¹ Singer, Peter, The Most Good You Can Do: How Effective Altruism Is Changing Ideas About Living Ethically (New Haven and London, 2015)Google Scholar; MacAskill, William, Doing Good Better: How Effective Altruism Can Help You Make a Difference (New York, 2015)Google Scholar.

² David Brooks, ‘The Way to Produce a Person’, The New York Times, 3 June 2015, <http://www.nytimes.com/2013/06/04/opinion/brooks-the-way-to-produce-a-person.html>.

³ MacAskill, Doing Good Better, ch. 5.

⁴ Hume, David, A Treatise of Human Nature, ed. Selby-Bigge, L. A. ([1739] Oxford, 1978)Google Scholar, bk. 3; An Enquiry Concerning the Principles of Morals, ed. L. A. Selby-Bigge ([1777] Oxford, 1982); Bentham, Jeremy, An Introduction to the Principles of Morals and Legislation, ed. Burns, J. H. and Hart, H. L. A. ([1789]; Oxford, 1996)Google Scholar, ch. XI; Moore, G. E., Principia Ethica (Cambridge, 1903)Google Scholar, sec. 103; Driver, Julia, Uneasy Virtue (Cambridge, 2001)CrossRefGoogle Scholar; Bradley, Ben, ‘Virtue Consequentialism’, Utilitas 17.3 (2005), pp. 282–98CrossRefGoogle Scholar. One might also respond that virtue ethicists are only concerned with particular virtues and not with what makes persons good overall. While I do not think this is the case, not much hangs on this question for the purposes of this article.

⁵ Hume, Enquiry, p. 289.

⁶ Hume, Treatise, bk. 3; Hume, Enquiry.

⁷ Driver, Uneasy Virtue, p. 82.

⁸ See Bradley, ‘Virtue Consequentialism’ for a critique of Driver's view and a discussion of alternatives.

⁹ Driver's neutrality w.r.t. axiology is one feature that differentiates her from Hume. But there are others. First, for Hume, moral judgements depend on how one responds to a character trait upon surveying it. Driver rejects this sentimentalist aspect of Hume's theory, because a trait having good consequences and people typically perceiving it positively can often come apart (Driver, Uneasy Virtue, pp. 63–6). Second, Hume gives virtues a more central role in general. He holds that when we morally assess acts, we do so only as indicative of character traits (Hume, Treatise, pp. 348–9, 477, 575). Driver, on the other hand, holds that there is an independent assessment of acts as right and wrong to be distinguished from questions of praiseworthiness/blameworthiness and personal evaluation (Driver, Uneasy Virtue, pp. 73–4).

¹⁰ Pettit, Philip, The Robust Demands of the Good: Ethics with Attachment, Virtue, and Respect (Oxford, 2015)CrossRefGoogle Scholar, ch. 5.

¹¹ Pettit, The Robust Demands of the Good, p. 140.

¹² A second difference is that for Pettit virtues are only a subset of the disposition-dependent goods necessary for doing good. To be a good person requires, besides the virtues traditionally identified by virtue ethicists, dispositions of attachment and respect.

¹³ Rawls, John, A Theory of Justice (Cambridge, MA, 1971)Google Scholar, ch. 9; Daniels, Norman, Justice and Justification: Reflective Equilibrium in Theory and Practice (Cambridge, 1996)CrossRefGoogle Scholar.

¹⁴ Also, by not defining ‘being a good person’ in terms of having a good character, I avoid the need to provide an analysis of when something is or is not part of a person's character.

¹⁵ A scalar evaluation is useful even for (some) positions that hold that being a good or bad person is a binary notion. For example, we could hold that someone is good, if she falls into a certain range of ‘betterness’, and bad, if she falls below (or even allow a third, neutral category in between).

¹⁶ I will assume that personal identity is sufficiently stable diachronically for us to make such judgements. Alternatively, we can restrict such judgements to time spans during which personal identity is sufficiently stable.

¹⁷ The Objective Act Principle, of course, requires an account of act-individuation applied consistently across individuals. See Davidson, Donald, ‘Actions, Reasons, and Causes’, Journal of Philosophy 60 (1963), pp. 685–700 CrossRefGoogle Scholar; Goldman, Alvin I., ‘The Individuation of Action’, The Journal of Philosophy 68 (1971), pp. 761–74CrossRefGoogle Scholar; Carter, Ian, A Measure of Freedom (Oxford, 1999)CrossRefGoogle Scholar, ch. 7, on possible ways to do so.

