Rawls and Racial Justice

Politics, Philosophy & Economics (2017), Vol. 16, No. 3. Rawls and Racial Justice D. C. Matthew York University, Canada Abstract This paper discusses the adequacy Rawls’ theory of justice as a tool for racial justice. It is argued that critics like Charles W Mills fail to appreciate the both the insights and limits of the Rawlsian framework. The paper has two main parts spread out over several different sections. The first is concerned with whether the Rawlsian framework suffices to prevent racial injustice. It is argued that there are reasons to doubt whether it does. The second part is concerned with whether a Rawlsian framework has the resources to rectify past racial injustice. It is argued that it has more resources to do this than Mills allows. This second part of the paper centers on two Rawlsian ideas: ideal theory and the fair equality of opportunity (FEO) principle. It is argued that ideal theory is essential for the kind of rectificatory work that Mills wants nonideal theory to do, and that where there is a socioeconomic legacy of past injustice, it is hard to see how FEO could be implemented if it did no rectificatory work, a result which means that there is less need to turn to nonideal theory at all. Keywords: Rawls, racial justice, Charles W Mills, rectificatory justice, fair equality of opportunity. Introduction In recent years, the potential of Rawls’ work to address issues of racial justice has become the object of growing attention. Besides the work cited below, other work substantially about Rawls and race include Charles (2004), McCarty (2004), Shelby (2007), Goodman (2007), Carcieri (2010), Graham (2010; esp. Ch. 3), Dombrowski (2011; Ch. 4), and Lebron, (2013; Ch. 1). Given the continuing prominence of this work, this is not surprising. Indeed, given the explosiveness of race, what is perhaps more surprising is why it took so long. In any case, that it is now, belatedly, receiving this attention is due in no small part to the efforts of Charles Mills. In a series of papers and parts of a book, Mills has repeatedly criticized Rawls for his lack of attention to race and argued that a Rawlsian framework is of limited use in this area, at least without radical revision. See, for example, Mills (2005), Pateman and Mills (2007), Mills (2008), Mills (2009), Mills (2013), and Mills (2015). For Mills, it is most telling that the person widely regarded as the most important political philosopher of the 20th century produced a “huge body of work on questions of social justice… that has nothing to say about racial injustice, the distinctive injustice of the modern” (2009: 162). As a result, any discussion of Rawls and race must perforce have much to say about Mills and his work. As for Rawls, even if it is true that he had little or nothing to say about racial justice, it does not follow that there was nothing that he could have said. As Rawls himself noted, though his work omits discussion of issues of race, “an omission is not as such a fault. Whether fault there be depends on how well that conception articulates the political values necessary to deal with” issues of racial justice (2001: 66). Here Rawls is actually speaking about both racial and sexual justice. Accordingly, then, in examining the adequacy of justice as fairness as a tool for racial justice, it is necessary to consider the possible theoretical elaborations that could be made, in addition to Rawls’ explicit theorizing. This is what this paper intends to do. I will argue that critics like Mills fail to appreciate the both the insights and limits of the Rawlsian framework. The discussion that follows can be thought of as having two main parts. The first is concerned with the question of whether a Rawlsian framework suffices to prevent racial injustice. It is argued that there are reasons to doubt whether it does. The second, longer part is concerned with whether a Rawlsian framework has the resources to rectify past racial injustice. It is argued that it has more resources to do this than Mills allows. The paper proceeds as follows. In the first section, after explaining the two parts of racial justice theory, I lay out the Rawlsian position on racial justice, and criticize Mills’ argument that justice as fairness suffices to prevent racial injustice. The second section discusses Mills’ critique of Rawlsian ideal theory. I defend, expand on and add to others’ arguments that we can’t condemn something as racially unjust unless we have on hand ideal principles of racial justice. The following section, the third section, briefly summarizes both Tommie Shelby’s fair equality of opportunity (FEO) argument in support of Rawls and Mills’ critique of this argument and then presents my own argument that though Mills may be right that Rawls did not intend for FEO to be used for rectificatory purposes, where there is a socioeconomic legacy of past injustice, it is hard to see how this principle could be implemented if it did no rectificatory work. I therefore argue that while ideal theory is essential for nonideal theory to do the rectificatory work that Mills wants it to do, implementing the reforms called for by Rawls’ ideal theory would perform some of the work that it seems any plausible rectificatory theory would require, and consequently there is less need to turn to nonideal theory at all. The fourth section briefly discusses Shelby’s recent claim that he never proposed that FEO be used for rectificatory purposes. Rawls on Distributive Racial Justice Racial justice, as branch of social justice, is concerned with what conditions must obtain for there to be no arbitrary distinctions between racial groups in the assignment of basic rights and duties, and for each racial group to have its due claim to the advantages of social life. Cf. Rawls (1999: 5). Like Rawls, I take it that while people may differ in their conceptions of racial justice, they can share the same concept of ‘racial justice.’ The theory of racial justice aims to identify the principles or requirements of justice that are needed to secure this ideal. It has two component parts: the theory of distributive racial justice and the theory of rectificatory racial justice. This distinction obviously draws on the traditional distinction between distributive and rectificatory justice, a distinction that was first discussed by Aristotle; see Aristotle (1973: Bk V). Distributive racial justice is concerned with those measures that aim to prevent Cf. Mills’ talk of “pre-emptive” measures; see, e.g., Mills (2009: 162). Elsewhere, however, he has wrongly tried to downplay the preventive component of racial justice, writing that while “one could choose to subsume anti-racial-discrimination measures under the category of racial justice,” racial justice “is preeminently a matter of rectificatory justice” (2013: 11). I discuss this issue below. the introduction or continuation of injustice, where the ‘continuation’ of injustice is understood in terms of a failure to end ongoing injustice and not in terms of the non-rectification of past injustice. In other words, it does not rectify injustice created by past acts (including omissions); it prevents the creation or continuation of injustice made possible by future acts (including omissions). Rectificatory justice, by contrast, is concerned with those principles needed to repair the injustice created by past acts (including omissions). The rectificatory measures called for by these principles might include, at the very least, compensation for harms created by the injustice or (if possible) restoration to the pre-injustice situation, formal apology and punishment. Here I follow Roberts (2002). Note, however, that it seems to me that whatever package of these measures called for by a theory of rectificatory justice is what “rights the wrong.” Contra Roberts, while an apology is typically part of what is required to right a wrong, it doesn’t right it all by itself. In terms of the categories that will play a major role in the discussion that follows, distributive racial justice falls under ideal theory, whereas rectificatory racial justice falls under nonideal theory. As is well-known, Rawls did not attempt to show that his theory of justice would specifically address rectificatory racial injustice. What is less-well known – or at least less well understood as such – is that his work was not thereby silent on matters of race. For though race is scarcely mentioned in Theory, there is a sense in which it is addressed, namely, by claiming that the principles of justice that would be selected would be free of racial distinctions. Cf. Swift (2008: 249). Theory, in other words, addresses distributive – but not rectificatory – racial justice. Rawls’ central argument is well-known. The most reasonable principles of justice are those that would be selected in an initial choice situation (called, in his version, the “original position”) that is fair. As far as distributive racial justice is concerned, the theory’s key feature has to do with the design of this initial choice situation: the fact that in the original position the parties are ignorant of their race (Rawls, 1999: 11, 118). Race also figures in the design of the original position through the process leading to “reflective equilibrium.” The conviction that racial discrimination is unjust forms a “provisional fixed point” that principles yielded by a specification of the initial situation will likely have to accommodate. See Rawls (1999: 17-18). Since their racial identities are behind the “veil of ignorance,” it would be irrational for them to select racially discriminatory principles; once the veil is lifted, they may not be a member of the favored group. By contrast, Rawls suggests that the principles that they would select – a lexically prior liberty principle, and a two-part second principle made up of FEO and the difference principle – would make no racial distinctions. Given this, Rawls thought that they would suffice to prevent racial injustice, suggesting that in a well-ordered society in favorable