¹ Rawls, when saying that inequality is justified only if the worst off group is better off in terms of primary goods than in any other possible distribution, assumes that the correct unit over which to compare distributions is typical life spans. So do writers like Ronald Dworkin and Thomas Nagel. For Dworkin's views, see his ‘What is Equality? Part 1: Equality of Welfare’, Philosophy and Public Affairs, x (1981)Google Scholar. A statement of Nagel's views can be found in his ‘Equality’, Mortal Questions, Cambridge, 1979Google Scholar.

² See McKerlie, Dennis, ‘Equality and Time’, Ethics, xcix (1989), 283–345Google Scholar, and Temkin, Larry, Inequality, Oxford 1993, ch. 8Google Scholar.

³ For simplicity I discuss only equality with respect to well-being. However, most of what is said below applies to other metrics of egalitarian concern, such as resources.

⁴ Parfit, Derek in his Lindley Lecture ‘Equality or Priority?’, University of Kansas, 1995Google Scholar. See also the recent discussion in Temkin, ch. 9.

⁵ This description covers only what Parfit in ‘Equality or Priority?’ calls Strong Egalitarianism. Another position, Weak Egalitarianism, claims that we should never lower the level of well-being for the sake of equality, since the loss in utility will always out-weigh the gain in equality. I shall be concerned only with the former view.

⁶ The priority view should not be confused with the law of diminishing returns, i.e. the often true empirical fact that the more resources a person possesses, the less good more resources will do him. The priority view does not concern what brings more or less well-being about. It concerns the different issue of the moral value of making someone better off.

⁷ For a detailed discussion of some of the basic assumptions behind one version of the levelling down objection, see Temkin, ch. 9.

⁸ McKerlie, 479. As McKerlies goes on to say, this has obvious analogues in our society. Elderly members of our society may be much worse off than younger members without this necessarily giving rise to objectionable inequality, according to the lifetime equality view. Similarly, Temkin imagines Jobl and Job2: ‘[Job!] and his family are healthy and wealthy. He has the love and respect of all who know him. In addition, his plans are realised, his desires fulfilled, and he has complete inner peace. Job2, on the other hand, has led a wretched life. His health is miserable, his countenance disfigured. He has lost his loved ones. He is a penniless beggar who sleeps fitfully in the streets, and whose efforts and desires are constantly frustrated.’ This will go on for forty years, after which ‘the situations of Jobl and Job2 will be reversed during the second half of their lives, so that in fact the overall quality of their lives, taken as a whole, will be completely equal’ (Temkin, pp. 235–6).

⁹ There are two complications which should be set aside. First, it might be suggested that if a life contains both times of great suffering and times of great happiness, this life is worse than if the same total sum of benefits and burdens were distributed more evenly. Someone might suggest that what is objectionable about changing places egalitarianism is that it involves this unfortunate distribution of well-being. I suspect that this line of reasoning is confused, and I shall ignore it in what follows. Secondly, there is an issue about individual responsibility. If simultaneous inequality is the product of peoples' free and rational choices, there may for that reason be nothing objectionable about it. Here I shall merely assume that we consider only cases of changing places egalitarianism in which facts about individual responsibility are not reasons for claiming that there is nothing objectionable about them.

¹⁰ See McKerlie, ibid.

¹¹ This is of course not to deny that if one can now foresee that future segments will contain inequality, then one should now do something to avoid it, if possible.

¹² McKerlie, 483.

¹³ These two ways of aggregating inequality are mentioned by both McKerlie and Temkin.

¹⁴ This is the view that McKerlie seems to accept, at least when applied to equality between simultaneous segments.

¹⁵ This is elaborated from an example of Derek Parfit's; see his ‘Comments’, Ethics, xcvi (1986)Google Scholar. And the particular version presented here has been helped by Parfit's comments on an earlier draft of this paper.

¹⁶ McKerlie also discusses this example as an argument against the lifetime equality view, and it is part of the reason that he finds a revision of that view plausible.

¹⁷ McKeriie's view indeed seems to be that it is because of the simultaneous inequality between A and B that A rather than B ought to receive the treatment.

