¹ I say ‘greater than one’ rather than ‘not smaller than one’ to ignore the following position: it is impermissible for you to violate C even though you will thereby prevent exactly one violation by another person; if, however, your violating C reduces the overall number of violations of C (through reducing the number of violations by others), then it is permissible for you to violate C. I ignore this position for two reasons. First, no one has actually defended it. Secondly, and more importantly, the ‘one exactly’ position can safely ignore the challenge that I focus on in this paper (and which a defender of the ‘greater than one’ position must face): namely, how can it be impermissible for someone to violate C when her doing so will lead to fewer violations of C overall? (Of course, friends of the ‘one exactly’ position must respond to a related challenge: how can it be impermissible for someone to violate C when her doing so will not affect the number of violations of C overall?)

² See, for instance, Nagel, T., The View From Nowhere, Oxford, 1986, p. 178Google Scholar; Nagel, T., ‘Personal Rights and Public Space’, Philosophy & Public Affairs, xxiv (1995)Google Scholar; Nozick, R., Anarchy, State and Utopia, Oxford, 1974, pp. 28–30Google Scholar; Parfit, D., ‘Is Common-Sense Morality Self-Defeating?’, Journal of Philosophy, lxxvi (1979)Google Scholar, repr. in Consequentialism and Its Critics, ed. S. Scheffler, Oxford; Parfit, Reasons and Persons, Oxford, 1984, p. 98; Scheffler, , The Rejection ofConsequentialism, Oxford, 1982, pp. 80–114Google Scholar; Scheffler, ‘Agent-Centred Restrictions, Rationality, and the Virtues’, repr. in Consequentialism, ed. S. Scheffler.

³ Nozick, pp. 28 f.

⁴ Ibid., p. 30.1 take it that if A has a right against B that B does not V, then there is a constraint against B's V-ing. Also, if there is a constraint against B's V-ing, then for any occasion on which B may V there is some person who has a right against B that he does not V.

⁵ I say ‘a (that is, a single) question’ because, presumably, Nozick accepts the following bi-conditional. If, and only if, there is a rationale for accepting the-side-constraint view, one would not be irrational in accepting the view.

⁶ Nozick, p. 30.

⁷ Scheffler, , ‘Agent-Centred’, p. 252Google Scholar.

⁸ Ibid., p. 252.

⁹ I take it that if a moral theory says that a particular kind of state of affairs is good, then it ascribes to agents the goal of bringing about states of affairs of that kind. Scheffler says (in response to an attempt by Philippa Foot to defend constraints against the charge of paradoxicality: Foot, P., ‘Utilitarianism and the Virtues’, Mind, xciv (1985)CrossRefGoogle Scholar, repr. in Consequentialism, ed. S. Scheffler) that the paradox of constraints can ‘be formulated without using the notion of an “overall state of affairs”’, since we can simply ask: ‘How […] can it be rational to forbid the performance of a morally objectionable action that will have the effect of minimizing the total number of comparably objectionable actions that are performed and will have no other morally relevant consequences?’ (Scheffler, , ‘Agent-Centred’, p. 250Google Scholar). Now this shows that we can formulate the paradox without using the words ‘overall states of affairs’. But, surely, the effect which Scheffler describes in his simpler question falls under the concept of an overall state of affairs. (Compare: ‘the marathon ended two hours ago’ involves an occurrence of the event-concept though the word ‘event’ does not appear in the sentence.) And it would seem that Foot's criticism is directed against not the use of a particular word, but against the concept of one overall state of affairs being better than another.

¹⁰ Nozick, pp. 30–3: Darwall, Stephen L., ‘Agent-Centered Restrictions From the Inside Out’, Philosophical Studies, 1 (1986)Google Scholar.

¹¹ Of course, they may (but need not) have ends that will commit them to hold that the world is better if N* is achieved (e.g., because in most cases fewer people suffer if N* is achieved, as opposed to because there are fewer violations as such). So a friend of constraints may achieve his goals less well by conforming to constraints. But these goals are not goals that he is committed to in virtue of his commitment to constraints.

¹² Scheffler's argument presupposes that constraints ascribe to each agent the agent-neutral goal that no violations occur. Parfit seems to think this is begging the question against commonsense morality (Parfit, , ‘Is Commonsense’, p. 184Google Scholar). But he also thinks that a moral theory that includes constraints is directly, collectively self-defeating (i.e., that ‘it is certain that, if all rather than none successfully obey [constraints], we will thereby cause the [constraint]-given aims of each to be worse achieved’: ibid., pp. 174, 179 f.) and, thus, should be rejected. The constraint-given goal that Parfit has in mind here is that one harms as few innocent people as possible. My objection to the Conflict View applies to Parfit's argument as well. Constraints tell us, say, not to harm innocents. They do not ascribe to us as a goal that we minimize the number of innocents that we harm. By way of support of this claim, consider the fact that constraints are time-relative. That is, they forbid that an agent harms, say, one innocent to reduce the overall number of violations of innocents by him – see Kamm, Frances M., ‘Harming Some to Save Others’, Philosophical Studies, lvii (1989), 255Google Scholar.

