A. Intimate associations

Let me begin with what are called “intimate associations.” Suppose the government were to engage in picking people's spouses. “Preppy Jewish guys must marry pierced Wiccan girls.” Or: “Unfashionable secular academics may not marry Evangelical Yalies.” You get the picture. There is, after all, some precedent here. Southern states forbade interracial marriages until the Supreme Court ended this in Loving v. Virginia (1967).Footnote ¹ And, of course, as of the time I write this, most states forbid same-sex marriages, polygamous marriages, marriages of siblings, and so on.

Such laws by themselves, in denying the right to attain the status of marriage, primarily deny the legal incidents of that status (for example, with respect to intestate inheritance, income taxation, or guardianship in case of incompetency). Beyond that, they do not forbid freedom of association. But they might be supplemented by other laws that encroach on freedom of association more considerably.

Most obviously, such laws might forbid sexual contact outside of formal marriage. Adultery, cohabitation, and fornication statutes did this, though in modern times they were rarely enforced, at least against heterosexuals.Footnote ² But if the legal status of marriage is narrowly restricted in terms of choice of spouse, and this restriction is buttressed by restrictions on sexual contact outside of marriage, freedom of one kind of association—sexual relations—will be effectively curtailed.

But the government might conceivably go further still than governments have traditionally gone in the past. It might restrict not only sexual contact outside of marriage but also emotional attachments outside of marriage—assuming government does not believe that social norms are doing that job sufficiently. After all, many consider a spouse's emotional attachment to another to be as grave a breach of the obligation of marital fidelity as a spouse's sexual relations with another. And whatever governmental interest supports its enforcement of spouses' obligations of sexual fidelity presumably also supports its enforcement of their obligations of fidelity more generally.

However, here things get very interesting. Sexual infidelity might be relatively easy for government to define—Bill Clinton notwithstanding—and to punish, but what acts would constitute the crime of emotional infidelity? Presumably the emotions themselves are not a proper basis for criminal punishment, so government would have to punish acts that (1) were expressive of emotional infidelity, and (2) were under the voluntary control of the actor (and thus fairly punishable). We might imagine, for example, that government would have to forbid phone calls of more than five minutes or more often than once a day to someone other than one's spouse. But what if the recipient of the call were a professional colleague? Of course, one can have an affair with a colleague. Or what about conversations with one's mother, or childhood best friend? These messy details would have to be sorted out, and sorting them out might prove practically impossible.

The government might then be tempted to enforce the spousal association by placing affirmative duties on the spouses rather than those difficult-to-define negative ones. It might affirmatively require, with the backing of sanctions, that the spouses “love, honor, and obey” one another, as they have vowed to do. But the affirmative-duties approach to enforcing marital relations is even messier than the negative-duties approach. Love cannot be commanded. And what affirmative acts would express true honor? Obedience might be easier, though not without spouses ordering each other about in code-like fashion, consistently with the principle of legality.

Government could demand that spouses speak to each other a certain amount each day (in terms of sentences? words? syllables? duration? topics?), have sex with each other a specified number of times each week (what kind? what quality?), or give each other gifts (of what amount?) or flowers (which?).Footnote ³ And government could demand more than that spouses share whatever incomes they earn during their marriage, as governments currently do (although without micromanaging how the household budget is actually allocated or how allocation decisions are actually made); governments could go further and demand that spouses work at a specified occupation for a specified number of hours to ensure the other spouse is not deprived because the first has independent means.

What all of the preceding demonstrates is that even with respect to an “association” that the government has traditionally regulated, and regulated for reasons that I shall take to be proper bases for governmental regulation, restricting freedom to associate and freedom not to associate is an immensely complex and difficult task. Even where the choice of one's legal spouse is restricted, and even with a government prepared to enforce fidelity and other marital vows, government's task is daunting, especially if social norms are not supporting its efforts. And that is true not only with respect to affirmative duties, as one might expect, but also with respect to negative ones if one prohibits more than specific adulterous sexual acts. And imagine the difficulties if government ventured beyond marriage and tried to compel certain friendships or to disallow certain ones. What specific acts would it have to forbid or compel, and could it possibly achieve whatever purposes might motivate it?

