Defeasible Luck Equality

In my third chapter, “Defeasible Luck Equality,” I argue that luck egalitarianism is compelling if understood as a theory of a particular value (substantive fairness), but that it’s implausible if understood as a theory of all-things-considered institutional rightness. In support of the above pair of conditionals, I argue that luck egalitarianism does an excellent job of cohering with some of our core judgements of fairness, e.g., the judgement that it’s fair for an unlucky gambler or for a leisure-loving surfer to have fewer resources, or the judgement that it’s unfair for someone to be made less well-off by a congenital disability. Luck egalitarianism does not cohere with all of our moral judgements, but it doesn’t have to. Luck egalitarianism, as a theory of substantive fairness, need only cohere with our judgements of fairness: judgements belonging to other values, e.g., compassion, efficiency, etc., should be omitted. This means that, at times, luck egalitarianism must cohere with judgements that come apart from what is morally right. Subsidizing lifestyle choices and bailing out the imprudent is unfair, and an adequate theory of fairness must recognize this. The fact that compassion sometimes trumps fairness in cases of imprudence just means that a theory of fairness won’t always be an adequate guide to action.^{Footnote 1}

In his commentary, Matthew Palynchuk criticizes me for drawing upon the judgement that it’s unfair for someone to be made less well-off by a congenital (or otherwise brute luck) disability. I invoke this judgement in support of luck egalitarianism, and Palynchuk argues that, by doing so, I presuppose the medical model of disability, i.e., the idea that disability is an intrinsic property of the person who possesses it. Though Palynchuk acknowledges that a person may be either physically or cognitively impaired, and that impairment is an intrinsic property of the person who possesses it, he maintains that various social (and thus extrinsic) conditions must be met for an impairment to qualify as a disability. According to Palynchuk, an impairment becomes a disability when it’s stigmatized, when institutions are set up in a manner that makes goods inaccessible to those who are impaired, etc. Social factors like these are allegedly what make an impairment disabling, and thus disadvantageous, on the social model of disability that Palynchuk endorses.^{Footnote 2}

Palynchuk offers a couple of different reasons for why he thinks my discussion of disability presupposes the medical model. His most compelling reason, however, is that I appeal to disability when explaining why luck egalitarianism is a better theory of substantive fairness than Rawls’s theory is. In my explanation, I note that Rawls’s theory fails to acknowledge that the disadvantageous effects of disability extend beyond disability’s effects on one’s socio-economic class. Relative to other members of the worst-off class, disabled members face additional hardship in the form of either a lesser opportunity for welfare or fewer physical (or cognitive) resources, but Rawls’s theory of justice is not responsive to this (it treats all members of the worst-off class the same). Luck egalitarianism, by contrast, is capable of recognizing the unfairness of the additional burdens people who are disabled face, and it is thus a superior theory of substantive fairness, even if it’s also an inferior theory of institutional rightness.^{Footnote 3} As Palynchuk correctly notes, however, my argument doesn’t work on a purely social understanding of disability’s disadvantages. If impairments are not disadvantageous in and of themselves but rather because of social factors, then, in a Rawlsian society, there would be no brute luck inequality between impaired members of the worst-off group and unimpaired members of the worst-off group. Fair equality of opportunity eliminates any social sources of disadvantage, leaving intact only disadvantages traceable to talent or ability, and to one’s choices. From the perspective of the social model, then, which (according to Palynchuk) claims that an impairment is only disadvantageous because of social factors, impairment could not be a source of luck inequality between members of the worst-off group, at least not when fair equality of opportunity is realized.

