, Theories of Justice, Vol I of A Treatise on Social Justice,Berkeley: University of California Press, 1989.Barry, Norman P., On Classical Liberalism and Libertarianism, London:Macmillan, 1986.Benjamin, Martin, Splitting the Difference, Compromise and Integrity inEthics and Politics, Lawrence, Kansas: University Press of Kansas, 1990.Benson, Peter, “The Priority of Abstract Right, Constructivism, and thePossibility of Collective Rights in Hegel’s Legal Philosophy”, TheCanadian Journal of Law and Jurisprudence, Vol IV, No 2, July 1991,pp. 257-91.Berlin, Isaiah, Four Essays on Liberty, New York: Oxford University Press,1969.292Bird, Otto A., The Idea of Justice, New York: Frederick A. Praeger, 1967.Blackstone, W.T., and R.D. Heslep, eds., Social Justice and PreferentialTreatment: Women and Racial Minorities in Education and Business,Athens, Georiga: University of Georgia Press, 1977.Blocker, H. Gene, and Elizabeth H. Smith, ed., John Rawls’ Theory of SocialJustice, Athens: Ohio University Press, 1980.Bonjour, Lawrence, The Structure of Empirical Knowledge,Cambridge: Harvard University Press, 1985.Bradie, Michael, and David Braybrooke, Social Justice, Bowling Green Studiesin Applied Philosophy, Vol IV, 1982.Bradley, F.H., Ethical Studies, 2nd ed., Oxford: Clarendon, 1959.Brandt, Richard B., ed., Social Justice, Englewood Cliffs, N.J.: Prentice-Hall,1962.Brandt, Richard B., A Theory of the Good and the Right, Oxford: ClarendonPress, 1979.Braybrooke, David, Three Tests for Democracy: Personal Rights, HumanWelfare, Collective Preference, New York: Random House, 1968.Braybrooke, David, Meeting Needs, Princeton, N.J.: Princeton UniversityPress, 1987.Braybrooke, David, “Michalos’ Comments on Meeting Needs Disposed of”,Dialogue, Vol XXVII No 3, Autumn 1988, pp. 517-522.Brown, Grant, “Critical Notice of Jan Narveson, ‘The Libertarian Idea’,Canadian Journal of Philosophy, Vol 20, No 3, September 1990, pp. 417-447.Brown, Harold I., Rationality, London: Routledge, 1988.Buchanan, Allen, Ethics, Efficiency, and the Market, Totawa, N.J: Rowman &Allanheld, 1985.Buchanan, Allen E., “Assessing the Communitarian Critique of Liberalism”,Ethics, Vol 99, July 1989, pp. 852-882.Buchanan, Allen., Secession: The Morality of Political Divorce from FortSumter to Lithuania and Quebec, Boulder: Westview Press, 1991.293Campbell, Tom, Justice, London: Macmillan, 1988.The Canadian Journal of Lazy and Jurisprudence, Topic: Collective Rights, ed.Michael McDonald, Vol. IV, No 2, July 1991.Cauthen, Kenneth, The Passion for Equality, Totowa, N.J.: Rowman &Littlefield, 1987.Christian, James L., Philosophy, An Introduction to the Art of Wondering,5th ed., Toronto: Hold, Rinehard and Winston, 1990.Cohen, G.A., Karl Marx’ Theory of History: A Defence, Princeton: PrincetonUniversity press, 1978.Cohen, Ronald L., ed., Justice: Viezvs from the Social Sciences, New York:Plenum Press, 1986.Copp, David and David Zimmerman, ed., Morality, Reason, and Truth: NewEssays on the Foundations of Ethics, Totowa, N.J.: Rowman &Allanheld, 1985.Coval, S.C. and J.C. Smith, Law and Its Presuppositions, London: Routledge &Kegan Paul, 1986.Coval, S., J.C. Smith and Simon Coval, “The Foundations of Property andProperty Law”, Cambridge Law Journal, 45(3), November 1987, pp. 457-475.Cunningham, Frank, Democratic Theory and Socialism, Cambridge:Cambridge University Press, 