¹⁸ I take these distinctions from Parfit's discussion of evidence-relative, belief-relative and fact-relative wrongness (Derek Parfit, On What Matters, vol. 1 (Oxford, 2013), ch. 7).

¹⁹ The topic of this article in general is of interest not only to consequentialists. But it is, of course, of particular relevance to consequentialists – particularly global consequentialists. Global consequentialism is the idea that consequentialism should be about everything. See Parfit, Derek, Reasons and Persons (Oxford, 1984), pp. 24–5Google Scholar; Kagan, Shelly, ‘Evaluative Focal Points', Morality, Rules, and Consequences: A Critical Reader, ed. Mason, Elinor, Hooker, Brad and Miller, Dale E. (Lanham, MD, 2000), pp. 134–55Google Scholar; Pettit, Philip and Smith, Michael, ‘Global Consequentialism', Morality, Rules, and Consequences: A Critical Reader, ed. Mason, Elinor, Hooker, Brad and Miller, Dale E. (Lanham, MA, 2000), pp. 121–33)Google Scholar.

²⁰ Another problem is that on this principle a person would sometimes be judged a better person simply because she can do more acts, and thus more right acts, than another. One can avoid this conclusion by using the proportion of right to wrong acts instead of the cardinality of right acts. However, such a revised principle would still fail to include all relevant goodness/badness. If, for example, the proportion of right to wrong acts is identical for P and Q, and P’s wrong acts do much more bad than Q’s wrong acts, then the revised principle would judge P to be as good a person as Q.

²¹ It will also result in more plausible judgements in other cases. First, on M₁ , persons whose worst possible course of action is already very highly positive would tend to be judged much more favourably than those whose worst possible course is positive but closer to zero. Second, the previous measure, M₁ , is not defined when the best option is equal to zero. This is different for M₂ . (Though M₂ is not defined in situations in which the denominator is zero. But, as I will show in section V.2, this is not a problem.)

²² I take the name from Carter, Ian, ‘Distributing Freedom over Whole Lives’, Arguing about Justice: Essays for Philippe Van Parijs, ed. Gosseries, Axel and Vanderborght, Philippe (Louvain-la-Neuve, 2013), pp. 135–43Google Scholar.

²³ But we only include those courses of actions people are able to do. See Jackson, Frank and Pargetter, Robert, ‘Oughts, Options, and Actualism’, The Philosophical Review 95 (1986), pp. 233–55CrossRefGoogle Scholar, on related questions.

²⁴ This is also why we consider ‘courses of action’ rather than acts. For if we focused on individual acts whilst also wanting to include the probabilities with which a person is going to act in certain ways, we would run into the problem that a person with bad dispositions to act would be judged much too favourably.

²⁵ A further measurement question might be whether we should be concerned not only with the best and the worst courses of actions but also with how many courses there are in between and their respective levels of expected goodness. More complicated measures could be devised to account for this. For most situations, and to convey the intuitive idea, however, the simple measure here is fully satisfactory.

²⁶ This is a mere intuition and not meant as a worked-out principle to assess persons. For example, one might also argue that doing something wrong correctly believing it to be wrong makes one a worse person than doing something wrong falsely believing it to be right.

²⁷ See Korsgaard, Christine, ‘From Duty and for the Sake of the Noble’, Aristotle, Kant, and the Stoics: Rethinking Happiness and Duty, ed. Engstrom, Stephen and Whiting, Jennifer (Cambridge, 1998), pp. 203–36Google Scholar; Herman, Barbara, ‘On the Value of Acting from the Motive of Duty’, The Philosophical Review 90.3 (1981), pp. 359–82CrossRefGoogle Scholar; Williams, Bernard, ‘Acting as the Virtuous Person Acts’, Aristotle and Moral Realism, ed. Heinaman, Robert (London, 1995), pp. 13–23 Google Scholar, for (comparative) discussions of Kant's idea of moral worth and Aristotle's idea of moral excellence.

²⁸ One complication is that acting with bad beliefs and motives might not only make the person but also her acts worse. For example, an otherwise identical act might be worse if done out of bad instead of good motives. The relationship I am interested in is whether a person's bad beliefs/motives might in themselves make her a worse person. To remain neutral with regard to questions of axiology, I will remain agnostic about whether motives/beliefs make acts worse or better.

