2. Anderson’s account

In her reflections on the theoretical differences between luck egalitarianism and relational egalitarianism Anderson picks out three substantive disagreements between social relations and luck egalitarians which (in a sense I shall indicate directly) she takes to be secondary. First: ‘The two camps disagree about how to conceive of equality; as an equal distribution of non-relational goods among individuals, or as a kind of social relation between persons – an equality of authority, status, or standing’ (Anderson Reference Anderson2010, 1). For instance, because ‘the freedom to appear in public as who they [homosexuals: KLR] are, without shame or fear of violence, the right to get married and enjoy benefits of marriage, to adopt and retain custody of children’ is not ‘a divisible, privately appropriated [good], such as income and resources, or privately enjoyed [good], such as welfare’ (Anderson Reference Anderson1999, 288), luck egalitarians ignore this good while it is crucial to social relations egalitarians. Call this the currency disagreement.

Second, this disagreement ‘generates’ another disagreement. Luck egalitarians think that ‘inequality is unjust when it is … caused by morally arbitrary factors’ whereas relational egalitarians take inequality to be unjust ‘when it reflects, embodies, or causes inequality of authority, status, or standing’ (Anderson Reference Anderson2010, 2). As a result of being preoccupied with responsibility, luck egalitarians tend to focus on issues such as whether people deserve compensation for disadvantages caused by the frustration of their involuntarily developed expensive tastes for fine wine and plovers' eggs – questions that are far removed from the concern of real-life egalitarians. Call this the arbitrariness disagreement.

Third, the two camps also disagree about the site of justice, i.e. that to which requirements of justice apply. Luck egalitarians think justice ‘consists in a desirable distributive pattern’, whereas relational egalitarians ‘identify justice with a virtue of agents (including institutions)’ (Anderson Reference Anderson2010, 2). Accordingly, for luck egalitarians justice may or may not obtain in a world of solitary Robinson Crusoe-like individuals without social relations, whereas for social relations egalitarians it is only in a social setting that the question of justice arises. Call this disagreement the site disagreement.

These issues are secondary because their ‘source’ (see below) is a more basic disagreement about:

Let us call this methodological issue, which is the fundamental one for Anderson, the justification disagreement.

As already indicated, there are significant insights to be gained from Anderson’s critique of luck egalitarianism (I say more about these in the last section). However, her work on the luck/relational divide does neglect important differences within the luck egalitarian position, and also within the relational egalitarian position; and on occasion she connects claims that can be seen when examined closely to be independent of one another (Lippert-Rasmussen Reference Lippert-Rasmussen, Minar and Lippert-Rasmussen2012). Specifically, I would deny that the substantive disagreements which have surfaced in the discussion between luck egalitarians and relational egalitarians reflect a radical underlying disagreement over justification. If I am right, social relations and luck egalitarianism do not have distinct sources, or foundations, whose incompatibility rules out the pluralist egalitarian position described in the previous section.

One important preliminary point: ‘underlie’ (Anderson Reference Anderson2010, 4), ‘generate’ (Anderson Reference Anderson2010, 1, 5–6) and ‘source of’ (Anderson Reference Anderson2010, 3) are Anderson’s own organizational terms, and obviously a great deal hangs on what she means by them. I will interpret them as follows:

1. (i) P and Q are inconsistent;

2. (ii) P^* and Q^* are inconsistent;

3. (iii) P entails P^* and Q entails Q^* but not vice versa;

4. (iv) P^* and Q^* are, respectively, suitably related to P and to Q;Footnote ² and

5. (v) those who accept P^* do so because they accept P (likewise Q^* and Q).Footnote ³

It is possible, of course, to construe Anderson’s organizational terms causally. The suggestion would then be that, however the propositions involved in the four disagreements are related logically, or conceptually, disagreement over the substantive secondary issues is caused by radical divergence in our assumptions about justification. On this interpretation Anderson’s article is not a contribution to the ongoing philosophical debate between luck and relational egalitarians, but an account of the sort provided by historian of ideas identifying a certain structure in the beliefs of various political philosophers subscribing to luck egalitarianism or social relations egalitarianism. Admittedly, this interpretation makes better sense than the entailment interpretation (and, for that matter, the coherence interpretation which is explained in the next paragraph) of Anderson’s remark that most relational egalitarians subscribe to the second-order view of justification. This remark becomes puzzling on the entailment interpretation. If some relational egalitarians do not adopt this standpoint and nevertheless disagree with luck egalitarians on the three other matters of contention that she mentions – and if they do not, why think of them as relational egalitarians? – then, presumably, the source of the disagreement between luck egalitarians and these relational egalitarians is not over the standpoint from which principles of justice are satisfied. However, this has to be weighed against the fact that on the causal interpretation Anderson could not claim to describe a difference between luck egalitarianism and social relations egalitarianism as such. She could only claim to be describing a difference between the actual views of some luck and social relations egalitarians – however narrow they are relative to the logical space plausibly afforded to the luck (and, for that matter, the social relations) egalitarian perspective. Most importantly, for present purposes, her analysis would pose no threat to my pluralist account of egalitarianism, of which luck egalitarianism is one component, since no amount of argument against the assumptions other luck egalitarians make without strictly being required to do so as luck egalitarians could impugn luck egalitarian views, including pluralist egalitarianism, as such. Hence, while I believe that, exegetically speaking, the entailment interpretation is the one with most going for it, the pluralist egalitarian option reminds us that there are also non-exegetical reasons to adopt that interpretation rather than the causal one.