conditions racial (and sexual) injustice would not exist (2001: 65-66). Minorities (and women) might justifiedly have less basic rights and opportunities only if (as the general conception of justice requires) this were to their advantage and acceptable from their point of view, but this, he argues, is not now the case, nor does it seem that it ever was. Thus he holds that if there is racial injustice, it will be because there is only partial compliance, something which raises issues of non-ideal theory, about which he has relatively little to say. Interestingly, Mills agrees with Rawls that the two principles suffice to prevent the introduction of racial injustice. If we are starting from what he elsewhere calls a “moral ground zero” (Pateman and Mills, 2007: 113) of no history of injustice, he writes, “all we need is appropriate antidiscrimination legislation to make sure that… injustice does not enter the basic structure” (Mills, 2009: 179) Mills goes so far as to suggest that such legislation would ensure not only “a racism-free polity,” but “a race-free polity” (2009: 179). After all, he argues, race is socially constructed, and so without “systemic discrimination” race wouldn’t exist as a social entity. But even if it is true that race wouldn’t exist but for systemic discrimination, a society with a discrimination-free basic structure might still contain such discrimination. See my discussion below. – does not enter, that is, “the way major social institutions distribute fundamental rights and opportunities and determine the division of advantages from social cooperation,” which is what Rawls says is the primary subject of justice (1999:6). Though he does indicate what he has in mind, he is probably thinking of antidiscrimination legislation such as laws banning discrimination in employment and housing. But even if this is true – even if all we need is appropriate antidiscrimination legislation that is to ensure that injustice does not enter the basic structure – questions can be raised about whether justice as fairness authorizes, or can authorize, such legislation. For example, in an important paper, Seana Shiffrin (2004) has argued that the two principles of justice need to be supplemented by an antidiscrimination principle that forbids discrimination in housing, the provision of public amenities, economic markets and the like. A “more creative reading of the two principles,” she suggests, cannot entirely dispel “the impression of the absence of an anti-discrimination principle” (2004: 1654). But I believe that Tommie Shelby has convincingly cast doubt on whether anti-discrimination norms should be enshrined as fundamental principles of justice. In Theory Rawls suggested that we should understand the application of justice as fairness as proceeding through a four-stage sequence in which the parties are progressively allowed more information as they address various questions of justice: the original position, a constitutional convention (where a constitution is chosen by delegates now allowed knowledge of the general facts of their society), a legislative stage (where the justice of laws and policies are assessed from the perspective of a representative legislator allowed knowledge of “the full range of general economic and social facts” (1999: 175) but who doesn’t know particulars about him or herself), and a final stage (in which all knowledge restrictions are lifted) where judges and administrators apply the rules to particular cases. Shelby suggests that it is at these later stages that the persons will have access to general facts relevant to racial justice (such as whether some racial groups in the society are or have been disadvantaged because of racism), and can make appropriate provision, if necessary, for their address (2004: 1706-08). It is here, he argues, that such special measures are appropriate, for our most fundamental principles of justice should be “sufficiently general in form and universal in application” (2004:1708) and thus should not include principles (like antidiscrimination principles) that are responsive to some but not all contexts. But even if there are reasons to be skeptical of Shiffrin’s argument, I believe that there are other, more compelling reasons to doubt whether justice as fairness can authorize appropriate measures to prevent the introduction of racial injustice. Note first that Mills’ antidiscrimination legislation would not prohibit all forms of discrimination; in particular, private racial discrimination might still exist: discrimination not by businesses, but by individuals in the choices they make about where to live, whom to befriend and how to treat people in informal contexts. Accounts of the harm of such discrimination more typically make it depend on past racial injustice (where the effects of this have not been adequately addressed), but I believe that over time it can inflict significant disadvantage even in the absence of past injustice (or – what amounts to the same thing – even if where past injustice has been fully addressed). I argue for this point in great detail in work in progress. For an example of an account of the harm of private discrimination that makes it depend on past injustice, see esp. Loury (2002). See also Anderson (2010). Such choices would be protected by the liberty principle, and yet if widespread would arguably introduce injustice, especially over time. Private discrimination can be thought to introduce injustice by violating either FEO or the liberty principle. I favor an approach that emphasizes FEO. For an argument of how it (or what seems very much like it) violates the liberty principle, see Richard Schmitt’s contribution to Nagel, et al (2003: 15-24). Furthermore, since the two principles are the object of citizens’ sense of justice, it is not necessarily the case that persons who have realized this sense will not discriminate in these ways. A somewhat different argument holds that “the members of a well-ordered society not only accept the two principles, but… also the grounds for them. If they accept the justificatory apparatus, then they accept that citizens are equal from a moral point of view and that features like race are morally arbitrary. If they accept this view, they would not treat race as relevant when it is irrelevant and would not engage in racial discrimination” (Shiffrin, 2004: 1656; see also her (2010: esp. 126-8); in a recent review essay on Anderson’s The Imperative of Integration, Shelby seems to express a similar view; see his (2014: 255n.7). But even if this argument works, it is not clear that it could be extended to private racial discrimination. Rather, the members of a well-ordered society would be constrained not to treat race as relevant when it is irrelevant in institutional contexts relevant to the principles, not in private life. A more promising – though, in my view, also ultimately unsuccessful – way of defending Rawls draws on his duties of mutual aid and respect; see Voorhoeve (2005: 5). At one point Mills writes that “In a well-ordered society, one that is regulated by Rawls’s two principles, structural racial subordination will not exist, nor the legacy of such subordination” (2013: 11). Note that this sounds more or less like what he attacks Shelby for saying (see below)! Either he is saying what Shelby is saying (that a well-ordered society will reduce if not eliminate disadvantages that are the result of past injustice), or he understands ‘well-orderedness’ in a way that differs from what Rawls intended. In my view, the latter explanation is more plausible. For discussion, see the next note. If by “structural racial subordination,” he means subordination enacted through the basic structure, then he is partly right. Only “partly” because of his claim that in a well-ordered society the legacy of racial subordination will not exist. The problem is that this is most plausibly understood as construing a well-ordered society as a society with no injustice, a not-uncommon misunderstanding of the concept (for example, see Espindola and Vaca (2014: 233): “by definition, a well-ordered society is a society in which no injustice occurs”). But a well-ordered society is just a society in which the principles of justice are accepted and effectively implemented. Whether they achieve justice is another matter. If this were not so, Rawls would be assuming that the two principles suffice for justice when in parts of Theory he assumes well-orderedness. But racial injustice might arise through acts that fall outside of the basic structure. Accordingly, a society with a discrimination-free basic structure might not suffice to prevent the introduction of racial justice. Now as just noted, Rawls argued that the two principles are to be limited to the basic structure. This is significant because in response to G. A. Cohen’s (2008) incentives critique (to the effect that in a just Rawlsian society talented people would not require incentives to exploit their talents, and cannot justifiably decide to work less hard in their absence) some critics raised what Cohen calls “the basic structure objection,” to wit, that his critique “misapplies principles designed for a structure to individual choices and actions” (125). And in the same way that the basic structure objection can be raised to Cohen’s incentives critique, so it can be argued that to apply principles of justice to private discriminatory choices is to similarly misapply principles designed for a structure to individual choices and actions. I raise this point not to discuss it here, but simply to note one possible response to the above argument. In fact, I believe I have convincingly shown elsewhere that there is no good reason for restricting the purview of justice in the way suggested by the basic structure objection (Matthew, 2014). What’s more, I have now come to believe that that objection probably does not have a basis in Rawls’ work. Scheffler has argued for the same conclusion, but the considerations that he adduces are different from what I have in mind. See, e.g., Scheffler (2006). In any case, the main point I want to emphasize here is that Mills’ confidence that the two principles suffice to prevent racial injustice is questionable. And this in turn suggests that even if racial justice is, as he says, preeminently a matter of nonideal theory, there remain unresolved questions of race that ideal theory must address. It also suggests that those, such as Valentini, who have gone further and claimed that “questions of racial and sexual discrimination fall outside of ideal theories,” are also mistaken (2009: 343). See also Boettcher (2009: 238). So far I have only considered one of these questions, namely, the question of what is required to prevent the introduction of racial injustice. This, as I have said, constitutes one part of distributive racial justice. Another part has to do with the discontinuation of injustice. Here the main question is: what measures are necessary to secure the discontinuation, or ending, of injustice, and does justice as fairness have the theoretical resources to provide it? This question, to be sure, might seem to be an inappropriate one to raise given some of what are often taken to be Rawls’ assumptions. It is frequently said that in focusing on ideal theory, Rawls developed a theory of justice for societies in which there is no history of (racial or other) injustice. See, e.g., Mills (2009: 179), and Waligore (2016: 48, 52). It is with this kind of view in mind that it is possible to write that Rawlsian ideal theory “does not tell you what to do if, as is almost always the case, you find yourself in an unjust society, and want to correct that injustice” (Nagel, 2003: 82). Nagel, however, does not explicitly say Rawls developed a theory of justice for societies with no history of injustice. But this view is plainly incorrect. Rawls’ whole point in developing his theory was to find a theory that societies could use to judge and, if necessary, reform their basic institutions. In fact, he took it for granted that few if any societies could rightly be said to be in conformity with his two principles. This means that as judged by justice as fairness, most societies would have to be considered unjust, or less than perfectly just. But if so, then in implementing the two principles, and thereby reforming unjust institutions, societies would be putting injustices to an end. The second part of the theory of distributive racial justice is concerned with what is required to end racial injustice. Does it make sense to distinguish between preventing injustice and ending injustice? I believe that it does; it is possible that a theory that can successfully prevent an injustice may fail to end the same injustice, once it exists. (Whether the reverse – whether a theory that can successfully end an injustice may fail to prevent the same – is also true is more questionable.) To see this, consider Shiffrin’s recently-mentioned argument that the members of a well-ordered society will not only accept the two principles, but also the grounds for them, and consequently would not engage in racial discrimination (2004: 1656). I have already expressed doubts about whether this works for private discrimination. But if it does work, it may work only preventively: it may be that if we start from Mills’ “moral ground zero” of no past injustice, implementation of the two principles will have this effect, but that if we start from a society that already has private racial discrimination, their implementation will not have this effect. Now admittedly, for those especially concerned with the rectification of injustice, there may be good reasons for some unease about what I am here claiming is a part of the theory of racial justice. The problem is that, in some cases at least, it is clearly inadequate to simply reform unjust institutions. To see this, consider the following. In many – if not all – cases, when an injustice is perpetrated it inflicts harm on the people involved, harm that may require restoration or compensation in addition to formal apology and, possibly, punishment, for its full remedy. To be sure, it seems to me that what should ideally be fully remedied is not just the harm, but the wrong, of which the harm is merely a part. In focusing more narrowly on the harm here I do not mean to suggest that this should be the exclusive focus of the rectification. But the nature of the harm inflicted might be of roughly two types, which it is useful to distinguish here. First, the harm might have a more evanescent character. It is inflicted by the injustice, but it does not long persist once the unjust acting ends. Alternatively, the harm might have a more enduring character. It is inflicted, and there is a lasting effect even after the injustice is no longer being actively inflicted. It should be clear that injustices that inflict harms of the second type raise different – and more difficult – issues than the injustices that inflict harms of the first type. Ending unjust acting in cases of harms of the first type does more to satisfy the requirements of justice – brings us closer to its realization – than ending unjust acting in cases of harms of the second type. To see this, take the right to marry, or even the right to vote. If these rights are denied on the basis of race, then there would seem to be racial injustice. But it is far from a foregone conclusion that there will be an enduring legacy of injustice if institutional reforms extend these rights to those previously denied them. Now compare that to a case where racial injustice produces a structure of racial inequality. In such a case reforming unjust institutions might leave unaddressed this lingering effect of the injustice. Accordingly, these reforms may not bring us as close to the ideal of justice as in the first case – in particular, they may fall shorter of the recificatory ideal (to whatever extent it applies) than the reforms of the first case. What this means is that, even if we set aside formal apologies and punishments, as I will here, mere reform of unjust institutions may still fall short of the rectificatory ideal when the relevant injustice produces more enduring harms. There is no attempt, therefore, to deny that in some cases it is clearly inadequate to simply reform unjust institutions. The claim is simply that all instances of past racial injustice are not necessarily like this. In some cases it may be perfectly acceptable from the standpoint of justice to merely end an injustice; in others the need for rectificatory measures may be partially pre-empted. The rest of the paper is concerned with the question of rectificatory justice, so I will not here be concerned with whether the two principles suffice to end racial injustice. I do want, however, to reiterate that the theory of racial justice is concerned with three kinds of measures – those having to do with preventing racial injustice (preventive measures), those having to do with ending it (reformative measures) and the last having to rectifying it (rectificatory measures) – and that the first two fall under distributive racial justice, whereas the third falls under rectificatory racial justice. Below I will argue that, as it has unfolded, the dispute over Rawls and race can be understood as a dispute not only about whether justice as fairness has the theoretical resources to justify the rectificatory measures needed to secure racial justice for racial groups that have suffered injustice, but whether it need have recourse to a rectificatory theory at all (and if so, to what extent). The Uses of Rawlsian Ideal Theory As I have indicated, Rawls’ most persistent critic on issues of race has been Mills. For him, it is very telling that the mainstream world of white political philosophy – of which Rawlsians occupy a central place – has had so little to say about the racial injustice. See esp. Mills (2008: 1382-86). But if for him demographic characteristics have inclined Rawlsians to avoid theorizing about race, the traditional Rawlsian focus on ideal theory has been a crucial factor in making this avoidance possible. As he describes (and ridicules) what he takes to be the thinking, “In a perfectly just society, race would not exist, so we do not (as white philosophers working in ideal theory) have to concern ourselves with matters of racial justice in our own society, where it does exist” (2008: 1385). The problem, as he sees it, is that racial justice is mostly a matter of rectificatory justice – of what is needed to rectify past racial injustice – but rectificatory justice falls under nonideal theory, which has received comparatively little attention. Mills’ work, then, can be understood as a clarion call for political philosophers in general, and Rawlsians in particular, to turn to the work of theorizing what rectificatory racial justice requires. However, he argues that once we to turn to this rectificatory work, we will find Rawls’ work, at least if not appropriately reconstructed, to be of little use. Rawlsian ideal theory, he argues, is useless for rectificatory purposes. The remainder of this paper is concerned with this issue. In this and the following section, I describe and critique Mills’ argument. Mills also argues that Rawlsian ideal theory is possibly counterproductive in that it lends itself to “retrograde political agendas” such as all-encompassing defenses of colorblindness (2009: 180). Though I will focus on the useless charge here, it is not clear why Rawlsians can’t say that an undiscriminating use of colorblindness is a misuse of the theory. Such views, they can argue, ignore a basic insight of Rawlsianism: the need to distinguish between ideal and nonideal conditions. Before explaining this argument, however, something should be said about Rawls’ distinction between ideal and nonideal theory. Theory, as is well-known, is primarily a work of ideal theory. To be sure, I believe that it is not very widely recognized just how much it is not a work of ideal theory. After all, aside from the discussion of civil disobedience and conscientious refusal in Ch. 