Why does the government regulate marriage, the only intimate association it regulates at all? That is a question currently much mooted in the context of the debate over gay marriage.Footnote ⁴ Fostering the best environment possible for the rearing of children is one oft-cited reason.Footnote ⁵ And protecting reliance interests of spouses who become vulnerable to economic hardship by forgoing careers is another.Footnote ⁶ Conceivably, however, the government might decide that individuals are poor at choosing the mates most conducive to their flourishing, or that they need external pressures to forge the most beneficial intimate relationships. Or the government might decide that certain marriages—for example, interracial ones—would be socially beneficial (perhaps by destroying racial identities). Those are the reasons a benignly motivated government might attempt to curtail freedom of intimate association. What the previous paragraphs illustrate is just how difficult and intrusive such a task would be.

B. Associations regarding voting and political expression

Governments have attempted in various ways to curtail the associative freedoms of their citizens qua voters and political advocates. In some cases, as in modern Germany with respect to the Nazi Party, political parties with certain ideologies are prohibited from appearing on the ballot.Footnote ⁷ Those with Nazi sympathies can presumably meet somewhere and discuss their common political views, but they cannot deem their organization a “political party” such that its candidates can appear on the ballot as its candidates. Essentially, modern Germans with Nazi sympathies, though they may associate in other ways, may not “associate” their ballots.

In the United States, the Supreme Court in recent years has tended to support political parties' freedom to organize themselves as they choose.Footnote ⁸ In other words, the party members may choose, free from governmental interference, the terms on which they associate qua party members. For example, if they wish to hold a closed primary (an election of the party's candidates that is restricted to registered party members)—or an open one (one open to voters of any party)—that is their prerogative.Footnote ⁹

By contrast, in the so-called White Primary cases, and especially in Terry v. Adams, Footnote ¹⁰ the last of those cases (decided in 1953), the Court itself restricted party members' freedom to associate on terms of their choosing. After the Fifteenth Amendment outlawed denials of the right to vote on racial grounds, many Southern states attempted to continue restricting blacks' votes through various other means, some subtle, some not. Overt discrimination moved from the general election, where it was constitutionally forbidden, to the primaries of the Democratic Party, which was the dominant political party in those states. In the White Primary cases, the Supreme Court held that the Democrats had to open their primaries to black voters.

In Terry, the Court prohibited a county in Texas from giving effect to what amounted to (successful) racial bloc voting. Democrats far outnumbered Republicans in the county, and thus the Democratic primary winners always prevailed in the general elections for county offices. Further, white Democrats appreciably outnumbered black Democrats within the county's Democratic Party. Although the Court in one of the earlier White Primary cases had forced the Democratic Party in Texas not to discriminate against black voters in its primariesFootnote ¹¹—which, after all, were run by the state itself out of tax revenues—and the Democratic primary in this county was indeed open to black voters, the white Democrats organized themselves into the Jaybird Club and held their own, privately supported “pre-primary primary.” The winning Jaybird candidates then ran in the Democratic primary, in which they always won, and then in the general election, with the same results. In other words, the Jaybirds were successful racial bloc voters. The Court, in a most unedifying and poorly reasoned opinion even by its standards, held that something was constitutionally rotten about this and ordered it stopped.

A moment's reflection will reveal how strange the Court's decision was. Presumably, the Court wanted the Jaybirds either to admit blacks, to disband, or at least not to hold pre-primary elections. Admitting blacks to the Jaybirds would be pointless, at least as members entitled to vote, as that would make the Jaybirds merely a duplicate of the county's Democratic Party. That is so unless the white Jaybirds would then form the Bluebird Club and hold pre-pre-primary elections under its auspices—which would replicate the problem of the Jaybirds.

So suppose the Jaybird Club disbanded. And suppose, in its place, the white Democrats communicated with each other by phone or post (or today, e-mail) and came to an agreement among themselves to support a particular slate of white Democratic candidates. The ultimate results of this racial bloc voting scheme would be the same as those produced by the Jaybirds. The Supreme Court would not like this, but what could it do? Presumably, it could not prohibit white Democrats from talking politics with one another, as this would explicitly violate the First Amendment. Perhaps it could order them not to “agree” with one another—form a cohesive bloc—but it is not clear that it could do so. Nor is it clear that racial bloc voting is different in any principled way from interest-group bloc voting or from ideological bloc voting. (The Court has never argued that prohibitions on the government's taking race into account in governing entail prohibitions on the citizens' taking race into account in electing the government itself; and surely a candidate's race as well as ethnicity, gender, age, and religion have been central considerations for both party officials and citizens.)

