Philosophers and lawyers often inquire when discrimination is wrong, or what makes for wrongful discrimination.Footnote ¹ With few notable exceptions,Footnote ² the question why discrimination might be wrong or bad is seldom asked. This is perhaps not all that surprising. Finding out when discrimination is bad is of obvious practical importance (e.g. for legal purposes). Inquiring into why discrimination is bad might not be of equivalent import. It does not tell us, at least not directly, what selectors are allowed or not allowed to do. Still, I hope this article will show that inquiring into why discrimination is bad is not only interesting in its own right, but also that it can tell us something about when discrimination is bad.Footnote ³

Why, then, is discrimination bad? Discrimination is bad as such, I want to claim, because and only because it undermines equality of opportunity. This statement calls for some qualification and clarification. I said that I speak of why discrimination is bad, and not of when it is wrong. This means that I do not search here for a comprehensive definition of discrimination. Rather, I try to trace what, if anything, is bad about it. One implication of this is that I search here for a necessary condition of discrimination, not a sufficient one. A second point of clarification follows: by searching for the root cause of the badness of discrimination I do not presume to be saying something about what makes discrimination unique. In other words, I am not searching for what distinguishes discrimination from other incidents of injustice, but rather for what, at bottom, makes it bad. Third, I refer to the badness of discrimination ‘as such’. Particular incidents of discrimination might be bad for a variety of reasons. In speaking of the badness of discrimination as such I search for what source of badness, if anything, is common to all incidents of discrimination. Fourth, by ‘equality of opportunity’ I mean not equality of opportunity for a certain job or position, but rather overall equality of opportunity (e.g. equality of opportunity for welfare). Fifth, I assume (without argument) equality of opportunity to be a component of fairness (or distributive justice). I shall therefore use ‘equality of opportunity’, ‘fairness’, and ‘distributive justice’ interchangeably, even though they are no doubt distinct. Finally, in speaking of ‘equality of opportunity’ I do not presume to know what the reference group for assessing it is: one's city, country or the globe as a whole. I claim that discrimination is bad when it undermines equality of opportunity, in whatever the relevant reference group is taken to be.

Discrimination, then, is bad when and only when it upsets equality of opportunity. The view may not seem all that controversial, perhaps. But in fact, it is almost universally rejected.Footnote ⁴ Indeed, some have gone so far as to say that if there is a ‘connection between anti-discrimination laws and equality it is at best negligible, and in any event is insufficient to count as a justification’.Footnote ⁵ The ‘equality of opportunity’ account seems in fact to be very much the minority view.Footnote ⁶ I shall begin my defence of that view by looking into alternative accounts of what makes discrimination bad, and showing that none of them is successful in accounting for the badness of discrimination as such. This is the concern of section I, and it thus lays the ground for the defence of the equality of opportunity account. Section II takes on the challenge (to that thesis) presented in cases which appear discriminatory despite not involving inequality of opportunity. The notable case here is that of segregation. Finally, in section III, I address some further potentially tough cases for the equality of opportunity account. These concern the confinement of discrimination to salient groups, ‘buying off’ discriminatees by means of financial compensation, ‘discrimination’ in the selection of life partners, and the implications of discrimination for the duties of employers.

