I. COMPARATIVE AND NON-COMPARATIVE NOTIONS OF JUSTICE

After years of relative neglect, the notion of desert is attracting serious philosophical attention.Footnote ¹ In particular, attention is given to the question of whether and to what extent this notion is comparative. Consider the following scenario. Adam and Eve die and fly up to Heaven. There stands God, reviewing their records at the Pearly Gates: ‘You two have amazingly similar, and amazingly horrible, records – records that suggest both of you should be cast down into the everlasting fires of Hell. But only Adam will suffer that fate. Eve, you will join Adam for only one year, and then an angel will show you the way out.’ Given Eve's similar record, Adam immediately protests that it is unfair to set only Eve free. To this, God replies: ‘You got what you deserve; you have no reason to complain.’

God's reply sounds powerful, almost irresistible. After all, why should Adam's punishment depend on Eve's punishment? Yet something unfair must be going on here; surely God is playing some trick. How can it be fair to send Adam to burn in Hell while largely sparing Eve with her identically spotty record? Is it any of Adam's business what God does with Eve? Or, in terms of the title, can a deserved-but-unequal punishment be just? The purpose of this article is to explicate the conceptions of justice that underlie Adam and God's competing outlooks and to present the best case – may God forgive us – for Adam's position, although we end up not endorsing it as such.

To unpack the philosophical problem under discussion, let us generalize the opening story. Assume that A1 and A2 hold similar moral records, and are also alike in all pertinent respects to punishment;Footnote ² hence they deserve the same punishment, P1. However, the punishment, P2, actually imposed upon A2, is less harsh than the punishment A2 actually deserves, and there is no practical way of making P2 any harsher. Given these circumstances, what punishment should be imposed upon A1? One can initially think of two different answers that express fundamentally different views about retributive justice, respectively:

• A Non-Comparative Account of Retributive Justice (hereafter: ‘Non-Comparative Justice’): Justice requires that A1 receives P1, the punishment he deserves, despite A2 receiving a more lenient punishment, P2.

This view is non-comparative in that it makes punishment depend solely on individual desert, unrelated to the punishment anyone else receives. The only justice considerations are those concerning retribution and they are strictly non-comparative, just as desert seems to be.Footnote ³ In this sense, Non-Comparative Justice might be seen as a ‘purist’ theory of justice in punishment. From the point of view of Non-Comparative Justice, there is no loss, in terms of justice, in imposing on A1 the punishment that A1 deserves, P1. Punishing A1 with P1 inflicts no wrong on A1. The only regret is that A2 does not get what she deserves, but this failure in retributive justice would not be corrected by adding an extra failure, i.e. by refraining from imposing upon Adam the punishment that he deserves.Footnote ⁴ Since Adam's punishment fits his individual moral desert, he has no grounds to argue that Eve's good fortune makes his treatment unjust. To be sure, he may protest that Eve did not get the punishment she deserves, but he is in no better position to voice this protest than anyone else, which is a different way of saying that his complaint cannot be understood as a complaint about an injustice done to him. A clear manifestation of Non-Comparative Justice can be found in Van den Haag, a leading supporter of the death penalty. In his view, the fact that some people who deserve the death penalty don't receive it ‘is hardly sufficient for letting anyone else found guilty escape the penalty . . . if the death penalty is morally just, however discriminatorily applied to only some of the guilty, it does remain just in each case in which it is applied.’Footnote ⁵

• A Comparative Account of Retributive Justice (hereafter: ‘Comparative Justice’): In case A2 receives a more lenient punishment than he deserves, P2, justice requires that A1 receives no more than P2 either.

