JUSTICE

Justice names not a thing, but a property of things. It makes sense therefore to focus the explication on the adjective "just"—or, better still, "unjust." Doing so facilitates clarification of how justice judgments are distinctive within the larger realm of moral judgments, and the even larger universe of evaluative judgments.

The application of ordinary empirical predicates, such as "tree" or "hard," is two-tiered: based on a definition and empirical facts. Any dispute about whether such a predicate applies thus reduces to linguistic and empirical differences. Such a dispute can be resolved by agreeing on a definition and settling the empirical disagreement.

Evaluative predicates, by contrast, have this special feature that their application is only conditioned, not determined, by their definition and the empirical facts. Thus, people can disagree about whether a painting is beautiful, even if they use this predicate in exactly the same sense and also agree about all empirical features of the painting. In such cases it may be said that they have different conceptions of beauty.

The same holds for moral predicates. Despite agreement on all relevant empirical facts, people disagree about whether something is praiseworthy or not. Such a disagreement could stem from one party's failure to understand the meaning of the word; but more typically the disputants know what the word means, and their disagreement shows then that the empirical facts and the meaning of the word together do not determine its correct application. Those who judge it praiseworthy to teach children through beatings are morally mistaken; they need not linguistic instruction to improve their understanding of "praiseworthy," but a good discussion about how children should be educated.

A dispute over the application of an evaluative predicate such as "just" may thus be due to differences of three kinds: linguistic differences about its meaning; theoretical differences about which substantive conception or criterion of justice should guide its application; and empirical differences about the evaluated object.

The boundaries between these three kinds of differences is not sharp and may shift over time, as Wittgenstein memorably describes:

It might be imagined that some propositions, of the form of empirical propositions, were hardened and functioned as channels for such empirical propositions as were not hardened but fluid; and that this relation altered with time, in that fluid propositions hardened, and hard ones became fluid. The mythology may change back into a state of flux, the riverbed of thoughts may shift. But I distinguish between the movement of the waters on the riverbed and the shift of the bed itself; though there is not a sharp division of the one from the other.… And the bank of that river consists partly of hard rock, subject to no alteration or only to an imperceptible one, partly of sand, which now in one place now in another gets washed away, or deposited. (Wittgenstein 1969, §§ 96, 97, 99)

Not discussed by Wittgenstein, the boundary between linguistic and theoretical differences (in the case of evaluative predicates) is fuzzy and fluid in part because the controversies among competing conceptions reflect back upon the concept. Each conception seeks to introduce a certain order and unity. And sometimes elements of such a conception are widely found to be so convincing that they harden into an element of the concept.

Conduct-Guiding Function

The predicates "just" and "unjust" have not merely an evaluative but also a normative—hence conductguiding—function. Calling a possible action or law unjust is to oppose its implementation. To be sure, people do make justice judgments about the distant past, about hypotheticals and fiction. But even these judgments imply oughts—for example: that the Athenians ought not to have attacked neutral Melos; that it would have been all right for them to attack a Melos allied with Sparta; or that Angelo (in Shakespeare's Measure for Measure ) ought not to have demanded Isabella's virginity for the life of her brother.

The words "rational" and "irrational" are also conduct-guiding in this sense. Yet there is a difference. Whether an action is rational or irrational depends on the ends of the actor. Whether it is just or unjust is independent of these ends. Insofar as even an agent's ultimate ends can be criticized as irrational (Parfit), this difference becomes less deep. Like morality, rationality can then deliver unconditional judgments: an action is irrational if it was performed in pursuit of an irrational end.

Generally, claims of injustice are meant to evoke emotional rejection: condemnation, outrage, resentment. But this may be part not of the meaning of the word but of the usual pragmatic context of its employment.

Judicanda

The meaning of predicates is partly determined by their domain of application. Anyone who understands the meaning of "beautiful" and "just" knows how the domains of these predicates differ—knows, for instance, that a painting can be (un)beautiful but not (un)just whereas a patent regime can be (un)just but not (un)beautiful.

Things to which evaluative predicates are applicable can thus be called judicanda, from the Latin judicandum : that which is to be judged. The judicanda of justice may be categorized under four headings:

(a) individual and collective actors; that is, individuals as well as organized and unorganized groups such as a family, firm, state, or mob;

(b) the conduct of such actors, their actions and omissions;

(c) social rules, such as laws, social institutions, and conventions;

(d) states of affairs and events, such as the fact that some are much worse off than others or that some good persons suffer while some bad ones enjoy good fortune.