I. WHY IS DISCRIMINATION BAD?

Discrimination could be bad for a variety of reasons. We may initially divide relevant accounts between those that locate the badness of discrimination in the harm to the person being discriminated against (the discriminatee), as opposed to accounts that locate it outside her.Footnote ⁷ The latter group of reasons, which I first want to address, may be further divided into two: badness which resides with the discriminator (or selector), and badness residing in society as a whole. Looking at the first sub-group, we may say that discrimination is bad because it corrupts the moral character of the discriminator. When discrimination is wrong (and in this article I shall always refer to ‘discrimination’ in the pejorative, rather than the literal and neutral sense, namely to imply wrongful discrimination) then we might think that to commit it is to undermine one's own moral virtue. This is familiar enough to us from characterizations of discrimination that focus on the morally repugnant intentions of selectors.Footnote ⁸ Bad intentions are therefore a necessary condition of discrimination under this account. But this is disputable. Think of the following example (introduced by Kasper Lippert-Rasmussen).Footnote ⁹ Consider an employer who, while not a Scandinavian herself, has a somewhat inflated view of Scandinavians. Consequently, whenever a Scandinavian applicant comes along, she gives careful consideration to all other candidates but then, perhaps somewhat unwittingly, judges the Nordic applicant to be superior, even when that is blatantly not the case. The selector has an inflated view of Scandinavians, while having a correct and unprejudiced view of all others. This seems like a case of discrimination, even though there is nothing repugnant about the employer's intentions (or beliefs). Corrupting the moral character of the selector is therefore a type of bad which, while plausibly associated with discrimination, does not seem to be true for discrimination as such.

Consider now accounts that locate the badness of discrimination in society as a whole. One such account purporting to explain why discrimination is bad concerns efficiency. Namely, discrimination is bad because it deprives society of the benefit of having a workforce made up of the best qualified.Footnote ¹⁰ We are, indeed, often repelled by discrimination for the waste which it typically brings about. But it is possible to think of cases of discrimination which are not necessarily inefficient. Think of cases such as discrimination against women for reasons of maternity, statistical discrimination against minority groups who are likelier to engage in absenteeism or crime, and discrimination on the basis of reactive attitudes (not hiring someone based on the judgement that one's clients prefer not to be served by members of these minorities). We think of these practices as discriminatory and yet they are not obviously inefficient.

A third source of badness is one which it is not so easy to categorize (at least not initially) in terms of whether the badness resides within or outside the discriminatee. This particular account of badness says that discrimination is bad because it manifests a lie.Footnote ¹¹ Discrimination represents the discriminatee as lacking in qualifications which in fact she does have. This may lead to depicting her as having a lower status than she in fact has. Notice that this claim is different from a related, and more common, account of discrimination (which we shall discuss shortly), according to which discrimination is bad because it is disrespectful. According to the claim under investigation here, the badness lies with the false representation, independently of any disrespect it may convey.

Now, this badness account could be interpreted in (at least) two ways. First, it could be said that it is harmful for individuals to be falsely represented. And second, it may hold that it is bad when actions (hiring) convey untruths, independently of any harm they may cause. In other words, a world which contains actions that manifest some untruth is worse than one which does not. Notice the following implication of the first variant of the claim. Since the badness resides in the lie (as distinct, I said, from a message of disrespect) this badness account is forced to condemn (in equal measures, mind you) falsely representing someone as having better qualifications than she actually has.Footnote ¹² (It might be said, in reply to this point, that representing Jones as better than she is invariably entails representing Smith as worse than he in fact is. But as we already saw, this does not follow: I may discriminate against Smith, having a correct view of her, by virtue of having an inflated view of Jones.)Footnote ¹³ Yet, it does not seem harmful to the person selected to be represented as better qualified than she really is. Harmful lies cannot therefore account for the badness of discrimination.

All this does not, however, undermine the second variant of the falseness account. That version of the claim says that it is simply bad for there to be acts which manifest lies (regardless of harms to the misrepresented individual). The problem with this claim, however, is that while it portrays a plausible enough bad, it is one which does not seem necessarily discriminatory. While it might be bad (because false) to represent Jones as more qualified than she really is, this does not seem to explain the badness of discrimination, of all things. If and when discrimination is bad it is surely so because someone else (Smith) has been discriminated against. And if that is the case, then surely the badness of the overall act must be somehow associated with that other (disadvantaged) person.