Another interpretative option is to weaken (c) so that it claims instead that P and P^* form a more coherent set of propositions than P and Q^*, and Q and Q^* form a more coherent set of propositions than Q and P^*. It less easy to dismiss this construal, but I shall nevertheless do so. For one thing, this makes Anderson’s claims stronger, more perspicuous, and thus more interesting. But also the equivocal terms at issue here – ‘source’, ‘underlie’ and ‘generate’ – are not suggestive of the kind of reciprocal adjustment of different kinds of belief that is generally taken to be involved in coherence and reflective equilibrium. I would add that most of my challenges are as effective even if the coherence interpretation is correct.Footnote ⁴

3. Second-versus third-person justification

As we have seen, Anderson thinks the fundamental disagreement between luck egalitarians and relational egalitarians hinges on the difference between a second- and a third-person perspective on justification. Unfortunately, however, she offers non-congruent characterizations of the two perspectives. Describing the third-person conception of justification, she writes: ‘In a third-person justification, someone presents a body of normative and factual premises as grounds for a policy conclusion. If the argument is valid and the premises are true, then the conclusion is justified’ (Anderson Reference Anderson2010, 2). Given this, one assumes that in second-person justification it is insufficient, to justify a policy conclusion, simply to present a valid argument for it invoking only true normative and factual premises – that more is needed. However, this is not what Anderson says. Instead she ties her characterization of second-person justifications to ‘vindicating claims on others’ conduct' (Anderson Reference Anderson2010, 3). This unexpected turn raises two problems at least.

First, we seem now to have four, not two, conceptions of justification to consider. Call justifications that conform to Anderson’s characterization of third-person justification ‘context-insensitive’ and those that do not ‘context-sensitive’. Call justifications meeting Anderson’s characterization of second-person justification ‘conduct claim dependent’ and those that do not ‘conduct claim independent’. These certainly appear to cross-cut, giving us the following theoretical matrix: (Table 1)

Table 1 Conceptions of justification.

As I say, this flows from Anderson’s conflation of two rather different issues. (a) and (b) are those Anderson would presumably be most likely to recognize: (a) is, as it were, double-defined as the third-personal sort of justification, and (b) is double-defined as the second-personal sort of justification. (c) and (d) are (for Anderson, again presumably) unwelcome hybrids.

Take first the context-insensitive, conduct-claim dependent combination. According to this, a distribution of goods, say, is unjust only if the following two conditions are met: (1) the relevant distribution reflects the fact that some people do not comply with the demands others can reasonably make on their conduct, and (2) the fact that these demands are reasonable can be given a context-insensitive justification. On this view, which may well be the position of Dworkinian luck egalitarians, unavoidable, irremediable inequality is not unjust, because it fails to satisfy (1).

Take next the context-sensitive, conduct-claim independent combination. On this view, something might be unjust even if (1) the relevant object of injustice does not reflect the fact that some people do not comply with the demands others can reasonably make on their conduct, and (2) it can be given a context-insensitive justification. On this view, unavoidable, irremediable inequality may or may not be unjust. While it does not reflect the failure of some people to comply with the demands others can reasonably make on their conduct, it may be that it is incapable of being given contextrelative justification. This might be so if some people (reasonably) reject the view that natural inequalities are unjust. Given that they hold such a view, and given that one cannot justify a certain view by appealing to a principle that one (reasonably) rejects, they (as opposed to people who subscribe to no such principle) cannot justify, or demonstrate, to others the injustice of such inequalities. As Rawls (Reference Rawls2000, 190) puts it: ‘A person’s right to complain is limited to violations of principles he acknowledges himself’. It is also arguable that one does not have authority, in Darwall’s sense, to hold someone else to account for failing to comply with a principle one does not accept oneself (Darwall Reference Darwall2006, 8).

