I. Prevailing Theories That Do Not Recognize Arbitrariness and Irrationality as Grounds of Discrimination

In a well-written book that aims to explain the nature and purpose of anti-discrimination laws, Tarunabh Khaitan observes that anti-discrimination laws are usually justified by the necessity to reconcile three, sometimes competing, core notions. The first is the notion of equality, the second is the notion of dignity, and the third is the notion of liberty.Footnote ⁶ Equality, or egalitarian conceptions of anti-discrimination laws, understand the commitment to anti-discrimination norms as an essential measure by which the law enables some kind of redistributive justice in society. Dignity, or dignitarian conceptions of anti-discrimination laws, explain them in terms of providing redress for material or symbolic harm done to one’s identity or personhood. The liberty conception relies on the obstruction discrimination constitutes to the promise of universal legal freedom and unimpeded autonomy afforded to each individual to justify the necessity for anti-discrimination laws. According to such readings, anti-discrimination laws ensure the autonomy and freedom of individuals by alleviating them from undue and unfair burdens uniquely imposed on them because of their group affiliation.

As I will point out, the common point of departure for all of these three accounts is the assumption that anti-discrimination laws must prohibit only specific grounds of discrimination. All three lay the burden of proof on the shoulders of the aggrieved party, designating specific legally recognized grounds for unlawful discrimination which must be proven for an individual to be considered a victim of discrimination. They stipulate discrimination as resting on the conceptualization of difference based on a special set of traits frequently called ‘common,’ ‘classic,’ or ‘suspicious,’ that is, traits of individuals that form ‘salient’ or vulnerable groups, such as race and gender, in explaining what discrimination is and why it is wrong. It would be foolhardy to attempt to do justice to all the theorists who adhere to these three conceptions in an overview such as this. I will, therefore, restrict my current discussion to the few contributions I find most convincing.

Amongst the egalitarians, Shlomi Segal, for instance, argues that discrimination is wrong because it exacerbates inequality of opportunity to access material resources and the means by which to acquire them, such as jobs or education, or symbolic resources such as respect and social and cultural capital. In other words, the aim of anti-discrimination law is to minimize inequality and to contribute to a legal code committed to ensuring a fair distribution of opportunities and resources in society.

Although Segal’s argument explains the idea of discrimination in terms of unequal access to opportunity, it is important for him to clarify that “perhaps the main reason why ‘social salient’ groups is central to discrimination is the fact that these groups are already disadvantaged in terms of opportunities. It is no wonder then that accounts of discrimination accord these groups a prominent place.”Footnote ⁷ In other words, salient or vulnerable groups feature prominently in Segal’s account of discrimination because they help us to identify those in a disadvantaged position who should not be subjected to discrimination exacerbating the challenges he or she faces due to their unfavorable positioning in the socio-economic, sexual, or racial hierarchy.

Deborah Hellman, a leading theorist identifying with the diginitarian perspective, argues that attributes such as race and gender define a group mistreated in the past or one which currently holds a disadvantaged position in the societal pecking order. Classifications made on the basis of these traits are more prone to be demeaning towards individuals, are more likely to be construed as derogatory, punitive, or otherwise as an expression of lesser treatment, pointing to attitudes that are unwilling to fully recognize a person’s humanity or moral worth.Footnote ⁸ To Hellman, classification on the basis of race and gender for instance are more likely to be demeaning than classifications based on traits unencumbered by similar socio-cultural significance, implying that the whole discussion must bring into consideration the socially constructed nature of the issues at hand. Whether the characteristic one uses as a basis for a distinction between individuals has the potential to demean is always determined by the role that characteristic plays in distinguishing between individuals in a social and political context.Footnote ⁹ A demeaning message is likely to be communicated when differentiating between individuals based on distinctions rooted in traits such as race and gender because they resonate familiar forms of discrimination towards members of vulnerable groups that have been used to disenfranchise them of rights, whether formally or only de facto, in the past.Footnote ¹⁰

One of the instances that Hellman provides for a distinction that is discriminatory because of the demeaning message it sends concerns employers who require only female employees to wear makeup. This distinction demeans women because, in our culture, the makeup requirement is associated with a certain understanding of women’s bodies as objects for adornment and enjoyment by others in a way that distinguishes it from other requirements that are not demeaning such as hair-length requirements.Footnote ¹¹

The liberty approach is well encapsulated by Sophia Moreau, who reminds us that anti-discrimination laws protect our freedom to choose our own path in life by insulating us from “normatively extraneous features of us, such as our skin color or gender.”Footnote ¹² Individuals must be free to make choices, such as where to live and where to work, regardless of the extraneous traits they happen to hold.Footnote ¹³ Although individuals may choose to imbue certain traits they possess with deep meanings to their personality, they must not be forced to factor them in as social liabilities when they consider their choices and plans. Even when individuals wish to factor in traits that are important to them, such as their religion or their race, when they consider questions about where to work, study, or live, they should not have to bear the price of externally imposed social costs, that is, as traits that would make them less valuable or less attractive in the eyes of their employers, teachers, neighbors, or negatively affect their rights and obligations as property owners.Footnote ¹⁴

Extraneous traits constitute prohibited grounds of discrimination. Moreau neither provides a definite or a single criterion by which to identify extraneous traits, nor does she attempt to provide an exhaustive list of them. She suggests, rather, that we think about them in a normative sense. To her, the prohibited grounds of discrimination reflect a normative judgment according to which individuals have a right to make decisions in the social contexts protected by anti-discrimination law, “in a manner that is insulated from the burdens imposed by these traits, or by other people’s assumptions about them.”Footnote ¹⁵ The normative reasons for including traits under the extraneous traits list are diverse and may vary from one society to another.Footnote ¹⁶

A common thread running through these three accounts is that a person or a group cannot be considered a victim of discrimination if they do not bear at least one of the traits that are listed a priori by the legislator as grounds potentially constituting prohibited discrimination. In the rest of this paper, I will explain how these accounts overlook specific kinds of wrongful distinctions and argue that such elaborations merit recognition as forms of discrimination. I will focus, primarily, on distinctions that are based on irrational or arbitrary reasons which are not included in the existing law. As Khaitan rightly observes, “almost no one who takes both law and philosophy seriously stands up for ‘rationality’ as a foundational value anymore.”Footnote ¹⁷ One of the aims of this paper, then, is to provide such an account for the value of rationality to anti-discrimination law and to explain why, and in which circumstances, irrational or arbitrary distinctions constitute discrimination.

