[Draft of November 8, 2002; not for citation; comments welcome]
I hope that most philosophers of law agree that stringent conditions need to be satisfied before the state is justified in enacting criminal laws and punishing persons who violate them. In this paper, I will examine one possible such condition: the criminal law should be used only as a last resort. What does this principle mean, how should it be applied, and what reasons might we have to accept it? These are some of the issues I propose to address. In Part I, I will examine some preliminary matters and place the last resort principle in a larger theoretical context. In Part II, I will discuss a number of possible interpretations of this principle. In Part III, I will explore some reasons to believe this principle is plausible. In Part IV, I will endeavor to apply this principle to the difficult case of drug offenses. If my reasoning is sound, a theory of criminalization should probably include the (appropriately construed) last resort principle. But this conclusion will prove disappointing to those who hope to employ the principle to bring about fundamental change in the substantive criminal law. I will argue that the principle may not help to reverse the growth of the criminal law to any degree that could not be achieved more directly and less controversially by other principles that a theory of criminalization is generally thought to include. Unless we reject other parts of conventional wisdom about crime and punishment, the application of a last resort principle is unlikely to bring about the sweeping reforms that theorists might have anticipated. But at least some of this conventional wisdom can be challenged, and the last resort principle might have a very limited role to play in criminal law reform after all.
Some philosophers of law have maintained, explicitly or implicitly, that the criminal law should be used only as a last resort.1 They have neglected, however, to provide a detailed analysis of what such a requirement involves.2 But a few preliminary matters are clear. The application of a last resort principle to the criminal law would be an exercise in what might be called idealized criminal theory. The project is aspirational in spirit, attempting to describe the criminal law we would prefer to have---the criminal law as it ought to be, rather than as it actually is. No one can seriously maintain that the criminal law incorporates a last resort principle at the present time.3 Nor can anyone reasonably claim that the common law provides much support for this principle.4 Anglo-American jurisdictions create offenses so casually and routinely that the criminal law is likely to be employed as a first resort.5 If the last resort principle turns out to be sound, its primary use will be to enable commentators to critique existing law, providing ammunition to those who hope to retard the seemingly inexorable trends toward enacting too many criminal laws and punishing too many persons.
Hopefully, legislators will heed these criticisms.6 By almost any measure, the United States today has enacted too many criminal laws and inflicts too much punishment. Commentators are reluctant to estimate the number of criminal offenses, but the figure is well into the hundreds of thousands.7 No living person can begin to describe more than a tiny fraction of the criminal laws that currently exist.8 And these numbers (whatever exactly they may be) are bound to rise, as criminal statutes are easily enacted but seldom repealed. Data about the extent of punishment are more reliable. At the present time, approximately 2.1 million Americans are incarcerated in jails and prisons, and 6.5 million are under the supervision of the criminal justice system---which includes probation and parole.9 These statistics are unprecedented in the history of democratic governance, and should shock our sense of justice.10 But no believes that these trends will be reversed in the near future. No existing political constituency (except for a handful of academics) favors a reduction in the size and scale of the criminal law.
A solution to these problems requires (inter alia) a theory of criminalization---a set of conditions that must be satisfied before the state may enact a statute that subjects offenders to punishment.11 Should the last resort principle be included among these conditions? In my previous work, I have defended a theory that imposes severe constraints on the authority of the state to enact criminal offenses.12 I will briefly summarize this theory, although it is crucial to recognize that a last resort principle can stand on its own merits; one need not accept my views to believe that the principle should be included in a theory of criminalization. My account of criminalization builds on the deficiencies of its competitors.13 The theory actually in place under existing constitutional law in the United States is woefully inadequate.14 Most laws burden (that is, limit or restrict) liberties. When the constitutionality of these laws is challenged, courts respond by dividing liberties into two kinds: fundamental and non-fundamental. 15 The constitutionality of legislation that restricts a fundamental liberty is subjected to “strict scrutiny” and is evaluated by applying the onerous “compelling state interest” test. Virtually all criminal laws, however, limit non-fundamental liberties, and are assessed by applying the much less demanding “rational basis” test. Under this test, the challenged law will be upheld only if it is substantially related to a legitimate government purpose. The legitimate government purpose need not be the actual objective of the legislation---only its conceivable objective.16 Since only those laws that lack a conceivable legitimate purpose will fail this test, courts almost never find a law to be unconstitutional when non-fundamental liberties are restricted.17 As a result, the state needs only some conceivable legitimate purpose to enact the great majority of criminal laws on our books today. Persons who break these laws can be punished simply because the state has a rational basis to do so.
