¹ Nozick, R., Anarchy, State, Utopia (Oxford, 1974), p. 169Google Scholar. I have argued elsewhere that Nozick is correct, at least within the constraints of a sufficientist theory of distributive justice: proponents of coercive taxation are committed to the (temporary) conscription of labour. But Nozick is not correct (in my view) in rejecting coercive taxation on those grounds. See my Whose Body is it Anyway? Justice and the Integrity of the Person (Oxford, 2006), esp. ch. 3.

² Unless otherwise specified, all sources in the text and notes are to Cohen's works. The latter will be referred to as follows: History, Labour and Freedom (Oxford, 1988) – HLF; Self-Ownership, Freedom and Equality (Cambridge, 1995) – SO; If You're an Egalitarian, How Come You're so Rich? (Cambridge, Mass., 2000) – If; Rescuing Justice and Equality (Cambridge, Mass., 2008) – RJE.

³ On that count, he finds himself in good company: thus, Ronald Dworkin is thought, by some, to be guilty of similar bias. See Christofidis, M. Cohen, ‘Talent, Slavery and Envy in Dworkin's Equality of Resources’, Utilitas 16 (2004), pp. 267–87CrossRefGoogle Scholar.

⁴ See the works mentioned in n. 5, as well as ‘On the Currency of Egalitarian Justice’, Ethics 99 (1989), pp. 916–44.

⁵ In that respect, Cohen agrees with both Scheffler and Nagel, who defend agents’ personal prerogative not to enslave themselves to the worst-off. See Scheffler, S., Consequentialism and its Critics (Oxford, 1988)Google Scholar and Nagel, T., Equality and Partiality (Oxford, 1991)Google Scholar.

⁶ As we shall see, it is not clear whether Cohen rejects coerced labour on grounds of justice, or on grounds of values other (and more important) than justice.

⁷ For a very interesting discussion of what counts as socially useful labour, see White, S., The Civic Minimum (Oxford, 2003), ch. 5CrossRefGoogle Scholar.

⁸ Carens, J., Equality, Moral Incentives and the Market: An Essay in Utopian Politico-economic Theory (Chicago, 1981)Google Scholar.

⁹ See Wilkinson, T. M., Freedom, Efficiency and Equality (Basingstoke, 2000)CrossRefGoogle Scholar, for a discussion of the informational constraints faced by egalitarians. According to Nir Eyal, there is a further problem with the ethical solution, to wit, that teaching individuals to behave in such thoroughgoing and all-encompassing altruistic ways would in all likelihood necessitate considerable levels of intrusion. I do not take a stand on this issue. See Eyal, N., ‘Poverty Reduction and Equality with Strong Incentives: The Brighter Side of False Needs’, in Ryberg, J., Petersen, T. S. and Wolff, C. (eds.), New Waves in Applied Ethics (London, 2008), pp. 182–216Google Scholar.

¹⁰ For a highly sceptical view on this point, see P. Casal, ‘Rawls, Cohen, Mill, and the Egalitarian Ethos’, unpublished typescript. For the claim, which I develop in the remainder of this section, that justice might well impose a duty to provide personal services to those in need, see Arneson, R., ‘Property Rights in Persons’, Social Philosophy and Policy 9 (1992), pp. 201–30CrossRefGoogle Scholar. For a critical discussion of Cohen's egalitarian ethos, and an interesting account of what a (less demanding) Rawlsian ethos would look like, see Titlebaum, M., ‘What Would a Rawlsian Ethos of Justice Look Like?’, Philosophy and Public Affairs 36 (2008), pp. 289–322CrossRefGoogle Scholar.

¹¹ Note that the ethical solution to the trilemma does not say that A is under moral duty (or lack thereof) to choose doctoring. All it says is that if A chooses to become a doctor out of concern for fellow human beings, then there is no trilemma. It would be wholly coherent, for an anti-egalitarian, to accept that point, and yet maintain that A is not under a duty so to act. I am grateful to Cohen for helping me clarify that issue.

¹² ‘The Structure of Proletarian Unfreedom’, Philosophy and Public Affairs 12 (1983), pp. 3–33, at p. 20.

¹³ At those junctures where he seems to reject coercion in deference to some value other than justice, he does not provide an argument for the primacy of that value over justice, which leaves those criticisms of coercion incomplete. I shall return to that point in section 3.4.

¹⁴ As far as I can see, Cohen does not separate those two reasons in his text, but I take my interpretation to be an accurate reading of the relevant passage at RJE, 219.

¹⁵ I owe the example to Cohen.

¹⁶ One might think that if taxation is a casualty of stringent epistemic requirements (a price which Cohen would find difficult to pay), then the principle that D must be convicted only if the jury has knowledge of his guilt beyond reasonable doubt is a casualty of rough knowledge. But there are two related reasons for rejecting that thought. First, as I noted in the text, punishment differs from other kinds of coercion. Thus, although we have very good reasons for permitting rough knowledge at the point of taxing people, we also have very good reasons for requiring knowledge beyond a reasonable doubt of their guilt, should they default on paying their taxes (thereby making themselves liable to punishment).

Second, the reason why we might want to allow for rough knowledge in the case of egalitarian distributive justice is that stringent epistemic requirements come at a serious cost for the less fortunate – too serious, I venture here, given that the plight of the less fortunate is central to egalitarian justice. But the case of coercive punishment is very different. For the justification for punishing D resides, not in whatever beneficial effect it may have on third parties, but in D's wrongdoing. In order therefore legitimately to punish D, we need to know whether or not D is indeed guilty of such wrongdoing; but given that we need not take into account the interests of third parties when deciding whether or not to coerce him, we need not know whether others would be better off as a result of his punishment. Accordingly, we lack a reason to permit rough knowledge (of his guilt) to guide our decision. By contrast, the implications for the less fortunate of coercing or not coercing A provide a reason, or so I claim in the text, to allow rough knowledge to guide our decision. (Note that I invoke here a retributivist justification for punishment. Retributivism strikes me as far more plausible than other candidates such as deterrence or, appositely here, arguments which appeal to victims’ right that D should be punished.) I am grateful to Daniel McDermott for helping me clarify those points.

¹⁷ Someone objected to me that there is a difference between the two kinds of risk, to which my argument here is not sensitive, to wit: that in risking wrongfully to make A worse off or to violate her personal prerogative, we risk harming her; by contrast, in risking unjustly to deprive the less fortunate of A's services, we ‘merely’ risk allowing harm to happen to them. In so far as harming is generally worse, other things equal, than allowing harm to happen, we ought to take the latter, rather than the former risk. By way of reply: I agree that it is generally worse to harm than to allow harm to happen; but I dispute that it is always worse. Moreover, things here are not equal, since the risks run by A are smaller than those run by the less fortunate. Accordingly, it is not clear to me that the doctrine of doing and allowing supports the claim that we ought to reject coerced labour.

¹⁸ I owe this point to Nir Eyal.

¹⁹ For a discussion of the relationship between the Kantian requirement and the thesis of self-ownership, see Steiner, H., ‘Self-Ownership and Conscription’, The Egalitarian Conscience – Essays in Honour of G. A. Cohen, ed. Sypnowich, C. (Oxford, 2006), pp. 88–101CrossRefGoogle Scholar. Interestingly in the present context, Steiner argues that the thesis of self-ownership is compatible with the view that one may conscript someone else's labour for the sake of preventing and redressing the violation of third parties’ own self-ownership rights.

²⁰ Otsuka, M., ‘Freedom of Occupational Choice’, Ratio 21 (2008), pp. 440–53CrossRefGoogle Scholar.