6, there are other places in the book where it is clear that the assumption of well-orderedness cannot fully be in force. For example, it is clearly not fully in force in Ch. 8, when the question is whether citizens who grow up in a well-ordered society will come to develop a sufficiently strong sense of justice. (This, however, doesn’t make this chapter a work of nonideal theory.) That is, it is a work that mainly assumes strict compliance – that everyone complies with the principles of justice – in the favorable conditions that make constitutional democracy possible (1999: 7-8, 32, 215-17). Here I pass over some tricky issues having to do with Rawls’ claim that what he calls “natural limitations” constitute unfavorable conditions. For discussion, see Simmons (2010: 12-17). Note that, contra what Valentini claims, the idealizations involved in the original position do not have anything to do with what makes Rawls’ ideal theory ideal. See Valentini (2009: 338, 353). This error also mars Kang (2016). Rawls seems to focus on ideal theory for two reasons. First, and most importantly, he believes that we can’t properly evaluate a theory of justice unless it is fully implemented. As he writes, “The evaluation of principles must proceed in terms of the general consequences of their public recognition and universal application, it being assumed that they will be complied with by everyone” (1999: 119). Cf. Simmons (2010: 8-9). Second, he believes that ideal theory is a necessary precursor to nonideal theory; it provides “the only basis for the systemic grasp” of its “pressing problems” (1999: 8). By contrast, if we take the sorts of situations in which we might need nonideal theory to correspond to the two conditions of ideal theory (full compliance and favorable conditions), then there are two situations in which we might need it: where there is only partial compliance or where unfavorable conditions obtain. Arvan argues that there are three conditions of ideal theory, and consequently that there are three circumstances in which we might need nonideal theory: where there are unfavorable conditions, partial compliance, and what he calls “no-circumstances-of-justice.” This last refers to the absence of “the normal conditions under which human cooperation is both possible and necessary,” such as roughly equal capacities, moderate scarcity and conflicting interests (Rawls, 1999: 109-10). For Arvan’s argument, see Arvan (2014: 97-100). But it seems to me that favorable conditions imply the circumstances of justice, that these circumstances are too minimal to be considered idealizations and that it is not clear if anything that can properly be called ‘Rawlsian nonideal theory’ can be applied where such circumstances do not obtain. Yet it seems that there may be at least one other sort of situation in which we might also need nonideal theory: a situation in which, because of past injustice, even the full implementation of correct ideal principles of justice will not suffice for justice. More about this possibility will be said below. To develop his argument, Mills distinguishes between what he calls “ideal ideal” theory and “rectificatory ideal” theory: the former is concerned with what justice requires in the ideal circumstance in which there is no past injustice, whereas the latter focuses on what the recification of injustice ideally requires (2009: 178). Mills then proceeds to make the by-now familiar point that ideal theory is supposed to fulfill a targeting role with respect to nonideal theory: it is supposed to identify the ideal that nonideal theory should aim at (see Simmons, 2010; Swift and Stemplowska, 2012). Mills understands this as meaning that ideal theory aspires to tell us which alternative policy options should be adopted in order to move closer to ideal justice. But he argues that ideal ideal theory cannot in general play “an adjudicative role in determining which” policy is superior as far as recificatory justice is concerned, for it “represents a goal located in a different conceptual space, on an alternate timeline to which we have no access” (2009: 179; see also Mills, 2015: 60). In other words, because its starting point is (so Mills says) an alternative society with no past injustice, ideal ideal theory cannot tell us which policy alternatives should be adopted in our nonideal society. This, Mills maintains, makes it useless for recificatory justice. As he writes, Ideal theory represents an unattainable target that would require us to roll back the clock and start over. So in a sense it is an ideal with little or no practical worth. What is required is the nonideal (rectificatory) ideal that starts from the reality of these injustices and then seeks some fair means of correcting for them, recognizing that in most cases the original prediscrimination situation… cannot be restored (2009: 180). So ideal theory is useless in identifying the rectificatory measures needed to take us to the ideal rectificatory justice aims at. Note that this implies that we don’t need to know what ideal justice requires to be able to know what rectificatory measures would be fair. In this respect, there is some similarity between Mills’ argument and Amartya Sen’s argument against Rawls. As Stemplowska has noticed; see Stemplowska (2008: 233 n.10). Sen argues that the Rawlsian search for perfect justice is neither necessary nor sufficient for comparative judgments of the kind that we are normally interested in in the unjust world in which we live (2006; 2009: esp. 15-18, 96-105). We do not need a theory of perfect justice to know, say, that slavery or famines are unjust, nor would such a theory be sufficient to tell us which of, say, a society that lacks public schools and one which lacks access to medical care is more unjust. Whereas for Sen we don’t need a theory of ideal justice to make comparative judgments of distributive justice, for Mills we don’t a theory of ideal justice to make comparative judgments of rectificatory justice. At this point attention should be brought to an easily overlooked aspect of Mills’ argument. That argument, as we have seen, assumes that since our actual world is marked by past injustice, rectificatory measures are needed to secure the rectificatory ideal. But why? Why not simply implement the two principles and so reform basic institutions? Mills’ implicit answer must be that where there is a legacy of past injustice, reforming basic institutions may not suffice for justice, and that in fact will not suffice in American context in which he is writing. Recall that I am setting aside formal apology and punishment and focusing exclusively on what is required for restoration or compensation (understood as making the person or persons to be compensated no worse off than they would have been but for the injustice). This, it is worth saying explicitly, is what I take Mills to be primarily concerned with (despite his brief mention, in Mills (2013: 13-14), of “symbolic measures”). To see this, take the formal equality of opportunity (EO) component of FEO. This component requires procedural fairness in the selection of candidates for positions of advantage; equally qualified candidates should have the same chances to obtain such positions. The problem is that where groups have suffered certain forms of past injustice, it is quite possible that their members will generally not be competitive in their qualifications. Thus securing formal EO would not suffice to rectify the legacy of past injustice (Pateman and Mills: 129; Mills, 2015: 54-55). And the same thing is true, Mills must argue, for substantive EO component of FEO as well as the other principles of justice, both individually and combined. In response, Shelby has argued that Mills misunderstands both ideal and nonideal theory, expecting too much from each. Ideal theory, he says, is supposed to serve as a guide for nonideal theory. But this does not mean that it is supposed to provide “a set of axioms from which theorems of rectification can be directly deduced” (2013: 155). On the other hand, the viability of any nonideal theory of justice depends on some associated ideal theory of justice; we must have on hand an ideal theory to make judgments of just and unjust, including judgments of racial injustice. On this view, unless we know what racial justice ideally requires, we can’t say which past acts were acts of racial injustice, and consequently we can’t say where action is needed if we are to move closer to racial justice, nor how injustice should be responded to (Shelby, 2013: 155-56). See also Simmons (2010: 33-4), and Stemplowska (2008: 230). In my view, there is merit in this defense of ideal theory: it is not its job to tell us which policies to adopt in order to move closer to ideal justice, and it does not seem that we can condemn something as (or know how to respond to something judged to be) racially unjust unless we have on hand ideal principles of racial justice. But I believe that more can be said. Recall that in his defense of the rectificatory ideal over the ideal ideal Mills argues that what we need is an ideal that “starts from the reality of these injustices and then seeks some fair means of correcting for them” (emphasis added). It is not clear, however, how we can determine “fair” means of correcting injustice without first having an ideal theory on hand. This is so for at least two reasons. First, ideal theory seems to be necessary to identify what counts as past racial injustice. Actually, what we need is not a full ideal theory of justice, but what Robeyns calls a “partial ideal theory” – in this case, an ideal theory of racial justice. See Robeyns (2008: 344). (More specifically, what we need is the second kind of partial ideal theory that she mentions.) This is Shelby’s main point. Different theories of justice will recognize different things as violations of justice, and consequently, they will recognize different things as requiring rectification. For example, consider the difference between a theory of justice which only recognizes formal EO, and one