Rethinking the Wrongness Constraint on Criminalisation

Orthodox thought holds that criminalisation should be subject to a wrongness constraint: that is, that conduct may be criminalised only if it is wrongful. This article argues that this principle is false, at least as it is usually understood. On the one hand, the wrongness constraint seems to rest on solid foundations. To criminalise conduct is to facilitate its condemnation and punishment; to coerce citizens against it; and to portray it as wrongful. All of these actions are presumptively impermissible when the conduct that they target is not wrongful. On the other hand, the article argues that the wrongness constraint is nevertheless unsound. Although it is presumptively impermissible to criminalise non-wrongful conduct, this might yet be permissible, given sufficient countervailing reasons. Moreover, there are realistic cases – specifically, some cases of over-inclusive criminalisation – in which such countervailing reasons exist.

especially important role in determining the proper scope of the criminal law. Thus, one leading work on the subject says that the wrongness constraint is 'implicitly assumed in debates about criminalisation', 2 while another describes it as the proper 'starting point' for such debates. 3 Despite this widespread endorsement, however, the case for the wrongness constraint has never been examined in detail. This should lead us to question whether the endorsement is deserved. What are the supposed normative foundations of the wrongness constraint? And how, if at all, do they support that principle?
The standard answer to these questions points to the criminal law's distinctive use of condemnation and punishment. Criminal sanctions are condemnatory and punitive; thus, to criminalise conduct is to subject it to potential condemnation and punishment. Common sense suggests that we may only condemn and punish those who have done something wrong -or, perhaps more to the point, that we shouldn't condemn and punish those who haven't done anything wrong. Seemingly, then, common sense demands that only wrongful conduct may be criminalised. For many people, this attractively simple line of argument is all we need to establish the wrongness constraint. As Antony Duff writes, for example: Or, in the more succinct words of Victor Tadros: Two of the central functions of criminal law are condemnation and punishment of offenders. But condemnation and punishment are justified only if the person has done wrong. Therefore, it is wrong to criminalize conduct that is not wrong. To do so would be to warrant condemnation and punishment of conduct that is not wrong. 6 At its simplest, then, the case for the wrongness constraint can be reduced to two claims. The first is a claim about the nature of criminalisation. As Tadros puts it: To criminalise conduct is to warrant the condemnation and punishment of that conduct. The second is a normative claim about condemnation and punishment: Legislators may warrant the condemnation and punishment of conduct only if that conduct is wrongful.
In this article, I assess this case for a wrongness constraint on criminalisation. I begin in Section II by clarifying the first of the two claims just identified: that to criminalise conduct is to warrant its condemnation and punishment. In Section III, I then turn to the second of these claims: that legislators may warrant the condemnation and punishment of conduct only if that conduct is wrongful. I examine three possible arguments for this view. These arguments hold that to criminalise conduct is to facilitate its condemnation and punishment; to coerce citizens against it; and to portray it as wrongful. All of these actions are impermissible, the arguments hold, when the conduct that they target is not wrongful. I conclude that each of these arguments is plausibly sound. In all three cases, criminalisation does involve an action of the relevant type, and that action is plausibly impermissible when the conduct targeted is nonwrongful. Therefore, to criminalise non-wrongful conduct is likewise impermissible.
Ultimately, however, I conclude that the wrongness constraint is false, at least as it is usually understood. In this context, 'impermissible' is best understood in its presumptive sense: Although there are strong reasons against criminalising non-wrongful conduct, this might yet be permissible, all things considered. In Section IV, I explain why this is so. First, I argue that the normative principles examined in Section III are best understood in this presumptive sense: It is permissible to breach them, given sufficient countervailing reasons. Second, I argue that there are realistic cases in which legislators have such reasons. Specifically, some types of over-inclusive criminalisation are probably justified, despite targeting nonwrongful conduct. Thus, while legislators have strong reasons against criminalising non-wrongful conduct, it is not the case that conduct may be criminalised only if it is wrongful. We should therefore rethink our commitment to the wrongness constraint.
Before we begin, a couple of quick clarifications. First, in the context of the wrongness constraint, 'wrongness' simply means moral wrongness. We could thus re-phrase the principle as follows: Conduct may be criminalised only if, morally, we ought not to engage in that conduct. 7 Second, we must assume here that criminality does not entail moral wrongness. The wrongness constraint is meant to guide legislators; however, it cannot do so if making conduct criminal necessarily makes it morally wrongful. Advocates of the wrongness constraint must thus presume that citizens have no general moral obligation to comply with the law. Of course, they needn't claim that criminalisation cannot affect moral wrongness: Surely, conduct can sometimes become wrongful as a result of its criminalisation. The point is that it doesn't necessarily do so. In short, the wrongness constraint requires moral wrongness, which is presumed to be independent of criminal wrongness. I conclude in Section V by revisiting this point in more detail.

II. CONDEMNATION, PUNISHMENT AND CRIMINALISATION:
CLARIFYING THE RELATIONSHIP What does it mean to say that criminalisation 'warrants' condemnation and punishment? In short, this is to describe some of the legal liabilities that criminalisation creates. By criminalising conduct, legislators make citizens liable to be convicted and sentenced for that conduct -liabilities that expose citizens to potential condemnation and punishment. These liabilities are most easily understood in terms of their correlative powers. Conviction and sentence are 7 Thus understood, the wrongness constraint doesn't imply anything about what moral wrongness consists in, or about what sub-set of moral wrongs the criminal law should target. It simply claims that moral wrongness is a necessary condition for permissible criminalisation. On why the principle is best understood in this way, see J. Edwards and A.P. Simester, 'Wrongfulness and Prohibitions', Criminal Law and Philosophy 8 (2014): 171-186. powers in the technical sense that their use changes citizens' legal position, including in relation to official condemnation and punishment. Normally, citizens have legal rights against these things: Officials are under legal duties not to condemn and punish citizens. But conviction and sentence release officials from these duties: Officials are permitted to condemn and punish offenders. Moreover, these officials will often be legally obliged to exercise this permission, because of the duties attached to their official roles. We can capture this combination of permissions and duties by saying that officials are authorised to condemn and punish offenders.
Courts authorise several forms of official condemnation and punishment by convicting and sentencing citizens. Take conviction first. Most obviously, a guilty verdict in a criminal trial is itself a form of public condemnation. But more significantly, convictions go on one's criminal record. Arguably, criminal records have punitive as well as condemnatory effects: For example, they permit others to discriminate against one in ways that would otherwise be wrongful. 8 Similar things are true of sentencing. Again, judges' remarks when passing sentence are often condemnatory in themselves. But more importantly, sentencing offenders authorises their punishment. Once offenders are sentenced, they may be treated in ways that they would otherwise have rights against: For example, their property may be confiscated, they may be restrained and locked away, or forced into unpaid work. Finally, many think that punishment is itself condemnatory. As Duff says, criminal punishments are not mere penalties: They have an expressive, condemnatory dimension. 9 In light of all this, conviction and sentence may each be seen as both condemnatory and punitive in character.
Under what conditions may courts exercise these condemnatory and punitive powers? The short answer is that criminal guilt is typically both necessary and sufficient for conviction and sentence: Courts may convict and sentence a defendant if and only if the defendant is guilty of the crime alleged. Moreover, criminal guilt usually obligates courts to exercise these powers: Courts have duties 8 Most notably and obviously, in decisions about employment. But criminal records can also be harmful in many less obvious ways. to convict and sentence the guilty. Strictly speaking, this last statement isn't always true. Under some conditions, courts might be able to grant guilty defendants a 'sentence' that is functionally identical to an acquittal. 10 Normally, however, criminal courts are both permitted and obliged to convict and sentence guilty defendants. To say that criminalisation 'warrants' condemnation and punishment is thus arguably apt.