Second, it is striking that in her actual arguments against luck egalitarianism Anderson nowhere offers an example of a justification that is context-sensitive. True, she offers examples of justifications that she thinks make unreasonable demands on some people, but these, if they fail, fail from a third-person perspective as well. Take her discussion of Salieri’s complaint that Mozart has a greater natural talent than he does (Anderson Reference Anderson2010, 10). Anderson thinks that the claim that the relevant inequality here is unjust cannot be established from a second-person point of view. However, she nowhere shows why any justification for the inequality would be context-sensitive. Instead, what she does is assert that it ‘is unreasonable to demand that people do things beyond human capacities’ (Anderson Reference Anderson2010, 10). However, an argument for the conclusion that there is no injustice in Mozart’s musical talents being greater than Salieri’s appealing to this premise comes across no differently in a third-person context – say, Brahms uttering the argument to Tchaikovsky – from the way it comes across in a second-person context – say, Mozart uttering the same argument to Salieri and appealing to the presumed fact that it is beyond human capacity to prevent his musical talents being greater than Salieri’s. The idea of context-relative justification is irrelevant to its assessment.Footnote ⁵ If it is sound, it can be employed by anyone to justify to anyone that the unequal distribution of natural musical talents across Salieri and Mozart is not unjust.

In her earlier work, Anderson also employs the ‘principle of interpersonal justification’ to rebut certain principles – not arguments, as happens in her later work – that she ascribes to luck egalitarians (Anderson 199, 322–325). As with the Mozart-Salieri argument, none of these involve context-sensitivity of the kind apparently rendering the relevant principles justified when formulated in the second-person, but not in the third-person, and vice versa. For instance, Anderson contends that the view that firefighters who voluntarily engage in ‘dangerous occupations’ from which consumers benefit have no claim to assistance from consumers fails to be interpersonally justified: ‘The principle “let us be served by occupations so inadequately compensated that those in them shall lack the means necessary to their freedom, given the risks and conditions of their work”; cannot survive the test of interpersonal justification’ (Anderson Reference Anderson1999, 323). I agree that this principle is implausible, but in my view it is as implausible in (pure) third-person form (i.e. ‘let people in non-dangerous occupations be served by occupations so inadequately compensated that those in them shall lack the means necessary to their freedom, given the risks and conditions of their work’) as it is in second-person form.Footnote ⁶

Interestingly, the interpersonal test which Cohen – someone who, qua luck egalitarian, is supposed to take a third-person perspective on justification – employs shows that context-variation plays an essential role in his critique of the incentive argument for inequality. In that critique Cohen argues that an unequal distribution is unjust precisely because, while it can be justified in some contexts (e.g. where untalented people are justifying to other untalented people the need for incentives for talented people), it cannot in other argumentative contexts (e.g. where talented people are attempting to justify, to untalented people, the idea that talented people should be given incentives) despite its validity and the truth of the premises appealed to. The asymmetry arises because it is the talented people who make it true that incentives are necessary for making the worst off as well off as possible: while non-talented people can see the incentive-seeking behavior of talented people as something that must be taken as given when determining what justice requires, talented people themselves must regard this behavior as something which they should abstain from in the light of their embrace of the difference principle. They cannot simply say: ‘Don’t complain about our insistence on incentives. The fact is that we do insist on them.’

It is unclear if this critique reflects non-luck-egalitarian commitments of Cohen’s – e.g. his commitment to a certain kind of justificatory community whose relation to luck egalitarian justice is unclear. (I return to the endorsement of community in the last section.) Hence, given the entailment interpretation of Anderson’s ‘source’, not too much weight can be put on this observation. Still, exegetically, it is an odd feature of Anderson’s distinction that her characterization of third-person justification, and therefore her implicit characterization of second-person justification, appears to have been adopted in large part from Cohen (Reference Cohen2008, 35–41, 64–68), not Darwall.Footnote ⁷ Indeed, on the causal interpretation of Anderson’s ‘source’ claim outlined in Section 2, the present point amounts to a decisive objection to this claim.Footnote ⁸ Substantially, it is an odd feature that something very much like the ideal of second-person justification canvassed by Anderson – who, like other social relations egalitarians, is not concerned with socio-economic inequalities as such – is being used by a main contender of luck egalitarianism to criticize the only deviation from equality often believed to be allowed by Rawls' difference principle.

