The Presumption of Equality *

Many distributive egalitarians do not endorse strict equality of goods. Rather, they treat an equal division as having a special status such that departures from equality must be justified. They claim, then, that an equal division is “presumptively” just. Though the idea that equality is presumptively just and that departures from it may be just has intuitive appeal, making a case for this idea proves difficult. I argue, first, that extant “presumption arguments” are unsound. Second, I distill two general philosophical morals: luck egalitarians have not adequately defended the presumption of equality and they face serious obstacles in doing so; Rawls has defended it, but only indirectly via the contract apparatus. This approach narrows the presumption’s appeal. Third, I consider and reject two alternative ways of understanding the presumption of equality that might avoid the problems revealed by my examination of extant views. The first appeals to the idea of value pluralism. The second treats the presumption as a view about the burden of proof. I conclude, ultimately, that it is misleading to think of distributive egalitarianism as typically having the form of a presumption argument.


INTRODUCTION
Many distributive egalitarians do not endorse strict equality of goods. 1 Rather, they treat an equal division as having a special status such that departures from equality must be justified. They claim, then, that an equal division is "presumptively" just. Rawls, for instance, argues that inequalities in wealth are just only if they make everyone better off than they would be at a "benchmark of equality". Many luck egalitarians hold that distributive equality is the "moral default" and that departures from this default caused by brute luck 2 are unjust while those produced by choice are just.
Though the idea that equality is presumptively just and that departures from it may be just has intuitive appeal, making a case for this idea proves difficult. In this paper I first show that extant "presumption arguments", as I shall call them, are unsound. Some of the arguments I canvas are given in a critical vein, so not in all instances do I make an original case against the presumption. The point of this canvasing is to demonstrate the difficulties besetting presumption arguments. Second, I distill two general philosophical morals: luck egalitarians have not defended the presumption and they face serious obstacles in doing so; Rawls has, but only via the contract apparatus, which narrows the presumption's appeal. Third, I consider and reject two alternative ways of understanding the presumption of equality that might avoid the problems revealed by my examination of extant views. The first appeals to the idea of value pluralism. The second treats the presumption as a view about the burden of proof. I conclude that it is misleading to think of distributive egalitarianism as typically having the form of a presumption argument.

GETTING TO AND DEPARTING FROM EQUALITY
Here are the bones of the presumption arguments I will consider: 1. Equality is presumptively just because it eliminates the influence of luck on distribution. Departures produced by choice are just because they hold people responsible for their choices.
2. Equality is presumptively just because it distributes the effects of luck equally. Departures produced by choice are just because they hold people responsible for their choices. 1 I use "goods" here as a neutral term for whatever distribuenda particular theories endorse.
2 Luck egalitarians distinguish between brute luck and option luck. See Dworkin (1981: 293). Throughout my discussion, I will use "luck" to mean "brute luck".
LEAP 6 (2018) 3. Equality is presumptively just because it is demanded by the equal worth of persons. Departures produced by choice are just because they hold people responsible for their choices, which is necessary for recognizing their capacity for agency.
4. Equality is presumptively just because it contains no inequalities caused by morally arbitrary factors. Departures from equality are just when and because they increase everyone's wealth and maximize the wealth of the least wealthy.
5. Equality is presumptively just because it contains no inequalities justified by morally arbitrary factors. Departures from equality are just when and because they increase everyone's wealth and maximize the wealth of the least wealthy.
6. Equality is presumptively just because it is demanded by the equal worth of persons. Departures from equality are just when and because they increase everyone's wealth and maximize the wealth of the least wealthy.
Before outlining the arguments listed above, I must explain, in order to avert confusion, an idiosyncratic way in which "equality" is sometimes used by luck egalitarians. Ronald Dworkin (1981: 285-304), who is arguably the father of luck egalitarianism, maintains than an equal division of resources is not flatly equal but is, rather, one that meets the "envy test". According to this test, a division is equal if no one prefers someone else's bundle of resources -which includes both material goods and natural talents -to her own. This division, Dworkin states, ensures that each person pays the cost to others of her choices and is, to that extent, just.
One might think that changes over time to such an equal division are themselves just whenever people freely engage in production and exchange. However, Dworkin argues, this is not the case, for some will confront good luck and others bad, in particular with respect to their mental and physical powerssome will, luckily, have highly marketable talents and others less marketable talents. Because of the influence of luck, subsequent distributions will not be envy-free. Dworkin argues that those who suffer bad luck are owed compensation via a tax and transfer system, which is based upon a hypothetical insurance market, the details of which need not detain us. The point is that such a system is necessary to ensure that the distribution of resources in a market economy remains envy-free and, hence, equal and just.