It's important to notice, however, that 'guilt' is used here in a procedural sense. The powers to convict and sentence are not conditional upon whether one has actually committed a crime; rather, they are conditional upon guilt as established through the criminal process. Such guilt can arise only in the context of criminal proceedings and in only a limited number of ways. Put simply: The defendant must either plead guilty to the relevant charge or be found guilty beyond a reasonable doubt. 11 This is not to say that actual criminal conduct is irrelevant to the legitimacy of conviction and sentence. To convict someone of a crime they didn't commit is clearly wrong, on some level. 12 The point is rather that the legal powers to convict and sentence are not conditional upon actual criminal conduct. What matters to the legal permissibility of conviction and sentence -what legitimises condemnation and punishment in the eyes of the law -is criminal guilt, in the procedural sense.
This point is important, because criminal guilt and actual criminal conduct are logically independent: neither entails the other. Clearly, criminal conduct isn't sufficient for criminal guilt. For example, one's criminal conduct might never be detected, or one might benefit from attrition in the criminal process. Perhaps less obviously -but as we'll see, more importantly for our purposes -neither is criminal conduct necessary for criminal guilt. Clearly, one can plead guilty to a crime 10 In English law, for example, courts may sometimes grant guilty defendants an absolute discharge: Powers of Criminal Courts (Sentencing) Act 2000, ss. 12 and 14. Roughly, a conviction followed by an absolute discharge does not count as a conviction for any purposes beyond the immediate context of the proceedings in which the offender is convicted. 11 This is arguably an over-simplification. Some informal sanctions for criminal conduct -such as cautions and fixed penalty notices -might be seen as condemnatory and punitive in character, despite not being conditional upon a plea or finding of guilt. For ease of exposition, I leave these complications aside here. For discussion of the relative advantages and disadvantages of such informal sanctions, see e.g. A. Ashworth and L. Zedner, 'Defending the Criminal Law: Reflections on the Changing Character of Crime, Procedure, and Sanctions', Criminal Law and Philosophy 2 (2008): 21-51, pp. 24-28. 12 It might even be wrong on a legal level: for example, as a civil wrong for which one can claim compensation. Thanks to an anonymous reviewer for pointing this out. without having committed it. 13 But one can also properly be found guilty without having committed the crime alleged. This possibility arises from the standard of proof in criminal trials. If the evidence against one leaves no room for reasonable doubt, then one is legally guilty. This holds even when a doubt that is unreasonable, from the fact-finder's perspective, in fact discloses the truth. 14 Again, one is legally liable to be convicted and sentenced in such cases, despite not having committed a crime.
Informally, then, criminalising conduct can plausibly be described as warranting the condemnation and punishment of that conduct. However, it's important to be clear about what this means. To criminalise something is to give courts the power to convict and sentence those who are guilty of it -along with a duty to exercise these powers in most cases. Convicting and sentencing someone means, among other things, permitting and obliging officials to condemn and punish that person. For short, let's describe conduct that is subject to this kind of liability as punishable. To criminalise conduct, let's say, is to make it punishable in this sense. If the case for the wrongness constraint is to succeed, one must therefore show that conduct may be made punishable only if it is wrongful. Our task in the rest of this article is to assess this claim.

III. THREE ARGUMENTS FOR THE WRONGNESS CONSTRAINT
Why might we think that conduct may be made punishable only if it is wrongful? In this Section, I examine three possible answers to this question. All three answers have a similar structure: By making conduct punishable, legislators perform an action that is impermissible when the conduct being targeted is not wrongful. Specifically, to make conduct punishable is to facilitate its condemnation and 13 This is not a trivial risk: Modern criminal justice systems typically offer defendants numerous powerful incentives to plead guilty. Sentence discounts for guilty pleas, alongside practices of plea, charge, and fact bargaining, inevitably incentivise even innocent defendants to avoid going to trial. 14 To see this, notice that the law itself sometimes gives content to the 'reasonable doubt' standard. And on no conception of this content does 'proved beyond reasonable doubt' mean the same as 'true'. This is particularly obvious when the standard is made to depend on fact-finders' mental states: for example, on whether they are 'sure' that the defendant committed the crime alleged. But the point also holds for other elaborations of the standard. For discussion of the various legal meanings of 'reasonable doubt', see P. Roberts and A. Zuckerman, Criminal Evidence, 2nd ed. (Oxford: Oxford University Press, 2010), ch. 6.4. punishment; to coerce citizens against it; and to portray it as wrongful. All three arguments are plausibly sound. We thus have a strong case for thinking that it's impermissible to make nonwrongful conduct punishable. To see why, let's examine the arguments in turn.

A. The Argument from Facilitation
The most obvious argument for the wrongness constraint points to the permissibility of actual condemnation and punishment. Conduct may be made punishable, we might think, only if we may actually condemn and punish it; and conduct may be condemned and punished only if it is wrongful. 15 This argument relies on two normative claims. The first is a claim about permissible condemnation and punishment: one may condemn and punish others only for their wrongful conduct. The second is a claim about the permissibility of making conduct punishable: Legislators may make conduct punishable -and thereby facilitate the condemnation and punishment of that conduct -only if it would be permissible to condemn and punish it.
The first of these claims is highly plausible. To see why, consider first condemnation. Once we appreciate what condemnation involves, it quickly seems obvious that its permissibility depends on the wrongness of the conduct being condemned. To condemn others for their conduct is to express disapproval of that conduct, and to signal the appropriateness of certain reactions towards them: for example, attitudes of resentment. 16 Consequently, condemnation also tends to tarnish the target's reputation: It amounts to 'moral defamation' if not appropriately directed. 17 To inflict such responses on others seems clearly wrong, unless they've done something to merit them. And it's hard to imagine anything that might merit these responses besides the targeted person's wrongdoing. 18 Thus, the first claim just identified seems straightforwardly true of condemnation: only wrongful conduct may permissibly be condemned.
The first claim is also highly plausible in relation to punishment: That is, the permissibility of punishing others plausibly depends on the wrongness of the conduct for which they're being punished. One explanation for this is that, as we've already noted, punishment itself is usually thought to be condemnatory. Thus, if wrongness is a constraint on the permissibility of condemnation, then it's also a constraint on the permissibility of punishment. 19 However, it's also independently plausible that permissible punishment requires wrongdoing. While people disagree about the precise conditions for permissible punishment, most current theories take it for granted that only wrongdoers may be punished. 20 To see why this assumption seems so plausible, we can briefly examine two questions: first, why punishment requires justification; and second, why wrongdoing is probably necessary for any such justification to succeed.
Punishment requires an especially strong justification, because it harms people in ways that would otherwise infringe their rights. 21 Indeed, criminal punishment appears to infringe our rights in especially dramatic ways. For example, consider imprisonment, fines, and community service. Outside the context of (deserved) punishment, these things would clearly infringe our rights: They are materially identical to abduction, robbery, and enslavement. On reflection, then, the state seems to breach its duties to its citizens when it 19 This observation raises the issue of non-punitive penalties: sanctions that involve similar 'hard treatment' to punishment, but without its condemnatory aspects. Would a legal regime backed by penalties, rather than punishments, also have to be subject to a wrongness constraint? I leave this question aside here. punishes them: Punishing them is at least presumptively impermissible. 22 It follows that, if punishment is to be permissible, then it must also be justified: There must be factors counting in its favour that make it permissible, despite its ordinarily infringing rights. So why might wrongdoing be necessary to justify punishment, in this sense?
