The last two decades have seen a remarkable revival in academic political philosophy, particularly in the English-speaking world. A subject which was widely pronounced dead in the 1950’s has recently produced thousands of articles and numerous books of real importance. One indicator of the scale of this revival is the length of a recently published annotated bibliography about just one of those books (John Rawls’s A Theory of Justice). It runs to nearly 700 pages, and all the books and articles listed concern a book that was only published in 1971.

This resurgence has been fueled by a single-minded quest for an adequate theory of the just society—a picture of the ideal liberal society which would place all the untidy fragments of the liberal tradition in their appropriate priority relations. Such a theory would give us a clear picture of how we ought to value liberty compared to equal opportunity, compared to individual property rights and other familiar principles. Rather than merely weighing conflicting principles against each other, such an ideal solution would give us a single coherent vision of the just society—one which we should attempt to achieve, or at least approach, in our actual public policies over the long term. I believe that this quest for a single, adequate solution to the theory of justice asks too much. Any reasonable construction of the values central to liberalism will not give us a single Utopian vision in clear focus. Rather, it will give us patterns of stark conflict requiring hard choices among fundamental values. It will give us ideals without an ideal—conflicting partial visions, each of which would take public policy in an entirely different direction if given further emphasis.

I would like to develop this point by pursuing in detail a key part of the theory of justice—the problem of equal opportunity. In this country we have long been more concerned with giving people a fair chance to be unequal than with making them equal, as a matter of final result. If there is a specifically American contribution to the development of Western liberalism, it derives from our preoccupation with equal opportunity.

However, systematic solutions to the problem of equal opportunity lead to clearly illiberal conclusions. We have been blinded to the difficulties in a rigorous theory of the just society by our neglect of one crucial factor—the family and the issues of liberty surrounding it. Once we take full account of the family, our common assumptions about equal opportunity do not add up to a coherent scenario for the just society. Rather, they yield only a pattern of contradictory results, a pattern I call a trilemma—roughly, a dilemma with three corners.

This pattern is only disturbing if we expect too much. Despite our lack of a rigorous theory of equal opportunity, we have ample grounds for important substantive conclusions with immediate policy implications. The lack of a rigorous ideal solution does not disarm us from evaluating real problems in the real world. The trilemma of equal opportunity is a forced choice among three principles. The first principle, merit, requires procedural fairness in the evaluation of qualifications for positions and, then, selection of the most qualified. Whether from Civil Service exams, SAT tests, or the NFL draft, the basic idea of meritocratic assignment is widely familiar in this society.

However, meritocratic assignment, by itself, is not enough for a viable theory of equal opportunity. There must be some attempt to deal with inequalities in the conditions under which talents and other qualifications develop. The argument can be dramatized with an example which I will borrow from the English philosopher Bernard Williams. In order to acquire some perspective on the educational and job qualifications built into our daily lives, let us imagine a different kind of society—one dominated by a warrior class. The top positions are all occupied by skillful warriors who have perpetuated their positions from one generation to the next. Suppose, however, that advocates of equal opportunity achieve a reform. They decide that admission to the warrior class should, from that point on, be determined by a suitable competition. We might imagine a kind of fighting Olympics designed to select the best warriors.

As it happens, however, the present warriors’ children overwhelmingly dominate the competition, so that the reform makes no difference to the outcome. To simplify matters, children from the present warrior class all turn out to be extremely well nourished, while their competitors are so undernourished that they fail in the competition for lack of strength. The competition yields the spectacle of 300-lb. sumo-style wrestlers vanquishing 90-lb. weaklings. Only a cynic could contend that real “equal opportunity” had been achieved. While meritocratic assignment may have been achieved in the sense that the sumo wrestlers actually are better warriors, they have been permitted to develop into better warriors under such unequal conditions that the result cannot be presented as fulfilling equal opportunity.

A second principle of equal opportunity is needed to supplement the first. If we can reliably predict that the children of the present warrior class will become the dominant class, that is objectionable, even if they do so by winning a meritocratic competition. Equality of life chances requires that the prospects of children for eventual positions in the society should not vary systematically with their arbitrary native characteristics (their family background, race, class, sex, ethnic group, etc.). To the extent that we can predict where they will end up merely by knowing such facts about them (such as that they are children of the present warriors) an identifiable kind of injustice is being perpetuated.