which recognizes FEO. Clearly, all else being equal, the rectificatory measures required by the former will be less than those required by the latter. True, Mills has argued that a weak – and, so he thinks, uncontroversial – principle of equal opportunity, extendable from formal EO, can nevertheless be strong enough to justify strong corrections of past injustice (see Pateman and Mills, 2007: Ch. 4, esp. 127-30). Such a principle would disallow “differential and superior credentials” to count when “they arise out of [a] history of discrimination… by law or custom” (129-30). But there are problems with this argument. It seems doubtful whether Mills’ principle can in fact be justifiably extended from formal EO. It is not clear why persons concerned with present and future discrimination in the selection of candidates (procedural fairness) would be concerned about superior credentials that arise out of past discrimination. Or at least this is not clear unless they are also concerned with equal life chances, and so concerned about disadvantages created by past discrimination. In other words, it seems that it is precisely because it appeals to considerations that justify FEO that Mills’ principle seems appealing. (A further problem is that libertarians would not accept formal EO, and so even if Mills’ principle is extendable from it, it would be controversial to them.) In response Mills might argue that all plausible theories of justice will condemn things like slavery and Jim Crow. This is obviously quite right. But I’m sure Mills would agree that racial injustice extends to practices that go well beyond slavery and Jim Crow, and the less obvious the injustice the greater the need for an ideal theory of justice. Shelby makes just this point; see (2013: 156). Second, nonideal theory seems to be constrained by ideal theory. As Christine Korsgaard, Robert Taylor and others have argued, in guiding our choice between alternative nonideal policies, it seems that nonideal theory has to be guided by criteria other than effectiveness (Korsgaard, 1996: Ch. 5, 147-151), (Taylor, 2009), and (Simmons, 2010: 33-4). For example, it would presumably not be permissible to publicly execute persons convicted of racial discrimination, even if such a policy was most effective in leading us to full compliance (Taylor, 2009: 490). While I agree with Taylor that ideal theory must to some extent constrain nonideal theory, I do not accept his view that pure procedural justice constrains it. For discussion, see Matthew (2015). So again, ideal theory would be necessary to identify permissible rectificatory measures. So far I have argued that ideal theory is needed to determine what means of correcting injustice are fair. I now want to argue that ideal theory is necessary to determine whether past injustice should be corrected at all. Above I noted that there are two components of racial justice theory: distributive racial justice and rectificatory racial justice. What I did not say is that it is not a foregone conclusion that all theories of justice will have the rectification of past injustice as a part of their ideal. It is possible, after all, to take a ‘let bygones be bygones’ view of past injustice. For a recent argument that justice as fairness should regard the rectification of injustice as part of its ideal, see Espindola and Vaca (2014). See also Boxill (2014: 194-5). Mills, of course, would reject such an approach, and I would join him in doing so. But whether rectification is or is not an ideal to be pursued is a question of ideal theory. It cannot be a part of nonideal theory because nonideal theory is supposed to tell us how to get to the final destination of ideal justice (or how to ameliorate injustice, if viable paths to ideal justice are blocked See Matthew (2015).); it is not supposed to tell us what ideal justice is. So it seems that there is reason to believe that ideal theory is essential for the kind of rectificatory work that Mills wants nonideal theory to do. The question I now want to ask is whether there is any need to turn to nonideal theory at all. After all, Mills’ discussion assumes that there is nothing – not noncompliance or the presence of unfavorable conditions – blocking the reforms called for by ideal theory. To be sure, and as we have already seen, Mills is of the view that, given the legacy of the past injustice that he is concerned with, ideal theory cannot simply be implemented if the rectificatory ideal is to be achieved. Yet if ideal theory isn’t blocked, that raises the question of whether the reforms it authorizes might not do some – or perhaps even all – of the rectificatory work that Mills wants nonideal theory to do. The Rectificatory Uses of FEO Above I had occasion to mention Shelby’s central argument in response to Shiffrin’s argument that the two principles need to be supplemented by an antidiscrimination principle. The paper in which this argument appears is in fact a wide-ranging defense of the value of Rawlsian liberalism for racial justice. What I want to note here is that in the course of this defense, Shelby appeared to argue that Rawls’ FEO principle could be used for rectificatory purposes. FEO goes beyond mere formal EO in requiring that all sectors of society should have a fair chance of attaining positions. As Rawls writes, “those who are at the same level of talent and ability, and have the same willingness to use them, should have the same prospects of success regardless of their initial place in the social system. In all sectors of society there should be roughly equal prospects of culture and achievement for everyone similarly motivated and endowed” (1999: 63). In an argument consisting of less than a page and a half, Shelby claimed that when coupled with the liberty principle, the institutionalization of FEO would “mitigate, if not correct” socioeconomic disadvantages that are the result of past racial injustice (2004: 1711). Though he is not sure what “institutional reforms” this principle might require, he writes that “it seems clear that it would require, at a minimum, considerable redistribution of wealth, the expansion of educational and employment opportunities, and aggressive measures to address discrimination in employment, housing and lending” (2004: 1711). Though Shelby’s defense was wide-ranging, it was this FEO argument, in particular, that attracted Mills’ critical attention, and in 2013 he published an article-length critique of it (2013). It should be clear why this should not be surprising. After all, the implementation of FEO is something that is called for by Rawls’ ideal theory. Thus if Shelby is right, then it seems that at least some of the need to draw on a nonideal theory is obviated, and Mills’ argument that ideal theory is useless for rectificatory purposes would have to be scaled back if not abandoned. Let us turn, then, to Mills’ critique. Though he makes a number of points – that Shelby does not attach much weight to the ideal/nonideal theory distinction, that if Robert Taylor’s (2009) pure procedural justice argument against strong forms of affirmative action works in that case, then it also works against Shelby’s “more radical redistributivist program” (2013: 16) that he jumbles together the wrong of racial injustice and the wrong of class injustice, for example – I believe that his strongest arguments pertain to the question of what Rawls intended. In this respect Mills draws attention to a number of passages where “it would have been natural” for Rawls to make use of FEO if he shared Shelby’s rectificatory view of it, but did not do so (2013: 7-10). Since I do not go on to contest them, I will not detail most of Mills’ arguments here. (I will, however, return to his claim that Shelby jumbles together the wrongs of racial and class injustice.) It seems to me that even if they are correct, they are inconclusive. Whatever Rawls’ intention, there may be features of justice as fairness such that FEO can indeed – perhaps even must – perform rectificatory work. I will now argue for this conclusion. I will argue that where there is a socioeconomic legacy of past injustice, FEO has to perform rectificatory work if it is to be implemented. The issue turns on the important – if relatively neglected – question of what it is for a principle of justice to be ‘implemented.’ Here the relevant question is whether FEO counts as implemented by the realization of a certain state of affairs, or by a certain design of institutions. This question has come up in connection with Cohen’s dispute with Rawls. Citing the behavior of talented people described earlier, Cohen suggested that the difference principle (which requires that we maximize the position of the worst off) should be understood in a “strict” rather than a “lax” manner; inequalities ‘necessary’ to benefit the least well-off should be understood as necessary apart from the intentions of the talented. Moreover, he suggested that implementation of the strict difference principle requires an “egalitarian ethos” in which citizens “internalize, and… unreflectively live by, principles that restrain the pursuit of self-interest” (2008: 73). In the literature that subsequently developed, some critics seized on this point about ethos to argue that, as Joshua Cohen puts it, “it is entirely consistent with the Rawlsian version of egalitarianism to affirm that the social ethos matters to distributive justice” (2002: 383). Others who have endorsed this response include: Pogge (2000), Tan (2004: 341-2, 344), Scheffler (2006: 118), and Baynes (2005: 191). In her work endorsing and extending this claim, Miriam Ronzoni provides what as far as I know is the only extended discussion of what it means to “implement” principles of justice. Ronzoni’s main argument seeks to show that a just basic structure “is not blind to problematic social norms” (2008: 205). It starts by observing that there are two ways of understanding how principles of justice regulate the basic structure: the first has it that they apply to the “conduct” of institutions, and the second