1 Introduction

The law is a blunt instrument. Two guys walk into a bar. The bouncer asks for ID to prove they are old enough to drink. One produces his driver’s license to show he’s 35, but the other forgot his wallet. The owner is on bouncer duty tonight, and she waives them both in, as they both have scruffy beards and greying hair. However, following several deaths in the city of Collegetown caused by underage drinking, it has been made a criminal offense to admit any person into an establishment serving alcohol without individually confirming their age by examining a government-issued ID. There are (let’s say) good reasons for the existence of this offense. Still, few would say that what the bar owner did was truly wrong, and her conduct is not the sort of thing that the offense really aims to prevent. For the law to label the bar owner a criminal and punish her as one is very plausibly unjust.

This classic puzzle about malum prohibitum offenses illustrates a significant challenge for lawmakers, legal officials, and legal theorists, as well as for citizens who must navigate the demands of both law and morality on their conduct. There are sometimes good reasons to define a criminal offense in a way that is over-inclusive, in the sense that the definition will encompass conduct that is not otherwise wrongful. But are these reasons ever sufficient? When, if ever, can such laws justifiably be made and enforced? When, if ever, can they permissibly be violated? In The Realm of Criminal Law, Antony Duff tackles this challenge head on.Footnote 1 He softens the bluntness of this type of law with subtlety and insight, offering a strategy for reconciling over-inclusive offenses like the above malum prohibitum crime with the principle that wrongness is a prerequisite for legitimate criminalization. Characteristically for Duff, his attempt to answer what initially seems to be a narrow theoretical puzzle ends up illuminating a wide range of foundational issues about the criminal law.

We find Duff’s strategy promising in many ways as an effort to reconcile over-inclusive offenses with the wrongness constraint on criminalization. Nonetheless, we aim to move the discussion forward by raising questions about Duff’s solution and highlighting some limitations and costs. We begin in Part 2 by sketching the contours of Duff’s position; then in Part 3 we propose one refinement and offer two practical observations; and finally, in Part 4 we raise broader concerns. In particular, we question whether the problem of over-inclusive offenses is one that can or ought to be solved, or whether it is better conceived as a difficulty to be managed and mitigated. Of course, we should avoid undue harshness in the law where we can, and Duff’s approach is guided by this worthy ambition. But there may also be a limit to this. To the extent that the harshness cannot be avoided, perhaps this should be acknowledged and faced up to, rather than obscured or finessed.

We do not take our concerns to constitute a refutation of Duff’s view. Instead, our aim is to clarify the choice about whether to sign up to Duff’s approach or rather to opt for a less theoretically ambitious alternative. As with all of Duff’s work, we find that critically engaging with it pays hefty dividends.

2 Duff’s Solution to the Problem of Over-Inclusive Offenses

We begin by presenting and clarifying Duff’s solution to the problem of over-inclusive offenses—namely, how to justify crimes that include act tokens that are not otherwise wrongful, that is, not wrongful independently of their criminalization. The danger here is the dire one of subjecting morally innocent conduct to criminalization and, ultimately, punishment. Malum prohibitum offenses are one category of crimes that raise this challenge: they are assumed to encompass predominantly act types that are not wrongful independently of the law. Another closely related category of crimes that raise this challenge are overly broad “proxy” offenses that are defined to cover an act type that, on the one hand, does include some tokens that are mala in se (the target wrongs) but, at the same time, also encompasses numerous tokens that are non-wrongful.Footnote 2 The discussion to follow covers both problematic categories of over-inclusive crimes, though for simplicity we often focus on mala prohibita. In the rest of Part 2, we explain Duff’s preferred version of the wrongfulness constraint on criminalization (section 2.1), then clarify the puzzle to be solved (section 2.2), before sketching Duff’s solution (section 2.3).

2.1 Negative Legal Moralism and the Strong Wrongness Constraint

Duff aims to defend a version of negative legal moralism that is characterized by what he calls the Strong Wrongness Constraint (“SWC”). It holds that “we may legitimately criminalize a type of conduct only if it is wrongful independently of its criminalization.”Footnote 3 Four clarifications are in order regarding SWC.

The first concerns the sense of wrongfulness that SWC treats as a prerequisite for criminalization. For Duff, SWC should be taken to mean that criminalizing act type X requires not that X is pre-legally wrong, but that it is wrong independently of the criminal law—either prohibited by morality (i.e., mala in se) or prohibited by a justified non-criminal regulation. To satisfy SWC, as he puts it, what is required is that the conduct “is prohibited by a pre- or non-criminal regulation; what is then criminalized is the violation of that regulation, on the grounds that such violations are wrongful…. [T]his shows how a criminal law focused on wrongfulness can have room for mala prohibita.”Footnote 4 It should be noted, however, that while this move accommodates the existence of some mala prohibita, it cannot fully resolve the problem posed by over-inclusive mala prohibita (or other over-inclusive offenses). We return to this point below (in section 2.2).Footnote 5