Dworkin's argument is not a presumption argument because equality of resources is not presumptively just on his viewit is just full stop. Distributive justice is preserved, he thinks, so long as production and LEAP 6 (2018) exchange continually produce distributions that meet the envy test. I assume that "equality" in the presumption arguments I discuss below refers to flat equality and not to Dworkinian equality because Dworkinian equality is not, by the lights of luck egalitarianism, merely presumptively just. Susan Hurley (2003: 146-58; see also Eyal 2005) has proposed that luck egalitarians endorse the presumption of equality because they believe that an equal division eliminates the influence of luck on distribution, which is required by justice. They then endorse departures from this division that are produced by choice alone, as these inequalities hold people responsible for their choices, which is also required by justice. But, Hurley notes, this argument does not work because an equal division of goods might also be a matter of luck. She reasons as follows. If we redistribute equally all the goods that people have acquired as a matter of luck, then what share people have is still a matter of luck, at least insofar as people do not have the share for which they are responsible. It follows that an equal division does not extinguish the effects of luck, it merely rearranges them. So, an equal division cannot stand as the moral default on the ground that it neutralizes the effects of luck.

Eliminating the influence of luck
Suppose that the luck egalitarian can, contra Hurley, found the presumption of equality upon the demand to eliminate the influence of luck on distribution. In this case, the luck egalitarian would, nonetheless, not have a sound argument for the presumption of equality. 3 This is because (re)distributing equally all the goods that people have gotten as a matter of luck would not create flat equality. The resulting distribution would instead contain inequalities produced by choice; people's shares would be composed of whatever goods they acquired through their choices and an equal portion of the luck-tainted redistributed goods. It is hard to see how this division should serve, for the luck egalitarian, (merely) as a moral default. By luck egalitarian lights, this division does not require departing from since it contains inequalities that are just, namely those caused by choice, and (ex hypothesi) it extinguishes the effects of luck.
There are two problems, then, with founding the presumption of equality on the idea that this will neutralize the influence of luck. The first is that equalizing the goods gotten by luck does not necessarily eliminate the influence of luck. The second is that even if it did, equalizing the goods LEAP 6 (2018) gotten by luck does not produce equality overall. Samuel Freeman (2007: 120-21, 151; see also Vallentyne 2003: 170, 177) has defended the luck egalitarian against Hurley's charge by suggesting that the luck egalitarian is justified in demanding that the influence of luck be equalized. If we assume, he says, that everyone's natural endowments are equally a matter of luck, then a principle of equity requiring us to treat like cases alike enjoins us to distribute the goods produced by our natural endowments equally rather than to let them fall into the hands of people according to their particular talents. 4 The default justice of equality is founded, then, not upon the aim to neutralize luck, but rather on the demand to distribute the effects of luck equally. 5

Equally distributing the effects of luck
The problem with this amendment is that it does not address the second of the two problems identified above. Redistributing the effects of luck equally does not produce a flatly equal distribution. It produces one in which the goods initially acquired by luck are distributed equally and the goods attributable to people's choices are distributed unequally. And so, according to luck egalitarianism, there is no obligation of justice to depart from this division.

Equal moral status and responsibility for one's choices
Critics of Hurley claim that she proposes the luck neutralizing aim as a possible ground for the presumption of equality because she confuses the luck egalitarian demand to eliminate the effects of luck on inequality with a demand to eliminate the effects of luck on distribution (Vallentyne 2003(Vallentyne , 2006Lippert-Rasmussen 2005;Cohen 2006;Segall 2012). For luck egalitarians, what should be neutralized is not the effects of luck but rather the differential effects of luck. It follows that if equality were caused by luck, it would not be unjust on that ground. In this sense, equality has a special status for luck egalitarians: it is immune from moral condemnation 4 Presumably this argument would hold mutatis mutandis for other instances of luck that tend to affect people's distributive shares, such as their class position at birth, their health, etc. 5 Cohen (2006: 445) holds a similar view, although he does not offer it as an argument for an egalitarian default. He says, "[s]uppose…that we could divide goods up into those for which people are responsible and those for which they are not responsible. Then…we might use only the goods for which people are not responsible as the currency of justice… If it is fair for people to keep, before any redistribution is set in train, what and only what they are responsible for, because they are responsible for it, then the same conception of fairness also requires that the rest be distributed equally, because to distribute otherwise is to benefit people in disaccord with their exercises of responsibility". LEAP 6 (2018) even if it is caused by luck. Hurley calls this approach "the equality default view", suggesting that this immunity is unfounded. Kok-Chor Tan (2012: 89-90) counters Hurley's suggestion by providing a sketch of an argument for the presumption of equality and for departing from equality on grounds of choice. He reasons as follows: 6 1) Persons have equal moral worth as agents.
2) Therefore, distributive equality is the "moral default": it is the distribution from which departures must be justified.
3) Departures from equality are justified when they are expressions of agency.
4) Therefore, departures produced by choice are justified. 5) "Distributive arrangements" that reflect luck and not choice fail to treat people as moral equals.
6) Therefore, distributions that reflect luck and not choice are unjustified.