Because of the normativity of justice assessments, judicanda of the first two categories have a certain primacy. It is ultimately actors—and their conduct—who bear responsibility for the justice of social rules and, in part through these rules, for the justice of states of affairs. To be sure, sometimes people complain of injustice in states of affairs beyond human control. Such complaints may have a religious context; but they may also, where religion has faded, constitute a purely evaluative (nonnormative) use of "unjust."

One may think that this list is underinclusive, that human feelings constitute a fifth judicandum as exemplified by the anger of Achilles and its critique as unjust. But such locutions are better understood as meaning to assess the person and her or his conduct. It is Achilles whose justice is in question—on account of his tendency to get angry allegedly without good reason. Thus, talk of unjust feelings is imprecise, meant to call into question the justice of the person who feels this way.

One may think that the first category reduces to the second in a similar way: that justice assessments of actors really mean to judge these actors' conduct, and that the list is therefore overinclusive. But the nature of such a reduction would be controversial: Do assessments of actors refer to their actual conduct, to their conduct dispositions, to their intentions (toward conduct)? Such controversy shows that the proposed reduction does not hold as a matter of meaning. It can be a substantive element in a conception of justice. But, to state it as such, actors must be shall be a separate category of judicanda.

The same holds for the possible reduction of the third category to the second. One may think, for instance, that social rules are unjust if and only if it would be unjust to (help) impose them. But here, too, there are various other ways of formulating the reduction; and it is a substantive moral question which of these formulations, if any, is correct.

In view of the diversity of judicanda, one may dispair of Plato's grand ambition in the Republic and conclude that it is impossible to give a general characterization of the concept—let alone a general conception—of justice that plausibly covers all four categories of judicanda. Such doubt can only be dispelled by setting forth a plausible general structure for the concept (building on the general points already made) and some main hypotheses toward a unified conception of justice. This attempt is made in what follows.

A Unified Concept of Justice

In contrast to many one-place evaluative predicates (beautiful, good, conscientious, modest), "unjust" has an essential second place. It is indeed often used as a one-place predicate (as in "this man is unjust"). But reference to a second place is always implicit. In this respect, "unjust" is like the predicate "mother." To say that she is a mother is to say that she is the mother of someone. One cannot understand what it is to be a mother without understanding what it is to be the mother of someone. Likewise here: To call a man unjust is to say that he is unjust to others. One cannot understand what it is for a judicandum to be unjust without understanding what it is for it to be unjust to someone. Injustice conceptually requires recipients : those who receive unjust treatment from the judicandum.

To be sure, one may call a proposed law unjust even if it fails to pass and thus never treats anyone unjustly, and one may call a woman unjust, even if she treats no one unjustly, on account of her intentions or dispositions. But even in such cases an implicit reference to recipients is essential: one implies that the proposed law, if adopted, would treat some persons unjustly, and that the woman intends, or is disposed, to treat others unjustly.

Recipients need not necessarily be victims. It is possible that the injustice of an action or rule entails that some are treated better than they should be. In such cases, there may be other victims—people who, because of the action or rule in question, are unjustly treated worse than those who were treated too well. But when there are no such victims, can one then still speak of (victimless) injustice? To use an example from Kant's Metaphysics of Morals (p. 333): Would it be unjust if a society about to dissolve itself were to set free a convicted murderer from its jail? (Kant suggests that this would be unjust—but not because his execution is owed to the murderer himself, but because it is owed to his victim(s) and to all of humankind.) The concept of justice does not settle this question; different conceptions of justice will answer it differently.

The concept of justice involves an essential third place in that the notion of recipients, of those who receive just or unjust treatment, presupposes benefits and burdens that these recipients either should but do not have or do but should not have. What sorts of benefits and burdens these are, and whether they are understood in absolute terms or relative to what other recipients have, varies with judicandum and context.

Various aspects of justice have traditionally been distinguished. These distinctions can be displayed in three dimensions.

dimension one: first-order and procedural (in)justice

In a first dimension, one can distinguish assessments of a particular allocation of benefits and burdens (first-order justice) from assessments of the way in which such an allocation comes about (higher-order or procedural justice, sometimes also called fairness ). Thus a judicial divorce may be unjust on account of the ordered division of marital property, and it may also, and independently, be unjust because avoidably only one of the two parties was allowed to speak. The latter injustice—a violation of the classical precept audiatur et altera pars —involves (relative) second-order benefits and burdens: the advantaging of one party and the disadvantaging of the other in the decision-making process. In other cases, violations of procedural justice may involve absolute higher-order benefits or burdens, as when exculpatory or incriminating evidence or witnesses are arbitrarily excluded from a criminal trial.