This view is comparative because it assumes that justice in punishment is determined not only by individual desert but also by the punishments meted out to others. The Comparative Justice view is non-purist in the sense that factors other than the wrongdoer's desert determine the justness of his punishment. Comparative Justice shares with Non-Comparative Justice the conception that the only justice considerations are those concerning retribution, and not those concerning deterrence, incapacitation, etc. Yet, unlike Non-Comparative Justice, according to Comparative Justice, (retributive) justice is served not by strictly imposing upon people the punishment they deserve but also by taking into account what others receive for the same transgressions. Such a comparative view of retributive justice seems to be held by David Miller, Tom Hurka and Shelly Kagan, who argue that retributive justice has non-comparative as well as comparative aspects.Footnote ⁶

Both Non-Comparative Justice and Comparative Justice are rather radical views of justice in punishment. Under Non-Comparative Justice, if a criminal gets the punishment he deserves, he has no justice-based grounds for complaint about the huge discrepancy between his own punishment and that imposed on other wrongdoers with similar records. Comparative Justice, or at least some versions of it (see below), is even more radical. It implies that in so far as retributive justice is concerned, we should let out of jail all the criminals who received a more severe punishment than others with similar records, once the more severely punished complete the same term imposed on those less severely punished. Of course, there might be good social reasons (such as deterrence, rehabilitation or incapacitation) for not letting them out, but if these are ignored, as we do here, Comparative Justice may compel this mass release of prisoners.

Some philosophers, such as Hurka and Kagan, acknowledge the comparative aspects of desert and claim that these aspects should be balanced against its non-comparative aspects. Since, in their view, the comparative aspects of retributive justice stem from the importance of equality to justice, we shall refer to them as ‘egalitarian comparativists’. While we agree – contra Non-Comparative Justice – that comparison is necessary for implementing justice, we believe the reason for this is not based on equality per se. First, as argued by Harry Frankfurt and others,Footnote ⁷ equality in itself lacks any intrinsic moral value. Second, egalitarian comparativists are committed to balancing between two very different values – desert and equality. This balance presents a serious challenge to policymakers. It would thus be better if the value competing with desert could be shown to be conceptually closer to it, which is partially what we try to show below.

Accordingly, we develop a non-egalitarian comparativist account of retributive justice, which suggests that while comparison is vital to the imposing of just retribution, this is so only because of its contingent connection to two crucial aspects of retributive justice: the importance of respect and the importance of making sure that the punishment imposed is the one deserved.

II. RETRIBUTIVE JUSTICE AND RESPECT

Let's then go back to the opening example. God's claim sounds convincing because of its underlying assumption regarding God's perfect justice. With an omniscient and perfectly just God, one could be sure that if Adam is sent to Hell, then (a) this is the punishment he truly deserves and (b) the punishment was imposed with clean hands, so to speak. However, the example ignores the fact that if it is God imposing the punishment, then just as He cannot fail to impose upon Adam the punishment he deserves, He cannot fail to do so with regard to Eve either. An omnipotent, omniscient and perfectly just God cannot treat Adam and Eve differently, given their similar moral records. It is true that, under Non-Comparative Justice, God does not care about mere equality, which, ex hypothesi, has no intrinsic value. However, He does care about retribution: His eyes, as Jeremiah says, ‘are open upon all the ways of the sons of men, to give everyone according to his ways, and according to the fruit of his doings’ (32:19). Given His perfection, God cannot fail to give everyone what they deserve, and in both cases He would do it with clean hands. Viewed from this perspective, the example is incoherent.

In the real world, however, punishments are imposed by imperfect agents, and it is certainly possible that these agents impose on A1 the punishment that A1 deserves while imposing on A2 a more lenient one than A2 deserves. Our story of Adam and Eve should thus be reformulated as one about justice in the real world, justice on earth. Assume, in this vein, that a judge (hereafter: ‘Judge’) imposes a ten-year sentence on Adam for some crime, but only a five-year sentence upon Eve for that same crime. How could this happen? This could happen either because of some innocent error or because of some kind of immoral bias against Adam. In both cases comparison of the sentences is crucial for discovering that something went wrong.