On the other hand, the state needs an extraordinary rationale to punish persons who exercise fundamental liberties. The Constitution effectively precludes the state from criminalizing travel, prayer, or political speech, for example. Outside the narrow range of fundamental liberties, however, it is only a slight exaggeration to say that the state can decide to criminalize almost anything. A hypothetical case may help to demonstrate the extent of state power in the criminal arena---and the potential injustice of this power.18 Suppose that legislators become alarmed by the fact that too many persons are unhealthy and overweight. Initially, they decide to facilitate the efforts of consumers to eat a better diet by enacting legislation requiring distributors of fast foods to display nutritional information on their packaging. If the constitutionality of this law were challenged, it would seem appropriate for courts to defer to legislators by invoking the rational basis test. Suppose, however, that legislators came to believe (as is probably the case) that better information would have little impact on the problem of obesity. Imagine that they decided to prohibit---on pain of criminal liability---the consumption of designated unhealthy foods. Suppose that sausage were placed on this list. Once again, the rational basis test would be applied to assess the constitutionality of this law. This hypothetical crime is almost certainly constitutional, since the liberty to eat sausage does not seem to qualify as fundamental. The state has an uncontested interest in protecting health, and it is at least conceivable that proscribing the consumption of sausage would bear a substantial relation to this interest. Admittedly, many foods are more detrimental to health than sausage, and not all sausages are especially detrimental to health. But the fact that a criminal law is underinclusive and/or overinclusive is not regarded as a constitutional impediment under the rational basis test. 19 In other words, a statute may proscribe some instances of conduct that do not contribute to the statutory objective, and need not proscribe each instance of conduct that does contribute to the statutory objective.
What is remarkable about the foregoing approach is its complete indifference to the distinction between criminal and noncriminal legislation. It is one thing for noncriminal laws that burden non-fundamental liberties to be evaluated by the rational basis test. But it is quite another when criminal legislation is assessed by that same standard. The criminal law is different---importantly dissimilar from other kinds of law. The extraordinary procedural protections surrounding the criminal sanction are sensible only on the assumption that the criminal law is unlike other bodies of law.20 What is so distinctive about the criminal law? Later I will argue that the criminal law is different in that it subjects persons to state punishment. Punishments violate rights in the absence of a compelling justification. Contemporary constitutional law provides an inadequate theory of criminalization because it fails to provide a justification sufficient to override these valuable rights.
The deficiencies in this theory point the way toward a better account of criminalization. Why not require the state to have a compelling interest for each criminal law it enacts?21 The standards applicable to infringements of fundamental rights (like speech) should be invoked whenever persons become subject to punishment.22 This theory requires the law in question to be necessary to achieve a compelling government purpose. In other words, the government’s purpose must be essential, and the law must be the least restrictive means to attain that purpose. To qualify as the least restrictive means, the law must be narrowly tailored to serve the compelling state interest. The requirement of narrow tailoring has two dimensions. First, criminal laws should not be overinclusive, proscribing instances of conduct beyond those that serve the compelling state interest.23 Next, criminal laws should not be underinclusive, and must apply equally to each instance of conduct the state has the same compelling interest to proscribe.24 The state must treat us as equals in protecting our interest not to be punished; it should not punish some while sparing others if it has the same compelling reason to punish both. Of course, this theory cannot be implemented without criteria to decide which state interests are compelling; attempts to identify these interests are bound to generate enormous dispute.25 Moreover, since the law must be necessary to achieve the compelling government purpose, the state must show that its objective would be more difficult to attain without resorting to punishment. This latter requirement, it would seem, expresses the last resort principle. A criminal statute cannot be necessary to accomplish a purpose if other means could do so more easily.
Although the foregoing theory of criminalization seemingly includes a last resort principle, our decision to adopt this principle depends mostly on how it should be interpreted and applied. One would expect the implementation of this principle to require a laborious, case-by-case determination of the relative advantages and disadvantages of criminal and noncriminal approaches to given problems. A few matters, however, seem capable of resolution a priori. For example, can our current practice of creating a series of lesser-included offenses survive the last resort principle?26 Suppose that a defendant cannot possibly violate statute X without also violating statute Y; he commits Y whenever he commits X. Under these circumstances, how can offense X be necessary to achieve a compelling government purpose? As we have seen, the last resort principle is satisfied when the statutory objective would be harder to attain without punishing the conduct proscribed by X. But this conduct is already punished; no one would escape punishment if offense X did not exist. As long as Y exists, how can X be said to be necessary?