1987.Dahl, R.A., Dilemmas of Pluralist Democracy, New Haven: Yale UniversityPress, 1982.Daniels, Norman, ed., Reading Rawls: Critical Studies of A Theory of Justice,New York: Basic Books, 1974.Daniels, Norman, “Wide Reflective Equilibrium and Theory Acceptance inEthics”, Journal of Philosophy, Vol 76, No 5, May 1979, pp. 256-282.Danielson, Peter, “The Lockean Proviso”, in: Con tractarianism and RationalChoice: Essays on David Gauthier’s Morals by Agreement, ed. PeterVallentyne, Cambridge: Cambridge University Press, 1991, pp. 99-111.294Danielson, Peter, “Closing the compliance dilemma: How it’s rational to bemoral in a Lamarckian world”, in: Contractarianism and RationalChoice: Essays on David Gauthier’s Morals by Agreement, ed. PeterVallentyne, Cambridge: Cambridge University Press, 1991, pp. 291-322.Danley, John R., “Liberalism, Aboriginal Rights, and Cultural Minorities”,Philosophy and Public Affairs, Vol 20 No 2, Spring 1991, 168-185.Davidson, Donald, Essays on Actions and Events, Oxford: Clarendon Press,1980.Day, J.P., Liberty and Justice, London: Wolfboro, 1987.Deutsch, Morton, Distributive Justice: A Social-Psychological Perspective,New Haven: Yale University Press, 1985.Dworkin, Ronald, “The Original Position”, in: Reading Rawls: Critical Studiesof A Theory of Justice, ed. Norman Daniels, New York: Basic Books,1974, pp. 16-53.Dworkin, Ronald, Taking Rights Seriously, Cambridge: Harvard UniversityPress, 1977.Dworkin, Ronald, “What is Equality? Part 1: Equality of Welfare”, in:Philosophy & Public Affairs, Vol 10 No 3, Summer 1981, pp. 185-246.Dworkin, Ronald, “What is Equality? Part 2: Equality of Resources”, in:Philosophy & Public Affairs, Vol 10 No 4, Fall 1981, pp. 283-345.Dworkin, Ronald, ‘Liberalism’, A Matter of Principle, Cambridge, MA:Harvard University Press, 1985.Elster, Jon, “Rationality, Morality, and Collective Action”, Ethics, Vol 96, No1, October 1985.“Equal Opportunity”, Special Issue of Social Philosophy & Policy, Vol 5, Issue1, Autumn 1987.Ethics, Special Symposium Issue on “Rationality and Morality”, Vol 96 No 1,October 1985.Ethics, Special Symposium Issue on “Rawisian Theory of Justice: RecentDevelopments”, Vol 99 No 4, July 1989.Ewin, R.E., Liberty, Community, and Justice, Totowa, N.J: Rowman &Littlefield, 1987.295Feibleman, J.K., Justice, Law, and Culture, Dortrecht: Martinus Nijhoff, 1985.Feinberg, Joel, Social Justice, Englewood Cliffs, N.J.: Prentice-Hall, 1973.Feinberg, Joel, Rights, Justice, and the Bounds of Liberty, Princeton, N.J.:Princeton University Press, 1980.Fishkin, James, S., Tyranny and Legitimacy: A Critique of Political Theories,Baltimore: Johns Hopkins University Press, 1979.Fishkin, James, S., Justice, Equal Opportunity, and the Family, New Haven:Yale University Press, 1983.Frankel Paul, Ellen, Fred D. Miller Jr. Jeffrey Paul and John Ahrens, EqualOpportunity, New York: Basil Blackwell, 1987.Friedman, Milton, Free to Choose: A Personal Statement, New York: HarcoutBrace Jovanovich, 1980.Gaiston, William A., Justice and the Human Good, Chicago: University ofChicago Press, 1980.Gauthier, David, Morals by Agreement, Oxford: Clarendon Press, 1986.Gauthier, David, ‘Why Contractarianism?”, in: Contractarianism andRational Choice, ed. Peter Vallentyne, Cambridge: CambridgeUniversity Press, 1991, pp. 15-30.Gauthier, David, “Rational Constraint: Some Last