As I indicated above, I think that Palynchuk is correct to note that some of what I say in my book is incompatible with a purely social understanding of the disadvantages associated with disability. Unlike Palynchuk, however, I don’t think this is a flaw in the book. On the one hand, I agree that something along the lines of a distinction between impairment and disability is appropriate. There’s more to the concept of disability than possessing a cognitive or physical condition, and much of the disadvantage associated with disability is social in nature, e.g., it’s caused by stigmatization and by society’s inattentiveness to accessibility. On the other hand, the plausible claim that there’s more to the concept of disability than a person’s cognitive or physical condition is different from the, frankly, implausible claim that disability’s disadvantageousness is strictly social.^{Footnote 4} For one, it’s inappropriate to label the underlying cognitive or physical condition an ‘impairment’ if the underlying condition is not intrinsically disadvantageous. It would be better to instead use evaluatively neutral language. Additionally, a purely social understanding is unable to account for the disadvantages impairment creates in asocial situations, such as on a deserted island or in a tract of uninhabited wilderness. Most standard impairments, e.g., lacking the use of one’s legs, lacking the use of a sense, etc., would be disadvantageous in asocial situations like this, and social factors have nothing do with that.

The claim that disability’s disadvantages are partly asocial does not imply that a person who is disabled has a life less worth living, any more than being less strong or less intelligent makes one’s life less worth living. The value of a person’s life is distinguishable from how advantaged she is, and Elizabeth Anderson’s claim that luck egalitarianism expresses a disrespectful kind of pity towards people who are disabled^{Footnote 5} fails to make that distinction. In his discussion of Anderson’s claim, Palynchuk is right to draw a distinction between matters of status and matters of condition. Sometimes, what we mean when we say that a person’s life has value is specifically that she has moral status, i.e., that she is a member of the moral community whose interests matter just as much as the interests of other members of the moral community. We might, for example, contrast a human being with a house plant and note that the human being’s life has intrinsic value but that the house plant’s life does not. Other times, what we mean when we say that a person’s life has value is that her life is good, i.e., that she is living a flourishing life or is experiencing a high level of subjective welfare. Though it would be disrespectful to claim that a person who is disabled is not a member of the moral community, Palynchuk is wrong to think that the same is true of the claim that disability is a partly asocial source of hardship. Again, claiming that disability is a partly asocial source of disadvantage is analogous to claiming that having less strength or intelligence is a source of disadvantage, i.e., analogous to the claim that, all things holding equal, a person who is less strong is less (physically) resourced or has somewhat fewer opportunities for welfare than a person who is stronger, and I don’t think that there’s anything disrespectful about the latter claim. Claims like the latter one only become disrespectful when they’re exaggerated, e.g., it would be disrespectful to claim that disability is an insurmountable hardship or that a person who is disabled is incapable of flourishing. There’s no inconsistency between (a) claiming that disability is a partly asocial source of hardship, and (b) recognizing the significant talent and resilience that people who are disabled possess.

Cohen’s Equivocal Attack

In her article, Kristin Voigt defends G.A. Cohen against the criticisms I make in my fourth chapter, “Cohen’s Equivocal Attack on Rawls’s Basic Structure Restriction.” In that chapter, I argue that Cohen’s critique of the difference principle’s restricted scope, i.e., his critique of Rawls’s view that the difference principle applies to the design of institutions but not to the uncoerced choices citizens make within those institutions, equivocates across the distinction between fundamental values and regulatory principles. Since the difference principle is an institutional regulatory principle, the question at hand when discussing its scope of application is whether, within a Rawlsian framework, the regulatory principles justified for institutions are also justified for the personal context. When he replies to Rawls’s defenders, however, Cohen presupposes that the issue at hand is whether a fundamental value applies to the personal context. For example, when replying to the complaint that applying the difference principle to personal decision-making places excessive demands on citizens, Cohen offers two main responses. First, he claims that citizens have a personal prerogative to deviate (to some extent) from the demands of the difference principle for the sake of pursuing their non-justice projects. Second, he claims that subjective welfare ought to be included in the difference principle’s metric, and that doing so mitigates the distributive demands it places on the better-off, since the better-off would become the worst-off in the event that they became too miserable. Unfortunately for Cohen, subjective welfare is very difficult to measure, and it is thus not appropriately included in the metric of a principle that’s meant to guide action. Similarly, balancing the difference principle against self-interest and the personal values that comprise a personal prerogative treats that principle as if it were a fundamental one, i.e., it treats the difference principle as if it were an input in practical reasoning that must be balanced against other considerations when deciding what to do. All-in-all, Cohen’s replies are only effective if we assume that the question at hand is whether a particular fundamental value applies to the personal context, instead of a regulatory principle.^{Footnote 6}