C. Creedal organizations

One of the most common bases for association and non-association has been that of beliefs. Those who share certain beliefs associate with one another in order to discuss, strengthen, and promote such shared beliefs. The beliefs may be religious ones, and the associations may be churches, synagogues, mosques, campus ministries, and a multitude of orders, charities, study groups, and the like. Or the beliefs may be philosophical, political, or ethical.

Assuming the government is not attempting to stamp out the beliefs by forbidding speech that advocates or even merely transmits those beliefs, restrictions on this form of freedom of association make little sense. Yet governments have attempted to restrict such freedom. No government has, to my knowledge, ordered Catholics to admit women, homosexuals, or Jews to the priesthoodFootnote ¹²—or ordered the Ku Klux Klan to admit blacks into its ranks. But the state of New Jersey did forbid the Boy Scouts to revoke a scoutmaster's membership on the ground that he was a homosexual. In Boy Scouts of America v. Dale (2000),Footnote ¹³ the U.S. Supreme Court struck down this application of New Jersey's antidiscrimination law on the ground that forcing the Scouts to allow homosexuals to be members and scoutmasters would alter the Scouts' “messages” and thus violate their First Amendment right of freedom of expressive association. However, the New Jersey nondiscrimination law in question no more affected the Scouts' ability to send their preferred messages than any other law affecting a group's ability to control its membership. If antidiscrimination laws dictate that the New York Times may not discriminate against blacks or women in hiring reporters or choosing columnists, the content of what appears in its pages will undoubtedly be affected. Laws regulating membership in any organization—including commercial ones—will affect the content of that organization's expression. In that respect, the Dale decision was correct. That “message effect,” however, is only an incidental effect of a law that is not enacted for the purpose of affecting messages; and laws that are not aimed at messages and have only incidental message effects are not ordinarily violative of the First Amendment.Footnote ¹⁴ That is why Ollie's Barbeque could not rename itself Ollie's Racist Barbeque and then claim protection from public-accommodations laws on the ground that admitting blacks would denature its racist “message.”Footnote ¹⁵

The more basic interest at stake in Dale and in other collisions between antidiscrimination laws and membership criteria is the interest in associating with like-minded people around certain beliefs, including beliefs bearing on membership in the group itself. If that interest finds protection in a proper account of freedom of association, then the Black Panther Party should not have to admit whites on an equal basis even if those whites subscribe to all the tenets of the Black Panthers other than those regarding membership. Catholics should not have to admit women to the priesthood even if those women subscribe to all Catholic tenets other than that of a male priesthood. In short, a creedal organization should be able to organize around its creed, whatever the creed is, so long as it is not criminal, and even if that creed entails restrictions on membership. If it is part of the Boy Scouts' creed that homosexuals are not to be members, then it is logically impossible for a homosexual such as Dale to pledge adherence to the Scouts' creed and still wish to be a member. Government's forcing Dale's membership on the Scouts would be tantamount to government's denying the right to have a creedal association organized around a creed that endorsed discrimination of a prohibited type. The Supreme Court seems to agree up to a point: So long as it deems the organization to be an “expressive” organization rather than primarily commercial or civic, the Court will protect the organization's membership criteria from regulation under the First Amendment.Footnote ¹⁶ Nonetheless, as the next section illustrates, this places too great a burden on distinguishing creedal from noncreedal organizations and overlooks the “creedal” aspects of all voluntary organizations.

D. Clubs and other voluntary organizations

Once we move beyond religious and political groups, which are obviously creedal, we enter the vast realm of voluntary nonmarketplace associations. There may be, for example, ascriptive identity groups of great variety: the Downtown Women's Club; the Black Men's Fraternal Organization; the Greek-American Club; the Gay Men's Club; and so forth. There may be groups devoted to certain hobbies and activities—the Harley-Davidson Owner's Club; the Chess Club; the No-Carts Golf Association; the World Affairs Discussion Group; the Debating Society; and so forth. (The Boy Scouts and Girl Scouts and many other associations are both ascriptive—restricted on the bases of gender as well as sexual orientation—and also devoted to particular activities—as well as being creedal.) There may be groups devoted to certain virtues or accomplishments—the Veterans of the Korean War; the Abstainers Club; the 20,000 Footers Mountaineers Club; and so forth.