We start with the case of an innocent error. Here differential punishment is worrisome because of the doubt it casts on whether wrongdoers actually got what they deserved. If Adam and Eve commit the same crime yet receive different punishments (by the same judge, or even by different judges in the same legal system), a suspicion is raised that either Adam got a harsher punishment than he deserved, or that Eve received a lighter one than she deserved, or both. Investigating into the matter might reveal some (honest) mistakes about the deserved punishments.Footnote ⁸

But we are interested here in the opposite question of whether a clearly deserved punishment can nonetheless be unjust, and this is where the immoral bias we mentioned above plays a significant role. Suppose that Adam is punished more harshly than Eve because Judge is motivated by racism towards the social group to which he belongs. This would be a clear case of a punishment expressing disrespect towards Adam, and might well justify some type of intervention. Now imagine a different scenario: because of budgetary constraints, there is not enough room in jail for both Adam and Eve. Judge flips a coin and, unfortunately for Adam, it is he who loses the toss and has to go to jail.Footnote ⁹ Or, to take a more common example, consider the random assignment of cases to judges who preside in the same court, all of whom are competent and decent, yet apply different sentencing policies within the range of the law, as a result of which Adam ends up with a harsher punishment than Eve's. In these cases, we are far less troubled by the way Adam is treated and by the discrepancy between his punishment and that of Eve. Obviously there is an advantage in a legal system which speaks in one voice and imposes the same punishments for the same crimes and records. But given the individual differences between judges and the inescapable indeterminacy of desert-based-judgments, such an ideal is unattainable.

The distinction between differences in punishment due to prejudice and differences due to random factors is similar to that between racial profiling and random police patrols. In both cases, some criminals are caught while others are spared, which seems to suggest some unfairness. Yet the unfairness exists only in the racial profiling case and not in the random patrol case, because with a random police patrol everyone gets an equal ex ante chance of being caught, hence nobody can complain when he is arrested while others are not.Footnote ¹⁰ When, however, one is arrested as a result of racial profiling, one does have a basis for a complaint against the arrest. More generally, when one is arrested, accused, convicted or punished as a result of disrespect, one has a justified grievance against such treatment, which is another way of saying that one is wronged – by the police, the District Attorney or the court.

Note that it is not the immorality of Judge in general that makes the punishment morally problematic, but the immorality, i.e. the disrespectful attitude, that underlies the punishment she imposes. If Judge lies, commits tax fraud or drives recklessly, yet none of this motivates her punishment practice, then Adam has no basis for complaining against the justness of the sentence (although he may have a legitimate complaint against Judge, probably shared by many others). In short, it is not the corrupt character of Judge in itself that makes a punishment problematic, but the corrupt character as manifested in the imposition of punishment.

This explanation for the moral defect in imposing on Adam a harsher punishment than the one imposed on Eve – i.e. that it expresses disrespect towards him – is in the right direction but is still incomplete. If punishment P1 is what Adam truly deserves, how could it be the case that the moral value of imposing it diminishes just because of the imposer's flawed motivation? This indeed looks puzzling, which is why Non-Comparative Justice looks so attractive. Indeed if the puzzle cannot be solved, Non-Comparative Justice seems the only respectable view here. But it can be solved, or so we shall argue. Here is what we propose.

The sense of puzzle arises from the implicit assumption that the morality of an act does not depend on the morality of the agent, hence the moral value of imposing a deserved punishment is independent of the imposer's motivation. This view has been famously defended by Ross and Mill, followed by a fair number of philosophers in the twentieth century.Footnote ¹¹ However, it is implausible, as shown by Stocker,Footnote ¹² SverdlikFootnote ¹³ and Scanlon.Footnote ¹⁴ Sverdlik illustrates the point by reference to four motives. First, the desire for money: some actions become morally wrong just because they are done in the hope of being paid, such as having sex with someone, or getting married, while if these same actions are carried out for different motives they are perfectly acceptable. Second, what he calls ‘trifling’: carrying out a significant moral act, such as abortion, for trifling motives (e.g. mere convenience), would be wrong, though having an abortion for more serious motives might be permissible. Third, cruelty: acting on the desire to cause pain to a living thing is usually enough to make an act wrong, even though the same act might be permissible if done for other motives. Finally, and directly relevant to the present discussion, ‘there are the motives of racism, sexism, anti-Semitism, and the like’.Footnote ¹⁵ In Sverdlik's view, an otherwise permissible act, such as a refusal to sell one's house to another, ‘is made wrong when motivated by racism’.Footnote ¹⁶