In her defence of Cohen, Voigt argues that I conflate Cohen’s view about what Rawlsians are, as matter of consistency, committed to, with his view about what the actual relationship between justice and personal choice is. Cohen is not himself committed to the view that the difference principle correctly supplies the content of distributive justice. He thinks that justice is a fundamental value and that luck equality, not the difference principle, correctly specifies what distributive justice is. As such, Cohen presumably believes that the principle of justice to which citizens ought to commit themselves is the principle of luck equality. Voigt’s claim is that my critique of Cohen’s welfare-inclusive, prerogative-constrained difference principle is not a critique of the ethos that Cohen, external to his engagement with Rawls, believes is required of a fully just society.

I’ll start by noting that I agree with most of what Voigt states in her commentary. To a large extent, the goal of my fourth chapter is to show that Cohen’s canonical, internal critique of Rawls fails, and that the reason it fails is that his replies to Rawls’s defenders equivocate across the distinction between fundamental values and regulatory principles. External to his critique of Rawls, though, Cohen does believe that luck equality is the correct account of what distributive justice is, and I’m sure he would also insist that luck equality applies to the uncoerced choices citizens make. In fact, I myself use luck equality as an evaluative tool to explain why it would be bad to apply an unclear distributive regulatory principle to the personal context. Aside from the fact that the purpose of a regulatory principle is to guide action, and that a principle’s requirements must be reasonably clear for it to serve that function, I argue that applying an unclear distributive regulatory principle to the personal context exacerbates luck inequality. If better-off citizens are unable to determine what, precisely, a welfare-inclusive, prerogative-constrained difference principle requires of them, the result is that some citizens will do more than is strictly required, whereas others will do less than is required. In other words, lack of clarity results in an unequal distribution of the burdens associated with following a regulatory principle. Furthermore, because this inequality between citizens is the product of an epistemic difficulty that lies beyond their control, the inequality is not one for which they can be held responsible, i.e., it’s an instance of luck inequality.^{Footnote 7} On this basis, I argue that something clearer than Cohen’s version of the difference principle is needed for the personal context, e.g., something similar to the public standard of assistance that Peter Singer defends in The Life You Can Save.^{Footnote 8}

As far as I can tell, there are two matters that one might plausibly think are at issue between Voigt and myself. The first is whether, external to his debate with Rawls, Cohen thinks that regulatory principles should be adopted for the personal context. More specifically, does Cohen think that citizens should adopt and attempt to follow a welfare-inclusive, prerogative-constrained difference principle? Does he think that such a principle strikes a compelling balance between luck equality, efficiency, fairness, community, and citizens’ personal commitments? Or does Cohen think that citizens should refrain from adopting a regulatory principle and instead determine, on a case-by-case basis, how best to weigh the relevant values (luck equality included) in any given situation? Voigt thinks that Cohen, as a radical pluralist, would prefer the latter. For my part, I think it’s unclear whether Cohen, independent of his critique of Rawls, thinks that a qualified version of the difference principle is justified for the personal context. I also don’t know whether he thinks that some sort of regulatory principle should be adopted for the personal context, or whether he would prefer that citizens decide what to do on an entirely case-by-case basis.