The two principles just mentioned are principles of equality intended to ration fairly the chance to be unequal. But these two principles conflict with a third, one which expresses a kind of liberty central to the liberal tradition (and with broader appeal as well). It is, roughly, the freedom of families, acting consensually, to benefit their children, or what might be called family autonomy. Family autonomy does not apply to cases where the consensus within the family has broken down (such as issues of child placement) or to the freedom of families to impose disadvantages or harms on children. Rather, it is the freedom to benefit one’s children without coercive interference from the outside (and where, we might hypothesize, there is no real dispute about the benefits inside or outside the family). By family autonomy, then, we mean the least controversial core of family relations. The trilemma shows how this element of liberty (a part of liberty which affects most of our lives more directly than do our other liberties) conflicts with even modest constructions of equality (constructions concerned only with opportunities and not with results), even under ideal conditions. The resulting pattern lends plausibility to the notion that liberalism is a kind of continuing dialogue about the conflicts between liberty and equality.

It would not be surprising if our central principles clashed under difficult circumstances—a history of injustice, noncompliance with some of the principles being prescribed, or extreme scarcity where no plausible version of justice was possible. However, the trilemma argument does not require any of these pessimistic assumptions. Let us assume no legacy of injustice from the past; let us assume complete compliance with the principles being proposed; and let us assume only moderate scarcity as in a prosperous industrial society such as our own (rather than the extreme scarcity of, say, the Third World). To be realistic, however, we need one more assumption, namely, that there are inequalities, both social and economic in the adult generation. The point of equal opportunity is to ration chances to be unequal. If there were strict equality of outcomes, then there would be no point in worrying about equal opportunity. Let us assume that there are inequalities of result in the adult generation (background inequalities) of the kind which apply to every modern developed society, capitalist or socialist.

Given background conditions of inequality, implementing any two of these principles—merit, equal life chances, family autonomy—can be expected to preclude the third. It is as if we had a three-cornered stool but only two legs available to hold it up. It doesn’t matter which two we pick, but lack of the third is enough to undermine the whole structure (provided that we expect or require a stable structure to begin with). For example, implementing the first and third principles undermines the second. The autonomy of the family protects the process whereby advantaged families differentially contribute to the development of their children. Given background conditions of inequality, children from the higher strata will have been systematically subjected to developmental opportunities that can reliably be expected to advantage them in the process of meritocratic competition. Under these conditions, the principle of merit—applied to talents as they have developed under such unequal conditions—becomes a mechanism for generating unequal life chances.

Suppose one were to keep the autonomy of the family in place but attempt to equalize life chances? Fulfilling the second and third principles would require sacrifice of the first. Given background conditions of inequality, the differential developmental influences just mentioned will produce disproportionate talents and other qualifications among children in the higher strata. If they must be assigned to positions so as to equalize life chances, then they must be assigned regardless of these differential claims. Some process of “reverse discrimination” in favor of those from disadvantaged backgrounds would have to be applied systematically throughout the society if life chances were to be equalized (while also maintaining family autonomy).

Suppose one were to attempt to equalize life chances while maintaining the system of meritocratic assignment? Given background conditions of inequality, it is the autonomy of families that protects the process by which advantaged families differentially influence the development of talents and other qualifications in their children. Only if this process were interfered with, in a systematic manner, could both the principles of merit and of equal life chances be achieved. Perhaps a massive system of collectivized child-rearing could be instituted. Or perhaps a compulsory schooling system could be devised so as to even out home-inspired developmental advantages and prevent families from making any differential investments in human capital in their children, either through formal or informal processes. In any case, achieving both merit and equal life chances would require a systematic sacrifice in family autonomy.

Implementing any two of these principles precludes the third. While inevitable conflicts might be tolerated by systematic theorists in the nonideal world, these conflicts arise under ideal conditions. Given only moderate scarcity and strict compliance with the principles chosen, given that there is no aftermath of injustice from the immediate past, we are applying these principles in our thought experiment to the best conditions that could realistically be imagined for a modern, developed society.

Of course, liberalism has long been regarded as an amalgam of liberty and equality. And liberals and libertarians have long been fearful of the sacrifices in liberty that would be required to achieve equality of result. Equality of opportunity, by contrast, has been regarded as a weakly reformist, tame principle which avoids such disturbing conflicts. However, even under the best conditions, it raises stark conflicts within the one area of liberty which touches most of our lives most directly. Once we take account of the family, equal opportunity is an extraordinarily radical principle and achieving it would require sacrifices in liberty which most of us would regard as grossly illiberal.

