The Idea of Justice

by Amartya Sen

  d o c o m p a r a t i v e s i d e n t i f y

  t r a n s c e n d e n c e ?

  A transcendental identification is thus neither necessary nor sufficient for arriving at comparative judgements of justice. We should, however, examine a third type of connection that might conceivably link the comparative with the transcendental. Could it be the case that the comparative rankings of the different alternatives must inter alia also be able to identify the transcendentally just social arrangement? Would the transcendental invariably follow from the full use of the comparative? If that were the case, we could plausibly argue that in a somewhat weak sense there is a necessity for the tractability of the transcendental alternative. It would not, of course, imply that there is any need to go via the transcendental approach to comparative assessments, but it would at least give transcendental identification a necessary presence in the theory of justice, in the sense that if the transcendental question cannot be answered, then we should conclude that we cannot fully answer the comparative question either.

  Would a sequence of pairwise comparisons invariably lead us to the very best? That presumption has some appeal, since the superlative might indeed appear to be the natural end-point of a robust comparative. But this conclusion would, in general, be a non sequitur. In fact, it is only with a ‘well-ordered’ ranking (for example, a complete and 102

  v o i c e a n d s o c i a l c h o i c e transitive ordering over a finite set) that we can be sure that the set of pairwise comparisons must also always identify a ‘best’ alternative.

  We must, therefore, ask: how complete should the assessment be for it to be a systematic discipline? In the ‘totalist’ approach that characterizes the standard theories of justice, including Rawls’s, incompleteness tends to appear as a failure, or at least as a sign of the unfinished nature of the exercise. Indeed, the survival of incompleteness is sometimes seen as a defect of a theory of justice, which calls into question the positive assertions that such a theory makes. In fact, a theory of justice that makes systematic room for incompleteness can allow one to arrive at quite strong – and strongly relevant –

  judgements (for example, about the injustice of continuing famines in a world of prosperity, or of persistently grotesque subjugation of women, and so on), without having to find highly differentiated assessments of every political and social arrangement in comparison with every other arrangement (for example, addressing such questions as: exactly how much tax should be put on the sale of petrol in any particular country, for environmental reasons?) I have discussed elsewhere why a systematic and disciplined theory of reasoned evaluation, including assessment of social justice, need not take a ‘totalist’ form.* Incompleteness may be of the lasting kind for several different reasons, including unbridgeable gaps in information, and judgemental unresolvability involving disparate considerations that cannot be entirely eliminated, even with full information. For example, it may be hard to resolve the conflicting claims of different equity considerations, of which a very special case is the one chosen by Rawls in the form of lexicographic maximin, which gives total priority to the minutest gain of the worst-off group even when this entails huge losses for groups that are not the worst-off but

  * This was a central feature of the approach to social choice theory that I tried to develop in my book Collective Choice and Social Welfare (1970). The issue is revisited, with response to critical comments, in some of my recent essays, including: ‘Maximization and the Act of Choice’, Econometrica, 65 (1997); ‘The Possibility of Social Choice’, American Economic Review, 89 (1999); and ‘Incompleteness and Reasoned Choice’, Synthese, 140 (2004). See also Isaac Levi’s response to the last, in ‘Amartya Sen’, in the same number of Synthese, and his important book, Hard Choices (Cambridge: Cambridge University Press, 1986).

  103

  t h e i d e a o f j u s t i c e are very badly off, on which quite different reasonable positions can be taken by impartial observers. There may also be varying reasonable compromises in balancing small gains in liberty, which is given priority in Rawls’s first principle, against any reduction in economic inequality

  – no matter how large. The importance of recognizing the plurality of reasons of justice has already been discussed earlier in this book, and this issue will be further examined in later chapters.

  And yet, despite such durable ambiguity, we may still be able to agree readily that there is a clear social failure involved in persistent famines or in widespread exclusion from medical access, which calls urgently for remedying (thereby yielding an advancement of justice), even after taking note of the costs involved. Similarly, we may acknowledge the possibility that the liberty of different individuals may, to some extent, conflict with each other (so that any fine-tuning of the demands of

  ‘equal liberty’ may be hard to work out), and yet strongly agree that government-arranged torture of prisoners, or arbitrary incarceration of accused people without access to court procedures, would be an unjust violation of liberty that calls for urgent rectification.

  There is a further consideration that may work powerfully in the direction of making political room for incompleteness of judgements about social justice, even if it were the case that every person had a complete ordering over the possible social arrangements. Since a theory of justice, in the standard forms, invokes agreement between different parties (for example in the unanimous agreement that is sought in the ‘original position’ in the Rawlsian framework), incompleteness can also arise from the possibility that distinct persons may continue to have some differences in assessments (consistently with agreeing on a great many comparative judgements). Even after vested interests and personal priorities have been somehow ‘taken out’ of consideration through such devices as the ‘veil of ignorance’, there may remain possibly conflicting views on social priorities, for example in weighing the claims of needs over entitlement to the fruits of one’s labour (as in the example of the three children quarrelling about the use of a flute).