For that reason, I’ll move on to the second, more interesting issue, namely whether there’s good reason to adopt a distributive regulatory principle for the personal context. I think there is good reason to do so, and the reason is the same one I use to criticize Cohen’s welfare-inclusive, prerogative-constrained difference principle. As Voigt notes in her commentary, citizens committed to a luck egalitarian ethos should not rely solely on luck equality when deciding what to do. Luck equality is but one value that should be considered, and when it conflicts with, for example, the requirements of community or with citizens’ personal commitments, citizens must each decide for themselves how best to manage that conflict in the particular circumstances they face. As a result, some citizens will do more for the sake of luck equality (and other political values) than other citizens, simply because it isn’t obvious what’s morally required, and this in turn yields unchosen inequality in the distribution of moral burden. Eliminating unfairness in the distribution of moral burden requires adopting a clear, easy to follow regulatory principle of the sort my book recommends.

Narrow Justice, Internality, and Conflicting Values

A second sort of issue my critics raise concerns the manner in which we’re to adjudicate between conflicting values. On the one hand, Smolenski worries that identifying justice as a single, fundamental value exacerbates the complexity of political practical reasoning. How are we to weigh a single, defeasible value against conflicting, defeasible values in a manner that isn’t arbitrary? Colin Macleod, by contrast, worries that, by endorsing the use of procedural devices like Rawls’s original position, I fail to appreciate the extent to which value conflicts can be managed without appealing to procedural fairness. Macleod also notes that my use of the original position seemingly differs from Rawls’s use of it. In my book, the original position functions as a device for non-arbitrarily selecting regulatory principles that reflect a particular balance between conflicting political values. Since other reasonable weightings of those values are possible, some method for making a non-arbitrary choice is needed, and the original position, if successful at eliminating arbitrariness,^{Footnote 9} confers legitimacy upon the principles we’ll use to design our coercive institutions. In Rawls’s work, however, contractors in the original position do not explicitly weigh conflicting values. Instead, they select the principles that, given the limited information available, are most likely to benefit them.

I’ll take this opportunity to clarify what I’m up to. In my chapter entitled “Narrow Justice,” intuitionistic balancing and procedural fairness are combined, after a fashion. I argue that the function of procedural fairness in Rawls’s work, and in much of political philosophy, is to find a non-arbitrary way of striking a balance between conflicting political values. Furthermore, I argue that it’s this procedural solution to value balancing that makes it appropriate for contextualists to locate justice at the level of institutional regulatory principles. Institutional regulatory principles are fully fair, from a contextualist perspective, because they are the output of a fair procedure (rather than merely a balance between substantive fairness and other values). By conferring full fairness, procedural fairness makes it appropriate to think that the selected regulatory principles express what justice is, rather than what’s morally required, all things, including justice, considered.^{Footnote 10}

Though I end up arguing that the contextualist understanding of justice’s conceptual (referential) identity is flawed because procedural fairness cannot coherently confer full fairness, let’s leave that aside for now. Two points are worth noting. First, understanding justice as one value among many does not make political practical reasoning any more complex than it already was. Those who endorse the pluralist understanding of narrowness can make use of the same value-balancing devices contextualists do. We just don’t think that the output of those devices should be called ‘justice.’ Second, adopting a procedural device like the original position doesn’t mean that value balancing becomes a purely procedural matter. Before the original position can be employed, it’s necessary to draw up a list of reasonable options from which contractors are to choose, and the best way to draw up that list is by determining which principles represent reasonable weightings of the political values at stake. The list of principles needn’t be created entirely from scratch: to some extent, we can draw on principles that were put forward and defended at various points throughout the history of political thought. This is the best way to understand the list that Rawls actually uses in A Theory of Justice,^{Footnote 11} and the fact that contractors in the original position don’t explicitly focus on political values does not conflict with my interpretation. The specific question contractors in the original position are interested in is ‘Which of the principles on the list are rational to choose?’ Never at any point do they ask ‘Which principles are rational for us to place on the list?’ The contents of the list are determined prior to the original position, and the fact that they reflect reasonable weightings of our shared political values is what makes them appropriate for inclusion.^{Footnote 12}

In addition to raising a general worry about non-arbitrarily balancing values, Smolenski raises some specific worries about whether a luck egalitarian who thinks of justice as a single, fundamental value can consistently accommodate certain widely shared judgements of justice. One judgement that concerns him is the judgement that protecting minority cultural groups is a particularly important requirement of justice, and I’ll focus on it in particular.