As it stands, this argument is invalid: given that equality is a distribution that reflects luck (barring very unusual circumstances), it would, by the lights of this argument, be unjust and so it cannot be presumptively just. However, this version of the argument makes Hurley's mistake. It treats distributions as such, rather than unequal distributions, as susceptible to the luck egalitarian criterion. This problem can be solved with the following changes: 1) Persons have equal moral worth as agents.
2) Therefore, distributive equality is the "moral default": it is the distribution from which departures must be justified.
3) Departures from equality are justified when they are expressions of moral agency. 4) Therefore, departures produced by choice are justified.

5)
Departures from equality that reflect luck and not choice fail to 6 Tan (2012: 89-90) claims that he is not in fact arguing for the egalitarian default. He says, "[i]n explicating what I take to be the implicit starting point of luck egalitarianism, that of equal moral agency and how that ideal is interpreted to support equal distribution as a default independent of luck but subject to choice, I have not provided any argument for it. I take this ideal of equal moral agency and its egalitarian entailment to be a basic and starting intuition common to most accounts of luck egalitarianism". treat people as moral equals 7 . 6) Therefore, departures that reflect luck and not choice are unjustified.
This version of the argument is valid but not sound. The ultimate reason for this is that the same principle that grounds equality also grounds departing from equality to inequality caused by choice. So, the argument provides no grounds for departing (or not departing, for that matter) from equality. To see this, notice that the equal worth of persons does not entail distributive equality. In fact, the equal worth of persons is compatible with a number of distributions, including a winner-take-all lottery, the division that recognizes persons' rights of self-ownership, and the division in which each gets what he deserves. Indeed, Tan's argument implies that the equal worth of persons is also compatible with the ultimately just distribution to which he believes we should depart. This is because, given that luck and choice exhaust the possible cause-types of inequality, premise 5) entails that inequalities caused by choice treat people as moral equals. So, both equality and inequality caused by choice have the virtue of treating people as moral equals. However, if both distributions recognize persons' equal moral worth, it is not clear, without further argument, why one is merely presumptively just while the other is completely or ultimately just.
Perhaps the further argument runs as follows: even though an equal division and an unequal division caused by choice both treat people as moral equals, the latter has the additional virtue of recognizing people's moral agency. So, an unequal division caused by choice is, in the end, just. The problem with this suggestion is that if recognizing people's agency is a distinct virtue from recognizing their moral equality, then it is not clear how departures from equality caused by luck fail to recognize people's moral equality, as Tan proposes. Indeed, it turns out that they do this by ignoring people's moral agency. Tan (2012: 89) states, "[f]or luck egalitarians, a distributive arrangement that reflects not agents' free decisions and choices, but the circumstances that are forced on them, such as their good or bad luck, fails to treat them as moral equals". The manner in which departures produced by choice succeed in treating people as moral equals, then, is by recognizing their moral agency and the manner in which departures produced by luck fail to treat people as moral equals is by ignoring their moral agency. So, in the end, that it treats people as moral equals is what justifies equality as the default and what justifies departures LEAP 6 (2018) from equality that are traceable to choice. Thus, the puzzle of why inequality based upon choice is preferable to equality remains.
(One might wonder, at this point, how it is that equality treats people as moral equals given that it fails to recognize people's moral agency due to its lacking inequalities caused by choice. I believe the luck egalitarian answer must be this: the egalitarian moral default is immune from any condemnation that appeals to what causes it: it is not unjust even if caused by luck (as we saw above) and it is not unjust even if caused by a failure to incorporate choice. 8 ) To summarize, then, the problem with Tan's approach is that because the same principle grounds both the default and departures from it, there is no reason to think of the default as merely presumptively just and departures based on choice as just full stop. 9

Morally arbitrary factors and mutual benefit
Another presumption argument can be found in Brian Barry's interpretation of an argument given by John Rawls that G.A. Cohen calls "the Pareto argument for inequality". Cohen contends that this argument, which is not only reconstructed but also endorsed by Barry, is in fact internally inconsistent (Rawls 1971: 60-75;Barry 1989: 213-34;Cohen 2008: 87-97, 151-168; see also Cohen 1995 andShaw 1999).
Some background: Rawls offers the Pareto argument, which is distinct from his well-known social contract argument, to support the "difference principle". The difference principle governs the distribution of income and wealth and says that only inequalities that benefit everyone and maximize the income of the worst off are just. The argument identifies an equal division as a presumptively just "benchmark" and states that we should depart from this benchmark only when inequalities fulfill the difference principle.
Rawls assumes, in this argument, that there may exist strong Pareto improvements on equality so that it will be possible to increase the income of everyone if inequality is permitted. He assumes this on the ground that the sum total of wealth may be greater when inequality is permitted because these inequalities provide incentives for people to work more, or at harder tasks, than they would under equality. Rawls's aim in the Pareto argument is to single out the difference principle as the Pareto improvement on equality that qualifies as just. Though the reasoning is somewhat 8 For discussion, see Segall (2016: 48-73). 9 A related issue is whether the two "conjuncts" of the luck egalitarian ideal can be grounded in the same principle. These are the claims that departures produced by choice are just and that departures caused by luck are unjust. See Sher (2014: 2-19). murky, it is clear that the idea that people's natural talents are "arbitrary from a moral point of view" and hence should not "improperly influence" their incomes figures prominently.