In the present context, our claim is that a bad motive might convert an otherwise just action (the imposition of a deserved punishment) into a wrong one, or, in other words, might contaminate an otherwise just punishment. We are dealing with a three-stage tale: (a) causing suffering is prima facie wrong; (b) however, when carried out within a legitimate framework of punishment, the wrongness dissolves and the suffering imposed has positive moral value; and (c) yet, when imposed for the wrong kind of motive, namely from disrespect towards the criminal, the punishment loses its redeeming power, which means that the causing of suffering turns back to be morally wrong.

The notion of disrespect is central to our analysis although we cannot offer here a complete account of it. The paradigmatic cases of disrespect we have in mind are such in which a person is treated less favourably than she would have been had she not been a member of a perceived inferior group. Her perception as such means that she is treated as less worthy than other human beings, as not fully human. Such cases are familiar from the literature of many countries and are well-researched and documented. Since judges hardly ever have personal acquaintance with the criminals they try, if they impose a harsher punishment than they would do with other (equally blameworthy) criminals, it is very often because of the perceived group – ethnic, racial, national, religious – with which the criminals are associated. Typically this would happen when judges who belong to the majority culture discriminate against members of a disadvantaged group, but it might also happen when judges of a disadvantaged group rule against a criminal from an advantaged group.

Note that disrespect towards groups of people is a wider phenomenon than one might think because it influences us in sophisticated ways of which we are often not fully aware.Footnote ¹⁷ Judge – just as other officials – might impose harsh punishments out of a sincere belief that such policy is required to fight against a type of crime which is spreading among the youth, while the truth is that she would not adopt this policy if it were not the case that 90 per cent of those involved in this criminal activity were black.Footnote ¹⁸ This is why differential punishment between seemingly similar cases should always raise a red flag urging us to ensure that the different punishments were not meted because of prejudice or disrespect.

One might suggest a broader interpretation of disrespect beyond the paradigmatic cases just mentioned. Consider a case of what one might call ‘individualized disrespect’, in which Judge acts out of some personal hatred towards a criminal. We tend to say that such cases are close enough to our working example to justify including them in our analysis, although they are pretty rare, and hence of little practical relevance.

Consider next the case of bribery. Judge imposes what ends up being a just punishment on Adam, but only because someone else (Adam's enemy) bribed her to do so. In some sense, imposing a punishment because of a bribe seems an obvious case of disrespect towards Adam, who is not sentenced on the merits or demerits of his behaviour, but on totally external considerations. However, one might argue that the disrespectful message Adam receives in this case is less ‘personal’ – Judge would have agreed to take the bribe regardless of the identity of the convict – and thus less troubling than the case of discriminatory motives.

Lastly, consider circumstances in which Judge gives Eve a more lenient punishment only because Eve is a woman, and Judge believes women should not spend as much time in jail as men. In such circumstances, the discrepancy between the punishment meted out to Adam and that meted out to Eve reflects no negative attitude towards Adam; hence Adam seems to have no grievance about his (deserved) punishment.Footnote ¹⁹ We leave this an open question.