The answer is that punishment is justified only if it is deservedand one deserves punishment only if one has done wrong. 23 Deserved punishment is justified because the deserving have forfeited their rights against being punished: While punishing them would usually infringe their rights, it doesn't do so when the punishment is deserved. 24 It's widely assumed that desert is the only way that one can forfeit one's rights against being punished. Thus, undeserved punishment is always presumptively impermissible. Furthermore, it's widely assumed that no other justification can outweigh this right against undeserved punishment. This reflects the more general principle that one may not harm the morally innocent in order to promote the greater good. 25 On these assumptions, not only does punishing non-wrongdoers infringe their rights; it likely always infringes them unjustifiably. The first normative claim identified above therefore follows.
Consider next the second normative claim on which this argument relies: that it's permissible to facilitate the condemnation and 22 Throughout this article, I use the term 'presumptively impermissible' to refer to actions that one ought not to perform, in the absence of sufficiently strong countervailing reasons. This definition includes, but is not limited to, actions that we have duties not to perform. Breaching one's duties is presumptively impermissible in a special sense: Only a limited range of reasons can justify breaching one's duties. For the classic account of this idea, see J. Raz, Practical Reason and Norms (Oxford: Oxford University Press, 1975), ch. 2. 23 I use 'desert' here in the negative sense outlined in this paragraph: If one deserves punishment, then one lacks a right against being punished. Note, however, that negative desert needn't be explained in retributive terms: One can consistently believe both that negative desert exists, and that only consequentialist aims can justify punishment. For accounts of this sort -which also try to explain why wrongdoing should be necessary for negative desert -see e. 24 Note that the language of 'forfeiture' here is merely a helpful way of describing the normative effects of desert. It is a separate question whether the idea of forfeiture can also explain those effects, or serve to justify punishment. For a defence of the view that it can, see C.H. Wellman, 'The Rights Forfeiture Theory of Punishment', Ethics 122 (2012): 371-393. 25 For some people, it follows that all punishment aimed at consequentialist goals is impermissible. Others reply that consequentialist punishment can be permissible, so long as it is suitably constrained. punishment of conduct only if it would be permissible to condemn and punish that conduct. Again, this claim is plausible, for at least some relevant senses of 'facilitate'. Typically, by making conduct punishable, legislators cause some of that conduct to be condemned and punished. And if it's impermissible to do something, then it's also impermissible to cause it. 26 Of course, criminalisation is not a direct cause of condemnation and punishment. So perhaps it's more accurate to say that, by making conduct punishable, legislators procure or incite other state officials to inflict such sanctions. But as we've seen, to make conduct punishable is to procure its punishment in an especially direct way: If a defendant is guilty, then officials are both authorised and obliged to punish them. Legislators have little room to deny responsibility for conduct that they are, in effect, demanding from officials. 27 Thus, if it's impermissible to condemn and punish those who've done nothing wrong, then it's likely also impermissible to facilitate such responses in this way.

B. The Argument from Coercion
A second possible argument for the wrongness constraint points to the coercive, liberty-restricting character of criminalisation. By making conduct punishable (this argument goes), legislators coerce citizens against that conduct: They restrict the liberty to engage in it. But whether such coercion is permissible depends on the liberty being restricted. If conduct isn't wrongful, then citizens should be free to engage in it. Thus, it's impermissible to make non-wrongful conduct punishable -and the wrongness constraint therefore follows. 28 To assess this argument, we need to know two things.
Mainly, we need to know whether it relies on a sound account of permissible coercion: that is, whether it's true that coercing others against non-wrongful conduct is impermissible. Firstly, though, we need to know whether it's correct to describe criminalisation as 'coercive'. How, exactly, does making conduct punishable coerce citizens against that conduct?
Criminalisation is aptly described as coercive because it interferes with citizens' choices in a particular way. Specifically, by criminalising conduct, legislators issue a threat: If citizens are guilty of engaging in that conduct, then the state will condemn and punish them. Such threats exert a kind of pressure on citizens' choices that is characteristic of coercion. By issuing them, legislators pressurise citizens not to engage in the conduct criminalised: They make that conduct more difficult for citizens to choose. Of course, there are several ways of making options more difficult to choose, not all of which are coercive. For example, legislators might offer citizens a reward for not choosing the relevant option. But this result is surely coercive if achieved by means of a threat: that is, by claiming that, if citizens choose the option, they will be made significantly worse off. 29 Since condemnation and punishment normally infringe citizens' rights, criminalisation thus amounts to a coercive threat of such sanctions.
Such interference with others' choices is also what makes coercion morally troubling. By coercing others against an option, one restricts their freedom to choose that option. Threats might also be troubling for other reasons: for example, the fact that they involve a commitment to carrying them out. 30 But what makes them troubling as a form of coercion is their unwelcome impact on the threatened person's choices. 31 Interfering with others' choices in this way is potentially wrong, because it infringes their autonomy. There is some range of options for which we are each entitled to control over our choices among those options. And within this range, there are 29 This line between threats and offers is usually thought to mark the boundary between conduct that generally is and generally isn't coercive. limits to the ways in which others can permissibly interfere with our choices. 32 Put simply, there are some options that we should be able to choose, free from certain types of pressure from others. Coercive behaviour is impermissible if and when it infringes this right.
Do legislators act impermissibly, in this sense, in threatening to condemn and punish citizens for non-wrongful conduct? It's plausible that they do, for two reasons. First, the option that legislators are thereby interfering with is among those that citizens should remain free to choose. If conduct is non-wrongful, then the choice of that conduct falls within the protected range just mentioned. Wrongful conduct might not fall within this range -at least, not always. Plausibly, we lack a general right against others that they refrain from interfering with our choices to do wrong. 33 By contrast, this right does plausibly extend to our choices to act permissibly. If it's permissible for us to do something, then we should also be free to choose to do it; which is to say, we have rights against others that they refrain from certain interferences with that choice.
Second, and following on from this, threats of condemnation and punishment are among the impermissible types of interference. Such threats inherit their wrongness from the wrongness of what's threatened: If it's impermissible to do something to someone, then it's also impermissible to threaten coercively to do it to them. 34 As we've seen, it's impermissible to inflict undeserved condemnation and punishment on others. Hence, it's also impermissible to threaten others with such sanctions. This isn't necessarily a complete account of the wrongness of threatening, or of coercive conduct more broadly. 35 But it's a highly plausible part of such an account: If we're entitled to freedom from  35 For instance, some people dispute whether the wrongness of threatening requires that the conduct threatened be wrongful. Blackmail is often given as a counter-example: see e.g. Lamond, 'The Coerciveness of Law' (n. 30), pp. 48-51. Since this dispute is irrelevant to our enquiry, I set it aside here. certain interferences with a choice, then the prospect of wrongful sanctions on the relevant option is surely among these. Since legislators threaten such sanctions by criminalising non-wrongful conduct, it follows that it's impermissible for them to do so.
At this point, one might object that criminalisation does not necessarily have coercive effects. On the view just described, the wrongness of coercion is grounded in its effects on people's choices. However (one might observe), criminalising conduct needn't interfere with citizens' choices to engage in that conduct. For the criminal law's threats can fail, in several ways. 36 First, these threats often aren't communicated successfully: citizens tend to be unaware of the precise content of the substantive criminal law. 37 Second, these threats may not be credible, even when they are communicated: Some criminal laws are never enforced, or are enforced only sporadically. 38 And third, because of attrition in the criminal process, even regular enforcement doesn't guarantee credibility. If one can be confident of avoiding detection, arrest, or prosecution for one's conduct, then the criminality of that conduct needn't affect one's choices.