According to the Barry/Cohen interpretation, Rawls founds the benchmark of equality upon the morally arbitrary status of natural talents and he founds the difference principle upon the irrationality of prohibiting mutually beneficial inequalities. Rawls's reasoning for the benchmark, on this account, is as follows: inequalities in wealth that have morally arbitrary causes -including those caused by differences in natural talent -are unjust. Therefore, an equal distribution of wealth is prima facie just. 10 Cohen identifies a problem with this inference: a division lacking inequalities with morally arbitrary causes will nonetheless contain inequalities with morally non-arbitrary causes (which, as Cohen sees it, are inequalities caused by choice). So, prohibiting inequalities caused by something morally arbitrary does not produce equality. So, it looks like the benchmark of equality cannot be grounded on the claim that inequalities with morally arbitrary causes are unjust.
Cohen concludes from this problem with Barry's reconstruction not that the reconstruction is wrong, but that Rawls must hold that differences in income caused by choice are not inequalities. So, the benchmark, Cohen says, is, for Rawls, not a flatly equal distribution. (As odd as this idea might seem to a Rawlsian, the notion that a division containing differences in shares produced by choice is an equal division has a precedent in luck egalitarianism, as we saw above. This may explain Cohen's willingness to regard Rawls's benchmark of equality as containing choice-produced differences in shares.) Rawls reasons further, according to this interpretation, that it would be irrational to settle for equality if there exist (strongly) Pareto superior unequal distributions -unequal distributions, that is, that benefit everyone. Of these distributions, the one that maximizes the income of the least well off -who have (again, as this interpretation goes) the strongest complaint against inequality -is just. Hence the difference principle is just.
Cohen maintains that Rawls's case for the benchmark undermines his case for departing from it. If the benchmark is prima facie just on the ground that it contains no income inequalities with morally arbitrary causes, and if, as Rawls concedes, income differences that maximize the income of the least wealthy are caused by differences in natural talent, 10 This is Cohen's terminology. Rawls himself does not refer to the benchmark of equality as "prima facie just".
LEAP 6 (2018) then the inequalities sanctioned by the difference principle are unjust. In other words, if equality is prima facie just on the ground that it is devoid of inequalities with morally arbitrary causes, then a Pareto improvement on equality that contains inequalities with morally arbitrary causes cannot be all things considered just.
Cohen's observation about Rawls argument is, though, trivial given the way in which he sets the argument up. Because differences in shares of wealth with morally non-arbitrary causes are, by definition, not inequalities, then inequalities are, by definition, differences in shares with morally arbitrary causes. So, on Cohen's account, Rawls's argument for the benchmark amounts to the claim that equality is just because it contains no inequalities. Naturally, any departure from the benchmark would be unjust on this account. Only if the benchmark is flatly equal and all differences in shares, regardless of their cause, count as inequalities, does Cohen's objection have force. For in this case, it would be a substantive claim to say that the difference principle is inconsistent with the justification for the benchmark on the ground that the difference principle allows inequalities that have morally arbitrary causes.
In summary, then, there are two problems with Rawls's argument as Cohen interprets it (via Barry). The first is that eliminating inequalities with morally arbitrary causes does not in fact produce equality. The second is that if eliminating inequalities with morally arbitrary causes did (somehow) produce equality, the ideal justifying equality would be violated by departures to the difference principle.
There is a way of avoiding this second problem, Cohen says, but it ends up vitiating the case for the benchmark. Suppose we interpret Rawls as claiming not that morally arbitrary causes make inequalities unjust but that morally arbitrary causes cannot make inequalities just. If this is the case, then, departures from the benchmark to the difference principle are indeed just: though the inequalities allowed by the difference principle are caused by differences in natural talent, they are not justified by their being so caused. They are justified by the fact that they maximize the wealth of the least wealthy.
However, Cohen argues, this approach provides no ground for the benchmark of equality. If what makes a distribution prima facie just is that it contains no inequalities justified by their morally arbitrary causes, then equality is not the only distribution that can stand as a benchmark. Indeed, the principle of utility, Cohen says, can be the benchmark on this account, for it justifies inequalities on the ground that they are necessary to maximize the sum total of goods. So, on this second reconstruction of the Pareto argument, Cohen claims, departures from the benchmark of LEAP 6 (2018) equality to the difference principle are justified but the benchmark itself is not.
However, the problem is not merely that the ideal justifying equality does not single out equality as the benchmark, as Cohen observes, it is that the difference principle can also qualify as the benchmark since it shares the virtue of containing no inequalities justified by the morally arbitrary. As Cohen sets up this second version of the Pareto argument, it seems that there is no reason to prefer departures from equality to equality, as long as those departures contain no inequalities justified by their morally arbitrary causes, because it is sufficient for qualifying as just (albeit, prima facie just) that a distribution contains no such inequalities.
