by Gerald Gaus
I am not so naive as to suppose that this is not an attractive goal for many philosophers, but it does make rather puzzling the paraphernalia of the practice of justice: demanding, complaining, blaming, claiming, agitating, “the justice system,” punishments, editorials, social movements, consciousness-raising, and revolutions. Estlund’s theory, while allowing that “prime justice” may be hopeless, admits the category of “concessive” justice—the justice that applies to Alf when he knows that he will not conform to prime or ideal justice. A theory that is hopeless all the way down is truly a truncated theory of justice.46 Such a theory fails to make sense of the way justice enters into our form of life.
Again, this is not to say that ideal justice itself must be implementable; even in a nondreaming account, ideal justice still can be a hopeless (but not a useless) vision that orients, inspires, and guides us. Nor is it to say that for every well-formed judgment that one social state is more just than another, a theory of justice must yield a practical recommendation to move to the more just condition. Note the range of practical activities listed in the previous paragraph; claims and demands can be quite specific, but social movements and consciousness-raising guide action in much vaguer ways. The point is merely that, as I shall consider it here, ideal theory is not simply about theoretical evaluative judgments of social states. The orientation of judgments of justice cannot be the entire story; there must arrive a point at which a political philosophy concerned with justice yields recommendations about moving to more just social states. An ideal might be simply aspirational, but not the entire field of justice.
When a political philosophy focused on justice moves from judgments of justice to recommendations, it must move away from the sphere of pure moral philosophy, understood simply as the justification of normative judgments.47 At this point, political philosophy must draw on social and economic theory, political science, and psychology in seeking to formulate policies, recommendations, and demands.48 Depending on the context, this will require employing different social science toolkits and predictive models. A political theorist advocating a constitutional structure, one advocating a tax policy, and one advocating a change in particular laws (say, about school segregation), will obviously draw on different data, toolkits, and models, but it will be rare indeed that no significant social science will be drawn on. Even at the judicial stage, where it may seem that “recommendations” are simply derivations from legal requirements, courts regularly invoke social theory and science. Thus, for example, in the landmark Brown v. Board of Education of Topeka, which was without doubt about basic claims of justice, the Supreme Court noted the psychological harms of school segregation.49 Because the recommendation task of a theory of justice or political philosophy requires a set of tools, data, and models, analyzing this aspect of a theory of justice requires understanding the ways that political philosophy draws on social science.
2 SOCIAL REALIZATIONS AND THE IDEAL
2.1 Perfect Principle Conformity and Ideal Societies
I have thus far dealt with “the ideal” as an unspecified variable for optimal or perfect justice. Although I wish the analysis to remain reasonably general—it certainly should not turn on a precise understanding of the ideal—it is important to identify the constraints on characterizing an ideal as I will examine it. My aim is a sufficiently broad characterization such that it captures the essentials of an interestingly large group of theories while, at the same time, lending itself to a coherent and fruitful analysis.
G. A. Cohen’s account of justice is often depicted as a sort of “ideal theory,” indeed often as an instance of the most extreme type.50 Cohen—and in this respect he shares much with Estlund51—is adamant that a claim that some fundamental principle of justice J is the correct principle cannot be defeated by a demonstration that J is outside the bounds of what humans plausibly can be expected to do. That is, on his view J’s claim to be a fundamental principle cannot be defeated by the finding that, given psychology and other social sciences, it is highly improbable that people will conform to it. The fundamental truths of justice, it is said, do not depend on such practicality considerations, nor do they derive from any set of facts about a group’s current beliefs and convictions.52 This suggests a characterization of the ideal, the
Principle-Defined Ideal: Supposing that we have identified principles of justice J that do not depend on feasibility constraints or sociological facts (such as current beliefs about justice), for any social state, S, it is an ideal social state if and only if in S everyone affirms J and perfectly conforms to J. All other social states are nonideal.
Cohen’s text actually does not support this interpretation of the ideal. “Ideal theory,” he says, is “what transpires in the ideal society, and nonideal theory … applies to settings in which, among other things, citizens do not affirm and act upon the correct principles of justice.”53 The implication of the italicized phrase is that nonideal states are not merely defined by lack of endorsement and conformity to the relevant principles. In contrasting the ideal liberal and socialist societies, Cohen tells us that “in the ideal socialist society, equal respect and concern are not projected out of society and restricted to the ambit of an alien superstructural power, the state. If the right principles are, as Marx thought, the ones that are right for everyday, material life, and if they are practices in everyday life, as the socialist ideal utopianly envisages they will be, then the state can wither away.”54
We see here that the ideal is characterized by not only conformity to the relevant principles, but a specification of the aspects of the social world to which the principles apply (“the practices”); the socialist and liberal dispute about the ideal does not involve a dispute simply about principles, but about the parts of the world that are to be regulated by the principles. Whereas, Cohen suggests, the liberal sees ideal justice as regulating the state (and, we might add, more generally, the basic structure of social cooperation), the socialist holds that the principles apply to various practices of everyday life. This, though, means that the identification of the ideal society must involve specification of various types of institutions, practices, or spheres of life;55 if there is no state or no basic structure, for example, Cohen indicates that there simply cannot be liberal justice. But then a liberal ideal must provide an account of what constitutes state structures, or what institutions are part of the basic structure. In Rawls’s liberal ideal, for example, the family sits (somewhat uncomfortably) in the basic structure.56 Rawls’s liberal ideal thus differs from that of a liberal who would exclude the family, yet both may endorse Rawls’s two principles of justice.57 And conceptually, the liberal and socialist might agree on the same principles, yet their visions of the “ideal society” could still radically differ with regard to the parts of social life to which they apply.
