Agency, Authority, and Responsibility

Pierre Cloarec PhD candidate, Philosophy, University of Sherbrooke and Paris-Sorbonne University pi.cloarec@gmail.com Agency, Authority, and Responsibility This is a draft, not to be quoted without permission. A previous version of this paper was presented at Mancept Workshops in Political Theory, 2014. Abstract Responsibility is an important issue in discussions of justice, and luck egalitarianism owes much of its appeal to its making clear how it could be incorporated in an egalitarian conception of justice. Yet it is also a notoriously puzzling issue, that calls for a greater clarity if we are to clarify how it can rightly be used as grounding justicebased claims. It is appropriate, then, to begin by some conceptual clarifications on the many meanings of “responsibility.” This serves to understand how differing conceptions thereof could fit into a broader conception of justice. Here I propose to start from the question: “What is it that makes a distribution of responsibilities a fair one?,” rather than to assume, as luck egalitarians do, that the adequate distribution of responsibilities and associated burdens and benefits is one that tracks attributive responsibility as finely as possible. I attempt to show that, for political purposes, responsibility is best understood in relation to agency and authority over oneself, rather than as moral responsibility commonly understood. The bulk of my argument is that, if we understand equality in a relational sense, we should accept the idea that persons are to be viewed as agents possessing a presumptive authority over themselves, and hence that the way responsibilities are ascribed to them should reflect this view. Moreover, if equality and agency matter, the sense of one’s own worth ought to be socially secured. If my argument is right, strict luck egalitarianism is unable to do so, although a revised version thereof might be advanced that meets the requirements advanced here. Keywords Equality, responsibility, luck egalitarianism, relational egalitarianism, agency, authority Agency, Authority, and Responsibility Responsibility is an important issue in discussions of justice. It is important both to oneself and with respect to others. If we believe that a burdened person brought her plight onto herself through actions she was responsible for, we might rightly refuse to aid or otherwise support her. “You deserved it,” we might say. But it is also important to oneself in several respect. It is significant to us that we be held responsible, or that a responsibility be attributed to us, for some aspects of our life or of our actions, or that we hold some authority over what we do or what choices we make, or that what befalls us depend, to some extent, on what we did to bring it about. Yet responsibility issues are notoriously puzzling, and when we speak of responsibility we have to clarify what we are talking about, and what the scope of responsibility is. We have to ask: What is a fair distribution of responsibilities? Here I provide some arguments against a common characterization of responsibility, that embodied in the luck-choice distinction as used by luck egalitarians. Their answer to the question is: “A fair distribution of responsibilities is one that tracks individual choices, for which persons are responsible, but is insensitive to brute luck, for which persons are not responsible.” Rather, I suggest that what makes responsibility important lies in the way persons conceive of themselves and of each other, namely, as agents involved in a plurality of (broadly understood) social practices and expecting others to participate in those practices as responsible agents. A fair distribution of responsibilities is thus, or so I argue, not one that tracks precisely individual responsibility, but one that satisfies some reasonable egalitarian principles. This approach is relatively broad, and how it is to be applied to more specific issues (such as issues of distributive justice, or of gender relations) remains an open question, for dealing with these issues requires adding some empirical and conceptual elements that cannot be straightforwardly derived from the general framework itself. 1 Responsibility: Why care? Some brief indications of why responsibility is important are in order. Although it is not the place here to develop a full-blown argument for that hypothesis, it remains disputable whether responsibility is that much important. Simply claiming that it is would be question-begging, for one might reasonably deny that responsibility is that important, at least at the fundamental level. Or perhaps we need not use that vocabulary. Firstly, one might object, it is implicit in the idea that freedom matters for justice, and secondly, it only obscures an already complex discussion. (This paper is in part an attempt at defusing these potential objections.) Unless we have good reasons to Agency, Authority, and Responsibility 3 think that responsibility is important, any attempt at defending some form of responsibility-sensitive egalitarianism lacks appropriate motivation. Unfortunately, I cannot provide more than some sketchy observations on that matter. The basic reason why egalitarians should be concerned with responsibility is that we, as members of our society and participants in various social practices, take it to be a matter of concern. Now that claim is ambivalent. For that members of a society take something to be a matter of concern may not suffice to affirm that it really ought to be a matter of concern. We need to supplement that basic reason. Otherwise, it could be replied that, although many people do care about responsibility, political philosophers need not do so. On what ground, then, can it be claimed that we should care about responsibility? Three main arguments support this thesis. 1 Firstly, responsibility is part of what makes our social practices intelligible. Secondly, we cannot, so to speak, avoid issues of responsibility. Thirdly, responsibility is a non-negligible aspect of moral agency. (To be sure, these arguments, though distinct, are related.) Now, it is unfortunately not the place here to develop these arguments more fully. Suffice it to say that, if one accepts some variant of at least one of them, one has at least some ground to think that egalitarians should be concerned with responsibility. In the remainder of this paper, I propose a way to articulate this concern. 1. McTernan [Forthcoming] develops in more detail the first argument and some aspects of the third, building on Scanlon’s “Value of Choice” view. The idea of the inevitability of responsibility would need more extensive explanation than I am able to offer here. In short, the idea is that it sometimes happens that it is not possible not to assign a responsibility for something, regardless of whether the issue of responsibility arose in the first place. Think of a natural disaster: no one is responsible for its occurrence in the first place, but it is not possible not to assign the associated burdens to some agent, be it the victims (who are then held responsible for how they cope with post-disaster difficulties) or the government (which is then held responsible for providing disaster relief or funding reconstruction). Agency, Authority, and Responsibility 4 2 The concept of responsibility The concept of responsibility, as well as how it is to be articulated within a broader conception of justice, are notoriously puzzling issues on their own, and things only worsen when one explores the bearing of responsibility on questions of justice or fairness. Several distinct meanings are attached to the notion of “responsibility”, and these meanings do not always overlap. Let me first summarize the conceptions of responsibility one could appeal to. The most basic is causal responsibility: roughly speaking, a person is causally responsible for an outcome as soon as some actions of hers figure substantially in the best causal explanation of this outcome. 2 This generic definition is of course subject to discussion, as it seemingly implies that one could be causally responsible for some outcome even if she did not originate the causal chain that brought this outcome about. Setting aside the possibility that no one could ever be causally responsible for anything, that is, that no one could ever be said to have originated any causal chain, there remains serious problems with causal responsibility. However I do not discuss these difficulties here: suffice it to notice that causal responsibility is itself a puzzling issue, and that it is neither sufficient, nor necessary that one be causally responsible for an outcome for this person to be responsible in some other sense. I now turn to more clearly normative conceptions of responsibility. One is said to be morally responsible for an action or for an outcome in the attributive sense if this action or outcome has certain features that exposes one to moral appraisal, congratulation, or blame. 