5
Immanent Criticism of an Essentially Contested Practice

1. Introduction

In chapter 3 we examined two conflicting accounts of why we punish, of the justification of the practice; the utilitarian believes we punish to augment social utility, whereas the retributivist believes we punish to express society's condemnation of acts it regards as blameworthy, to vent its righteous anger, and to vindicate right—we punish for justice. We said that in choosing between the two principles we are deciding which ideal on the whole better accords with the purpose of the practice, and that ultimately to decide that question we must step inside the practice and see what is done. In chapter 4 we did just that. There I argued that central features of legal punishment are best accounted for by the retributivist ideal. In chapter 4 we also saw how other features of the actual practice diverge from the retributivist principle, that these features are inconsistent with the retributivist account of the purpose of the practice. The retributivist argues that punishing a person who cops a plea, or someone who is insane, or a person whose act of wrong was done in the name of some higher law or as a political act, or with an extended sentence because the criminal is a repeat offender, is not really "punishment." In suggesting

this we were supposing that there is a practice called legal punishment that has a certain meaning and accords with a certain principle, and that an action that is nominally part of the practice is essentially or by right excluded from it if the action can't be accorded with that principle. We assumed that the features that accord with the retributivist ideal are somehow "central," a claim that itself demands justification. In other words, we assumed that there is a practice of legal punishment that is essentially retributive, and that the utilitarian is somehow mistaken about the meaning and purpose of punishment. In this chapter we shall consider what is wrong with this assumption and what its rejection means to those who would defend retributive (or, for that matter, utilitarian) immanent criticism.

The retributivist claims that certain features of the actual practice aren't really part of the practice, aren't really punishment. This leads to the question: How do we know what counts as a practice and determine what the practice includes, or which features are "central"?

2. What Counts as a Practice: The Turn to Interpretation

What might we mean when we call the spanking of a child or the arrest, conviction, and incarceration of a person a practice?^[1] The Oxford English Dictionary tells us that the word "practice" can mean "a habitual doing or carrying on of something." But not everything we do habitually is a practice, and sometimes to do something out of habit is not to engage in it as a practice. I might say I have a bad habit of biting my nails, but I would be unlikely to say that I engage in the practice of nail-biting. It seems, then, that a practice is more than an action we do over and over again. The notion that a practice

[1] In chapter 1 we considered John Rawls's account of a practice: a practice is defined by fixed rules. But we criticized this account, on the grounds that many practices don't have fixed rules.

is something "carried on" implies that it is intended; we don't call the earth's rotation about the sun a practice, even though it occurs repeatedly. A practice, rather, seems to be an action or set of related actions that is intended or has some purpose. It makes sense to ask of a practice, why do we do it?

When someone engages in a practice with which we are familiar, we can recognize this and guess at why that person acts so. Of course, we might be wrong: we might see an adult spank a child and guess that this was punishment, but they might just be playing some game. Some sort of principles or standards are involved in practices, so that when we observe a spanking, or an execution, we can suppose that certain things led up to this action and were the reason for it. We could go up to the adult and say, "Why are you punishing this child? Did she do something wrong?" And the adult might say either yes and explain what mischief the child had been up to, or no and perhaps explain the game they are playing that looks like but isn't punishment (we are assuming that this is some fictitious world where adults don't say "None of your damned business").

There is a funny thing about practices, though. They are the sorts of things we recognize as having a meaning or purpose, but often we can't easily say what that meaning or purpose is. Spanking a child and arresting, convicting, and incarcerating a person are practices in our society; but how do we know what meaning or purpose these practices have? What if we disagree?

The view I shall take is that we "know" the meaning or purpose of a practice by an act of interpretation.^[2] It is a non-foundationalist view.^[3] A foundationalist believes that practices

[2] Several issues are raised by the claim that we "know" by interpretation: Knowledge in what sense of the word? Can interpretation ground practices (or, in the language of hermeneutics, can interpretation ground signs)? Why say we can only interpret the meaning of a practice—can't we understand a practice's meaning, i.e., truly know it? The line I shall take is that in the case of legal punishment the consensus is that we can't know; punishment is an essentially contested practice.

[3] Here I borrow terms used by Don Herzog in Without Foundations: Justification in Political Theory (Ithaca: cornell University Press, 1985), pp.. 24, 225.