It is again unclear why a government would want to regulate membership in such organizations. Would it make sense to require a Women's Club or a college sorority to admit males? A Black Men's club to admit Chinese? The Harley-Davidson Owners Club to admit Kawasaki owners? The Greek-American Club to admit Italian-Americans? The Veterans of the Korean War to admit nonveterans? The Debating Society to admit nondebaters? And so on. Were government to do so, it would be tantamount to declaring that one may not have a women's club, a Greek-American club, or a veterans club. But so long as the activities of these associations are themselves noncriminal, why would government wish to thwart voluntary associations formed around these activities or characteristics? (I suggest some answers to this question in Section IV below.)

Moreover, with a flashback to the earlier discussion of Terry v. Adams, Footnote ¹⁷ what exactly would government be demanding by requiring these associations to admit unwanted members? Would the old members have to talk to the new members? (If so, how much, and on which topics?) Be nice to them? Take their opinions into account? If the new members must be allowed to vote, must the old members refrain from bloc voting? Must they refrain from organizing activities among themselves (i.e., forming a club within a club)? Governmental regulation of membership in otherwise voluntary associations carries with it the impossible task of government's micromanaging the activities of those associations to make good its promise of nondiscrimination.

E. Games and activities

The multitude of sports, games, and similar recreational activities affect both how associations are defined and whom they permit to participate. The 2001 Casey Martin case is famous for having put the U.S. Supreme Court in the position of questioning whether the PGA's “no cart” rule was essential to professional golf tournaments.Footnote ¹⁸ Justice Antonin Scalia, in his dissent, scoffed at the idea that there is some Platonic form of golf that does or does not require that competitors walk.Footnote ¹⁹ But the Court's majority felt compelled to decide the question. (It decided that the “no cart” rule was inessential and thus that Casey Martin, a golfer with a disability that made walking difficult, could use a golf cart in PGA competition.)

Justice Scalia's point remains and appears to expose the Court majority's position as completely arbitrary. As Scalia noted, the Court would be hard-pressed to justify a baseball league's refusal to allow a fourth strike to the myopic. He might also have pointed to Major League Baseball's rule barring midgets, a rule instituted after an owner, Bill Veeck, placed Eddie Gaedel, a midget, on the St. Louis Brown's roster and inserted him as a pinch hitter in order to draw a base on balls (which he did, on four pitches).Footnote ²⁰

The rules of sports and games are entirely artificial. But whatever they are, they will advantage some and disadvantage others. Were the basket in basketball four feet lower, the diminutive former NBA player Muggsy Bogues would be advantaged and giants like Shaquille O'Neal disadvantaged.

Moreover, sports and games frequently define not only how they are to be played but who can play. Most colleges and universities, for example, compete in “women's basketball” or “women's volleyball” or “women's soccer.” Men are not eligible to play. In contrast, no university or conference of universities in the modern era would even think about offering “White basketball” or “Asian volleyball.” Yet the typical rationalizations for gender-identified sports teams turn on average physical differences between men and women, and average physical differences are equally available to rationalize having racially-identified teams. When this point is pressed, the fallback explanation for the difference (between gender-identified sports teams and racially-identified ones) is that women's sports are just “different” from the men's versions, even if the rules are the same. But I imagine that one could make similar claims about racially-identified sports. Indeed, one frequently hears claims that black basketball players play a “different game” from that played by whites, or perhaps the less controversial claim that Latin American or European soccer differs from North American soccer.

The Casey Martin case came to the courts as a case of discrimination against the disabled (Casey Martin) by a place of public accommodation (a PGA tournament). Even though that interpretation of federal civil rights statutes was strained,Footnote ²¹ it is nonetheless true that sports and games have a commercial aspect, and that how they define themselves and who is eligible to compete can look very much like ordinary marketplace criteria defining what products and services are available and who is best qualified to produce them.

F. Marketplace associations

Let me turn now to marketplace (a.k.a. commercial) associations. We are quite familiar with governmental interferences with individuals' freedom of association in the marketplace, principally through various antidiscrimination laws. Government does not tell employers specifically whom to hire. Rather, it tells them only on what grounds they may not refuse to hire (or on what grounds they may not fire), grounds such as race, ethnicity, gender, age, and religion.