3 Whether causal responsibility is required for one to be morally or attributively responsible crucially depends on what account of causality we rely on, but I will not discuss this further. It is simply worth noting that causal responsibility may not be required in order to ascertain moral responsibility; for example, if a person, call her Annie, fails to prevent someone else (Bernie) to perform an action that Annie knows is or perceives as wrong in some sense, Annie may be morally responsible, even though she is not involved in the causal chain that involves Bernie. When we say “You deserved it,” we usually rely, at least implicitly, on moral responsibility, in that we take some aspects of a person’s action to expose her to moral appraisal. It is plausible to include not only the performance of an action, but the process through which it was performed, in our evaluation: say, if we consider that a person has failed to acquire the knowledge that we think would have been relevant to make a more adequate decision. Notice that attributive responsibility is also an explanatory conception, in that it is used to identify who is responsible for what, while it does not entail, in itself, that one should bear the consequences of an action for which she can reasonably be held responsible. Finally, it is important keeping in mind that, in part due to its explanatory aspect, the ascription of attributive responsibility is subject to conflicting “stories,” to the effect that one and the same event may be described in several ways as regards attributive responsibility. Substantive responsibility, by contrast, reflects who should bear the burden for a given event or action. As such, it is not of great help, for it might as well be arbitrarily ascribed. It need not reflect attributive responsibility, although the former can be used as a justificatory support for the ascription of substantive responsibility. We could 2. Dworkin [2011: 103]. 3. See Miller [2001] and Scanlon [1998: 248]. Agency, Authority, and Responsibility 5 nonetheless imagine that substantive responsibility be ascribed randomly. Clearly, this would be pretty curious, and the usual case is that substantive responsibility is ascribed with reference to a more defensible criterion, one, that is, that is rationally intelligible, even if it turns out to be erroneous. One important thing to notice is that there is no “natural” link between attributive and substantive responsibility, to the effect that it is conceivable that one be held attributively responsible for something without having to bear the burden associated to, or without enjoying the benefit stemming from this thing (whatever it is). Finally, it is worth noting that, although substantive responsibility is socially distributed, it is distinct from responsibility as a social attribution, for reasons I briefly present below. Responsibility as a virtue is a moral concept, too, but it is necessary to distinguish it from moral responsibility in the above sense. To make the distinction clear, a simple example is enough. Sometimes a person can be morally responsible for an action or an outcome, in that what she does exposes her to moral appraisal, and yet be said to be irresponsible in that she does not act virtuously. Consider a person who knowingly drives too fast and puts other people at risk. It is clear that she is morally responsible for this behavior under the assumption that the person cannot claimed to have been ignorant of speed limits and other traffic rules, or of how dangerous a relentless driver can be for other people. Yet we may say of this person that she has an irresponsible behavior. When we say so, we rely on responsibility as a virtue: a responsible behavior is one that meets a standard of adequacy (for a given practice). The last two usual conceptions of responsibility could be confused. These are responsibility as assignment and responsibility as social attribution. One is responsible in the assignment sense when one is assigned a specific task, be it through direct order (“You are responsible for keeping record of this meeting”) or through a more general rule (“The last person to leave the office is responsible for switching off the lights”). 4 By contrast, responsibility as social attribution is used to determine how some decision powers are distributed in a society, or who is attributed the authority required to make certain choices. In this sense I am responsible for the choice of my clothes. 5 In both cases, whether one is responsible for something is dependent upon a social decision or convention, hence the possible confusion. In either case, it is reasonable to fear that responsibilities could be arbitrarily distributed. Neither view can, on its own, serve as a basis for the distribution of responsibilities: it is necessary that, however responsibilities are distributed or assigned, this distribution be justified. I can now make clear what the distinction is between substantive responsibility and responsibility as a social attribution. Both are socially distributed, and could be confused, but they are nevertheless different. Substantive responsibility, as I said earlier, reflects the way the consequential burden of responsibility is ascribed to persons in a society. It is thus a consequential notion, and not one that reflects what power or authority one possesses. By contrast, responsibility as a social attribution reflects the way decision powers are distributed, and is not, in itself, a consequential notion. It would be abusive to assume that the two are completely disconnected or independent, but it remains important to notice how they are distinct. 4. Responsibility as assignment is so defined by Dworkin [2011: 103]. 5. The definition of responsibility as social attribution and the “choice of clothes” example are borrowed from Fleurbaey [1995: 44]. Agency, Authority, and Responsibility 6 I do not discuss here the detail of each conception. 6 Neither do I examine how they could fit into a more fully articulated theory of justice. Rather, I wish to highlight two issues related to the normative conceptions of responsibility: that of agency, and that of authority. To each of the above normative definitions (that is, all but that of causal responsibility) corresponds an account of agency and of authority, be it, with respect to the latter, authority over oneself or over others. These relations I can only sketch here; note that this is not intended to provide a definitive account of how agency and authority relate to responsibility. This will serve to make clear what difficulties a responsibility-sensitive account of justice faces and has to deal with. Two last observations are in order. First, to say that one is morally responsible for an action or an outcome is to say that, at some time, she has had (and perhaps still has) the authority to perform or not to perform this action or to bring out this outcome. 7 If one does not have such an authority, that is, if what action one performs, what outcome she brings about, or what she achieves does not depend on what she is willing to do, she cannot be said to be morally responsible. Notice that this need not entail acceptance of any thesis regarding the free will problem: that one has authority over a decision could be compatible with the absence of free will. I do not provide a further argument here: suffice it to say that one’s having the authority to make a decision makes her morally responsible for what she decides to do. To clarify: if one is in position of control over something (which is to say that she possesses an authority over this thing), then it follows that how she acts, and what results therefrom, can rightly be attributed to her. Here however the link between authority as legitimacy and authority as control is, it should be noted, weakened. Second, responsibility as a virtue need not, contrary to what its labeling suggests, entail a perfectionist conception of agency or the endorsement of a virtue ethics. Although this might be the case, I focus here on a nonperfectionist understanding of the virtue of responsibility, which corresponds to what we can reasonably expect from each other, and is compatible with a wide array of nonmoral conventions. In this sense we say that one is a responsible driver when one does not drive too fast, or does not lack the kind of prudence expected from a driver. So it corresponds to a kind of attitude or behavior, but these are to a large extent conventional; this kind of the virtue of responsibility is moral only to the extent that we expect other people to act according, or comply to, or abide by, certain rules. It is thus not defined against the background of virtues, but by reference to the social practice at stake and its norms. Although it can be defined in abstract and general terms for a given practice, the virtue of responsibility is closely linked to the practice considered, and cannot be described in an entirely abstract manner. 6. My objective, I should make clear, is not to devise a conceptual map of the meaning of “responsibility.” I take this to be important as a preliminary step, not as a self-standing analysis. 