This would make the problem with the second Pareto argument similar to the problem with Tan's account. On his view, recall, the same principle -the equal worth of persons -justifies both equality and departures from it and, therefore, there is no reason to see one distribution as presumptively just and the other as fully just. On this second Pareto argument, separate principles justify equality and inequality: equality is justified by the fact that it contains no inequalities justified by something morally arbitrary and inequality is justified by the ideal of mutual benefit. Hence it is not as clear as it is on Tan's account that there is no reason to see an equal division as presumptively just and see the proposed departure from equality as ultimately just; perhaps the ideal justifying inequality can defeat the ideal justifying equality.
Yet the second Pareto argument is vulnerable to Tan's problem, but for a different reason. The reason is that what makes equality just is the same thing that makes departing to the difference principle just: the absence of inequalities justified by their morally arbitrary causes. Indeed, that the difference principle shares the just-making property of equality is what keeps the second Pareto argument from making the mistake of the first Pareto argument. What we learned from that argument was that in order for a departure from equality to be just, the departure distribution cannot violate the principle that justifies the equal division. So, if equality is just because it contains no inequalities justified by the morally arbitrary, then the just unequal division must also contain no inequalities justified by the morally arbitrary. (If it contained such inequalities it would be unjust.) It follows that, on the second Pareto argument, it is not obvious how we can distinguish, from the point of view of justice, between the benchmark of equality and the difference principle because they both contain a property that is sufficient for making them just. Hence the second Pareto argument, like Tan's argument, appears not to be able to explain why equality is merely presumptively just while the proposed departure from equality is LEAP 6 (2018) ultimately just.
However, Cohen's second Pareto argument may be salvageable on grounds not available to Tan precisely because separate principles are invoked to justify the benchmark and difference principle inequality. Perhaps the argument Cohen has in mind is this: equality is, let us say, sufficiently just because it contains no inequalities justified by differences in natural talent, and so no inequalities justified by something morally arbitrary. Difference principle inequality is more just than equality because it also contains no inequalities justified by something morally arbitrary and, on top of that, it is mutually beneficial relative to equality. (Another option would be to say that difference principle inequality is not more just than equality but is in some other way morally preferable. This interpretation is suggested by Barry's claim that it would be irrational to insist on equality if mutually beneficial inequality that maximally benefits the worst off is available.) On this interpretation, then, the second Pareto argument is stronger than Tan's argument but it is not immune from Cohen's original criticism of that argument, namely, that other divisions (besides equality) that lack inequalities justified by something morally arbitrary (though not the division to which we should depart) can qualify as the benchmark.

Equal moral standing and moral arbitrariness
A further problem with the first version of the Pareto argument outlined above is not noticed by Cohen. It is that if the benchmark of equality contains inequalities produced by choice, as Cohen claims it must, it is unclear why strong Pareto improvements on equality might be possible and it is unclear why departing from equality is required to avoid leveling down. After all, if people already have different shares of wealth at the benchmark that are produced by their choices about work, they do not need the promise of incentives inequality in order to work harder. The incentives inequality is already present at the benchmark. This observation suggests that the Barry/Cohen account of Rawls's presumption argument LEAP 6 (2018) is incorrect. 11 So, I offer below an alternative account of Rawls's argument for the benchmark of equality. However, before outlining that account, which is different in its structure from the arguments I have thus far examined, let me identify three general desiderata for constructing presumption arguments that can be gleaned from the analysis above. It turns out that these cannot be simultaneously fulfilled, so presumption arguments must have a different structure than those discussed above.
1. The thing that allegedly produces and justifies equality must actually produce equality. Otherwise genuine equality is not the default, and in some cases, it is not clear why the alleged default requires departing from.
2. The thing that produces and justifies equality must entail equality. Otherwise, it is not clear why equality is the default, rather than some other distribution, including in some cases, the distribution departure to which is recommended.
3. The unequal departure distribution must fulfill the principle that justifies equality. Otherwise, that distribution cannot be just.
The second two of these cannot be mutually satisfied. If what justifies equality entails it, then no departures can be justified. And if what justifies equality does not entail it, then departures can be justified, but equality is not uniquely justified as the default. As we will see below, this problem is traceable to the following feature of the above arguments: the presumptive justice of equality and the ultimate of justice of inequality are thought to hold in all circumstances. Rawls's Pareto argument, though not sound as it stands, avoids this problem. It runs (as I interpret it) as follows: 1) Persons have equal moral worth.
2) Therefore, an equal division stands as a "benchmark for measuring improvements".