have fixed grounds which justify them; that we can prove an action is right by absolute principles about which we can have certain knowledge. According to the foundationalist, we can be mistaken about why we do what we do, about the meaning of our practices. A practicing Jew or Muslim who refuses to eat pork on the grounds that doing so would bring forth divine retribution, when in fact the taboo arose for ecological and economic reasons (see chapter 2, section 5), is, according to the foundationalist, wrong about why she does not eat pork. Nonfoundationalists take a different position. Practices, in their view, have no absolute grounds; the origin of a practice, its natural cause, needn't be connected with the purpose the practice has now.^[4] The nonfoundationalist claims that ultimately a justification of a practice ends with the fact that this is what we do. In answering the question, Why do we do this? all we can do is construct reasons until eventually those who share in the practice are satisfied and stop asking "why" questions. If someone persists in asking, "But why is that a good reason?" eventually we can only shrug our shoulders. The nonfoundationalist believes there are no transcendent principles to which we can point that ultimately ground our practices and force an end to "why" questions. The nonfoundationalist does not deny that there can be justification; she claims, not that there is no sense in which we can say someone is right or wrong, that there are no grounds for justifying what we do; but, rather, that ultimately justification comes to an end, and that that end is a set of convictions and beliefs which do provide grounds, but which themselves cannot be further justified.^[5]

[4] We saw that this was Nietzsche's position; see chapter 2, section 2.1.

[5] A particularly intriguing version of nonfoundationalism is provided by the philosopher Ludwig Wittgenstein, a nonfoundationalist with respect not only to social practices but also to claims in the natural sciences and mathematics. In On Certainty , trans. Denis Paul and G. E. M. Anscombe (New York: Harper and Row, 1969), Wittgenstein takes the position that "to be sure there is justification; but justification comes to an end" (par. 192). Our justifications ultimately rest on a shared system of convictions, what he calls the "inherited background," that itself cannot be justified (cf., for example, pars. 94, 102, 105, 141–44, 298). Wittgenstein writes: "We all believe that it isn't possible to get to the moon; but there might be people who believe that this is possible and that it sometimes happens. We say: these people do not know a lot that we know. And, let them be never so sure of their belief—they are wrong and we know it. If we compare our system of knowledge with theirs then theirs is evidently the poorer one by far" (par. 286). But all Wittgenstein says is that "we say" that "they are wrong," not that in fact "they are wrong." For Wittgenstein, justification rests on "persuasion" (pars. 262, 609–12, 669).

Commitment to nonfoundationalism means that in thinking about the justification of legal punishment, no account in itself is privileged; what does privilege an account is its ability to persuade to a satisfactory degree those who view the practice differently.^[6] As a nonfoundationalist, I can't resolve the debate between utilitarians and retributivists by appealing to some transcendental standard by which we can say that either utilitarianism or retributivism is wrong , but only by arguing that retributivism offers a better interpretation of what punishment is.

The view that interpretation is the way to come to the meaning of practices and even to a conception of what counts as and constitutes a practice is central to the recent work of Ronald Dworkin.^[7] Dworkin argues that in interpreting a practice we begin with some pre-theoretical, intuitive notion of it. The act of interpretation itself begins when we imagine this act or set of actions we assume is a practice to be a set of rules with a value or purpose or principle that guides the rules and would warrant a change of actual practice should its practitioners begin to act counter to this purpose or principle. Dworkin argues that we give constructive interpretations of practices, interpretations that are concerned with the purpose not

[6] What counts as a satisfactory degree will depend on the particular context in which we disagree and are drawn to the activity of justifying.

[7] Ronald Dworkin, Law's Empire (Cambridge, Mass.: Harvard University Press, 1986).

of the author but of the interpreter of the practice.^[8] The practice is a "form of life" and has a vocabulary that is shared by people, who may disagree about the purpose of the practice while nevertheless sharing in it as a form of life.^[9]

Dworkin emphasizes that what is at stake in our interpreting practices is our ability to criticize and change the practice. Our interpretation may lead us to change our view of what the practice requires. Upon our interpretive effort we may find that the practice makes unjust demands that turn out not really to belong to the practice; they are condemned by principles needed to justify other demands the practice imposes.^[10] The act of interpretation is in a sense constitutive of the practice itself. In Dworkin's view there is a dialectic of sorts: how we view the practice shapes what we think it requires; but, also, how we view what is required may shape how we interpret the practice. The implication of this view is that our practices have no foundation other than the interpretations we give to them.

We might think that this view that we know the meaning or purpose of practices, and even what constitutes the practice, on the basis of interpretation is a rather subjectivist account: aren't practices out there, aren't their purposes waiting to be discovered and understood? But if we hold a dialectical account of interpretation, we can meet the charge of subjectivism by noting that our interpretation is of a practice—all that we say about the practice is about something of which we have some pre-theoretical ideas whose source is "out there." It is in this respect that the critical work our interpretation does is immanent criticism. in saying that we can only interpret the meaning of the practice, we are owning up only to the idea that we can have practices that don't obviously serve any single purpose and may serve competing purposes, and that we may disagree about what the purpose(s) is (are) though we still all engage in them.^[11] There is no single purpose

[8] Ibid., p. 52.

[9] Ibid., p. 63.

[10] Cf. ibid., p. 204.