Second, and relatedly, SWC is to be contrasted with what Duff calls the Weak Wrongness Constraint (“WWC”). This is the claim that “‘[i]t is permissible to criminalize some conduct only if that conduct is wrong either independently of its being criminalized or as a result of its being criminalized.’”Footnote 6 Duff sets aside this weak version of the constraint because, as he reasons, ultimately either it imposes no constraint on what act types can permissibly be criminalized, or else there is no meaningful difference between WWC and SWC. No constraint at all is imposed on what can be criminalized if WWC is read to mean simply that anything that has been criminalized becomes, in virtue of that fact, wrongful, and thus permissibly criminalized. On the other hand, if WWC is understood in a more robust way to recognize that specific, additional conditions must be satisfied in order for conduct to become wrongful in virtue of its prohibition by law (as is the case with traffic rules or other regulations designed to solve coordination problems), then it is not clear how WWC differs from how Duff understands SWC. For as we have noted, Duff understands SWC to mean that any wrongness that is independent of criminalization is enough to satisfy the wrongness constraint, even if this wrongness is itself explained by the operation of the law (“[e]nacting a regulation can… make a normative difference…. It would therefore be consistent with [SWC] to criminalize breaches of the regulation, so long as they are wrongful qua breaches of a regulation that ought to be obeyed”Footnote 7). Thus, Duff sets aside WWC and focuses just on SWC. We follow him in this.

The third clarification is this. Presumably, SWC cannot mean that legitimately criminalizing an act type, X, requires that absolutely every token of X is wrongful pre-criminally. Theft surely can be legitimately criminalized. But not every instance of the act type theft is wrongful. Sometimes acts of theft are justified or excused. Thus, as Duff recognizes,Footnote 8 SWC must be understood to mean that act type X cannot be criminalized unless the unjustified and unexcused instances of X would be pre-criminally wrongful. Accordingly, we suggest the following canonical statement of SWC:

SWC (official): The state may legitimately criminalize an act type X only if all tokens of X that do not satisfy one of the justifications or excuses that would be properly recognized in our criminal law system are pre-criminally wrong (i.e., either mala in se or prohibited by a justified non-criminal regulation).

One final clarification: Is SWC an all-things-considered constraint on criminalization or merely a presumptive constraint—that is, a weighty but defeasible reason not to criminalize conduct that is not pre-criminally wrongful? Some theorists like Andrew Cornford and James Edwards think it must be only presumptive.Footnote 9 Duff, however, is interested in defending SWC as an all-things-considered and “categorical constraint on the criminalization of non-wrongful conduct.”Footnote 10

2.2 SWC and the Problem of Over-Inclusive Offenses

With SWC clearly in view, the remaining challenge that Duff confronts is this. If SWC is true, how can mala prohibita, or other crimes that are over-inclusive in relation to their target wrongs, be justified? In particular, what is the adherent of SWC to say about acts that contravene a malum prohibitum or over-inclusive criminal prohibition where the actor knows her act will not be harmful or bad (i.e., will not bring about the evil the statute seeks to prevent)? SWC would seem not to allow such acts to be justifiably criminalized.

Consider an example of Cornford’s that Duff discusses: statutory rape.Footnote 11 Suppose it’s a crime in our jurisdiction for an adult over 18 to have sex with someone under 16. This offense targets the underlying wrong of exploitative sex, but there are good reasons—to do with clarity, determinacy, and giving people notice so they can avoid criminal sanctions, among other things—for the legislature to target this wrong indirectly via simple age limits and in a potentially over-inclusive way, by defining the age of consent using the bright-line cutoff of 16 years old.Footnote 12 Suppose, then, that 18-year-old Jack is in a committed relationship with a very mature 15-year old named Jill. Jack and Jill have loving, consensual, non-exploitative sex. Plausibly, what Jack did is not morally wrong (and is not otherwise justifiably prohibited by an independently existing, non-criminal regulation). How can it nonetheless be justifiable to criminalize Jack’s conduct? Doesn’t this violate SWC?Footnote 13

Duff offers the following pointed summary of the difficulty:

[U]nless we can appeal, as I do not, to some general obligation to obey the law[,] there will be people who commit the criminal offence, without what the law recognizes as a defence, but who do not act wrongfully in doing so: not because the law that defines the crime is badly drafted, or unjustifiably over-broad or strict; but because although we have good enough reason to formulate both the regulation and the criminal law in those over-broad, over-strict terms, some citizens will also sometimes have good enough reason to violate the regulation and to commit the offence. Their conduct will thus be morally permissible, but the law cannot afford to recognize that permissibility by defining their conduct as legally permissible (whether by means of a narrower offence definition, or by allowing them a formal defence).Footnote 14

As this passage highlights, Duff declines to solve this quandary by appeal to a general obligation to obey the law. Why? Most simply, because Duff does not believe in such a general obligation, and indeed offers arguments against recognizing one.Footnote 15 But even setting aside these arguments, there are several reasons to follow Duff in declining to appeal to such an obligation. First, the existence of a general obligation to obey the law is controversial. This makes it risky for those who are committed to SWC to place too much weight on it. Second, if overbroad offenses were consistent with SWC because of a general obligation to obey the law, there would—contrary to initial appearances—be no puzzle of the kind Duff is trying to solve. If there were a general obligation to obey any law that is passed, we simply cannot end up with a scenario where one breaks the law but this is not wrongful (as such law-breaking would violate the general obligation to obey the law). Duff’s project in this context is to assume there is a genuine challenge here—as supported by the sense of unease provoked by the statutory rape example discussed above, among others—and then to try to identify a solution. By contrast, it would not be satisfying to brush aside the apparent unease we feel about such malum prohibitum offenses just by insisting on a general obligation to obey the law.