Words’, in:Contractarianism and Rational Choice, ed. Peter Vallentyne,Cambridge: Cambridge University Press, 1991, pp. 323-330.Gibbard, Allan, “Constructing Justice”, Philosophy and Public Affairs, Vol 20,No 3, Summer 1991, 264-279.Gibbard, Allan, Wise Choices, Apt Feelings, A Theory of NormativeJudgment, Cambridge, Mass.: Harvard University Press, 1990.Ginsberg, Morris, On Justice in Society, Ithaca, N.Y.: Cornell University Press,1965.Goldman, Alan H., “Justice and Hiring by Competence”, AmericanPhilosphical Quarterly, Vol 14 No 1, January 1977, pp. 17-28.296Goldman, Alan, Justice and Reverse Discrimination, Princeton: PrincetonUniversity Press, 1979.Goodin, Robert B., Reasons For Welfare: The Political Theory of the WelfareState, Princeton, N.J.: Princeton University Press, 1988.Goodin, Robert E., and Andrew Reeve, eds., Liberal Neutrality, London:Routledge, 1989.Gordon, Scott, Welfare, Justice, and Freedom, New York: ColumbiaUniversity Press, 1980.Graham, Gordon, Contemporary Social Philosophy, New York: BasilBlackwell, 1988.Grant, George, English-speaking Justice, Toronto: Anansi, 1974, 1985.Greene, Thomas H., Comparative Revolutionary Movements: Search forTheory and Justice, 2nd ed,, Englewood Cliffs, N.J.: Prentice-Hall, 1984.Griffin, James, “Some Problems of Fairness’, in: Ethics, Vol 96 No 1, October1985, pp. 100-118.Gross, Barry, ed., Reverse Discrimination, Buffalo, N.Y.: Prometheus, 1977.Gutman, Amy, Liberal Equality, Cambridge: Cambridge University Press, 1980.Gutman, Amy, “Communitarian Critics of Liberalism”, Philosophy andPublic Affairs, Vol 14 No3, Summer 1985, PP. 308-322.Gutman, Amy, and Dennis Thompson, “Moral Conflict and PoliticalConsensus”, Ethics, Vol 101 No 1, October 1990, pp. 64-88.Hayek, Friedrich A., Law, Legislation, and Liberty, Vol 2: The Mirage of SocialJustice, London: Routledge & Kegan Paul, 1976.Haksar, Vinit, Equality, Liberty, and Perfectionism, New York: OxfordUniversity Press, 1979.Hardin, Russell, Morality Within The Limits Of Reason, Chicago: Universityof Chicago Press, 1988.Hare, R.M., “Rawis’ Theory of Justice”, Part I and II, Philosophical Quarterly,Vol. 23, April 1973 (144-155) and July 1973 (241-252).297Hare, R.M., Moral Thinking: Its Methods, Levels, and Point, Oxford: OxfordUniversity Press, 1981.Harman, Gilbert, “Moral Relativism Defended”, in Relativism Cognitive andMoral, Michael Krausz and Jack W. Meiland, eds., Notre Dame:University of Notre Dame Press, 1982.Harsanyi, John C., “Morality and the theory of rational behavior”, inUtilitarianism and Beyond, eds. Amartya Sen and Bernard Williams,Cambridge: Cambridge University Press, 1982.Haslett, D.W., Equal Consideration: A Theory of Moral Justification, Newark:University of Delaware Press, 1987.Heller, Agnes, Beyond Justice, Oxford: Basil Blackwell, 1987.Hobbes, Thomas, Leviathan, Parts I and II, Indianapolis: Bobs-Merrill, 1958.Hobhouse, L.T., Liberalism, New York: Oxford University Press, 1911, 1964.Hobhouse, L.T., The Elements of Social Justice, London: Allen & Unwin,1922.Hochschild, Jennifer, What’s Fair?, Cambridge: Harvard University Press,1985.Honderich, Ted, Conservatism, London: Hamish Hamilton, 1990.Hume, David, A Treatise of Human Nature, ed. L. A. Selby-Bigge, 2nd ed., ed.P.H. Nidditch, Oxford: Clarendon Press, 1978.Hume, David, An Enquiry Concerning the Principles of Morals, repr. fromthe edition