We can give two responses to this objection. First, for the argument from coercion to succeed, it would be enough that criminalisation has some coercive effects. True, criminalising a given type of conduct won't exert pressure on every choice to engage in that conduct. But even the mere risk of sanctions will be enough to exert some pressure on some people's choices. 39 Admittedly, however, this is only a partial response to the present objection. For this objection shows that criminalisation needn't have any coercive effects. If citizens are totally unaware of a crime's existence, or if they 36 By 'fail', I mean fail to exert some pressure on citizens' choices. I assume that such pressure can be wrongful even if it doesn't lead to actual deterrence of non-wrongful conduct. 38 It's hard to say exactly how often this occurs -partly because we know so little about the exact scope of the criminal law. However, there are well-known examples of criminal laws falling into disuse, whether through gradual neglect or deliberate choice by officials. A good example from English law is the crime of blasphemy, which was finally abolished by s. 79 of the Criminal Justice and Immigration Act 2008 after almost a century of disuse. 39 The same is true of legal regulation generally: At most, it tends to coerce. Sometimes it will coerce successfully; sometimes it will merely be intended to coerce; sometimes it will risk coercion as a sideeffect. Whether any particular coercive measure can be justified depends on the relative difficulty of justifying each of these things. See e.g. Lamond, 'The Coerciveness of Law' (n. 30), pp. 52-56. think it will never be enforced, then their choices will remain totally unimpeded. It's debatable whether any such cases exist in realityand if they do, how frequently they occur. 40 But regardless of this, we should at least account for their possibility. Ultimately, the coercive effects of criminalisation are contingent upon independent factors, such as enforcement practices. Thus, even if criminalisation typically has such effects, it needn't always do so.
A second response to the objection addresses this issue: It's sometimes wrong to make coercive threats, even when doing so does not have coercive effects. Here are just two ways in which criminalisation might be wrong qua coercive threat, despite not pressurising citizens' choices. First, making threats might be wrong because it is inchoate to actual coercive effects. By enacting new crimes, legislators are often attempting to coerce citizens against the targeted conduct. And even in the absence of coercive intent, criminalisation risks coercive effects -precisely because failures of coercion will be due to factors that are independent of decisions to criminalise, and hence beyond legislators' control. Second, special normative considerations arguably apply to coercion by law. For example, perhaps citizens should be able to rely on the substantive criminal law as an authoritative guide to what they may and may not do. If so, then even completely unknown and unenforced criminal laws affect citizens' choices in a sense: their choices would be pressurised, were they to use the law as they are entitled to use it. 41 By criminalising non-wrongful conduct, then, legislators make an impermissible coercive threat. Perhaps there are cases in which this is false -in which legislators expect the relevant crime to go completely unenforced, and to do nothing to guide citizens' conduct. But since there would be little reason to enact such crimes, we can expect these cases to be exceptional. For most imaginable cases of criminalisation, the argument from coercion is plausibly sound. (this argument goes), legislators portray that conduct as wrongful. Obviously enough, it seems wrong to portray conduct as wrongful if it's not actually wrongful. Hence, to criminalise non-wrongful conduct also seems wrong. To see whether this argument succeeds, we need to consider two points. First, we need to consider what's wrong with portraying non-wrongful conduct as wrongful. But second, we need to consider whether criminalisation actually does portray conduct as wrongful. Let's begin with this second issue. What, if anything, does criminalising conduct say about its wrongness?

C. The Argument from Communication
The answer must be that criminalisation portrays conduct as morally wrongful: It communicates a judgement that one ought not to engage in the conduct criminalised. 42 Two promising lines of argument can be given to support this view. The first relates to the nature of legal wrongs generally, and relies on two assumptions that are widespread in legal discourse. The first is that crimes are legal wrongs: That is, that criminalising a given type of conduct creates a legal duty not to engage in that conduct. 43 The second is that legal duties are putative moral duties. As some would put it, the law claims legitimate authority: It claims the power to make our compliance morally obligatory. 44 On these assumptions, by criminalising conduct, legislators at least imply that that conduct is morally wrongful. For since crimes are legal wrongs, they are also putative moral wrongs: In the eyes of the law, it's wrong for citizens to engage in the conduct that they target.
A second promising argument for this view points to the specific institutional features of the criminal law. According to this argument, many of these features make sense only if we see crimes as putative moral wrongs. Some of these features are part of the very nature of the criminal process. Criminal conduct plays a justificatory role in this process, at every stage: It is criminal conduct for which one is arrested, charged, found guilty, and ultimately punished. 45 42 Duff gives the most explicit endorsement of this claim: see generally Duff, Answering for Crime Others are pervasive features of current practice. For example, the doctrines of the substantive criminal law often cast wrongness in an inculpatory role, and the absence of wrongness in an exculpatory role. 46 The seriousness of offences is a key factor in sentencing decisions. 47 And officials' behaviour often suggests that they regard crime as something to be avoided: 'crime prevention' and 'law enforcement' are considered to be good uses of public resources. Against this institutional background, one might think, criminalisation cannot help signalling that the targeted conduct is morally wrong.
Despite the plausibility of these arguments, however, they remain open to challenge. To repeat, the conclusion is that criminalisation communicates a judgement about the moral wrongness of the targeted conduct: It sends a message that that conduct is wrong. But how exactly could criminalisation send this message? The answer can't be that legislators intend to portray conduct as wrongful by criminalising it, for that is by no means necessarily true. 48 Rather, the answer must be that criminalisation has a certain symbolic meaning: We share the understanding that criminalisation conveys a judgement that the targeted conduct is wrongful. However, whether such a shared understanding exists is an empirical question. Thus, the above arguments may lead us to infer that criminalisation portrays conduct as wrongful. But we must bear in mind that this is just an inference, which further evidence might ultimately disprove. 49 Consider next the claim that it's impermissible to portray nonwrongful conduct as wrongful. On one level, this claim seems obviously true. To portray conduct as wrongful when it's not is to tell a lie; and lying is generally wrong, because it risks deceiving others. 50  conduct. By sending the message that conduct is wrongful, legislators can lead people to believe that that conduct is wrongful -and hence, can make choosing that conduct less attractive than it should be. Again, we may question whether criminalisation is necessarily deceptive in this way. Evidence suggests that people aren't so easily fooled. 51 But as with the argument from coercion, such evidence doesn't fatally undermine the present argument. Criminalisation might still lead to some deception. And even when it doesn't, the fact that legislators are attempting or risking deception might be enough to make it impermissible.
One might further argue that such deceptive communication has unwelcome side-effects in the criminal law context. For example, Simester and von Hirsch argue that it can undermine the criminal law's distinct value as a type of regulation. As we've repeatedly noted, the criminal law is distinctive in its use of condemnatory sanctions. But its ability to condemn is undermined when it targets conduct that doesn't actually deserve condemnation: This '[blurs] the moral voice' and 'gunks up the censure machine'. 52 Over time, such deception might even lead to worse consequences: For example, it 'risks undermining the moral authority of the criminal law'. 53 Once again, these are empirical claims that require empirical support. And the available evidence here gives us reason for caution: Whether criminalising non-wrongful conduct weakens the law's moral standing in these ways is probably contingent upon other factors. 54 Nevertheless, the potential for such side-effects shows us that deceptive communication needn't always be a trivial matter. At 51 Contrary to widespread assumptions, it appears that blind obedience of the law is rare. People don't always comply with legal obligations that they don't endorse on content-dependent grounds. And even when they do, their compliance may not be motivated by a belief in the law's moral authority: see e.g. least in those cases where it would risk these side-effects, legislators have relatively strong reason to avoid it.