This appears to be, and indeed has proven to be, a fairly manageable project. The employer may not refuse to hire, say, an otherwise qualified black or woman for a particular job. But suppose the employer replies, “Okay, I'll ‘hire’ so-and-so, but I won't put her to work or pay her.” The government's response will undoubtedly be this: The employer can leave the worker in question idle if the employer so wishes, but the employer must pay her at the rate that the employer has set for the particular job in question. The employer will therefore either take a loss (by paying without receiving labor in return) or utilize the worker for the job in question.

But this point about whether the obligation not to discriminate on specified grounds in employment entails further obligations, such as to pay (yes), or to give work to (immaterial, given the obligation to pay), points out the ambiguities inherent in the obligation not to discriminate, ambiguities that have been bedeviling courts in the U.S. for several years. For example, does the obligation not to discriminate against Xs entail an obligation not to discriminate on grounds that have an adverse disproportionate impact on Xs?Footnote ²² Does it entail an obligation not to discriminate on the basis of traits highly correlated with Xs or with a subset of Xs?Footnote ²³ And does it entail obligations regarding behavior toward Xs other than work and pay, such as refraining from making snide or condescending remarks to them, telling jokes about them, shunning them, engaging in speech that offends them or their religious beliefs, and so on?Footnote ²⁴

These reflections on freedom of association and antidiscrimination laws in the marketplace can lead in a multitude of directions. If government is really concerned with stamping out discrimination against Xs in the marketplace, the implications might be quite extreme. Consider discrimination on the bases of race, ethnicity, gender, age, and religion, the principal forbidden grounds in American antidiscrimination law. What should they entail for the following: Playboy clubs (which hire only young women)? Hooters restaurants (same)? The cover girls for Vogue magazine (same)? Beauty pageants (same)? The cover girls for Ebony magazine (same, plus they must be African-American)? The Miss Greek-American beauty pageant (open only to young women of Greek ancestry)? Business prayer meetings (which may make employees of certain faiths or of no faith feel like outsiders)? Hiring only Native-Americans to satisfy consumer demand for jewelry “handcrafted by Native Americans”? Hiring only Shakers to satisfy consumer demand for “furniture made by Shakers”? And so on, virtually ad infinitum. Race, ethnicity, gender, and religion often appear to be part of the product that is being sold in the marketplace. The courts usually do not allow employers to discriminate merely to satisfy the discriminatory preferences of their customers or their other employees, but both the courts and academic commentators have considerable difficulty with the kinds of examples I have just given.Footnote ²⁵ Much of the value of goods in the market is bound up with aesthetics, notions of authenticity, aspects of the purchaser's identity, and so on. Authentic Shaker-made furniture or Native-American-handcrafted jewelry are different products from otherwise indistinguishable furniture and jewelry. Even more clearly, an old Chinese man on the cover of Ebony would change the magazine's message and appeal. And few go to Hooters for their hamburgers.

Indeed, even though, as I mentioned, the courts refuse to allow employers to discriminate against Xs merely because their other employees discriminate against Xs and will have their morale affected if Xs are hired, it is probably the case that the demographic make-up of the staff of a business has some effect on the staff's productivity. Indeed, one often hears the productive virtues of “diverse” staffs extolled. If demographic diversity might in some cases improve productivity, by the same token in other cases it might impede productivity.

If antidiscrimination laws threaten the loss of real value beyond that of a mere “taste for discrimination,” one must ask what their ultimate goal is. They coexist uneasily with a marketplace that is otherwise governed only by consumer sovereignty on the demand side and the profit motive on the supply side. The foregoing examples illustrate this uneasy coexistence.

At this time, government principally intrudes on freedom of association and non-association in the marketplace in only two ways: by regulating employment practices through its antidiscrimination laws, and by barring discrimination on the same forbidden grounds by places of public accommodation in serving customers. Nonetheless, there are many choices within the marketplace where those antidiscrimination laws do not (now) apply, but could apply. For instance, why, if employers are prohibited from discriminating against potential employees on certain grounds, should employees not be similarly prohibited from discriminating against potential employers? Or, if restaurants and other places of “public accommodation” may not discriminate (on certain grounds) against potential customers, why should customers be allowed to discriminate against restaurants and other such businesses on those same forbidden grounds? (If I refuse to patronize restaurants owned by Moldovans, I am acting within my legal rights as they currently exist; I am not doing so if the restaurant I own refuses to serve Moldovans.) Likewise, a business that cannot discriminate in those ways in its employment practices may presumably do so when it comes to choosing its suppliers and other contractors. But given the quite slippery (if not illusory) distinctions between employing someone, on the one hand, and purchasing services from him, on the other hand, or between purchasing services and purchasing goods, the writ of freedom of marketplace association as a matter of principle would seem to cover either all such instances of discrimination or none at all.

