Printer Friendly

The antidiscrimination paradigm: irrational, unjust, and tyrannical.

Part I. Antidiscrimination Is Irrational

In this article, I examine the antidiscrimination paradigm and the phenomenon of discrimination on the grounds of demographic characteristics such as race, sex, and so on. For brevity and concreteness, I focus mainly on race discrimination, although my thesis applies equally to discrimination on other grounds, such as sex, age, disability, and sexual orientation.

Race and Racism

To understand this discussion, we need not delve into the complex world of anthropology to tease out biological definitions of race. (1) Although anthropological racial classification is an interesting subject in its own right, for the purposes of discussing race discrimination and the antidiscrimination paradigm, the prevailing scientific definition of race, the question of borderline cases, and the precision of different racial categorizations based on biological or other characteristics are of little relevance. (2)

In fact, what is usually called race discrimination has little to do with scientific racial classification. Rather, it is discrimination based on clear observable characteristics that are widely regarded as attributes of race--things such as skin color, face and body structure, hair texture, and other external physical characteristics. Thus, whether these things are essential to the anthropological definition of race and whether race has any precise anthropological definition are of no consequence. (3) What is relevant to cases of race discrimination is that different people have different-colored skin, different body structure, different kinds of hair, and so on, and that these characteristics are often used to infer a person's race and to discriminate on this basis.

Although there is much disagreement over the meaning of the term racism, I take racism to be a thesis that asserts that the contents of the mind are determined by genetic heredity--or, in simple terms, that "biology causes behavior." This view is the essence of the explanation advanced by Ayn Rand, who explains racism as follows:
 Racism is the lowest, most crudely primitive form of collectivism.
 It is the notion of ascribing moral, social or political
 significance to a man's genetic lineage--the notion that a man's
 intellectual and characterological traits are produced and
 transmitted by his internal body chemistry. Which means, in
 practice, that a man is to be judged, not by his own character and
 actions, but by the character and actions of a collective of

 Racism claims that the content of a man's mind (not his cognitive
 apparatus, but its content) is inherited; that a man's convictions,
 values and character are determined before he is born, by physical
 factors beyond his control. (1964, 147)

Racism as a Causal Thesis

As this explanation of the concept makes clear, racism is a causal thesis. This thesis does not merely assert that racial characteristics happen to be correlated with intellectual or moral characteristics (though such correlation is obviously a consequence of the causal thesis), but rather also holds that genetic factors cause these characteristics. (4)

Although correlation between racial characteristics and other characteristics may occur in the absence of a causal genetic influence on character, mere correlation between racial and other characteristics is occasionally adopted as the defining feature of racism, so that any assertion of an empirical relationship between beliefs or character traits and race is regarded as racist. Indeed, it is common to hear any empirical assertion about differences in racial groups decried as racist, without regard to whether it is a causal assertion or not. David Stove uses the term in this sense and therefore accepts the thesis of racism as true, saying: "'Racism' is the belief that some human races are inferior to others in certain respects, and that it is sometimes proper to make such differences the basis of our behaviour towards people. It is this proposition which is nowadays constantly declared to be false, though everyone knows it to be true; just as everyone knows it to be true that people differ by age, sex, health, etc., and that it is sometimes proper to make these differences the basis of our behaviour towards them" (1999, 147). Thus, to Stove, unlike to Rand, an assertion that some race of people is inferior to another in some respect because of a particular cultural practice rather than a genetic cause would be regarded as racist (of course, Stove would also accept that such an assertion might be true).

This latter definition of racism is highly problematic. First of all, it multiplies concepts beyond necessity because it essentially equates racism with race discrimination. Second, it does not sit well with the historical evolution of the concept of racism, which emerged during the Enlightenment as a result of attempts to explain sociological and cultural differences between racial groups in terms of emerging and influential genetic ideas of causation. (5) For these reasons, the conception of racism as a causal thesis is the correct approach, and the notion of racism as merely a thesis about the empirical relationship between characteristics should be rejected. (6)

Racism as the Belief that "Biology Causes Behavior"

According to Rand's definition, racism is a thesis about the contents of a person's mind, not a thesis about his physical characteristics. Thus, it is not racist to believe that the biological characteristics of white people cause them to have lighter skin than black people. Indeed, it is obvious that this difference exists and that the biological characteristics of race do cause different physical characteristics in people, including different skin, hair, and eye color, different face and body structure, and so on. In fact, precisely these physical differences are the primary basis on which we place people in different racial groups in the first place.

Moreover, Rand stresses that racism is not a thesis about the cognitive apparatus of different races of people--it is about the contents of their mind, including their convictions, values, and character. In this view, the wider scientific issue of cognitive development and its relationship to genetic factors relates to the theory of racism only insofar as it hypothesizes whether genetic factors cause people's convictions, values, and character. (7) A theory that asserts that genetic factors are a substantial cause of these things can properly be regarded as racist, whereas a theory that asserts only that genetic factors influence cognitive development in some other way cannot be so regarded.

Although it is useful to elucidate the meaning of race and racism in discussions of race discrimination, one of my main aims is to demonstrate that causal theories of genetic determination are irrelevant in determining the rationality or irrationality--and the morality or immorality--of race discrimination in cases of passive observation. We will see that statistical dependence between characteristics is what is important for inferences from passive observation. Causal theories are relevant only insofar as we wish to know people's actual underlying nature rather than merely to predict how they will behave in particular circumstances. Thus, whether race is causally related to a person's convictions, values, and character or not--that is, whether various racist theories are true or not--is beyond the scope of this article and will be shown to be irrelevant to the inferential problem of discrimination (for a refutation of the theory of racism, see Rand 1964, 147-57).

The Antidiscrimination Paradigm

Most people now reject the thesis of racism; in other words, they reject the view that a person's genetic lineage determines his beliefs and character. However, many people have concluded not merely that racism is false, but also that in the absence of irrational discrimination, no correlation should exist between racial characteristics and other characteristics, regardless of the cause. Indeed, antidiscrimination activists' dogma has long been that all race discrimination is necessarily arbitrary and irrational--a manifestation of ignorance, fear, and prejudice.

In fact, causal theories of racism have often been regarded as the sine qua non of race discrimination, so much so that many commentators define all race discrimination as racism. For example, Kevin Reilly, Stephen Kaufman, and Angela Bodino take this position when they state that "[r]acism is prejudice or discrimination against other people because of their 'race' or because of what is thought to be their race (their biology or ancestry or physical appearance)" (2003, 15, emphasis added). Even with this definition, they maintain the claim that "[r]acism involves the assumption that people's birth or biology determines who they are: that behaviour is based on biology" (15). In other words, they regard the causal thesis of racism as a necessary condition for any form of race discrimination.

According to many people who adopt this prevailing view, to engage in any form of race discrimination is to identify oneself instantly as an ignorant bigot, a primitive relic, not conversant with modern notions of reason and justice. And yet this view of discrimination can be entirely discredited by an examination of some of the most basic findings of the science of statistical inference and prediction.

Statistical Inference and Discrimination

The discipline of statistics is concerned with the logic of inductive reasoning--with determining what inferences and predictions about the unknown follow logically from observed data. Statistical inference applies the mathematics of probability theory to decision problems involving unknown but relevant information. It is grounded in quantitative coherence theories, (8) which demonstrate that inferences made in accordance with the rules of probability theory are the only ones capable of avoiding certain behavioral inconsistencies, such as Dutch book inconsistencies (9) or other irrational intransitivities in our preferences or beliefs. In short, probability theory and the discipline of statistics furnish us with inferences and decisions that avoid irrational behavioral inconsistencies.

Among the insights of the discipline of statistics is the demonstration that we may rationally infer unknown characteristics of interest to us from any observable characteristics that either tend to exist or tend not to exist with the characteristic of interest (as opposed to observable characteristics that are statistically independent of the characteristic of interest). In statistical parlance, we say that we may rationally infer unknown characteristics of interest from any observed characteristics that are statistically dependent--not to be confused with causally dependent--on the characteristic of interest, conditional, of course, on whatever other information is at hand.

It is well known among scientists, statisticians, and philosophers that statistical dependence is a necessary but not a sufficient requirement for a causal relationship-that is, that correlation does not imply causation. (10) The principle is widely recognized as the repudiation of the logical fallacy encapsulated in the Latin maxim cum hoc ergo propter hoc, meaning "with this, therefore because of this." However, although this principle is well known, it is rarely appreciated that for the purposes of inference and prediction about unknowns, causal relationships are relevant only where we wish to know the effect of an action that we might take. In the absence of an intervening action--that is, in the case of passive observation--statistical dependence rather than causal dependence is all that is relevant for rational prediction, and causal connections are of no concern (for a detailed treatment of the relationship between cause and probability, see Pearl 2000).