A fourth potential source of badness is the already mentioned disrespect account. This account, again, might locate the badness of discrimination either in the discriminator or in the discriminatee. Discrimination might be bad because it is bad for selectors to be disrespectful.Footnote ¹⁴ Alternatively, discrimination is bad when and because it causes the discriminatee loss of (self-)respect. We already saw that concern for the moral virtue of the selector cannot account for all cases of discrimination (the case of discrimination in favour of Nordic candidates), so let us turn to the badness residing in the person being discriminated against, and the loss of respect she allegedly suffers. This badness account is vulnerable to the obvious counter-example of covert discrimination of which the discriminatee is unaware. Think of a policy whereby only male employees receive a bonus, and where the female employees never find this out. They are unlikely to experience a reduction of self-respect and yet we probably think that they are nevertheless discriminated against.Footnote ¹⁵ It therefore cannot be the case that the underlying badness of discrimination lies with some demeaning impact.Footnote ¹⁶

A fifth account locates the badness of discrimination squarely with the discriminatee. It says that discrimination is bad for reasons of desert, namely when and because it deprives the most qualified person of the job to which she is entitled.Footnote ¹⁷ This account relies, obviously, on a series of controversial premises. First, it must assume that desert is, in fact, a principle of justice, which is far from obvious.Footnote ¹⁸ Second, it assumes (the no less controversial view) that individuals deserve rewards for (at least in part) their natural talents. And third, it assumes that these rewards must take the shape of jobs, of all things. Notice that this account locates the source of badness not simply in the inefficiency that society must endure owing to not recruiting the best qualified (which we have already examined and dismissed), but rather in the wrong committed against the talented who are arguably deserving of those jobs.

Even setting aside the series of controversial premises that one would have to swallow in order to accept this account, it is still doubtful that the principle of desert successfully explains the badness of discrimination as such. Think, for example, of a particular job for which there are several qualified candidates, and suppose that one of them stands out in such a way which makes it pretty obvious from the start that she is the most qualified for the job. We might still think it wrong, because discriminatory, at least in some cases, not to open up the competition for the job and not to give each application at least some consideration. If this is so, then it would suggest that the badness of discrimination is motivated by reasons other than rewarding desert (e.g. by respect for applicants, or by some idea of equal chances to compete for the job).Footnote ¹⁹ To give a perhaps even stronger case, suppose (to use Andrew Mason's example) that the best qualified person for the job is very shy, and consequently would refrain from applying for it. Or suppose, further, that even if she did apply she would do badly on the interview due to her shyness.Footnote ²⁰ On a strict desert view, this would seem to call for headhunting rather than for an open competition. But once again, it seems discriminatory (at least in some cases) to deny the other candidates the chance to compete in this case. The desert of the talented, thus, does not seem to capture the badness of discrimination.

The final competing badness account I can think of is the view that discrimination is bad because it is bad for people's lives to be determined by other people's prejudice. Paradigmatic cases of discrimination, we may notice, typically do involve prejudice (e.g. on the basis of race or sex). Thus, some philosophers hold that discrimination is bad because we do not ‘want to see people succeeding or failing because of other people's prejudice’.Footnote ²¹ Now, it would be hard to dispute that for people to fail because of other people's prejudice is bad. But why exactly is it bad for people to succeed on the basis of someone else's prejudice? Some of the things we have already mentioned may account for this: it is inefficient to hire on the basis of prejudice; prejudicial hiring does not reward talent; and it is potentially demeaning to be hired for the colour of your skin rather than for your ability to perform the job. But even if all these reasons do explain why it is bad to allow someone to succeed on the basis of prejudice, still none of them explains why it is discriminatory to do so. In other words, on all these occasions, hiring someone based on prejudice can indeed potentially be bad but is not obviously discriminatory (at least not on its own). Once again, one reason why hiring X on prejudice might be both bad and discriminatory is that it implies the failure of Y to get the job, and through prejudice, which is certainly bad (and arguably discriminatory). But if this is where the badness of discrimination resides then we are back at a concern for equality of opportunity. Hiring X on prejudice is bad, on this account, because Y did not really get an equal opportunity to compete for the job. So the claim that discrimination is bad for reasons of prejudice in fact boils down to a concern for equality of opportunity.Footnote ²² Prejudice, then, does not provide an independent (let alone convincing) badness account.