In view of these considerations, I propose that we understand Anderson’s notions of second- and third-person justification as follows:

Within the family of second-person conceptions of justification various accounts of the claims we can reasonably make on one another’s conduct are on offer. For instance, on a Scanlonian account we cannot reasonably make claims on each other on the basis of impersonal reasons (Scanlon Reference Scanlon1998, 218–223). Such an account might cohere well with Anderson’s rejection of the possibility of purely natural, unjust inequalities. However, her preferred account is, so she writes, Darwall’s (Reference Darwall2006) (though, strictly speaking, Darwall’s account is an account of second-person reasons, not second-person justification); and in any case my reconstructed relational egalitarian position on the justification disagreement has a similar feature. Here I address Anderson’s Darwallian account of the claims one can reasonably make on others as she presents it, and I shall not try to vindicate her map of the disagreement between luck and relational egalitarians by constructing a Darwallian or Scanlonian account of the justification disagreement that is independent of her own description. But for the record, with the stated exception, discussed in Section 4, I doubt such an endeavour would help. One reason is this. Darwall (Reference Darwall2006, 28) notes that if there is, say, an obligation to protect ‘cultural treasures’ independently of the way this would promote the ‘needs and interests of free and rational individuals’ – something which he does not take a stand on – then such protection is something ‘free and rational individuals have the authority to demand of one another’. Evidently, then, Darwall thinks the very idea of second-person reasons is compatible with having reasons to promote what is impersonally valuable.

If the reconstructed justification disagreement is indeed the radical disagreement Anderson regards as the gravamen of the debate, the question becomes: Does it, as she maintains, generate the three secondary issues over which luck and relational egalitarians vie. I turn to this question now.

4. Source of the arbitrariness disagreement?

There are at least four ways in which one can show that the justification disagreement is not, as Anderson claims, the source of the three substantive disagreements: (1) by showing that it is possible to take the luck egalitarian position on one or more of the substantive disagreements while adopting a second-person perspective on justification; (2) by showing, analogously, that it is possible to take a relational egalitarian position on one or more of the substantive disagreements while adopting a third-person perspective on justification; (3) by showing that, without inconsistency, one can combine luck egalitarian and relational egalitarian positions on at least one of the substantive issues;Footnote ⁹ or (4) by showing that an unmixed set of positions on the three substantive issues (whether luck egalitarian or relational egalitarian) is inconsistent.

Section 7 explores (3) and (4), while this and the following two sections concentrate on three versions of (1) and, much more briefly, (2). Sections 5 and 6 argue that (1) succeeds, respectively, with respect to the currency and the site disagreement. In this section, I concede that (1) fails, strictly speaking, in the case of the arbitrariness disagreement, but that it does so in a way that undermines other parts of Anderson’s position.

Let us distinguish, then, between two forms of inequality deriving from bad, brute luck. The first I shall refer to as pure natural inequality. This inequality would have obtained regardless of what any human being did. Contrast this with impure natural inequality. This kind of inequality is a result of differential bad brute luck, but it is such that it could have been avoided had some agents acted differently or/and it can now be eliminated if some agents act in a certain way.

Now, the notion that somebody ‘could have exercised her agency differently’ is a tricky one. Strictly speaking, any inequality between two people is likely to be an impure natural inequality, since, had some of their ancestors not conceived when they did, this pair of individuals would not even have existed, and as a result the inequality would not have obtained. Should we therefore modify the condition pure inequalities must satisfy, e.g. by saying that they are pure where no one could successfully have exercised their agency with the intention of preventing the relevant inequality? This would broaden the scope of pure, natural inequalities. It would mean that the inequality between persons born with congenital untreatable and uncompensable diseases and people born without qualify as purely natural despite the worry about conception mentioned above. However, for present purposes I can set aside the complicated issues of where exactly to draw the line between purely and impurely natural inequalities and between natural inequalities and inequalities that are not natural in the relevant sense, i.e. social inequalities.Footnote ¹⁰

The key point is that very many inequalities with perfectly natural-looking causes (e.g. those arising from difference in musical talent between Mozart and Salieri) are not, or do not give rise to, purely natural inequalities in the present sense. As Anderson notes, Mozart could have crippled his musical talents (which, however, would not have eliminated the inequality in the value of opportunity prior to leveling down).

If we apply this insight to the arbitrariness disagreement, we see that Anderson misconstrues the implications of the views involved in the justification disagreement. Suppose we accept the relational egalitarian stance on the reconstructed justification disagreement. That is, we say that if something is unjust it reflects the fact that some agent has failed to comply with claims on her conduct that others could reasonably make. Assuming that no one could have conducted themselves in such a way that the relevant inequality in natural musical talent would not have arisen, this inequality cannot be unjust. This point, however, has little significance, because no luck egalitarian believes that the – or, more modestly, a – relevant equalisandum is natural musical talent. They might think resources overall are the equalisandum, but nothing in the story of Mozart and Salieri suggests that they cannot be equal in resources overall.

