¹ Ronald Dworkin holds a version of the equality view (Dworkin, Ronald, ‘What Is Equality? Part 1: Equality of Welfare’, Philosophy & Public Affairs, x (1981), 244–6Google Scholar). But it is not clear that he believes that equality makes an outcome better. He seems to think instead that we violate a duty—the duty of treating people as equals—if we cause, or merely permit, inequality. Inequality is not bad in itself in the way that suffering is, but it is a sign that people have been treated unequally and so—arguably—unfairly. However, Dworkin's view is concerned with equality understood in terms of the overall distribution of resources in an outcome.

² Thomas Nagel and (I will argue) John Rawls accept the priority view. The difference between the two egalitarian ideas was briefly explained in my paper ‘Egalitarianism’, Dialogue, xxiii (1984), pp. 230–7Google Scholar. It is discussed at greater length in Derek Parfit's unpublished manuscript ‘On Giving Priority to the Worse Off’. I have borrowed Parfit's name, the ‘priority view’, for the second kind of egalitarianism.

³ Parfit puts it nicely: the basic idea in the view is not the diminishing marginal utility of resources but the diminishing moral importance of utility itself (‘On Giving Priority to the Worse Off’).

⁴ The difference principle is discussed in Rawls, John, A Theory of Justice, Cambridge, Mass., 1971Google Scholar. Subsequent page references to this book will be included in the text.

⁵ It is important for the combined view to be able to say that in many cases the utilitarian reasons will outweigh the value of reducing inequality in this way. This is a problem for Dworkin's egalitarianism. He dismisses utilitarianism as either an implausible teleological principle or a misinterpretation of his own principle of moral equality (Dworkin, , pp. 244–6)Google Scholar. Dworkin does not say that equality has absolute priority over utility, but this seems to be a consequence of his view.

⁶ How would the combined view deal with the objections to the difference principle? In the first example the equality view would never tell us to help the more advantaged group, regardless of the size of the benefits and the number of people in the two groups. But the combined view also contains the utilitarian principle. When the example is most persuasive as an objection to the difference principle—when the more advantaged group is much larger, and the benefits that they would receive are also much larger—the combined view can decide that the gain in terms of utility outweighs the loss in terms of equality. In the second example, it will sometimes be true that helping other badly-off groups instead of the very worst-off group will reduce the total amount of inequality. So the combined view can also account for our judgment about this example.

⁷ I am not suggesting that we should only accept the equality view or the priority view if there are independent arguments that equality is valuable or that the interests of the worse off should have priority. Both views might be regarded as fundamental moral principles. But if a particular way of specifying either view is intuitively implausible, I think that it is reasonable to ask for arguments before accepting it.

⁸ Barry, Brian (Theories of Justice, Berkeley and Los Angeles, 1989, pp. 217–25Google Scholar) and Kymlicka, Will (Contemporary Political Philosophy, Oxford, 1991, pp. 55–70Google Scholar) both think that the intuitive reasoning in A Theory of Justice § 12 is a more important source for Rawls's egalitarianism, and a stronger argument, than the contractualist argument. They suggest that Rawls argues for equality by appealing to the more fundamental idea that it is regrettable if people's lives are influenced by morally arbitrary factors like natural ability. Equality would eliminate the influence of these factors. The argument is unconvincing, if it is intended to derive equality from some other value. Suppose that the world contained just one person who used natural ability to gain certain advantages. No one would object because the advantages result from morally arbitrary factors. At most we object to relative advantages—advantages that some people have and others lack—based on morally arbitrary factors. This is just to say that we object to inequalities that result from those factors. The argument does not appeal to some other value to support equality; it must start from the idea that a certain kind of inequality (or inequality with a certain cause) is objectionable.