None of the potential sources of badness reviewed so far captures what is bad about discrimination. Now, it might be suggested that discrimination is bad for a multitude of compatible, overlapping reasons, and that there is no point searching for one source of badness that captures all incidents of discrimination.Footnote ²³ If this is true, it would follow that it is wrong to dismiss this or that account of badness when it fails to account for the badness of a particular incidence of discrimination. I find this implausible. It is certainly true that particular incidents of discrimination will be bad for a variety of reasons. But there must be at least one reason, I maintain, underlying why discrimination is bad as such. Here is why.

Consider the implications that would follow were discrimination to have only overlapping, non-exhaustive sources of badness, some accounting for the badness of incidence X and some accounting for the badness of incidence Y. In many particular incidents of discrimination the different sources of badness clash. Suppose an employer adopts a consciously racist hiring policy, preferring members of his own race just for that reason. Virtue-wise the action seems bad. It could also be bad for reasons of efficiency: preferring members of his own race, the employer overlooks other, perhaps better qualified candidates. (And the same could be said for the deceit and desert accounts: the action does not truthfully represent the fitness of the appointee for the job, and it does not reward the desert of those truly suitable for the job.) Suppose, however, that the action is desirable on account of equality of opportunity. The employer in question is hiring disproportionately more (sufficiently qualified) blacks (like herself). To decide such a case we would first need to know which source of badness here trumps, as it were, the others. Ought the bad intentions, inefficiency, deceit and desert trump the concern for equality of opportunity? In a case similar to this (United Steelworkers v. Weber), the US Supreme Court decreed that ‘private affirmative action’ is not discriminatory.Footnote ²⁴ We may deduce from this that, at least in the Court's view, equality of opportunity was the weightier concern, even when all other concerns were stacked up against it. What this case shows, quite apart from the centrality of equality of opportunity (henceforth EOP) for discrimination, is that even if there are overlapping sources of badness (which I doubt), there must be some hierarchy between them. That is, there must be some way of adjudicating between the different bads entailed in a particular incident of discrimination. And, if there is one account of badness which can capture all incidents of discrimination then surely it deserves to be at the pinnacle of that hierarchy. The bottom line, then, is that we are searching for a badness account that successfully explains all incidents of discrimination. We want to know, in other words, why discrimination is bad as such.

In the rest of the article, then, I want to focus on the view that discrimination is bad because and only because it is unfair, that is, because it undermines equality of opportunity. To suffer unequal opportunity (that is, to be disadvantaged relative to others) is the one and only bad accounting for the badness of discrimination.

II. DISCRIMINATION WITHOUT INEQUALITY?

To scrutinize the equality of opportunity (EOP) account we must examine cases which are (or seem) discriminatory while not involving inequality of opportunity. These would pose an obvious challenge to the view defended here. Think of the following example (discussed by Deborah Hellman).Footnote ²⁵ A high school headmaster addresses his students and orders the black students to sit on the right of the hall and the white students to the left.Footnote ²⁶ The headmaster's order appears offensive.Footnote ²⁷ Importantly, there seems to be no disadvantage here. Seats on the left are no better than seats on the right. Still, given the historical and social context of racism, the headmasters’ order strikes us as offensive.

This seems a difficult case for the EOP account, but is easily explained by its rival accounts. For example, it might be suggested that the headmaster's motive in this case is racist (which would suggest corruption of moral virtue as the source of badness). Then again, he might do so because he likes the aesthetic effect of this form of seating, not an obviously wrong motive.Footnote ²⁸ Alternatively, and more plausibly, the order is bad because demeaning to the black students.Footnote ²⁹ Other accounts, then, seem to fare better than the EOP account on this particular case. Still, we might be able to understand the badness of such acts of segregation also in terms of equality of opportunity, I suggest. The headmaster's order is wrong in that it separates two groups and, given the social context, brands one of them as inferior.Footnote ³⁰ ‘Separate but equal’ treatment is thus bad, we may say, because given a certain context it affords one group a lesser opportunity for respect. Respect is (obviously) a social good, and what segregation (when it is bad) does is to lower the social standing and opportunity for respect of the vulnerable of the two (or more) groups.