Anderson might respond that the demand involved in leveling down – in casu that Mozart cripples his musical talents – is malicious, and that no one can reasonably make malicious demands of others (cf. Anderson Reference Anderson2010, 9).Footnote ¹¹ However, this response is ineffective in the present context, because it does not follow from a second-person conception of justification alone that justice does not require leveling down.Footnote ¹² In the case where leveling down is possible, there is someone who could conduct themselves in such a way that equality is realized, and nothing in the reconstructed second-person conception of justification as such rules out the idea that a demand, made of their conduct, to the effect that they realize equality by giving up benefits that as a result are simply eliminated is unreasonable. Now, Anderson thinks that a demand to this effect is unreasonable. But, first, luck egalitarians can agree that it is an unreasonable demand when it is taken to express a rule of conduct, but deny that this casts doubt on the claim that it would make things fairer. Second, people who subscribe to a second-person conception of justification can obviously hold different views on the sorts of demand one can reasonably make on others; and it is certainly possible to hold that for something to be unjust (a) there has to be someone who can prevent (or could have prevented – I return to this below) the relevant unjust state of affairs from obtaining, and (b) that it is not unreasonable to demand of a better-off person with a good fortune that she levels herself down. Accordingly, even people who share Anderson’s second-person perspective on justification may disagree over whether the inequality between Mozart and Salieri, and similar inequalities involving remediable differential brute luck, are unjust. This means the arbitrariness disagreement, as Anderson construes it, does not derive from the reconstructed justification disagreement.

Anderson might reply that her view of the demands one can reasonably make is clearly superior to any that would allow demands for leveling down. But, first, she does not set out a general account of the kinds of demand one can reasonably make on others. Instead she gives examples (involving maliciousness and acts that surpass human capacity), but it is unclear what, if anything, unifies these. She says that a demand on another has to involve an ‘interest’, but it is unclear why one cannot have an interest deriving from valuing some state of affairs considered impersonally valuable, such as equality (cf. Scanlon Reference Scanlon1998, 222). This makes it hard to assess whether Anderson’s view on reasonable demands is superior to competitors. Second, as a pluralist egalitarian I do not think leveling down really is unreasonable even where it would not be morally wrong not to level down all things considered.

Admittedly, there is one disagreement, or difference of focus, Footnote ¹³ between friends of third- and second-person conceptions of justification which remains unaffected by the argument so far: namely, whether inequalities that no one can do, or ever could have done, anything about can be unjust. It is open to friends of a third-person conception of justification to say that such pure natural inequalities are unjust (although they need not do so), whereas it is not open to friends of a second-person conception to do so. For more than one reason, friends of third-person conceptions of justification need not claim that pure natural inequalities are unjust. To start with, they might not be egalitarians. More importantly, however, the fact that a state of affairs of inequality can be unjust even if no one could reasonably have avoided it does not entail that it is unjust. Hence, even the case of disagreement over pure natural injustice is not one that neatly sorts luck egalitarians into one group and relational egalitarians into another, opposing group. Deontic luck egalitarians, for instance, do not object to purely natural inequalities.Footnote ¹⁴

Because luck egalitarians can acknowledge the injustice of such inequalities and relational egalitarians cannot, there is a disagreement between them. The problem is that this disagreement does not have the shape Anderson takes it to have; nor does it pertain to what one should do, since ex hypothesi the inequalities in question are such that no one can do, or ever could have done, anything to avoid them. Accordingly, the disagreement is purely axiological and not one that bears on action-guidingness. For this reason, the disagreement seems much less significant than an earlier one Anderson ascribed to luck egalitarians and relational egalitarians, indicting the former for ignoring the concerns of real-life egalitarians (Anderson Reference Anderson1999, 288).

5. Source of the currency disagreement?

I have conceded that the justification disagreement is the source of a version of the arbitrariness disagreement, at least, as far as pure natural inequalities are concerned, but I have also argued that pure inequalities of this sort are not the broader arbitrary inequalities whose assessment separates luck egalitarians from relational egalitarians. In this section I argue, albeit in a little less detail, that the justification disagreement is not the source of the currency disagreement – in other words, it is not at the root of the debate over whether equality concerns the ‘distribution of non-relational goods’ or a certain ‘kind of social relation between persons – an equality of authority, status, or standing’ (Anderson Reference Anderson2010, 1).

Note, first, that Anderson’s characterizations of the rival currencies here are again oddly non-congruent. Some relational goods, i.e. goods one’s possession of which depends on one’s relation to others, are natural – say, the good of being taller than others. Some relations between persons, such as the fact that we live higher up the mountain than you do, are not social. So Anderson’s distinction cannot be exhaustive. People who hold one of the following three views are neither relational egalitarians, nor luck egalitarians, on Anderson’s account: (1) that justice is an equal distribution of relational goods; (2) that justice is a certain natural relation between persons; or that (3) justice is an equal distribution of non-relational goods and a certain kind of social relation between people. Given that luck egalitarians are concerned with the distribution of non-relational goods (e.g. welfare, or resources), one would expect relational egalitarians to be concerned with the distribution of relational goods; but this is not the concern Anderson attributes to relational egalitarians. Perhaps, in a sense, she could make this very attribution. But if this is so, it is hard to see why luck egalitarians should not be concerned with the distribution of relational goods.Footnote ¹⁵