7. It is not required that one still has this authority at the time of performance. Frankfurtstyle cases are an example of cases where we can assert that one is morally responsible for an action or an outcome, even though she lost the possibility to do otherwise (Hurley [2003] analyzes such a case). More generally, it is conceivable that one locks herself, through a pattern of past actions for which she was responsible (over which she had the relevant authority), in a situation where she has no alternate possibility. This holds regardless of the free will problem itself. Agency, Authority, and Responsibility 7 3 Luck egalitarianism: some objections Any responsibility-sensitive account of justice may embody several of the above conceptions, suitably refined. Yet, curiously, moral responsibility has received much more attention than the virtue conception and variants of the social attribution thesis. It is the sense in which persons are said to be responsible when one works out a theory of fairness that relies on the luck-choice distinction. Following the now common usage, I will call this kind of theories luck egalitarian conceptions. 8 It is not, for the present purpose, necessary to distinguish sharply between the various conceptions that have been attributed this common label. 9 Generally speaking, a theory of fairness can be called luck egalitarian whenever it rests primarily on a distinction between one’s initial circumstances or opportunities, over which one has no control, and one’s willful choices, for which she is responsible. That the conception of responsibility embodied in this approach to fairness is moral responsibility should be clear enough, since one important principle of luck egalitarianism is that a person is to held liable for the actions she performs, the choices she makes, and the consequences thereof, insofar as she had the relevant opportunities to do otherwise. 10 Following Tan, I simply call this the luck-choice principle. This principle, and more broadly luck egalitarianism, have received various criticisms. 11 Most of the discussion has focused on luck, 12 on the one hand, and what makes a choice genuine (including the free will problem), on the other hand. I do not discuss this at length. Rather, I focus on the account of agency and authority that a responsibility-sensitive conception of justice should embody. To be sure, the luck-choice principle is appealing, and it is indeed difficult to resist it.13 Yet it should at least be amended. How, then, should we account for responsibility, if we are to avoid reliance on the luck-choice distinction? Before sketching a more positive account, it is important to note that the luck-choice principle is ambivalent, for 8. The label has been attributed by Anderson [1999]. Fleurbaey [1995] uses the term “factorsensitive egalitarianism” instead, which is a bit larger. The latter includes luck egalitarianism, but is not limited to it, in that it includes the endowment-insensitive, ambition-sensitive view advocated by Dworkin [1981a, 1981b], as well as pure resource egalitarianism. 9. See Anderson [2007], where a distinction is made between desert-catering luck egalitarianism and responsibility-catering luck egalitarianism. Here I focus on responsibilitycatering luck egalitarianism, but some of my observations could, with adequate revisions, apply to desert-catering luck egalitarianism as well. 10. So writes Arneson [1990: 178]: “When persons enjoy equal opportunity for welfare, any actual inequality of welfare in the positions that they reach is due to factors that lie within each individual’s control and hence is nonproblematic from the standpoint of distributive justice.” See also Arneson [1989]. At the most general level, a similar idea is formulated by Cohen [1989: 931] when he states that “a large part of the fundamental egalitarian aim is to extinguish the influence of brute luck on distribution,” while “genuine choice excuses otherwise unacceptable inequalities,” as well as by Roemer [1993]. Tan [2008] has recently defended an approach that he calls “institutional luck egalitarianism” that relies on the same fundamental thesis. See also the replies by Sanyal [2012] and Schemmel [2012]. 11. See inter alia Anderson [1999], Fleurbaey [1995], Hurley [2003], Scheffler [2003, 2005], Spitz [2008]. 12. See for example Barry [Barry 2006]. 13. Wolff [2010: 348] thus suggests that “there seems to be a larger element of luck egalitarianism in [Anderson’s] own view than she is prepared to own up to.” Agency, Authority, and Responsibility 8 what counts as (brute) luck and what counts as choice may be difficult to determine. It is necessary to draw a boundary between the two, but wherever such a boundary is drawn, there will be reasons to doubt that it is adequately located. At the extremes, one could say either that most, if not all, that befalls us stems from luck, or that we should be held fully responsible of the vast majority of our actions. Neither of these extreme views is really appealing, but I believe that even more moderate or intermediate views are ultimately flawed. Throughout this argument I consider that equality of opportunity, in the luck egalitarian sense (including equivalence of each person’s array of available options), is implemented. One problem with the luck-choice principle is that it does not reflect an adequate approach to agency. Although it is a defensible attempt to introduce a concern for responsibility into distributive justice, it treats individual actions as more separate than they actually are. Indeed, the outcome of one’s action is not determined solely by this action, and thus the very same action may lead to different outcomes, depending on how others act. What, under such conditions, is the point of being responsible for what befalls one? Arneson has suggested that one should make the most prudent choice. The main problem with this is that there are several ways to determine what the most prudent choice is, given that, for any choice, there is a wide array of possible outcomes. Suppose we had cardinal values and probabilities for each outcomes. Then we could propose many decision making rules, such as maximin, arithmetic mean, and so on, to determine which choice is the most prudent, the second most prudent, and so on. But then, provided we have several ways of ranking one’s possible actions, we would have several rankings and no way to determine which is the best for a given individual. This could of course be left to individual choice, but, ironically, this would also start a curious regression. Is this individual responsible for the ranking she favors? Now suppose she always acts against a maximin rule, i.e. always chooses the action whose worst possible outcome is the highest possible. There are two consequences to this. First, it is plausible that this person will, more often than not, choose to act in a way that, all things considered (that is, given what actions others have performed), is worse than what she might have done. She simply avoids many risky choices, and, because of her prudence, fails to accomplish much. So the recommendation that one acts in the most prudent manner will often cause harm to individuals. That persons are diverse, and thus that some will judge that the most prudent choice should be determined otherwise, does not change much for those who act according to the maximin rule. But there is a second consequence. If some persons systematically follow the maximin rule, they will affect everybody else’s outcomes as well. For we can admit that outcomes for each person are not independent of other persons’ choices. So it is enough that some persons choose the maximin to induce losses to many other persons. In this sense, the most prudent choice for one may reduce the overall outcome, as well as individual outcomes. To this a proponent of luck egalitarianism might reply that, if persons can know that some courses of actions will not be chosen by others, they can be pretty sure that some outcomes will be avoided. Thus they could apply the maximin rule to a reduced set of possible actions, even though some of these actions might be less prudent overall than the remaining ones. This is a serious possibility, but such cases are unlikely to be the most common, and it is easy to imagine situations in which it does not work. Therefore Agency, Authority, and Responsibility 9 a reply like this cannot be appealed to as a general reply to the worry that the most prudent choice will more often than not not be the most intelligent choice. (I use “most intelligent” in a general, nonconceptual sense here.) Now, a luck egalitarian could also reply that what others do is a matter of luck, and thus that one is not responsible for the part of her action’s outcome that depends on others’ actions. For this reply to work, however, it is necessary to add that one cannot possibly what others are likely to do. But if one knows that, for a given action, the possible outcomes are such and such, she is still responsible for the choice of this action. She knows that, if others act in a certain way, a specific outcome will be brought about, even though she has no foreknowledge of what others will actually do. Thus, the fact that, from each person’s point of view, an action’s outcome partly depends on luck is not fully relevant. Not having such foreknowledge does not prevent one from knowing what will befall her if she chooses to act in this or that way. Therefore, she is still responsible for the outcome insofar as she knew she could have acted in a way that guaranteed her a better outcome. Now a proponent of luck egalitarianism might propose a background proviso in the form of an equalization of the worst possible outcomes of every possible course of action. Under