Our view about the conditions under which an otherwise deserved punishment becomes unjust is importantly different from that offered by Stephen Nathanson more than twenty years ago. Nathanson's purpose was to explain how a deserved death penalty might nonetheless be unjust. In a formulation similar to Comparative Justice, he says that ‘whether one is treated justly or not depends on how others are treated and not solely on what one deserves’.Footnote ²⁰ But when he fleshes out this idea, it appears that looking at how others are treated is required in order to reveal irrelevant factors affecting punishment, but not necessarily discriminatory. For Nathanson, giving the death penalty for murders committed on Monday, Wednesday and Friday, while setting life imprisonment for murders committed on other days,Footnote ²¹ would be unjust because it would mean that some people would be sentenced to death based (inter alia) on factors that are totally irrelevant to their crime. From the perspective of retributive justice, punishments should be determined by the severity of the crime, not by the day of the week on which it was committed. The language of arbitrariness has indeed been quite widespread in the debate about the death penalty, especially after the US Supreme Court decision in Furman,Footnote ²² where the court held that the death penalty as then administered was unconstitutional because of its arbitrariness.

Describing some choice as arbitrary, or even worse as ‘capricious’,Footnote ²³ arouses an immediate negative response towards that choice in a way that makes other arguments against it seem redundant, hence the strength of Nathanson's proposal. However, recall the scenario mentioned earlier: he prison is full with prisoners, all of whom deserve to be there more than Adam and Eve, and it has room for only one more convict. Judge must decide whether Adam or Eve will go to prison or go home (no other punishment is available). She tosses a coin and unfortunately for Adam he loses and goes to prison. In Nathanson's view, this would be a case of an ‘irrelevant factor’ (the side the coin fell on) determining Adam's punishment, so Adam would have a justice-based complaint against it. But given that the punishment Adam got was what he in fact deserved, and given that the procedure was free of discrimination, in our view he has no claim against it.Footnote ²⁴ The same goes for Nathanson's example mentioned above. Suppose that each of A, B, C, D, E, F and G is a murderer who truly deserves the death penalty. Suppose – as seems to be the case in practice – that executing criminals is much more expensive than imposing life-imprisonment upon them. So the state decides randomly to assign 3/7 of the criminals the punishment they deserve, namely, a death penalty, and 4/7 of them a less harsh punishment, namely life-imprisonment. The choice between convicts is made on grounds which are completely arbitrary, namely according to the day on which the murder was committed. Since A, B and C committed their crime on Monday, Wednesday and Friday, respectively, they are executed. Since D, E, F, and G committed murder on other days they go to jail. This procedure seems no less arbitrary than tossing a coin. What complaint might A, B or C have? They would not be able to complain that the punishment they got was undeserved because ex hypothesi it was deserved. Neither could they claim that the actual imposition of the death penalty upon them was an expression of disrespect or discrimination. They could claim that they had bad luck, as no doubt they did, but suffering bad luck is not the same as suffering injustice, a point that weekly losers in lotteries find hard to accept.

Contrary to Nathanson, then, in our view the irrelevance of the factors determining the actual punishment is not sufficient to disqualify an otherwise deserved punishment. Such disqualification requires irrelevance of a special kind, namely, an attitude of disrespect towards the criminal (which is causally effective in determining the punishment). Only then would the punishment lose its moral value and become wrong.Footnote ²⁵ Let's call this view ‘respectarianism’, which we offer as an alternative to comparative egalitarianism.

Note that while negative motives contaminate otherwise just punishment, no especially positive motives are required to make a punishment just. In particular, Judge need not explicitly intend to realize justice when she sentences some criminal to jail. She might be so burdened by her work that she has no time to construct any clear goal and will just be ‘doing her job’. Things are more ambiguous when Judge acts on motives which are improper from a retributive justice perspective, e.g. to promote deterrence or rehabilitation. On the one hand, because of this impropriety – using the criminal as a means to achieve social goals – one would think that the punishment should be invalidated. On the other hand, there seems to be a sea of difference between wrong motives of the kind discussed above such as racist ones and motivation to advance worthy social goals. While it is true that, for the retributivists, deterrence and rehabilitation might be seen as illegitimate concerns in punishment, a judgment based on such grounds is surely not as morally corrupt – and potentially not corrupt at all – as a judgment stemming from racist ones. Answering this question requires one to have a more fully developed theory than we offer here regarding the type of improper motives that invalidate otherwise legitimate acts.