Luck egalitarians have reason to maintain that belonging to a culturally threatened minority group is a source of injustice. Whether one’s born into a majority or a minority culture is a matter of brute luck, and insofar as access to one’s culture is an important good, being born into a minority culture is specifically bad brute luck when the continued existence of one’s culture is uncertain. On one prominent account, the importance of access to culture is derivable from the relationship between culture and autonomy. Enculturation provides one with a context of choice within which one may decide how to live one’s life as one sees fit. Without culture, living an autonomous life is very difficult or impossible, as choice requires that one have an array of culturally furnished options from which to select. If this account is right, then protecting minority cultures is important because doing so protects the autonomy of individuals born into them.^{Footnote 13}

Pointing out that the circumstances of one’s birth is a matter of brute luck fits well enough with luck egalitarianism, but Smolenski worries that there’s a tension between the pluralist understanding of narrowness and an autonomy-based account of culture’s importance. On the view that I defend in my book, justice is a single, defeasible value: specifically, substantive fairness. Other political values matter too, and must be taken into consideration when deciding what to do, but they are external to justice. As such, there are many things that should be done, all things considered, even though justice does not require it, e.g., assisting the imprudent when they’re culpably unable to satisfy their own needs,^{Footnote 14} or refraining from levelling down.^{Footnote 15} In light of this, and with respect to the relationship between autonomy and culture, one might wonder how it’s possible to characterize the issue of minority rights as an issue of justice if the importance of minority rights is grounded in autonomy. After all, fairness and autonomy are different values, and if the importance of protecting minority rights is primarily grounded in autonomy, and not so much in fairness, then mustn’t the pluralist view label group rights a requirement of all-things-considered institutional rightness, and not so much a requirement of justice itself? And isn’t it strange to say that group rights are not primarily a matter of justice? Furthermore, when comparing the importance of protecting group rights to other, seemingly less important matters of justice, e.g., to compensation for unchosen expensive tastes, Smolenski contends that pluralists are unable to consistently claim that group rights are more important from the standpoint of justice. At best, pluralists might say that, in light of non-justice considerations, group rights are more morally important, but they wouldn’t be able to say that failing to protect minority cultures is a greater injustice than failing to compensate expensive tastes. If Smolenski is right, then the pluralist use of the word ‘justice’ is rather out of sync with common usage, not just among political philosophers but in everyday political discourse too.

Smolenski is wrong to claim that the pluralist view cannot account for the special, justice-related importance of group rights. Explaining why he’s wrong requires delving into the nature of value, however. In my book, I distinguish between two kinds of value concept.^{Footnote 16} The first category is ‘goods’ concepts, and it’s comprised of concepts that one might choose to incorporate into the metric of a distributive theory, e.g., resources, liberties, opportunities, subjective welfare, flourishing, etc. What’s distinctive about goods concepts is that they’re ‘good’ for the person who enjoys them, i.e., a person benefits from having more resources, more opportunities, etc. The second category is ‘function’ concepts, and functions indicate what we’re to do with goods. Fairness, efficiency, and respect are all function concepts. Fairness requires that we fairly distribute goods, efficiency requires that we increase the (aggregate or average) amount of goods, and respect places deontic constraints on what we can do with goods, e.g., respect for a person’s property requires that we refrain from stealing it.

Considering the logic of the relationship between them, it’s clear that goods concepts are not external to function concepts. Any theory of fairness, efficiency, or respect, will have to draw upon goods concepts when specifying what it is that we’re to respect, fairly distribute, or efficiently promote. What’s external to a function concept is other function concepts, i.e., fairness, efficiency, and respect are all different, sometimes conflicting values that should be kept separate from each other when discerning their content. Furthermore, were we to categorize it, autonomy would certainly fall under the ‘goods’ category. Autonomy is something of value to the person who enjoys it (it’s good for her), and it’s something that can be increased and indirectly distributed (think of how the distribution of educational opportunities affects the distribution of autonomy). It’s also something that should be respected.