Equal but separate treatment, then, may count as discrimination when it disadvantages a vulnerable group in terms of its access to respect. A similar sort of disadvantage, we may notice, can occur also when the vulnerable group is isolated for the purpose of superior (and not merely equal) material treatment. Deborah Hellman cites from Nelson Mandela's biography his recounting how black prisoners in South Africa were compelled to wear shorts, whereas white prisoners were ordered to wear long trousers. The intention behind this policy, we may plausibly speculate, was to infantilize the black prisoners. However, given the South African climate, being allowed to wear shorts rather than long trousers could actually be considered an advantage.Footnote ³¹ The fact that black prisoners were being advantaged in terms of comfort is nevertheless consistent with them being disadvantaged in terms of social standing. It follows that superior treatment which isolates a socially vulnerable group can be discriminatory also on the equal opportunity account.

I say that separate but equal (or even better) treatment ‘can be’ discriminatory on the EOP account because not all such treatment is discriminatory to begin with. For one thing, separate treatment must be pursued for no good reason to qualify as discriminatory. The qualifier ‘for no good reason’ (known in legal terminology, confusingly I think, as the ‘bona fide requirement’)Footnote ³² is obviously important here. A sign at the front of a bus saying ‘these front seats are reserved for passengers who are black’ is offensive in a way that a sign saying ‘these front seats are reserved for the elderly and handicapped’ is not. Of course, we need not deny that even the latter could potentially be stigmatizing. Yet, when balanced with other effects of the policy (allowing better and quicker access to those whose movement is restricted) the sign does not seem repugnant. The same, of course, goes for separation of toilets between whites and blacks compared to such separation between men and women or between those who are disabled and those who are not. Another indication that the qualifier ‘for no good reason’ is essential here is found in affirmative action. That practice can imply the isolation (for the purpose of a favourable treatment) of a historically disadvantaged group, something which is potentially stigmatizing. Still, the practice might not be discriminatory precisely because the temporal isolation might better facilitate future integration. (We think affirmative action is desirable even though it might expose black medical students, say, to potential stigma, because we hope the policy would lead to a more diverse health-care workforce in the future.) The temporal potential loss of opportunity for respect is thus balanced against a greater increase in opportunities, both immediate ones (in employment) and future ones (in respect).

Let us turn to other prominent challenges to the thesis that discrimination is bad always and only for reasons of inequality of opportunity. Elisa Holmes explicitly rejects that view and offers the following argument in support. Suppose someone accidentally gives her daughter more pocket money than she gives her son (and suppose that for some reason this cannot be rectified immediately). We do not normally think of this as an act of discrimination. This shows, says Holmes, that discrimination is not about equality (but rather about having the correct attitude or intention, say).Footnote ³³ Contra Holmes, this example does not refute the EOP account of discrimination, for the simple reason that that account never purported to claim that every instance of inequality of opportunity is thereby discriminatory. The claim, recall, was that inequality of opportunity is a necessary, not a sufficient, condition for discrimination. The suggestion that inequality-inducing accidental gifts are not discriminatory therefore does not undermine the EOP account of discrimination.

Here is another challenge to the EOP view of discrimination. In the famous Palmer v. Thompson case a mayor in the south of the United States, following Federal desegregation legislation, decided to close down a public swimming pool, stating that he wanted to end what to him was the offensive spectacle of whites and blacks splashing in the same waters. This seems to present a challenge to my thesis since there is no obvious inequality of opportunity here and yet the mayor's action is manifestly condemnable. But we need to know more about the case to see if it represents a challenge to the EOP thesis. Suppose that prior to the closure the public pool was used disproportionately more by blacks (say, because, proportionally, whites had greater access to private swimming pools). In that case the order to close the pool does undermine equality of opportunity, and would thus be (potentially) discriminatory also on my account. If the closure did not undermine equality of opportunity (suppose whites and blacks were truly equally harmed by it) then the mayor's decision might be repugnant (simply for the racist views it manifests), but not obviously discriminatory.Footnote ³⁴ This touches on an obvious point, but one which is nevertheless worth repeating: not every repugnant policy is a discriminatory one. Policies can be morally wrong for reasons other than discrimination.