of 1777, La Salle, Illinois: Open Court, 1966.Hurka, Thomas, “Why Value Autonomy?”, Social Theory and Practice, Vol13, Fall 1987, pp. 361-382.Jonsen, Albert R. and Stephen Toulmin, The Abuse of Casuistry: A History ofMoral Reasoning, Berkeley: University of California Press, 1988.Joseph, Keith and Jonathan Sumption, Equality, London: John Murray, 1979.Jouvenal, Bertrand de, The Ethics of Redistribution, Cambridge: CambridgeUniversity Press, 1952.298Kant, Immanuel, Groundwork of the Metaphysics of Morals, London:Hutchinson, 1956.Kant, Immanuel, The Metaphysical Elements of Justice, New York: BobbsMenu, 1965.Kamenka, Eugene, and Alice Erh-Soon Tay, ed., Justice, London: EdwardArnold, 1979.Kaufmann, Walter, Without Guilt and Justice: From Decidophobia toAutonomy, New York: Delta, 1973.Keibley, Charles A., ed., The Value of Justice, New York: Fordham UniversityPress, 1979.Kluegel, James R., and Eliot R. Smith, Beliefs About Inequality: Americans’Views Of What Is and What Ought To Be, New York: Plenum Press,1980.Krausz, Michael, and Jack W. Meiland, eds., Relativism, Cognitive and Moral,Notre Dame: University of Notre Dame Press, 1982.Kuhn, Thomas, The Structure of Scientific Revolutions, Chicago: Universityof Chicago Press, 1970.Kukathas, Chandran and Philip Pettit, Rawls: A Theory of Justice and itsCritics, Cambridge: Polity Press, 1990.Kymlicka, Will, “Rawis on Teleology and Deontology”, in: Philosophy andPublic Affairs, Vol 17, No 3, Summer 1988, PP. 173-190.Kymlicka, Will, Liberalism, Community, and Culture, Oxford: ClarendonPress, 1989.Kymlicka, Will, Contemporary Political Philosophy, Oxford: Clarendon Press,1990.Lerner, Melvin J., The Belief in A lust World, New York: Plenum Press, 1980.Levine, Andrew, Arguing for Socialism: Theoretical Considerations, London:Routledge & Kegan Paul, 1984.Locke, John, The Second Treatise of Civil Government and A LetterConcerning Toleration, ed. J.W. Gough, Oxford: Basil Blackwell, 1946.299Lomasky, Loren E., ‘Liberal Obituary?”, Ethics, Vol 102, No 1, October 1991,140-154.Luban, David, “Bargaining and Compromise: Recent Work on Negotiationand Informal Justiceu, Philosophy & Public Affairs, Vol 14 No 4, Fall1985, pp. 397-416.Lucas, J.R., On Justice, Oxford: Clarendon, 1980.Maclntrye, Alasdair, After Virtue: A Study in Moral Theory, London:Duckworth, 1981.Maclntrye, Alasdair, Whose Justice? Which Rationality?, Notre Dame,Indiana: University of Notre Dame Press, 1988.Mackie, J.L., Ethics: Inventing Right and Wrong, New York: Penguin, 1977.Maguire, Daniel C., A New Amercian Justice: Ending the White MaleMonopolies, New York: Doubleday, 1980.Mapel, David, Social Justice Reconsidered: The Problem of AppropriatePrecision in a Theory of Justice, Chicago: University of Illinois Press,1989.Martin, R., Rawls and Rights, Lawrence, Kansas: University Press of Kansas,1985.McClennen, Edward F, “Justice and the Problem of Stability”, Philosophy andPublic Affairs, Vol 18 No 1, Winter 1989, pp. 3-30.McDonald, Michael, “Justice in Hard Times”, in: Social Justice, ed. MichaelBradie and David Braybrooke, Bowling Green Studies in AppliedPhilosophy, Vol IV, 1982, pp. 34-43.McDonald, Michael, “After Virtue, Taking Rights Seriously”, Journal ofBusiness Ethics, Vol 5, 1986, pp. 21-28.McDonald, Michael, “The Personless Paradigm”, University of Toronto LawJournal, Vol XXXVJI, No 2, Spring 1987, pp. 