¹³ Ibid., 252.

¹⁴ It is not always clear from Kamm's writings who is the subject that expresses respect and concern for rights. Is it the agent who performs a minimizing violation, people who adopt the view that minimizing violations are permissible, or the very permissibility of minimizing violations? I think the first two claims raise the same problem and I shall discuss only the first. The third I do not understand.

¹⁵ Strictly speaking, I should say ‘quasi-rights’, for I have defined constraints in such a way that one cannot permissibly minimize violations of constraints (for some n greater than one) and rights and constraints are intimately connected; see n.5.

¹⁶ Kamm, , ‘Harming’, 252Google Scholar. In fact, Kamm offers two different answers to the puzzle: the irrationality argument and the inviolability argument (see ibid., 254 f.; Kamm, , ‘Non-consequentialism, the Person as an End-in-itself, and the Significance of Status’, Philosophy & Public Affairs, xxi (1992)Google Scholar; Kamm, , Morality, Mortality, 2 vols., Oxford, 1996, i. 259–89)Google Scholar. I shall discuss only the irrationality argument. For a challenge to the inviolability argument, see Lippert-Rasmussen, Kasper, ‘Moral Status and the Impermissibility of Minimizing Violations’, Philosophy & Public Affairs, xxv (1996)Google Scholar.

¹⁷ When Kamm says that ‘[d]isregarding the instructions of [a] right shows disrespect for the right, not respect for it’ (Kamm, , Morality, p. 269Google Scholar) she may, of course, be interpreted simply as saying how she intends to use the verb ‘to respect’. But if that is so, then she ‘solves’ her puzzle by definitional fiat.

¹⁸ Kamm, , ‘Non-consequentialism’, 384Google Scholar.

¹⁹ Ibid., 384.

²⁰ Ibid., 383. What makes us inviolable, she says, are certain properties – she does not tell us which, though she suggests that having a rational will is one of them – that we have as individuals (ibid., 383).

²¹ Nagel, , ‘The Value of Inviolability’, ms., 23 05 1992, 8 fGoogle Scholar.; published as ‘La valeur de l'inviolabilité’, Revue de Metaphysique et de Morale, xcix (1994)Google Scholar.

²² Nagel, , ‘Personal’, 87 fGoogle Scholar.

²³ See Nagel, , View, pp. 175–85Google Scholar.

²⁴ In ‘Value’ Nagel wrote that Kamm explains – in a way that Nagel endorses and elaborates – constraints ‘in terms of the agent-neutral value of inviolability itself’ (my emphasis; Nagel, , ‘Value’, 11Google Scholar: ‘La valeur’, 155 f.). But I think this a misstatement of his views. In the corresponding passage in Nagel, ‘Personal’, he writes that Kamm (and Quinn) explain constraints ‘in terms of the universal but non-consequentialist value of inviolability itself’ (Nagel, , ‘Personal’, 89Google Scholar).

²⁵ True, the sort of state of affairs that Nagel has in mind is different from the sort of state of affairs most people have in mind when they deny that the permissibility of actions is solely a function of the goodness of states of affairs. This is so because Nagel considers moral facts states of affairs.

²⁶ Nagel seems to recognize something like this requirement when he says that ‘(t)he defense of noninstrumental rights depends on the judgment that the possession of certain sorts of inviolability, as a matter of moral status, is valuable enough to outweigh the logically correlated possibility of actually suffering the corresponding violation. It also depends on the judgment that this can outweigh some increase in the empirical probability of suffering the corresponding type of harm, considered apart from whether it is a moral injury or not’ (Nagel, , ‘Value’, 23Google Scholar: ‘La valeur’, 164). See also Nagel's remarks about thresholds in Nagel, , ‘Value’, 25Google Scholar: ‘La valeur’, 165.

²⁷ Nagel, , ‘Personal’, 91Google Scholar.

²⁸ Ibid., 92.

²⁹ The premises would still be different even if there is in fact nothing good which is not good for someone.

³⁰ Kamm makes both claims as well. She believes that if we are inviolable, then ‘the world is, in a sense, a better place, as it has more important creatures in it’ (Kamm, , ‘Non-consequentialism’, 386Google Scholar). She also believes that ‘such a higher status [i.e., inviolability] is itself a benefit to us’ (ibid., 386).