has it that they are to bring about the social conditions that would realize the distributive pattern called for by the principles. To illustrate this difference, take FEO. The first interpretation understands the institutions that make up the basic structure as a fixed and given set of institutions – that is, an unchanging set of identifiable institutions (Ronzoni 2007: 72). For Rawls, an ‘institution’ is “a public system of rules which defines offices and positions with their rights and duties, powers and immunities, and the like” (1999: 47). Now imagine a society in which an informal pattern of rules exist in families that exerts pressure on girls and women not to apply to competitive schools. In this case, we might imagine that the given institutions of the basic structure (as this interpretation sees it) promote women’s equal educational opportunity as much as they can (consistent with familiar limits of legal coercion); for example, they might establish preferential access to women, offer certain female-only services to prevent or reduce female attrition, and so on. But all of this might not be enough to counter the relevant informal norms. Let’s suppose that they are not. Because the given institutions have been optimally arranged to secure FEO for women, the basic structure is just on the first interpretation. Yet the outcome is arguably unjust nevertheless. Ronzoni agrees that the outcome is unjust, but argues that what makes it unjust is that the basic structure is unjust (2008: 208-10). Like the first interpretation, informal social norms are not part of the basic structure on the second interpretation. But unlike the first, “their nature and effects are crucial to determining whether the basic structure is just or not” (Ronzoni, 2008: 210). This is because the justness of a basic structure is determined by how well it responds to the specific societal context to create the conditions that would realize the pattern called for by the principles. For informal social norms are “not conceived of as fully independent of the general background created by the basic structure” (Ronzoni, 2007: 75), For some evidence that informal social norms are influenced by major social institutions, see, e.g., Bowles (1998). and we create institutions to (and assess them by how well they) “realize certain standards of justice” (Ronzoni, 2008: 210). To ask “whether an institutional setting respects certain principles of justice without scrutinizing its structure, scope and strategy” in light of its context, Ronzoni argues, is “not an intelligible enterprise” (2008: 210). And if the justness of a basic structure is determined by how well it responds to its specific context, the basic structure described above is unjust because it responds badly, for (let us suppose) there are alternative institutions that could have been more effective in securing FEO for women, more effective, that is, in affecting its context to change informal social norms. Thus it is only on the first, naïve understanding of how principles of justice regulate the basic structure that it might be thought that a just basic structure is blind to social norms. It might be wondered how Ronzoni’s account relates to Rawls’ four-stage-sequence argument discussed earlier. In particular, it might be wondered if that four-stage process does not preclude her account since it is here that historical context is to be taken into. The answer is that it doesn’t. The four-stage process can be used to address, to some extent at any rate, the historical context while the institutions of the basic structure are understood in the way that Ronzoni argues we should reject. On this view, while the particular context is to be taken into account at the legislative stage, for example, legislation still has to work within the constraints set by given and fixed institutions. I thank an anonymous reviewer for bringing this issue to my attention. If Ronzoni is right, then her view has important implications for the debate over FEO’s rectificatory potential. For if FEO can only count as implemented if a certain state of affairs is realized, then where there is a socioeconomic legacy of past injustice, the institutional measures required to realize the relevant state of affairs may be different from what is required where there is no such legacy of injustice. Espindola and Vaca seem to accept a similar view; see Espindola and Vaca (2014: 237). Thus suppose that, as a result of injustice, the members of a group adopt values and practices that tend to militate against their economic success. FEO would seem to call for institutional measures aimed at cultural reform. If this can’t be accomplished through a different design of existing institutions, different institutions may be required. So take our educational institutions. It may be that to achieve the kind of wide-ranging transformation of cultural values and practices needed here, we should replace the system of schooling with which we are familiar with more a comprehensive and holistic system that starts earlier, instructs differently and encompasses far more than just the education of students (e.g. various social and health services). The system that I am imagining here is modelled on the Harlem Children’s Zone. For a discussion of its goals, methods and the need it addresses, see esp. Tough (2008). It should be noted that questions can be raised here about what changes an institution must undergo for it to change its identity. What, for example, must change about schools for them to become some other kind of institution? Answering these questions, I believe, will lead us too far afield. I will say, however, that I am not sure that a general answer can be given, that there are some features of institutions (e.g. their purposes) the changing of which will always have the effect of changing their identity. I thank an anonymous reviewer for bringing this issue to my attention. Or suppose that the legacy of past injustice consists, not in self-defeating cultural practices and values, but in inequality of wealth between racial groups. FEO would seem to call for wealth redistribution between the groups involved, and here too if this can’t be accomplished through a different design of existing institutions, different institutions may be required. Now since FEO is demanded by Rawls’ ideal theory, the measures its implementation requires would have to be considered reformative, as opposed to rectificatory. They would be measures required to reform the society so that it puts an end to injustice. And yet, it seems clear that these reformative measures would at the same be performing rectificatory work; they would clearly perform the work of undoing the disadvantages created by past injustice. Rectificatory measures are typically understood as falling under nonideal theory. So if we understand what it means to implement principles of justice in the way suggested by Ronzoni, it turns out that implementing Rawls’ ideal theory would do much to carry out some of the rectificatory work that it seems any plausible nonideal theory would require. This means that there is less need to turn to nonideal theory than it may at first appear – less need, not no need. Even if implementing FEO would address all of the advantages related to equality of opportunity, this would still fall short of a plausible full rectificatory ideal. At this point consideration should be given to an issue that I mentioned above, Mills’ claim that Shelby jumbles together the wrongs of racial and class injustice. In Mills’ view, it is not just that Rawls never intended for FEO to cover race; it is also that attempts to use FEO to rectify racial injustice represent a ‘category mistake’ (2013: 19-22). To be disadvantaged by the normal workings of a capitalist economy, he argues, is very different from being disadvantaged because of racial oppression. As he writes, “Disadvantaging from social oppression in a non-ideal racist society is categorically different from disadvantaging because of class membership in an ideal society” (2015: 62; see also Mills, 2013: 20-21). My reply is that these are indeed different, but this does not show that implementing FEO would not perform rectificatory work. FEO is concerned with realizing a certain state of affairs (i.e. one where equally talented and motivated citizens from all sectors of society have the same prospects of success); it does not differentiate between disadvantages based on their source. Indeed, it is precisely this feature that allows its implementation to perform both reformative and rectificatory work. As for Mills’ claim that Rawls refrains “from using FEO to cover race, since this is a problem of partial compliance theory” (2015: 62), Waligore seems to have a similar view; see Waligore (2016: 51). what explains this refraining is Rawls’ claim that in a well-ordered society race would not define a point of view from which we should assess the basic structure, but this could only be true if FEO addresses race at the level of ideal theory, if, that is, implementing FEO addresses raced-based inequality of opportunity (Rawls, 2001:66). Returning to Ronzoni, I believe that her argument is importantly right. Principles of justice can only be considered implemented if the state of affairs called for by them is realized – or, at least, realized as much as possible. But is there any evidence that this was actually Rawls’ view? Although Ronzoni claims that she is primarily “interested in showing how ‘certain core ideas of Rawls’ theory of justice can be developed in a way that makes them worth defending,’” she also claims that there is textual evidence that it was indeed Rawls’ view (2008: 213, quoting Thomas Pogge). Such evidence, she says, is provided in the following passage: It is conceivable that a social system may be unjust even though none of its institutions are unjust taken separately: the injustice is a consequence of how they are combined together into a single system. One institution may encourage and appear to justify expectations which are denied or ignored