These principles are not demanding by themselves; they are demanding in combination. Each of the trilemma scenarios fully implementing two of the principles leads to drastic sacrifice of the third. One reasonable, but unsystematic response, is to trade off small increments of each without full realization of any.

But this is to live without a systematic solution. The aspiration fueling the renewal of liberal theory has been that some single solution in clear focus can be defined for ideal conditions and then policy can be organized so as to approach this vision gradually. But if trade-offs are inevitable, even for ideal theory, then we have ideals without an ideal, conflicting principles without a single unifying vision.

My position is, first, that equal opportunity offers a prime case for this result: second, that despite the lack of a systematic solution for ideal theory, there are significant policy prescriptions which result without any necessity for solving the priority relations among these principles. We can live with moral conflict and still offer significant prescriptions.

Having sketched the first point, let us turn to the second, the issue of policy implications. There are two kinds of policy implications we can evaluate without having to solve the problem of priority relations among these three principles for ideal theory—without, in other words, employing the model of a single unified and coherent ideal which we should, as best we can, approach through partial realization. The first kind of policy implication may be seen wherever we can achieve a major improvement in the realization of one of these values without a major loss in any of the others (or in any other new values which the proposal impinges upon). The second kind of policy implication applies to any proposal that would impose a major loss in one of these values without any comparable gain in any of the other values (or in any new values which the proposal impinges upon). The first kind of policy implication is exemplified by all those things we could do, in the second term of the Reagan Administration, to improve equality of life chances, family autonomy, or meritocratic assignment.

We are far from the possibility frontier in achieving any of these values, and, in some cases, we have moved further away rather than closer in recent years. We blithely tolerate the perpetuation of an urban underclass; a whole generation of urban youth is growing up with blighted life chances and with few opportunities to make it into the mainstream economy, and with few policy initiatives now focused on their problem. Family autonomy is protected for middleclass families, but poor families have far greater difficulty in forming and maintaining themselves intact. By neglecting job prospects among the poor, this Administration has also affected the incentives for family formation as well as the ability of poor families to provide essential prerequisites for child development and socialization. We are also far from achieving meritocratic assignment. In my view, discrimination persists against blacks, Hispanics, women, and other minorities. There is no justification for tolerating job discrimination on the basis of arbitrary factors, irrelevant to the roles in question. In other words, despite Mr. Reagan’s talk of protecting “the family” and of creating an “opportunity society,” his policies have promoted middle-class families and middle-class opportunities at the expense of the disadvantaged.

The second kind of policy implication is exemplified by the major quick-fix for the first set of problems—preferential treatment based merely on arbitrary native characteristics. When it is applied in competitive meritocratic contexts, this policy yields a major sacrifice in one of our three values, meritocratic assignment, without a significant gain in any of the others. The difficulty is that preferential treatment, when it is based merely on arbitrary native characteristics, is mistargeted as a policy which could have any effect on equality of life chances. It is mistargeted because, in competitive meritocratic contexts (e.g., admissions to graduate and professional schools), there are strong institutional pressures to accept the most qualified applicants with the specified arbitrary native characteristics. Just as family background provides disproportionate opportunities to develop qualifications among advantaged white children, it does so among relatively advantaged minority children. This policy only serves to widen the gap between the urban underclass and the black middle class—despite the fact that it is typically justified as special consideration for those who are from disadvantaged backgrounds.

My objection does not apply to policies which apply preferential treatment to those who are, actually, from disadvantaged backgrounds. In that case, meritocratic assignment is sacrificed for a gain in equal life chances. Rather, my objection applies to programs which are applied merely on the basis of arbitrary native characteristics, so as to reward the most qualified members of the group (who will, as a statistical matter, tend to come from its more advantaged portions). Hence, the irony of the De Funis and Bakke cases. De Funis, a Sephardic Jew from a relatively poor background, was not admitted to the University of Washington Law School, while most of the minority students admitted on the basis of preferential treatment were, apparently, from more advantaged backgrounds (or, at least, were the children of black professionals). On the other hand, the program at the University of California at Davis which the Court struck down in the Bakke case was unusual for having procedures in place to direct special consideration to those who actually came from economically disadvantaged backgrounds. Theoretically, the program which the Court struck down in Bakke was defensible within our framework, while the program it avoided a decision on in De Funis (providing preferential treatment for race as such) would not be.