This proposition is a general principle of rational inference, admitting no special cases. In particular, the principle applies to race discrimination as much as to any other form of discrimination in decision making. Racial characteristics are characteristics like any others used in rational inference, and no special status should be accorded to them. Statistics is a value-flee science, concerned with rational inference and prediction. It knows no boundaries of political correctness and is not constrained by political or social expedience or popularity.

To illustrate this rather esoteric statistical argument by a concrete example, consider the case of taxi drivers in Washington, D.C.--themselves of various races--who often refuse to pick up young black men for fear of being assaulted. (11) This case involves a prediction problem like many others encountered in statistics. The unknown information of interest to the driver in such a situation is whether he will be assaulted by his prospective passenger, and the available factors for prediction are restricted to what can be determined by observing the potential passenger for the small amount of time available before making the decision whether to pick him up.

In this context, the driver can observe the prospective passenger's race, sex, approximate age, clothing, and appearance and perhaps also gain a quick glimpse of his current behavior and demeanor. On the basis of this information, the driver may infer a high or low risk of assault and use this inference to avoid passengers that he predicts would pose a high risk of assaulting him. This prediction is done on the basis of a procedure that is common to statistical inference: using statistical dependence between known and unknown characteristics to predict the latter.

In judging these actions, few deny that a correlation exists between race and crime, sex and crime, and age and crime, such that it is demonstrably true that a young black man chosen at random (from the population of Washington, D.C.) is far more likely to commit a crime during a taxi ride than, say, a randomly chosen elderly Asian woman. Crime statistics coupled with elementary arguments in statistical inference clearly reveal as much. Walter Williams observes that "[w]hether we like it or not, race and crime are highly correlated. And, more important for dorm guards and taxi drivers, violent criminal acts are highly correlated with race. Black people know this, as do white people" (1995, 18). And yet commentators such as Andrew Hacker regard such discrimination practiced by taxi drivers as "patently racist" (1992, 20). Presumably then, it is also patently sexist, patently ageist, and even, in cases where the driver takes account of the prospective passenger's clothing, patently "fashionist."

To the taxi driver, why a correlation exists between race and crime makes no difference at all. It makes no difference whether a young black man is more likely to assault him because of some inherent genetic predisposition to violence (as would be the case under a racist account) or because of some alternative cause, such as poverty, a lack of education, or a lack of positive male role models. The only thing relevant for the purpose of rational prediction is the fact that a correlation exists.

Some commentators have correctly recognized discrimination in taxi service as a manifestation of rational discrimination rather than of racism (D'Souza 1995, 25053, 1999; Williams 1995, 17-19). It is rational in the context of the limited information available to the taxi driver, who is not in a position to inquire into his prospective passengers' criminal history. He is instead confronted with an inferential problem involving very little available information. Of course, we cannot know the taxi driver's thought processes for certain (because of our limited information of having observed them only in this situation of highly limited information), and the driver may indeed be an insufferable racist. But what we do know is that in this situation his discrimination is not evidence of racism or irrational behavior.

The Effect of More Information

This lack of information is crucial to the phenomenon of rational discrimination on the basis of noncausal characteristics and, indeed, is the entire basis for the existence of the inferential problem in the first place. Our nonomniscience compels us to make inferences about the unknown on the basis of the known.

However, as we gain information that is more directly related to the characteristics of interest to us, the situation changes. In particular, if no causal relationship exists between the characteristic of interest to us and an observable characteristic used for prediction, then the relationship between the two must operate through some third, intermediary characteristic. Information on this intermediary characteristic may render the previously employed characteristic redundant for prediction.

This switch occurs in the case of race discrimination in situations where the decision maker acquires information that bears more directly on the characteristic of interest. Thus, in the case of employment, there may be less reason or even no reason to use race as a factor to predict criminality. Instead, a prospective employee may be assessed on the basis of a police background check, character references, educational references, previous employer references, demeanor at the interview, and so on. This kind of inquiry may be sufficient to ensure that there is little or no remaining statistical dependence between race and crime conditional on this additional information. (12) In this case, the decision maker would have no rational reason to use race as a factor to predict the likelihood of criminal conduct. Discrimination on the basis of a factor that does not have a causal connection to the characteristic of interest arises from a lack of information rather than from any more enduring judgment. Thus, the endpoint of rational race discrimination occurs when we gain sufficient information to see that race is conditionally independent of the characteristics of legitimate interest to us.

An Objection--That Discrimination Is Based on Collectivism

One common objection to race discrimination (even where it is done rationally, as described previously) is that it focuses on the characteristics of racial groups instead of on the characteristics of the actual individuals that are the focal point of a decision. Because such discrimination involves making a decision about an individual on the basis of the known characteristics of others who share the same race rather than on the basis of the unknown characteristics of the person himself, some argue that this action manifests collectivism. Thus, in cases of race discrimination by taxi drivers, we sometimes hear the objection that the driver should focus his attention on whether the prospective passenger is a criminal or not rather than on his race and that to do otherwise constitutes a failure to judge the individual on his own merits.

This kind of objection completely misses the point of the inferential problem in that it advocates the impossible--namely, concentrating on a characteristic that is unknown. The inferential problem arises because we cannot "focus our attention" on an unknown characteristic. Suffice to say, proposing that one take into account unknown variables in an inferential problem is not a solution to the problem, but merely an attempt to wish it away.

This objection also misconstrues the nature of collectivism, a pernicious doctrine that takes the group to be the relevant methodological unit for decision making. The doctrine of collectivism does not direct us to use group characteristics as a means of predicting an individual's characteristics. Rather, it holds that the individual's characteristics, known or not, inferred or not, are irrelevant. It is not concerned at all with any direct or inferential inquiry into an individual's characteristics, but holds that the group's characteristics are the relevant characteristics of interest.

It is not a manifestation of collectivism to use known information about groups in an attempt to predict relevant characteristics about an individual. Rather, the attempt to predict an individual's characteristics is a correct and proper manifestation of methodological individualism.

An Objection--That There Are Other Important Factors in Prediction

A similar objection is that race is not the only factor of predictive value in inference problems--that people have other characteristics of potential predictive value that we may observe with only a cursory inspection, such as their sex, apparent age, what they wear, how they walk, how they look at us, and so on (for an example of this objection, see the remarks in Corlett 1993, 164-65).

Although it is certainly true that these kinds of characteristics may also have substantial predictive power, this condition is no objection to rational race discrimination. The presence of other characteristics does not necessarily render race redundant for prediction. Only when these other factors are sufficient to ensure that no remaining statistical dependence exists between race and the characteristic of interest, conditional on this other information, does race become irrelevant to inference and prediction.

Moreover, as a critique of actual cases of rational race discrimination that occur in practice, this objection is a straw man. Practitioners of rational discrimination rarely use race as the only factor in their decisions. Indeed, in the case of discrimination by taxi drivers, factors such as the sex, age, clothing, and the apparent demeanor of the prospective passenger have been reported to be factors in the driver's assessment of whether the person is a safe passenger to pick up (D'Souza 1995, 250-53).

An Objection--That Rational Discrimination "Gives Comfort" to Racists, Sexists, and So Forth

Perhaps the perennial objection to any argument in favor of rational discrimination--or, indeed, to any argument challenging the dogma of the antidiscrimination paradigm in any way--is that such talk "gives comfort" to racists, sexists, bigots, misogynists, and so forth.

This objection has several problems. First of all, it is not factually correct. It is unlikely that racists gain any comfort or ammunition from arguments for rational discrimination based on noncausal factors, let alone agree with them. The racist's defining characteristic is his belief in the causal determination of a person's character and actions by that person's genetic lineage. This belief is neither affirmed nor contradicted by arguments for rational discrimination on the basis of statistical dependence. Indeed, as we have seen, such causal judgments are irrelevant to prediction and therefore irrelevant to the phenomenon of rational discrimination.

Nor should racists gain comfort from statistical arguments in favor of rational discrimination because it is incorrect to claim that the actions of racists are indistinguishable from those of nonracist practitioners of rational discrimination. Discrimination on the basis of a judgment of racism (that is, of a causal connection between race and crime, and so forth) can be clearly distinguished from discrimination on the basis of noncausal statistical dependence by looking at the effects of additional information.

The most fundamental flaw in this argument, however, is that regardless of the its correctness or incorrectness, it is nonetheless an irrational appeal to the fallacy of guilt by association. The merits of arguments for rational discrimination do not depend on whether such talk "gives comfort" to racists, sexists, the boorish, the bigoted, the stupid, the wicked, or anyone else.