Of course, conduct X could still be punished more severely than conduct Y, even if X and Y no longer exist as distinct crimes. Increased punishments for the more serious conduct could be imposed through sentencing provisions, rather than by convicting persons of separate offenses.27 Little, then, would change in actual criminal practice---except that defendants could no longer be charged and convicted of both X and Y. At the present time, lesser-included offenses give prosecutors enormous leverage in securing guilty pleas from defendants. Prosecutors often agree to drop one charge in exchange for a plea of guilty to the charge that remains. I am not unmindful of the advantages that such provisions secure. But it is hard to see how these advantages could be preserved within a theory of criminalization that satisfies the exacting conditions I have described.28
Apart from these kinds of situations, however, an empirical task of Herculean proportions would be needed to implement a last resort principle in the criminal law. My central project is to decide whether such a project is worth undertaking, and whether it offers a realistic hope of reversing our tendency to criminalize too much and to punish too many. Before addressing this topic directly, I want to make four additional preliminary observations about the last resort principle itself.
First, how should this principle be conceptualized within a theory of criminalization? A comprehensive theory will include an exhaustive list of positive and negative reasons---reasons in favor of enacting criminal offenses, and reasons against doing so. Most theorists have focused on the former considerations---reasons to criminalize. Joel Feinberg’s seminal work on the moral limits of the criminal law is the most familiar example. His several liberty-limiting principles each begin: “It is always a good reason in support of a proposed prohibition that…”29 Of course, a criminal prohibition may not be justified even though it satisfies a liberty-limiting principle that Feinberg endorses.30 He claims that the decision not to criminalize conduct that satisfies a liberty-limiting principle should be “determined by such practical matters as the use of available resources, court facilities, police time, enforcement costs, effects on individual expectations, and the like.”31 These latter “practical matters”---included among the reasons not to criminalize---receive far less attention from Feinberg, presumably because they are less interesting to a philosopher of law. Questions about police resources, for example, while unquestionably important to criminologists, do not seem to raise concerns that are central to Feinberg’s project to probe the moral limits of the criminal law.
Although no one should insist on too sharp a dichotomy between positive and negative reasons in a theory of criminalization, the last resort principle should probably be categorized among the latter.32 That is, we should probably test the last resort principle by assuming that a given type of conduct satisfies whatever positive conditions are needed to justify the imposition of criminal sanctions. The conduct in question, for example, should be assumed to be wrongful, cause harm, involve culpability, and the like. Even when each of these positive conditions obtains, however, there still may be persuasive reasons not to criminalize. The fact that the criminal law is not the last resort is the reason not to criminalize I will explore here.33
Next, I will assume that we are able to draw any crucial distinction on which the application of the last resort principle depends. The principle invites us to suppose that the state has a number of possible means (to which it might “resort”) to attain its ends. Some of these means involve the criminal sanction, and others do not. The last resort principle offers guidance about which means to choose. Therefore, interpretations of the last resort principle presuppose an ability to distinguish criminal from noncriminal approaches to given problems. This distinction can be very difficult to draw.34 A fine that is disproportionate to the injury inflicted, I am sure, is an alternative to incarceration as a mode of punishment; it is not an alternative to punishment. The last resort principle, as I construe it here, offers no advice about what kind of punishment to prefer.35 It tells us that punishment (of any kind) should not be imposed when nonpunitive alternatives are better in attaining the objective of the legislature. As we will see, however, the crucial contrast---between alternative modes of punishment and alternatives to punishment---can be elusive.36 The concept of punishment, like most concepts in ordinary language, is vague and allows for borderline cases. Perhaps we must eventually resort to stipulation in deciding whether some novel way to deal with a social problem amounts to punishment. In such cases, it will be hard to decide whether the criminal law is employed as a last resort.
Moreover, I will assume that the last resort principle is a part of a theory of criminalization, and not a component in a theory of sentencing. The difference, as I understand it, is crucial. We might believe that the decision to punish a particular defendant should be a function of his peculiar circumstances. Perhaps we have reason to think that we can attain our goals more effectively by cautioning him. Or perhaps we believe that he will disregard our warnings, so punishment is needed to achieve our objectives. Punishment, it might seem, should only be inflicted on given offenders as a last resort. This way of understanding the last resort principle has some plausibility, and might help to alleviate the problem of imposing too much punishment. It would not, however, address the problem of enacting too many criminal laws, and is not the interpretation I adopt here. Punishment is an option for the state only when conduct has been criminalized. I propose to construe the last resort principle as applying to this prior decision---whether to make punishment an option. It does not render assistance to sentencing authorities about whether to punish conduct that has already been criminalized, but offers guidance to legislators about whether conduct should be criminalized in the first place.