  Even when all the parties involved have their own complete orderings of justice that are not congruent, the ‘intersection’ between the rankings – that is the shared beliefs of the different parties –

  104

  v o i c e a n d s o c i a l c h o i c e will yield a partial ranking, with different extents of articulation (depending on the extent of similarity between the orderings).17 The acceptability of evaluative incompleteness is indeed a central subject in social choice theory in general, and it is relevant to theories of justice as well, even though Rawlsian ‘justice as fairness’ and other such theories firmly assert (and it is an assertion rather than something that is actually established by the arguments presented) that a full agreement will definitely emerge in the ‘original position’ and in other such formats.

  Thus, for reasons both of incomplete individual evaluations and of incomplete congruence between different individuals’ assessments, persistent incompleteness may be a hardy feature of judgements of social justice. This can be problematic for the identification of a perfectly just society, and make transcendental conclusions difficult to derive.* And yet, such incompleteness would not prevent making comparative judgements about justice in a great many cases – where there might be fair agreement on particular pairwise rankings – about how to enhance justice and reduce injustice.

  Thus the hiatus between the relational approach and the transcendental approach to justice seems to be quite comprehensive. Despite its own intellectual interest, the question ‘what is a just society?’ is not, I have argued, a good starting-point for a useful theory of justice.

  To that has to be added the further conclusion that it may not be a plausible end-point either. A systematic theory of comparative justice does not need, nor does it necessarily yield, an answer to the question

  ‘what is a just society?’

  * On a mathematical point, it must be acknowledged that a transitive but incomplete ordering over a finite set will invariably yield one or more
‘maximal’ elements, in the sense of there being one or more alternatives that are undominated by any other element. A maximal set must not, however, be confused with a set of ‘best’ elements, since maximality does not guarantee the existence of a best element (only one that is no worse than any other). On the far-reaching relevance of the distinction between maximality (needed for an acceptable choice) and optimality (needed for making a perfect choice), see my ‘Internal Consistency of Choice’, Econometrica, 61 (1993), and ‘Maximization and the Act of Choice’, Econometrica, 65 (1997). The foundational nature of the mathematical distinction involved can be seen in N. Bourbaki, General Topology, Parts I and II, English translation (Reading, MA: Addison-Wesley, 1966), and Theory of Sets (Reading, MA: Addison-Wesley, 1968).

  105

  t h e i d e a o f j u s t i c e s o c i a l c h o i c e a s a f r a m e w o r k f o r r e a s o n i n g

  What, then, are the points of relevance of social choice theory for the theory of justice? There are many connections, but I will focus here on seven points of significant contribution, in addition to the focus on social realizations (already discussed).18

  (1) Focus on the comparative, not just the

  transcendental

  Perhaps the most important contribution of the social choice approach to the theory of justice is its concern with comparative assessments.

  This relational, rather than transcendental, framework concentrates on the practical reason behind what is to be chosen and which decisions should be taken, rather than speculating on what a perfectly just society (on which there may or may not be any agreement) would look like. A theory of justice must have something to say about the choices that are actually on offer, and not just keep us engrossed in an imagined and implausible world of unbeatable magnificence. As I have already discussed this contrast fairly extensively, I shall not comment further on it here.

  (2) Recognition of the inescapable plurality of competing principles

  Social choice theory has given considerable recognition to the plurality of reasons, all of which demand our attention when issues of social justice are considered, and they may sometimes conflict with each other. This inescapable plurality may or may not lead to an impossibility result, yielding an impasse, but the need to take note of the possibility of durable conflicts of non-eliminable principles can be quite important in the theory of justice. In the chapters to follow, this plurality will be more fully explored.

  106

  v o i c e a n d s o c i a l c h o i c e (3) Allowing and facilitating re-examination

  Another feature of some importance is the way social choice theory has persistently made room for reassessment and further scrutiny.

  Indeed, one of the main contributions of results like Arrow’s impossibility theorem is to demonstrate that general principles about social decisions that initially look plausible could turn out to be quite problematic, since they may in fact conflict with other general principles which also look, at least initially, to be plausible.

  We often think, if only implicitly, of the plausibility of principles in a number of specific cases which focus our attention on those ideas –

  the human mind cannot often enough grasp the immense reach of general principles. But once the principles are formulated in unconstrained terms, covering inter alia a great many cases other than those that motivated our interest in those principles, we can run into difficulties that were not foreseen earlier, when we signed up, as it were, on the dotted line. We then have to decide what has to give and why. Some may find social choice theory to be too permissive and indecisive (Condorcet saw his results as the beginning of a discussion, not the end of it), but the alternative, well illustrated by mainstream theories of justice, like Rawls’s or Nozick’s, of inflexible insistence on exacting and highly demanding rules does not give the idea of justice its due.