In light of the above, pluralist luck egalitarians can provide a fairly straightforward reply to Smolenski’s worry. Pluralists can acknowledge that cultural protection is of special importance from the perspective of justice for two reasons: (a) autonomy is an especially important good, and (b) securing access to culture is necessary for securing equal access to an autonomous life. If we think that unchosen expensive tastes are unimportant relative to the threat of cultural disintegration, the reason is that having an expensive taste is less disadvantageous than lacking access to one of the necessary conditions for autonomy. The unfairness caused by cultural disintegration is thus more significant than the unfairness caused by unchosen expensive tastes.

Luck Equality at the Procedural Level

One of the novel features of my book is that, in addition to using luck equality at the level of value trade-offs, I suggest that it should be used at the procedural level for purposes of fairly specifying a particular description of the original position. Multiple descriptions of the original position are possible, after all, and if the original position is to succeed at fairly selecting a set of regulatory principles, then we need some means of determining which description is the fairest. Arbitrarily choosing a description of the original position would rob it of its ability to confer legitimacy upon the principles it selects. Citizens subject to the coercive force of the state could reasonably object that the state’s principles fail to justify its use of coercion, as different principles would have been selected had some other description of the original position been chosen. Unless we can justify the claim that our chosen description is the fairest, the original position won’t be able to perform its function, i.e., it won’t be able to eliminate arbitrariness and confer legitimacy upon the principles used to design coercive institutions.

Rawls has his own way of specifying a particular description of the original position. He specifically uses the idea of reasonableness to accomplish that task. For example, he argues that reasonableness requires placing knowledge of citizens’ talents and social positions behind the veil of ignorance, as it would be unreasonable for people to accept proposed principles because they were “dominated or manipulated, or under the threat of an inferior social position.”^{Footnote 17} Though reasonableness goes some way towards specifying a particular description of the original position, my book argues that there’s still some indeterminacy. In particular, reasonableness is indeterminate between withholding knowledge of talents and social positions, and permitting knowledge of those particulars but requiring that contractors bargain in a reasonable manner.^{Footnote 18} In the latter scenario, hypothetical contractors know what their talents and social position are, but they are prevented from coercing or pressuring each other into an agreement, or from offering each other morally arbitrary considerations. This latter scenario is not inconsequential either: reasonable but better positioned contractors have greater leverage, and with knowledge of their superior talents and social position in hand, it would no longer be rational for them to agree to a principle that maximizes the position of the least well-off. I argue that a principle which promises some, but not maximal benefit for the least well-off, would be selected in the latter scenario.^{Footnote 19}

To break the tie between withholding knowledge of talents and social positions, and permitting said knowledge but requiring that contractors be reasonable when bargaining, I argue that substantive fairness, understood as luck equality, is needed. As noted above, better positioned contractors have greater leverage or bargaining power, and said leverage, though not contrary to reasonableness, is contrary to luck equality. One’s talents and social position are largely the product of brute luck, so any bargaining advantage secured though knowledge of those particulars is unfair, from a luck egalitarian perspective.^{Footnote 20} Substantive fairness’s role in eliminating indeterminacy is important for securing the legitimacy of the principles selected, and it’s conceptually important. First, it suggests that the output of the original position, though legitimate, cannot possess full fairness, as its requirements deviate somewhat from the principle of fairness that the original position itself depends upon. Using substantive fairness at the procedural level thus preserves the claim that the original position’s output is not justice itself, but is rather what’s institutionally required, all things, including justice, considered.^{Footnote 21} Second, employing substantive fairness at the procedural level tangibly connects it to the notion of legitimate coercion, thereby supplying a particular way in which we can understand it as the ‘first virtue of institutions.’ Though substantive fairness lacks the normative weight of an institutional regulatory principle, it’s nonetheless needed to specify the fairest description of the original position, and thus to confer legitimacy upon the principles selected.^{Footnote 22} In combination, these points provide powerful support for the claim that justice is one value among many, rather than institutional rightness.