Finally and perhaps most importantly, we should not take the last resort principle too literally. We cannot really believe that the criminal law should be a last resort in addressing social problems in the same sense that we might believe that war, for example, should be a last resort in resolving international problems. I have contended that we suffer from rampant overcriminalization. But we should not pretend that this is the worst affliction that can befall a state; we could come to utilize modes of social control that would make the criminal law seem benign by comparison.37 Such means of social control are a staple of science fiction.38 Even when they are feasible, some devices never occur to us, and pass beneath our radar screen. Imagine a system of social control in which persons were permitted to smoke cigarettes if they agreed to accept a stigmatizing brand on their foreheads.39 Such options should probably not be conceptualized as a rival form of the criminal law; they seem to involve alternatives to punishment altogether.40 If we are assessing the justifiability of a possible criminal prohibition against smoking and incorporate the last resort principle into our theory of criminalization, it is doubtful that we would need to include such alternatives in our deliberations. Any inclination we may have to embrace a last resort principle for the criminal law derives from our assurance that such options are not serious candidates for implementation.41
With these four preliminary observations out of the way, the fact remains that no theorist has explicated the last resort principle in any detail. I have suggested that it provides a reason not to criminalize even when the positive conditions in a theory of criminalization are met. Beyond this observation, what does the principle mean? How should it be applied? Under what circumstances, if any, should we accept it? Does it create any serious constraints on the imposition of the criminal sanction? I now turn to these matters.
In this Part I will attempt to interpret the last resort principle. I will suppose that an interpretation casts a principle in its best light; it helps to explain why theorists might be attracted to it.42 But despite my efforts to construe the last resort principle as plausible, I will have difficulty finding an interpretation that is likely to be helpful in retarding the trend toward overcriminalization I described above. Any such interpretation will have to reject other parts of conventional wisdom about crime and punishment that are more widely accepted than the last resort principle itself.
A superficial interpretation of the last resort principle might be as follows: If noncriminal alternatives are preferable to the criminal law in attaining the legislative objective, the former should be employed. I describe this interpretation as superficial because, as so construed, the principle says nothing distinctive about the criminal law, and seems equally applicable to any purposeful endeavor. In this respect, the principle is comparable to claims like “the criminal law should not be used if it is not effective [in controlling conduct],”43 or “punishment ought not to be inflicted [where it] would produce more evil than the offence would.”44 Presumably, these kinds of principles apply with equal force to all goal-directed activities; it is hard to see why any means should be employed that is known to be inferior to an alternative, ineffective, or counterproductive.45 These principles appear to be little more than requirements of practical reasoning. If a carpenter wants to drive a nail, why would he knowingly adopt a strategy that will not succeed, would do more harm than good, or is worse than a means he rejects? I assume that the last resort principle is not simply a requirement of practical reasoning.46 A sensible interpretation of this principle must explain why it has a special (although not necessarily unique) application to the criminal law. This difficulty can be overcome by construing the last resort principle as a tie-breaker. According to this simple modification, the principle states that a criminal law should not be enacted when other means are equally effective at attaining the legislative objective. Even when all other things are equal, alternatives to the criminal law should be preferred.
Other difficulties of interpretation and application are not so easily resolved. We cannot begin to implement the last resort principle (as so construed) without identifying the objective of penal legislation. Unless we understand the nature of this objective, we are in no position to decide whether given alternatives are better or worse at attaining it. This issue plunges us directly into some of the deepest quagmires of criminal theory. All philosophers agree that the criminal law has an objective, but disagree radically about what that objective is. Suppose we believe that the sole objective of the criminal law is to prevent (or, in Feinberg’s words, eliminate or reduce) whatever conduct has been criminalized.47 This belief gives rise to what I will call the preventive interpretation of the last resort principle: the criminal law should be used only as a last resort to prevent given kinds of conduct. If noncriminal means to prevent the conduct in question succeed as well or better, the criminal sanction should not be employed.
Of course, we know that the criminal law cannot really eliminate; some criminal conduct will persist, whatever the law may say. In light of this realization, we may be tempted to endorse the following two premises. First, there is an acceptable level of criminality for any given offense. Second, the objective of the criminal law is to reduce the incidence of the conduct in question to whatever level is deemed acceptable. If these premises were true, we could try to ascertain whether we could produce acceptable levels of given kinds of conduct without utilizing the criminal sanction. If we discovered that we could do so, the preventive interpretation of the last resort principle would entail that no criminal offense should be enacted.