[11] On the idea that some of our concepts are essentially contested—that is, that we all use the concept (or engage in a practice) although we disagree about what the concept (or practice) is all about—see W. B. Gallie, "Essentially concepts," in Gallie, Philosophy and the Historical Understanding (London: Chatto and Windus, 1964).

that everyone would (or anyone is compelled to) agree accounts for everything significant about the practice. Any claim about the purpose of such practices must be an interpretation. Our discussion of the practice of legal punishment in chapters 2–4 suggests that it is such a practice. To further illustrate this point, that legal punishment is a practice with conflicting principles immanent in it, I include an appendix that discusses how, in what appear to be two instantiations of the practice of legal punishment, aspects of the practice can be accounted for only by distinct and mutually exclusive interpretations of its purpose.

3. An Objection to the Retributive Interpretation of Punishment

I have defended the retributive interpretation that we punish for justice and have even advanced the argument that to punish people in a way that violates the retributive ideal is to engage in some practice other than punishment. But there is a problem with the claim that features of legal punishment that violate the retributive ideal aren't part of the practice: they are . It's unconvincing to say that punishing a person who cops a plea, or who is insane, or whose act of wrong was a political act of protest is not really punishment. To call any of these "punishment" would be good English. Slightly more convincing is the retributivist's argument that grammatically we can't punish someone who merely violates a regulation but does nothing blameworthy; we can only inflict upon this person a penalty.^[12] But ordinary language is ambiguous even here: we

[12] See Joel Feinberg, "The Expressive Function of Punishment," in Doing and Deserving: Essays in the Theory of Responsibility (Princeton: Princeton University Press, 1970); and R. A. Duff, Trials and Punishments (Cambridge: Cambridge University Press, 1986), who argues that if doesn't fulfill the Kantian requirement that the defendant "participate" in the punishment by the understanding why she's being punished, then it's not "punishment" (p. 28; cf. pp. 37–38: the insane "can't" be punished).

speak of the murderer receiving the death "penalty." And the claim that it's impossible to punish someone for something she did not do is not at all nonsense, yet nevertheless there is precedent in ordinary language for saying the opposite. The ambiguities in the word "punishment" reflect the tensions in the practice of legal punishment.

The difficulty the retributivist immanent critic faces is that legal punishment is an essentially contested practice: both utility and retribution are principles immanent in the practice. Recent philosophical discussions about utilitarian and retributive theories of punishment have arrived at what might seem to be a similar conclusion: both principles are seen to be "there" in the practice, sharing in what might be called a division of justificatory labor; and neither the utilitarian nor the retributive theory alone is sufficient—we need some optimal mix. In the next section I shall consider the major compromise approaches and discuss why I think that although they are right to acknowledge that both principles are at work in legal punishment, in trying to reconcile the two they fail to take seriously enough how each principle lays claim to be the sole principle of the practice.

4. The Contemporary Utilitarian-Retributive Debate

Contemporary theorists of punishment have sought to reconcile utilitarianism and retributivism. Historically, they did this in the context of defending utilitarianism against the objection made by retributivists that the principle of utility, consistently applied, recommends in principle the punishment of the innocent, for surely in some cases doing this would be "optimific."^[13] Most utilitarians saw this charge—violation of

the retributive demand to punish only the guilty—as a valid concern and sought to preserve their utilitarian account while recognizing the retributive demand.^[14] By taking what has been called "the middle way,"^[15] they understood the two principles as sharing in a division of justificatory labor. In one version, the retributivist is said to justify how we punish in particular cases by pointing to the rules of the practice, one of which is to punish only the guilty, whereas the utilitarian is said to justify why we have the practice to which the rules belong.^[16] According to this rule-utilitarian account, the utilitarian avoids the unacceptable position of justifying punishment of the innocent, without violating the principle that we should do what has the best consequences, by claigling that in the long run it's best to follow (the best) rules.^[17] In another version, the

[13] See Igor Primoratz, Justifying Legal Punishment (Atlantic Highlands, N.J.: Humanities Press International, 1989), chs. 2–3; J. J. C. Smart, An Outline of a System of Utilitarian Ethics (Adelaide, Australia: Melbourne University Press, 1961); C. L. Ten, Crime, Guilt, and Punishment (Oxford: Oxford University Press, 1987). Some utilitarians deny that their principle justifies punishment of the innocent. To punish someone who has not been found guilty of a crime would not be really to punish them. See Anthony Quinton, "On Punishment," Analysis , vol. 15 (June 1954), p. 138; K. Baier, "Is Punishment Retributive?" Analysis , vol. 16 (1955), p. 27; C. K. Benn, "An Approach to the Problems of Punishment," Philosophy , vol. 33 (October 1958), p. 331. But see Arnold Kaufman, "Anthony Quinton on Punishment," Analysis , vol. 20, no. 1 (October 1959), p. 13; and K. G. Armstrong, "The Retributivist Hits Back," Mind , vol. 70 (October 1961).

[14] One exception is J. J. C. Smart, who faithfully abides to his utilitarian tenets: if it would augment social utility, he demands, why not punish the innocent? "Extreme and Restricted Utilitarianism" in Michael Bayles, ed., Contemporary Utilitarianism (Gloucester, Mass.: Peter Smith, 1978), p. 113.