My commentators raise a number of worries about my use of substantive fairness at the procedural level. First, Louis-Philippe Hodgson is doubtful that the main issue at stake between pluralist and contextualist understandings of justice is whether fairness is primarily procedural. He notes that Rawls has his own, non-luck-egalitarian but seemingly substantive understanding of fairness: fairness as reciprocity. Furthermore, he notes that Rawls, via his use of reflective equilibrium, directly applies the idea of reciprocity to the principles his procedure selects. Accordingly, Hodgson is doubtful that employing substantive fairness at the procedural level, or evaluating the output of the original position in light of it, has the conceptual implications I claim it does. The issue, if there is one, is whether reciprocity or luck equality provides the better understanding of substantive fairness, rather than whether substantive fairness is more fundamental than procedural fairness.

Before I reply to Hodgson, a clarificatory point is in order. Though Rawls often uses the word ‘fairness’ to refer to reciprocity, the concept I refer to as ‘substantive fairness’ throughout my book is a rather different concept. Substantive fairness, as I understand it, is at bottom a distributive concept. It’s about who has what and why. Luck equality is a particular conception of it, but other conceptions, such as desert, are possible too. To be a conception of substantive fairness, though, a theory must primarily be concerned with the evaluation of distributions, and it must do so in light of particular information about a distribution’s individual shareholders, e.g., in light of information about the relative size of their shares and/or about the relative deservedness of each shareholder.^{Footnote 23} As a result, substantive fairness is also essentially a comparative concept. Meeting its requirements involves comparing shareholders and ensuring that, relative to one another, the size of their share is appropriate.

Reciprocity, by contrast, is not primarily distributive. Reciprocity, as Rawls understands it, is a social relationship between citizens, and it allegedly obtains when the principles that regulate shared institutions are mutually acceptable to all, and when all citizens engaged in social cooperation adequately benefit from that cooperation. In other words, Rawls thinks of reciprocity as a combination of mutual acceptability and mutual benefit. Though mutually acceptable regulatory principles that also secure mutual benefit will, of course, have distributive consequences, the ideas of mutual acceptability and benefit are not, in and of themselves, distributive ideas. Mutual acceptability is a justificatory requirement. It requires that regulatory principles be interpersonally justifiable. And mutual benefit, though a comparative concept, is about comparing what citizens have to what they would otherwise have under other circumstances, e.g., in the state of nature or under alternative institutional arrangements. Mutual benefit fails to obtain when a citizen has less than what she would have had under “a suitable benchmark of comparison,”^{Footnote 24} rather than when she has less than other citizens (though inequality and the absence of mutual benefit can, of course, coincide).

In light of the difference between substantive fairness and reciprocity, I don’t think we can infer much about my book’s main claim from the fact that Rawls appeals to reciprocity when designing the original position. Additionally, I think we should take note of an important difference concerning the relationship between, on the one hand, a fundamental principle of substantive fairness and distributive regulatory principles, and on the other hand, reciprocity and distributive regulatory principles. Distributive regulatory principles, because they represent a compromise between substantive fairness and other political values, always fall somewhat short of substantive fairness. In other words, from the perspective of substantive fairness, a regulatory principle such as the difference principle will always contain a degree of unfairness. However, optimal distributive regulatory principles do not fall short of reciprocity, as reciprocity is not one of the values that the original position adjudicates between. Instead, reciprocity is a criterion used to determine what the appropriate balance of political values is: the appropriate balance is one that’s mutually acceptable and mutually beneficial. Though a society’s institutions may themselves fall short of reciprocity by falling short of the difference principle, the difference principle is not itself an imperfect realization of reciprocity (assuming that the difference principle is indeed the principle citizens would adopt in the original position).^{Footnote 25} As such, evaluating the difference principle from the perspective of substantive fairness is importantly different from evaluating it from the perspective of reciprocity. The difference principle, if the output of the original position, simply is what reciprocity requires. But from the perspective of substantive fairness, the difference principle is only what’s morally required, all things, including substantive fairness, considered. And if a deviation from fairness is also a deviation from justice, it follows that the difference principle cannot be what distributive justice is.