212-226.McDonald, Michael, “Culture, Political Obligation and Ideology”, Philosophyand Culture, Proceedings of the XVIIth World Congress of Philosophy,Vol II, ed. Venant Cauchy, University of Montreal, 1988, pp. 319-323.300McDonald, Michael, “Should Communities Have Rights? Reflections onLiberal Individualism”, The Canadian Journal of Law andJurisprudence, Vol IV, No 2, July 1991, pp. 217-237.Mendus, Susan, Toleration and the Limits of Liberalism, Atlantic Highlands,NJ: Humanities Press, 1989.Mettrick, Alan, Last in Line, Toronto: Key Porter, 1985.Michalos, Alex C., ‘Meeting Current Needs’, Dialogue, Vol XXVII No 3,Autumn 1988, PP. 507-516.Miii, J.S., On Liberty, New York: Crofts, 1947.Miii, J.S., Utilitarianism, Indianapolis: Bobs-Merrill, 1971.Miii, J.S., Considerations On Representative Government, Indianapolis, Bobs-Merrill, 1958.Miller, David, Social Justice, Oxford: Clarendon Press, 1976.Moon, J. Donaid,ed., Responsibility, Rights, and Welfare: The Theory of theWelfare State, Boulder, Col.: Westview Press, 1988.Moore, Margaret, “Justice for Our Times”, Canadian Journal of PoliticalScience, XXXIII:3, September 1990.Moser, Paul K., Empirical Justification, Dordrecht: D. Reidel, 1985.Mosovitch, Allan, and Glenn Drover, eds., Inequality, Toronto: University ofToronto Press, 1981.Nagel, Thomas, The View From Nowhere, Oxford: Oxford University Press,1986.Narveson, Jan, The Libertarian Idea, Philadelphia: Temple University Press,1988.Narveson, Jan, “Collective Rights?”, The Canadian Journal of Law andJurisprudence, Vol. IV, No 2, July 1991.Nielsen, Kai, “Arguing about Justice: Marxist Immoralism and MarxistMoralism”, Philosophy and Public Affairs, Vol 17, No 3, Summer 1988,pp. 212- 234.301Nielsen, Kai, Equality and Liberty: A Defense of Radical Egalitarianism,Totowa, N.J.: Rowman & Allanheld, 1985.Nozick, Robert, Anarchy, State, and Utopia, New York: Basic Books, 1974.Nozick, Robert, The Examined Life, Toronto: Simon and Schuster, 1989.O’Neill, Onora, “Between Consenting Adults”, Philosophy and Public Affairs,Vol 14 No 3, 1985, pp. 252-277.O’Neill, Onora, Faces of Hunger, London: Allen & Unwin, 1986.Pal, Leslie A., Public Policy Analysis, Toronto: Methuen, 1987.Paul, Jeffrey, ed., Reading Nozick, Totowa, N.J.: Rowman and Littlefield, 198LPerelman, Chaim, Justice, New York: Random House, 1967.Pettit, Philip, Judging Justice: An Introduction to Contemporary PoliticalPhilosophy, London: Routledge & Kegan Paul, 1980.Phillips, Derek L., Tozvard a Just Social Order, Princeton, N.J.: PrincetonUniversity Press, 1986,Plato, The Republic, in: The Dialogues of Plato, transL B. Jowett, New York:Random House, 1937.Pogge, Thomas W., Realizing Rawls, Ithaca: Cornell University Press, 1989.Posner, Richard A., The Economics of Justice, Cambridge, Mass.: HarvardUniversity Press, 1981.Rachels, James, The Elements of Moral Philosophy, Philadelphia: TempleUniversity Press, 1986.Railton, Peter, “Moral Theory as a Moral Practice”, Nous, Vol XXV, No 2,April 1991, pp. 185-191.Ratner, Robert S. and John L. McMullen, eds., State Control: Criminal JusticePolitics in Canada, Vancouver: University of British Columbia Press,1987.Rawis, John, A Theory of Justice, Cambridge, Mass.: Harvard University Press,1971.302Rawis, John, “Fairness to Goodness”, Philosophical Review, Vol 84, Oct. 1975,pp. 536-554.Rawls, John, “The Basic Structure