Jonathan Schonsheck appears to subscribe to these two premises, and construes the last resort principle accordingly. In attempting to limit the scope and reach of the criminal sanction, he asks: “Is there some technique of social control which will be successful in reducing the incidence of that action to an acceptable rate---but which is less intrusive, less coercive than a criminal statute?”48 If the answer is affirmative, Schonsheck insists that the alternative technique should be used, and the criminal law should not.
Unfortunately, both of the foregoing premises should be rejected---and Schonsheck’s interpretation of the last resort principle along with them. First, there is no “acceptable” rate of criminality in any sense that would be useful to a criminal theorist. Not surprisingly, Schonsheck offers no guidance as to how we might decide whether a given rate of criminality is acceptable. Even though we can be confident of failure, the state should strive to eliminate core crimes like murder and theft---to reduce their incidence to zero.49 In fact, a number of commentators have appealed to this aspiration in attempts to draw the elusive boundary between the criminal and civil law. According to this train of thought, the civil law, unlike the criminal law, should aim only for optimal rather than for total deterrence. In other words, the civil law should price, while the criminal law should prohibit.50
To be sure, economists speak meaningfully of an optimal rate of crime in at least two contexts: when referring to ideal expenditures on law enforcement, or when trying to identify appropriate levels of punishments. Consider the first of these contexts. No one recommends that a society should invest all of its resources in crime prevention. At some point, the marginal cost of deterring one additional theft is greater than the occurrence of the undeterred theft. When this point is reached, it is inefficient to allocate further resources to theft prevention. It does not follow, however, that the thefts not prevented are somehow “acceptable.” They remain unacceptable because we would seek to prevent them were greater resources to become available. Economists also speak of an optimal rate of crime when deciding how severely given offenses should be punished. For at least two reasons, the state must be careful not to set punishments too high. First, the state should not depart from a principle of rank ordering in implementing a principle of proportionality.51 More serious offenses should not be punished with greater severity than less serious offenses, even if deviations from this principle proved effective in preventing crime. Next, punishments must not be so severe that they deter people from engaging in productive and beneficial activities.52 If punishments for the offense of careless driving were exorbitantly high, risk-averse people might be deterred from driving at all, since they could only guarantee that they would not drive carelessly only by abstaining from driving altogether.53 But the existence of these two contexts does not presuppose that there is an acceptable level of crime in a society. Since no rapes should be tolerated, for example, we cannot decide whether noncriminal means of reducing rape would produce an acceptable rate. Therefore, the second premise in the above argument should be rejected as well.
But we need not believe that there is an acceptable level of crime to salvage the preventive interpretation of the last resort principle. All we need to believe is that the function of the criminal law is to reduce the incidence of the conduct to be criminalized. The criminal law is only one of many possible devices to prevent a state of affairs from obtaining, and it is always a contingent matter whether a noncriminal alternative would succeed as well or better. As long as we construe the purpose of the criminal law wholly in instrumentalist terms, we must remain open to the possibility that some other instrument will attain this purpose more effectively.
This interpretation of the last resort principle seems plausible. As Neil Komesar points out, evaluating one means of social control without simultaneously evaluating its competitors distorts policy analysis in the same way that ignoring opportunity costs distorts economic analysis.54 “Evaluation” is typically explicated in preventive terms, and several competitive means of prevention are available. Some of these means involve legal regulation, and some do not; of those that do, some of these means involve criminalization, and others do not. Feinberg includes “withholding licenses, withdrawaing professional certification, refusing to enforce certain kinds of contracts, job dismissals by public agencies, suspending governmental subsidies or financial support, child custody discontinuances, and so on.”55 Michael Bayles distinguishes six distinct legal techniques in addition to penal sanctions. He describes grievance-remedial, administrative-regulatory, public benefit-conferral, public burden-imposing, private-arranging, and compulsory treatment strategies.56 Peter Schuck includes informal social norms and markets as the main non-legal means to “organize human effort, advance well-being, and maintain social order.”57 Andrew Ashworth lists “morality, social convention, and peer pressure” as the modes of social control that rival law.58 And, as Herbert Packer reminds us, we always have the option of doing nothing.59 How should we choose which mode of social control to employ? If any combination of these competitors works as well or better than the criminal law in reducing the incidence of the conduct in question, the preventive interpretation of the last resort principle entails that punitive sanctions should not be employed.