Although, then, we might start off thinking of “ideal justice” as simply universal endorsement of, and action in accordance with, the correct, true, or best principles of justice, to make sense of ideal justice we need to suppose a set of institutions, practices, or spheres, and the agents that act in them—and all this requires some account of how institutions or agents (could possibly, are apt to, can without excessive strain) operate, and how various spheres are demarcated.58 As Henry Sidgwick observed, “even an extreme Intuitionist would have to admit that the details of Justice and other duties will vary with social institutions.”59 Even the idea of agency itself needs specification. For example, it needs to be determined whether groups are agents subject to principles of justice and, if so, what are group agents, and for what can they possibly be held accountable.60 At the very minimum, our vision of Paradise must specify features of the social world to which the principles apply, and some suppositions about the workings of the agents regulated.61
2.2 Justice and Its Social Realization
I have been stressing that even theories that, prima facie, identify the ideal in terms of perfect conformity to principles also include reference to
social institutions and the way they function. More fundamentally, though, any plausible notion of the ideally just society must take seriously the way such a society would actually work out. Recall Sen’s charge that Rawls succumbs to “institutional fundamentalism.” Although he focuses on institutionalism and rules, the general category Sen is attacking is rather broader, encompassing much of what is called “deontology,” whether that be focused on institutions, rules, or fidelity to principles.62 Sen rejects such “fundamentalism” because “justice is not just a matter of judging institutions and rules, but of judging societies themselves.”63 I believe that Sen is correct that there is something extraordinarily severe and unattractive about a conception of justice that is entirely blind to the social realization of conformity to its principles, as if the only thing relevant to justice is that a certain set of principles be obeyed, come what may. Sen is right that such a view would be implausible. What is wrong, however, is his charge that Rawls or other plausible versions of principle-based justice are blind in this way.64 It is certainly true that Rawls sought to include “an important element of pure proceduralism” in the design of institutions: to a significant extent he aims to discover a set of institutions whose outcomes, whatever they might be, would be just.65 But these institutions are designed in light of their known and predicted effects: the aim is a “fair, efficient, and productive system of social cooperation” that is “maintained over time.”66 Recall Rawls’s characterization of teleology and deontology: “Utilitarianism is a teleological theory whereas justice as fairness is not. By definition, then, the latter theory is deontological, one that does not specify the good independently from the right, or does not interpret the right as maximizing the good. (It should be noted that deontological theories are defined as nonteleological ones, not as views that characterize the rightness of institutions and acts independently from their consequences. All ethical doctrines worth our attention take consequences into account in judging rightness. One that did not would simply be irrational, crazy).”67
A sharp distinction between an ideal theory focused on principles and one focused on social structures is, ultimately, untenable. All principle-based theories “worth our attention” are sensitive to their social realizations.68 And this includes Kantian principled theories as well. We can better appreciate how this is so by looking at Rawls’s interpretation of Kant’s universal law formulation of the categorical imperative, which he models in terms of a four-step “CI procedure.” The first three steps of the CI procedure are fairly straightforward:
(1) I am to do X in circumstances C in order to bring about Y. (Here X is an action and Y a state of affairs).
The second step generalizes the maxim at the first to get:
(2) Everyone is to do X in circumstances C in order to bring about Y.
At a third step we are to transform the general precept at (2) into a law of nature to obtain:
(3) Everyone always does X in circumstances C in order to bring about Y (as if by a law of nature).69
In the fourth step we are to consider the “perturbed social world” that would result from the addition of this new law of nature; we seek to understand the new “equilibrium state” on which this perturbed social world would settle. We are then to ask ourselves whether, when we regard ourselves as a member of this new social world, we can “will this perturbed social world itself and affirm it should we belong to it.”70
Thus on Rawls’s analysis of Kant’s “deontology,” our endorsement of a moral imperative takes place against the background of a model of the social world and the effects of introducing a maxim. To characterize an ideally just condition is not simply to posit that certain principles are in fact followed; we must consider the social realization of conformity and consider whether we can maintain our judgment that the realized world would be ideally just. Consider a case that worried Rawls: we come to the conclusion that a particular principle is the best principle of justice, and we reflect on the social world where everyone conforms to it. Assume additionally, however, that when we consider the social realization of this perfect compliance individuals become deeply alienated from their sense of justice.71 Perhaps they curse their very sense of justice, and bemoan the day they discovered true justice, which sets them at odds with their deepest convictions about their place in the universe. Cynicism and alienation are rife. We might wonder: is this a just world, in which those devoted to justice so suffer for their devotion? If we think not, we have at least two options: either conclude that under these social realizations, perfect compliance with this principle is not ideally just, or posit a social world where the social realization of universal justice is not so dire and identify that as the ideal.72 Either option acknowledges that the social realization of universal compliance with our proposed principle is relevant to a claim that it regulates the ideal social world.