(There are some other instances of governmental interference with freedom of marketplace association that are primarily designed to keep the marketplace competitive and free of fraud. Antitrust laws prohibit a narrow range of associations that have as their object price-fixing or monopolization.Footnote ²⁶ The CEOs of corporations A and B may associate on any number of bases, but they may not agree not to compete with one another. And conflict-of-interest laws prohibit another narrow slice of associations as prophylaxes against fraud or deceptive practices.Footnote ²⁷ All such laws do indeed restrict freedom of association in the marketplace in order to keep the marketplace functioning competitively and nondeceptively. And then there are, of course, laws that interfere with freedom of marketplace association precisely to counter marketplace competition, such as laws governing unions and their rights vis-à-vis employers and employees.)Footnote ²⁸

G. The nonexistent boundaries between these types of association

Having divided associations into various classifications, I now want to emphasize how arbitrary and misleading that division is. A great many, if not most, associations fall into more than one of the classifications I have mentioned. Social clubs, for example, are venues for close friendships, recreational activities, and commercial intercourse. The same is true of creedal groups, such as religions. Attempting to find the essential nature of any particular association in the terms that I have used to describe associations generally is a bootless quest, a misguided Platonism regarding human associations.

Because much of my attention in the remaining pages will be devoted to marketplace associations, my emphasis here will be on how permeable the boundary is between marketplace enterprises and creedal, social, recreational, and intimate forms of associations. The marketplace is not a sphere detached from these other human pursuits. The New York Times and Time Warner are businesses that compete in the marketplace, but they are also creedal. They sell a point of view in their editorials, their selections of stories to report, and in myriad less obvious ways. (I leave aside the question of who—which human beings—the “they” actually refers to in these cases.)

Universities, private academies, and other such institutions are both businesses and creedal (and recreational, social, and frequently somewhat intimate) associations. Recreational and social clubs compete in the marketplace for members; and their members form friendships, ideological commitments, and commercial relationships within their confines.

When parents choose a babysitter for their adolescent daughter and, in so doing, prefer a single, heterosexual female for the job over a heterosexual young male and a homosexual young female, are they to be regarded as associating commercially or as associating intimately? Likewise, if they are choosing someone to take their daughter camping? Likewise, if they choose an all-female private high school for her to attend? To ask these questions is to reveal how artificial the boundary is between commerce and the rest of human affairs. And if the parents are not acting commercially, then is an enterprise set up to provide such parents same-sex heterosexual babysitters or scoutmasters, or sex-specific schools, acting any more “commercially” than are the parents?

The same point can be made about many of the examples I gave earlier. If, to take one of those examples, Ebony magazine wants to target African-American women—and thus prefers to hire only such women to appear on its cover—there is surely a creedal and social aspect to its choice and the choice of its audience. The same may be true of the customers who want genuine Shaker-made furniture or Native-American handcrafted jewelry. And, as I pointed out earlier, neither employers nor their employees leave their creeds behind when they enter the office.

I cannot emphasize enough how impossible it is to divide any particular domain of associations from the other associative domains.Footnote ²⁹ And from this point follows another: Any governmental interference with freedom of any form of association, including commercial association, will perforce result in interference with the other forms of association.

A. Social engineering

On their face, antidiscrimination laws appear inconsistent with classical liberalism. However, there is a plausible case that they are not—or, at least, that they are not when applied to: (1) a relatively large, public corporation; (2) that is solely engaged in commerce; (3) that is producing a good or service with no ascriptive or ideological content (say, rivets); and (4) the law prohibits intentional discrimination on a few ascriptive grounds and does not extend to disparate impact discrimination (discrimination based on criteria that correlate more highly with some ascriptive groups than others), trait-based discrimination (discrimination based on criteria typically identified with a particular ascriptive group), (some) hostile-environment discrimination, and perhaps statistical (“rational”) discrimination. So limited, antidiscrimination laws might be justifiable attempts to overcome collective action problems that existed in various parts of the country during the last century and that prevented employers, employees, and consumers from hiring and dealing with African-Americans, women, and sometimes homosexuals. In many areas of the South, for example, even after the demise of the Jim Crow laws, there was tremendous social pressure not to cross the “color line” in commerce, pressure often backed by fear of violent retaliation and undependable law enforcement protection. To a lesser extent, this may also have been true of hiring homosexuals. And there was surely social pressure, if not the threat of violence, deterring the hiring of women. Antidiscrimination laws may have been welcomed as the means for overcoming this collective action problem.