⁹ It might seem that Rawls intends A Theory of Justice § 12 to explain why the principle about equality on p. 60 is converted into the difference principle stated on p. 83. But § 12 does not say anything to justify the difference principle specifically, as opposed to other egalitarian principles. It contrasts the democratic equality interpretation of the two principles on p. 60 (the democratic equality interpretation includes the difference principle) with interpretations that accept the inequalities generated in a competitive economy. If Rawls has intuitive arguments for the difference principle itself they are to be found in §§ 13 and 17.

¹⁰ In fact Rawls frequently says that to be justified inequality must benefit the worst-off—they must be better off with the inequality than in any other outcome with less inequality. This stronger condition is incompatible with the lexical difference principle. I will assume (although this is controversial) that Rawls's real view is expressed by the weaker condition that inequality must not harm the worst off. One reason for thinking so is that the brief discussion of the lexical difference principle (pp. 82–3) seems to indicate that Rawls would use it in cases where it would differ significantly from the simple difference principle.

¹¹ Rawls is sensitive to this question about the difference principle. He says (pp. 78–9), considering the same example, that the initial outcome is just throughout but the outcome that would result from the change is the best just outcome. Here Rawls's conclusions about justice seem to be based on a view that might indeed result from starting with equality and putting a condition on permissible inequality. It would say that inequality is unjust if it disadvantages the worst off, if they could be helped by reducing the inequality. Rawls recognizes that this view is not equivalent to the difference principle.

¹² Barry, , pp. 229–34.Google Scholar

¹³ Barry thinks that Rawls permits inequality that would benefit the worst off because the independent principle about Pareto superiority overrides the initial commitment to equality. He regards the Pareto principle as a requirement of rationality. Rawls does sometimes present the move from equality to the difference principle as a matter of rationality—see Rawls, , p. 546Google Scholar. But other passages—especially pp. 100–1—suggest that Rawls thinks that the view that inequality is acceptable if it benefits (or at least does not harm) the worst off is itself a fair response to the moral arbitrariness of natural endowment and social position. It is no less fair than requiring strict equality. The apparent commitment to strict equality in § 12 is provisional, and it is revised in § 17 — seemingly as a result of thinking more about fairness, not an appeal to Pareto optimality—in a way that is intended to lead to the difference principle.

¹⁴ Perhaps, rather than being derived from equality, the difference principle should be understood as a distributive principle that is an alternative to the equality view. It is not a version of the priority view, and it claims that the best, or the fairest, overall distribution is the one in which the worst-off group is best off. But I think it is difficult to regard the claims made by the lexical difference principle as motivated by considerations of fairness in distribution. It may not be possible to combine, in a single principle, a concern for fairness in distribution between different people and a concern with improving the quality of every life.

¹⁵ There is a reason for not simply identifying the difference principle with a version of the priority view. As I have explained it, the priority view assigns value to outcomes. An affluent society might fail to satisfy the difference principle while a poor society satisfies it, even though the worst-off group in the first society is better off than the worst-off group in the second society. Rawls's view would say that the first society is unjust and the second is not. The corresponding version of the priority view would say that the first society is a better outcome than the second society. This version of the priority view could be turned into a theory of justice that would agree with Rawls about the justice of the two societies. But it is not clear that Rawls intends the difference principle to lead to the conclusion that the first society is a better outcome than the second society, as opposed to just saying that distribution is unfair in the first society but fair in the second society.

¹⁶ Nagel, Thomas, ‘Equality’ in Thomas Nagel Mortal Questions, Cambridge, 1979, pp. 106–27, see p. 125Google Scholar. Subsequent page references to this article will be included in the text.

¹⁷ The differences that are important for the issues I am discussing will become apparent later. To list some others: Nagel's principle does evaluate actions by assigning values to outcomes, it applies to distribution between individuals not social classes, and it is directly concerned with peoples' interests rather than primary goods.

¹⁸ Nagel suggests a deeper source for egalitarianism than Rawls. Rawls begins with the idea that there is a need for principles to govern the fair distribution of the goods produced by social cooperation. He then argues, in intuitive and contractualist ways, that these principles will be egalitarian. On this view the scope of egalitarian principles is determined by the limits of social cooperation; if people do not cooperate, egalitarianism does not apply. Nagel locates the source of egalitarianism in a method for moving from the moral claims of individuals to evaluations of overall outcomes. If he is right, egalitarian principles have unlimited scope.