The next step is to clarify what I mean by the terms ‘irrational’ or ‘arbitrary’ distinctions. In the next sub-section, I will present three illustrations of irrational or arbitrary distinctions, which, to my mind, constitute discrimination. These illustrations will later help me propose a principled definition of irrationality and arbitrariness that I believe should apply to anti-discrimination law.

II. Towards a Definition of Arbitrariness and Irrationality: Three Illustrations of Irrational or Arbitrary Distinctions

1. (1) Steve is a young, promising, and well-published scholar who applies for a tenure-track position at a law faculty. Though they readily acknowledge his academic achievements and his potential to contribute to the faculty, some of the young faculty members attempt to convince the dean against hiring Steve. They do so because they are afraid that hiring him will pose a high threshold of publications and other academic achievements that all faculty members will now need to meet in order to get tenure. The dean, wary of starting a conflict with his colleagues, decides not to offer Steve a position in the faculty. Steve claims he has been discriminated against by the faculty. Steve does not belong to any salient minority group.

2. (2) A government decides to support citizens who reside in peripheral areas by granting them certain income tax exemptions. The government decides to implement this decision in a special provision to be included in the main statute that governs tax issues. For this matter, the parliament assembles a committee that decides what areas are to be considered peripheral and therefore eligible for the new exemptions. Four peripheral areas send powerful lobbyists to the committee to ensure that their areas are included in the tax exemption program. As the committee drafts its conclusions, these four are listed as eligible for tax exemptions. One peripheral area, ‘the Tamar region,’ resembles all the other four areas in all aspects except one: it did not engage lobbyists to represent it in the parliament committee. It does not make it to the list of regions to benefit from the program. The Tamar area argues that it has been discriminated against.Footnote ¹⁸

3. (3) Dan is a senior lawyer at a big law firm. Dan is responsible for the training of all incoming interns engaged by the firm. The interns’ main job is to perform research on legal questions and summarize their findings. The firm expects Dan to assess the interns’ quality of work and recommend hiring the ones who display the best performance as junior lawyers at the firm. Dan tends to favor and promote interns who stay long hours in the office over interns who do not, not because he was most impressed by the findings they handed to him, but rather because his parents always taught him that employees who stay long hours in their workplace are the best employees.Footnote ¹⁹ Dan therefore recommends the firm to hire the interns who have stayed the longest hours at the office once the internship period is completed. The interns whom Dan did not recommend claim the firm discriminated against them. Although the practice of deciding in this way might have a disparate impact on those with childcare obligations, for instance, let us assume that the interns whom Dan did not recommend do not belong to any socially salient group.

In all of these examples, distinctions are made between individuals or groups in distributing goods. The parties disfavored by these distinctions do not necessarily belong to vulnerable or salient minority groups. Under most anti-discrimination laws, they will not be acknowledged as victims of discrimination. In the rest of this paper, I will argue that we should acknowledge them as victims of discrimination since they were all subjected to distinctions based on arbitrary or irrational criteria. Before explaining why arbitrary or irrational distinctions amount to discrimination, I will first proffer my own definitions of the terms arbitrariness and irrationality, and argue how we should understand them in the contexts of discrimination.

III. Arbitrariness and Irrationality in the Context of Discrimination

Arbitrariness is a tricky concept, especially in the context of discrimination. Any given distinction is prone to be perceived as arbitrary from the point of view of the party disfavored by it.Footnote ²⁰ For politically aware fifteen-year-old high-school students well-informed about current political affairs and hoping to affect the political climate of their country, it seems arbitrary that the state restricts voting rights to citizens eighteen years old and older. The same feeling of unfavorable arbitrary treatment would also hypothetically hold similar allure to fifteen-year-olds who drive well and consider themselves mature enough to understand the dangers of careless driving but are still barred from acquiring a driving license until their eighteenth birthday. This is not the kind of arbitrariness or irrationality that I argue is wrong.

Many distinctions with regards to the distribution of goods and services are based on variables such as age, which serve as proxies for assumed capacities, such as political maturity or having sufficient responsibility and ability to drive safely. A proxy is a specific kind of generalization designed to predict the abilities and capacities of individuals by using a particular variable. There are neither perfect generalizations nor perfect proxies. Any chosen proxy will always fall short of infallibly predicting the capacities and abilities of certain individuals or groups. Yet the bar always needs to be set somewhere, qualifying some while disqualifying others.Footnote ²¹ We may think that the ones disqualified by such proxies are arbitrarily treated because they are denied the autonomy to present their own individual characters and capacities.Footnote ²² However, the fact is that we live with, and continue to tolerate many such imperfect proxies because they are indispensable to regulating many contexts of our lives. Most importantly, proxies are reliable when they establish a strong empirical correlation between the proxy and the specific relevant characters we are looking for. The effort to find reliable proxies is actually an effort to respect our autonomy as much as possible. This effort compensates, to some degree, for the violation of our autonomy because we know our autonomy was not violated in vain. The ones who rely on reliable proxies have tried to minimize the violation of our autonomy by establishing a strong empirical correlation between the proxy and the relevant characteristic they are looking for. In that sense, beyond the practical necessity to use reliable proxies in some cases, reliable proxies are not arbitrary. Therefore, curtailing some individual’s autonomy through the imposition of reliable proxies that do not reflect one’s capacities most accurately cannot be the kind of wrongful distinctions that anti-discrimination law purports to remedy.Footnote ²³

I therefore suggest concentrating on different kinds of irrational or arbitrary distinctions. I would like to suggest a general working definition of arbitrary or irrational distinctions, and later on break it to two sub-categories. A distinction is arbitrary or irrational if it satisfies two (combining) conditions. First, it is neither instrumental to nor constitutive of the purpose of distributing the goods at handFootnote ²⁴ —that is, it cannot be accounted for as a tool pertinent to furthering the distribution of the good or goods. Second, there is no normatively acceptable justification for the one who draws the distinction to rely on that ground.Footnote ²⁵ In other words, a distinction that is neither instrumental to nor constitutive of the purpose of distributing the goods at hand, can still be saved as a non-arbitrary distinction if it satisfies the second condition of a normatively acceptable justification.