  (4) Permissibility of partial resolutions

  Social choice theory allows the possibility that even a complete theory of justice can yield incomplete rankings of justice. Indeed, the incompleteness in many cases can be ‘assertive’, yielding statements such as x and y cannot be ranked in terms of justice. This contrasts with an incompleteness that is tentatively accepted, while awaiting – or working towards – completion, on the basis of more information, or more penetrating examination, or with the use of some supplementary criteria.

  The theory of justice has to make room for both kinds of incompleteness, assertive and tentative. Tentative incompleteness may reflect operational difficulties, rather than any deeper conceptual or 107

  t h e i d e a o f j u s t i c e valuational deadlock. The operational problems may relate to limitation of knowledge, or complexity of calculation, or some other practical barriers in application (the kind of considerations that have been illuminatingly and powerfully explored by Herbert Simon, leading to his important notion of ‘bounded rationality’).19 Even when incompleteness is in this sense tentative, it may still be hardy enough to demand incorporation in a functioning theory of justice, combined with room for re-examination and possible extension. In contrast, with assertive incompleteness the partial nature of the resolution is an integral part of the conclusions advanced by a theory of justice, even though that theory itself could remain open to further scrutiny and revision.

  (5) Diversity of interpretations and inputs

  The formal structure of social choice theory, which often takes the form of exploring functional connections, guided by sets of axioms, between individual rankings and priorities on the one hand, and social conclusions on the other, is open to alternative interpretations. For example, there has been considerable interest within the discipline in the distinction between the aggregation of individual interests and that of individual judgements.20

  A person’s voice may count either because her interests are involved, or because her reasoning and judgement can enlighten a discussion.

  Also, a person’s judgement may be seen as important either because she is one of the parties directly involved (this may be called ‘membership entitlement’), or because the person’s perspective and the reasons behind it bring important insights and discernment into an evaluation, and there is a case for listening to that assessment whether or not the person is a directly involved party (this can be called ‘enlightenment relevance’).21 In the Rawlsian universe of justice as fairness, it is the membership entitlement that seems to get all the attention at the political level (though Rawls devises the original position with the aim of cutting out the influence of their vested interests in the choice of principles of justice), whereas in the approach advanced by Adam Smith, invoking ‘impartial spectators’, distant voices may be given a very important place for their enlightenment relevance, for example 108

  v o i c e a n d s o c i a l c h o i c e to avoid parochialism of local perspectives. This contrast will be more fully explored in Chapter 6.

  Sometimes the so-called ‘individual’ rankings and priorities can be seen not as those of distinct persons, but of different approaches by the same person to the decisional issues involved, all of which could command some respect and attention. Another variation relates to the possibility that the individual rankings may not be those of individual preferences at all (in any of its various senses), as is usually presumed in mainstream social choice theory, but diverse rankings yielded by different types of reasoning. In general, social choice theory as a discipline is concerned with arriving at overall judgements for social choice based on a diversity of perspectives and priorities.

  (6) Emphasis on precise articulation and reasoning There is some general merit in the explicitness of fully stated axioms and carefully established derivations, which make it easier to see what is being assumed and what exactly they entail. Since the demands that are linked to the pursuit of justice in public discussion, and sometimes even in theories of justice, often leave considerable room for clearer articulation and fuller defence, this explicitness can itself be something of a contribution.
r />   Consider, for example, the Rawlsian claim that in the original position there would emerge a contract with the priorities that he specifies, including the overall priority of liberty under his first principle, and the conditional priority of the interests of the poorest group, judged by the holdings of primary goods, under his second principle.*

  But there are other alternative contracts that also have appeal, and there may or may not be any clear agreement on this even in the circumstances of the original position. Rawls’s conviction that his two principles would unanimously emerge in the original position is not backed by any kind of definitive reasoning, and it is not even fully clear which normative premises would lead to that exact choice or

  * Rawls presents several arguments in his Theory of Justice (1971) on why these principles may appeal in the original position, and backs them up with somewhat broader arguments in his later writings, particularly Political Liberalism (1993).

  109

  t h e i d e a o f j u s t i c e would be consistent with it. In fact, a number of fairly detailed investigations in social choice theory have actually identified the axiomatic basis of these Rawlsian presumptions,22 and helped to clarify what the debates are about. Even though the axiomatic correspondences do not resolve the difficult issue of being sure about what should be chosen, they show on which lines the normative debates may fruitfully proceed.

  Given the complex nature of human values and social reasoning, they may often be hard to capture in precise axiomatic terms, and yet the need for explicitness, to the extent that can be achieved, must have much dialogic merit. How far to go towards axiomatization cannot but be, to a considerable extent, a matter of judgement in dealing with the competing claims of precise characterization, on the one hand, and the need to take note, on the other, of the complexities that may be hard to axiomatize but which are nevertheless significant concerns that can be usefully discussed in more general – and somewhat looser