by another. These distinctions are obvious enough. They simply reflect the fact that in appraising institutions, we may view them in a wider or narrower context (1999: 50). What is important, Rawls says, is how institutions look when viewed in a wider context. But it is hard to see what support this provides for the view that principles of justice can only be considered implemented if the state of affairs called for by them is realized. Still less, it is worth noting, does it provide support for the view that the institutions of the basic structure should not be understood as fixed. Recall that on the interpretation of the basic structure that Ronzoni rejects the institutions of the basic structure are a set of fixed and given institutions. The above passage provides us with no reason to reject this interpretation; indeed, nothing Ronzoni says provides us with a good reason to reject it. Since I have already indicated that I think Ronzoni’s argument is importantly right, it might be surprising that I would say this. But someone who accepts that principles of justice have to realize certain of states of affairs to be implemented – or rather, who accepts that they have to realize certain states of affairs as much as possible to be implemented – is not thereby committed to the view that the institutions of the basic structure cannot be fixed. To see this, note that the fixed view does not require that there be no changes over time. It is possible that fixed institutions can be designed differently with a view to realizing a certain state of affairs. That is to say that the design of given institutions can vary due to different conditions without the institutions themselves varying. For example, both private-property and socialist regimes can make use of the same institutions (such as a political constitution defining rights and market institutions), though they are designed differently (Cf. Rawls, 1999: 239-42). Ronzoni seems drawn to the view that the institutions of the basic structure cannot be fixed by her view that, on her account of what it means to implement principles of justice, “no basic structure is just if it fails adequately to respond to the context it is meant to be applied to” (2007: 74). The idea is that if the institutions of the basic structure are fixed, then this raises the possibility that, because they are fixed, no possible configuration of them will produce a just basic structure – that is, no possible configuration will realize the distributive pattern called for by the principles. And so her solution is to say not just that the institutions of the basic structure are variable, but that they are as variable as circumstances may require to produce the relevant pattern. But it is not at all clear on what Rawlsian basis she can demand more of institutions than they can deliver. Moreover, Ronzoni may have come to see this, retreating from the passage just quoted from “Two Concepts of the Basic Structure” (2007) to a weaker view in “What Makes a Basic Structure Just?” (2008). In private communication, Ronzoni expresses some uncertainty about her view, but says that she “probably” holds the weaker view. In the latter paper she seems to accept that no deployment of institutions may affect some particular context so as to create the conditions that would fully realize the principles of justice, and that this would not make the basic structure unjust (2008: 217). In fact, there is a suggestion of this view in (2007); note the quotation on p. 75 quoted above, paying special attention to the word ‘fully.’ On this second view, a basic structure is unjust if there are alternative configurations of it (whether by means of a different design of a set of institutions, or a different set of institutions) that would better realize the principles; on the first, a basic structure is unjust if it fails to realize the principles. But having retreated from the stronger view to the weaker view, there would no longer seem to be as powerful a reason to hold on to the view that the institutions of the basic structure cannot be fixed. Yes, if the institutions of the basic structure are fixed, then that structure will be less flexible in being able to respond to particular contexts to create the conditions that would realize the principles. But this should be easier to accept if one retreats from the view that the institutions of the basic structure must be limitlessly variable. But even if the institutions of the basic structure are not limitlessly variable – even if we should reject Ronzoni’s initial position on this issue – and even if her ultimate view does not require that they be variable at all, I do not believe that Rawls regarded them as fixed either. For one thing, in Theory he explicitly says that the institutions of a just society may change over time, presumably as conditions change (1999: 401). More importantly, in Justice as Fairness: A Restatement he notes that his characterization of the basic structure “does not provide a sharp definition,” and defends this decision by arguing, among other things, that a sharp definition would “risk wrongly prejudging what more specific or future conditions may call for, thus making justice as fairness unable to adjust to different social circumstances” (2001: 12). It seems reasonable to suppose that different social circumstances matter because different institutions, or different designs of existing institutions, might be required in them if there are to be certain effects. On this view, the institutions of a basic structure are whatever institutions collectively have certain effects: presumably, the effect of exerting a certain kind of “profound influence” on citizens’ life prospects (Rawl, 1999: 7), and the effects that constitute maintaining “background justice” (Rawls: 2005: Lecture VII). This provides support for Ronzoni’s central insight. For if the institutions of the basic structure are determined by their effects, then they are variable, and so a change of institutions may be demanded if the effects called for by the principles are not realizable by existing ones. FEO as a Principle of Rectificatory Justice? I have just argued that where there is a socioeconomic legacy of past injustice, FEO has to do some rectificatory work if it is to be implemented. This goes against Mills’ view that ideal theory is useless for rectificatory purposes, and supports Shelby’s position. But Shelby has argued that he never proposed FEO as a principle of rectificatory justice. In this section I discuss this claim. Shelby’s strategy in denying that he proposed using FEO as a principle of rectification is to quote passages from “Race and Social Justice” and explain what they do and do not mean. In a key passage he writes: The key point is that these socioeconomic disadvantages could be mitigated so that members of historically oppressed racial groups were not materially disadvantaged in the competition for opportunities and valued positions in society. This is not the same as calling for compensation for past wrongs. It would be a forward-looking measure used to bring society closer to the ideal of a well-ordered society. It would not replace the need to make amends for or acknowledge past racial injustice (2013: 158). The implementation of FEO would mitigate the socioeconomic disadvantages left by past racial injustice, but this is not the same as compensating for past wrongs. On its face, Shelby’s argument seems odd; if the implementation of FEO would mitigate the socioeconomic disadvantages left by past injustice, it would seem to be doing rectificatory work. So what is he thinking here? The answer lies in how he understands principles of rectification. To understand his view here, it is necessary to understand his distinction between two different sets of principles. He lists and explains them as follows: 1. Principles of reform and revolution: the principles that should guide efforts to bring an unjust institutional arrangement more in line with justice such that the society’s members have a more just (though not necessarily perfectly just) society within which to live. 2. Principles of rectification: the principles that should guide the steps a society takes to remedy or make amends for the injuries and losses the oppressed have suffered as a result of past injustice (2013: 154). Shelby actually discusses four sets of principles, but only two of them are relevant for my purposes. (The other two are penal principles and principles of political ethics.) With these principles so distinguished, Shelby suggests that principles of reform and revolution (1) and principles of rectification (2) can be thought of as jointly constituting corrective justice (2013: 154). Further, he acknowledges that though Rawls provides some guidance for (1) and principles of political ethics, he “provides almost no help” with (2), principles which he identifies as Mills’ “chief concern” (2013: 154). Despite distinguishing between principles of reform and revolution, on the one hand, and principles of rectification on the other, and saying that Mills’ chief concern is with the latter, Shelby goes on to understand Mills’ claim that the two principles “can help us prevent injustices from occurring but cannot help us rectify injustices after they have already occurred” (2013: 157) as a claim that these principles are useless as principles of reform and revolution (“I see no reason to believe that Rawls thought a well-ordered society could never be created out of an unjust one” [157]). More to the point, he explains (1) and (2) as follows: Principles of type (1) have to do with altering the basic structure of a society so that it better approximates a well-ordered society. Type (2) principles address the need to make amends to those burdened and harmed by unjust basic structures. Type (1) principles are forward looking, oriented toward establishing a just society. Type (2) principles are backward looking, oriented toward settling unpaid moral debts (2013: 154). Type (1) principles are concerned with altering the basic structure