Allow me to explain the definition I provided of irrational distinctions in further detail. The first condition concerns grounds neither instrumental to, nor constitutive of, the purpose of distributing the goods at hand. Grounds are instrumental to the purpose of distributing goods, for instance, when they are strong proxy for a quality, “the real basis for distributing the benefit in issue.”Footnote ²⁶ Denise Réaume uses the Canadian Supreme Court case of Andrews v Law Society of British Columbia Footnote ²⁷ as an example of grounds not instrumental to the distributing purpose at hand. Andrews, a British subject and a permanent resident in Canada, met all the admission requirements to make the provincial bar, with the exception he was not a Canadian citizen. Andrews brought a motion to strike down the requirement for citizenship, arguing it violated the equality provisions stipulated in section 15 of the Canadian Charter of Rights and Freedoms.Footnote ²⁸ The question was whether the possession of citizenship is an appropriate proxy by which to better realize the objective of producing good lawyers, that is, legal practitioners both knowledgeable about Canadian political and social affairs and committed to contributing to the country. The generalized assumption that citizens could normally be expected to possess better knowledge of current Canadian political and social affairs and would be more likely to exhibit a strong commitment to the country than non-citizens is inaccurate. Still, as Réaume observes, “we all know citizens who know little about Canadian affairs and feel even less commitment to the country, and we can certainly imagine a highly committed and knowledgeable non-citizen.”Footnote ²⁹ Citizenship is therefore not a reliable proxy for these qualities. By contrast to proxies such as age in the cases of a driving license and the right to vote, the proxy of citizenship does not rest on a strong empirical correlation. Unlike permanent residency, for instance, which establishes a strong connection between the length of residency in Canada and the knowledge of current Canadian political and social affairs, as well as for exhibiting a strong commitment to the country, citizenship is not necessarily a strong indication for those qualities. It is not instrumental in producing them nor does it guarantee their retention. Using the terminology of the definition I provided for irrational or arbitrary distinctions, the fact that citizenship is not instrumental to the distinction’s purpose marks the distinction as irrational or arbitrary. The court in Andrews did not use the terms ‘irrational’ or ‘arbitrary,’ but the imprecision of citizenship as a dependable proxy for good, committed lawyers led the court to conclude that the distinction between citizens and non-citizens constitutes discrimination.

Grounds are constitutive to the purpose of distributing goods when they are “part of the relevant objective—it wouldn’t be this objective if we didn’t use this distinction.”Footnote ³⁰ Réaume provides an example for a constitutive ground by recounting a debate between Justice Wilson and Justice McIntyre in Andrews. McIntyre J. argued that lawyers are required to be citizens because practicing law is exercising a governmental function, an activity which must be guarded as the exclusive privilege of citizens. Wilson J. contested the argument by noting that not all lawyers exercise governmental functions. However, she rightly observed that if the object of the distinction was electing lawyers for the positions of professional judges, the grounds of citizenship would be relevant. This is because judges are indispensable government functionaries. That is, citizenship is constitutive for the purpose of electing judges. In using the terminology I expounded on above, if citizenship was not a constitutive ground for the purpose of distinguishing between candidates for judicial positions, the distinction would have been irrational or arbitrary.

Let us now proceed to the second condition that concerns the nature of the justification one has for relying on grounds neither instrumental to nor constitutive of the distinction’s object. If one relies on such grounds for a normatively acceptable reason, then the distinction one made using these grounds is not irrational or arbitrary. A common normatively acceptable reason for relying on grounds neither instrumental to nor constitutive of the distinction’s object is that the one who draws the distinction does not have any other available grounds to rely on. Suppose that I have no available information on candidates for a baby-sitting job, except the fact that the potential candidates have or do not have children of their own. Suppose also that having or not having children of their own is not a reliable proxy for being a good babysitter. If I rely on this proxy only because I do not have any other available proxies that are more reliable, then I have a normatively acceptable reason to rely on it. Therefore, in this instance, the second condition of the definition provided above has not been met. The distinction is still arbitrary, but because it has a normatively acceptable reason, it is permissibly arbitrary, and therefore cannot be deemed irrational or arbitrary according to the definition I provided. If I have other available proxies for assessing the candidates’ quality that are more reliable, such as reference letters, then I do not have a normatively acceptable reason to rely on the unreliable proxy, such as the fact that the candidates have or do not have children of their own. In this case, when I fail to rely on an available reliable proxy, the distinction I made is irrational or arbitrary.

At this point, one may ask to what extent a person has to put in time and effort to find better or more reliable proxies. In which circumstances is the second condition of having a normative justifiable reason to rely on an unreliable proxy satisfied? How much time do I need to dedicate in order to find a reliable proxy for choosing a baby-sitter? Answering this question exceeds the scope of this paper. The second condition, according to which one needs to provide a normatively acceptable reason for relying on unreliable grounds, is a minimal condition which aims to prevent only bluntly arbitrary or irrational distinctions. Therefore, the answer is that non-arbitrary or rational distinctions require only minimal efforts for seeking better or more reliable proxies. That is, arbitrary or irrational distinctions are made when there are reliable proxies easily available that are overlooked or ignored. For instance, if candidates for a baby-sitter position provide me with letters of references and rather than read them I prefer concentrating on the fact that they have or do not have children of their own, I am acting irrationally.