Many people are uncomfortable with the notion of rational race discrimination. The idea of discrimination conjures up images of atrocities such as outright genocide. Indeed, Stove amusingly observes that "[i]t seems to be often believed that, if you admit truths of the kind which I have listed above [namely, that it is sometimes proper to adjust our behavior toward people on the basis of racial differences], consistency requires that you try to murder entire races of people. I do not know what one can say of a belief as ridiculous as this, except that it is extremely ridiculous" (1999, 149). As we can see from the principles of statistical inference, rational race discrimination requires no such thing. It is an inferential procedure that usually results in tentative conclusions based on the absence of more direct information.

Theories of rational discrimination obviously should not be used as the entering wedge for irrational discrimination or racism, but, more important, truth should not be feared and howled down because of the danger that some may try to misrepresent it for ulterior purposes. So long as the notion of statistical dependence based on conditioning information is well understood, and so long as we are sensible in our acquisition of additional information, theories of rational discrimination cannot be used convincingly to justify irrational discrimination.

How to Reduce Rational Discrimination

Rational race discrimination can be diminished in frequency by two means without the imposition of coercion or appeals to irrationality. The first is by gaining sufficient additional information so that race is rendered conditionally independent (or close to independent) of the characteristic of interest to us--that is, by finding ways to reduce the cost, difficulty, or inconvenience of acquiring additional information in decision problems. The second is by bringing about a long-term change in our data--that is, by encouraging long-term changes in racial groups' negative behavior so as to reduce the statistical dependence between membership of these groups and the negative behavioral characteristics that are of interest to us. As Dinesh D'Souza puts it, "[r]ational discrimination against young black men can be fully eradicated only by getting rid of destructive conduct by the group that forms that basis for statistically valid groups distinctions" (1995, 287).

Rather than proposing either of these two courses of action, the antidiscrimination paradigm instead directs us to use even less information--to ignore knowledge of race or racial characteristics. Moreover, the philosophies of multiculturalism and cultural relativism that often accompany the antidiscrimination paradigm reject any criticism of negative behavior among racial groups and essentially reject the existence of negative group traits: they hold that a cultural group may be judged only by the standards of the culture itself, and therefore they reject the contention that we can even judge negative behavior for a cultural group. This approach aggravates the problem rather than solving it (for discussion of problems in the philosophy of multiculturalism and its similarities with the theory of racism, see Welsh 2008).

The use of known proxy characteristics to predict unknown characteristics of interest is a well-known practice in statistical inference and prediction. Its usefulness lies in exploiting known statistical dependencies between characteristics to gain information on unknowns. The antidiscrimination paradigm, which holds all discrimination to be irrational, effectively denies basic principles of rational inference and prediction. Far from identifying irrational behavior based on ignorance and prejudice, the antidiscrimination paradigm is itself irrational and is itself based on ignorance and prejudice--ignorance of the principles of rational inference and baseless prejudice that empirical equality prevails among racial groups.

Refusal to accept the rationality of certain forms of race discrimination poses great danger. If racism is regarded as the sine qua non of race discrimination, then we will forever misdiagnose rational discrimination as evidence of racism, thereby creating or sustaining hostility and antagonism between racial groups.

Part II. Antidiscrimination Is Unjust

We have seen that contrary to the antidiscrimination paradigm, race discrimination may be rational even in the absence of a causal link between race and the characteristics that directly interest us when there is a statistical dependence between these characteristics, conditional on the limited information available to us. Notwithstanding that such discrimination may be rational in certain contexts, is it nonetheless unjust to discriminate against a person on the basis of his race? Or, to put it another way, is it morally legitimate to engage in rational discrimination, or is the antidiscrimination paradigm the quintessence of justice that its advocates would have us believe?

It is certainly clear that discrimination, whether rational or irrational, harms the interests of those who are discriminated against. Or rather, we speak of discrimination against a person only when the discrimination does harm their interests. This harm is often no fault of the persons subjected to the discrimination-even in cases of rational discrimination--because many are the victims of rational but nonetheless incorrect inferences about their individual characteristics (although some are indeed judged correctly). In some contexts, such as discrimination against young black men in the provision of taxi services, the majority of those discriminated against are innocent of any wrongdoing and would not be subjected to discrimination if not for the limited information available to the decision maker.

In considering the fate of African Americans who are the object of race discrimination, D'Souza contends:
 It cannot be denied that African Americans suffer slights in terms
 of taxi-drivers who pass them by, pedestrians who treat them as a
 security risk, banks that are reluctant to invest in black
 neighbourhoods, and other forms of continued discrimination. Some
 of this discrimination is irrational, motivated by bigotry or
 faulty generalization. Much of it, as we have seen, is behaviour
 that is rational from the point of view of the discriminator and at
 the same time harmful for black individuals who do not conform to
 the behavioural patterns of their peers. (1995, 525-26)

It is perfectly legitimate to sympathize with the victims of incorrect inferences that lead to their being feared as criminals and excluded from the provision of certain services. Indeed, this form of inference may permeate much deeper than the mere loss of amenities. To a young black man, his first encounter with any new person begins with an inescapable warning sign from the information-beating characteristics of his age, sex, and race. Before he can even open his mouth to speak, he has given a signal to those who meet him: "Look out, I am statistically more likely to assault and rob you!" Such a person is indeed a victim of his unchosen characteristics, and we may rightly sympathize with him as he endures this predicament.

For this reason, many who have recognized the rationality of certain instances of race discrimination have nonetheless baulked at the actual practice of rational discrimination, fearing that it may be logically valid but morally unacceptable. For example, despite his clear recognition that certain forms of race discrimination are rational, D'Souza nonetheless states: "Just because discrimination is rational, however, does not mean that it is moral.... [R]ational discrimination based on unalterable characteristics is problematic in a way that discrimination against high school dropouts and convicted felons, who are in a position to reform their circumstances, is not.... The public policy dilemma is that such discrimination forces a choice in which the claims of morality are on one side, and the claims of productivity are on the other" (1995, 286-87).

Of course, many go much further than D'Souza, regarding discrimination as a mortal sin, rational or not. So common is this attitude that the eradication of all forms of race and sex discrimination has become the core of modern programs of alleged "social justice," including the United Nations Declaration on the Elimination of All Forms of Racial Discrimination (United Nations 1963). But is the claim of morality really on one side (antidiscrimination) against the claim of rationality? Is rationality really opposed to morality and justice?

Justice as the Rational Objective Judgment of Other People

In order to understand whether instances of race discrimination are unjust, we must be clear about the nature of justice and, in particular, about what rights and legitimate moral expectations apply to those affected by our decisions. This statement may seem trite, but it is a logical step that is rarely taken in discussions of the alleged injustices of discrimination. In such discussions, it is often presumed that any harm to the interests of those discriminated against is sufficient to demonstrate injustice. In other words, it is presumed that others are morally entitled to the fulfillment of certain desires, and our failure to fulfill these desires on the grounds of race discrimination violates their alleged rights.

There is, however, no such thing as the right to a taxi cab ride, a bank loan, or a job. Indeed, there is no such thing as the right to have people provide us with any services or opportunities, even if they make these things readily available to others. We have a right to have services supplied to us or opportunities granted to us only when we can find a willing supplier. Similarly, no one has a right to have a woman refrain from avoiding him on the street or clutching anxiously at her handbag, even if she is willing to walk past others contentedly. We may expect her to pass gracefully only when she feels safe and pleased to do so.

If others' actions that benefit us cannot be regarded as rights, can they nonetheless be regarded as legitimate expectations, especially when similar actions are taken to benefit other people? To answer this question we must ask how others should act to fulfill their own desires and why they offer beneficial services and opportunities to some people but not to others--that is, why they discriminate. But we have already seen the answer to this question: we should expect a person to discriminate in decision making when it is rational to do so.

In fact, contrary to the whole morality versus rationality dichotomy adopted by D'Souza and others, justice is nothing more than the rational objective judgment and treatment of others. This view is implicit in the widely accepted notion that justice consists of giving every person his due, a view expressed in expositions of justice by philosophers such as Aristotle, Aquinas, Spinoza, Comte-Sponville, and Rand, among others. (13) According to Rand, justice requires that "you must judge all men as conscientiously as you judge inanimate objects, with the same respect for truth, with the same incorruptible vision, by as pure and as rational a process of identification" (1957, 933). In fact, there is no disparity between the rational and the just--the rational is the just. So it follows apodictically that rational discrimination is just, by virtue of its rationality.

This conception of justice as the rational judgment of people focuses attention on the process by which humans are evaluated; it regards justice as a property of human decision making. This view clashes with the notion that people are entitled by moral right to the equal fulfillment of certain desires and the attainment of certain outcomes from others' decisions. This latter view reflects the idea of justice as a property of the outcome of decisions rather than as a property of the decision making itself. It is a manifestation of the insidious idea of egalitarianism or "social justice," which holds that justice consists of the equality of results and that the process by which these results are attained is mere window-dressing.