[15] Primoratz, Justifying Legal Punishment , ch. 6.

[16] John Rawls, "Two Concepts of Rules," in Bayles, ed., Contemporary Utilitarianism , pp. 59–98. See also R. M. Hare, Moral Thinking:Its Levels, Method and Point (Oxford: Oxford University Press, 1981), for another effort to incorporate retributivism, understood as rule-following, into a utilitarian theory.

[17] I understand Rawls to put forth a different argument: to engage in the practice is to justify actions of the practice by appealing to its rules; we do this, not because these rules are best, but because we must appeal to them if we are to engage in the practice. Rawls says his point is "a logical point" and that "[t]here is no inference whatsoever to be drawn with respect to whether or not one should accept the practices of one's society" ("Two Concepts of Rules," p. 97).

retributivist is said to justify whom (only the guilty) an`d how much (an amount commensurate to the evil) we may punish, whereas the utilitarian is said to justify why we punish at all (to augment social utility).^[18] By seeing each principle as fulfilling a different function, these theorists claim to give a utilitarian justification of punishment that respects the retributive demand that we not punish the innocent.

These attempts at an attractive and neat reconciliation are widely accepted.^[19] Their various proponents share a widespread prejudice among us that contradiction and conflict among principles are bad. But I believe that the reconciliation these theorists have achieved is illusory. There is no simple division of justificatory labor such that utility and retribution are compatible principles, both at work fulfilling different functions within the practice. Legal punishment is an essentially contested practice. To say, as does Rawls, that utilitarians justify why we have the rule-guided practice while retributivists justify actions of the practice by appealing to those rules is to give an inaccurate account of the arguments actually

[18] H. L. A. Hart, "Prolegomenon to the Principles of Punishment," in Stanley Grupp, ed., Theories of Punishment (Bloomington: Indiana University Press, 1971). Unlike Rawls, Hart casts his theory as a compromise theory, in that he acknowledges that our practices have conflicting principles, and so we need a compromise (H. L. A. Hart, Punishment and Responsibility [Oxford: Clarendon Press, 1968], p. 10). I shall nevertheless speak of his theory as an attempt at reconciliation.

[19] Cf. Ernest van den Haag, Punishing Criminals (New York: Basic Books, 1975), pp. 25–26; Andrew von Hirsch, Doing Justice (Westford, Mass.: Northeastern University Press, 1986): "[W]hile deterrence accounts for why punishment is socially useful, desert is necessary to explain why that utility may justly be pursued at the offender's expense" (p. 51); and Ten, Crime, Guilt, and Punishment . Ten adopts a "revised" compromise theory of punishment, the details of which needn't concern us here. Von Hirsch and Ten both take a pluralist approach, where both principles are given due weight.

made. Rawls's argument, which is seminal to the development of rule utilitarianism, has often been criticized as an incoherent version of utilitarian theory.^[20] My point is different. It is that Rawls and others who have taken "the middle way," by focusing on the philosophical problem of how to defend utilitarianism against the retributive charge that it justifies punishment of the innocent, give an account of punishment that is untrue to our practice, to arguments actually made by practitioners. Retributivists do not justify only whom or how much we may rightly punish; some retributivists declare that we must punish, as Kant does with his soon-to-be-deserted-island example. Both retributivists and utilitarians offer accounts of why we punish at all (to express condemnation of blameworthy acts and mete out justice; to augment social utility); and both counsel us on what rules should guide us regarding the distribution of punishment within the practice. Sometimes we require mens rea to convict a defendant; many acts are made crimes because society regards these acts as morally wrong; in sentencing we, generally try to fit the punishment to the moral gravity of the offense; and we allow for mercy and pardons. Although some of these features of our practice have been given utilitarian rationales, all accord unambiguously only with the retributive principle.^[21] But sometimes we hold people accountable to a strict liability standard; we punish acts that are clearly not morally culpable (such as unknowingly selling adulterated food);^[22] and in determining a criminal's sentence we sometimes take into account the threat of

[20] Cf. Primoratz, Justifying Legal Punishment ; Ten, Crime, Guilt, and Punishment .

[21] On the point that utilitarians can offer no account of mercy or pardon, because these presuppose a wrong deserving punishment, see Primoratz, Justifying Legal Punishment , pp. 36–37; and Alwynne Smart, "Mercy," in H. B. Acton, ed., Philosophy of Punishment (London: Macmillan, 1969), pp. 212–28.