One may also inquire into what unreliability means. In other words, can any unreliable proxy be saved by the second condition of a normatively acceptable reason for relying on it? The answer is no. There are proxies that are unreliable (or not necessarily reliable) and proxies that are totally flawed because they do not track the relevant facts at all. Examples for unreliable proxies are candidates having children of their own as a proxy for good baby-sitters, or interns that stay long hours in the office as a proxy for good lawyers. The factors of having or not having children might be relevant to a baby-sitting job because parents generally have more experience with children. The factor of spending long hours at the office might be relevant for lawyers because it at least indicates a strong commitment to one’s job as a lawyer. These proxies are unreliable or not necessarily reliable because they leave too many outliers (that is, too many candidates who are wonderful babysitters despite the fact that they do not have children of their own, or lawyers that are talented and efficient despite the fact that they did not spend long hours at the office as interns). Examples for totally flawed proxies include, for instance, choosing only red haired candidates as baby-sitters or lawyers. The second condition applies only to unreliable proxies. Not to totally flawed proxies. The second condition entails that it is sometimes permissible to rely on unreliable proxies when there is a normatively acceptable reason to do so. However, it is never normatively permissible to rely on totally flawed proxies since they do not track the relevant facts at all.

The following two sub-categories are representative examples of this compounded definition:

1. (1) A distinction based on a factor or a variable deemed rational and relevant to the distributed goods only from the perspective of promoting the self-interest of the one who draws the distinction. That is, a distinction promoted which, from a neutral point of view,Footnote ³¹ would not appear to be normatively justifiable as relevant to the goods being distributed. The first example from section II, that of Steve not being offered the position due to the fears his high qualifications arouse amongst the department’s junior faculty, is a good illustration of this category. Another good illustration of this category is the second example provided, which concerns a government that wishes to support all peripheral areas but perhaps unwittingly awards income tax exemptions only to those peripheral areas who invested in lobbying the relevant committee.

   The idea is that we observe the relevance of the ground not through the eyes of the one who draws the distinction. Rather, the relevance of the ground is examined in a neutral moral manner. In this sense, a ground is relevant only if, morally speaking, it can be justified by reasons that exclude the promotion of the interests of the one who draws it. This, of course, opens a debate about which grounds are morally permissible.Footnote ³² I cannot do justice to this debate here. My point is that the question whether a ground is morally permissible and therefore potentially relevant, or morally objectionable and therefore irrelevant, cannot be determined only by examining the subjective self-interests of the one who draws the distinction.

2. (2) A distinction based on an unreliable proxy. That is, a proxy that has no empirical foundation: a proxy imperfect to such an extent that it is irrevocably flawed.Footnote ³³ A distinction reliant on such a proxy for the distribution of goods is epistemically blameworthy when it is motivated by reasons that cannot be normatively justified. For instance, when it jumps to conclusions, carelessly overlooks counter evidence, indulges in wishful thinking, dogmatism, or is reliant on sloppy arithmetic.Footnote ³⁴ The third example from section II might be appropriate here. Dan, the senior lawyer who recommends that the company hire those interns who stay the longest at the office, promotes a distinction not based on any relevant empirical data, but rather based on his parents’ assertion that working longer hours without being required to do so are necessarily good employees. Assuming that Dan does not provide a more convincing justification for his distinction, arguing, perhaps, that he does not have any other commensurate way to assess the quality of the interns’ work, it is fair to say that Dan has treated them arbitrarily or irrationally.

IV. The Argument: Why Arbitrary or Irrational Distinctions Should Be Understood as Discrimination

Irrational or arbitrary distinctions are discriminatory because they disadvantage individuals without fair cause, which can be objectively understood as relevant to the distributed goods at hand. The idea of relevancy is not new to the doctrine and philosophy of discrimination law. Many accounts define discrimination as the accounting for considerations that are “morally irrelevant”Footnote ³⁵ or “normatively extraneous.”Footnote ³⁶ Such accounts, it seems, assume a moral and legal obligation to act rationally towards human beings. Although irrational treatment may be rooted in distinctions based on seemingly endless differences, such as height, intelligence, marital status, and so on, these accounts prohibit only irrational treatment based on differences stemming directly or indirectly from an individual’s affiliation to a salient social group.

Seeing as a moral and a legal obligation to provide rational excuses for giving worse treatment to certain groups of human beings is widely recognized, why ought such an obligation apply only to those whom we disadvantage due to their affiliation to socially salient groups? Deborah Hellman tackles this issue by arguing that traits such as race and gender, which ostensibly allow us to identify members of socially salient groups, are different from other traits, such as eye color, because they are the defining attributes of groups historically disadvantaged or presently occupying the lower rungs of society. Irrational, and even rational, classifications on the basis of these traits are more likely to be derogatory to individuals who bear them because they coincide with and replicate familiar forms of commonly practiced discrimination.Footnote ³⁷

One of the scenarios that Hellman offers to demonstrate the nature of demeaning discriminatory distinctions concerns employers who require only female employees to wear makeup. This distinction demeans women because, as Hellman rightly points out, in our culture, the wearing of makeup requirement is associated with a certain understanding of women’s bodies as ornamentations, objects for pleasure and enjoyment. By requiring only female employees to wear makeup, employers make a distinction which is discriminatory in a way different from similar but non-demeaning physical stipulations such as hair-length requirements.Footnote ³⁸

Let us trail Hellman’s line of thought and imagine another example. A new law school is established. During the first year of its establishment, the law school requires only eight faculty members. All eight faculty members initially hired are placed in tenure-track positions. The senior ones are hired as associate professors while the juniors are hired as assistant professors. All senior professors sign a contract that assures them an annual salary of $300,000. All junior professors sign a contract that assures them an annual salary of $200,000. The chief accountant of the law school, who negotiates the contracts on behalf of the law school, insists against any deviation from this rigid hiring policy. Two years later, the accountant leaves the law school. Responsibility over employment contracts of incoming employees is placed with the dean. The following year, the law school hires two more junior faculty members as assistant professors. Unlike the chief accountant, the dean thinks that the assistant professors’ salaries should be determined with regards to their personal academic competence (that is, publications and teaching evaluations). Because the new assistant professors are very well published relative to the early stage of their careers, he decides to offer each of them an annual salary of $250,000. For the sake of the example, let us assume that the salary of the assistant professors who were hired when the law school had been established was not raised since.