such a scheme, the threat described above would cease to exist, since no single course of action would threaten any person to end up less well-off than any other course of action. Although this proposition would avoid the threat described above, it is implausible, since it is also plausible that some courses of action would have a worst possible outcome higher than the highest possible outcome of many other courses of action. Under circumstances of this kind, one would face a serious dilemma: either it would be necessary to level down the worst possible outcomes of every action, or it would be necessary to level up the outcomes of some actions up to an implausible point, that is, beyond any adequate threshold. On the one hand, the first possibility significantly reduces the appeal of luck egalitarianism, and so luck egalitarians should avoid it. On the other hand, the second possibility is more plausible than it may seem. Indeed, it need not be a requirement of luck egalitarianism that, if some possible courses of action are dominated by some other ones, that is, if some possible courses of action are always “worse” than some other ones, and thus if the former are never rationally chosen, then the former be “improved” so that they are at least as good as the latter. A luck egalitarian could rightly affirm that, if some course of action is such that it will not rationally be chosen, then it is not required to compensate individuals as long as they have available to them some better courses of action. Still, it might be replied that, since an action’s outcome depends on what others do, or on how others react, they have a share of responsibility of their own in the overall result. If so, persons could be, on a luck egalitarian outcome, attributively responsible for their actions, and nonetheless not be responsible for some of these actions’ outcomes. Each person would be partly responsible for everybody else’s fate, to the extent that each would have contributed to the distributive outcome. Here it is possible to construct an argument similar to that of Cohen 14 against the “incentive argument.” If the responsibility for the overall distributive outcome is shared by many persons (presumably the whole population), in such a way that anyone’s situation depends on many other persons’ actions and reactions (not only on her own actions), then many of the resulting distributive inequalities ought to be compensated, because it is only 14. G.A. Cohen [1997, 2008]. Agency, Authority, and Responsibility 10 occasionally the case that they can be attributed to any specific person. Consequentially, the luck egalitarian principle allows much less distributive inequality than it may be thought. In a parallel manner, it is arguable that the brute v. option luck distinction does not entail harshness toward those who fare less well due to “their” choices. Nicholas Barry provides half of this argument. The idea is that, even when one seemingly makes choices of one’s own, or, in other words, when one seemingly faces option luck, there remains an important part of brute luck, since any resulting distributive inequality may be owed, all things considered, to others’ actions or reactions. Barry’s argument does need to be significantly modified, because he proceeds from actual social circumstances, where “two persons may make different decisions as a result of their very different backgrounds.”15 The reply I assume a luck egalitarian could propose is different in that I suppose, rather, counterfactual circumstances, where luck egalitarian equality of opportunity (for welfare, for advantage, and so on) holds. When such circumstances hold, already existing inequalities play no role, since, by hypothesis, they do not exist (or, in the case of “natural” inequalities, such as inequalities of innate talent, are appropriately compensated) in the first place. Still, the reply is similar in its principle, and reaches the same conclusion as Barry’s argument. For even when equality of opportunity holds, option luck may not play such a great role, as soon as outcomes do not depend solely on “individual” choices. Therefore, as before, the luck egalitarian principle allows much less inequality than it may be thought. 16 Such a reply comes at a price, though. For the appeal of luck egalitarianism stems from its sensitivity to individual responsibility. Should a luck egalitarian advance this reply, she would face a severe dilemma. On the one hand, the reply avoids the harshness objection, and so defuses an important objection to luck egalitarianism. On the other hand, it diminishes the importance of individual responsibility, since in many cases, a person’s fate will turn out to be owed to brute luck, rather than individual choice. Much of luck egalitarianism’s appeal is lost in the process. In other words, luck egalitarians seemingly have to make a hard choice: either they cannot face the harshness objection, or they cannot account for personal responsibility. If I am right, luck egalitarians will rather abandon this reply, and adopt a mixed view in order to face the harshness objection.17 There remains two ways in which these replies are not satisfying. One is that they do not fit well into a more general conception of justice, that relies on decentralized markets as a means to produce and distribute goods. The other is that they do not fit well with an otherwise appealing view of agency. The former objection has been developed by Anderson,18 and I need not develop it further here. I advance an argument in support of the latter objection in the following section. 15. Barry [2006: 101]. 16. See also, for a similarly intended argument, Voigt [2007]. 17. Arneson [2000] acknowledges the strenght of Anderson’s criticisms, and proposes a deeply revised version of luck egalitarianism, which he labels “responsibility-catering prioritarianism,” as a reply these criticisms. Since he pushes in favor of a prioritarian proviso, he is able to defuse the harshness objection without running afoul of the luck egalitarian view of personal responsibility. His response does, however, bring further difficulties, since it turns out to rely on moral merit as a ground for distributive justice. 18. Anderson [2007]. Agency, Authority, and Responsibility 11 4 Agency and security What problem does this induce as regards agency? The first thing to notice is that, under the conditions assumed above, it is plausible that persons may enjoy a great freedom as non-interference, and yet face serious practical constraints. Indeed, they are likely not to choose many possibilities that would have been better from their own point of view, had not these possibilities entailed serious risks. It is enough that, as I have assumed so far, they know these risks in order for them to be responsible for their fate, should these risks realize. In other words, they lack a form of background security.19 The point here is not only that persons will miss otherwise interesting opportunities due to this lack of background security, but also that their agency is thwarted when this security is lacking. Why is this so, or, otherwise stated, why should a lack of security thwart one’s agency? Admittedly, persons have some abilities to choose the course of action they want to perform. This course of action may fail to produce the outcome the person desires, and it may so fail for many reasons. This can be due to a lack of understanding or mistakes on the part of the person, and these are, I believe, pretty common cases, but here we may rightly focus on cases where such failures stem from others’ actions. In such cases, a person may act in a clever and adequate way and yet fail to reach the desired outcome because other persons have not performed the actions necessary for this outcome to be brought about. Think of a retailer who examines carefully the possibility to establish a shop in a town. She has examined the local circumstances and has serious reasons to believe her shop will be successful. She supplies goods people in the town appreciate at affordable prices, and she sees no reason to believe her business will fail. However, the shop turns out to be unsuccessful and, despite her efforts, the retailer runs out of business. Under a luck egalitarian scheme, she could have avoided bankruptcy by not opening the shop, and thus she is responsible for her failure. Even if it is admitted that people in the town hold a share of responsibility in this failure (by not attending the shop), the retailer is likely to hold the financial responsibility for her failure. Many persons, I contend, would rather not take such a risk, despite their will to perform a variety of possibly risky, yet not inconsiderate, 20 actions. Furthermore, in case of failure, it is likely that their authority over several aspects of their life will be diminished in otherwise unjustifiable ways. If agency matters, then, security is important. Security supports agency in that one’s ability to make choices is better supported when she is not, as a matter of practice, constrained to act in ways she would rather have avoided. Notice that this constraint is implicit: it depends on whether the person is willing to expose herself, and possibly others (her relatives, friends, or colleagues), to risks that may cause her and others serious harms, or induce an implicit cost (that is, cost her the possible gain she could not earn because of the associated risk). What kind of security is required in order to 19. There is an obvious parallel between “background security” and “background justice” in the Rawlsian sense (see Rawls [1999]). Background security could rightly be embodied within background justice taken as a broader approach. 