[22] The example may be misleading, since many legal theorists call such an offense a "public welfare offense," which they distinguish from strict liability offenses. See Richard Wasserstrom, "Strict Liability and the Criminal Law," Stanford Law Review , vol. 12 (July 1960), pp. 731–45.

future harm posed by the criminal, regardless of the moral gravity of the offense of which he has been convicted. These features of our practice accord only with the utilitarian principle. Two principles are immanent in our practice; they pervade it through and through, and sometimes they conflict. By insisting that the retributivist speaks only to questions concerning the distribution of punishment, questions inside the practice, and that the utilitarian speaks only to the question of why we punish at all, Rawls and other compromise theorists fail to take seriously enough the claims of either the retributivist or utilitarian immanent critic, both of whom employ their distinct conceptions of why we punish at all to criticize or justify actions taken within the practice. By acknowledging that legal punishment is essentially contested, we are forced to take seriously both retributive and utilitarian immanent critics.

I believe (though I do not know how to go about proving) that legal punishment is not merely contested, conflict-ridden, or filled with inconsistencies, but that it is essentially contested.^[23] There are several senses of "essentially contested," all of which I believe apply to the practice of legal punishment. When we say a practice is essentially contested we might mean that one doesn't adequately comprehend the practice unless one comprehends each of the essentially contested understandings. Each of these understandings is an essential part of an adequate comprehension of the practice. In this sense punishment is essentially contested, for if by punishment we understood only retributive punishment or only utilitarian punishment, our understanding would be deficient—it would leave out many features which both utilitarians and retributivists would have to admit are part of our present practice of punishment. To say that a practice is essentially contested might also be to say that the core of the practice is as up for grabs as the peripheries or boundaries of the practice.^[24] In a

[23] x is "essential" to y if lacking x, y would not be y .

[24] On the idea of "core" and "periphery" with respect to practices, see James Griffin, Well-Being: Its Meaning, Measurement, and Moral Importance (Oxford: Clarendon Press, 1986).

practice that by this criterion is not essentially contested, there may be actions which some think appropriate to the practice and others think inappropriate, not properly part of the practice, not consistent with its purpose; but all will agree what that purpose is. By contrast, they will not so agree in a practice that by this criterion is essentially contested. In this sense, too, punishment is essentially contested. Utilitarians and retributivists do not merely disagree about whether this or that action properly belongs to the practice; they disagree about the fundamental purpose of punishment. A third meaning of essentially contested is "necessarily or inevitably contested." This is a very strong criterion, but still one by which punishment might be said to be essentially contested.

As a practice, legal punishment has underlying it some idea; it fulfills some purpose. But this purpose is realized, in the case of legal punishment and many other practices, only through the establishment of institutions, and there are important consequences following from the institutionalization of a practice. Some theorists have suggested that one such consequence is a perhaps inevitable tension between the original idea underlying the practice and the requirements imposed by the institutions needed to realize that idea. Hanna Pitkin distinguishes the institutionalized existence of practices, which she calls their "form," from the essential purpose or idea underlying these practices, which she calls their "substance."^[25] For example, in the debate, in Book I of Plato's Republic , between Socrates and Thrasymachus regarding what justice is, Socrates refers to the idea (or substance) of justice, while Thrasymachus refers to what justice in its institutionalized existence has come to mean, namely, "the advantage of the stronger." Pitkin argues that both Socrates and Thrasymachus are talking about the same concept, justice, but that

[25] Hanna Pitkin, Wittgenstein and Justice (Berkeley: University of California Press, 1972), p. 187; cf. pp. 186–92.

the conflict between the two reflects a typical process: when ideas or purposes become institutionalized, they become transformed. Justice takes on a new meaning. The original meaning is not lost, however; it is preserved and is in tension with the transformed conception.^[26] Pitkin, then, suggests that the contested character of justice derives from the tensions that result when ideas and purposes are institutionalized. Such a process might account for the essentially contested character of legal punishments.^[27]

Several theorists have argued that the original purpose of punishment was to mete out justice, which some think was originally manifested as the "primitive" urge to avenge.^[28] Hegel notes that the German word for justice, Gerechtigkeit , is related to the German word for revenge, Rache .^[29] Some argue that responsive attitudes underlie punishment: we punish as a response to blameworthy actions (in response perhaps to a "primitive urge," perhaps to an abstract conception of just

[26] Cf. Pitkin, Wittgenstein and Justice , ch. 8. I do not mean to suggest that Pitkin thinks ideas come from nowhere. She describes two processes: from concept to practice, and vice versa.

[27] Pitkin herself suggests the application to punishment in Wittgenstein and Justice , p. 188.

[28] See Karl Menninger, The Crime of Punishment (New York: Viking Press, 1966); Hans von Hentig, Punishment: Its Origin, Purpose and Psychology (1937; reprinted Montclair, N.J.: Patterson Smith, 1973); Emile Durkheim, The Division of Labor in Society , trans. George Simpson (1893; reprinted Glencoe, Ill.: Free Press, 1933). René Girard, in Violence and the Sacred , trans. Patrick Gregory (1972; reprinted Baltimore: Johns Hopkins University Press, 1977) sees legal punishment as a functional equivalent to primitive sacrifice, which puts an end to the chain of revenge violence. Nietzsche says that the idea that the criminal deserves punishment came later; originally punishment's roots were in "anger at some harm or injury, vented on the one who caused it," but that this anger was "held in check" by the idea of paying back the injury through equivalent pain, an idea that has its roots "in the contractual relationship between creditor and debtor" (Friedrich Nietzsche, On the Genealogy of Morals , trans. Walter Kaufmann and R. J. Hollingdale [New York: Vintage, 1969], essay 2, section 4).