The egalitarian theory of justice, despite its ancient lineage, is entirely bereft of any rational basis. (14) This theory destroys rationality in decision making and destroys rational distinctions between people, thereby undercutting genuine rational justice. To condemn the injustice in the antidiscrimination paradigm, we must reject this egalitarian theory of justice and uphold the virtue and morality of rationality.

Rational Discrimination as a Just and Moral Act

By the very fact that rational discrimination involves the rational, objective assessment of others, albeit based on very limited information, it is thereby a just and moral act. As Rand indicates, it involves as conscientious a judgment as one might make of any inanimate object, "with the same respect for truth, with the same incorruptible vision, by as pure and as rational a process of identification." In particular, it involves the use of all available information that has any predictive value in the inferential problem. It does not ignore information-bearing characteristics such as race, sex, or age. Rational discrimination does not involve any injustice or any weighting exercise "in which the claims of morality are on one side, and the claims of productivity are on the other," as D'Souza puts it.

It may be objected that rational discrimination can lead to an unfair result when our inferences about a person are incorrect. For instance, a young black man who is refused service by a taxi driver may not be a criminal--indeed, most are not. This young man suffers a difficulty that might have been avoided had the taxi driver ignored race, sex, and age in making in his decision. We have already seen, however, that this young man has no inherent right to a taxi cab's service or to any other service. The decision to provide this service is the prerogative of the taxi driver, who stands to suffer serious harm if he incorrectly judges a passenger not to be an assailant. (15) In assessing the results, the driver must weigh the loss from a false positive against the danger of a false negative. Because rational discrimination according to the rules of statistical inference produces the best inference of the risk of these errors, this assessment is best suited to an optimal weighing of risk versus reward. (16)

More important, although errors in judgment are unfortunate and do indeed lead to detrimental effects, they are the product of limited information, not of a defect specific to rational discrimination. To object to rational discrimination because it may lead to error is to regard omniscience as the standard of justice--a standard that any other inferential method, including adoption of the antidiscrimination paradigm, also fails.

Rational discrimination is not an insult to the affected individual, as those who champion the antidiscrimination paradigm often claim. A taxi driver's decision to decline service to a young black man who is hailing his cab is not a malicious snub or an unwarranted attack on his character. His assessment of the man means what it means: that he does not know the man's character, but he does know that a higher proportion of people with this man's observable characteristics are violent thugs who commit assaults against taxi drivers, and, hence, based on what the taxi driver knows, the risk that this man may be such a person is unacceptably high. This decision does not involve any direct data on the man's actions or character, which is precisely the complaint of those who regard the decision as an unwarranted insult. For this very reason, however, neither is it an implicit insult.

As already argued, rational discrimination does not manifest collectivism: the subject individual is not harmed for the sins of the group per se (although the driver's inference does depend on group behavior). The assessment involved in rational discrimination is an inference about the individual, albeit based on knowledge of statistical dependence between characteristics that is obtained from wider observations of groups. It can be clearly distinguished from collectivist judgment because rational discrimination springs solely from the absence of more direct information about the individual--as soon as we observe the characteristics of the individual that are more directly related to the characteristic of interest to us (in the sense that conditioning our judgment on this new information sufficiently reduces or eliminates the statistical dependence between race or sex and the characteristic of interest to us), the inferential connection between the individual and the group is severed or at least diminished.

The Injustice of the Antidiscrimination Paradigm

To engage in irrational discrimination is clearly unjust. If a person bases his assessment of others on irrelevant information or on demonstrably false theories (whether they are racist theories or not) and engages in discriminatory actions that are not warranted by the information available, then an action based on this assessment is indeed an injustice. It is so not because discriminatory action involves a violation of rights (it rarely does) and not because such actions harm or disadvantage the person discriminated against. No one has a right to the absence of detrimental treatment unless this treatment involves an aggression against person or property. Rather, irrational discrimination is unjust because it reflects the decision maker's failure to assess objectively and rationally the person discriminated against, using all the evidence available.

Not only is rational discrimination just, but the failure to engage in rational discrimination is itself unjust because it involves a failure to judge others rationally and objectively. In particular, the failure to discriminate rationally involves an evasion of the available evidence or of the inferential process. (17) Thus, we see that the antidiscrimination paradigm, which forbids all race discrimination, is itself unjust because it mandates the irrational.

This statement may seem a surprising claim. After all, rational discrimination often involves a negative judgment against a person, as in the case of taxi drivers who assess young black men as having a higher likelihood of criminality than the general populace. How then can the absence of negative judgment of others involve injustice? The answer is that to refrain irrationally from negative judgment does not differ in principle from irrationally refraining from positive judgment. Any irrational judgment of another person is an injustice, regardless of whether the judgment is negative or positive. As Rand puts it, "to withhold your contempt from men's vices is an act of moral counterfeiting, and to withhold your admiration from their virtues is an act of moral embezzlement" (1957, 933).

Of course, unlike judgments of a person's character based on more direct evidence, such as observation of his actions, inferences based on race discrimination are likely to be fairly weak. They are unlikely to be sufficiently reliable to warrant "contempt for a man's vices." Rather, they are likely to warrant only caution at the higher likelihood of a man's vices. Nonetheless, what is true for a person's known vices and virtues is also true (though to a lesser degree, owing to the greater likelihood of error) for a person's vices and virtues that are rationally inferred from the available evidence, even if this judgment includes inferences from characteristics that have only a correlation and not a causal link with these vices and virtues. (18)

An Objection--That Race and Sex are Unchosen and Unchangeable Characteristics

The fact that race is an unchosen characteristic and cannot be changed means that a person is not responsible for and cannot avoid rational discrimination on the basis of his race. The only way a person can diminish rational discrimination is to provide additional information to those who are attempting to judge him. For example, in the case of discrimination by taxi drivers, this demonstration may involve wearing more "respectable" clothing, smiling and displaying a friendly demeanor, and adopting some other observable characteristic that is negatively correlated with criminal conduct.

D'Souza's contends that "rational discrimination based on unalterable characteristics is [morally] problematic in a way that discrimination against high school dropouts and convicted felons, who are in a position to reform their circumstances, is not" (1995, 286-87), but if the moral is simply a species of the rational, then this contention is not true. Because rational discrimination is not predicated on an assessment of causal relationships between characteristics, it makes no difference whether a characteristic is changeable or not. Its presence is all that matters.

Although chosen and unchosen characteristics are treated identically for the purpose of statistical inference, they have one important empirical difference. Because chosen characteristics are based on the decisions of the individual we are assessing, they often have a stronger statistical relationship with the unknown characteristics about the individual that are of interest to us. Therefore, as a practical matter, chosen characteristics are often more important in making assessments about an individual, even in situations of very limited information. Thus, in trying to assess whether a prospective taxi passenger is likely to assault or rob the taxi driver, the person's clothing and demeanor may be more relevant than his race (though it is unlikely to be more relevant than his sex or age). Some black men report that they have difficulty catching taxis while wearing casual clothing, but that they can do so if they are wearing more formal clothing, such as a suit and tie (D'Souza 1995, 250). This difference arises from the driver's rational inference: he knows that men wearing suits are statistically less likely to assault him, even if they are also black. If, however, a person has chosen to adorn himself in gangland tattoos or act in a menacing or intimidating manner, then perhaps it is best for the driver to seek passengers elsewhere.

Contrary to fashionable theories of "social justice," rational discrimination is not unjust, and hence the total elimination of discrimination is not just. Although irrational discrimination is indeed an unjust phenomenon, rational discrimination is a just and moral act, involving as it does an objective and rational assessment of people on the basis of all available evidence. In truth, the antidiscrimination paradigm is unjust because it mandates the irrational suppression of relevant evidence. It is an attempt to supplant truth with egalitarianism.

Part III. Antidiscrimination Laws Are Tyrannical

To see the full meaning and consequence of antidiscrimination laws, we must first understand that discrimination in its widest sense--that is, the ability to draw relevant distinctions between things--is the basis of concept formation, which is the basis of reason. (19) Discrimination of some kind is therefore the basis of all rational thought. Not only is this statement true for discrimination in its widest sense, but it is also true mutatis mutandis for discrimination on the basis of persons' demographic or other attributes. That is, the ability to discriminate between different people on the basis of their observable characteristics is the basis for forming and using anthropic concepts--concepts pertaining to man. Discrimination is the means by which we identify and evaluate different people.