Procedural justice plays an important role even outside jurisprudence. Rules and decisions about the awarding of honors, contracts, jobs, promotions, and university admissions may violate procedural justice. There are blatant cases, as when a coveted job goes to an insider's spouse without advertisement or search. In other cases it is controversial what the requirements of procedural justice are, exactly. The most important such controversies in recent decades have centered around the question of whether certain selection processes may or should favor people of a particular color, gender, or ethnicity so as to compensate for, or to help overcome, group disadvantages due to past or present discrimination.

In the realm of politics, especially, there may be third-order and fourth-order judgments of justice. This is the case when the rules or the participants for some procedure are selected through a metaprocedure. Thus, the electoral law of a country may be unjust independently of whether this has any effect on the composition of its legislature and, thereby, on its legislative output.

Even in private life one may find violations of procedural justice. It may be unjust if on the basis of a nasty rumor one discriminates against a person, or perhaps even repeats the rumor to third parties, without giving this person an opportunity to respond to the allegation. Insofar as the injustice of such conduct is independent of the truth of the rumor in question, it involves once more a higher-order burden.

dimension two: formal and material (in)justice

Perpendicular to this first dimension is the common distinction between formal and material (in)justice. Formal justice requires that relevantly similar cases be treated similarly. There is much room for controversy about whether particular cases are relevantly similar and about what is to count as similar treatment. The requirement of formal justice is therefore primarily a demand for justification. The complaint that relevantly similar cases are being treated in a dissimilar way demands a justification, showing that the treatment was not dissimilar or else was appropriately responsive to dissimilarity among cases. This requirement of justification holds across the first dimension: The procedures through which benefits and burdens are allocated to recipients as well as these allocations themselves must satisfy formal justice. It must be justifiable that one accused was acquitted and another convicted (first-order justice), say, and also that one accused but not another was provided an attorney at public expense (procedural justice).

Judicanda can be gravely unjust even when they clearly do treat similar cases alike. Thus, parents who beat all their children without cause violate material justice, as do judicial systems that deny all accused rights to speak, to appeal, and to consult legal counsel.

One might think that justice reduces to material justice: When dissimilar treatment of cases cannot be justified, this merely shows that the treatment of some of these cases is materially unjust, independently of how the others are treated. But this view is surely not implicit in the concept of justice. And it is substantively implausible. In many cases, justice is comparative. To illustrate, there are indefinitely many schedules according to which individuals and businesses might be taxed. While the demands of material justice disqualify a wide range of these options, they do not mandate one uniquely just tax code. Still, it would clearly be unjust to impose diverse materially just tax codes—some more advantageous than others—upon the various households or businesses of a single jurisdiction. Unjust is here the unequal treatment of taxpayers, which is a violation of formal justice.

dimension three: various domains of material (in)justice

Material justice imposes various requirements whose content is controversial. This multiplicity can be ordered in a third dimension by distinguishing various domains. Thus distributive justice deals with access to scarce resources—from the division of a pie to the structure of an economic order that regulates access to raw materials and the distribution of the jointly created social product. Commutative justice governs exchanges, which may be faulted for first-order flaws, as when the items exchanged are not equivalent, or for higher-order (procedural) flaws, such as excessive inequalities in information or bargaining power. Corrective or restitutive justice is concerned with how to make up for violations of social and moral rules and how to deal with the costs such violations cause. Retributive justice, finally, deals with the ascertainment and punishment of such violations.

In each of these four domains there are procedures, which may be formally or materially unjust, as well as particular allocations of benefits and burdens, which also may be formally or materially unjust. The traditional three-dimensional schema, as here reconstructed, thus contains sixteen boxes for sorting particular justice assessments. In the first dimension, the assessment is either first-order or higher-order. In the second dimension, it concerns either formal or material justice. And in the third dimension, it falls into the domain of distributive, commutative, corrective, or retributive justice.