We began this Section by asking: why might we think that conduct may be made punishable only if it is wrongful? We can now see that there are (at least) three good answers to this question. The arguments from facilitation, coercion and communication all plausibly establish that it's impermissible to make non-wrongful conduct punishable. At first glance, the wrongness constraint on criminalisation seems to follow from these conclusions. As we saw, to criminalise conduct is to make it punishable. So doesn't the discussion in this Section imply that only wrongful conduct may be criminalised? I turn to this question in the following Section.

IV. WHY CRIMINALISING NON-WRONGFUL CONDUCT MIGHT SOMETIMES BE PERMISSIBLE
By making non-wrongful conduct punishable, legislators act impermissibly. But if that's true, how could the wrongness constraint be false? The answer lies in an ambiguity in the language of permissibility -an ambiguity that I've only hinted at thus far. When we describe something as 'wrong' or 'impermissible', we could be saying one of two things. First, we could be saying that it's impermissible, all things considered: That is, that we have decisive reason not to do it, taking all relevant factors into account. Second, we could be saying that it's presumptively impermissible: That is, that we have good reason not to do it, albeit a reason that may be defeated by countervailing factors in some cases. 55 Accordingly, when we say that conduct may be made punishable only if it is wrongful, we could be saying one of two things. We could be making the strong claim that it's impermissible to make non-wrongful conduct punishable, all things considered. Or we could be making the weaker claim that this is presumptively impermissible. This weaker claim leaves open the possibility that legislators are sometimes permitted to criminalise non-wrongful conduct, given sufficient countervailing reasons. In this Section, we'll see that this possibility is a realistic one -and thus, that the wrongness constraint is false, at least as it is usually understood. To demonstrate this, we need to establish two things. First, we need to establish that the normative principles discussed 55 See further n. 22 above on presumptive impermissibility. above are best understood in the presumptive sense, rather than the all things considered sense. We need to establish, that is, that the types of action examined in the previous section are only presumptively impermissible when the targeted conduct is nonwrongful. Second, we need to establish that there are some cases in which legislators are justified in criminalising non-wrongful conduct -in which the reasons in favour of this course of action outweigh the reasons against it. Let's begin with the first of these points.

A. Criminalising non-wrongful conduct is only presumptively impermissible
The normative claims examined in Section III are best understood as claims about presumptive impermissibility. As we saw there, these claims are highly plausible; yet once understood as 'all things considered' claims, they quickly seem highly implausible. To see this, consider first the claim that it's impermissible to facilitate the condemnation and punishment of non-wrongful conduct. According to an all things considered version of this claim, legislators have decisive reason against such facilitation. But if this claim were true, then criminalisation would never be permissible. For criminalisation always and inevitably facilitates the condemnation and punishment of some non-wrongful conduct. Partly, this result is due to the practical realities of criminal justice. Officials inevitably make errors in enforcing the criminal law, meaning that some non-wrongdoers are condemned and punished. However, this result is hard to avoid even at the level of the 'law in the books'. This is because crimes need to be defined so as to apply generally, over whole populations and diverse sets of facts. Such generalised rules tend to be over-inclusive, relative to the wrongs that they target. No matter how much we try to refine the definition of a crime, it will thus tend to catch some people whom we do not wish to condemn and punish as wrongdoers of the relevant kind. 56 More fundamentally, however, criminalisation facilitates some undeserved condemnation and punishment as a matter of law. conviction and sentence is conditional not upon criminal conduct, but upon criminal guilt. And one may plead guilty to a crime, or even properly be found guilty of that crime, despite not having committed it. Thus, even if we could define crimes in such a way that they only caught wrongful conduct, we could not similarly limit the resulting liabilities. Criminalisation inevitably authorises some condemnation and punishment of non-wrongdoers -and hence, is always impermissible, on the all things considered version of the present claim.
Similar objections apply to an all things considered version of the claim about coercion. According to this claim, legislators have decisive reason to avoid coercing citizens against non-wrongful conduct. Again, though, if this version of the claim were true, then criminalisation would never be permissible. 57 Partly, this result is also due to the fact just highlighted: Liability to conviction and sentence is conditional upon criminal guilt, rather than actual criminal conduct. Hence, the threats of condemnation and punishment involved in criminalisation attach to conduct beyond that which is criminalised. In relation to coercion, however, the problem may also run deeper than this. For liabilities to conviction and sentence are not the only liabilities that criminalisation creates. For instance, consider the liability to be prosecuted for the relevant crime. This liability is conditional not upon guilt, but its mere probability. 58 Plausibly, however, if we haven't engaged in wrongful conduct, then we have a right against being formally accused of having done so. This suggests that the creation of liability to prosecution could also amount to an impermissible coercive threat. If this is correct, then again, the present claim would render all criminalisation impermissible.
In making these points, I've assumed that there is no material difference among the different effects of criminalisation -in particular, between its effects on the conduct actually criminalised and its inevitable implications for a wider range of conduct. But perhaps this is false. Perhaps, for example, there is a difference between what 57 Advocates of similar claims about permissible coercion are often explicit in supporting the presumptive version of them. See e.g. Edmundson, 'Is Law Coercive?' (n. 34), p. 93. 58 In England, for instance, all that's required is that the prosecution is more likely than not to succeed, and that the public interest does not speak against it. For critical discussion, see J. legislators intend to do by criminalising conduct and what they do thereby as a side-effect. One might argue that this difference is significant for the permissibility of criminalisation. Admittedly, it cannot be impermissible, all things considered, simply to coerce citizens against non-wrongful conduct or to facilitate its punishment. But these actions are plausibly impermissible, all things considered, when performed intentionally. Thus, there is a moral difference between the actual criminalisation of non-wrongful conduct and the implications that all criminalisation has for such conduct. Only in the former case are the relevant actions intended -and thus impermissible, all things considered. 59 We can give two responses to this objection. First, it's doubtful whether intention can play the normative role that the objection requires of it: namely, rendering it impermissible, all things considered, to coerce citizens against non-wrongful conduct or to facilitate its punishment. For one thing, it's doubtful whether these actions are necessarily impermissible, all things considered, even when performed intentionally. Mightn't they be justifiable if, for example, they were necessary in order to avoid some sufficiently great harm? 60 More fundamentally, however, it's doubtful whether intention can affect the permissibility of actions at all. Exactly how actors' mental states could make their actions harder to justify -that is, could give them stronger reason against performing an action -is notoriously difficult to explain. Indeed, it's especially difficult to explain in relation to collective actors such as states or legislatures, for whom autonomy and individual responsibility are relatively unimportant concerns. 61 While we can't assess these claims fully here, they at least give us reason for caution about the above objection.
The second response to the objection is more decisive: It's not necessarily true that, when legislators criminalise conduct, they intend to facilitate its punishment or to coerce citizens against it. Of course, these things are often true. By criminalising conduct, legislators are often trying to deter citizens from engaging in it, and to 59 Thanks to two anonymous reviewers for suggesting this objection. 60 Some find this view plausible. For example, for many deontologists, it is generally impermissible intentionally to harm others in order to bring about the greater good. However, this constraint does not apply when the harm to be averted thereby is sufficiently grave. For critical discussion, see L. ensure that those who do so are prosecuted and punished. But whether this is true of any particular new crime is a contingent matter. For example, criminalisation might be intended purely as a symbolic gesture or as a way of winning votes. Indeed, legislators might not intend that their new law be enforced at all. 62 Given this, it is also contingent whether, in criminalising non-wrongful conduct, legislators act impermissibly, all things considered. On the present interpretation, coercing citizens against such conduct, and facilitating its punishment, are impermissible in this stronger sense only when they are intended. But it's not necessarily true that, in criminalising non-wrongful conduct, legislators intend such effects. Thus, the present objection cannot save the wrongness constraint, as an all things considered principle.