The antidiscrimination paradigm holds that certain human characteristics may not properly be used as the basis for discrimination--that they are not legitimate means by which to differentiate one person from another. At its root, however, the paradigm does not confine itself to particular demographic characteristics, such as race or sex. Rather, at its strongest, it condemns the act of discrimination per se, without reference to any particular characteristics.

This position has profound implications for human cognition. Without the mental act of differentiation on the basis of the different human characteristics--that is, without discrimination--no anthropic concepts can exist. Without such concepts, we are forced to treat each person as a distinct phenomenon, rationally incomparable with others. This condition does not entail that we are blind to the observable characteristics of those we meet, but only that we cannot abstract from these perceptual characteristics to gain conceptual knowledge about people. (20) We can still see the difference between dark and light skin because our eyes function in this way, but without the mental act of discrimination, we cannot abstract from this specific perception to the notion of race. As Rand observes, "Man's sense organs function automatically; man's brain integrates his sense data into percepts automatically; but the process of integrating percepts into concepts--the process of abstraction and of concept formation--is not automatic" (1964, 21).

The philosophy of antidiscrimination is designed to cripple this conceptual faculty. The antidiscrimination paradigm's goal and logical consequence are not merely political egalitarianism, but epistemological egalitarianism. The doctrine seeks not merely to secure the egalitarian provision of certain services and opportunities, but also to prevent the use of any anthropic conceptual distinctions, with political egalitarianism as a consequence of this wider aim.

Because the antidiscrimination paradigm cannot attack the perceptual faculty, which functions automatically, it attacks instead the conceptual faculty, particularly the higher-level concepts. It achieves egalitarianism at a conceptual level by attacking the epistemological and moral legitimacy of forming and using anthropic concepts. It cannot completely obliterate elementary concepts such as "dark," "light," "tall," "short," "man," "woman," and so on because these concepts are too easily derived from perceptions. So it attacks instead the legitimacy of the use of these concepts in decision making, with the ultimate goal of making conceptual distinctions between humans totally anathema.

Explicit Statements of the Goal of Epistemological Egalitarianism

Although the previous claim may seem a radical or even embellished one, it is substantiated both by the express statements of the backers of antidiscrimination laws and by the progression of these laws from initially innocuous assertions to increasingly extensive and repressive systems of government control. Indeed, an explicit statement of the full import of the antidiscrimination paradigm, as it pertains to race discrimination, can be found in the United Nations Declaration on the Elimination of All Forms of Racial Discrimination, which holds that "any doctrine of racial differentiation or superiority is scientifically false, morally condemnable, socially unjust and dangerous and has no justification in theory or practice" (United Nations 1963, preamble, para. 5, emphasis added).

To understand the full absurdity and wickedness of this statement, notice that the very concept of race used by the United Nations is itself based on a "doctrine of racial differentiation." This concept is the object of the United Nations' attack. The declaration regards any use of this concept as "scientifically false, morally condemnable, socially unjust and dangerous," which says more about this institution's conceptions of science, morality, and justice than it does about race discrimination.

Nor is this epistemological egalitarianism confined solely to matters of racial discrimination. The United Nations International Covenant on Civil and Political Rights directs that "the law shall prohibit any discrimination and guarantee to all persons equal and effective protection against discrimination on any ground such as race, colour, sex, language, religion, political or other opinion, national or social origin, property, birth or other status" (United Nations 1966, art. 26, emphasis added). This statement's absurdity is palpable. If applied consistently, it would grind the entire process of inference and cognition to an immediate halt.

Some may object that my interpretation is too literal. However, it should be clear from the immense amount of thought and effort the United Nations has applied to its various declarations and covenants that these statements have been carefully crafted and stated with full awareness of their profound epistemological implications. The statements' full absurdity and malevolence should not be read down, because an idea's progression is not determined by our fuzzy notions of when we wish to stop applying it; the progression is determined by the idea's logical consequences. So long as an idea is accepted, even if only on the basis of intellectual evasion or passive acquiescence, it will march inevitably toward its logical end, slowly grinding down all barriers of common sense or "pragmatism" placed in its way. In fact, this progression is precisely what we see when we observe the antidiscrimination paradigm's actual materialization in the evolution of antidiscrimination laws and the attendant intellectual culture of political correctness.

The Progression of Antidiscrimination Laws

Although it is useful to hear the antidiscrimination paradigm's actual materialization in explicit form--straight from the horse's mouth, as it were--we do not need to rely solely on abstract statements of principle to see the goal and logical progression of antidiscrimination laws. The evolution of these laws in practice demonstrates the goal.

From modest beginnings, antidiscrimination laws have expanded substantially in scope, breaking through every barrier of principle set against them, even the principles that have been espoused as the basis for the legitimacy of the initial antidiscrimination laws. These laws have metastasized far beyond the anticipations of those who have failed to appreciate the principle of epistemological egalitarianism at their root. Moreover, so long as this principle is accepted or at the very least remains immune to serious criticism, these laws will continue to progress further and further beyond the alleged goals of antidiscrimination activists and the expectations of shortsighted "pragmatists."

Beginning with equal-pay laws and legislation prohibiting race discrimination, antidiscrimination laws have extended to prohibit discrimination on grounds of sex, marital status, pregnancy, potential pregnancy, family responsibilities, age, disability, and the possession of palliative or therapeutic aids, interpreters, and guide dogs. These prohibitions often include prohibitions against "indirect discrimination," which occurs when discrimination on some other ground has a differential impact on the groups protected by discrimination laws. This progression of legislative prohibition has been the general pattern in many countries, including the United States, the United Kingdom, and Australia. (21)

To compound the tyranny of these laws, antidiscrimination legislation has often been constructed to allow accusations of "unlawful" conduct, which, unlike the concept of genuine criminal conduct, allows a complainant to bring a complaint before an executive government commission rather than a court, where the defendant is judged on the civil rather than the criminal standard of proof and can be (and often is) denied the right to legal counsel, the right to cross examine the complainant, the right to avoid self-incrimination, and other safeguards guaranteed to defendants in criminal prosecutions. Moreover, the stacking of these commissions with antidiscrimination activists also means that many defendants do not enjoy even the presumption of innocence. Of course, when such a commission publicly finds that a defendant engaged in "unlawful" conduct, the distinction between this finding and a judicial finding of criminal guilt is lost on most observers, which is precisely the intention.

Even ignoring the irrationality of the underlying antidiscrimination paradigm, the already wide ambit for claims, and the lower standards of proof, antidiscrimination law is rife with further abuse and vexatious claims--vexatious, that is, even considering the already lax standards for findings of guilt (McIntyre 1998).

Antidiscrimination laws continue to expand in scope to prohibit more and more kinds of discrimination in more and more contexts, with no sign of any end to the expansion. Moreover, these laws have broken through many of the principles that have been used to justify laws prohibiting race discrimination in the first place. Many advocates of antidiscrimination laws complain that race is not causally related to the characteristics that are important in our judgments of people. Yet the laws they support also prohibit discrimination on the basis of age and disability, both of which are causally related to a great many important characteristics. They also complain that race is an unchosen and unchangeable characteristic. Yet they support laws that also prohibit discrimination on the basis of marital status and pregnancy, characteristics that are chosen and changeable.

The inferential logic of rational discrimination has been raised in the courts in antidiscrimination cases, and the possibility of correlation between race and other characteristics has been recognized in the case law. However, this recognition has not led to any alleviation of prohibitions on rational race discrimination. Instead, it has had the opposite effect, leading to a de facto prohibition of discrimination on the basis of many nonracial characteristics that are correlated with race. In particular, antidiscrimination laws have prohibited many instances of decision making having a "disparate impact" on particular racial groups, even where this discrimination is based on nonracial characteristics and is perfectly rational. Of course, such cases will occur whenever nonracial characteristics are correlated with race, so this prohibition may potentially apply to a wide class of decisions.

Case law on disparate impact demonstrates that statistical evidence is often regarded as the substance of the case and that, as a general principle, disparity from an assumption of statistical independence of racial characteristics and treatment will be treated as prima facie evidence of unlawful discrimination, with defendants then being required to demonstrate the "legitimacy" and "predictive validity" of any test or rule that has a disproportionate impact on people of different racial groups (Jones 1982; Kaye 1982a, 775-76). (22) This requirement imposes substantial costs on defendants. Moreover, even though discrimination on the basis of nonracial characteristics that are correlated with race may be acceptable in cases where "validity" is demonstrated, it is clear from the terms of antidiscrimination legislation that one is not permitted to discriminate on the basis of race as a proxy for discrimination on the basis of some other characteristic of interest, even if it is rational to do so. If the basic goal of the antidiscrimination paradigm were not already clear, cases of disparate impact, which focus explicitly on the outcomes of decision making rather than on the process, demonstrate these laws' manifestly egalitarian goal and destroy any pretence that they are concerned with enforcing rational inference.

