¹ Throughout, when I talk about an agent ‘not existing’, I will mean ‘never exists’.

² McDaniel, Kris and Bradley, Ben, ‘Desires’, Mind 117 (2008), pp. 267–302CrossRefGoogle Scholar, at §§ 5–6 argue that desires are best understood as involving a three-term relation between an agent who has the desire, a proposition that is the object of the desire, and a proposition that is the condition of the desire. So for instance, Karen desires that she buys a book at the shop, on the condition that no-one has given it to her first. A desire of this sort is satisfied if the condition is true (no-one has given the book to Karen first) and the object of the desire (Karen buys the book at the shop) is true. It is frustrated if the condition is true and the object of the desire is false. And the desire is cancelled if the condition does not obtain. This analysis, which is prompted by the need to accommodate conditional desires, is elegantly extended to unconditional desires as a special case where the condition is trivial: that something or other is the case. Although McDaniel and Bradley formulate desires quite differently than I have done here, the division of states of affairs that I have suggested above can readily be modified to fit McDaniel and Bradley's understanding of desires. Simply substitute McDaniel and Bradley's definitions of satisfaction and frustration for my eponymous states of affairs. The definition of Absence remains unchanged, and thereby includes the case of cancelled desires.

³ Roberts, Melinda A., ‘Can it Ever Be Better Never to Have Existed At All?’, Journal of Applied Philosophy 20 (2003), pp. 159–85CrossRefGoogle ScholarPubMed; Holtug, Nils, Persons, Interests, and Justice (Oxford, 2010)CrossRefGoogle Scholar.

⁴ Johansson, Jens, ‘Being and Betterness’, Utilitas 22 (2010), pp. 285–302CrossRefGoogle Scholar.

⁵ See, e.g. Keller, Simon, ‘Welfare and the Achievement of Goals’, Philosophical Studies 121 (2004), pp. 27–41CrossRefGoogle Scholar. Keller advocates a particular theory of the value of goal-satisfaction for welfare, while leaving it somewhat open what is the relative value of simply lacking a given goal (p. 38).

⁶ Here are some further technicalities which I will be ignoring in the substance of the argument: first, strictly speaking, I should make some accommodation for the strength of a desire, and perhaps also for the time at which the desire is held. I ignore these complications for simplicity, but they could easily be incorporated for completeness, later. Second, you may have concerns about the status of past desires. Do past desires count just as much as present or future ones? I will assume that the status of a desire as past, present or future is irrelevant, on the assumption that much the same arguments would apply even if there is temporal bias in the valuation of desire-satisfaction. For a recent defence of the claim that no time-bias is required, see Bykvist, Krister, ‘The Moral Relevance of Past Preferences’, Time and Ethics: Essays at the Intersection, ed. Dyke, H., (Dordrecht, 2003), pp. 115–36CrossRefGoogle Scholar. See also McDaniel and Bradley, p. 291.

⁷ , Broome, Weighing Lives (Oxford, 2004), pp. 165–6Google Scholar. Broome himself is following Raz, Joseph, ‘Value Incommensurability: Some Preliminaries’, Proceedings of the Aristotelian Society 86 (1985), pp. 117–34CrossRefGoogle Scholar. Some use the term ‘incomparable’ to refer to the relationship of being incommensurate in value. See Nien-hê Hsieh, ‘Incommensurable Values’, Stanford Encyclopedia of Philosophy, ed. E. N. Zalta, <http://plato.stanford.edu/archives/fall2008/entries/value-incommensurable/,2008>, § 2 for further discussion.

⁸ Thanks to Lloyd Humberstone for assistance in formulating this. Humberstone has pointed out to me that this definition is not adequate if we apply it to a domain in which only two things are of value, and they are putatively incommensurate. I overlook this complication here, because I am interested in a much larger domain than that.

⁹ Broome, p. 165.

¹⁰ Fehige, Christoph, ‘A Pareto Principle for Possible People’, Preferences, ed. Fehige, C. and Wessels, U. (Berlin, 1998), pp. 508–43CrossRefGoogle Scholar. Fehige appears to overlook the possibility of the fourth disjunct. Hence he concludes that the third disjunct is true.

¹¹ Fehige frequently refers to ‘preferences’, rather than ‘desires’, and says he intends these terms interchangeably with ‘wants’ and ‘wishes’ (p. 509 n.). Putting wants and wishes aside, I want to resist the thought that preferences and desires are interchangeable, and believe that Fehige's argument is best taken to be an argument about the value of satisfying desires. Henceforth I refer only to desires. If there is any prospect of reducing one to the other, Krister Bykvist, ‘Changing Preferences: A study in preferentialism’ (PhD Thesis, Uppsala University, 1998), ch. 1 provides a promising approach to the project.

¹² Fehige, p. 510.

¹³ The example is Fehige's, pp. 511–13.

¹⁴ Fehige, pp. 513–14.

¹⁵ Strictly speaking, Fehige is asking the wrong question here – at least for my purposes, since I am focusing on impersonal value. It should not be ‘Have we done her a favour?’, but ‘Have we made things better, simpliciter?’ I take it the question still has rhetorical force, in either case.