Another challenge to the view that discrimination is bad for reasons of equality of opportunity is offered by Matt Cavanagh. Suppose, he says, that an island is hit by some natural disaster and that you are a captain of a rescue ship that can save people in either of the towns in that island. To simplify matters, assume that all relevant features (number of inhabitants, distance from the rescue ship, etc.) are equal between the two towns. Suppose also that the captain has personal attachment to neither town. Under those circumstances, fairness requires, it is safe to assume, that the captain should quickly flip a coin to decide which town to head for. All the same, says Cavanagh, we do not think it impermissible if the captain opts to skip the coin toss and rather to decide on a whim to head to one of the towns. If we accept the story so far (which I do), Cavanagh claims, then it follows that the moral requirement underlying the limits placed on the captain's conduct are not those of equality of opportunity. What we care about in this case is that the fate of the two towns would not depend on the prejudice of the selector. We would object, for example, if the captain chose one town over another because it contained fewer Jews. Our underlying concern here is thus the avoidance of prejudice and not ensuring that the towns’ inhabitants have some equal opportunity to survive the disaster. This is evidenced by our willingness to forego the coin toss, or so says Cavanagh.Footnote ³⁵ I do not find this persuasive, and for the following simple reason. Although a coin toss is, to be sure, a good way of ensuring equal chances, it is by no means the only way. Allowing a selector to act on a whim, when it is truly a whim, also accords equal chances. If the captain has no access to information that might bias his decision (anything from the composition of races to the number of individuals in each town) then his choice seems to be truly made on a whim. And choosing on a whim, crucially, is a procedure that distributes equal chances. It is, in this sense, precisely like flipping a coin.Footnote ³⁶

Consider, finally, the following counterexample. Think of a society made up of five ethnic groups, whereby group A discriminates against group B, group B discriminates against C, and so forth. There is no inequality of opportunity here but the state of affairs nevertheless seems repugnant.Footnote ³⁷ This proves, some may say, that the badness of discrimination is owed to reasons other than some unequal distribution of opportunities.Footnote ³⁸ One thing we could say here is that this state of affairs involves (avoidable) dignitary harm. While each is equal to others in terms of opportunities, each has also suffered dignitary harm. If this state of affairs is bad, then, it is so because the treatment is demeaning. Crucially, however, and in difference from the paradigmatic cases of segregation, here the loss of respect is suffered equally by all. The EOP account therefore fails to detect anything bad about this state of affairs. In my view, however, the example represents a case which is bad (or even wrongful), but not discriminatory. It is bad because it entails some universal reduction in respect, but it is not discriminatory.

To see this, consider the following variant of the above story. Suppose that group A is viewed as quirky and idiosyncratic by group B, but its judgment is highly esteemed by groups C to E. And suppose that group B is viewed as quirky and idiosyncratic by group C (the group it will end up discriminating against), but highly valued by everyone else, and so forth. It might be the case that when A discriminates against B (and B against C, and so forth) it causes a slight reduction in B's self-respect but an increase in C's through E's self-respect (and one whose accumulated amount is larger than the decrease in B's). It is easy to see where this is going. We have here what seems like multiple discrimination (consistent, recall, with equality of opportunity) which results, on the whole, in an increase in the absolute amount of self-respect in society (or, at the very least, no reduction in that absolute amount). Is there something bad about this state of affairs? Perhaps there is, but if so, it does not seem to me to be for reasons of discrimination (but if anything, to something like social cohesion), and thus does not pose a problem for the EOP account.

III. SOME (FURTHER) TOUGH ISSUES

Defending the view that the badness of discrimination resides exclusively in inequality of opportunity entails, mainly, fending off objections and counterexamples. Before resuming that undertaking let me quickly point out three, perhaps rather obvious, advantages of the thesis. These concern disparate impact, affirmative action and so-called age-discrimination.