3) Suppose strong Pareto improvements on equality (e.g., unequal 11 The idea that the moral arbitrariness claim is deployed by Rawls in the argument for the benchmark is not well supported by the text of the Pareto argument. But it may be supported by Rawls's reason for rejecting the bargaining theorist's use of the nonagreement point as the "status quo": "[i]t is to avoid the appeal to force and cunning that the principles of right and justice are accepted. Thus, I assume that to each according to his threat advantage is not is not a conception of justice. It fails to establish an ordering in the required sense, an ordering based on certain relevant aspects of persons and their situation which are independent from their social position and their capacity to intimidate and coerce". Rawls says, further, in a footnote to this passage, "[w]hat is lacking is a suitable definition of a status quo that is acceptable from a moral point of view. We cannot take various contingencies as known and individual preferences as given and expect to elucidate the concept of justice (or fairness) by theories of bargaining. The conception of the original position is designed to meet the problem of the appropriate status quo" (1971: 134).
LEAP 6 (2018) distributions that increase everyone's share) are possible. 4) Then, of the two main candidates for the just strong Pareto improvement -laissez-faire and the difference principle -the one that minimizes the influence of morally arbitrary factors on people's income shares is just.
5) The difference principle minimizes the influence of morally arbitrary factors on people's income shares.
6) So, the difference principle is just.
This is merely a skeleton of Rawls's account, but the details are not important for assessing its form qua presumption argument. I wish to make two points about its structure. The first is that, as in Tan's case, the second premise does not follow from the first without further argument. As I have already observed, many distributions can be grounded in the equal standing of persons. Indeed, if this were not case, there would be little disagreement among theorists of distributive justice.
The second point about the structure of Rawls's argument concerns the status of the presumption. The sense in which equality is presumptively just is that it is just unless inequality can be mutually beneficial (Rawls 1971: 62,76; see also Cohen 2008: 156-60). For Rawls (1971: 78), it is an open question as to whether or not this is the case. The answer depends upon the plausibility of certain economic theories and on controversial ideas about human motivation -for instance, the claim that people, as a rule, will work less hard in the absence of incentives to acquire extra wealth. 12 At the end of the day, for Rawls, whether or not equality or the difference principle is just depends on the circumstances. 13 For this reason, an entailment relation between the justifier of equality and equality, which Rawls does not provide and which is necessary to single out equality as the benchmark, does not prevent departures: we can say that justice demands equality, and only equality, in some circumstances and it demands a departure from equality in others. Compare this approach to Tan's. On his account, both equality and inequality generated by choice are just (in the way in which they are just) in any circumstance. The difference is in the way in which they are just -presumptively or ultimately.
This gives us a clue as to how we might salvage Tan's argument. It can be reinterpreted to have the same structure as Rawls's. 14 This interpretation   12 For criticism, see Cohen (1997). 13 For discussion, see Cohen (2003). 14 Thanks to Christopher Wellman for pointing out that the luck egalitarian argument might have this structure.
LEAP 6 (2018) solves the main problem with Tan's view, which is that because the same principle grounds both equality and choice-generated departures from it, there is no reason to treat equality as merely presumptively just and departures generated by choice as ultimately just. If circumstances change, however, then the same grounding principle might entail a different distribution. It might be that equal moral worth demands equality if choices do not produce inequality and otherwise demands inequalities produced by choice.
Nevertheless, it is unlikely that this interpretation is what luck egalitarians have in mind when they invoke the presumption, because it commits them to the idea that it is an open question as to whether or not choice will produce inequality. But there are strong reasons to think that inequalities produced by choice are simply inevitable. This is because differences in shares based upon choice depend, not on complex theoretical claims or controversial empirical claims, but on nothing more than the laws of physics: if you and I have the same capabilities and are gathering nuts in the same place, if I choose to gather for ten minutes and you choose to gather for twenty, you will (ceteris paribus) have more nuts than I. If inequality produced by choice is inevitable, then it is implausible to treat the presumption as endorsing equality on the condition that choice might not produce inequality.
To summarize, then, Rawls's Pareto argument contains a plausible justification for departing from equality, to the extent that that departure precludes levelling down. (One may not agree with his account of which unequal distribution is just, but the idea that mutually beneficial inequality is, at least pro tanto, preferable to equality is reasonable.) However, Rawls's account fails insofar as it does not single out equality as the just benchmark.
We can now draw two general conclusions about presumption arguments. First, luck egalitarians have not successfully defended the presumption of equality, because their distinction between presumptively and ultimately just describes two different ways in which distributions are just, rather than two different circumstances under which distributions are just. They are therefore vulnerable to the dilemma identified above: whatever justifies equality must entail it, or else equality is not uniquely justified, but if it does entail it, all departures are precluded.
Second, Rawls (1971: 118-83) is entitled to the presumption of equality, but the Pareto argument does not do the necessary work. In order to move from the claim of equal moral standing to the benchmark, Rawls needs the contract argument. In this argument, he models the equal standing of persons in his description of the parties to the contract and in his description of the hypothetical circumstances (the "original position") in LEAP 6 (2018) which the parties deliberate about the distribution of wealth (among other "primary social goods"). The parties would opt for equality he argues, given their equal claim to social goods (derived from their equal status) and their lack of knowledge of their particular capabilities, unless they can all have more under inequality. In this case, Rawls argues, they would opt to maximize the wealth of the least wealthy, given their inability to predict their ultimate place in the distribution of wealth. While this approach arguably supplies a justification for equality as the benchmark (and, indeed, a justification for departing to the difference principle) it is controversial qua hypothetical consent argument. 15 One last consideration: perhaps there are alternative ways of understanding the presumption of equality that I have not considered. Below, I briefly discuss two. I argue that neither is a strong candidate.