The preventive interpretation suggests that we compare two jurisdictions that differ in only one respect: the first includes an offense proscribing a given kind of conduct, whereas the second employs noncriminal means of prevention. Since real experiments of this sort are difficult or impossible to perform, philosophers must rely on thought-experiments. In the relevant thought-experiment, we compare two possible worlds that only differ in that the first contains an offense to reduce the occurrence of a given kind of conduct, while the second relies solely on alternative strategies. We then identify the possible world in which less of the conduct occurs. If fewer (or as many) instances of the type of conduct occur in the latter world, the criminal sanction would not be justified.
But this interpretation is problematic. The first problem I will cite is perhaps the least interesting philosophically, but among the most difficult to solve in practice. How can we hope to identify the possible world in which less of the conduct to be criminalized occurs? How confident must we be that an alternative to the criminal law would be more effective as a preventive?60 We might be tempted to say, with Peter Alldridge, that the last resort principle requires that the criminal law may be used only “when all other methods of the legal regulation of the phenomenon in question have been canvassed and found wanting.”61 But this statement cannot be taken literally; the last resort principle is not intended to produce paralysis. Alternative strategies are almost never exhausted. In most cases, we can at least imagine rival devices that might succeed but have not been attempted. In addition, we can always provide some reason to believe that we did not try hard enough when a given alternative was deemed to have failed. In short, unless we know where the burden of proof lies in judging success or failure, and how that burden can be discharged, we cannot begin to implement the last resort principle.
Even if this difficulty can be overcome, two additional problems with the preventive interpretation are seemingly insurmountable. First, we should challenge the assumption that reducing the incidence of given types of conduct is the sole objective of the criminal law. Obviously, the preventive interpretation of the last resort principle is jeopardized if the criminal law has central functions other than reducing crime. Many, and probably most theorists assign a different function to the criminal law. Consider, for example, the extreme version of retributivism forcefully defended by Michael Moore. According to Moore, “the only function of the criminal law is the achievement of retributive justice.”62 Crime prevention is utterly irrelevant on this model, since retribution “is the intrinsic good that is the function of Anglo-American criminal law.”63 Moore argues that “the achievement of retributive justice gives a legislature good reason to criminalize all immoral behavior, for this is the only way the good of retributive justice can be achieved.”64 Since criminalization is the only way to achieve retributive justice, Moore has no need to invoke a last resort principle.65 For those whose conduct is immoral---culpable wrongdoers---the criminal law is the only game in town; no alternative to punishment can possibly do as well or better.66 The criminal law is not the last resort; it is the only resort.
Without going as far as Moore in contending that the prevention of given kinds of conduct is wholly immaterial to the justification of criminal legislation and the rationale of punishment, we should agree that the criminal law has important objectives other than prevention. Feinberg has persuasively argued that punishment has an expressive function. More specifically, punishment “is a conventional device for the expression of attitudes of resentment and indignation, and of judgments of disapproval and reprobation, on the part either of the punishing authority…or of those ‘in whose name’ the punishment is inflicted.”67 To be sure, theorists have challenged Feinberg’s views about exactly what it is that punishment expresses, and how the expressive and hard treatment components cohere in a single rationale of punishment.68 Hopefully, these difficult issues need not be resolved. For present purposes, we need only make the tentative assumption that an adequate account of the nature and justification of punishment must include both components.69 Thus I will suppose that a normative defense of punishment must defend our practice of subjecting offenders to hard treatment as well as to censure.70 In what follows, I will make no attempt to defend this assumption. My immediate aim is to explore its implications for the last resort principle. Later, I will consider the consequences of rejecting or qualifying this assumption.
Although expressive views are typically advanced as theories of punishment, they also have profound implications for the content of the substantive criminal law itself---for issues of criminalization. The reason should be clear.71 Punishments must be justified, and justified punishments must be deserved. Persons can deserve censure and condemnation only if their conduct merits these responses.72 If so, expressive theories provide guidance about what conduct we ought not to criminalize, while also helping us to decide what offenses we should enact. Dan Kahan has argued that sanctions other than incarceration for serious crimes tend to be resisted by the public because they fail to express stigma.73 When white-collar offenders are fined rather than imprisoned, for example, citizens think the state does not regard their offenses as sufficiently serious.74 But the content of the message to the public would be even more misleading if this conduct were not criminalized at all. The failure to proscribe and punish spousal rape, for example, would indicate that this behavior is less deserving of censure than other kinds of nonconsensual sex.75 Thus I conclude that expressive theories pertain just as much to criminalization as to punishment.