³¹ Henceforth I shall call arguments that instantiate this form ‘the-best-is-the-actual-arguments’.

³² It is not clear what Nagel means by ‘form’ here. But presumably he thinks that two arguments can have the same form (and both have true premises) and yet it need not be the case that either both of them are cogent or both of them are non-cogent.

³³ Nagel, , ‘Personal’, 92Google Scholar.

³⁴ Ibid.

³⁵ Perhaps this one: ‘A morality with rights and only a morality with rights recognizes the value of inviolability. If one and only one system of morality recognizes the value of inviolability, then that system of morality is true. Thus, a morality with rights is true. Thus, there are rights.’ Surely, offering this argument to opponents of rights is circular in the second way? For it is hard to see what reason one could have for accepting the argument's second premise other than the conclusion of the argument.

³⁶ What I am alluding to here is a version of what constitutes a general problem for moral theory: namely, that not all human beings (say, one-year-old children) may have the properties in virtue of which we believe human beings deserve special treatment. In this specific case, it may turn out that some human beings (e.g., one-year-old children) are violable.

³⁷ The same ambiguity can be found in Kamm's writings; see, for instance, Kamm, , ‘Non-consequentialism’, 386Google Scholar and Kamm, , Morality, p. 272Google Scholar.

³⁸ Nagel thinks that ‘if it were not wrong to torture us to prevent a greater evil, being tortured would be less bad than if it were wrong’ (Nagel, , ‘Value’, 22Google Scholar: ‘La valeur’, 163). So, presumably, he would accept that a similar point can be made with respect to the valuefor-us argument: namely, that in a non-benign world things may in fact turn out to be worse for us if we are inviolable, since the harms we suffer would then be worse for us.

³⁹ Nozick, pp. 45 f.

⁴⁰ In fact, Nozick seems to believe that we can permissibly sacrifice human beings by imposing a slight discomfort on them for the sake of saving (many more) animals from excruciating pain (ibid., p. 41). This suggests that the hierarchal scale should be calibrated for kind of sacrifice as well as kind of being.

⁴¹ Kamm operates a scale of moral status also. She claims, for instance, that if persons could permissibly harm each other, then they would ‘come down on the ladder of significance (as lions are lower on this scale than persons)’ (Kamm, , ‘Non-consequentialism’, 387Google Scholar). It is noteworthy that this claim presupposes that we operate an interval scale of moral status (i.e., that equal differences between the numbers that we assign to different levels of status correspond to equal differences in moral status, something which is not true of ordinal scales). Suppose the scale is ordinal. Suppose also that it is still impermissible to harm persons for the sake of animals. In that case, persons retain their ranking in the scale of moral status of different kinds of beings and no step downwards has been taken. Are our intuitions about moral status really so fine-grained as to allow us to operate an interval scale of moral status?

⁴² Nagel, , ‘Personal’, 93Google Scholar.

⁴³ Ibid..

⁴⁴ See Nagel, , Equality and Partiality, Oxford, 1991, p. 149Google Scholar.

⁴⁵ Nagel, , ‘Value’, 24 f.Google Scholar: ‘La valeur’, 164.

⁴⁶ It also ignores the situation in which we are violable but no one performs minimizing violations (though they could). In the following, I shall ignore this situation as well as situations in which some, but not all, of those who can perform minimizing violations actually do so.

⁴⁷ It seems that Kamm would say that (c) is worse than (b) (Kamm, , Morality, p. 282Google Scholar). In a discussion of a world in which 1) there is ‘in reality’ a constraint, 2) we (except one) believe there is one, and 3) this exceptional and deluded person performs a minimizing violation, she suggests that this world is worse than one in which the deluded person had not been deluded and had not violated the constraint. The former world is worse because events in this world ‘are not in accord with the truths of moral reality’. This claim is puzzling, since neither are events ‘in accord with the truths of moral reality’ in a world where the deluded person had not performed a minimizing violation. In fact, more events in such a world seem not to be ‘in accord with the truths of moral reality’.

⁴⁸ I need to say ‘presumably’, because I have argued that it is not clear what the necessary and sufficient conditions of showing respect for a constraint are.

⁴⁹ See Foot, ‘Utilitarianism’, and Dancy, Jonathan, Moral Reasons, Oxford, 1993, p. 221Google Scholar. G. A. Cohen generously wrote extensive and highly revision-inducing comments on the penultimate draft of this paper. I am also grateful to G. A. Cohen, Roger Crisp, Nils Holtug, Karsten Hint Jensen, Klemens Kappel, Jesper Ryberg and Peter Sandøe for some very helpful discussions of earlier drafts. Thanks are due to the Carlsberg Foundation for its financial support.