VALUE PLURALISM
The idea of value pluralism is frequently invoked by egalitarians, especially luck egalitarians. 16 This doctrine says that equality is one among many values that bear upon the assessment of distributive arrangements. 17 One proposal is that the presumption of equality is simply an expression of the notion of value pluralism. To say that an equal distribution is presumptively just is simply to say that, while there is something to be said for equality in assessing distributive arrangements, there is something to be said for other values as well and that these values permit or require deviations from equality. The presumption of equality, then, simply expresses the sensible view that equality is not the only thing that matters.
But notice that this way of putting the point presupposes the idea that 15 For discussion, see Dworkin (1975); Stark (2000); Enoch (2017). 16 Larry Temkin (2003: 63), for example, says, "any reasonable egalitarian will be a pluralist. Equality is not the only thing that matters to an egalitarian. It may not even be the ideal that matters most. But it is one ideal among others that has independent moral significance". See also, Temkin (2002); Parfit (2002); Segall (2007);Eyal (2007);Cohen (2008: 4). 17 There are several notions of value pluralism proposed by egalitarians and several contexts in which value pluralism is said to apply. Some theorists, such as Temkin, Parfit and Cohen, hold that equality is one among many values bearing upon the goodness or the badness of a distribution. Others, such as Segall, hold that distributive justice (understood as the fulfillment of the luck egalitarian ideal of equality) is one among many values bearing upon the justice of the design of social institutions. Segall holds, further, that justice is one among many values bearing upon the morality of institutional design. Another view, advanced by Eyal, claims that luck egalitarian equality is one among many values bearing upon both the moral and non-moral goodness of a policy of compensating victims of bad luck. (See the references in the previous footnote.) equality has a special status. The proposal states that because equality is not the only thing that matters, we are sometimes justified in departing from it. But that assumes that equality has some sort of priority -it is the division that other values might justify deviating from. If equality really were simply one among many values pertinent to distributive ethics, then there are no grounds for positioning it as the distribution that other values might defeat. On a genuinely value pluralist view, we would take equally into account the ideals of, say, utility, mutual benefit and equality. 18 That is to say, we would not assign a special weight or status to any of these values. A genuinely pluralist approach, then, represents an alternative to the idea that equality is presumptively just. Hence this approach cannot stand as an interpretation of the presumption.

THE BURDEN OF PROOF
Another way to interpret the notion that distributive equality is presumptively just is as a claim about the burden of proof: the burden of proof, the argument goes, is upon those favoring an unequal division and not upon those favoring equality. Equality simply requires no justification. 19 As a preliminary point it is worth noting that as a matter of social practice, the burden of proof tends to be assigned simply to those who hold the more unusual view. For instance, vegetarians are often expected by meat-eaters to justify their refraining from eating animals, where meat-eaters generally do not see themselves as owing anyone a justification for their practice, in spite of the fact that their practice is arguably more harmful. Though most people may not believe in distributive equality, most political philosophers do. 20 Yet this would surely be a flimsy ground for assigning the burden of proof to the non-egalitarian.
So, let us assume that claims about the burden of proof can stand on sturdier ground than the mere prevalence of a view. To assess the idea that distributive equality demands no justification let us compare that claim to some views about the burden of proof that are widely accepted. One such view is the legal presumption of innocence in some systems of criminal law. Another is the idea that the atheist is not required to disprove the existence of God in order to be justified in believing that God does not exist. The notion that the burden of proof falls upon those who support distributive inequality, I argue, is relevantly different from these two cases. 18 See Parfit (2002: 87-88.) 19 Thanks to David Rondell for proposing this interpretation. See Gosepath (2011); Wollheim and Berlin (1955-56). 20 Many theorists of distributive justice are either Rawlsians or luck egalitarians.
LEAP 6 (2018) So, we have reason to doubt the claim that distributive equality demands no justification. I argue, further, that even if the presumption of equality were relevantly similar to the presumption of innocence, the idea that equality should be presumed just demands justification, for those who say that the accused should be presumed innocent can offer reasons for their view. Yet providing a justification for the idea that distributive equality requires no justification is tantamount to providing a justification for the idea that equality should be presumed to be just. And, depending on how that argument goes, it will be subject to the pitfalls I identified above in arguments for the presumption.