Expressive theories are incompatible with the preventive interpretation of the last resort principle. This version of the principle states that noncriminal alternatives should be employed when they are as good or better at preventing given kinds of conduct. Once we understand the criminal law to have both preventive and expressive functions, we need to provide a new interpretation of the last resort principle. Even though alternative modes of social control may do a better job in reducing the incidence of criminality, they may fail to achieve an indispensable objective of the criminal sanction: expressing the censure conveyed by punishment. To decide whether alternatives to the criminal law are equally effective, we would have to evaluate not only their ability to reduce crime, but also their efficacy as expressions.76
The following example may be helpful in recognizing this point. Suppose that criminal sanctions for domestic violence were not effective in reducing the incidence of batteries that take place behind closed doors. Imagine that empirical research demonstrated that a criminal law against domestic violence led significant numbers of battered wives to be less inclined to notify the police because they did not want their husbands to be arrested and prosecuted. Suppose the state could adopt noncriminal means to prevent domestic violence that were just as effective as an approach that invoked the criminal sanction. This empirical finding, although interesting and important for many purposes, would not provide a decisive objection to criminalizing acts of domestic violence.77 The point of this law is not only to reduce the incidence of the proscribed conduct, but also to condemn those who engage in it. If we used these empirical studies to conclude that laws against domestic violence should not have been enacted, we would frustrate an objective of the legislature by failing to retain our conventional methods of condemning persons who are guilty of spousal abuse.78
I suspect that theorists who endorse the last resort principle would concur with my judgment about domestic violence. Since they have not explicated the principle in detail, however, it is hard to be certain. But commentators who provide examples of the kinds of criminal laws they believe would be jeopardized by the last resort principle indicate that they have different kinds of targets in mind. Since they tend to confine the last resort principle to conduct they almost certainly do not believe to merit censure at all, it is doubtful that they would invoke the principle to demand the repeal of criminal penalties for harmful conduct they find reprehensible. Schonsheck, for example, explicitly applies his principle only to drug offenses and to paternalistic interferences like seat bet requirements.79 Clearly, one need not appeal to a last resort principle to appreciate the controversy that surrounds these offenses; various kinds of objections have been brought against them.80 Since these laws are problematic on many grounds, it is hard to know to what extent our reservations are derived from the last resort principle rather than from other sorts of normative considerations.
Of course, the last resort principle could be applied to the expressive function of the criminal law as well as to its preventive function. According to this suggestion, conduct should be criminalized only when no alternative device conveys condemnation as well or better. If punishment were the only way to express censure, we could agree with Moore’s conclusion that criminalization would be the sole means to attain the function of the criminal law.81 On this assumption, the last resort principle would never provide a good reason to repeal (or not to enact) an offense designed to express condemnation. The last resort principle would not become false; it would become trivial and unimportant in a theory of criminalization, and could not retard the trend toward enacting too many offenses and punishing too many persons. Its application to the criminal law would achieve nothing that was not already accomplished by insisting that crime and punishment be expressive--a less controversial (although hardly indisputable) piece of conventional wisdom about our criminal justice system.
Punishment, however, is not the only possible way to stigmatize. It is a contingent fact about contemporary societies that the hard treatment imposed on persons who violate the criminal law is the conventional means by which expressive functions are attained. One can certainly imagine rituals other than hard treatment by which condemnation could be expressed.82 In closely-knit societies, devices other than punishment (understood to include hard treatment) may have been sufficient to stigmatize.83 Most sentencing theorists, however, oppose as demeaning those alternative modes of punishment that clearly express stigma but do not involve deprivations---such as “shaming sanctions.”84 But it is important to remember that these sanctions are alternative modes of punishment, not alternatives to punishment. If these alternative modes of punishment are deemed unacceptable, it is hard to believe that stigma could be expressed effectively if punishment were abandoned altogether. In the diverse liberal societies of today, punishment is perhaps uniquely suited to condemn. No formal mechanism seems able to convey censure as directly and forcibly. Although we should remain open-minded that other forms of social control may do a better job than the criminal sanction at preventing given forms of conduct, it seems less likely that alternative means of social control may be more effective at expressing condemnation.