20. What counts as “inconsiderate” is subject to discussion, if not controversy. Whether one, for instance, should be rescued at no cost for her if she attempts to climb a mountain under bad weather conditions (which she should be aware of), is open to controversy, but I will not try to settle this here. Agency, Authority, and Responsibility 12 protect agency and preserve a sense of responsibility is, to be sure, a difficult issue. The problem shows up as a rather common dilemma. On the one hand, the requirement that persons be held responsible for their actions entails that they should not be protected against the risks they face because of these actions. On the other hand, a lack of security entails that many risky, yet potentially beneficial actions will not be performed. Given that people cannot usually anticipate how others will act, and that at least some people will prefer security (in this context, maximin) to an uncertain benefit, a lack of security entails a loss both for individuals and for society as a whole. However, security may, though not systematically, reduce the importance of personal responsibility to an unacceptable extent. This is true both with respect to other people, who may have to bear excessive costs if one acts in an irresponsible manner (or if they are not held substantively responsible for their actions), and to oneself, in that one may, as I believe many people do, feel that what she does has much less effect on her own life than she could expect. In this sense, many people would be deprived of a sense of achievement, because they would be unable to determine whether their fate resulted from their own actions. This need not be related to any kind of reward, although my argument is compatible with that. This argument is a bit more complex than it may seem. Having a sense of achievement means that (a) persons can legitimately expect that the benefits they bring about for themselves remain in part theirs, (b) they can legitimately expect that others’ failures or losses will not induce similar failures or losses for themselves, (c) they will be able to acknowledge their own mistakes or failures. These are, to be sure, pro tanto claims. Being responsible, to some extent, for one’s own fate does not entail that one is entitled to keeping all the gains for herself, while possibly leaving others in dire straits. Similarly, expecting not to bear too great a cost because of others’ mistakes or failures does not entail that one can refuse to provide them some form of security or compensation. Acknowledging one’s mistake or failure does not entail either that one cannot enjoy the benefits of security, or that others are entitled to deny one these benefits. Finally, acknowledging one’s mistake or failure need not be a humiliating or otherwise demeaning action. This last observation is, as Anderson 21 stresses it, of great importance, and should not be overlooked. It is a central tenet of an egalitarian theory that treating people as inferiors, in this case by humiliating them through the recollection of their mistakes, is a strongly non-egalitarian and unjust practice. So, that one should be able to acknowledge her mistakes or failures should not entail, from an egalitarian standpoint, that she should undergo a humiliating practice of exposing them to, say, government officials. Such “shameful revelations,” as Wolff 22 calls them, ought to be avoided, for they constitute a serious threat to the sense of one’s worth. This consideration partly trumps the claims that persons should go with the consequences of their voluntary actions, that is, that they should be held fully substantively responsible for the consequences of the actions for which they are attributively responsible.23 This is not to say that responsibility plays no role in distributive justice. Rather, it sets a principled limit (the exact position of which is a 21. Anderson [1999]. 22. Wolff [1998: 113sq]. 23. See Anderson [2007: 243–246] and Scanlon [1998]. Agency, Authority, and Responsibility 13 matter of actual social arrangements) on the degree to which one’s being attributively responsible can be turned into a distributive loss for her. Now it is also necessary to provide a positive account of what it means to acknowledge one’s mistakes or failures. This need not entail viewing oneself as personally failed. Rather, it means acknowledging that, as human beings, we are prone to making mistakes, wrongly evaluating our chances of success, and so on, as well as the possibility to learn from our mistakes. There is nothing inherently humiliating in this process. It may become humiliating, but this results from specific social practices, and is not an internal characteristic of a learning process of this kind. Acknowledging one’s mistakes, or more generally one’s proneness to making mistakes, amounts simply to acknowledge that one is, to some extent, attributively responsible for these mistakes. Indeed, acknowledging oneself as an agent entails the acknowledgment that one’s actions may be erroneous, even though these actions may be socially influenced. To what extent one’s actions can be explained solely with reference to one’s various dispositions, desires, and so on, is not highly relevant here. The issue at hand is not so much an explanatory one as it is a normative one, and it can easily be admitted, without also engaging in a metaphysical or sociological controversy, that most persons hold some attributive responsibility in their own actions, that is, that those actions possess features that make it appropriate to attribute them to specific persons. 5 How is equality fundamental? As I noticed earlier, the principle that persons should be treated as equals partly trumps the requirement that justice should be sensitive to responsibility. This calls for more precision and clarity. Treating each other as equals is, as I see it, a basic requirement of egalitarian justice, and one that takes precedence over other requirements of justice.24 The problem here is that not all egalitarian principles are taken to trump other considerations. So let me clarify in what sense equality is fundamental, and when considerations other than egalitarian ones can trump a requirement of equality. Here it is important to notice the contrast between relational and distributive equality. That relational equality trumps other considerations does not entail that distributive equality, or the reduction of economic equality, ought to be the norm. Yet it is pretty common to consider that justice requires the reduction of 24. It could be objected to this that I beg the question in favor of relational egalitarianism, or even of egalitarianism itself. Why, it may be asked, should one ever accept the principle of relational equality? What can be said in support of this principle? There is nothing so obvious in the principle that it would be exempt from discussion. Unfortunately, I cannot devote more than a footnote to this difficulty. The most readily available response is to affirm that there can be no presumption of inequality prior to actual social arrangements, that is, that any personal trait taken to entail a relational inequality can be so construed only within a particular social arrangement, while it could be utterly irrelevant in some other social arrangement. Some of these personal traits do not even exist outside of a specific social arrangement. Starting from actual relational inequalities without accounting for the social arrangement where they exist thus misses the point. Otherwise stated, the objection seems to beg the question in favor of current social arrangements, and therefore it fails to address correctly the principle of relational equality. Agency, Authority, and Responsibility 14 inequalities, if not their elimination in specific cases. As I see it, the principle of relational equality is more fundamental than requirements of distributive equality, but its distributive implications are more indirect. From the principle of relational equality, it does not follow that the distribution of the relevant assets ought to be egalitarian. However, that persons ought to be treated as equals supports the considerations advanced by non-intrinsic egalitarians. 