The Rise of Political Correctness

Enactment of laws against discrimination has been accompanied by a wider moral crusade against discrimination that has made it highly taboo to call attention to even the most obvious distinctions between groups of people. Truth has increasingly been subordinated to political correctness. As Murray Rothbard observes,
 In academic and literary circles "Political Correctness" is now
 enforced with an increasingly iron hand; and the key to being
 politically correct is never, ever, in any area, to make judgments
 of difference or superiority. Thus, we find that a Smith College
 handout from the Office of Student Affairs lists ten different
 kinds of "oppression" allegedly inflicted by making judgments about
 people.... [They include] "lookism" (or "looksism"), defined as the
 construction of a standard of beauty-attractiveness. "Oppression"
 is also supposed to consist not only of discriminating in some way
 against the unattractive, but even in noticing the difference.
 (2000, 290)

In light of the explicit philosophical statements by the United Nations and others as well as the consequent evolution of antidiscrimination laws, the antidiscrimination paradigm's goal should be clear: to make categorical distinctions between people unthinkable.

Using Anthropic Concepts to Identify and "Correct" Discrimination

Although the antidiscrimination paradigm's ultimate goal is to prevent the use of conceptual distinctions between people--to establish epistemological egalitarianism by destroying the legitimacy of anthropic concepts--this goal must necessarily involve the use of anthropic concepts, because one cannot eliminate what one cannot identify. Those who seek to prevent conceptual distinctions must use these concepts to identify when others are making them. Just as animals lack the conceptual ability to understand or even notice the use of higher-level concepts by humans, so, too, the actual destruction of anthropic concepts would make it impossible to prevent or even identify discrimination. The antidiscrimination paradigm therefore countenances one and only one use of anthropic concepts: in identifying acts of discrimination by others and "correcting" them. Thus arises the final irony of antidiscrimination laws: while advocating alleged color-blindness in evaluating people, they must require the establishment of an implicit quota system in which all people must be coerced into representative demographic groups in all areas of human activity, subject only to statistically insignificant disparities.

How Antidiscrimination Laws Establish an Implicit Quota System

Antidiscrimination laws have often been distinguished from quota programs and other forms of affirmative action on the grounds that quotas involve an explicit attempt to forcibly establish a particular demographic breakdown in human activities, whereas antidiscrimination laws allegedly do not do so. In practice, however, antidiscrimination laws must lead to an implicit quota system just as dangerous as the most explicit quota program, regardless of its supporters' alleged contrary intentions.

Because we cannot directly observe another person's reasoning, we must infer his reasoning from his actions. This inference can be a difficult process, especially if the person has an incentive to disguise his reasoning, as is the case where we are attempting to infer a motive that is punishable by law. In cases of unlawful discrimination, we expect a person who has breached the law to concoct plausible alternative reasons for his decision and to deny culpability. Thus, it will rarely be possible to determine his reasoning on the basis of his testimony.

To establish unlawful discrimination in this context, all antidiscrimination laws must rely on a counterfactual theory of human decision making, a theory of how decisions would have been made in the absence of discrimination. This theory is not merely a useful evidentiary tool, but also a necessary component of antidiscrimination law. Because the counterfactual theory tells us how things would occur in the absence of discrimination, deviation from the results predicted by this theory is regarded as prima facie evidence of discrimination, especially when the deviation is statistically significant. (23) Indeed, because the counterfactual theory is regarded as an objective means of inferring motive, whereas all other evidence is potentially veiled in deceit, deviation from the counterfactual theory is regarded virtually as proof of discrimination.

David Kaye discusses this issue and finds that statistical evidence may be sufficient to establish prohibited discrimination when coupled with a demonstration that the decision-making process is "susceptible

to abuse," which means merely that it can be used to discriminate on the basis of race or another prohibited characteristic (1982a, 776; see also Brilmayer 1982 and Kaye 1982b). As a matter of practical application, deviation from statistical norms determined by counterfactual assumptions will usually be treated as establishing a prima facie case for all but the most objective and transparent decision-making processes, and sometimes for them, too.

The notion that one requires a counterfactual theory of discrimination in order to judge whether people are engaging in prohibited discrimination is not merely a speculative insight. It is borne out by the actual legal reasoning used in antidiscrimination cases. In practice, deviation from alleged statistical norms based on a counterfactual theory of demographic breakdowns in human activity is indeed regarded as evidence capable of establishing that unlawful discrimination has occurred. Statistical evidence of deviation from counterfactual expectations has become so central to antidiscrimination law that in Otero v. Mesa County School District (408 F Supp 162 [1975], District Court of Colorado), Justice Winter of the District Court of Colorado was led to comment that Title VII cases under the Civil Rights Act are often "contests between college professor statisticians who revel in discoursing about statistical theory."

In practice, antidiscrimination laws must necessarily establish an implicit quota system based on the prevailing counterfactual theory of demographic breakdowns in human activity. The statistical norms set by the counterfactual theory form an implicit set of quotas that must be realized forcibly by law in all areas of human activity. Failure to adhere to these quotas is regarded prima facie as unlawful discrimination, and unless this failure is shown to have arisen from an objectively "legitimate" and "validated" procedure (which is very difficult and costly to demonstrate), it is punished accordingly.

Whether a political program of this kind, which seeks to achieve a counterfactual demographic breakdown in human activities, proceeds using the nomenclature of quotas, goals, factors, or something else, it must necessarily impose implicit quotas by virtue of the demographic theory at its base. Although other affirmative-action programs may take more obvious steps toward establishing demographic quotas, this difference is solely one of degree, rather than of kind. Because antidiscrimination laws enforce adherence to a particular demographic theory, they are themselves a form of affirmative action, imposing their own implicit quota system.

Proportional Representation and Demographic Uniformity in Human Activities

It should come as no surprise that the theory of demographic breakdowns in human activity that is used in antidiscrimination law is the egalitarian theory of proportional representation. This theory rests on the idea that without unlawful discrimination, people would engage in activities, employment, and so forth in proportion to their numbers in the population, subject to normal statistical variation. The result is that cases of disparate impact require little more than a demonstration that demographic outcomes differ from the general population. Kaye summarizes the evidential procedure for a claim of disparate impact as follows:
 In essence, a claim of disparate treatment is presented, and
 statistical evidence of the pattern of selection is therefore
 appropriate.... Where direct evidence of discrimination is
 unavailable, statistical methods have been pressed into service.
 Reasoning that if selections are independent of race the racial
 composition of the [people selected by the decision maker] can be
 treated as the result of a random process, Finkelstein (1978) and
 Zeisel (1969) have shown that a comparison of the results of the
 actual selection process with those predicted by a Bernoulli
 process [a process that assumes statistical independence between
 race and other characteristics] can be quite revealing. This is not
 to say, however, that serious questions cannot arise in testing,
 with statistical or other evidence, the null hypothesis that random
 (or, more precisely, race independent) selection is at work.
 (1982a, 775-76)

Although Kaye allows that serious questions can arise, they do so only with respect to testing whether the selection has been made in such a way as to be statistically independent of race (the null hypothesis to which he refers). The assumption that in the absence of discrimination the selection should be statistically independent of race is regarded as beyond question.

Given this counterfactual theory, with regard to any human activity, statistically significant deviations from the demographic proportions in the general population are taken to be the result of discrimination on the basis of these demographic characteristics or are regarded as cases of disparate impact. Thus, if black women form x percent of the general population, then, absent demographic discrimination, they should also form x percent of all bus drivers, schoolteachers, beer brewers, Supreme Court judges, and company executives (subject to statistically insignificant fluctuations).

One complication of this theory follows from the fact that some decisions involve selection from a subset of the general population rather than from the entire population (Kaye 1982a, 776). For example, in selecting a bus driver, the relevant population of people for selection may be the population of all minimally qualified applicants (say, all people with a bus driver's license) rather than the entire population. In this case, the logic of the antidiscrimination paradigm would require that if black women formed x percent of the general population, but only y percent of the population of minimally qualified applicants for bus-driving jobs, then absent demographic discrimination they should also form only y percent of all bus drivers.

This qualification may appear to allow demographic breakdowns in specific human activities to differ from the demographic breakdowns in the general population. After all, the proviso means that a bus company can legitimately have a lower proportion of black women as bus drivers than are in the general population without being found to have engaged in race discrimination. However, under the counterfactual theory of proportional representation, this situation is only temporary and must be regarded as a result of race discrimination occurring farther down the chain. So if the proportion of black women with a bus driver's license differs from their proportion in the general population, this difference must mean that discrimination is occurring in the licensing system rather than in the hiring of bus drivers, and so on farther back down the chain. Regardless of where the disparity arises, under the counterfactual assumption of proportional representation it must be regarded as a result of discrimination. Although disproportionate outcomes may be temporarily allowed farther up the chain of selection in cases where the pool of applicants is already disproportionate, the logic of the proportional-outcome assumption requires that disproportionate outcomes in the pool of applicants must itself eventually be corrected. Complete demographic uniformity in all areas of human activity is the ultimate goal and expectation.