A helpful way of bringing to light the difference between comparativist-egalitarianism and respectarianism is through Elisa Holmes's distinction between strict equality principles and anti-discriminatory ones.Footnote ²⁶ A strict equality principle is of the form ‘all As should be treated as some As are treated in respect of x’, e.g. ‘give all children the same pocket money you give to some children’.Footnote ²⁷ An anti-discriminatory principle forbids treating people on certain grounds; it prohibits denying benefits from them because of their race, sex, age, and so on. To show the difference between these principles, Holmes suggests the following example. Suppose that by mistake a parent gives one of her children more pocket money than she gives another. This would be a violation of the strict equality principle that requires that all children get the same as some do and would provide a sound basis for the child who received less to complain. The discrimination principle, however, provides the child with no basis for complaint because it is not the case that he received less on a forbidden ground; it was a pure mistake. Hence, viewed from the perspective of discrimination, the parent would have no reason to redistribute the pocket money.

This means, explains Holmes, that the anti-discrimination principle ‘uses a comparison only heuristically’,Footnote ²⁸ as a way to find out whether forbidden grounds were at play. If they were not, then the very fact that the result is unequal (one child receiving more pocket money than his sibling) is not morally troublesome in so far as discrimination is concerned.

Strict equality fits comparativist-egalitarianism, which requires that all criminals who have committed some crime C (with more or less the same criminal intent, same lack of exonerating conditions, etc.) should be treated, i.e. punished, as some of them are. If some receive five years in jail, others should receive no more – regardless of how this five-year sentence came about. In contrast, the view we propose can be seen as an anti-discriminatory one. It prohibits treating criminals on grounds like race, sex, age, and so on, and calls for cancelling, or reversing, any hardship imposed on them on such grounds. Often, we come to know about such grounds only by comparing the punishment that criminal A1 received to that A2 received, but the comparison is used only heuristically and has no intrinsic value. In any case, if there are no traces of illicit grounds, the actual difference in punishment is probably not discriminatory and therefore not as troubling.

In conclusion, inequality in punishment is disturbing for two reasons: (a) because of the doubt it casts on whether the relevant criminals really got what they deserved; and (b) because of the disrespect it might express towards the offender receiving the harsher treatment. These two worries correspond to the two fundamental conditions for just punishment: that it fits the crime and that it is not imposed with improper motivation.Footnote ²⁹

Let us point out what we see as the advantages of the view we just laid out over that of egalitarian comparativists:

1. 1 Our view helps to understand why inequality in punishment is worrisome even in the circumstance in which one gets the punishment one truly deserves. It is because it expresses disrespect, which is a forbidden ground for punishment. The explanation for the mysterious fact that a deserved punishment loses its moral value when applied because of the wrong motivation has to do with a more general phenomenon in moral life in which the motivation behind an act affects its moral value.

2. 2 That respect is crucial for just punishment is not surprising given the fact that theories of retributive justice assign a crucial role to respect, whether in justifying the very practice of punishment (which, à la Kant, shows respect for the wrongdoer's free choicesFootnote ³⁰) or in justifying its most basic constraint (never to ‘punish’ the innocent).

3. 3 Relatedly, if punishment is essentially about respecting the free choices of the criminal, then, in the cases under discussion, we are not balancing two different values – desert and equality – as assumed by egalitarian comparativists, but ultimately one value, respect.

4. 4 Our proposal accounts better than egalitarian comparativists for the wrongness of levelling up. If the reason that Adam should receive the same punishment as Eve (‘levelling down’ his punishment) is the importance of equality in punishment, then this importance must also lead to levelling up, in case Eve got a harsher (and of course undeserved) punishment, say fifteen years in jail. Hurka objects to this implication,Footnote ³¹ but does not provide an argument for his objection. In our view, the explanation is straightforward: the purpose of levelling down is to defuse the effects of disrespect, which led to Adam receiving a harsher (though deserved!) punishment than the one he would have received had he not been, for instance, a black person in the US. In contrast, nothing will be gained by imposing upon Adam a harsher punishment than he deserves just because such injustice was done to Eve.Footnote ³² That would be doubling the injustice (giving an undeserved punishment to both Eve and Adam),Footnote ³³ not diffusing it.