We should mention one persistent counterargument to this conclusion about preferential treatment. Preferential treatment based merely on race (or on other arbitrary native characteristics) is sometimes supported, not as a remedy for developmental disadvantages in the present, but as a form of compensation for injustice in the past. However, the mistargeting objection has force here, as well, but with additional complications.

First, the list of groups which were, historically, victims of discrimination is much broader than the groups now demanding compensation. Consistent pursuit of this argument would produce a host of other ethnic claims—Irish, Polish, and Italian, Catholic as well as Jewish, in addition to the more familiar arguments made on behalf of Hispanics, Native Americans, and Orientals. This proliferation is not fanciful. The Anti-Defamation League discovered one American law school which had not less than 16 racial and ethnic categories for admissions classification.

Second, the very notion of compensation raises conceptual challenges—unacknowledged by its proponents—when applied to this kind of problem. An individual X is supposed to be compensated by returning him to the level of well-being he would have reached, had some identifiable injustice in the past (committed against his forebears) not occurred. In tracing back through the generations, however, it soon becomes clear that X would usually not now exist, were it not for the historical injustice. If we try to imagine the world which would have existed, had the historical injustices not occurred, we cannot return X to the level he would have reached, because he would not have reached any level at all.

For example, let us take the case of Kunta Kinte. If Kunta Kinte, Alex Haley’s ancestor in Roots, had not been brutally kidnapped and sold as a slave, there is no likelihood that the author of Roots would have come to exist in the 20th century. The mating and reproduction of each generation, in turn, depends on a host of contingencies, if the chain were to have been broken at any point, by a parent, grandparent, great-grandparent, etc., we would get a different result in this generation. If it were not for the initial injustice, Kunta Kinte’s descendants might well have been native Africans, perhaps residents today of Juffure (Kunta Kinte’s village in West Africa).

Hence, we cannot employ the straightforward notion of compensation (returning people to the level they would have reached had the injustice not occurred) when the injustice spans several generations. Perhaps some compelling version of the argument might be created which confronts this difficulty. Rather than deny this possibility I wish merely to claim that such an argument, if it were developed, should, at the least, accept my objection to mistargeting.

Compensation is not compatible with the mistargeting which results from preferential treatment applied merely to racial categories in competitive meritocratic contexts. Compensation cannot plausibly mean benefiting some blacks (already well-off) for earlier injustices to other blacks—particularly when those who do not, by and large, benefit from the compensatory argument (the urban underclass) are experiencing extremely disadvantaged conditions. To take a provocative analogy, would it not have been outrageous if the German government had paid “compensation” to well-off American Jews after World War II, ignoring the orphans and other direct victims of the Holocaust? It would not have been real compensation to benefit Jews indiscriminately or, even worse, to benefit disproportionately those who were untouched by the injustices at issue. For this reason, I conclude that the compensation argument does not alter the general conclusion about preferential treatment reached earlier. When it is directed at those who are, themselves, from disadvantaged backgrounds, it is admissible within our framework (for then the gain in equal life chances may balance the loss in strictly meritocratic assignment). But when, in competitive meritocratic contexts, it is applied merely on the basis of arbitrary native characteristics, then we have grounds for objecting to it.

Once we take account of the family, equal opportunity demonstrates the plausibility of a limited liberalism—one which offers ideals without an ideal, one which offers principles in conflict without a single unified vision. However, these moral conflicts do not disenfranchise us from offering significant prescriptions. If we simply embrace substantial gains without a loss and reject substantial losses without a gain, we can already see important results for the contemporary scene. While these results do not add up to a systematic theory, they should assist us in rethinking important policies.

Suggested Reading:

Bruce Ackerman, Social justice in the Liberal State (New Haven and London: Yale University Press, 1980)

James S. Fishkin, Justice, Equal Opportunity and the Family (New Haven and London: Yale University Press, 1983)

John Rawls, A Theory of Justice (Cambridge, MA: Harvard University Press, 1971)

Bernard Williams, “The Idea of Equality” in Peter Laslett and W.G. Runciman (eds.) Philosophy, Politics and Society, Second Series (Oxford: Basil Blaekwell, 1962)