as Subject”, American PhilosophicalQuarterly”, Vol 14 No 2, April 1977, pp. 159-165.Rawis, John, “Kantian Constructivism in Moral Theory”, Journal ofPhilosophy, No 77, September 1980, pp. 515-572.Rawls, John, “Social Unity and Primary Goods”, Utilitarianism and Beyond,ed. Amartya Sen and Bernard Williams, Cambridge: CambridgeUniversity Press, 1982.Rawls, John, “Justice as Fairness: Political not Metaphysical”, Philosophy &Public Affairs, Vol 14 No 3, Summer 1985, pp. 223-251.Rawis, John, “The Basic Liberties and Their Priority”, in: Liberty, Equality, andLaw, Selected Tanner Lectures on Moral Philosophy, ed. Sterling M.McMurrin, Salt Lake City: University of Utah Press, 1987.Rawis, John, “The Idea of an Overlapping Consensus”, Oxford Journal ofLegal Studies, 7, 1987, pp. 1-25.Rawls, John, “The Priority of Right and Ideas of the Good”, Philosophy andPublic Affairs, Vol 17, No 4, Fall 1988, pp. 251-276.Raz, Joseph, The Morality of Freedom, Oxford: Clarendon Press, 1986.Reiman, Jeffrey, Justice and Modern Moral Philosophy, New Haven: YaleUniversity Press, 1990.Rescher, Nicholas, Distributive Justice: A Constructive Critique of theUtilitarian Theory of Distribution, New York: Bobbs-Merrill, 1966.Resnick, Philip, The Masks of Proteus: Canadian Reflections on the State,Montreal & Kingston: McGill-Queens University Press, 1990.Ripstein, Arthur, “Foundationalism in Political Theory”, Philosophy andPublic Affairs 16, No 2, Spring 1987, pp. 115-137.Rousseau, Jean-Jacques, Of the Social Contract and Discourse on PoliticalEconomy, New York: Harper & Row, 1984.Rule, James B., Insight and Social Betterment, New York: Oxford UniversityPress, 1978.303Runciman, W.G., Relative Deprivation and Social Justice, London: Routledge& Kegan Paul, 1966.Sandel, Michael, Liberalism and the Limits of Justice, Cambridge: CambridgeUniversity Press, 1982.Sandel, Michael, ed., Liberalism and Its Critics, Oxford: Basil Blackwell, 1984.Sayre-McCord, Geoffrey, ed., Essays on Moral Realism, Ithaca: CornellUniversity Press, 1988.Scanlon, T.M., “Contractualism and utilitarianism”, in: Utilitarianism andbeyond, eds. Amartya Sen and Bernard Williams, Cambridge:Cambridge University Press, 1980.Scheffler, Samuel, The Rejection of Consequentialism, Oxford: ClarendonPress, 1982.Scruton, Roger, The Meaning of Conservatism, Harmondsworth, Eng.:Penguin Boooks, 1980.Segers, Mary C. et al., Elusive Equality: Liberalism, Affirmative Action, andSocial Change in America, Port Washington, N.Y.: Kennikot Press,1983.Sen, Amartya, and Bernard Williams, eds., Utitarianism and Beyond,Cambridge: Cambridge University Press, 1982.Shapiro, Ian, The Evolution of Rights in Liberal Theory, Cambridge:Cambridge University Press, 1986.Sher, George, Desert, Princeton, N.J.: Princeton University Press, 1987.Sher, George, “Three Grades of Social Involvement”, Philosophy and PublicAffairs, Vol 18, No 2, Spring 1989, pp. 133-157.Shklar, Judith N., The Faces of Injustice, New Haven: Yale University Press,1990.Shue, Henry, Basic Rights: Subsistence, Affluence, and U.S. Foreign Policy,Princeton: Princeton University Press, 1980.Sidgwick, Henry, The Methods of Ethics, Chicago: The University of ChicagoPress, 1962.304Sikora, Richard I., ‘Six Viewpoints for Assessing Egalitarian DistributionSchemes”, Ethics, Vol 99, No 3, April 1989, pp. 492-502.Singer, Peter, ed., A Companion to