²⁹ See Crisp, Roger, ‘Utilitarianism and the Life of Virtue’, Philosophical Quarterly 42 (1992), pp. 139–60CrossRefGoogle Scholar, for arguments to the effect that a life of virtue will have the highest expected value.

³⁰ A different response will be to widen the theory of the good such that having a certain motivation and certain dispositions in itself makes the world a better place. As explained earlier, on Pettit's view being good implies doing good, and doing good implies doing certain things robustly, that is, across a set of relevant possible worlds (Pettit, The Robust Demands of the Good). In Racist Prisoner, we could say that the prisoner is not a good person, because the good things he does are accidental and not modally robust. Committing to this type of axiology is possible and would make the Proportionality Principle a more plausible candidate as an all-things-considered principle. However, the aim of this article has been to see whether ‘being good by doing good’ is plausible for a broad range of axiological theories. But if Pettit is right about disposition-dependent goods, then I think the Proportionality Principle would be the most plausible principle to flesh out his theory. If Pettit's theory of the good is false, however, my above suggestion to combine the Proportionality Principle with other principles would be the adequate response.

³¹ There are alternatives to assuming that the Proportionality Principle has varying inter-principle strength. We could, for example, hold that the Proportionality Principle only applies under the condition that a person has sufficient abilities to do good or bad. Such a move would be in line with ‘holism about reasons’, according to which for a moral reason to have a certain valence, it requires the presence of enabling conditions (or the absence of disabling conditions) ( Väyrynen, Pekka, ‘Moral Generalism: Enjoy in Moderation’, Ethics 116.4 (2006), pp. 707–41CrossRefGoogle Scholar; Dancy, Jonathan, Ethics without Principles (Oxford, 2006 Google Scholar), pt. I). The absence of sufficient abilities to do good/bad would then function as a disabling condition. Whether we should go with the variable ‘inter-principle strength response’ or the ‘holism about reasons response’ depends, first, on how we spell out the rest of the principles and, second, how we decide to move from individual principles to judgements about how good a person is overall. It is, of course, beyond the scope of this article to defend all principles and to develop an all-things-considered account for balancing different principles in cases of conflict. The precise form of such a theory will also depend on one's first-order normative commitments about right actions. For example, the right way to move from individual principles to overall judgements might not necessarily be to ‘weigh’ principles and trade them off with each other. For example, if deontological duties make certain acts impermissible, then either the Proportionality Principle does not apply to many such situations or only applies when the deontic duties have been discharged first. Or it might mean that other principles of personal evaluation take lexical priority over the Proportionality Principle. The ‘variable inter-principle strength’ response makes more sense, if we use a balancing approach. It makes less sense, for example, if we use a lexical order (in which case, the ‘holism response’ would be more plausible).

³² The example is, of course, conditional on empirical assumptions about the effectiveness of aid which I cannot discuss here (Deaton, Angus, The Great Escape: Health, Wealth, and the Origins of Inequality (Princeton, 2013)Google Scholar, ch. 7). See Singer, Peter, The Life You Can Save: How to Play Your Part in Ending World Poverty (New York, 2009)Google Scholar and Singer, The Most Good You Can Do on the ethics of giving and global poverty.

³³ The extent to which this implication holds depends on a number of philosophical and empirical questions. For a discussion of the ethics of career choice see MacAskill, William, ‘Replaceability, Career Choice, and Making a Difference’, Ethical Theory and Moral Practice 17.2 (2014), pp. 269–83CrossRefGoogle Scholar.

³⁴ It is sometimes objected that even if non-human animals have moral standing, individual acts of buying meat do not matter at all in terms of expected goodness. See Matheny, Gaverick, ‘Expected Utility, Contributory Causation, and Vegetarianism’, Journal of Applied Philosophy 19.3 (2002), pp. 293–7CrossRefGoogle Scholar; Mark Budolfson, Bryant, ‘The Inefficacy Objection to Consequentialism and the Problem with the Expected Consequences Response’, Philosophical Studies (2016 Google Scholar, forthcoming), for discussion.

³⁵ I would like to thank Hasko von Kriegstein, Roger Crisp, Johannes Himmelreich, Benjamin Wald and two reviewers for generous comments. For helpful discussions, I thank Francis Dennig and the audience at a talk at the University of Toronto.