Now, let us complicate the scenario further still by assuming that all the assistant professors who were hired when the law school was established were women, and that the two new assistant professors are men. In keeping with Hellman’s argument, it would follow that, in light of this complication, the women were discriminated against by the dean. This is because the practice of providing women with a lower salary than the one that is given to men for the same job is demeaning. It replicates the prevalent cultural understanding that women’s work is worth less than men’s work.

Let us now consider a different variation of this scenario. In this reversed version, all the assistant professors who were hired when the law school was established are white men who do not belong to any socially salient group while the two new assistant professors are women. According to the definition of arbitrariness that I have outlined in section III, the male assistant professors who were hired when the law school was established were subjected to an arbitrary distinction. The law school made a distinction between the two groups of assistant professors based only on brute luck,Footnote ³⁹ that is, the fact that the accountant charged with negotiating the employment contracts when the first group of assistant professors was hired, left the school, and the dean, who was responsible for negotiating the later employment contracts, is more inclined towards generosity when negotiating the new professors’ salaries. The question is whether the assistant professors who were hired when the law school was established may be considered to be discriminated against in light of subsequent hiring policy changes which are immediately irrelevant to the agreed upon terms of their own employment.

To my mind, the answer is affirmative. In all three scenarios, individuals are disadvantaged because of irrelevant or arbitrary reasons. The only difference between the three is that both in Hellman’s example and in the first example I gave, the disadvantaged individuals have been not only disadvantaged, but have also been demeaned. A demeaning message may be considered as such only when distinctions are based on classic or suspicious grounds. But this fact alone does not mean that the disadvantaged individuals in the third example have not been wronged in any legally meaningful way, nor that they were not discriminated against.

Hellman argues that irrational or arbitrary distinctions not based on classic grounds are wrong, but not the kind of wrongs that anti-discrimination law addresses or should aim to address.Footnote ⁴⁰ For example, if a university adopts a policy that requires only math and science majors to complete a swimming class prior to graduation, it makes an irrational distinction by relying on an unreliable proxy, which assumes that math and science majors are likely to be poor swimmers. However, this irrational distinction is not based on classic or suspicious grounds. It therefore does not send a demeaning message towards these students, and does not qualify as discrimination. In Hellman’s own words, “treating people as equals does not require that distinctions be rational.”Footnote ⁴¹ Still, Hellman only explains why the students are not demeaned by this extra requirement. That is, she only explains why they are not subjected to a discrimination that is overtly wrong. She does not explain why the wrong the students suffer does not amount to discrimination.Footnote ⁴²

In order to understand the kind of wrong the disadvantaged male professors suffered in the third example, and why it amounts to discrimination, we should think deeper about what it means to be treated irrationally or arbitrarily. The dean, in the third example I have provided above, fails to take into consideration a crucial fact: the lower salary the law school provides for professors who were hired when it was established. By providing a higher income to the new professors, the dean disadvantaged existing junior faculty members. He also did nothing to correct or otherwise address this wrong. He did not offer to give a raise to the original junior faculty members, for instance, in order to assure that their income would reflect the management’s appreciation of their capacities and contribution to the law school’s core mission.

Completely ignoring relevant information or characteristics of the disadvantaged individuals is also a common denominator between the two categories of irrational or arbitrary distinctions presented in section III. Category (2) refers to irrational distinctions based on unreliable proxies. Take Dan, the senior lawyer who systematically hires interns able to stay longer hours at the office and passes over interns who do not, regardless of their other proclivities or competencies. Assuming that Dan does not have any acceptable normative justification to base his distinction on the length of hours the interns stay in the office, such as that he does not have any other feasible way to assess their quality of work, it is fair to say that Dan ignored relevant information about the actual contribution of each intern to the law firm.

The same failure to take into consideration relevant facts about the disadvantaged individual is clearly visible in category (1), concerned with distinctions based on factors or characters that are rational and relevant to the goods being distributed only from the perspective of promoting the self-interest of those who draw the distinctions. That is, they cannot normatively be justified as relevant to the establishment of how the goods at hand should be distributed.

The dean who chooses not to offer a tenure-track position to Steve, a young well-published scholar, only because young scholars in the faculty are threatened by his accomplishments, blatantly ignores Steve’s professional capacities and therefore his potential to contribute to the faculty’s core missions. The dean fails to take into consideration the determinant facts in relation to the goods being distributed—a position at an institution supposed to be committed to promoting research and knowledge. The same is true for the government that overlooks the demonstrated economic and social needs of a peripheral region, and fails to include it in a program of income tax exemptions, only because it does not effectively lobby the committee in charge of implementing the program.

What is wrong, we might ask, with not taking into account information or facts relevant to the distributed goods? Knowledge of all relevant facts is unquestionably important to insuring efficient and equitable distribution. Were relevant facts overlooked, in other words, goods would be distributed inefficiently, harming society as a whole. But efficiency is not my primary motivation in arguing that arbitrary distinctions constitute discrimination. Anti-discrimination litigation is not the correct instrument to promote efficiency in the job market, or elsewhere. Judges might consider themselves fit to make principled determinations regarding which distinctions promote the task at hand and which do not, however, they certainly lack the means to decide which kind of distinctions are most appropriately used in the interest of promoting efficiency. Bearing in mind that employers, for instance, are strongly incentivized to use efficient employment tests essential to the accruement of their profits, I am inclined to agree that strategies dictated by free market rationales are better suited to achieve and ensure efficiency, not litigation.Footnote ⁴³ Failing to consider relevant information is a wrong that constitutes discrimination for a different reason.

The relevant information in all the examples I presented concerns individuals who have different abilities, such as successfully performing a job as a professor in a university or as an intern in a law firm, and individuals with different needs, such as residents of peripheral areas. In other words, the information that is not taken into account relates to the characteristics and traits that make individuals different from each other in a manner that is relevant to the good or goods that are being distributed. For instance, legal interns have different abilities to perform their jobs. Some interns have the ability to produce more work per hour than others do. Some spend more hours at the office than others do. The difference between their ability to perform their job is relevant for the assessment of their productivity, whereas the mere difference between the numbers of hours they spend at the office is not.