This conceptual structure leaves open the possibility that some of these boxes may contain no substantive constraints. For example, there may be no first-order material demands of commutative justice if any exchange, no matter how lopsided, is morally acceptable so long as it has been performed freely by the exchanging parties without unfair inequalities in bargaining power or information. But this is not a conceptual point, but rather a substantive claim, affirmed by some (especially modern) and denied by other (especially medieval) conceptions of justice.

This complexity can be increased to sixty-four because—perpendicular to the three dimensions in which the character of diverse justice assessments can be differentiated—there is still the distinction between four categories of judicanda (introduced earlier) to which any such justice assessments can be applied: namely, individual and collective actors, their actions and omissions, social rules and institutions, and states of affairs and events.

In some accounts, international justice is given as a separate domain that, traditionally, is heavily focused on the use of force and, in particular, on just and unjust causes for going to war (ius ad bellum ), and on just and unjust ways of fighting a war (ius in bello ). (In his Metaphysics of Morals [pp. 343, 347–349], Kant adds a third theme: the just way of concluding a war so as to lay the foundation for a stable peace [ius post bellum ]) It seems more appropriate, however, to think of states as one class of collective actors among others. It may indeed be claimed that states are a special class of actors and that their conduct and relations are subject to distinctive justice requirements. But this is a substantive moral claim that should not be prejudged in the conceptual explication.

A state is understood not merely as a collective actor, but also as a comprehensive system of social rules and institutions enforced in a particular territory. A state can therefore be criticized as unjust not merely on account of its conduct toward or institutional relations with outsiders, but also on account of its internal institutional order—for instance, on account of how it distributes rights and duties, regulates and taxes economic cooperation, and enforces its laws against its members. So understood, states exemplify one kind of institutional scheme among others, whose special status once again should not be prejudged in the conceptual explication.

Toward a Unified Conception of Justice

The preceding sections were meant to sketch the structure of the contemporary concept of justice and thereby to characterize the linguistic consensus that underlies current debates about questions of justice. Even these general thoughts may not be wholly uncontroversial. But they are much less controversial than the thoughts to follow, which are meant to describe plausible elements of a conception of justice. Should these elements be found convincing, they might gradually become elements of a clearer and more unified concept of justice. But they do not now fully accord with the various understandings of this concept that are dominant in contemporary public and academic discussions.

It is possible to begin once more from the question, briefly raised in (1): What is specific about justice, how justice judgments are distinctive within the larger realm of moral judgments; what is being said with the complaint that some judicandum is unjust, over and above the claim that this judicandum is immoral (morally flawed)? What follows are four hypotheses toward answering this question.

first hypothesis: injustice involves abuse of morality itself

Morally flawed judicanda are unjust only if they involve an abuse of morality itself—that is, only if they appear with a moral pretension they do not live up to. Unjust is someone who is prepared to violate moral principles she herself likes to appeal to. Unjust is someone who allows his official conduct to be influenced by bribes even while he pretends to be an impartial judge, umpire, or mediator. Unjust is a beating falsely presented as deserved punishment. And unjust is legislation designed for the benefit of a small minority and yet claimed to impose moral obligations on the oppressed population. In all these cases, the injustice comes about through the false moral claim: Without the pretense of deserved punishment, the beating is wrong but not unjust. Without the claim of moral authority (that compliance is morally required), coercive rules can be wrong, but not unjust.

According to this first hypothesis, then, justice is a part of morality that defends the authority and dignity of morality itself. This would explain the central place justice is thought to occupy within morality. Unjust are only those judicanda that do not merely—openly or covertly—violate morality, but also appear under color of morality.

second hypothesis: justice may in principle be enforced

A second essential mark of injustice may be that it involves a violation of a right and, more precisely, of a moral right. (Injustice need not violate positive law. To the contrary, because positive law may be unjust or otherwise immoral, conduct it permits or even requires may still be unjust.) Violations of rights are those moral infractions that may in principle be averted through the use or threat of coercive force. The main implication of the second hypothesis therefore is that any injustice may in principle be forcibly averted. Justice may be enforced.