¹⁶ Fehige formulates this argument more rigorously than I have done here, using a precisely defined Pareto principle for possible desires. However, his set-up has the distracting feature of using a material conditional sentence ‘If A desires that p, then p’. Such sentences, being true when A has no desire that p, or when p, provide a neat device for explicating Fehige's view that both Absence and Satisfaction are equal in value, and are both better than Frustration. But in presenting Fehige's argument to others, I have found audiences become overly distracted by the semantics of conditionals, and fail to engage with the content of the view. So I am forgoing some rigour in the interests of avoiding distraction. Those interested may consult Fehige's original paper (pp. 519–23).

¹⁷ Fehige, p. 524.

¹⁸ Benatar, David, Better Never to Have Been (Oxford, 2006)CrossRefGoogle Scholar, defends a similar view, but in a broadly hedonistic framework, rather than a preferentialist/desire-satisfactionist one. In his generally approving discussion of Fehige's argument, Benatar claims it is important to distinguish the relative value of two different sorts of absent desire: those which are absent, but would have been satisfied and those which are absent, but would have been frustrated. Benatar claims that would-be frustrations are better than actual frustrations, and that would-be satisfactions are merely no worse than actual satisfactions (pp. 56–7). I note that classifying Absences in this way arguably renders Benatar's view incomplete, since it is not the case that, for any absent desire, it either would have been satisfied or would have been frustrated: some desires might be either.

¹⁹ Fehige, p. 522.

²⁰ The idea of maximizing consequentialism as opposed to optimizing is due to Sen, Amartya, ‘Maximization and the Act of Choice’, Econometrica 65 (1997), pp. 745–79CrossRefGoogle Scholar; ‘Consequential Evaluation and Practical Reason’, Journal of Philosophy 98 (2000), pp. 477–502. See Hsieh, Nien-hê, ‘Is Incomparability a Problem for Anyone?’, Economics and Philosophy 23 (2007), pp. 65–80CrossRefGoogle Scholar, for an elaboration and defence of this idea in the context of incommensurate value.

²¹ It is possible to object to Fehige by assuming the principle I mentioned earlier: person-affecting supervenience. If matters of impersonal value depend on personal value in that way, an obvious question that the stronger claim (7) invites is: who benefits in the empty world, which is allegedly a better world than this? Surely it is an embarrassment to say that it is better for non-existent people! But I think there are similar difficulties in the vicinity for anyone who thinks that existence can be a harm, so I will not press this worry against Fehige here.

²² Much the same argument is made in Gustaf Arrhenius, ‘Future Generations: A Challenge for Moral Theory’ (Ph.D. thesis, Uppsala University, 2000), pp. 83–4.

²³ Fehige's view, whereby Satisfaction and Absence are equal in value, he calls ‘anti-frustrationism. Arguably that name could apply equally well to limited incommensurability, but I introduce a new name so as to keep the positions distinct.

²⁴ Parfit, Derek, Reasons and Persons (Oxford, 1985), pt. 4Google Scholar.

²⁵ The project has been taken further by Arrhenius who gives an impossibility proof of developing an axiology that meets the sorts of constraints that Parfit expects any adequate theory to meet.

²⁶ Arrhenius, p. 83.

²⁷ Contra, for example Cowen, Tyler, ‘What Do We Learn from the Repugnant Conclusion?’, Ethics 106 (1996), pp. 745–75, at 757–8CrossRefGoogle Scholar, who says that views that – like mine – embrace incommensurability, constitute ‘giving up’ on normative population theory. This sort of attitude is seemingly based on the thought that, if the apparatus of decision theory cannot be made to fit our practical problems in a straightforward way, then no moral guidance can be given. But our problems are real, and decision theory is just a tool: a tool which manifestly does not do the job. So we need to change the tool, not deny the nature of reality.

²⁸ On the analysis of harm in this connection see, e.g. Hanser, Matthew, ‘Harming Future People’, Philosophy & Public Affairs 19 (1990), pp. 47–70Google ScholarPubMed; Harman, Elizabeth, ‘Can we Harm and Benefit in Creating?’, Philosophical Perspectives 18 (2004), pp. 89–113CrossRefGoogle Scholar; Emilio Mora, ‘Understanding Harming Actions: A Defense of the Harm-State Theory’ (MA Thesis, Monash University, 2008); Parfit, app. G; Shiffrin, Seana V., ‘Wrongful Life, Procreative Responsibility, and the Significance of Harm’, Legal Theory 5 (1999), pp. 117–48CrossRefGoogle Scholar.

²⁹ Broome, p. 167.

³⁰ If this were never the case, then the stark incommensurability theorist would have to endorse a weak version of the repugnant conclusion, in which a much larger population, in which no-one has a high level of welfare, is not worse than a smaller population in which everyone has a high level of welfare.

³¹ Thanks to Daniel Cohen, Patrick Emerton, Christoph Fehige, Caspar Hare, Lloyd Humberstone, Simon Keller, Neil Levy, Emilio Mora, Robert Sparrow and an anonymous referee for helpful comments relating to this article.