UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Rational disagreement about social justice Preinsperger, Kurt 1992

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1992_fall_preinsperg_kurt.pdf [ 5.65MB ]
Metadata
JSON: 831-1.0086610.json
JSON-LD: 831-1.0086610-ld.json
RDF/XML (Pretty): 831-1.0086610-rdf.xml
RDF/JSON: 831-1.0086610-rdf.json
Turtle: 831-1.0086610-turtle.txt
N-Triples: 831-1.0086610-rdf-ntriples.txt
Original Record: 831-1.0086610-source.json
Full Text
831-1.0086610-fulltext.txt
Citation
831-1.0086610.ris

Full Text

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 criteri