There are other similar “social engineering” rationales that can be offered in support of limited antidiscrimination laws and their interference with freedom of association. One rationale takes us back to the discussion of cosmopolitan liberalism and the perhaps paradoxical fact that liberalism in its classical form is a big tent housing “illiberal” choices and the illiberal associations that are their outgrowths. It may be argued that a liberal polity cannot exist if certain liberal attitudes are not cultivated and sustained, and that racism, sexism, and similar prejudices must be extirpated for the liberal project to succeed. Freedom of association must be limited to secure a regime in which freedom of association can flourish.

There is, of course, some plausibility to this argument, particularly where African-Americans in the United States in the mid-twentieth century were concerned. There may also be some plausibility to it with respect to women, homosexuals, and members of certain religious groups. But notice that there is an ever-present danger with such arguments, a danger that in the name of saving liberalism, we shall have to destroy it.

Antidiscrimination laws might also be supported on welfare-maximizing grounds, apart from the collective-action market imperfection mentioned previously. Normally, in a free market economy, if we put aside market imperfections, including information deficits, wealth is maximized by consumer/producer sovereignty. Redistribution of wealth, to the extent it is justified, is better accomplished by direct transfers through taxation/spending than by interfering with market choices. Moreover, the standard concern of liberal redistributionists is the income of individuals, not the income of ascriptive groups.Footnote ³⁴

However, antidiscrimination laws might be viewed as aimed at affecting social attitudes regarding race, religion, and the like, with the ultimate goal of improving the welfare of certain groups and, perhaps as a consequence, increasing the welfare of most individuals. In the marketplace, assuming a free, competitive economy, if there were no change of racial attitudes among employers, employees, and customers, antidiscrimination laws would merely transfer wealth from some individuals to others and would cause a net decline in welfare. The hope—and the hope has been realized almost completely—was that the existing racist, sexist, and similar attitudes would be extinguished and that the total welfare would ultimately be increased—because the allocation of resources would be improved by antidiscrimination laws once those old attitudes disappeared.

Notice that in terms of overall social welfare, as opposed to a reallocation of welfare among groups, antidiscrimination laws would have been a gamble. They would have been a bet that racial and like attitudes could be altered by forced associations of certain kinds, and that, so altered, the individuals whose attitudes they were would attain more satisfaction and fulfillment in the newer forms of association than they would have if the old forms had persisted. With the racial premium removed, for example, not only could blacks find employment commensurate with their skills, and opportunities to consume commensurate with their wealth, but employers would receive more productivity from employees and provide more satisfaction to consumers. And the same is true regarding gender, religion, and the other “forbidden grounds.”

As I said, the bet was won. But some are dissatisfied with the results, as dramatic as they have been. They are dismayed that blacks—at least those not from immigrant families—have made less economic progress as a group than they would have hoped. Or they are dissatisfied that women, though now nearly equal in numbers to men in medicine and law, are still disproportionately underrepresented in the sciences, engineering, and top corporate management, and overrepresented in unpaid household and childcare labor.

Those dissatisfied with the results of current antidiscrimination laws sometimes argue for compelled proportionate representation, usually maintaining that the tastes and attitudes that result in the current underrepresentation of certain groups in certain occupations can give way with no net loss of aggregate welfare. (Sometimes the mantra that these attitudes are “socially constructed” is a shorthand way of making this claim.)

If government were to bet on such claims, I strongly doubt that the bet would pay off. The kinds of blatantly racist and sexist attitudes that the current antidiscrimination laws eroded were pretty thin and superficial, at least for most who held them. And the amount of welfare that was sacrificed to honor those attitudes was quite high. The attitudes and tastes that result in “underrepresentation” are quite different in all relevant respects. The preference for, say, good math skills is thin and superficial only for those who do not care whether the walls of their house will meet at right angles, whether their monthly bank balance is correct, or whether the plane they are flying in is safe. Moreover, the social groups whose identity and salience as groups were themselves the products of racist and sexist attitudes have lost their identity and salience with the eradication of those attitudes and the practices that embodied them, the identity-politics hustlers to the contrary notwithstanding.