Macleod also raises worries about my use of substantive fairness at the procedural level. He specifically worries that distributive principles don’t have the right form to be directly applied to procedures. After all, distributive principles are specifically meant for distributions. Though using a distributive principle to indirectly assess a procedure is easy enough, i.e., one can use the principle to assess the distributive effects of the procedure’s output, using it to directly assess the procedure is puzzling. Additionally, Macleod wonders why specifically substantive fairness should be used to specify a fair contract situation. If we can apply substantive fairness at the procedural level, then can’t we apply other values too? And if we can apply other values, shouldn’t we do so?

The answer to Macleod’s worries lies in the conceptual relationship between substantive fairness and procedural fairness. Though these two concepts are not identical, I don’t think, upon reflection, that they’re entirely separable either. More specifically, any kind of fair procedure is at least partially constituted by a fairly distributed good. With respect to the original position, the relevant good is bargaining power. The original position is a fair contract situation in part because it involves a fair distribution of bargaining power among contractors. With respect to democratic procedures (a subject I discuss at length in Chapter 6), the relevant good is voting power. Democratic procedures are a fair way of legislating laws in part because citizens are given an equal say over political outcomes.^{Footnote 26} With respect to procedures like flipping a coin or drawing straws, the relevant good is whatever likelihood the chance of winning represents. Flipping a coin is a fair way of determining who gets to, for example, take the first turn in a board game because flipping a coin fairly distributes the likelihood of going first.

Understanding procedural fairness this way presupposes a distinction between procedural goods and non-procedural goods. Substantive fairness pertains to the distribution of goods in general, and procedural fairness specifically pertains to the distribution of procedural goods. One property that distinguishes procedural goods from (at least many) non-procedural goods is that procedural goods are entirely comparative. A person can have a better, equal, or lesser chance of taking the first turn than another person; and a person can have a greater, equal, or lesser amount of bargaining power than another person. There is no such thing as absolute bargaining power or absolute chance, though. To say that I have a lot of bargaining power in a particular situation is the same thing as saying that those with whom I’m bargaining have little bargaining power, and saying that I have a high chance of taking the first turn is the same thing as saying that my opponent has a low chance. Non-procedural goods aren’t like this. Though terms like ‘rich’ and ‘poor’ are often used in a comparative manner, it’s perfectly coherent to speak of ‘absolute poverty’ or cardinal (and not just ordinal) increases in a person’s level of wealth. Similarly, we can talk about a person’s utility or welfare level in absolute/cardinal terms. A term like ‘happy’ doesn’t just mean ‘happier than others.’

If any fair procedure is at least partially constituted by a fairly distributed good, then there’s nothing odd about directly applying a distributive principle to a procedure. Distributive principles are for assessing distributions, and procedures are partially constituted by distributions. My account of procedural fairness also helps show why substantive fairness is better suited than most other political values are for specifying a fair procedure. If I’m right that any fair procedure is partially constituted by a fairly distributed good, then there’s an analytic connection between procedural fairness and substantive fairness. Put another way, a fair procedure is, by definition, substantively fair. Additionally, though, I think that the understanding of procedural fairness sketched above suggests that at least some political values cannot intelligibly be applied to it. For example, efficiency, or at least the sort of efficiency I discuss in my book, is primarily about absolute amounts. Depending on one’s conception of efficiency, institutions and distributions are efficient when they maximize either the aggregate amount of something, the average amount of something, or when no Pareto improvements are possible. Since procedural goods are strictly comparative, though, efficiency is an inapplicable concept.