[29] G. W. F. Hegel, Vorlesungen über Rechtsphilosophie (1818–1831) , ed. Karl-Heinz Ilting (Stuttgart-Bad Canstatt: Friedrich Fromman, 1973), vol. 4, p. 294.

deserts); and these responsive attitudes are not utilitarian. But we translate these attitudes into institutional contexts; once we have done this, utilitarian considerations force themselves upon us.^[30] In this view, when punishment became institutionalized we of necessity became utilitarians (or at least consequentialists).^[31]

But perhaps this is not the right story. It is difficult to say what is original idea or "core justification" and what is institutional consequence. Bentham argues the opposite. For Bentham, the original idea underlying punishment, its core justification, is utility. For Bentham, it is the demands on which retributivists typically insist that are institutional consequences. Bentham suggests that as a result of having a practice engaged in by many people, demands emerge for fairness, equal treatment, due process. One of these demands is that we punish only for acts stipulated by law as mischiefs. Once we codify laws as part of a practice—once we declare certain commonly occurring mischiefs to be crimes—the demand emerges that we punish only crimes; and Bentham implicitly acknowledges this retributive demand.^[32]

No matter which story is right.^[33] In either case the contested character of legal punishment arises from the interplay between ideas and institutional consequences.

[30] Joel Feinberg argues this in Doing and Deserving , pp. 82–83.

[31] As I argued in chapter 4, retributivism needn't be nonconsequentialist or deontic; the retributivist can acknowledge that institutions are required to attain her ideals at all and can therefore insist, as a condition of her retributivism, on the maintenance of the institutions, even if that requires violating the ideals.

[32] See Jeremy Bentham, An Introduction to the Principles of Morals and Legislation (1789; New York: Hafner Press, 1948), ch. 15, section 25: in discussing the accidental punishment of a person innocent of an offense, Bentham uses the phrase "justly punished" to describe the punishment deserved by someone guilty of an offense. Bentham thus implicitly acknowledges the retributive principle that we may punish only those who commit offenses—justice demands this. But see Primoratz's discussion of Bentham, Justifying Legal Punishment , p. 26, for counterevidence.

[33] An appeal to ordinary language might suggest that retribution is core and that utilitarian considerations emerged as institutional consequence. it makes sense to argue that we can't punish someone who hasn't done a wrong (though ultimately this is an unconvincing argument). It would be far more odd to try to argue that when we punish someone convicted of a crime who is neither deferrable nor rehabilitatable nor a future threat, we really aren't "punishing."

5. Retributive Immanent Criticism in an Essentially Contested Practice

If I am right that legal punishment is an essentially contested practice, then the effort of contemporary theorists to bring together utilitarianism and retributivism by claiming that the two principles share in a division of justificatory labor, or that each is to be given due weight in a pluralist approach, is misdirected. Truth lies not in false reconciliation but in acknowledgment of incompatibility and difference.^[34] But how will this truth help the practitioner of an essentially contested practice? The immanent critic uses the principle(s) underlying the practice as a guide for action within the practice. But if both utility and retribution are principles immanent in, indeed essential to, legal punishment, what is the immanent critic to do? Adherence to one principle implies a resolution to a practical problem that is precisely the opposite of the resolution implied by adherence to the other principle.

Of course, one possibility for the immanent critic of an essentially contested practice is to argue that the practice ought not to be essentially contested; for example, the retributivist can argue that punishment is essentially retributive and all aspects of the practice that violate that principle, such as plea-bargaining, extended sentences for repeat offenders, or use of a strict liability standard of accountability, should be stricken from the practice as inconsistent with its purpose, which is the condemnation of blameworthy action. The retributivist, in other words, can try to persuade us that retribution

[34] There are still whole-hearted utilitarians (Jeremy Bentham and J. C. C. Smart, for example) and whole-hearted retributivists (Kant—with respect to moral, not legal, punishment; R. A. Duff; and Igor Primoratz). But their views, too, are inadequate, for they also fail to see that punishment is essentially contested.

is the core justification of legal punishment. In practice a defense attorney or judge wanting to justify, for example, acquittal for a crime lacking intent might argue just that: the point of punishing is to mete out justice, and where a wrong was done unintentionally, justice is not served by inflicting punishment.