25 Following O’Neill, I use the label “nonintrinsic egalitarianism” for this kind of view. According to it, (a) distributive equality is motivated by a set of egalitarian reasons, including, but not limited to, the alleviation of suffering, the elimination of status harm, and so on; and (b) pluralist egalitarianism allows that these considerations be trumped by, say, liberty, economic efficiency, and so on, including some principle of personal responsibility (O’Neill uses this to avoid the well-known leveling down objection). Although I believe relational egalitarianism is itself a pluralist view, and although non-intrinsic egalitarianism provides a powerful rationale for distributive equality, it is worth noticing the contrast between the two views. This should also shed light on the way responsibility can be embodied in an egalitarian theory. Non-intrinsic egalitarianism allows that demands of equality be overridden by other considerations, while, as I indicated earlier, relational equality is not to be overridden by other considerations. Nevertheless, the distance between relational and non-intrinsic egalitarianism should not be overestimated. Both are pluralist views in that both allow for a plurality of values beyond equality, but they differ in that the latter is more directly concerned with distributive justice. Importantly, however, non-intrinsic egalitarianism can embody relational egalitarianism, and indeed relational egalitarianism provides a relevant support for the non-intrinsic view (one that O’Neill simply suggests, without developing it any further). My thesis, for short, is that (a) the views are compatible, (b) the principle of relational equality is more fundamental than the non-intrinsic view, (c) it contributes to reinforce the non-intrinsic view, and (d) it helps understanding the contrast between non-intrinsic and deontic egalitarianism. I cannot provide a complete argument in favor of these four claims here. The following observations are only partial and a fuller account of the relation between the two views remains necessary. It is to be noted that proponents of non-intrinsic egalitarianism (here we can include, with adequate precaution, Rawls and Scanlon) take the preservation of self-respect or the avoidance of status harm as motivating the quest for distributive equality (or for less inequality). These are, it seems to me, important aspects of relational egalitarianism. In this sense, if one accepts the principle of relational equality, one should also support non-intrinsic egalitarianism. Whether accepting an intrinsic principle of equality, such as that of relational equality, is compatible with a “non-intrinsic” view, might seem puzzling, but this is misleading. Indeed, non-intrinsic egalitarianism is intended as a distributive view, and thus it is the distributive requirements of the view that are labeled “non-intrinsic.” That an intrinsic non-distributive principle of equality could support a non-intrinsic distributive principle need not surprise us. Interestingly, this implies that relational equality is more radical in its implications than it might seem, and that it has been argued against Anderson’s view. Indeed, if my argument gets things right, (a) relational equality is a basic value that can be trumped only by extremely weighty and urgent considerations, 25. These include, for instance, Rawls [2001], Scanlon [2003], and recently O’Neill [2008]. Agency, Authority, and Responsibility 15 if there are any; (b) it is compatible with a wide array of other social values (that free and equal persons can reasonably endorse), and can be embedded in a plurality of distinct social arrangements; (c) it strongly supports non-intrinsic egalitarianism, which is itself a pluralist egalitarian view. Furthermore, it is to be noted that relational egalitarianism avoids the leveling-down objection, since this objection is opposed to strict, unqualified distributive egalitarianism (a view that few persons actually defend). 6 A presumption in favor of authority over one’s own life The combination of agency and equality entails a presumption in favor of authority over one’s own life. Indeed, as soon as we view each other as equals, and treat each other accordingly, there can be no presumption in favor of authority of one over the other. Such a presumption cannot avoid appealing to the idea that there are wiser persons, whose judgment ought to be followed by those who have fewer such abilities. Were there such a presumption, persons would not be treated as equals. This is not to say that, in the context of some particular social practices, no one can be said to have a more accurate, or wiser, or otherwise better judgment. It seems pretty obvious that, say, an engineer’s judgment is better than a layman’s as regards bridges or railroads. Such cases do not, however, entail a relational inequality. The engineer’s superior authority as regards bridges or railroads stems from a series of social practices, including the division of labor displayed in any relatively complex society. It does not extend beyond that, and does not stem from the presumption that engineers are superiors and laymen inferiors. Some explanations are required as regards the sense of “authority” in this context. The term is indeed ambiguous, for it may refer, on the one hand, to the degree of control one possesses over something, and, on the other hand, to who is in a legitimate position to issue a demand, make a choice, or perform an action. So one may be in position to control something, and yet not have the relevant legitimacy to do so, while, in a so to speak symmetrical manner, one may have the required legitimacy and yet not be in position to control something. What is it that is involved if we claim that persons have a presumptive authority over themselves? Both conceptions, it seems, could be appealed to, but it is arguable that the “legitimacy” view should enjoy priority in this case. Indeed, in a society of equals, no one having a presumptive authority over others means that no one can legitimately issue a demand on others, or make choices on their behalf, out of a mere presumption that they are less wise, or less intelligent, or morally inferior. As a matter of principle, it does not make a relevant difference that someone be in position to control something if she cannot do so legitimately. In other words, the question we must address is: Who possesses the presumptive legitimacy to issue demands, make choices, and so on? In a society of equals, this presumptive authority is that of each person over herself, but, as I note below, this presumption is not one that cannot be overridden. More generally, the presumption in favor of authority over one’s own life indicates that, unless a relevant and powerful enough reason is presented, persons are not to try to rule other persons’ life, unless they possess a legitimate authority. Not much need be said here about this kind of legitimate authority, namely, democratic authority; suffice Agency, Authority, and Responsibility 16 it to say, as a short digression, that it is not, in itself, incompatible with treating each other as equals.26 What matters is that the presumption in favor of authority over one’s own life is not one that cannot be overridden. It follows from the principle of relational equality, but it does not entail that political authority can never be justified. Relational equality does put a stringent requirement of justification on the exercise of political authority, but it does not rule it out as illegitimate or unjustifiable altogether. This justification, of course, is to be addressed to equals, and the resulting authority must remain compatible with the principle of relational equality. Note that, so understood, the principle supports, if not leads to, the principle of political equality as advocated by Gaus, namely, that All inequalities of political authority must be justified; inequalities within the law-making institution must be conclusively justified. Inadequately justified differences in political authority are unjust. Differential political power granted on the basis of superior moral or political competence cannot be publicly justified.27 Indeed, given that justification is addressed to equals, not to inferiors, Gaus’s principle needs to rely on the view that persons are equal, and that they ought to treat each other as equals. If this is granted, any differential in political authority must be justified in an egalitarian manner. But, since inequalities of political authority may, and do turn out to be necessary for the proper functioning of a large and complex society, such inequalities stand in need of justification. In this sense, it appears that there is no contradiction between treating persons as equals and allowing inequalities of political authority, to the extent that the latter are adequately justified. Here, obviously, the issue at hand is what is taken to be an adequate justification of such inequalities of political authority. A more stringent standard of justification is likely to lessen the extent of inequalities of political authority, while a somewhat loose standard is likely to allow greater such inequalities. A relational egalitarian view entails, I believe, a rather demanding standard of justification, since a looser standard would allow differentials in authority that would lead to treating some (perhaps many) persons as inferiors in several ways, including disrespecting their judgment or browbeating them into some specific kinds of behavior. How does this presumption bear on issues of responsibility? Authority over one’s own life, as I see it, is claimed by many persons. In this sense, persons claim a share of substantive responsibility, in that they claim a share in what we may call the social division of labor that expresses their status as an equal and their authority over their own lives. The adequate share of responsibility is not determined solely along the lines of attributive responsibility: it is not enough, as the difficulties luck egalitarians face show, that one be attributively responsible for something for her to bear substantive responsibility for this thing. More importantly, the adequate share of responsibility 26. See Christiano [2008]. 