of society to create a well-ordered (or better-ordered) society, whereas type (2) principles are concerned with settling moral debts created by past unjust institutions. The problem with this is that Shelby wrongly understands rectification as solely concerned with “making amends” for past wrongdoing, where this does not involve reforming institutions so as to “bring an unjust institutional arrangement more in line with justice,” even when this has the effect of lessening the disadvantages of groups that have suffered injustice. But this is not the only way one can understand rectificatory justice; one can also understand it as an ideal that also encompasses institutional reform, as Shelby himself acknowledges (2013: 154-55). This may perhaps be done by maintaining that, in cases of social injustice, fully making amends for unjust losses or injuries cannot be done in the absence of institutional reform. Although I won’t try to defend it here, on this view, it is a mistake to think that “one could fully pay reparations to the victims of past racial injustice and yet their society remain … racially unjust” (2013: 154). Still, even if all of this is right, it might be said that none of it establishes that Shelby proposed FEO as a rectificatory principle; even if it is true that one can understand rectification as an ideal that encompasses institutional reform, Shelby himself expressly does not so understand it. This, it must be admitted, is true enough. But my aim is not to argue that Shelby proposed FEO as a rectificatory principle. Instead, my argument is the less ambitious one that since the reformative measures called for Rawlsian ideal theory take us some way towards the rectificatory ideal that it seems any plausible theory would endorse, there is less need to theorize a nonideal theory to justify the measures needed to achieve this ideal, a result which means that ideal theory is not completely useless for rectificatory purposes. Conclusion In this paper I have been concerned with advancing the debate over the value of Rawls’ work for issues of racial justice. I have defended three main claims: first, that there is some reason to doubt to whether justice as fairness can authorize all of the measures needed to prevent the introduction of racial injustice; second, that ideal theory is essential for the kind of rectificatory work that Mills wants nonideal theory to do; and finally, that where there is an economic legacy of past injustice, FEO could not be implemented if it did no rectificatory work, a result which means that however at fault Rawlsians are for generally avoiding the theorizing of nonideal theory, there is less need to turn to it than it may at first appear. I assume that Rawlsians’ have not avoided rectificatory racial justice because of the rectificatory work FEO would do if implemented. Far from it being the case that Rawls does not tell you what to do if you find yourself in an unjust society and want to end ongoing injustice, in implementing the two principles, and thereby reforming unjust institutions, one would be doing much not just to end injustice, but rectify it as well. Acknowledgments I would like to thank two anonymous referees and Louis-Philippe Hodgson and Miriam Ronzoni for reading and providing helpful comments on at least one version of this paper. References Anderson E (2010) The Imperative of Integration Princeton: Princeton University Press. Aristotle (1973) Nicomachean Ethics. In: McKeon R (ed.) Introduction to Aristotle Second Ed. Chicago: University of Chicago Press. Arvan M (2014) First Steps Towards a Nonideal Theory of Justice. Ethics & Global Justice 7: 95-117. Baynes K (2005) Ethos and Institution: On the Site of Distributive Justice. Journal of Social Philosophy 37: 182-195. Boettcher J (2009) Race, Ideology, and Ideal Theory. Metaphilosophy 40: 237-259. Boxill B (2014) Compensation and Past Injustice. In: Cohen A I and C H Wellman (eds.), Contemporary Debates in Applied Ethics Second Edition. Oxford: Wiley-Blackwell, pp. 191-202. Bowles S (1998) Endogenous Preferences: The Cultural Consequences of Markets and other Economic Institutions. Journal of Economic Literature 36: 75–111. Carcieri M D (2010) Rawls and Reparations. Michigan Journal of Race and Law 15: 267-316. Charles G E (2004) Colorblindness from the Original Position. Tulane Law Review 72: 2009-2036. Cohen G A (2008) Rescuing Justice and Equality. Cambridge, MA: Harvard University Press. Cohen J (2002) Taking People as They Are? Philosophy & Public Affairs 30: 363-386. Dombrowski D A (2011) Rawlsian Explorations in Religion and Applied Philosophy. University Park, PA: Pennsylvania University Press, Ch. 4. Espindola J and M Vaca (2014) The Problem of Historical Rectification for Rawlsian Theory. Res Publica 20: 227-243. Goodman C C (2007) Beneath the Veil: Corollaries on Diversity and Critical Mass Scholarships from Rawls’s Original Position on Justice. Washington and Lee Journal of Civil Rights & Social Justice 13: 285-347. Graham K (2010) Beyond Redistribution: White Supremacy and Racial Justice. Lanham, MD.: Lexington Books. Kang H R (2016) Can Rawls’s Nonideal Theory Save His Ideal Theory? Social Theory and Practice 42: 32-56. Korsgaard C (1996) Creating the Kingdom of Ends. Cambridge: Cambridge University Press, Ch. 5. Lebron C J (2013) The Color of Our Shame: Race and Justice in Our Time. New York: Oxford University Press, Ch. 1. Loury G C (2002) The Anatomy of Racial Inequality. Cambridge, MA: Harvard University Press. Matthew D C (2015) Rawlsian Affirmative Action: A Reply to Robert Taylor. Critical Philosophy of Race 3: 324-343. Matthew D C (2014) Purview and Permissibility: The Site of Distributive Justice and the Case of Private Racial Discrimination. Social Theory and Practice 40: 73-98. McCarty T (2004) Political Philosophy and Racial Injustice: From Normative to Critical Theory. In: Benhabib S and N Fraser (eds.) Pragmatism, Critique, Judgment. Cambridge: MIT Press, pp. 147-168. Mills C W (2015) Racial Equality. In: Hull G (ed.) The Equal Society: Essays on Equality in Theory and Practice. Lanham, MD: Lexington Books/Rowman & Littlefield, pp. 43-71. Mills C W (2013) Retrieving Rawls for Racial Justice? A Critique of Tommie Shelby. Critical Philosophy of Race 1: 1-27. Mills C W (2009) Rawls on Race/Race in Rawls Southern Journal of Philosophy Supplement 47: 161-84. Mills C W (2008) Racial Liberalism. PLMA 123: 1380-1397. Mills C W (2005) ‘Ideal Theory’ as Ideology. Hypatia: A Journal of Feminist Philosophy 20: 165-184. Nagel M E, et al. (2003) Symposium on Charles Mills’s The Racial Contract. In: Stokes C and T Meléndez (eds.) Racial Liberalism and the Politics of Urban America. East Lansing, MI: Michigan State University Press, pp. 11-50. Nagel T (2003) John Rawls and Affirmative Action. Journal of Blacks in Higher Education 39: 82-84. Pateman C and C W. Mills (2007) Contract and Domination. Cambridge: Polity Press. Pogge T W (2000) On the Site of Distributive Justice: Reflections on Cohen and Murphy Philosophy and Public Affairs 29: 137-169. Rawls J (2005) Political Liberalism Expanded Ed. New York: Columbia University Press. Rawls J (2001) Justice As Fairness: A Restatement. Kelly E (ed.) Cambridge, MA: Harvard University Press. Rawls J (1999) A Theory of Justice Revised Ed. Cambridge: Harvard University Press. Roberts R C (2002) Justice and Rectification: A Taxonomy of Justice. In Roberts R (ed.) Injustice and Rectification. New York: Peter Lang, 7-28. Robeyns I (2008) Ideal Theory in Theory and Practice. Social Theory and Practice 34: 341-62. Ronzoni M (2008) What Makes a Basic Structure Just? Res Publica 14: 203-218. Ronzoni M (2007) Two Concepts of Basic Structure, and their Relevance to Global Justice. Global Justice: Theory Practice Rhetoric 1: 68-85. Scheffler S (2006) Is the Basic Structure Basic? In: Sypnowich C (ed.) The Egalitarian Conscience: Essays in Honour of G. A. Cohen. Oxford: Oxford University Press, pp. 102-130. Sen A (2009) The Idea of Justice. Cambridge, MA: Harvard University Press. Sen, A (2006) What Do We Want from a Theory of Justice? Journal of Philosophy 103: 215-38. Shelby T (2014) Integration, Inequality and Imperatives of Justice: A Review Essay. Philosophy & Public Affairs 42: 253-285. Shelby T (2013) Racial Realities and Corrective Justice: A Reply to Charles Mills. Critical Philosophy of Race 1: 145-162. Shelby T (2007) Justice, Deviance, and the Dark Ghetto. Philosophy & Public Affairs 35: 126-160. Shelby T (2004) Race and Social Justice: Rawlsian Considerations. Fordham Law Review 72: 1697-1714. Shiffrin S V (2010) Incentives, Motives and Talents. Philosophy & Public Affairs 38: 111-142. Shiffrin S V (2004) Race, Labor, and the Fair Equality of Opportunity Principle. Fordham Law Review 72: 1643-1675. Simmons A J (2010) Ideal and Nonideal Theory Philosophy & Public Affairs 38: 5-36. Stemplowska Z (2008) Comment A: Worth the Paper it’s Written on. In: King D (ed.) Symposium: Contract and Domination By Carole Pateman and Charles W. Mills. Journal of Political Ideologies 13: 228-233. Swift A and Z Stemplowska (2012) Ideal and Nonideal Theory. In: Estlund D (ed.) The Oxford Handbook of Political Philosophy. Oxford: Oxford University Press, pp. 373-389. Swift A (2008) Comment F. In: King D (ed.) Symposium: Contract and Domination By Carole Pateman and Charles W. Mills. Journal of Political Ideologies 13: 246-250. Tan K (2004) Justice and Personal Pursuits. Journal of Philosophy 101: 331-362. Taylor R S (2009) Rawlsian Affirmative Action. Ethics 119: 476-506. Tough P (2008) Whatever It Takes: Geoffrey Canada's Quest to Change Harlem and America. New York: Houghton Mifflin. Valentini L (2009) On the Apparent Paradox of Ideal Theory. Journal of Political Philosophy 17: 332-355. Voorhoeve A (2005) Incentives and Principles for Individuals in Rawls’s Theory of Justice. Éthique et économique/Ethics and Economics 3: 1-7. Waligore T (2016) Rawls, Self-respect, and Assurance: How Past Injustice Changes What Publicly Counts as Justice. Politics, Philosophy & Economics 15: 42-66. PAGE \* MERGEFORMAT 1