This use of discrimination to "correct" disparities with counterfactual expectations explains why many so-called civil rights groups and other supporters of the antidiscrimination paradigm have become militant advocates of multiculturalism and attendant systems of affirmative action and demographic quotas, whether explicit or implicit. There is no mystery in this position: if one expects the elimination of racism and the rational and just treatment of individuals to result in demographic uniformity in human activities, then the absence of such uniformity must demonstrate ongoing racism. If the racism is not observable, it must be taking the insidious but undetectable form of "institutional" racism.

Although highly fashionable, the theory of proportional representation is entirely baseless and has been thoroughly debunked by empirical evidence. Discussing empirical evidence on this point, Thomas Sowell finds: "Even an approximate equality of 'representation' of different groups in different occupations, institutions or income levels has been a very rare--or nonexistent--phenomenon, except where such numerical results have been imposed artificially by quotas" (2004, 6-7). Far from forming an accurate counterfactual theory for antidiscrimination law, proportional representation in human activities has always been the exception rather than the rule, even in the absence of demographic discrimination.

The fundamental problem with antidiscrimination laws, however, is not the particular theory of proportional representation. Although this false theory aggravates the injustice of these laws, the implicit imposition of quotas by antidiscrimination laws follows from any counterfactual theory of demographic breakdown in human activities. As Sowell has observed, it is ridiculous to suppose that we can determine such counterfactual outcomes (1990, 129). However, even if we were able to do so accurately, any such counterfactual theory, no matter how accurate or absurd it might be, imposes an implicit quota system by force of law.

The malevolence of the antidiscrimination paradigm should not be underestimated. It is an irrational and unjust theory that aims for nothing less than the imposition of complete epistemological egalitarianism by gradually making anthropic concepts illegitimate. On the basis of the counterfactual presumption of proportional representation, it seeks to impose demographic quotas in all areas of human activity by the use of legalized thuggery. The intellectual culture of political correctness that it has spawned is an attempt to enforce this objective by moral intimidation, and the laws that it has spawned and will continue to spawn constitute a tyrannical attack on human reason itself.

The antidiscrimination paradigm is, however, a truly ingenious means of propagating egalitarianism, nothing short of Machiavellian in its brazen ambition--an ambition its supporters openly admit. Despite the paradigm's repugnancy, irrationality, and immorality, one cannot help but stand genuinely in awe of its ambitious nature and the success it has already achieved in placing certain anthropic concepts off-limits in polite discussion. After all, how can one ensure equal outcomes for different groups of people any better than by destroying the concepts that allow them to be differentiated in the first place? Whatever else the antidiscrimination paradigm demonstrates, it attests to the immense capacity to neuter the minds of large numbers of people with little more than belligerent moral intimidation. Whenever a person is too anxious to think about or mention an obvious difference between racial groups, lest he be called a racist or a bigot, the antidiscrimination paradigm is at work, crippling his conceptual faculty.


Bamshad, M., and S. E. Olson. 2003. Does Race Exist? Scientific American Magazine (December): 78-85.

Bamshad, M., S. Wooding, B. A. Salisbury, and J. C. Stephens. 2004. Deconstructing the Relationship Between Genetics and Race. Nature Genetics 5: 598-609.

Bernardo, J. M., and A. F. M. Smith. 1994. Bayesian Theory. Chichester, U.K.: John Wiley.

Brilmayer, L. 1982. Statistical Evidence of Discrimination: Comment. Journal of the American Statistical Association 77: 789-90.

Chabris, C. F. 1998. IQ Since "the Bell Curve." Commentary 106, no. 2: 33-40.

Corlett, J. A. 1993. Racism and Affirmative Action. Journal of Social Philosophy 24: 163-75.

de Finetti, B. 1980. Foresight: Its Logical Laws, Its Subjective Sources. In Studies in Subjective Probability, edited by H. E. Kyburg and H. E. Smokier, 93-158. New York: Dover.

Devlin, B., S. E. Fienberg, D. P. Resnick, and K. Roeder, eds. 1997. Intelligence, Genes, and Success: Scientists Respond to the Bell Curve. New York: Springer.

Dretske, F. 1969. Seeing and Knowing. London: Routledge and Kegan Paul.

D'Souza, D. 1995. The End of Racism. New York: Free Press.

--. 1999. When Discrimination Makes Sense. Hoover Digest 4. Available at:

Edwards, A. W. F. 2003. Human Genetic Diversity: Lewontin's Fallacy. BioEssays 25, no. 8: 798-801. Also available at:

Finkelstein, M. 1978. Quantitative Methods in Law. New York: Free Press.

Fischer, C. S., M. Hout, M. S. Jankowski, S. R. Lucas, A. Swidler, and K. Vos. 1996. Inequality by Design: Cracking the Bell Curve Myth. Princeton, N.J.: Princeton University Press.

Fishburn, P. C. 1981. Subjective Expected Utility: A Review of Normative Theories. Theory and Decision 13: 139-99.

Fishkin, J. 1975. Justice and Rationality: Some Objections to the Central Argument in Rawls's Theory. American Political Science Review 69, no. 2: 615-29.

Hacker, Andrew. 1992. Two Nations: Black and White, Seperate, Hostile, Unequal. New York: Ballentine.

Herrnstein, R. J., and C. Murray. 1994. The Bell Curve: Intelligence and Class Structure in American Life. New York: Free Press.

Jones, D. H. 1982. Statistical Evidence of Discrimination: Comment. Journal of the American Statistical Association 77: 787-88.

Kaye, D. 1982a. Statistical Evidence of Discrimination. Journal of the American Statistical Association 77: 773-83.

--. 1982b. Statistical Evidence of Discrimination: Rejoinder. Journal of the American Statistical Association 77: 790-92.

Lad, F. 1996. Operational Subjective Statistical Methods: A Mathematical, Philosophical, and Historical Introduction. New York: Wiley Interscience.

Lewontin, R. C. 1972. The Apportionment of Human Diversity. In Evolutionary Biology 6, edited by T. Dobzhansky, M. K. Hecht, and W. C. Steere, 381-98. New York: Appleton-Century-Crofts.

Lieberman, L., R. C. Kirk, and A. Littlefield. 2003. Perishing Paradigm: Race, 1931-99. American Anthropologist 105, no. 1:110-13.

Martin, M. G. F. 1992. Perception, Concepts, and Memory. Philosophical Review 101: 745-64.

McIntyre, A. 1998. Anti-discrimination Law and Its Abuses. Proceedings of the XIXth H. R Nichols Society Conference (August). Available at: vol19/vol19-7.php.

Murray, C. 1998. Income Inequality and IQ. Washington, D.C.: American Enterprise Institute.

Nolan, R. 1994. Distinguishing Perceptual from Conceptual Categories. In Philosophy and the Cognitive Sciences, edited by R. Casati, B. Smith, and G. White, 221-31. Vienna: Holder-Pichler-Tempsky. Available at: distinguishing_perceptual_from_conceptual_categories.pdf.

Pearl, J. 2000. Causality. Cambridge, U.K.: Cambridge University Press.

Rand, A. 1957. Atlas Shrugged. New York: Signet.

--. 1964. The Virtue of Selfishness. New York: Signet.

--. 1984. Philosophy: Who Needs It? New York: Signet.

--. 1990. Introduction to Objectivist Epistemology. Expanded 2d ed. New York: Meridian.

Reilly, K., S. Kaufman, and A. Bodino, eds. 2003. Racism: A Global Reader. New York: M. E. Sharpe.

Ridley, S. E., J. A. Bayton, and J. H. Outtz. 1989. Taxi Service in the District of Columbia: Is It Influenced by the Patron's Race and Destination ? Washington, D.C.: Lawyers' Committee for Civil Rights under the Law.

Risch, N. 2000. Searching for Genetic Determinants in a New Millennium. Nature 405: 847-56.

Rothbard, M. N. 2000. Egalitarianism as a Revolt Against Nature and Other Essays. Auburn, Ala.: Ludwig von Mises Institute.

Smith, T. 2006. Ayn Rand's Normative Ethics. New York: Cambridge University Press.

Sowell, T. 1990. Preferential Policies: An International Perspective. New York: William Morrow.

--. 2004. Affirmative Action around the World: An Empirical Study. New Haven, Conn.: Yale University Press.