¹⁹ If the size of the benefits for the two groups were roughly the same, the outcome in which the worst off were helped would be the least individually unacceptable outcome. Nagel mentions the possibility that our conclusion about this example depends on utilitarian reasons rather than egalitarian reasons. He does not use this possibility to defend his view, although he does not explain why. We can suppose that the policy that benefits the very worst off would also help a large group of affluent people. It might be the best choice in utilitarian terms. If we would still help Nagel's badly-off group, our decision must be based on egalitarian reasons.

²⁰ I have not attempted to answer Nagel's argument from unanimity to egalitarianism, apart from emphasizing his own worry that it commits the egalitarian view to innumeracy. There are other problems with the argument. It might be an important merit in a moral principle that it achieves literal unanimity—finds outcomes which everyone prefers to all the alternatives. But egalitarianism does not do this, and it is not clear that achieving Nagel's substitute for unanimity—the least individually unacceptable outcome—is a decisive merit in a moral principle. The special value of unanimity might not carry over to the closest approximation to it. Also, the outcome that would be acceptable to the most people has a competing claim to be considered the closest approximation to unanimity. Finally, I do not agree with Nagel that the fact that moral claims belong to individuals itself points us towards pairwise comparison rather than aggregation as the appropriate method for combining those claims.

²¹ Nagel, Thomas, Equality and Partiality, Oxford, 1991, pp. 63–74Google Scholar. References to this book will also be included in the text.

²² In ‘Equality’ Nagel appeals to unanimity to explain why we should choose the least individually unacceptable outcome and help the handicapped child in his example. This part of the argument is not developed in Equality and Partiality. There Nagel uses unanimity to explain how we should balance impartial egalitarian moral reasons against partial agent-relative moral and non-moral reasons (pp. 41–53), not to explain why impartial moral reasons will be distinctively egalitarian.

²³ In this respect the view is different from Rawls's difference principle and Nagel's principle in ‘Equality’. They would make the priority depend on the relational fact that the badly off are worse off than others. As a result, they would give priority to the worst off regardless of their actual level of well-being, and the strength or extent of the priority given would not depend on their actual level of well-being (in Nagel's case the actual level of well-being of the worst off might affect which outcome counts as the least individually unacceptable outcome). Perhaps some people will feel that priority for the badly off should depend on their relative position. They think that we understand why it is appropriate to give a special value to benefits for the badly off (a value that is greater than the value that utilitarianism would assign to those benefits) only when we compare their lives to the lives led by more fortunate people. If we feel strongly that a preference for the badly off requires these comparisons with other lives, it is possible that we are being influenced by a concern for equality between different people rather than by a version of the priority view.

²⁴ We could formulate a principle of egalitarian priority that applied only to choices between different people and gave weight to numbers. It would give priority to the worse off when different people are concerned, but—unlike the principle that I have de scribed—it would say nothing about a case involving just one person. This view could be said to give importance to the difference between lives. It also fits Nagel's view of egalitarianism as involving a method for combining the claims of different people, although the method is not as simple as pairwise comparison. In fact it seems that we can only characterize the method as combining claims by giving priority to the claims of those who are worse off. To be intuitively plausible the view must reach the complex conclusion that we should give some priority, but not absolute priority, to the worse off, where the extent of the priority does depend on their absolute level of welfare. However, I think that these claims about priority between people will seem unsupported unless they are explained by a view about the value of a gain for a single person at a certain level of well-being.

²⁵ This consequence would not follow if the strength of the egalitarian reasons for helping the badly-off group also increased in strength as we added more members to the better-off group. But according to most views about how we should give moral weight to equality, this would not happen.

²⁶ If this difficulty is a real one it might lead egalitarians to state the priority view in terms of objectively valuable features of people's lives rather than utility or welfare.