The expression "in principle" flags two qualifications. First, it is not meant that anyone is permitted forcibly to avert any rights violations. In a well-ordered society, for instance, only the police and the courts are permitted to use force in response to rights violations—ordinary citizens may use or threaten force only in urgent emergencies. Second, there is no permission to use force regardless of the morally significant costs of doing so. Defensive force may be grossly disproportionate and therefore impermissible. And efforts forcibly to avert a rights violation may also be impermissible when they carry the risk of triggering much graver violations of the rights of third parties (a point well illustrated in Heinrich von Kleist's story "Michael Kohlhaas"). That any rights violation may in principle be forcibly averted thus means only that those responsible for it and those profiting from it have no moral right against having this violation blocked by force. Here justice contrasts with other moral qualities. No moral rights are violated by judicanda lacking in generosity, civility, charity, humanity, decency, kindness, courtesy, mercy, or beneficence. Enforcing these virtues is wrong in principle and violates moral rights of those against whom they are enforced.

third hypothesis: injustice harms in violation of negative duties

Related to the preceding, another essential mark of injustice may be that it always involves a violation of negative duties on the part of those who are responsible for the moral quality of the judicandum in question. What is at issue here, roughly speaking, is the traditional distinction between moral infractions that harm recipients and those that merely fail to help them. However this distinction may be made precise, the third hypothesis is simply that a claim of injustice always involves a claim of undue harm.

This hypothesis is related to the second hypothesis insofar as the two distinctions may wholly or largely coincide. That they coincide precisely would mean that people in every case have a right not to be harmed but in no case a right to be helped. Therefore, only harmings may be forcibly averted, refusals of help must not be.

It is doubtful that the two distinctions coincide precisely. Many civil-law countries have "Good Samaritan" laws proscribing failure to render assistance. These statutes presuppose, as seems plausible, that there are cases in which morally mandatory assistance may be compelled through threat of criminal sanctions. Conversely, there may also be harmings that are so trivial that in principle they must not be forcibly averted.

Still, the two distinctions largely coincide, and some moral predicates fall entirely on one side or the other. Judicanda lacking in morally required or recommended generosity, civility, charity, humanity, decency, kindness, courtesy, mercy, or beneficence fail to help people without harming them. Such infractions may not be forcibly averted. Any injustice in a judicandum, however, does harm people and thus may in principle be forcibly averted.

The last two hypotheses differ in their informative value for the investigation of justice. The second hypothesis is illuminating: something important is learned about justice, if one finds that in principle every injustice may be forcibly averted. The third hypothesis would be similarly informative if it started out with a clear justice-independent notion of harm. The claim would then be that a judicandum is unjust only if it harms persons (in this independent sense). But the connection postulated by the third hypothesis can also be taken in the opposite way: Any unjust treatment is to count as a harming. So understood, this claim would be using a harm-independent notion of injustice to illuminate the notion of harm—rather than the other way around. Still, although the latter definitional sequencing seems more adequate, it is moderated by the need to preserve the core meaning of harm. Coming back to Kant's example, failure to give a criminal his deserved punishment cannot plausibly be presented as an instance of harming him. If one wants to call such an act of grace unjust (while maintaining the third hypothesis), then one must identify another harmed party—other criminals arbitrarily excluded from the amnesty, perhaps, or future victims of crimes that would not have occurred but for this failure to punish. Given this constraint imposed by the core meaning of harm, the third hypothesis is not then entirely uninformative.

fourth hypothesis: justice is not purely recipient-oriented

The last hypothesis is negative in character. It merely rejects a hypothesis that has come to dominate Anglophone academic discussions of justice. The refutation of this approach eliminates only one of many possibilities and may thus seem to make little progress. But because the rejected approach is so elegant and influential, it is interesting nonetheless to explore how it fails.

The approach to be criticized has developed out of utilitarianism, which holds that the moral assessment of actors, conduct, and social rules should be based solely on each such judicandum's relative impact on the world, and on human happiness in particular.

If such a view is accepted, the ordinary distinctions among moral predicates lose much of their significance. Whatever moral predicate may be used to criticize a rule, action, or person—cowardly, unjust, evil, indecent, and so on—the complaint always boils down to the judicandum's failure to be optimally happiness-promoting. Utilitarians have little use for the received panoply of moral predicates because it ultimately does not matter how a suboptimal judicandum is squandering potential human happiness.

Those who reduce the traditional multiplicity of moral defects in this way may simply want to do away with the surplus predicates. Alternatively, they may prefer to redeploy these predicates—using them not (as traditionally) for different defects of the same judicanda, but instead for defects in different kinds of judicanda. In this way, justice has come to name the specific moral virtue of social rules.