Consider first disparate impact. This occurs when a seemingly neutral selection procedure ends up disadvantaging a vulnerable group. A paradigmatic example is the inclusion of a physical strength test in hiring for firefighters, thus ending up excluding women from the job. Cases of disparate impact are normally seen as presenting a challenge to intent- or respect-based (or in general, deontological) accounts of discrimination.Footnote ³⁹ And it is not difficult to see why. Such cases show that some action may constitute wrongful discrimination even if there was no discriminatory intent and even if there was no loss of respect, either manifested or incurred. On my account, in contrast, discrimination is wrong because it exacerbates inequality of opportunity, which is precisely what disparate impact does and, perhaps more pertinently, why we think it to be bad.

Think, next, of affirmative action. Affirmative action arguably poses a challenge to desert-based accounts of discrimination.Footnote ⁴⁰ In contrast, affirmative action is normally pursued for the sake of equality of opportunity, and is thus easily accounted for on my account. Consider finally so-called ageism or age-discrimination. My account sides with the practice of many European countries which opted not to follow the United States in abolishing mandatory retirement. Mandatory retirement is not discriminatory precisely because, over their whole life, the old are not disadvantaged (compared to the young) with respect to employment opportunities.Footnote ⁴¹

As promised, I want to end my inquiry by presenting four issues that pose a potential challenge for the EOP account. These are: the confinement of discrimination to salient groups, ‘paying off’ discriminatees through material compensation, ‘discrimination’ in the selection of life partners, and the duties of employers. Here is the first issue. Accounts of discrimination often rely on what is called ‘socially salient groups’ or ‘groups with HSD (history of social disadvantage) traits’.Footnote ⁴² Socially salient groups are groups that are recognizable by their distinguishing trait, and that have a history of being disadvantaged on account of it. The appeal to social saliency helps explain why discriminating against a person based on salient group membership (race, say) is much more offensive than rejecting her on the basis of a non-salient trait (having freckles, say). Arguably, speaking of the badness of discrimination exclusively in terms of equality of opportunity fails to explain why discrimination on the basis of salient features is so much worse than discrimination based on mundane features (even when both of them are arbitrary).Footnote ⁴³ There are two points to be made in reply here. First, perhaps the main reason why ‘socially-salient groups’ is central to discrimination is precisely the fact that these groups are already disadvantaged in terms of opportunities.Footnote ⁴⁴ It is no wonder, then, that accounts of discrimination accord these groups a prominent place. Second, as mentioned in the previous section, one of the opportunities that individuals potentially care about is the opportunity to gain the respect of others. Crucially, discrimination would typically cause a greater disadvantage (in terms of respect) when it is directed at members of salient groups compared to when it is directed at members of ad hoc groups (such as ‘the untalented’, ‘people who are not family members of mine’ and ‘people whose surname contains seven letters’). In other words, disadvantageous hiring on the basis of salient group membership presents a much greater undermining of equality of opportunity with regard to a particular good, namely access to self-respect, compared to discrimination on the basis of non-salient traits.

Here is a second difficult issue. Suppose someone is being denied a job based on her skin colour but is then offered some cash as compensation. And suppose that that package of compensation makes her as well off as others in that given society. Since the discriminatee is not made worse off here, and given that my account makes inequality of opportunity a necessary condition, it would imply that there is no discrimination involved. Still, the practice seems quite repugnant. I bite the bullet on this: if, given the choice, Smith would not turn away the package of compensation in favour of the denied job (thereby indicating that there is no inequality of welfare) then this indeed does not strike me as a case of discrimination. Note, of course, that we are assuming an ideal situation, whereby Smith's preference for the compensation package is fully informed and voluntary.Footnote ⁴⁵