When we form judgments about an individual without paying attention to their unique distinguishing features that are relevant to the distributed goods that they seek, we fail to fully recognize them as autonomous human beings. Full respect to one’s autonomy, as one scholar put it, means paying attention to the specific life one has chosen for themselves, or to the specific evidence of the ways in which they have constructed their life.Footnote ⁴⁴

Take unreliable proxies for instance: interns surely exercise their autonomy in working hard by staying long hours in the office. However, the idea is not to respect autonomy per se. Rather, we should respect autonomy when its exercise relates to the distributed good at hand. Reliance on the number of hours interns spend in the office as a crude gauge by which to determine an intern’s productivity or quality, fails to take into consideration other available and more relevant facts, which may be more indicative of interns’ work. It is possible, of course, that some interns read and write very fast and otherwise preform their job well without needing to spend long hours in the office to complete their tasks, and vice versa. The unreliable proxy of the number of hours they spend in the office does not account for the relevant individual capacities of each intern, and therefore does not show respect for the interns’ autonomy.

However disruptive, failing to fully respect one’s autonomy by not paying sufficient attention to any character traits that are relevant to the distinction at hand cannot be the wrong that discrimination aims to address. This is because every time we draw on reliable proxies in order to form judgment about individuals, we also accept the principle that such generalizations, despite being unable to reflect the specific contexts and unique character of each individual, are adequate. When the law permits only individuals older than eighteen to acquire a driving license, it creates an assumption that any individual younger than eighteen would not be a safe driver. This is despite the widely accepted fact that some individuals who are seventeen or even sixteen years old might be as safe a driver as any of their compatriots eighteen or older. And still, the law does not entertain the possibility that such a crude distinction would constitute discrimination because, though inattentive to the myriad specific and distinguishing capacities of every individual, the distinction is attentive to a capacity that is relevant to the goods being distributed. As I mentioned earlier, proxies are reliable when they establish a strong empirical correlation that can predict the specific relevant characters we are looking for. The effort to find reliable proxies is actually an effort to respect our autonomy as much as possible.

What, then, is the essential difference between relying on reliable proxies and relying on unreliable proxies? The difference is that when a decision is made on the basis of what seems to be a reliable proxy, it strives to minimize disrespecting one’s autonomy. This is because we form our judgment by relying on relevant factors to the distributed goods. When we rely on relevant factors, such as reliable proxies, we meet minimal standards of respect towards individuals as autonomous human beings because we strive to take into account the relevant differences between them. It is not only “useful individually and socially for people to act based upon accurate representations of the world,”Footnote ⁴⁵ as Larry Alexander puts it, but also when we rely on unreliable proxies, we fail to show individuals the minimum respect they deserve as human beings. We therefore do not respect a person’s autonomy at all because we invest our attention in deriding the irrelevancy of facts which prevent a discussion of the relevant facts we should consider. In other words, it is not the mere failure to fully respect a specific person’s autonomy that is wrong in taking irrelevant facts into consideration. We may try to take into account relevant factors, such as reliable proxies, and still fail to fully respect a person’s autonomy when the reliable proxies do not reflect their specific characteristics. It is the total disrespect of an individual’s autonomy that is wrong with considering irrelevant facts. When we do not minimally respect autonomy, we completely ignore its value to every human being. The threshold of minimal respect for autonomy is considering relevant considerations to the distributed good at hand. When we draw distinctions that rely on irrelevant considerations, such as unreliable proxies, we fail to meet the threshold of minimal respect.

When one forms judgments about individuals based on arbitrary or irrational reasons, one denies them equal treatment because one fails to produce plausible justification for giving them different treatment. The failure to produce ‘plausible justifications,’ what I refer to in section III as ‘normatively acceptable justification,’ stands at the very core of the equation of arbitrary or irrational treatment with discrimination. Without insistence on explanations that seriously and objectively consider relevant individuals’ needs or capacities relative to the distributed goods at hand, one soon devolves into a society in which a host of subjective interests or whims of the person or the body with the power to draw distinctions demolish individual autonomy.

At this point one may ask why the total failure to respect one’s autonomy ought to be conceptualized it terms of discrimination. What is wrong, for instance, with conceptualizing it in terms such as unfairness? The answer is that most legal systems do not acknowledge legal claims concerning unfair treatment. The wrong of being treated unfairly is very vague and it is not clear who is responsible for correcting it. Discrimination is a common legal claim in every legal system. It is perceived as the opposite of equality or equal treatment, which is enshrined in constitutions around the world, as well as in statutes that are specifically designated to address discrimination. More importantly, the common legal doctrine defines discrimination in terms of distinctions that rely on irrelevant grounds. Because irrelevancy and arbitrariness are strongly connected, it makes sense to conceptualize arbitrary distinctions in terms of discrimination, even when they are not related to the grounds that are usually listed in anti-discrimination statutes such as race, sex, and the like.

Up until now, I have answered two questions: why arbitrary distinctions amount to discrimination and when they are morally wrong. I have asserted that arbitrary distinctions amount to discrimination because they fail to minimally respect a person’s autonomy. Such discriminations are morally wrong when they do not provide a normatively acceptable reason for disrespecting a person’s autonomy. However, a third question should be answered at this stage: why and when should arbitrary distinctions that are morally wrong be also legally prohibited? The fact that an arbitrary distinction is morally wrong may be relevant to the question of whether it should be legally prohibited, but it is not the only relevant consideration with respect to the issue of prohibiting it by law.