In his famous A Theory of Justice (1999 [1971]), John Rawls adopts this redeployment. His theory is focused exclusively on social institutions and, even more narrowly, on a society's basic institutional design. And, presenting justice as the first virtue of social institutions, he recognizes no further moral predicates by means of which such institutions might be subjected to potentially competing moral judgments.

Rawls further follows the utilitarians by adopting a broadly consequentialist mode of moral assessment, and one that focuses specifically on the well-being of individuals. Thus he shares with utilitarianism the purely recipient-oriented mode of moral assessment. The unique characteristic of this approach can be expressed as the assumption that the only information needed in the moral assessment of any judicandum is information about its relative impact on persons, that is, information about how persons fare with the judicandum as it is versus how they would fare if the judicandum were different.

Unlike the utilitarians, Rawls employs the purely recipient-oriented approach on two distinct levels and moreover rejects, on both levels, utilitarian conceptions of well-being as happiness or desire fulfillment. Central to his theory is a criterion ("the two principles") of justice that assesses alternative institutional designs on the basis of the distribution of social primary goods each would produce. The just design is the one that would produce the best feasible distribution among citizens of basic rights and liberties, income and wealth, powers and responsibilities of office, and other social bases of self-respect. This proposed criterion of justice is justified, again in a purely recipient-oriented manner, on the ground that its public adoption would lead to a better fulfillment of citizens' three higher-order interests than the public adoption of any alternative criterion of justice. The argument on this second level is presented in terms of a contractualist thought experiment ("original position") in which representatives of citizens, informed only that their clients have the three higher-order interests but given no further specific information about them, come to agree on a particular public criterion of justice.

Rawls's theory, with its particular assessment standards on the two levels, has received much lively critique. Under the Equality of What? label, academics debate whether human well-being should be conceived in terms of happiness, welfare, desire fulfillment, Rawlsian social primary goods, Dworkinian resources, or capabilities à la Sen or Nussbaum. And academics also debate how such well-being information about individuals should be aggregated: Should well-being simply be averaged or should special weight be given to equality, sufficiency, or the worse off? Yet, beneath all this disagreement, the purely recipient-oriented approach is largely taken for granted in the Anglophone countries.

The fourth hypothesis is that any such purely recipient-oriented conceptions of justice are untenable.

defense of the fourth hypothesis

When it comes to actors and their conduct, the fourth hypothesis is—at least outside academic philosophy—hardly controversial. Here any conception of the form "to be just is to promote a good distribution among one's recipients" is bound to be unacceptable because it disregards morally relevant information about how a judicandum has its effects. It is widely taken to make a great moral difference whether some conduct brings about the death of an innocent person or merely fails to prevent such a death. The world at large may be worse, perhaps even less just, if two good persons die prematurely than if a single less deserving person dies in their stead. And yet, it is not morally required, nor even permitted, to save the former by killing the latter.

An analogous point applies to social institutions. This is clearest when one reflects on the criminal law. Here Rawls's first priority rule (A Theory of Justice 1999 [1971], p. 266) holds that basic liberties may be restricted whenever such restriction, by making the remaining basic liberties more secure, is a gain on balance for the basic liberties of the representative citizen. This claim fits well with his purely recipient-oriented approach as enshrined in the contractualist thought experiment of the original position. But it is inconsistent with the considered judgments his theory was meant to accommodate and to unify. One of the examples Rawls gives (A Theory of Justice 1999 [1971], pp. 212–213) involves a strict-liability criminal statute that permits conviction without a showing of mens rea. Although it violates citizens' basic liberties, this law is nonetheless said to be permissible as "the lesser of two evils" if it is necessary to block even greater dangers to citizens' basic liberties (a danger of civil war, in his example).

Similar arguments could plausibly be made in regard to other aspects of the criminal law. Constraints on searches, seizures, and interrogations should be relaxed if this would entail, through more effective crime fighting, a net gain for citizen's basic liberties. Standards of evidence ("beyond a reasonable doubt") should be lowered, and less than unanimity be required for a jury conviction, if this would, through increased deterrence and through disabling more repeat offenders, produce a net gain for citizen's basic liberties. And draconian punishments (e.g., execution) should be imposed for high-elasticity crimes (e.g., drunk driving) if citizens' overall risk of a premature violent death can thereby be reduced.