The third tough issue (for the EOP thesis) that I want to examine concerns discrimination in private spheres of life, where the paradigmatic case is that of selecting life partners. Few would endorse the view that the state should regulate the way we choose life partners (beyond, say, prohibiting us from marrying an underaged person, or a blood-relation of ours). But importantly, that does not mean that the choices individuals make in that sphere are beyond reproach.Footnote ⁴⁶ In particular, we may think that there is something morally wrong about individuals choosing a priori never to marry individuals of a certain race. If this is a form of discrimination (allowing that it is one that the state should, nevertheless, not attempt to curtail), then it is one that does not concern equality of opportunity, at least not obviously so.Footnote ⁴⁷ If we think there is something repugnant about individuals making such choices then the problem does not seem to rest with some putative inequality of opportunity (to marry) but is rather grounded in the morally reprehensible motive of the discriminator. I concede, in response, that there is something morally wrong in the conduct of racist grooms and brides-to-be. And I further concede that these cases do not concern equality of opportunity. But I deny that they represent discrimination. The reason is that in these cases, typically, the rejected individuals cannot be said to be disadvantaged (as distinct from being harmed), with regard to marriage opportunities or otherwise. This is so because a priori the selector is not in any particular position of power over the discriminatee.Footnote ⁴⁸

The fourth and final challenge to the EOP account shifts the focus back to the discriminating agent. My claim in this article has been that discrimination is bad because and only because it is unfair (because it undermines equality of opportunity). This seems to suggest, the critic might say, that what is wrong about employers who discriminate is that they fail to promote distributive justice. To put this differently, discriminators are at fault for not using their power (the ability to award jobs) in a way that would curtail inequality of opportunity (and moreover, for welfare!). And this, the claim goes, is counterintuitive, and in at least two ways. On the one hand, this objection reveals my account to be potentially too wide. For it seems to suggest that employers are under a greater burden to advance distributive justice than are other citizens, which is implausible.Footnote ⁴⁹ On the other hand, the EOP account is revealed by this objection to be, in a different way, too narrow: it seems to miss the point of what is wrong about the employer's conduct. Our intuition is that she is at fault not for her failure to promote a worthy social goal (equality of opportunity) but for something else and much more repugnant (e.g. showing disrespect to the applicant, allowing prejudice to affect her decision, etc.).

One thing to note about this objection is that it may apply to other badness accounts as well. The view that discrimination as such is bad because of efficiency reasons (it fails to appoint the person who would be the most useful one in carrying out the job) would suggest, arguably, that the sin of discriminators is the failure to promote the greater good. This equally seems to miss the point of what is repugnant about the very act of discrimination. This does not, of course, get the EOP account off the hook. Recall, then, that my claim was that discrimination as such is bad because and only because it undermines equality of opportunity. I do not deny, I have said, that particular incidents of discrimination might be repugnant for additional reasons (such as disrespect or inefficiency). A discriminatory act might be repugnant for a variety of reasons, but there is only one reason, I said, that explains why discrimination as such is bad. This is why, I think, the EOP account is innocent of the narrowness objection: it never purported to, and need not, capture everything that is condemnable about each and every instance of discrimination. Because it never purported to exhaust everything which is bad about particular acts of discrimination, it also never meant to exhaust discriminators’ culpability. To sum up the point: the badness of discrimination as such is tied up with the person being discriminated against and is always one and the same. The extent of the culpability of the selector, on the other hand, will vary depending on her intent, knowledge, the grounds on which she chose to discriminate, etc. This, in my view, shows the strength rather than weakness of the EOP account. For, it categorizes as discriminatory even actions that appear to have a rather minimal amount of culpability on the part of the selector (such as those concerning disparate impact, say).

Let us turn, finally, to the wideness objection. The objection said that locating the badness of discrimination in equality of opportunity implies that employers are under a greater duty, compared to non-employers, to promote distributive justice. But we can now see that this also does not follow. The accusation would have been true had my claim been that undermining equality of opportunity is a sufficient condition for discrimination. This would have entailed that the sin of the selector was in not promoting social justice. But my claim, recall, was that it is only a necessary condition. Undermining equality of opportunity therefore does not purport to exhaust why discriminating selectors are culpable; what exactly they are culpable of; nor what their duty in terms of hiring was to begin with.