Similar responses will also apply to other versions of this objection: that is, to other attempts to distinguish the actual criminalisation of non-wrongful conduct from the implications that criminalisation inevitably has for such conduct. On the one hand, if we adjust the relevant normative claims to refer to legislators' mental states, then we face responses analogous to those just given. For example, consider claims referring to legislators' beliefs. It's not necessarily true that, when legislators criminalise non-wrongful conduct, they believe that they are facilitating its punishment or coercing citizens against it. And in any event, the normative significance of such beliefs is doubtful. On the other hand, claims referring to objective factors will fail to distinguish the two kinds of effects. For example, consider claims referring to the evidence available to legislators. According to this evidence, criminalisation clearly has implications not only for the conduct criminalised, but also for a wider range of conduct. Thus, on an all things considered version of such a claim, we would again be left with the implausible result highlighted above: all criminalisation would be impermissible.
All of this suggests that the normative claims about facilitation and coercion cannot plausibly be understood in the all things considered sense. But is the same true of the claim about communication? According to an all things considered version of this claim, legislators have decisive reason to avoid portraying non-wrongful 62 I assume here, along with the anticipated objection, that there is a satisfactory way of determining what legislators collectively intend. This may be false -in which case, the objection fails along with my response.
conduct as wrongful. This claim doesn't suffer from the same problem as the two claims just considered. Only the conduct actually criminalised is portrayed as wrongful through criminalisation; thus, this claim doesn't apply to the wider set of liabilities thereby created, and doesn't have the implausible result that all criminalisation is rendered impermissible. 63 Might it therefore provide a sound basis for the wrongness constraint, as an all things considered principle?
In fact, an all-things-considered version of the claim about communication is also implausibly strong. Telling untruths to others is surely easier to justify than wrongfully coercing them, or wrongfully facilitating their condemnation and punishment. Yet as we've just seen, criminalisation is sometimes permissible despite such effects. Whatever justifies criminalisation in these cases can surely also justify legislators in communicating at least some falsehoods. Indeed, as we noted above, legislators' reasons to avoid such falsehoods can be relatively weak. Portraying non-wrongful conduct as wrongful might sometimes have unwelcome side-effects, such as diminishing the law's moral standing; however, these effects are probably contingent and incremental. In their absence, such communication is wrong mainly because it risks deception -a risk that, in reality, may be relatively small. It's therefore unlikely that legislators' reasons to avoid such communication are necessarily decisive against it. 64 It follows that the arguments examined so far do not suffice to establish the wrongness constraint. The normative claims on which these arguments rely are best understood in the presumptive sense: They establish that it's presumptively impermissible to make nonwrongful conduct punishable. But this claim supports only a weaker principle of criminalisation than the wrongness constraint: that it is presumptively impermissible to criminalise non-wrongful conduct.
This principle leaves open the possibility that it is sometimes permissible to criminalise non-wrongful conduct, all things consideredand hence, that the wrongness constraint is false, as it is usually understood. 63 Even this might not strictly be true. As we just noted, crimes should ideally be general and clear in their definition; but this can be achieved only at the cost of over-inclusiveness. Ironically, then, welldefined crimes might well end up portraying some non-wrongful conduct as wrongful. 64 Compare P.H. Robinson, Intuitions of Justice and the Utility of Desert (Oxford: Oxford University Press, 2013), ch. 10, arguing that the criminal law may sometimes depart from community intuitions about retributive justice, despite the unwelcome effects to which such departures can lead.

B. Justifiably criminalising non-wrongful conduct: the case of over-inclusive crimes
Moreover, there are indeed realistic cases in which it is permissible to criminalise non-wrongful conduct, all things considered. To illustrate this point, I'll examine the case of over-inclusive crimes. Over-inclusive crimes target a genuine underlying wrong: At least some of those who commit them genuinely deserve condemnation and punishment through the criminal law. As the name suggests, however, they are over-inclusive relative to that wrong: Some of the conduct that they catch does not constitute wrongdoing of the relevant kind. Due to constraints of space, we can't hope to establish here that any particular over-inclusive crime is definitely legitimate. But we can establish a more general claim: that some over-inclusion in criminal legislation is probably legitimate. To do this, we can identify some good reasons that legislators might have for enacting over-inclusive crimes. Taken together, these reasons can plausibly justify legislators in making some non-wrongful conduct punishable.
To aid discussion, let's begin with an example of an over-inclusive crime: the crime of sexual activity with a child, under England's Sexual Offences Act 2003. Adults commit the actus reus of this offence if they sexually touch another person, and this other person is under sixteen. 65 This provision targets a genuine underlying wrong: the exploitation of children who are not yet mature enough to make fully informed decisions about sexual activity. 66 However, it is overinclusive relative to this wrong. Some people who are under sixteen are mature enough to make well-informed decisions about sexual activity. And some adults who engage in sexual activity with persons under sixteen do not exploit them. Imagine, for example, an eighteen-year old who has a mutually consensual and non-exploitative sexual relationship with a relatively mature fifteen-year old. Given these observations, this provision apparently falls foul of the wrongness constraint: It criminalises some non-wrongful conduct. If so, its creation was presumptively impermissible. Might it be justified despite this? And if so, how? The answer is that over-inclusion in crimes of this kind can serve valid goals. The most familiar of these goals relate to determinacy. Age-of-consent crimes are over-inclusive relative to the wrong of exploitation that they target. But they are also much more determinate in their scope than crimes defined explicitly in terms of exploitation. This increased determinacy yields several related benefits. First, it makes it easier for actors to predict when their conduct will fall within the scope of the crime. Second, it promotes efficient and consistent decision-making by courts and other officials. And third, it can help to reduce errors in such decision-making. Admittedly, these benefits are of relatively modest weight in the criminal law context. Compared to the avoidance of undeserved punishment, the virtues of determinacy seem modest: There's relatively little value in the efficient, consistent, and accurate imposition of undeserved sanctions. 67 Nevertheless, these are genuine virtues -virtues, indeed, that are commonly identified with the ideals of legality and the rule of law. 68 Hence, they may do at least some modest work in justifying over-inclusive crimes.
More significantly, though, over-inclusion can enhance the preventive potential of criminalisation. Unlike determinacy, prevention can provide relatively strong reasons in favour of over-inclusive crimes. Over-inclusion might enhance the preventive potential of criminalisation in several ways; in the context of child sex crimes, two are particularly important. First, age-of-consent crimes might be a more effective deterrent than crimes explicitly targeting exploitation. By stating clearly that any sexual activity with children is punishable, we can expect them to deter more of the exploitation that they target. 69 Second, age-of-consent crimes cause less secondary victimisation through the criminal process than would a crime based on exploitation. By putting exploitation in issue in criminal cases, legislators would oblige courts to examine intimate 67 For further discussion of all these claims, see Schauer, Playing by the Rules (n. 56), ch 7. 68 The ideal of the rule of law is that the law should be capable of guiding citizens' conduct. As the discussion here suggests, legislators needn't, and shouldn't, give absolute priority to this ideal. But they must give it some weight: Compliance with the rule of law is necessary to maximise citizens' freedom and autonomy under coercive systems like the criminal law. See Raz, The Authority of Law (n. 44), ch. 11. 69 There are good reasons generally to be sceptical about the marginal deterrent effects of changes to the substantive criminal law: see e.g. Robinson and Darley, 'Does Criminal Law Deter?' (n. 37). However, age-of-consent crimes overcome at least one of the obstacles to deterrence that Robinson and Darley identify: Unlike many other crimes, the standards of conduct that they impose are widely known. details of complainants' lives and behaviour: for example, their relative maturity or their conduct towards the defendant. As is well known, putting complainants 'on trial' in this way can seriously harm them. 70 Legislators can avoid this by defining offences in terms of a more easily ascertained fact, like age.