More precisely, it seems less likely that alternative means of social control that are acceptable may be more effective at expressing condemnation. It is crucial to keep in mind that the foregoing conclusion---the last resort principle is of little significance in the criminal law---is defensible largely because we would never dream of invoking noncriminal alternatives that might serve as well or better at preventing given kinds of conduct while expressing condemnation. The fact that the criminal law strikes most of us as the only resort in expressing censure may be an indication of the moral progress we have made. Perhaps, then, my conclusion thus far should be expressed as a challenge. If the last resort principle is to be helpful in retarding the trend toward overcriminalization and expressive theories of criminalization and punishment are correct, we must identify an acceptable means to stigmatize whatever conduct the principle precludes us from criminalizing. Although I am skeptical, we should not be dogmatic about whether such a device can be found.
But a different possibility is worthy of consideration. We might try to salvage the preventive interpretation of the last resort principle, and thus its potential usefulness in a theory of criminalization, by attempting to limit the scope of expressive theories.85 Although it is clear that persons who commit the core offenses of the criminal law qualify for censure, one might contend that the objective of some criminal laws is and ought to be wholly preventive. Of course, the failure of these offenses to include an expressive dimension is precisely what many commentators deem to be objectionable about them.86 As I have indicated, however, existing criminal law has already expanded far beyond its core, and no longer seems to require moral blame as a condition for liability.87 Theorists need not resign themselves to this development; they can help to retard it without relying on expressive theories by including a last resort principle in their theory of criminalization. This solution seems especially tempting in the case of many white-collar crimes. Consider, for example, the provisions of the mail and wire fraud statutes that punish political powerbrokers who use their influence to obtain government jobs for their friends.88 It is hardly obvious that this conduct merits censure.89 Should these provisions therefore be repealed? Theorists who hope to preserve the expressive function of the criminal law will probably answer in the affirmative. Commentators who are less confident about expressive theories of criminalization and punishment might be able to appeal to a last resort principle to reach the same conclusion. At the very least, the last resort principle seems far more plausible when applied to such examples than when invoked against proscriptions of domestic violence or other cases of core criminality.
Arguably, then, a preventive interpretation of the last resort principle might be salvaged---for that class of criminal laws that are not designed to give rise to censure or condemnation, but are wholly preventive. At least some such laws may exist.90 I am unsure whether the distinction between offenses that are designed solely to prevent and those that (also) involve censure or condemnation parallels other distinctions frequently drawn---between malum in se and malum prohibitum offenses, for example, or between regulatory and non-regulatory offenses. It is hard to know how closely these distinctions overlap, partly because no theorist has ever succeeded in providing an adequate account of them.91 Regardless of how these latter contrasts are ultimately drawn, however, it is unlikely that they will correspond closely to the distinction between offenses that are designed solely to prevent and offenses that (also) involve censure or condemnation. Many (apparent) regulatory offenses---such as those pertaining to the handling of radioactive materials---are sufficiently dangerous to have an expressive dimension. The same is true of many (apparent) malum prohibitum offenses, such as driving on the wrong side of the road. Hence it is probably best not to equate the contrast I am drawing with these older, more familiar (albeit notoriously unclear) distinctions. In the absence of a convenient term to describe the distinction I have in mind, we might simply say that some offenses are wholly preventive in function. When the justifiability of these kinds of offenses is called into question, the last resort principle, understood in preventive terms, seems plausible and important.
Thus far, I have described a possible avenue to salvage the preventive interpretation of the last resort principle. Some offenses, I will tentatively suppose, are wholly preventive. But a second and much more straightforward challenge to the preventive interpretation remains. The noncriminal means to reduce a crime like rape---such as courses in self-defense for potential victims---can always supplement, but need not replace the criminal law. In fact, it is barely possible to imagine a world in which only criminal prohibitions were used to reduce the incidence of conduct like rape. What would such a world be like? Would we not impose tort liability on rapists? Would we not educate students to believe that rape was wrongful? The criminal law can hardly be expected to be an effective deterrent unless it works in tandem with other mechanisms of social control.92 In other words, the relevant thought-experiment must include a third possible world for our consideration: a world in which both criminal and noncriminal means are used. In almost every case, this third possible world, which combines criminal and noncriminal strategies, would probably reduce the incidence of the conduct in question more effectively than either of the previous two.93 No sensible interpretation of the last resort principle would call for the repeal of a criminal law that contributes to the prevention of harmful conduct, just because criminal and noncriminal alternatives would be equally effective when used alone. When preventing given kinds of conduct is imperative, we have good reason to utilize all means at our disposal, including both criminal and noncriminal strategies.94 If I am correct, the last resort principle, interpreted preventively, offers little hope of alleviating the problem of overcriminalization with which we began.