In an important series of papers, David Wiens has also cast doubt on whether a plausible ideal theory can be cast entirely in terms of a set of substantive principles, with comparative justice determined simply by comparative satisfaction of the correct principles.73 Wiens distinguishes evaluative criteria from directive principles. On his interpretation, for example, “Rawls’s principles are meant to characterize an ideal balance of certain basic values, that is, the balance of basic values realized by a fully just institutional scheme.”74 The original position is thus understood as a way in which a complex set of criteria—liberty, equality, reciprocity, fairness, and so on—can be employed to yield directive principles for a set of institutions. “These moral conditions modeled by the original position operationalize certain ‘commonly shared’ basic values and, thus, represent basic evaluative criteria (as I understand that concept).”75 Wiens argues that there are social worlds where the set of values that lie behind Rawls’s two principles of justice would not be best satisfied by recommending worlds whose institutions most closely satisfy the two principles. Consider, for example, social worlds a and b. In social world a the difference principle is best satisfied; in social world b it is not quite as well satisfied, but b’s institutions guarantee much more equality than in world a. Wiens posits the relations in figure 1-1.
Let us stipulate that there are no effects on the satisfaction of the first principle of justice or fair equality of opportunity: they are equally well met in a and b. Wiens’s insight here is that the values that the original position seeks to model would plausibly favor world b over a even if, ex hypothesi, the two principles are better met in a. The two principles of justice, we might say, only adequately model Rawls’s underlying evaluative standards in a (proper) subset of the social worlds to be evaluated. This reading is confirmed by Rawls, who argued that the “General Conception of Justice,” not his famous two principles, is the best articulation of the underlying values modeled in the original position in some empirical circumstances (essentially those of pressing scarcity).76 Thus Rawls explicitly acknowledged the inadequacy of evaluating the justice of all social worlds in terms of the two principles (which he called the “Special Conception” of justice). It must follow, then, that if he advances a coherent theory, which sometimes employs the two principles to evaluate social institutions and in other cases the General Conception, there must be deeper underlying evaluative standards than the Special Conception.
Figure 1-1. Distributive shares under two principles (modified from Wiens, “Against Ideal Guidance,” p. 443)
One might dispute the importance of the case of Rawls. One might grant that both his “four step CI-procedure” and his recognition that the two principles are inappropriate in some circumstances do indeed indicate that Rawls tests principles in terms of the acceptability of their social realizations given fundamental values or evaluative standards. But one might argue that Rawls is driven to this because of his conviction that justice is the preeminent virtue of social institutions: if, as for Rawls, the overall acceptability of institutions essentially turns on their justice, then the requirements of justice must c
onverge with our judgments of the overall acceptability of a social world and its institutions. Because Rawls holds justice to be such a central virtue of social life and institutions, a social world’s overall evaluation must closely track its justice. Others, however, insist that justice is simply one of many virtues:77 even from the perspective of political philosophy, they claim, our overall evaluation of a social world depends on many values, of which justice is but one. Think again of Wiens’s worlds a and b, and suppose we agree with him that b is morally superior to a, even though a better satisfies Rawls’s difference principle. But if justice is simply one of many virtues of social institutions, one might say that, while b is better, it is not “through-and-through just”78—it fails to meet the principles of justice as well as does a.
This is a conceivable view. However, to the extent that justice is one of many virtues of the morally relevant social world, “truths” about justice do not tell us a great deal about the acceptability of our social world. Just as one might say, “Yes, yes, this social world is pretty inefficient, but inefficiency isn’t that important” one could then say, “Yes, yes, this social world is pretty unjust, but justice isn’t that important.” Ideal justice is simply about everyone conforming to some principle come what may, but it may not be especially praiseworthy that such an ideal is achieved. In this case, while the ideal may still orient our thinking about justice, justice no longer orients our evaluation of social and political worlds. Justice is insulated from its social realizations by diminishing its significance.79 Of course between these radical extremes—of justice being preeminent and it being just one of many virtues of social worlds—justice may be very important, but not quite the first virtue of social institutions. But the more important justice is to our overall evaluation, the more important becomes its social realization, for the closer we expect our evaluations of justice to track our overall evaluations.