Importantly, this kind of over-inclusion is also relatively easy to justify to potential actors. While age-of-consent crimes do authorise undeserved sanctions, they at least make those sanctions relatively easy to avoid. Of course, these offences do constrain actors' choices in one sense: They restrict some liberties to engage in non-wrongful sexual conduct. As we've seen, this gives us genuine cause for concern. For most people, though, the crime of sexual activity with a child will have only a modest impact on their liberties. And for those whose liberties are restricted, this provision makes their choices very clear: It gives them very clear guidance on when their conduct will be punishable. 71 In short, while the over-inclusion involved in age-ofconsent crimes is not benign, neither is it cynical. Not only does this over-inclusion pursue valid goals; it does so in a way that's relatively fair to those whose liberties are restricted. 72 A possible reply to all of this is that there are ways of achieving these goals that do not involve the criminalisation of non-wrongful conduct. For example, when legislators enact over-inclusive offences, perhaps they could also create 'no wrongness' defences: that is, defences exempting citizens from liability where they have not committed the relevant underlying wrong. In the context of sexual activity with a child, for example, we could allow defendants to avoid liability by pleading that their conduct was not actually exploitative. This solution seems like an attractive compromise. It enables us to preserve the advantages of over-inclusive offences, 70 This effect has most often been discussed in the context of sexual offences, where courts are obliged to ascertain the complainant's non-consent. For discussion, and for critique of the evidential rules governing such cases, see J. Temkin, Rape and the Legal Process, 2nd ed. (Oxford: Oxford University Press, 2002), ch. 4. 71 Provisions like these can also provide another sort of valuable guidance: guidance on how to avoid the wrong underlying the crime. Age-of-consent crimes might provide more effective guidance on how to avoid exploiting children than a crime framed explicitly in terms of exploitation. See e.g. Alexander and Ferzan, Crime and Culpability (n. 1), pp. 310-311; Simester and von Hirsch, Crimes, Harms, and Wrongs (n. 1), pp. 76-77. Whether this is true of any particular crime is a contingent matter. 72 In this respect, age-of-consent crimes differ from many other over-inclusive crimes. Some types of over-inclusive criminalisation are unabashedly cynical, aimed mainly at making it easier to prosecute without thereby making non-wrongful conduct punishable. And if such a compromise option is available, then legislators surely ought to prefer it to an option under which non-wrongful conduct is made punishable. Hence (the reply goes), it remains impermissible for legislators to criminalise non-wrongful conduct, despite their admittedly good reasons for doing so. 73 Unfortunately, however, 'no wrongness' defences are an imperfect compromise. While they will lead to fewer breaches of the normative principles underlying the wrongness constraint, they are also likely to set back the goals that over-inclusive offences pursue. We should expect, therefore, that legislators will sometimes have stronger reason simply to enact an over-inclusive crime, than to compromise by adding a no-wrongness defence. To see why, let's remain with the example of sexual activity with a child. How might legislators set back the goals that are served by the over-inclusive definition of this crime, if they were to introduce a 'no-exploitation' defence?
Consider first the goals of determinacy. The problem here is that, by introducing a 'no-exploitation' defence, legislators would be introducing precisely the kind of indeterminacy that over-inclusion here serves to avoid. This is because, as a result of this defence, exploitation would become a condition of liability for the crime. 74 The law would thus inherit the problems of this concept's indeterminacy -such as inefficiency, inconsistency, and consequent loss of predictability. Admittedly, a no-exploitation defence would be less problematic in this respect than an offence defined in terms of exploitation. Exploitation would be in issue only if the defendant chose to introduce it; it wouldn't need always to be considered. For our purposes, however, it's enough that exploitation would always be potentially in issue. Indeed, we may expect that it would frequently be in issue: To have a chance of avoiding liability, defendants would need only to produce some evidence that their conduct was not exploitative. 75 Given this, a no-exploitation defence is likely to 73 Thanks to an anonymous reviewer for suggesting this reply. 74 More precisely: it becomes a condition of liability for the crime that the absence of exploitation is not proved. 75 Or so we may assume: A requirement that the defendant prove the absence of exploitation would, I take it, infringe the presumption of innocence.
result in significant problems of indeterminacy. Legislators thus have good reason to avoid its creation.
Consider next how such a defence might set back the goals of prevention. Take first the prevention of secondary victimisation through the criminal process. A no-exploitation defence would set back this goal in much the same way that it would set back the goals of determinacy: Exploitation would always potentially be in issue in criminal cases, so complainants could always potentially end up 'on trial'. By contrast, in the absence of such a defence, this need never happen. Take next the goal of prevention through improved deterrence. Again, the problem here is that, with such a defence in place, exploitation becomes a condition of liability for the crime. Hence, fewer people may be deterred from conduct that is actually exploitative, when they believe (incorrectly) that it is not exploitative. 76 Moreover, it's hard for legislators to avoid this effect without engaging in deception. If the law is to be honest with citizens about the criteria for their liability -including the criteria found in defences -then it cannot avoid the risks of citizens interpreting those criteria incorrectly. 77 Since those risks are greater for an exploitation standard than for an age-of-consent standard, the latter may deter citizens more effectively from the underlying wrong.
We should conclude, then, that legislators are sometimes justified in criminalising non-wrongful conduct. It follows that the wrongness constraint, as an all things considered principle, is false. To emphasise, it doesn't follow that the criminalisation of non-wrongful conduct is unproblematic. The state has good reason to avoid condemning and punishing such conduct; hence, even when legislators are justified in enacting over-inclusive crimes, perhaps they should seek to avoid this result in other ways (such as through 76 The risk of this effect is especially high in cases like the present, in which people are prone to over-estimate their ability to judge their own behaviour. To be clear, however, there will doubtless remain some individuals who are justified in trusting their own judgement. The question is whether the benefits of increased deterrence among the former group outweigh the costs of over-deterrence among the latter group. Compare Duff, Answering for Crime (n. 1) pp. 166-172; L. Alexander and E. Sherwin, The Rule of Rules: Morality, Rules, and the Dilemmas of Law (Durham, NC: Duke University Press, 2001) pp. 55-56. 77 One might reply that the law need not be so honest. It could employ some measure of 'acoustic separation': The age of consent could be advertised as the rule governing citizens' conduct, while courts employ the no-exploitation defence covertly. However, such strategies are both morally dubious and practically difficult to maintain. See e.g. Alexander and Sherwin, The Rule of Rules policing or prosecution policies). However, our concern here is with criminalisation: with making conduct punishable. 78 And legislators can have good reason to make non-wrongful conduct punishable, despite their reasons against doing so. As a principle of criminalisation, the wrongness constraint must therefore be rejected.

V. DOES JUSTIFIABLY CRIMINALISING CONDUCT MAKE IT WRONGFUL?
Advocates of the wrongness constraint, as an all things considered principle, might make one further argument in its favour. They might concede the point just made: that is, that legislators might sometimes be justified in criminalising non-wrongful conduct. However, they might argue that this point is consistent with the wrongness constraint. For there remains a possibility that we have so far overlooked: that by justifiably criminalising conduct, legislators make that conduct wrongful. This argument is not premised on the claim that citizens have a general duty to comply with the law. As we saw at the outset, advocates of the wrongness constraint presumably deny that there is such a duty. 79 Rather, it is premised on the narrower, more plausible claim that citizens have a duty to comply with justified laws. If this claim is true, then there is no such thing as conduct that is non-wrongful but permissibly criminalisedand so the wrongness constraint is salvaged.