To conclude this section, let us return to the opening example of Adam and Eve.

According to our approach, if Adam's punishment was driven by disrespect then it should be reduced. But by how much? In the definition of Comparative Justice in section I, we assumed that A1 should not receive a punishment that is harsher than P2, presumably even if P2 is much less than the deserved punishment. In this view, retributive justice requires ‘levelling down’ A1's punishment all the way to A2’s, which means that if A2 was spared from any time in jail, A1 should be spared as well. We might now call this view ‘strict respectarianism’. It holds that the value of neutralizing the effects of disrespect trumps the value of imposing upon wrongdoers the punishment they deserve, regardless of the absolute and comparative levels of P1 and P2. (By absolute level we mean the non-comparative deserved levels of punishments and by the comparative levels we mean the magnitude of the relative inequality between these punishments.)

However, a more moderate reading of Comparative Justice is possible, which we shall name ‘moderate respectarianism’. According to the reading, since both desert and respect are fundamental requirements of retributive justice, one must strike a reasonable balance between them rather than let the concern for respect prevail at all costs. The moderate version of respectarianism also calls for levelling down of A1's punishment but not necessarily all the way down to A2's punishment. It attempts to balance the loss in desert involved in giving A1 less than he deserves against the gain in respect involved in levelling down A1's punishment.

These two versions can be formulated as follows:

The strict version seems to lead to an unwelcome implication. It would require that everyone with x level of culpability receive the lowest punishment that anyone under the same legal regime has ever received for committing an act with that level of culpability. Moreover, since there is no reason to make punishment levels time-bound, if anyone ever received an especially low punishment with that level of culpability, it would be hard ever to correct the problem.Footnote ³⁴ Of these two views, we lean towards the moderate one. First, if the harsher punishment imposed on A1 is a result of relatively minor prejudice against him and if, ex hypothesi, this is the punishment he truly deserves, it seems too radical to require that A1's punishment would be levelled all the way down to A2. It makes more sense to try to strike some kind of balance between desert-based respect and motivation-based respect, which in practice would mean a significant levelling down of A1's punishment, though not necessarily all the way down to A2.

III. SUMMARY AND CONCLUSIONS

The main purpose of our article was to explain how a deserved but unequal punishment might be unjust. We said very little about the implications of our proposal for public policy. We definitely do not suggest customizing ‘motivation tests’ for all judges and public officials in order to make sure that the actions they carry out within their capacity are not tainted by disrespect. Nonetheless our proposal is not merely theoretical. When some group is (and perhaps even if it were in the past) a clear target of prejudice and discrimination, its members have a strong claim against state officials not to be treated more harshly than other citizens and, in case they do suffer such treatment, to have the results of such treatment undone. Given the over-representation of blacks in US death rows, and given the continuous discrimination against blacks in the US in general, and in the criminal justice system in particular, our argument in this article implies a strong reason to convert their death penalty sentence to life imprisonment.

As indicated above, unequal treatment is suspected to indicate lack of desert and lack of respect not only at the sentencing phase, but at other junctures in the criminal process as well. We should now add that it plays such a role in other contexts too. In the legal sphere, one naturally thinks of administrative law. Suppose the mayor closes down some store on the basis of the claim that it failed to comply with some regulations regarding the environment, but she avoids doing so to other stores that behaved the same way. Suppose further that the only store closed down on these grounds belongs to a black owner. It follows from the above discussion that such an act, though formally within the mayor's authority, might be wrong and, if so, should be redressed. One way of explaining why this is so is that when public officials use power against people in their jurisdiction, it is not their own power they are using but the power entrusted to them by the state. When they act upon racial or other improper motives or when they aim at settling a personal feud with some rival, they thereby abandon their role as officials, the result being that their acts cannot enjoy the protection of the law and thus lose their legitimacy.