In order to answer this question, we should first note that the account I have just presented, which focuses on distinctions made on arbitrary or irrelevant grounds, is consistent with the practice of anti-discrimination law. As Re’em Segev rightly indicates, although anti-discrimination law does not prohibit distinctions made on every irrelevant ground and only prohibits distinctions based on a common index of legally-precedented and publicly-recognized prohibited grounds, such as race and sex, one of the reasons to single out these grounds is the fact that they are usually morally insignificant to the distinction at hand. Therefore, a common defense to an alleged discriminatory act is that it was based on morally significant facts.Footnote ⁴⁶

The account presented here is entwined with the actual practice of anti-discrimination law but it differs from it in an important manner. It reads arbitrariness or irrationality as a sufficient condition to consider an act discriminatory in itself. It treats the common prohibited grounds of discrimination, such as race and sex, as aggravating factors, but not defining features of discrimination.Footnote ⁴⁷

At this stage, one may argue that arbitrary distinctions, on their own, do not provide a sufficient reason for prohibition by law exactly because they lack the aggravating factors of the common prohibited grounds of discrimination. That is, one may argue that the law does not prohibit arbitrary discriminations as such because they do not meet the necessary threshold of wrongdoing as opposed to distinctions that are based on the common prohibited grounds. In order to delve into this claim we should first understand why the common list of grounds add aggravating factors and what their nature is.

In the next subsection, I will therefore delve into these aggravating factors and argue that while arbitrariness or irrationality should be a sufficient condition for establishing discrimination, they do not constitute, by themselves, the worst kind of discrimination. The common or classic list of prohibited grounds of discrimination adds important factors that make irrational and arbitrary distinctions based on them the worst kind of discrimination. These important factors are also the reason to prohibit distinctions that are made on these grounds even when they are relevant to the distributed goods at hand.

V. Why Arbitrary or Irrational Distinctions Are Not the Worst Kind of Discrimination

Arbitrary or irrational distinctions based on one of the common or classic list of prohibited grounds of discrimination are the worst kind of discrimination. Not only are they entirely oblivious of the need to respect a person’s equality and autonomy, but by their very nature they are also demeaning and offensive to him or her. In addition to inflicting individual harm, such discriminatory behavior also affects the way vulnerable groups are perceived in society as whole.

Let me begin by explaining the notion of being demeaning and its relation to irrational or arbitrary distinctions based on the common list of prohibited grounds, such as race and sex. Deborah Hellman defines a demeaning act as an act that “puts down, subordinates, diminishes, denigrates, a person, treats him or her as lesser, as not fully human or not of equal moral worth.”Footnote ⁴⁸ Classification based on traits such as race and gender are more likely to demean than classifications based on other traits because demeaning is socially and culturally contingent. History, conventions, culture, and social understanding all play a role in determining which actions carry a demeaning message.Footnote ⁴⁹ Whether the characteristic one uses as a basis for distinction between individuals could potentially be construed as demeaning is determined by how that characteristic has been used in the past to distinguish between individuals and the relative social status of the group defined by the characteristic today.Footnote ⁵⁰ Characteristics such as race and sex delineate collectives mistreated in the past and struggling against systematic inequality in the present. A distinction riding on the wavelengths of race and gender is more likely to be demeaning because it replicates widespread forms of discrimination towards members of these vulnerable groups that have been historically used to push them into disadvantageous positions in the social, economic, political, and cultural spheres of public life.Footnote ⁵¹

For instance, ordering people of color to sit in the back of the bus is demeaning because it perpetuates the history of racial injustice that included segregation in public transportation. It conveys the message that people of color are inferior to whites. ^{Footnote 52} Employers who require only female employees to wear makeup similarly demean women because it evokes misogynist notions of women’s bodies as objects for adornment and pleasure. ^{Footnote 53} It is not enough for a classification to be based on traits such as race and gender, which replicate past discrimination, in order to be demeaning however. In order to demean one must occupy a position of power or status relative to the afflicted party. That is, it is particularly in a potentially subordinate or domineering environment, such as within hierarchical dyadic relationships in one’s place of employment, that a distinction potentially becomes an illegal demeaning action.Footnote ⁵⁴

Hellman’s singling out the affiliation to vulnerable groups as a crucial factor for arbitrary and irrational distinctions to be perceived as discriminations normally co-resides with another argument that stands in favor of discrimination law oriented only towards vulnerable groups. As Kelman explains, unlike individuals belonging to vulnerable groups and subjected to irrational distinctions, those who are subjected to similar irrelevant distinctions but do not belong to subordinated groups, will suffer much less from these distinctions, partially because, in the rare occasions they may face disadvantages due to such distinctions, they are more likely to make a ‘market exit’ to another place that will not treat them irrationally.Footnote ⁵⁵

I agree with Hellman that the argument about demeaning distinctions should not restrict itself to irrelevant distinctions that disadvantage individuals who belong to vulnerable groups. Hellman’s argument also helps explain why relevant distinctions that disadvantage individuals who belong to vulnerable groups also constitute discrimination. Even when distinctions are based on reliable proxies that accurately connect the character of belonging to a vulnerable group with a certain behavior or capacity, they are demeaning because they perpetuate dominant biases against vulnerable groups. It is inescapable that the discussion over which factors or characters are relevant to qualifying a distinction at hand as illegal discrimination is inextricable from dominant perceptions and structural inequalities in society, which many times disadvantage individuals who belong to vulnerable groups.Footnote ⁵⁶ Recall, for instance, the issue of make-up. Many customers expect female employees to wear make-up because the dominant perception in society is that well-cultivated women wear make-up. Thus, the dominant perception that sees women’s bodies as objects for adornment and enjoyment by othersFootnote ⁵⁷ considers such a requirement on behalf of employers to their female employees as relevant to the distinction between male and female employees.

The same is true for reliable proxies. It is true, for instance, that women are generally physically weaker than men. We can therefore safely assume that sex is a reliable proxy for physical power, and that sex is an ostensibly relevant factor in making a distinction that distributes goods according to an index of physical power. We will never know, however, if females are indeed born this way or if society educates and constructs them to be receptive towards their physical inferiority vis-à-vis men and never challenge its veracity from the very first day of their existence.Footnote ⁵⁸ This exercise comes to show that if we continue to determine the relevance of factors or characters according to dominant perceptions in society, we are doomed to reproduce, and indeed extend, current hierarchies between strong and vulnerable groups, all of whom camouflage themselves in the semblance of normality or alternately masquerade as ‘natural.’ There will be no real chance for a change. That is where the common or classic grounds of discrimination come into play. When they place every distinction based on one of them, relevant or not, arbitrary or not, under suspicion of discrimination, the common or classic grounds of discrimination fuel social change.