Here is how the presumptive innocence of the accused differs from the presumptive justice of distributive equality: where the former is a strictly epistemic notion, the latter is not. The demand to presume the accused innocent says that we must treat the accused as though they are innocent until there is sufficient evidence of their guilt. That is, until we know whether or not they are innocent, we treat them as though they are. We are not required to believe that they are innocent, or, alternatively, we do not ascribe to them the property of being innocent. To presume that an equal division is just, however, is not to treat equality as though it is just until we have sufficient evidence to think it unjust. Rather it is to say that equality is just, but that its justice can be defeated by other considerations. The presumption of innocence tells us what we should do when we do not know what to believe. The presumption of equality tells us what we should believe, namely that equality is just in a certain circumstance or that equality is initially just.
Even if I am mistaken about this difference, however, it seems reasonable for someone to demand reasons for assigning the burden of proof in a particular way. And, in fact, proponents of the idea that the burden of proof should be borne by the prosecutor to establish the guilt of the accused have offered reasons for their view. For example, one line of defense appeals to the serious harm of wrongful conviction. Placing the burden of proof on the prosecution to establish the guilt of the accused tends to produce more improper acquittals than improper convictions. The improper acquittals are seen as the legitimate price of avoiding wrongful convictions given the power imbalance between the accused individual and the state and the serious consequences of wrongful conviction.
Another view, proposed by Hamish Stewart (2014: 410), is that the accused have a right to be presumed innocent simply in virtue of being persons. The basic idea is that the moral status of persons includes being "without reproach"; it includes the right to not be "…found to have done wrong merely on the basis being a person". Therefore, to judge someone legally in the wrong that person must have done something legally wrong. So, persons have a legal right to be presumed innocent until proven guilty. The continuity between this argument and both Tan's and Rawls's arguments for the presumption of equality is striking and lends credence to the notion that the presumption in favor of distributive equality must itself be justified; it is not adequate to simply assert that distributive equality needs no justification.
The idea that the burden of proof falls upon the theist is similar in one respect to the idea that the burden of proof falls upon the non-egalitarian. Both are views about what one should take to be true. One should believe that God does not exist in the absence of evidence that he does exist and one should judge equality to be just in the absence of reasons that it is not just. However, the case involving God's existence hinges on the fact that the theist formulates the thesis that God exists in way that ensures that it cannot be disproved and then claims that the fact that God's existence cannot be disproved justifies belief in God. This is the point of Russell's teapot analogy. Russell says that if he were to claim that a china teapot, too small to be detected by the most powerful telescopes, is orbiting the sun, it would be ludicrous to claim that the teapot non-believer must disprove the existence of the teapot in order to be justified in believing in its non-existence. 21 The dis-analogy between this case and the presumption of equality is plain. The theist says that because we cannot prove the non-existence of a thing the non-existence of which is virtually impossible to prove we must believe in the existence of that thing. This is indeed a strong reason for thinking that the burden of proof does not rest upon the atheist. Yet no such sleight of hand is present in the case of the presumption of equality. We cannot say that the burden of justification falls upon advocates of distributive inequality because those individuals have formulated their account of distributive justice in such a way that there are no reasons that count against it and then claim that egalitarians must accept their view on the ground that they (egalitarians) can provide no reasons against it. So, just as the presumption of equality is not analogous to the presumption of innocence, it is likewise not analogous to the presumption of the non-existence of God. 21 To this, theists claim that absence of evidence is not evidence of absenceeven if evidence cannot be provided for the existence of something it might still exist. And, moreover, theists say, the basis for their belief in God is not the absence of evidence for his non-existence but rather evidence for his existence. Just as there is evidence that there is not a China teapot orbiting the sun, there is evidence that there is a God, for the postulate that God exists can explain e.g., the origins of universe, the complexity of life on earth and so on. To this the atheist replies that the things that the postulate allegedly explains can be explained without the postulate.

SUMMARY
The idea that distributive equality enjoys a special status such that departures from it require justification is indeed attractive. It allows us to endorse equality, in some sense, and, at the same time, avoid some major criticisms of equality, including the claim that it requires levelling down or fails to hold people responsible for their economic choices. Nevertheless, characterizing this special status and showing how it can be overridden or defeated proves difficult. I argued above that despite their endorsement of the presumptive justice of equality, luck egalitarians have not successfully shown equality to be presumptively just. This failure is due to the structure of their presumption arguments. In order to assign a special status to equality, then, luck egalitarians must fashion an argument that takes a different form.
Rawls, on the other hand, can assign a special status to equality. Its special status takes this form: it is the just division if, as a matter of fact, inequality does not increase the social pie in a way that benefits everyone. Otherwise it is unjust and the mutually beneficial division that maximizes the wealth of the least wealthy is just. To make this case, however, Rawls cannot rely upon the Pareto argument alone because that argument does not explain why equality stands as a benchmark. It explains only why departing from the benchmark to the difference principle is justified. Rawls must rely on the social contract argument to justify the special status of equality: equality is the benchmark because it is what people who know nothing of their natural talents and initial social position would choose (unless mutually beneficial inequalities are an empirical possibility). A weakness of Rawls's approach is that it is limited in its appeal, given the controversial nature of social contract arguments.