But both competing principles are so deeply embedded in our practice and in our language that this argument, and in general the strategy of declaring legal punishment to accord with one principle and all aspects diverging from this principle no longer to be part of the practice, is unlikely to prevail as a rule. I believe we are stuck with our essentially contested practices. But rather than lament our fate of eternal disputation, I prefer to marvel at our ability to share a form of life that accommodates difference. For the fact that it is essentially contested might explain the persistence of a practice about which we are so deeply divided. The practice of legal punishment, just like the word "punish," gives expression to different voices while remaining a shared practice.^[35] We are pulled toward two irreconcilable poles, both of which are incorporated into one practice. To say that legal punishment is essentially contested is to say that it is legitimate, or that there are good reasons, to be drawn to either pole. Both utilitarians and retributivists accept, if grudgingly, that actions guided by the principle of the other really are part of the practice.

It might be that an essentially contested practice persists precisely because it accommodates competing views to which we are strongly drawn. It does not accommodate conflicting principles by having each share in a division of justificatory labor; or by giving due weight to each principle—for example, by treating retributivism as a constraint on a utilitarian calculation; but by availing itself of an understanding of the practice as on the one hand utilitarian through and through, on the other hand retributive through and through. But because

[35] Cf. Gallie's discussion of the concept of "champion" in "Esentially Contested Concepts."

both accounts of the practice are possible, neither is thoroughly persuasive. Neither a purely utilitarian nor a purely retributive practice of punishment would be acceptable to many of us. That it encompasses both principles makes the practice acceptable.

Suppose it is true that practices such as punishment are essentially contested in the strongest sense—"necessarily or inevitably contested." Of what use would knowledge of this be to those within the practice who must decide whether to do this or that? It would seem that practitioners are being told that there are, and always will be, good reasons to take either of two conflicting actions. Is there any value in knowing that not a single, self-consistent principle, but conflicting and at times irreconcilable principles, are immanent in (some of) our practices?

In one sense, gaining knowledge that a practice is essentially contested leaves the practice unchanged—two conflicting principles are still immanent in it; all that is different is that we now recognize this. But there is a sense in which this knowledge affects the practice. W. B. Gallie argues that though knowing a concept is essentially contested means we must live up to the fact that it's impossible to find a general principle for deciding which of two contested uses is best, it also means there are "intellectually respectable conversions," to be distinguished from those of "a purely emotional, or a wholly sinister kind."^[36] Our knowledge might lead to a "marked raising of the level of quality of argument in the disputes of the contestant parties." Gallie suggests also that knowledge that a practice is essentially contested lets us exclude the "intrusions of lunatic voices."^[37] Frank Michelman argues that awareness of fundamental contradiction is liberating:

lt is after all a crucial point about "the fundamental contradiction" that it is a form of order, not chaos. It signifies

[36] Gallie, "Essentially Contested Concepts," pp. 184–85.

[37] Ibid., pp. 188–89. Some of us might think this a very suspect goal: who gets to identify and exclude the "lunatics"?

not random confusion—as such ineffable, indescribable, unexaminable—but a formation of thought that can be named, perceived, cognized. And while surely there is nothing potentially enabling about the coming to consciousness of random disorder … appreciation of a hitherto denied contradiction can be emancipating.^[38]

Both Gallie and Michelman are, I think, right to see an advantage in knowing that conflicting principles are immanent in a practice. The alternative against which we must judge the value of this knowledge is that we reject the demand that we be guided by any principle(s) and instead simply muddle through. Gallie and Michelman suggest that at the very least this knowledge focuses our debate. It also lets us see the cost there sometimes is of advocating one principle over the other. If the utilitarian and the retributivist can be persuaded that the practice of which they speak is essentially contested, they may shed a dogmatism that keeps them from acknowledging arguments with which they might concur were they not so steadfastly determined to consider only utilitarian or only retributive considerations. For example, with respect to the problem of whether we should legalize crack, the utilitarian who limits herself to the empirical issue of what the consequences of legalization would be fails to see the rather compelling point that perhaps we should want to make crack use illegal regardless of the consequences, because we regard crack use as wrong. Similarly, the diehard retributivist who takes the position "justice at all costs" fails to see that sometimes such a position is irresponsible: at times we simply must take consequences into account.^[39]

Much of the debate regarding legal punishment has been characterized by self-conscious dogmatism. Bentham and

[38] Frank I. Michelman, "Justification (and Justifiability) of Law in a Contradictory World," in J. Roland Pennock and John W. Chapman, eds., Nomos , vol. 28: Justification (New York: New York University Press, 1986), p. 86.

[39] As the retributivist Duff writes: "I would act unjustly to prevent the heavens falling" (Trials and Punishments , p. 298).