However, for Hellman, the argument about what constitutes a demeaning distinction is all the more reason to prohibit discrimination only when it is based on the commonly stipulated grounds, because they are the only characteristics that flag one’s affiliation to vulnerable groups in society. In my eyes, the requirement that a case meet one of the eligibility criteria as detailed in the common list of prohibited grounds to establish it as discrimination does not necessary follow from the argument about being demeaning. Had we taken as a point of departure the assertion that treating any person in an arbitrary or irrational manner is tantamount to ignoring his or her equal value as an autonomous human being, we could have arrived at an entirely different conclusion. Our conclusion in such a hypothetical situation would be that distinctions based on the commonly established prohibited grounds constitute the worst kind of discrimination because apart from discriminating against them, they also are demeaning and deeply offensive to their victims. However, there is another, less severe form of discrimination caused when distinctions are drawn arbitrarily or irrationally in total disregard for individual autonomy. This stripping of autonomy of any meaningful value occurs even when the victims of these distinctions do not belong to any vulnerable groups in society.

For this reason, I think that in principle, the law should prohibit arbitrary distinctions on their own in cases of legal systems that justify prohibiting discrimination in terms of protecting individual autonomy. From a conceptual point of view, distinctions that are based on the common listed grounds are only one instance of irrelevant distinctions. In practice, there are maybe good reasons to focus only on irrelevant distinctions that are based on the common list of prohibited grounds because they are the worst kind of irrelevant distinctions. Whether every legal system should actually prohibit arbitrary distinctions on their own, and whether it should distinguish between private and public actors in this regard, are questions that relate to many issues that exceed the scope of this paper. Such issues may include, for instance, the structure of anti-discrimination laws (for instance, whether they include a general clause that prohibits discrimination as such or whether they include an exhaustive list of prohibited grounds), the general legal approach towards interference in the private sector, the actual possibilities of enforcing prohibition against arbitrary distinctions, and so on.

In the next section, I will address counter arguments that object to recognizing arbitrariness and irrationality as sufficient conditions for discrimination.

VI. Addressing Arguments Against Recognizing Mere Arbitrariness and Irrationality as Discrimination

Perhaps one of the central arguments levelled against recognizing mere arbitrariness and irrationality as the basis for discrimination litigation is the ‘liberty interest argument,’ according to which individuals have a strong interest in being free to associate with whomever they choose, even if it involves irrational or arbitrary distinctions which de facto exclude disadvantaged persons.Footnote ⁵⁹ In other words, the liberty interest argument highlights the principle of the liberty of individuals to act arbitrarily or irrationally. Their autonomy, so the argument goes, may only be curtailed in cases where the disadvantaged individual belongs to one of several pre-defined vulnerable groups and is being excluded based on that belonging. If the exclusion is not based on any recourse to the factors that make the victim vulnerable, the argument goes, the liberty of the ostensible victims of arbitrary and irrational distinctions should not be prioritized over that of the ones who draw the distinctions to go about their business. The desire to correct social ills, commendable as it is, does not warrant nor excuse such legal interference.Footnote ⁶⁰

Mark Kelman, who rightly argues that the liberty interest is rather weak in most, if not all, cases of irrelevant and arbitrary distinctions, has put forward a convincing answer to such a libertarian approach. According to Kelman, the liberty interest carries serious weight only in those cases where an intimate relationship or association exists between the one who draws the distinction and the one who is disadvantaged (or advantaged) by it. Thus, people are free to choose whomever they want as their lifetime partners, lovers, or prom dates, even when their choices distinguish between other people in an arbitrary or irrational manner.Footnote ⁶¹ However, intimate relationships of this sort are not the kind of irrelevant distinctions legal scholars are generally concerned with. The irrelevant distinctions we are talking about here are fundamental parts of ensuring equal access to employment, collective and private resources, and services.Footnote ⁶² In other words, people expect to be treated in a professional, respectful manner when there are impersonal relationships at stake, and demand that the law ensure their right to equal treatment and access to collective goods.

A second argument brought against acknowledging all irrelevant distinctions as discriminations, regardless of their connection to the commonly protected grounds, is that such an acknowledgment might prove too universal and therefore too obtuse in practice and application. ^{Footnote 63} Such trivializations of the far-reaching effects of historically rooted discrimination subordinated groups could be answered by a forthright acknowledgement, that discriminations on the basis of merely irrational grounds, are less severe than discrimination perpetuating the disenfranchisement of historically subordinated groups.Footnote ⁶⁴ In other words, although I argue that we should reject all irrational distinctions as discriminations, it does not follow that we cannot allocate more resources to allow the justice system to combat discriminations based on protected grounds. It is highly reasonable, and common practice throughout law-enforcement, to allocate resources to combat the most pervasive and egregious forms of discrimination. ^{Footnote 65}

It is also important to note two significant advantages of such an expansive account of discrimination. First, if adopted, it could put a stop to the ceaseless efforts to extend legal protections to accommodate victims of the ‘new’ forms of discrimination.Footnote ⁶⁶ Because the list of protected grounds of discrimination is limited in most jurisdictions, there are scholars and practitioners who try to graft additional characteristics, such as appearance and weight, which are the subject of irrational distinctions in many societies but have not yet been formally recognized as legally acknowledged grounds for illegal discrimination.Footnote ⁶⁷ Second, adopting the approach I suggest could put a stop to the ceaseless efforts to define who belongs to the protected minority group and who does not. Scholars such as Ford and Ferretti have pointed out the impossible mission of defining the exact characters of individuals who are supposed to fit into the category of minority members.Footnote ⁶⁸ In adopting the approach laid out here, any irrational distinction would be recognized as discrimination without having to campaign for its inclusion in the ‘prestigious’ list of recognized grounds of discrimination, and without having to adopt an essentialist definition of membership in a recognized protected group.