27. Gaus [1996: 252]. A weaker version of this claim would only require that there be an undefeated, even if not conclusive, justification for the inequalities at hand. It is beyond the scope of this paper to discuss whether we should adopt Gaus’s original version or the weaker form just suggested; both are compatible with the wider egalitarian framework I use here. Agency, Authority, and Responsibility 17 must allow persons to take credit for their successes, as well as to acknowledge their mistakes, without it leading either to humiliating others or to humiliating oneself. The task of defining the adequate social distribution of responsibility is a difficult one, but, if my argument is right, it is more promising that the luck egalitarian perspective. One of its objectives is to avoid the dilemma I have indicated luck egalitarians face, namely, that either they cannot avoid the harshness objection, or they must run afoul of their defense of individual responsibility. At this point it remains plausible that a mixed view, taking relational equality as a basic principle (one that cannot be trumped) and luck egalitarianism as a pro tanto principle, be developed. 28 But from this is cannot be inferred that persons are not entitled to the relevant form of background security. Relational equality motivates both respect for judgment, or for authority over one’s own life, and security to an extent compatible with the former. Were the status of persons as equals not adequately protected, their authority over themselves would not be adequately secured either. In this respect, authority over one’s life in the egalitarian sense does not parallel unhindered freedom in the libertarian sense. It is important to notice the distinction. Libertarians do require that persons have authority over their own lives, but for this to hold it is enough that no one be prevented to act as she sees fit (as long as she respects others’ libertarian rights). Yet if one acts in such a way as to fare much worse than others, libertarianism does not guarantee her any further protection. She is left with whatever she can earn, and remains free to do whatever helps her to survive, even if this involves humiliating herself in some way, or to become dependent on others.29 By contrast, relational egalitarianism is more pluralist, since it accounts not only for liberty, but also for security. Precisely what kind of security this recommends is open to controversy and debate, and many different schemes could be proposed. Relational egalitarianism does not define any such scheme; all it does is to set a standard such a scheme ought to satisfy. It is perfectly fit that how the standard is to be satisfied be left to democratic decision; and indeed this is more adequate. For defining a precise scheme a society would be forced to endorse would not fit with common principles of democracy. Let me briefly clarify this point. First of all, defining a precise scheme for satisfying a common standard of security exposes us to the objection that that scheme is not, by any means, the only possible one, and thus that, in doing so, we are making an arbitrary choice. Second, avoiding arbitrariness by providing relevant supporting reasons would almost inevitably imply advancing even more controversial motivations. Third, this is, I believe, incompatible both with egalitarianism and with democracy, for it would expose us to the threat of authoritarianism, which is contradictory both with equality and with democracy. Indeed, authoritarian rulers cannot but claim a superior judgment without adequately motivating it (if this is even possible), which makes authoritarianism incompatible with equality understood as equal standing (in the 28. See, for example, Fleurbaey [2008]. Sen and Nussbaum’s capability approach might, to some extent, be assimilated to such a mixed view, in that it is an attempt to combine an opportunity (capability) egalitarian view of distributive justice with a fundamental entitlements conception of human dignity. See inter alia Nussbaum [2000, 2002, 2003, 2006, 2011a, 2011b] and Sen [1982, 1990, 1992, 2009]. Several commentators have seen Sen’s approach as close to luck egalitarianism, for it seemingly relies on a similar conception of responsibility, but whether this needs to be the case is disputable. 29. See Van Parijs [1995: 14]. Agency, Authority, and Responsibility 18 political realm). It is also incompatible with democracy understood as the rule of a democratic assembly, rather than that of a non-elected ruler who rejects democratic discussion. It remains likely that how background security is implemented in some given society will not fit with our own views, but, as democratic citizens, and insofar as the selected scheme was democratically established, it is still our duty to comply to this scheme, even though we remain free to criticize it publicly and to advocate a modification of this scheme. As long as the scheme does not run afoul of the most basic and commonly (reasonably) accepted principles of justice, there is no reason to believe that only one of many possible schemes would be acceptable, 30 even more so given the diversity of situations different societies face around the world, as well as the variety of ways in which the values at stake could weigh into the balance. The challenge we face seems pretty clear. Conflicting values are at stake, all of which appear to be important enough not to be discarded. Too great an emphasis on personal responsibility threatens to thwart agency, while, as I have shown previously, security may diminish the importance of responsibility to an unacceptable extent. It is more difficult to say to what extent equality conflicts with those values (if it does), since, if my argument is right, it is supportive of both. A conflict may still arise, for egalitarianism puts limits on the extent to which they may be realized, and is not to be trumped (unless, as I said earlier, extremely weighty reasons, if there are any, are presented) by other values. But should such a conflict arise, it is worth noting that it is not owed to an already present incompatibility between the values at stake. The idealnonideal distinction notwithstanding, we can observe, quite simply, that the most common ground of such a conflict is human behavior within given institutions. Greed, corruption, lie, and other misdeeds are all too common, however the institutions are arranged. Egalitarianism recommends that we avoid distrusting persons on no relevant basis, for distrust of this kind is likely to be demeaning to many; and yet it is true that some persons turn out not to be trustworthy. Could this be a rationale for not trusting people in the first place? I believe the answer is a qualified “No.” Distrust is not in itself inegalitarian. It becomes so when it leads to pervasive suspicion on the part of institutions, with the underlying indication that ordinary citizens are not to be trusted. In extreme cases, this could entail systematic and intrusive inquiries into persons’ lives, in order to determine whether they are trustworthy, or even whether they are morally deserving of the security they enjoy. 31 Rather, basic security should not be conditional if it is to remain egalitarian. It could be objected to my argument that, in many cases, persons will enjoy the benefits of security, despite being clearly responsible for their fate, and so the view I defend does not, in fact, track responsibility in an adequate manner. This is true, but let me explain why this objection is much less powerful than it might seem. Firstly, it does 30. In other words, namely those of Gaus [1996]: it is reasonable to believe that (a) no such scheme can be conclusively justified, and (b) several of them are both inconclusively justified and undefeated (that is, each is justified as against no scheme at all). It could be asked: Is there really no conclusively justified scheme for implementing security? I believe there is not. For any justificatory attempt faces the “burdens of judgment [Rawls 1996: 54–58],” and thus involves controversial weighting of otherwise reasonably acceptable values. Under these conditions, the non-democratic imposition of a given scheme would be authoritarian. 