Posner both explicitly reject the idea that the criminal law embodies moral values or social judgments about right and wrong^[40] and that we punish for reasons other than social utility or economic efficiency. Their accounts are unpersuasive precisely because they refuse to acknowledge a truth about our practice: we do insist on punishing the rapist or murderer, not on the basis of some calculation of social utility (either of punishing the particular rapist or murderer, or of as a rule punishing rapists or murderers), but because we regard rape and murder as abominable, as deserving our condemnation. Similarly, Kant, in his discussion of moral punishment, explicitly rejects not merely utilitarian but any consequentialist considerations. Kant's account, too, is unpersuasive, insofar as it refuses to acknowledge a truth about our practice: sometimes we must take consequences into account if we are to maintain the institutions that let us mete out justice at all.

We might think, given my view that both utility and retribution are principles immanent in punishment, that we should say there is no single practice of legal punishment, but a mixture of distinct practices, of retributive punishment and utilitarian punishment; and that what I offer in chapter 4 is an immanent criticism not of legal punishment but of retributive legal punishment. But in maintaining that legal punishment is an essentially contested practice, I mean that it is a single practice that has conflicting principles. The judge who refuses to accept a plea bargain on the principle of justice is not practicing "retributive punishment" but simply "punishment" as he understands it. The suspect brought into the station house to be questioned by the police and eventually to be tried before a judge, convicted, and sentenced to prison experiences one practice. The prosecutor, judge, parole officer, sentencing commission members, and prison warden all engage in one and the same practice; but there are conflicting interpretations

[40] For Bentham, there is no such thing as right apart from that which is good or that which maximizes pleasure and minimizes pain. Cf. chapter 3, section 1.1, above.

of what that one practice is. Similarly, readers often disagree about their interpretation of a novel; but we say, not that each interpretation is of its own novel, but that they are different interpretations of one and the same novel.

By my declaring that the practice of punishment is essentially contested, that it has not one essential purpose but, rather, a set of not always consistent purposes, we might think that I am defending a combination of retributive and utilitarian approaches. But that is not my position. I defend the retributive interpretation of the practice, while acknowledging that the practice has immanent in it conflicting principles. The fact that a novel has more than one interpretation, and that I know this, needn't deter me from advancing a particular interpretation of the novel which I think is best. Similarly, that I acknowledge legal punishment to be a practice with competing interpretations, none being "correct," needn't deter me from favoring one in particular, and not merely on a whim, but because I have good reasons for thinking it is the best.

I hear and empathize with the cries of injustice and oppression of the marginalized who are punished for violating laws they don't regard as their laws; I would be outraged by the punishment of a person who lacked the mental capacity to know what he was doing in committing a crime; I am troubled by the use of plea bargains that fail to acknowledge the crime committed and instead profess to punish crimes not committed. I see hypocrisy in using the criminal justice system to keep a person behind bars not because of what he has done in the past but because society fears what he may do in the future. These judgments both draw on and ground my understanding of why we punish, an understanding that, while but an interpretation, I believe provides the best account of, and makes coherent, the complex but connected actions and institutions which constitute our practice of legal punishment.

To declare that legal punishment is an essentially contested practice needn't be completely to undermine our convictions and commitments, either to utilitarian, or, in my case, to retributive accounts of the practice. It is, rather, to acknowledge

what seems to me an unavoidable, though perhaps disappointing,truth about the human condition: our practices lack absolute grounds, and their purposes can be established only by interpretation, about which we may disagree.^[41] Given the strategic advantage for someone wanting to justify doing y for reason x , of being able to say we already do y for reason x , and so long as we are aware of the need sometimes to ask why we have our practices at all, to leave open the possibility of radical criticism, immanent criticism is a strategy well worth pursuing. But in a practice acknowledged to be essentially contested in the strong sense of necessarily or inevitably contested, the honest and perspicacious immanent critic can hope only to draw us to one of two legitimate poles, to win this or that battle, but never to prevail.

The observation that legal punishment is essentially contested is not merely a meta-theoretical, detached, empirical assessment of and prediction concerning discourse about punishment. It is meant to shape that discourse. I believe that the retributive account for which I have argued is stronger for being informed by this observation. By acknowledging that as a complex practice legal punishment incorporates conflicting values and fulfills competing purposes, the retributivist is forced to adjust her conception of the ideals of the practice so that it more adequately reflects the facts. By taking seriously the problems prosecutors, trial court judges, and other practitioners actually confront, the retributive immanent critic may have to rethink her ideals—"Do I really want to insist that we abolish plea-bargaining because it violates the retributive ideal, even though the practitioner insists that doing so will either lead to the collapse of the criminal justice system or else necessitate a clandestine system of bargaining that might

[41] Cf. Charles Larmore, Patterns of Moral Complexity (Cambridge: Cambridge University Press, 1987), p. xi: "[W]e do best to see morality, at its deepest level, as a motley of ultimate commitments. As a result we should acknowledge that moral conflict can be ineliminable."

be even more objectionable?" But rather than giving up her ideals in the face of a nonconforming actuality, the retributive immanent critic who accepts the nonfoundationalist view of practices and who is therefore no longer dogmatically committed to her retributive ideals can with adjusted ideals persevere in the face of a less than ideal reality.