These theory-produced judgments collide with people's intuitive sense of justice, which is responsive not merely to the magnitudes, probabilities, and distribution of morally significant goods and ills, but also to the attitude social rules manifest and to the causal pathways on which they affect their recipients. Even if these recipients themselves have no reason to care whether burdens falling on them (such as hunger or risks of premature death) are imposed by social rules or merely not averted by them, this difference is nonetheless morally significant. The purely recipient-oriented approach cannot then fulfill Rawls's stated ambition of accommodating and unifying his compatriots' considered judgments.

The reason is that judgments of justice take account not only of the passive perspective of recipients, of citizens as governed by social rules, but also of the active perspective of authors, of citizens as co-responsible for these rules. For citizens as recipients, all threats to their basic liberties are indeed on a par—whether they arise from criminal or crime-fighting activities. But for citizens as co-legislators it makes a considerable difference whether people are roughed up by criminals or by police interrogators, are killed prematurely by a drunk driver or through execution for drunk driving. As co-legislators, people take greater moral responsibility for harms they mandate or authorize through their institutional order than for equal harms they could prevent through it. That some institutional arrangements would be better for citizens as recipients is not sufficient to show that they are morally permitted—let alone required. Purely recipient-oriented conceptions of justice are bound to fail because they systematically ignore the active perspective of those who bear responsibility for a particular judicandum.

Plausible justice assessments of social rules must be sensitive to both citizen perspectives. What matters in the moral assessment of social rules is not merely how these rules affect, but how they treat their recipients. The importance of this point extends well beyond the criminal law. A rule under which those suffering from some genetic defect are not entitled to life-saving treatment may be exactly as bad for them as a rule mandating execution for those with certain physical features is for the executed. Though comparable in their relative impact on recipients, these two rules are nonetheless worlds apart in how they treat such recipients.

Once such differences in attitude and causal pathways have been restored their proper moral significance, people may well find that justice is not the only moral virtue of social institutions. If a society's institutional order provides little funding for public health services, more citizens will avoidably die or suffer. And yet, such an order is surely less unjust than one under which equal suffering and deaths are explicitly imposed by the rules, perhaps on members of a certain ethnic or religious minority. In fact, in a poor society with other urgent needs, low funding for public health services may not be unjust at all. And even in an affluent society, such low funding may not be unjust, so long as society helps treat the medical conditions it causes (e.g., through pollution). Still, if the institutional order of an affluent society ignores the plight of citizens with congentital medical conditions, say, it may well be morally flawed in other ways: be ungenerous, mean, or inhumane.

This result supports the fourth hypothesis by showing it to be a step toward a conception of justice that covers all judicanda in a unified way: In regard to actors and their conduct people are already familiar with the possibility of actors being beyond reproach in terms of justice and yet morally flawed in other ways.

See also Affirmative Action; Civil Disobedience; Distant Peoples and Future Generations; Feminist Social and Political Philosophy; Just War Theory; Natural Law; Punishment; Racism; Rawls, John; Reflective Equilibrium; Rights; Social Contract; Terrorism; Utilitarianism.

Bibliography

Ackerman, Bruce. Social Justice and the Liberal State. New Haven, CT: Yale University Press, 1980.

Aristotle. Nicomachean Ethics. [c. 330 BCE]. Translated by Terence Irwin. Indianapolis: Hackett, 1985.

Barry, Brian. Theories of Justice. Berkeley: University of California Press, 1989.

Beitz, Charles. Political Equality. Princeton, NJ: Princeton University Press, 1989.

Bentham, Jeremy. An Introduction to the Principles of Morals and Legislation [1789], edited by J. H. Burns and H. L. A. Hart. New York: Methuen, 1982.

Berlin, Isaiah. Four Essays on Liberty. Oxford: Oxford University Press, 1969.

Cohen, G. A. If You're an Egalitarian, How Come You Are So Rich? Cambridge, MA: Harvard University Press, 2001.

Daniels, Norman. Just Health Care. Cambridge, U.K.: Cambridge University Press, 1985.

Dworkin, Ronald. Sovereign Virtue. Cambridge, MA: Harvard University Press, 2000.

Dworkin, Ronald. Taking Rights Seriously. Cambridge, MA: Harvard University Press, 1978.

Elster, Jon. Local Justice. New York: Russell Sage Foundation, 1992.

Galston, William. Justice and the Human Good. Chicago: University of Chicago Press, 1980.

Gauthier, David. Morals by Agreement. Cambridge, U.K.: Cambridge University Press, 1987.