Take, for instance, G. A. Cohen’s formulation of egalitarianism: ‘its purpose is to eliminate involuntary disadvantage, by which I (stipulatively) mean disadvantage for which the sufferer cannot be held responsible, since it does not appropriately reflect choices that he has made or is making or would make’ (Cohen Reference Cohen2011, 13). With this formulation it is an entirely open question whether involuntary disadvantages include disadvantages in relational goods, such as having a lesser authority, status, or standing than others. Admittedly, the currencies of equality that Cohen discusses are welfare, resources, and capabilities. However, the fact that he does not discuss relational goods as a currency of egalitarian justice has nothing to do with his luck egalitarianism – his ‘right reading of egalitarianism’. What defines luck egalitarianism is a commitment to responsibility-catering equality.

To propose that the currency disagreement separates luck egalitarians from relational egalitarians on this ground is like proposing, on the basis of a reading of Singer and Hare, that consequentialists and deontologists disagree over the claim that it is only welfare maximization that matters morally. True, both are consequentialists, and both think welfare maximization is all that matters morally, but many consequentialists maintain that things other than welfare are non-instrumentally valuable, and so construing the disagreement between consequentialism and deontology on this basis is confusing. Of course, Anderson might be right that the equalisanda proposed by Cohen and other luck egalitarians are deficient because they leave out important dimensions, but this is a criticism of luck egalitarian theorists, not luck egalitarianism as such. Equally, she could deny that the egalitarian social relations with which Cohen and others are concerned are valuable only as part of a greater pattern of distributive equality. This may be right, but it would not count as an objection to my pluralist egalitarian view.

At this point some might object that these considerations show at most that luck egalitarians are not committed to Anderson’s particular luck egalitarian position on the currency disagreement; they do not establish that the currency disagreement does not derive from the justification disagreement.Footnote ¹⁶ In support of this Andersonian source-claim the following consideration can be offered. Whenever objectionably inegalitarian social relations exist someone must have failed to comply with claims on their conduct that others could reasonably make. However, this is not true of unequal distributions of non-relational goods. Arbitrary inequality of welfare between, say, non-interacting Robinson Crusoes could arise through differential bad brute luck.

However, this contrast between inegalitarian social relations and inegalitarian distributions of non-relational goods is false. First, people might comply with all reasonable demands and yet, together and unintentionally, bring about objectionably inegalitarian social relations. Second, it could be true – I am not saying that it is – that we are genetically hardwired to bow and scrape before the stunningly beautiful and intelligent. If that were so it would be beyond human capacity not to bring about the objectionably inegalitarian social relations this involves and accordingly, in the light of Anderson’s own view, no one can reasonably require others not to bring them about. Hence, the focus on reasonable claims on others' conduct does not motivate a focus of social relations.Footnote ¹⁷ I conclude that the currency disagreement does not derive from the justification disagreement.

6. Source of the site disagreement?

Consider, finally, whether the site disagreement derives from the justification disagreement. Anderson takes the site disagreement to consist in the fact that whereas luck egalitarians hold that justice ‘consists in a desirable distributive pattern’, relational egalitarians ‘identify justice with a virtue of agents (including institutions)’. There are at least two reasons for denying that this disagreement is an upshot of the justification disagreement. First, nothing in the third-person conception of justification rules out a deontic conception of equality. Take, for instance, the view that justice obtains whenever agents do whatever they reasonably can to minimize the degree to which some people have less welfare than others through no choice or fault of their own. This view, focusing as it does on neutralizing differential luck, is luck egalitarian. It is incompatible with Anderson’s luck egalitarian position on the site issue. It also concerns the non-relational good of welfare. Yet it is perfectly compatible with a third-person conception of justification, which, after all, only says it is not a necessary condition of injustice that some agent has failed to comply with claims on her conduct that others could reasonably make. The third-person conception of justification does not say that a correct view of justice will always accommodate the idea that something can be unjust even if no agent has failed to comply with claims on her conduct others could reasonably make.Footnote ¹⁸ Moreover, the view that ‘being unjust’ is a property of distributions, but one that those distributions can have only if some agent has failed to comply with claims on her conduct that others could reasonably make, is compatible with a second-person conception of justification, yet it falls on the luck egalitarian side of the site disagreement.

Second, consider saintly virtues like the virtue displayed in selfless devotion to benefiting the worst off. On Anderson’s account we cannot reasonably demand that others possess or manifest these.Footnote ¹⁹ This means that the second-person view of justification as such is neutral on the question whether the site of justice is virtues or distributive patterns, since whichever of these two one favors, justice, as one construes it, might involve unreasonably demanding claims.