¹⁸ This argument is of course just another instance of the levelling down argument.

¹⁹ One is obviously not forced by the severe suffering objection to give up the lifetime equality view. One might simply posit that the relief of suffering (or prioritarian concerns more generally) and lifetime equality are two separate issues, both of which are to be given some weight. Or, one might insist that simultaneous inequality as such is bad, besides lifetime inequality being bad. My only general objection against these proposals is that they might be less economical than merely assuming the priority view.

²⁰ McKerlie discusses this option.

²¹ This is Parfit's example. See Reasons and Persons, pp. 165–6.

²² Notice that the appeal to temporal asymmetry is quite different from the appeal to temporal distance. Temporal distance is neutral with respect to whether something is in the past or in the future; only the temporal distance as such matters, irrespective of direction (or, as explained, certain psychological relations that correlate with temporal distance matter, irrespective of whether they extend into the future or into the past). Temporal asymmetry, on the other hand, concerns only whether something is in the past or in the future; how distant in the past or the future something is located is irrelevant. Obviously, the views could be combined.

²³ This view, or a view very much like it, has recently been defended by Jeff McMahan. See Jeff McMahan, ‘Preferences, Death, and the Ethics of Killing’, Preferences, ed. C. Fehige and U. Wessels, forthcoming. This view need not deny that numerical identity as such matters for egoistic concern. Numerical identity as such might be a necessary but not sufficient condition for selfish concern. Thus the view under discussion differs from that defended by Parfit in Reasons and Persons.

²⁴ Notice that this view still takes the lifetimes of people to be relevant, but it denies that the goodness of peoples' lives as wholes should be compared.

²⁵ As this remark implies, the view under consideration shares a feature with the view that simultaneous inequality as such is bad. Where that view tells us to compare time-relative levels of well-being, the present view tells us to compare time-relative and discounted sums of well-being. Hence, they agree that simultaneous inequality with respect to some time-relative notion is bad. However, the two views disagree about what sort of simultaneous inequality matters.

²⁶ This again raises the question about how to aggregate across time simultaneous inequality with respect to time-relative and discounted sums of well-being between per-sons. Do these instances of simultaneous inequality add up, or do they rather cancel out?

²⁷ In addition to what is clear from Parfit's example, Quinn, Warren has shown how these two notions are importantly different in his paper ‘The Puzzle of the Self-Torturer’, Morality and Action, Cambridge, 1993Google Scholar.

²⁸ There are other versions of future directed priority views, i.e. ones that do not focus on total amounts of well-being.

²⁹ Another problem concerns the possible instability of the future directed priority view. Assume that we should at any time do most for those who from that time are worse off. This view may be hard to implement in a policy because it gives us new aims all the time. McKerlie discusses a similar objection to the future directed equality view.

³⁰ Nils Holtug suggested this excellent reply.

³¹ This is what Parfit argues in Reasons and Persons, pt. iii.

³² Another example is this. Suppose that C will suffer a decline in well-being tomorrow. Either we prevent this and thereby bring about a certain increase in well-being. Or we produce just as great an increase in well-being some days later in C's life. Again, those sympathetic to the priority view will, I think, believe that we should prevent C's decline in well-being rather than merely increase his level of well-being by the same measure, even when our efforts and the effects in terms of well-being are the same.

³³ One should not confuse this with diminishing marginal utility. To claim that the priority view must be sensitive to distributions of well-being is a general claim about the moral value of well-being. It is not a claim about what brings more or less well-being about.

³⁴ This reply was suggested by Derek Parfit.

³⁵ Numerous people have commented on earlier versions of this paper. I would in particular like to thank Derek Parfit for his extensive written comments, and Roger Crisp for suggesting many improvements. Also, I would like to thank Ingmar Persson, Dennis McKerlie, Nils Holtug, Kasper Lippert Rasmussen, Peter Sandøe, Govert den Hartogh and his workparty on practical philosophy at Amsterdam University, and audiences at Utrecht University for comments and suggestions on earlier versions of this paper.