31. When this hypothesis holds, pervasive distrust is in fact one manifestation of moralism; see Scheffler [2005: 14–16]. Agency, Authority, and Responsibility 19 not follow from one’s being morally responsible for an outcome that she should be held substantively responsible for that outcome. Secondly, it is not that important to track individual responsibility so closely, as a matter of social justice. It is not the state’s role to determine accurately who is responsible for what: this is a Herculean task, and one that is likely to involve unjust practices. Furthermore, anyone facing this task also faces many cases of indeterminacy. It remains possible, privately, to claim that one was in fact responsible for this or that, but from this one cannot infer that the state ought to act similarly. In addition, it is worth noting that, although a conflict may arise between security and responsibility, the former is also, in an important respect, supportive of the latter. Indeed, a lack of security is likely to encourage many persons to shift their responsibility onto others, so as to avoid the costs they would incur, should they be held responsible for the action or outcome at hand. Here we encounter a variant of the second horn of the dilemma noted above, which luck egalitarians face unless they adopt a mixed view: if persons are willing to avoid the costs of, or the loss of wellbeing induced by substantive responsibility, it is quite likely that many of them will attempt to put it on others’ shoulders, instead of accepting their own flaws or acknowledging their own mistakes in a non-moralistic way. Security means that persons can face their weaknesses and acknowledge their mistakes without it leading to destitution, self-humiliation, or other kinds of hardships that harms one’s sense of her own worth. 7 Concluding remarks If my argument thus far is right, a fair distribution of responsibilities should meet the following criteria: (a) It treats persons as equals; (b) It starts from a presumption of authority over oneself; (c) It secures the sense of one’s own worth; (d) It is democratically agreed upon. All these criteria are interrelated. Treating persons as equals entails that no one has a presumptive authority over any other, and requires that persons securely enjoy the sense of their own worth, and of their life plans’ worth. The presumption of authority over oneself rules out political authoritarianism, and thus recommends that, however responsibilities are distributed, and as long as this distribution meets the other criteria, it be the outcome of an open, democratic process. We must ask, of any proposed distribution of responsibilities: “Does it meet these criteria?” Luck egalitarianism, in its most “brute” version, does not meet (a) and (c), because it starts from the assumption that a fair distribution of responsibilities is one that tracks individual responsibility, and equalizes or compensates brute luck or circumstances, should this lead to treating people as inferiors or to degrading their sense of their own worth. It could, however, meet (b), for parallel reasons: one can be responsible for something only insofar as she possesses a presumptive authority over this thing. Finally, it is difficult to evaluate whether luck egalitarianism could be democratically agreed upon. On the one hand, it is supported by reasonable arguments and so a reasonable democratic assembly could Agency, Authority, and Responsibility 20 accept it as a principle of distributive justice. On the other hand, given that it does not adequately treat persons as equals, it does not appear to be a principle that an egalitarian democracy could rightly accept. Now, it remains plausible that a deeply revised proposal, that meets the above criteria and embodies some important aspects of the luck-choice principle, be advanced. In this sense, if my argument is right, its conclusion is not that luck egalitarianism ought to be rejected in its entirety, but that luck egalitarians ought to adopt a pluralist view from the start, rather than make ex post concessions. Given these observations, we can envision several ways in which to distribute responsibilities. All these are, to a great extent, conventional and rely on democratic agreement, but that does not mean that they are arbitrary or that any distribution of responsibilities is acceptable (as long as it is democratically agreed upon). So the fear that the social attribution view be exposed to arbitrariness must not be overstated. My argument does not entail that the social attribution view could, at any rate, be used alone: rather, if it is right, it sets principles constraints on what makes a conventional distribution of responsibilities a just one. Importantly, this means that we should not start from any given hypothesis about how responsibilities ought to be distributed, as luck egalitarians recommend. This is not limited to substantive responsibility: other kinds of responsibilities can be socially distributed as well. A further examination of this problem is not possible here, for it would require much more space than I can use here, but it is to be hoped that the framework proposed here could serve such a purpose. This, of course, remains an open-ended framework, and its application to further problems requires a careful analysis. I have said little of how it could fit into a broader conception of justice, and nothing of how it could fit into a conception of global justice. I believe it could be so used, but here is not the place to engage in such an analysis. This is left for later essays. Bibliography ANDERSON Elizabeth. 1999: What Is the Point of Equality?, Ethics 109(2): 287–337. —. 2008: How Should Egalitarians Cope with Market Risks?, Theoretical Inquiries in Law 9(1): 239–270. ARNESON Richard J. 1989: Equality and Equal Opportunity for Welfare, Philosophical Studies 56(1): 77–93. —. 1990: Liberalism, Distributive Subjectivism, and Equal Opportunity for Welfare, Philosophy & Public Affairs 19(2): 158–194. —. 2000: Luck Egalitarianism and Prioritarianism, Ethics 110(2): 339–349. BARRY Nicholas. 2006: Defending Luck Egalitarianism, Journal of Applied Philosophy 23(1): 89–107. CHRISTIANO Thomas. 2008: The Constitution of Equality, Oxford: Oxford University Press. COHEN Gerald Allan. 1989: On the Currency of Egalitarian Justice, Ethics 99(4): 906– 944. —. 1997: Where the Action Is: On the Site of Distributive Justice, Philosophy & Agency, Authority, and Responsibility 21 Public Affairs 26(1): 3–30. —. 2008: Rescuing Justice and Equality, Cambridge (MA): Harvard University Press. DWORKIN Ronald. 1981a: What Is Equality? Part 1: Equality of Welfare, Philosophy & Public Affairs 10(3): 185–246. —. 1981b: What Is Equality? Part 2: Equality of Resources, Philosophy & Public Affairs 10(4): 283–345. —. 2011: Justice for Hedgedogs, Cambridge (MA): Harvard University Press. FLEURBAEY Marc. 1995: Equal Opportunity or Equal Social Outcome?, Economics and Philosophy 11(1): 25–55. —. 2008: Fairness, Responsibility, and Welfare, Oxford: Oxford University Press. GAUS Gerald. 1996: Justificatory Liberalism, Oxford & New York: Oxford University Press. HURLEY Susan L. 2003: Justice, Luck, and Knowledge, Cambridge (MA): Harvard University Press. MCTERNAN Emily. Forthcoming: How to be a Responsibility-Sensitive Egalitarian: From Metaphysics to Social Practice, Political Studies MILLER David. 2001: Distributing Responsibilities, The Journal of Political Philosophy 9(4): 453–471. NUSSBAUM Martha C. 2000: Women and Human Development, Cambridge: Cambridge University Press. —. 2002: Capabilities and Social Justice, International Studies Review 4(2): 123–135. —. 2003: Capabilities as Fundamental Entitlements: Sen and Social Justice, Feminist Economics 9(2-3): 33–59. —. 2006: Frontiers of Justice, Cambridge (MA): Harvard University Press. —. 2011a: Capabilities, Entitlements, Rights: Supplementation and Critique, Journal of Human Development 12(1): 23–37. —. 2011b: Creating Capabilities, Cambridge (MA): Harvard University Press. O’NEILL Martin. 2008: What Should Egalitarians Believe?, Philosophy & Public Affairs 36(2): 119–156. RAWLS John. 1996: Political Liberalism, New York (NY): Columbia University Press. —. 1999: A Theory of Justice, Cambridge (MA): Harvard University Press. —. 2001: Justice as Fairness: A Restatement, Cambridge (MA): Harvard University Press. ROEMER John E. 1993: A Pragmatic Theory of Responsibility for the Egalitarian Planner, Philosophy & Public Affairs 22(2): 146–166. SANYAL Sagar. 2012: A Defense of Democratic Egalitarianism, The Journal of Philosophy 109(7): 413–434. SCANLON Thomas. 1998: What We Owe to Each Other, Cambridge (MA): Harvard University Press. —. 2003: The diversity of objections to inequality, in The Difficulty of Tolerance, Cambridge: Cambridge University Press. SCHEFFLER Samuel. 2003: What Is Egalitarianism?, Philosophy & Public Affairs 31(1): 5–39. —. 2005: Choice, circumstance, and the value of equality, Politics, Philosophy & Economics 4(1): 5–28. SCHEMMEL Christian. 2012: Luck Egalitarianism as Democratic Reciprocity? A Response to Tan, Journal of Philosophy 109(7): 435–448. Agency, Authority, and Responsibility 22 SEN Amartya. 1982: Equality of What?, in Choice, Welfare and Measurement, Cambridge (MA): Harvard University Press. —. 1990: Justice: Means versus Freedoms, Philosophy & Public Affairs 19(2): 111– 121. —. 1992: Inequality Reexamined, Oxford & New York: Oxford University Press. —. 2009: The Idea of Justice, Cambridge (MA): Harvard University Press. SPITZ Jean-Fabien. 2008: Abolir le hasard?, Paris: Vrin. TAN Kok-Chor. 2008: A Defense of Luck Egalitarianism, The Journal of Philosophy 105(11): 665–690. VAN PARIJS Philippe. 1995: Real Freedom for All, Oxford: Oxford University Press. VOIGT Kristin. 2007: The Harshness Objection: Is Luck Egalitarianism Too Harsh on the Victims of Option Luck?, Ethical Theory and Moral Practice 10(4): 389–407. WOLFF Jonathan. 1998: Fairness, Respect, and the Egalitarian Ethos, Philosophy & Public Affairs 27(2): 97–122. —. 2010: Fairness, Respect, and the Egalitarian “Ethos” Revisited, The Journal of Ethics 14(3/4): 335–350.