I conclude, then, that the justification disagreement is not the source of two of the three substantive disagreements Anderson lists: it does not generate the currency and site issues. I have allowed that it is the source of the arbitrariness disagreement when this is taken, more narrowly than Anderson assumes, to involve the possibility of unjust, pure natural inequalities. I will return to this matter in the conclusion.

7. The substantive issues: combined positions and inconsistent sets

At the beginning of Section 4, I listed four ways to establish that the justification disagreement is not the source of the three substantive disagreements that Anderson identifies. Having examined the first two of these, I turn now to the third and fourth. These were: (3) a demonstration that, without inconsistency, one can combine luck egalitarian and relational egalitarian positions on at least one of the substantive issues, and (4) proof that an unmixed set of positions on the three substantive issues (whether luck egalitarian or relational egalitarian) will contain an inconsistency.

I start with (4), and I want to focus on the relational egalitarian positions on currency and site. On currency, the relational view is that equality (read: justice) is ‘a kind of social relation between persons’ (Anderson Reference Anderson2010, 1). On site, the relational view is that justice itself consists in certain principles ‘that regulate [agents’] conduct' such that if they conform to the principles their conduct is desirable in a certain way.

These two positions are incompatible for the following reason. If justice consists in people acting in conformity with certain principles, it cannot consist in a certain relation, given the following assumptions made in Section 4: (a) that people might act in conformity with the relevant principles and yet create unjust unequal social relations inadvertently – say, because we are genetically hardwired to bow and scrape before the stunningly beautiful or amazingly intelligent people; and (b) that people might violate the relevant principle and yet not create unjust unequal relations – say, because they infringe the relevant principles by seeking to subordinate each other, but are unsuccessful, and therefore the social relations retain their required egalitarian character.Footnote ²⁰

If this is correct, and if the relational stance on the currency and the site issues are incompatible, we have a further reason to deny that the currency and the site disagreements have their source in the fundamental methodological disagreement over justification Anderson identifies. Moreover, the discovery that the luck egalitarian stance on the three substantive issues is a consistent one and the relational egalitarian one is not ought to embarrass Anderson, because that settles the issue, as she construes it, in the luck egalitarian’s favor!

Even if the relational egalitarian stance on the currency and site issues were compatible, it would need to be shown that Anderson’s conception of the debate can survive challenges of type (3). I will now argue that, in one case, at least – that of arbitrariness – it cannot. The luck egalitarian position on arbitrariness is that ‘inequality [in the distribution of non-relational goods] is unjust’ if ‘it is caused by morally arbitrary factors’ (Anderson Reference Anderson2010, 2). The relational egalitarian position is that ‘inequality [in the distribution of non-relational goods] is unjust’ if ‘it reflects, embodies, or causes inequality of authority, status, or standing’ (Anderson Reference Anderson2010, 2).

Both positions state sufficient conditions of unjust inequality and, thus, are compatible. But if, as Anderson states them, the luck egalitarian and the relational egalitarian positions on arbitrariness are compatible, the justification disagreement cannot be the radical source of the three substantive disagreements. Admittedly, this problem can easily be solved by insisting that the ‘if’ in the characterization of the relational egalitarian stance should be read as ‘if and only if’ (or ‘only if’). However, the fact that Anderson’s characterization requires this revision, and the strength of the objections canvassed elsewhere in this article, strongly support the conclusion that Anderson’s mapping of the dispute between luck and relational egalitarians, and in particular her hierarchy of disagreement, is inaccurate.

8. Conclusion

It is not my view – and I would like to emphasize this – that Anderson fails to shed significant light on the debate between luck and relational egalitarians. While I think her mapping of the structure and source of the debate should be resisted, I believe the map itself has considerable value. With suitable warnings, it offers a helpful heuristic to the actual debate between a number of influential luck and relational egalitarian theorists; it is just that it is not an accurate map of the conceptual relations between different theories (Lippert-Rasmussen Reference Lippert-Rasmussen, Minar and Lippert-Rasmussen2012; Reference Lippert-Rasmussen2015).Footnote ²¹ It is striking that Anderson declines to define both luck egalitarianism and relational egalitarianism, Footnote ²² preferring instead to list various claims to which luck and relational egalitarians are committed. If her map of the debate was theory-focused, her job would presumably be to show that the various luck and relational egalitarian stances on the four disagreements with which we have been concerned here follow logically from the propositions defining luck and relational egalitarianism. If, instead, her map serves to depict some views subscribed to by theorists who think of themselves, or are thought of by others, as luck/relational egalitarians, and some issues that have actually surfaced in the philosophical debate between luck and relational egalitarians as it has actually unfolded – whether or not this unfolding accurately tracks the internal logic of the luck egalitarian and relational egalitarian positions – this is not necessary. The limitation of such an endeavor is that in refuting views luck egalitarians have actually held one does not necessarily refute luck egalitarianism as such. Everything turns on whether the accounts presented leave room for one or more variants that have not yet been elaborated. Notably, the approach does not engage the pluralist egalitarian position I espoused above in the first section. The value of the endeavor lies in the fact that it provides a useful overview of the debate and, ideally, pinpoints core issues of contention. As such it can serve as a guide to the points at which further philosophical reflection is required.