Before assessing this argument directly, we should firstly be clear about what is and isn't at stake here. The issue is not whether justifiable criminalisation can sometimes make conduct wrongful; it's whether it necessarily does so. Most would agree that, by enacting regulations targeting a given type of conduct, legislators sometimes give citizens new reasons against engaging in that conduct. Classic examples are regulations designed to solve problems of coordination 78 One could also frame this point in terms of the distinction between formal criminalisation ('the law in the books') and substantive criminalisation ('the law in practice'). The wrongness constraint is unsound as a principle of formal criminalisation. But it could yet be, in some sense, a sound principle of substantive criminalisation. or collective action. In such cases, legal regulation can turn otherwise permissible conduct into wrongful cheating or free-riding: consider tax evasion, insider trading, or driving on the hard shoulder. 80 By contrast, the present argument doesn't just claim that there are some such cases. It's more ambitious than this, in two respects. First, it claims that justified criminalisation always creates reasons against the targeted conduct. And second, it claims that these reasons are always decisive against that conduct. Only if these more ambitious claims are true will the argument save the wrongness constraint, as an all things considered principle.
At first glance, however, these claims simply seem too ambitious. Of those crimes that target otherwise non-wrongful conduct, many are unlike the classic examples just mentioned: Their aim is not to give citizens new reasons against engaging in the relevant conduct. 81 To illustrate this, return once again to the example of sexual activity with a child. As we saw, this offence is over-inclusive, relative to the wrong of exploitation that it targets: It catches some conduct that is not otherwise wrongful. But the point of defining it in an overinclusive way is not to give citizens new reasons against this nonwrongful conduct. Rather, it is to deter citizens more effectively against the pre-existing, underlying wrong, and to improve our response to that wrong. Hence, not only can one commit this crime without committing the underlying wrong; one can commit it without endangering the values pursued by its over-inclusive definition. 82 The potential for such cases of 'harmless disobedience', even of justified regulations, suggests that the present argument is unsound. Ostensibly, Wellman agrees with the present objection: It's wrong to disobey just laws. In the end, however, his discussion implies that 'just laws' are those grounded in obligations of fairness. Crimes like sexual activity with a child are unjust because they lack any such grounding: Those who don't commit the underlying wrong should not be criminally liable. 83 Moreover, for legal rules of this kind, the potential is ever-present. As we saw in the previous section, the very features of such rules that make them attractive for legislators are likely to result in cases where breach is permissible for citizens. See further Alexander and Sherwin, The Rule of Rules (n. 76), ch. 4; Schauer, Playing by the Rules (n. 56), ch. 6.4. Why might one think that it's wrong to commit justifiably enacted crimes, even in cases of harmless disobedience? One possible answer is that the criminal law has moral authority: It has a normative power over citizens, such that, by justifiably enacting new crimes, legislators obligate citizens morally not to commit those crimes. Some advocates of the wrongness constraint appear to endorse this answer. 84 By itself, however, it is unsatisfactory: It doesn't explain how the criminal law gets this power or why citizens would have the corresponding obligation to comply. Moreover, in the context of legal authority and obligation more generally, these things have proved difficult to explain. Indeed, they've proved difficult to explain partly because of the ever-present potential for harmless disobedience. 85 It's therefore unlikely that we can solve this problem simply by invoking the authority of the criminal law.
Nevertheless, citizens may yet have other good reasons to avoid committing justifiably enacted crimes. Most obviously, a system of just criminal laws, with which citizens generally comply, has great social value. By failing to comply with such laws (one might think), citizens endanger this value -even when they do not endanger the values underlying the specific crime that they are committing. For one thing, such non-compliance arguably displays an objectionable arrogance towards one's fellow citizens. 86 But more significantly, it tends to endanger the system itself -and correspondingly, to erode citizens' confidence that others will comply with the relevant rules. 87 Such damage to citizens' expectations of security and just treatment is bad, and it is wrong to risk its infliction. Therefore (the argument goes), it is wrong for citizens to commit justifiably enacted crimes, even when such commission is otherwise harmless.
Unfortunately, however, this argument still can't save the wrongness constraint, because it fails to establish the required conclusion. Remember how ambitious the argument needs to be in order to do this: It needs to show that citizens always have reason to avoid conduct that has been justifiably criminalised, and that these reasons are always decisive. Plausibly, citizens do have some reason to avoid undermining a system of just criminal laws, of the kind that the argument suggests. 88 But even if so, these reasons are neither always applicable nor always decisive: It is possible to commit justifiably enacted crimes without undermining that system, and without one's conduct being wrongful on that ground.
To illustrate this, return to our example from the previous section, of the relatively mature teenage couple who have a non-exploitative sexual relationship. We've already seen that this couple needn't endanger the values underlying the crime of sexual activity with a child. But might their conduct be wrongful on the basis that it tends to undermine a system of just criminal laws, with which citizens generally comply? Surely this is a contingent matter. First, it's unlikely that the couple's conduct risks undermining that system. We'd need to establish a causal link from their non-compliance to the required undermining effectwhich seems difficult, given that their conduct presumably takes place in private. Second, even if their conduct does carry this risk, this doesn't suffice to make it wrongful. Probably, any such risk would be relatively small. And the couple might have good reasons for their conduct -such as expression of their feelings or their sexual autonomy -that justify their non-compliance. In short, despite the fact that our couple are committing a justifiably enacted crime, their conduct is not necessarily wrongful. Even though criminalisation sometimes makes conduct wrongful, the wrongness constraint thus remains false.
These responses to the present argument won't convince everyone. Whether this argument succeeds depends on the extent of the law's ability to obligate our compliance. And we can't hope to resolve such a contested issue in the space available here. However, the responses should convince many advocates of the wrongness constraint. For as we've noted, these advocates tend to deny that 88 Although even this might be disputed. Such reasons to comply with just rules arise most obviously in cases where the rules are designed to co-ordinate subjects' behaviour -and thus, to facilitate expectations of compliance. It's harder to see how they would arise in relation to rules of other kinds. Compare Alexander and Sherwin, The Rule of Rules (n. 76), pp. 61-68. there is any general obligation to comply with the law. If they are correct to deny this, then citizens can commit crimes without acting wrongly -without endangering either the values underlying the specific crime, or the more general value of a system of just criminal laws. Coupled with the discussion in the previous section, this implies that conduct can permissibly be criminalised, even though it is non-wrongful both before and after its criminalisation. Such advocates of the wrongness constraint, as an all things considered principle, should therefore rethink their commitment.

VI. CONCLUSIONS
Orthodox thought holds that criminalisation should be subject to a wrongness constraint: that is, that conduct may be criminalised only if it is wrongful. In this article, I've argued that this claim is probably false. Criminalising non-wrongful conduct is indeed presumptively impermissible. But it might yet be permissible, all things considered. Moreover, there are examples of crimes targeting non-wrongful conductsuch as the crime of sexual activity with a child -the creation of which does seem justified. Nevertheless, we've also seen that the wrongness constraint rests on solid normative foundations. To criminalise conduct is to facilitate its condemnation and punishment; to coerce citizens against it; and to portray it as wrongful. These actions are presumptively impermissible if the conduct that they target is not wrongful. Thus, we can grant that legislators may sometimes depart from the wrongness constraint. But they must nevertheless have good reasons for doing so.

OPEN ACCESS
This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
School of Law, University of Edinburgh, Old College, South Bridge, Edinburgh, EH8 9YL, UK E-mail: a.cornford@ed.ac.uk