The fact that the preferences that result in “underrepresentation” are now quite different from those that animated “no blacks or women need apply” forms of discrimination explains the difficulties the courts have had with “trait-based” preferences (those based on a trait usually identified with a particular ascriptive group), “sex-plus” preferences (those directed at a subgroup of one sex), and other forms of preferences that have a local, though not necessarily a global, adverse disproportionate impact on certain groups. Requiring women to wear dresses, men to shave and to wear non-baggy pants fastened at the waist rather than the groin, and similar requirements are sometimes assimilated by advocates and more than a few courts to the “no Xs need apply” forms of discrimination. But although the genesis of these preferences might pass through certain race- or sex-linked ideas of propriety, they will, I suspect, prove hardier and more difficult to dislodge than the blatantly sexist and racist ones. The amount of re-engineering of our preferences necessary to rid them of all connections to ideas with a racial, ethnic, national, religious, or sexual component would be staggering if not impossible even to conceive. Our preferences for authentic Native-American-made pottery or Shaker furniture are just the tips of the iceberg of preferences that undergird our intimate, ideological, recreational, and other associations in addition to our marketplace ones.

I say that the current antidiscrimination laws have proven successful as social engineering. Nonetheless, it should be pointed out that although these laws are often thought of as harm-preventing intrusions on freedom of association—because they are thought of as shielding racial and religious minorities and women from harm—they are surely violative of libertarian principles and also, in their wager on overall welfare improvement, more than a little paternalistic. Although the employer, employees, and customers with racially or sexually bigoted views were, in discriminating, acting within their libertarian rights (assuming their wealth was justly acquired), the antidiscrimination laws interfered with their associating with whom they chose on the ground of overall social good, not a ground that a libertarian would accept as a justification. Moreover, the overall social good in no small part depended for its achievement on the paternalistic assumption that a forced change in attitudes would enhance the welfare of those forced to change.

B. Paternalism: Compelling associations that are beneficial to flourishing and forbidding those that are self-destructive

A second principal reason government might have for interfering with freedom of association is its concern that its citizens choose the good and eschew the bad. Most of the laws that currently interfere with freedom of association are not paternalistically justified; but there are hints of paternalism in some of the rationales for extending (or “interpreting”) antidiscrimination laws in various directions. For example, Kimberly Yuracko argues that gender discrimination laws can best be rationalized and justified by reading into them a “perfectionist” rationale, that is, a concern for women's flourishing.Footnote ³⁵ Certain jobs are degrading to women, and employers should not be allowed to offer them to women (and, necessarily, women should not be allowed to accept them on the terms offered).

Yuracko's basic perfectionist approach can easily be extended to racial and other forms of discrimination. Moreover, it is not confined to employment. If some jobs are bad for us—they represent shallow or base values and thus take us off the path to our true flourishing—then so are some noncommercial activities, organizations, creeds, and intimates.

Now I am not a skeptic about whether there are better and worse paths to flourishing, or whether some jobs, activities, organizations, creeds, and friendships are shallow, base, or degrading. What I am skeptical about is the competency of government to make judgments about the paths to flourishing. It has not been my experience that government is a repository for that kind of wisdom. Nor do I find the processes of governmental decision-making in the modern mass democracy well-designed to produce that wisdom. But even if government were generally reliable with respect to identifying the paths to flourishing, there would remain the question of government's entitlement to act paternalistically.

C. Legal moralism

There is another rationale for antidiscrimination laws and their infringements of freedom of association apart from how they affect consumer satisfaction in the economy, the welfare of certain groups, or the flourishing of those who discriminate. That rationale is simply that certain forms of discrimination are just morally wrong, and government is entitled to forbid them for that reason alone.

Now the question of when and why private discrimination is morally wrong is a difficult one, as the literature on the subject reveals.Footnote ³⁶ But putting that question aside, there is the more basic issue of whether governments may legitimately prohibit conduct that is merely immoral. Joel Feinberg presented the most sustained case against legal moralism.Footnote ³⁷ Other liberals have taken the opposite point of view.Footnote ³⁸ Even they, however, have suggested that there is a strong presumption in favor of liberty and against prohibitions based on mere immorality.Footnote ³⁹