One such point, which Anderson’s heuristic identifies, is whether purely natural inequality is unjust. As I have noted in Section 4, the disagreement over whether it is, or can be, unjust, by definition, has no practical significance, because, by definition, the relevant inequalities are unaffected by what we do. However, the issue is important to our understanding of what justice is, and, as Anderson notes, more needs to be done to defend the luck egalitarian contention that the claim that there is something bad about natural inequalities really is a claim about justice and not some other, perhaps axiological, issue. One possibility that should be taken seriously here is that luck egalitarians and relational egalitarians are simply asking different questions (Vallentyne Reference Vallentyne and Kaufman2015). It may well be that luck egalitarians are asking when people’s relative positions are fair, while relational egalitarians want to know what duties we owe each other (to put it the way Scanlon does). Strikingly, this construal suggests that the different ‘answers’ they give do not amount to disagreements, but instead highlight divergent lines of enquiry – lines that may, for all that has been said, complement one another, as they do in my favored pluralist egalitarian account of justice. Interestingly, Anderson (Reference Anderson2010, 16) writes: ‘These points address the third disagreement between luck egalitarians and relational egalitarians. The two sides are asking different questions when they ask what (distributive) justice demands.’ If the two sides are asking different questions, perhaps for that reason the answers they give need not conflict. If I am right, and if Anderson’s suggestion in the passage just quoted is taken seriously, it seems likely that luck egalitarianism and social relations egalitarianism are not committed to distinctive conceptions of justification whose incompatibility rules out the conciliatory project between the two I favor.

It is worth pointing out that a similar conclusion applies to other pluralist accounts of egalitarian justice combining luck egalitarianism with an ideal of egalitarian social relations. I shall briefly mention two. First, it applies to Kok-Chor Tan’s (Reference Tan2012, 127) recent attempt to address Anderson’s concerns through an account of luck egalitarian justice whose site is institutions. This account deflects, among other Andersonian objections, Anderson’s complaint that luck egalitarians will want to compensate unattractive people for their bad luck in ways that inevitably will convey humiliating messages. Second, it applies to G. A. Cohen’s latest work, where he expressed the view that ‘certain inequalities that cannot be forbidden in the name of socialist equality of opportunity [e.g. inequalities resulting from differential outcomes freely engaged in] should nevertheless be forbidden, in the name of community’ (Cohen Reference Cohen2009, 37).Footnote ²³ One possibility, which he suggests, is that the ideal of community ‘define[s] the terms within which justice will operate’ (Cohen Reference Cohen2009, 37). Admittedly, his notions of community – to wit, the justificatory community which we encountered in relation to the interpersonal test in Section 3, community of life experience, and the communal ethos of desiring to serve others and to be served by them though not on a quid pro quo basis (Cohen Reference Cohen2009, 41) – are different from the one embodied in Anderson’s democratic equality and in my pluralist egalitarianism. Still, his position is like my pluralist egalitarianism in giving an account of egalitarian justice that incorporates concerns other than the elimination of differential brute luck.

It seems fitting to end on a note congenial to readers who want to know where some of the fine-grained points above leave us with regard to the concerns of real-life egalitarians. Suppose Pikkety’s worry that increasing inequality in income and wealth threatens democracy is unfounded. More generally, suppose that as a matter of fact inequalities in income and wealth can grow alongside increasingly egalitarian social norms. These suppositions are not idle speculation. Anderson (Reference Anderson1999, 312) mentions ‘racism, sexism, nationalism, caste, class, and eugenics’ as five inegalitarian ideologies that are incompatible with democratic equality. To the casual observer it seems that in the last four decades, which have seen dramatic increases in inequalities in income and wealth, these ideologies have weakened considerably (Lippert-Rasmussen Reference Lippert-Rasmussen, Minar and Lippert-Rasmussen2012, 122–123; cf. Scheffler Reference Scheffler2003, 38). On the assumptions stated, unlike social relations egalitarians, pluralist egalitarians have much to complain about when they consider the way things have evolved distribution-wise over these decades. In the present circumstances, this is quite significant and rules out what to many will seem a complacent response to the results of Pikkety.