Hart, H. L. A. The Concept of Law. Oxford: Oxford University Press 1961.

Hayek, Friedrich A. The Mirage of Social Justice. Chicago: University of Chicago Press 1976.

Hegel, G. F. W. Elements of the Philosophy of Right [1821]. Translated by H. B. Nisbet. Cambridge, U.K.: Cambridge University Press, 1991.

Hobbes, Thomas. Leviathan [1651], edited by C. B. MacPherson. Harmondsworth, U.K.: Penguin, 1968.

Hume, David. An Enquiry concerning the Principles of Morals [1751], edited by Jerome B. Schneewind. Indianapolis: Hackett, 1983.

Hurley, Susan. Justice, Luck and Knowledge. Cambridge, MA: Harvard University Press, 2003.

Kant, Immanuel. The Metaphysics of Morals [1797]. Translated by Mary Gregor, cited by Academy pagination. Cambridge, U.K.: Cambridge University Press, 1996.

Locke, John. Two Treatises of Government [1689], edited by Peter Laslett. Cambridge, U.K.: Cambridge University Press, 1960.

Marx, Karl. Selected Writings, edited by David McLennan. Oxford: Oxford University Press, 2000.

Mill, John Stuart: The Basic Writings of John Stuart Mill: On Liberty, The Subjection of Women and Utilitarianism [1859, 1869, 1861], edited by Jerome B. Schneewind. New York: Random House, 2002.

Miller, David. Principles of Social Justice. Cambridge, MA: Harvard University Press, 1999.

Nagel, Thomas. Partiality and Equality. Oxford: Oxford University Press, 1991.

Nozick, Robert. Anarchy, State, and Utopia. New York: Basic Books, 1974.

Nussbaum, Martha. Sex and Social Justice. Oxford: Oxford University Press, 1999.

Okin, Susan. Justice, Gender, and the Family. New York: Basic Books, 1989.

O'Neill, Onora. Toward Justice and Virtue. Cambridge, U.K.: Cambridge University Press, 1996.

Parfit, Derek. Reasons and Persons. Oxford: Oxford University Press, 1984.

Plato. The Republic. [ca. 380 BCE]. Translated by G. M. A. Grube. Indianapolis: Hackett, 1974.

Rawls, John. Political Liberalism. New York: Columbia University Press, 1993.

Rawls, John. A Theory of Justice [1971]. Cambridge, MA: Harvard University Press, 1999.

Raz, Joseph. The Morality of Freedom. Oxford: Oxford University Press, 1986.

Roemer, John. Theories of Distributive Justice. Cambridge, MA: Harvard University Press, 1996.

Rousseau, Jean-Jacques. The Social Contract [1762]. Translated by M. W. Cranston. Harmondsworth, U.K.: Penguin, 1968.

Scanlon, Thomas M. What We Owe to Each Other. Cambridge, MA: Harvard University Press, 1999.

Sen, Amartya. "Equality of What?" In Choice, Welfare and Measurement, edited by Amartya Sen. Cambridge, U.K.: Cambridge University Press, 1982.

Sen, Amartya. Inequality Reexamined. Cambridge, MA: Harvard University Press, 1992.

Sen, Amartya, and Bernard Williams, eds. Utilitarianism and Beyond. Cambridge, U.K.: Cambridge University Press, 1982.

Sidgwick, Henry. The Methods of Ethics [1874]. Indianapolis: Hackett, 1981.

Smith, Adam. The Theory of Moral Sentiments [1759], edited by D. D. Raphael and A. L. Macfie. Oxford: Oxford University Press, 1979.

Tuck, Richard. Natural Rights Theories. Cambridge, U.K.: Cambridge University Press, 1979.

van Parijs, Philippe. Real Freedom for All: What (if Anything) Can Justify Capitalism? Oxford: Oxford University Press 1995.

von Humboldt, Wilhelm. The Limits of State Action [1852]. Cambridge, U.K.: Cambridge University Press, 1969.

Walzer, Michael. Just and Unjust Wars. New York: Basic Books, 1977.

Walzer, Michael. Spheres of Justice. New York: Basic Books, 1984

Williams, Bernard. Ethics and the Limits of Philosophy. Cambridge, MA: Harvard University Press, 1985.

Wittgenstein, Ludwig. On Certainty, edited by G. E. M. Anscombe and G. H. von Wright. Oxford: Blackwell 1969.

Thomas Pogge (2005)