Stove, D. 1999. Racial and Other Antagonisms. In Against the Idols of the Age, by D. Stove, 137-52. Edison, N.J.: Transaction.

Thompson, A. 2003. Issues at Stake in Eighteenth-Century Racial Classification. Cromohs 8: 1-20.

United Nations. 1963. "Declaration on the Elimination of All Forms of Racial Discrimination." Resolution 1904 (XVIII) of the General Assembly, November 20. Available at:

--. 1966. "International Covenant on Civil and Political Rights." Resolution 2200A (XXI) of the General Assembly, December 16. Available at: menu3/b/a_ccpr.htm.

Welsh, J. F. 2008. After Multiculturalism: The Politics of Liberty and the Dialectics of Race. Lanham, Md.: Lexington Books.

Williams, W. 1995. Do the Right Thing: The People's Economist Speaks. Stanford, Calif: Hoover Institution Press.

Zeisel, H. 1969. Dr. Spock and the Case of the Vanishing Women Jurors. University of Chicago Law Review 37: 1-18.

(1.) For present purposes, it suffices to note that a "race" of people refers to a definite group of people sharing certain physiological characteristics. Similarly, a reference to the "characteristics of a race" of people is a reference to the aggregated characteristics of this group of people (only where individual characteristics can be meaningfully aggregated in some way). Because there is no a priori reason to suppose that aggregate differences between groups of people cannot exist, there is similarly no a priori reason to rule out the possibility that races may have different characteristics. In this view, racial differences are an empirical matter to be determined by observation rather than by pronouncement.

(2.) The particular physiological attributes used as the foundation for the concept of race in anthropology is a matter of ongoing debate, and there is extensive scientific literature on the subject. For overviews, see Risch 2000, Bamshad and Olson 2003, and Bamshad et al. 2004. For discussion on the declining acceptance of the previous classifications, see Lieberman, Kirk, and Littlefield 2003.

(3.) For an argument over the scientific legitimacy of human racial classification, see Lewontin 1972 contra Edwards 2003.

(4.) In this sense, racism probably did not exist as an explicit theory until the Enlightenment. Until then, there was no real conception of human genetics and hence no conception of causal genetic factors. Earlier beliefs about the causes of racial differences certainly existed, but most people viewed differences between racial groups as the effect of climate differences rather than of genetics (see Thompson 2003).

(5.) For discussion of the emergence of the thesis of racism during this period, see D'Souza 1995, 25-66.

(6.) Stove can probably be forgiven for adopting the definition of those hypersensitive hysterics who scream "racism" at even the most obvious empirical observations, especially because he uses this definition in a consistent manner to deliver an intellectual haymaker to these same hysterics. By adopting their own false definition of racism as a mere empirical relationship between race and other traits, he correctly identifies that "[i]t is this proposition which is nowadays constantly declared to be false, though everyone knows it to be true" (1999, 147).

(7.) For debate on the degree to which genetic factors and environmental factors influence cognitive development and intellectual capacity, see Herrnstein and Murray 1994; Fischer et al. 1996; Devlin et al. 1997; and Murray 1998. For an overview, see Chabris 1998. Despite the common accusations of racism in such debate, this issue is material to the theory of racism only insofar as it involves hypotheses about the genetic determinism of a person's convictions, values, and character (matters that are philosophical questions rather than matters of biology).

(8.) For a discussion of quantitative coherence theories that lead to the rules of probability theory, see Bernardo and Smith 1994, chap. 2, especially 23-33, 83-85, and Lad 1996. For a comprehensive review of statistical decision theories that lead to the application of probability theory based on the maximization of expected utility, see Fishburn 1981.

(9.) A Dutch book inconsistency occurs when our degree of certainty in various events is such that there exists some combination of bets that would lead us to certain loss regardless of the outcome of the bets. Bruno de Finetti (1980) used this idea as the basis for his quantitative coherence theory.

(10.) In this article, I refer to both statistical dependence and correlation. The latter measures only the strength of the linear relationship between variables and is only one type of statistical dependence (other types do not entail such correlation).

(11.) A study by the Lawyers' Committee for Civil Rights under the Law found that taxi cab drivers in Washington, D.C., are less likely to pick up black men than other people and are less likely to drive passengers of any race to areas of the city with larger proportions of the black population. See Ridley, Bayton, and Outtz 1989.

(12.) In statistical parlance, we would say that race and crime are marginally correlated (that is, in the absence of conditioning information), but are conditionally uncorrelated (based on any new information that renders the correlation to be nil).

(13.) Philosopher Tara Smith discusses at length the virtue of justice and its derivation from rationality, pointing out its lineage in Benedict de Spinoza's A Theologico-Political Treatise; Thomas Aquinas's Summa theologica, question 58, art. 1; and Andre Comte-Sponville's A Small Treatise on the Great Virtues. See Smith 2006, 135-75. Elements of this rationality aspect of justice also appear in Aristotle's notion of distributive justice based on merit in The Nicomachean Ethics.

(14.) For a discussion of the nature of egalitarianism, see Rothbard 2000, 1-20. For a demolition of John Rawls's egalitarian theory of justice, see Rand 1984, 102-19, and Fishkin 1975.

(15.) Of course, depending on the terms of the driver's employment, this decision may instead be the taxi company's prerogative. In any case, the prospective passenger has no inherent right to the service.

(16.) The magnitude of the damage to be suffered is also part of the consideration, so it may be objected that the taxi driver is concerned only with his own costs and benefits, ignoring the negative impact on the prospective passenger. However, in a free market, where the taxi driver is compensated for picking up passengers, the passenger would be free to offer a premium to the driver for the taxi ride, thus factoring in the value to the passenger. All that is lacking for such an arrangement is sufficient entrepreneurialism to solve the problem of limited information and the repeal of laws prohibiting such arrangements.

(17.) This statement should not be taken to mean that all errors in judgment are unjust. One must distinguish between shortcomings in knowledge and the willful evasions of available evidence, both of which may lead to error.

(18.) Indeed, many philosophers contend that all knowledge is only a matter of probability. Although I do not accept this view (in fact, I consider it self-contradictory because its alleged validity rests on the fact that it is implicitly a statement of absolute knowledge and not just a statement of probability), it is illustrative for certain empirical observations, particularly those involving character assessments of others. Many such judgments are indeed the result of an accumulation of evidence leading to a higher and higher degree of confidence, which we eventually designate as knowledge.

(19.) In particular, concepts are formed by differentiating between different existents and grouping these existents according to a process of measurement omission. For a detailed treatment of this topic, see Rand 1990.

(20.) The distinction between perception and conceptual understanding is detailed in Rand 1990 and in other works on epistemology, including Dretske 1969, Martin 1992, and Nolan 1994.

(21.) In Australia, the Racial Discrimination Act of 1975 (Cth) legally prohibited discrimination on the basis of race, color, descent, or national or ethnic origin within certain contexts such as the provision of employment, goods, services, and housing or accommodation. The scope of antidiscrimination laws was expanded by the Sex Discrimination Act of 1984 (Cth), which contains similar prohibitions against discrimination on the grounds of sex, marital status, pregnancy or potential pregnancy, and family responsibilities; then by the Disability Discrimination Act of 1992 (Cth), which contains similar prohibitions against discrimination on the grounds of disability or the consequent possession of palliative and therapeutic devices and auxiliary aids, interpreters, readers, assistants, guide dogs, hearing assistance dogs, and other trained animals; and then by the Age Discrimination of Act 2004 (Cth), which contains similar prohibitions on age discrimination.

(22.) In U.S. discrimination law, statistical evidence is more likely to be sufficient in cases under civil law rather than in constitutional cases (Brihnayer 1982).

(23.) Meaning that it is unlikely to have arisen merely by random deviation. In particular, statistical significance tests work by assuming the counterfactual theory to be true and then asking how likely it is that a deviation from the theory at least as large as the one observed would occur under the hypothesis of nondiscrimination (that is, assuming that the person did not discriminate). If this probability is sufficiently low, then the hypothesis of nondiscrimination is rejected as implausible, and the person is found to have discriminated.

Ben O'Neill is a Ph.D. student in the School of Finance and Applied Statistics at Australian National University.
COPYRIGHT 2009 Independent Institute
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:O'Neill, Ben
Publication:Independent Review
Date:Mar 22, 2009
Previous Article:Wal-Mart to the rescue private enterprise's response to Hurricane Katrina.
Next Article:Urban land-use controls and the subprime mortgage crisis.

Related Articles
New Paradigm Resources Group Study Highlights Companies Positioned to Capitalize on Local Network Broadband Shortage.
Fouere's messages for government are well-intended.

Terms of use | Copyright © 2017 Farlex, Inc. | Feedback | For webmasters