Ethics, Oxford: Basil Blackwell, 1991.Smart, J.J.C. and Bernard Williams, Utilitarianism; For and Against,Cambridge: Cambridge University Press, 1973.Sobel, Jordan Howard, “Maximizing, Optimizing, and Prospering”, Dialogue,Vol XXVII, No 2, Summer 1988, pp. 233-262.Soifer, Eldon, “On Justifying Liberal Neutrality”, Paper delivered at theCanadian Philosophical Association Congress, Kingston, May 1991.Soltan, Karol Edward, The Causal Theory of Justice, Berkeley: University ofCalifornia Press, 1987.Spitz, David, The Real World of Liberalism, Chicago: The University ofChicago Press, 1982.Sterba, James P., The Demands of Justice, Notre Dame, md.: University ofNotre Dame Press, 1980.Stout, Jeffrey, Ethics After Babel: The Languages Of Morals And TheirDiscontents, Boston: Beacon Press, 1988.Sumner, L.W., The Moral Foundation of Rights, Oxford: Clarendon Press,1987.Sumner, L.W., “Justice Contracted”, Dialogue, Vol XXVI No 3, Autunm 1987,pp. 523-548.Sumner, L.W., “Positive Sexism”, in Equal Opportunity, ed. by Ellen FrankelPaul et al., New York: Basil Blackwell, 1987.Tawney, R.H., Equality, London: Unwin, 1964.Tawney, R.H., The Acquisitive Society, London: G.Bell and sons, 1921.Taylor, Charles, Hegel and Modern Society, Cambridge: Cambridge UniversityPress, 1979.Tempkin, Larry S., “Inequality”, Philosophy and Public Affairs, Vol 15, No 2,Spring 1986, pp. 99-121.305Thakur, Shivesh C., Social Justice: An Informal Analysis??, in: InternationalPhilosophical Quarterly, September 86, pp. 213-22LThurow, Lester C., !?A Surge in Inequality’, Scientific American, Vol 254 No5, Mary 1987, pp. 30-37.Tucker, D.F.B., Marxism and Individualism, Oxford: Basil Blackwell, 1980.Vallentyne, Peter, ed., Contractarianism and Rational Choice: Essays on DavidGau thier ‘s Morals By Agreement, Cambridge: Cambridge UniversityPress, 1991.Waizer, Michael, Spheres of Justice: A Defense of Pluralism and Equality,New York: Basic Books, 1983.Wedeking, Gary, “Is Mandatory Retirement Unfair Age Discrimination???,Canadian Journal of Philosophy, Vol 20, No 3, September 1990, pp. 321-334.Weinreb, Lloyd L., Natural Law and Justice, Cambridge, Mass: HarvardUniversity Press, 1987.Wilhoit, Francis M., The Quest for Equality in Freedom, New Brunswick,N.J.: Transaction Books, 1979.Williams, Bernard, Problems of the Self, Cambridge: Cambridge UniversityPress, 1973.Wittgenstein, Ludwig, Philosophical Investigations, 3rd ed., New York:Macmillan, 1968.Winfield, Richard Dien, Reason and Justice, New York: State University ofNew York Press, 1988.Wolgast, Elizabeth, The Grammar of Justice, Ithaca: Cornell University Press,1987.Young, Michael, The Rise of the Meritocracy, New York: Random House,1959.Zinn, Howard, Justice In Everyday Life: The Way It ReallyWorks, New York: William Morrow & Company, 1974."@en ; edm:hasType "Thesis/Dissertation"@en ; vivo:dateIssued "1992-11"@en ; edm:isShownAt "10.14288/1.0086610"@en ; dcterms:language "eng"@en ; ns0:degreeDiscipline "Philosophy"@en ; edm:provider "Vancouver : University of British Columbia Library"@en ; dcterms:publisher "University of British Columbia"@en ; dcterms:rights "For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use."@en ; ns0:scholarLevel "Graduate"@en ; dcterms:title "Rational disagreement about social justice"@en ; dcterms:type "Text"@en ; ns0:identifierURI "http://hdl.handle.net/2429/3128"@en .