Dilemmas of this kind exist outside the legal sphere too. Suppose a teacher tells her students that they must submit some paper by 1 March and that papers submitted after this date will not be accepted. Suppose further that some student is late in submitting her paper and consequently fails the course. On the face of it, this seems a legitimate use of the teacher's power. But now suppose that ten other students were also late in submitting their papers yet the teacher accepted them and treated them as if they were handed in on time. Finally, suppose that the student whose work was rejected (and only she) belonged to some minority group. According to the line of argument developed above, although the sanction imposed on the student was deserved and in some sense fair, the teacher's improper motive emptied it of any legitimacy it might have had and transformed it into a wrongful act. The student has a justice-based claim to have her paper accepted and to receive the deserved grade for the course, just like all the other students.

The same analysis would apply to non-legal sanctions on a national or international level, such as politically motivated boycotts or sanctions against organizations or countries. In general, such boycotts or sanctions are problematic because of their indiscriminate nature; they harm both the citizens who support and those who object to the government's policy. Nonetheless, in some extreme cases they might be justified. However, if they are used only against one country and not against others that deserve it no less, then the country suffering the sanctions has a strong justice-based claim against them. To justify such selective use, one might argue that it is impossible to fight simultaneously – by boycott or by other means – against all sorts of evil; hence there is nothing improper in focusing on only one corrupt country and not on all. This move is analogous to the case suggested above in which for practical reasons it was impossible to send both Adam and Eve to jail. But the solution we offered to that case applies here too, namely picking randomly between the available options by tossing a coin or some other means. If, because of resource constraints, boycotts or sanctions can be used only against one group, although ten groups really deserve it, then they definitely should not be intentionally imposed on the one group with a history of being oppressed and discriminated against. Selective boycotting, just like selective punishment, too often attests to improper motivation, which contaminates the relevant act and makes it wrong.

A central assumption throughout our article which we couldn't defend properly here concerned the relation between luck and justice. In our view, the very fact that some punishment is based on luck does not make it unjust. When there is no room in jail for two convicted criminals and a lottery determines who goes to jail and who goes home, this random procedure silences any complaints about injustice raised by the unlucky criminal. Moreover, from a retributive justice point of view, it is not unthinkable that the court should toss a coin to determine the exact sentence within the legal range of deserved punishments, for example, as a means for a self-conscious judge to neutralize her racial or other biases (although other considerations, such as the appearance of justice or propriety, may advise against such policy).

Finally, a similar argument to ours can be proposed in the realm of distributive justice, namely, that inequality has no intrinsic (negative) value but rather serves as a red flag for disrespect. If one person receives more than another, there is a good reason to suspect that this other person was denied her due share on the basis of illegitimate grounds; that she was discriminated against. As Frankfurt argued, the egalitarian impulse is often a disguised concern for respect. The existence of economic gaps is ‘heuristic’ rather than ‘criterial’.Footnote ³⁵ However, the analogy between retributive justice and distributive justice is more ambiguous with respect to the idea that inequality might indicate that somebody did not get what she deserves. As argued by Rawls, Feinberg, and more recently Scheffler, in the context of distribution, it makes less sense to talk about desert in a non-comparative manner, i.e. about the share in some good that one deserves regardless of the shares enjoyed by others.Footnote ³⁶ And without such a non-comparative notion there is no room for worry about whether one (or more) of the relevant subjects did not get what they deserve. However, this asymmetry in the role of desert between retributive justice and distributive justice has been recently challenged by others (like Hurka) arguing that desert does have a non-comparative dimension in distributive justice too.Footnote ³⁷ Whether this is so, and what this would mean about the relation between distributive justice and retributive justice, are questions that we leave for some other day.Footnote ³⁸