Third, and most fundamentally, even if there were a general obligation to obey the law, legislators would still face some residue of the question we started with. While such a general obligation to obey the law might have the effect of bringing over-inclusive offenses into conformity with SWC, we would still face the deeper question of whether this a justifiable legislative strategy. Conscientious lawmakers might still wonder whether the power of the state ought to be used to make otherwise innocent conduct wrongful, and to criminalize it, in this way.

2.3 Duff’s Solution: The De Minimis Principle

Duff therefore aims to meet the challenge directly, without appeal to a general obligation to obey the law. One can, he contends,

agree that there will be cases in which someone breaches a justified regulation whose breach the law formally, and justifiably, defines as a criminal offence, and for which [the defendant] can claim no legally recognized defence, but commits no wrong in doing so; but still insist that we must not criminalize conduct that is not wrongful prior to its criminalization.Footnote 16

How can this be done? The key is to appreciate “that we can be justified in formally criminalizing [] non-wrongful conduct—in defining it as criminal in the books” even when we are not “justified in substantively criminalizing it—in criminalizing it in action.”Footnote 17 In the case of Jack and Jill, for instance, even though Jack’s conduct is formally criminalized under the legal definition of statutory rape in his jurisdiction, SWC could still be satisfied if agents like Jack are not in practice actually subject to criminal conviction.

This can be accomplished, Duff proposes, by including in the criminal code a de minimis provision of the kind found in the Model Penal Code:

De Minimis Provision (DMP): The Court shall dismiss a prosecution if, having regard to the nature of the conduct charged to constitute an offense and the nature of the attendant circumstances, it finds that the defendant’s conduct … did not actually cause or threaten the harm or evil sought to be prevented by the law defining the offense or did so only to an extent too trivial to warrant the condemnation of conviction.Footnote 18

If the criminal justice system in which Jack finds himself reliably applies this principle, even though he would not have a defense to the charge of statutory rape, he would nonetheless at the end of the day not actually face conviction. After all, by hypothesis, he “did not actually cause or threaten the target harm or evil” of exploitative sex.

According to Duff’s proposal, DMP is not meant to function as an affirmative defense. It is not a justification or excuse that mitigates criminal culpability or otherwise exculpates. Instead, DMP is a requirement on the court in the first instance (perhaps to be applied sua sponte). It could also inform charging decisions by prosecutors. Why does it matter that DMP not function as a defense, to be raised by the defendant? Why not instead propose a defense or defenses that specifically exculpate non-wrongful conduct?

The main reasonFootnote 19 is that the same considerations that weigh in favor of over-inclusive offense definitions—clarity, determinacy, etc.—weigh against treating DMP as an affirmative defense. To introduce such a defense would reintroduce the same lack of clarity and indeterminacy that were the reasons for drafting the relevant offenses in an over-inclusive way in the first place.Footnote 20 For instance, a defense to statutory rape protecting defendants who engage in sex that is not exploitative or otherwise wrongful would reintroduce the same difficulties that the original bright-line age-of-consent rule was designed to avoid—e.g. difficult normative questions of what makes a sexual encounter exploitative and when the individual is sufficiently psychologically mature to render the encounter non-exploitative.Footnote 21 Duff contends that DMP operates differently, taking the question of the wrongfulness of the conduct out of the ordinary elements and defenses on which liability turns. His approach would retain the benefits of a bright line rule in the first instance, but also leave room for a more nuanced exercise of discretion by the officials of the system (whether prosecutors or judgesFootnote 22). As Duff puts it:

What the law (the law in action) says to those whose conduct is affected by the regulation is that they should obey the regulation (‘should’ as opposed to ‘must’); and that if they are found to have broken it, even for what they take to be good reasons, they will avoid prosecution, conviction, and punishment (substantive criminalization) only if they can persuade the relevant official that those reasons were indeed good enough. This, we might hope, will make possible a more informal and more productive examination of doubtful cases, whilst providing what should still be a reasonably effective disincentive for those who are tempted to violate the regulation without good enough reason.Footnote 23

Duff thus sees DMP as offering a superior way to remain on the right side of SWC while securing much of the benefit of well-justified over-inclusive offense definitions (e.g. bright line rules).

Having presented his solution to the challenge of over-inclusive offenses, Duff offers some insightful discussion of whether DMP should be a matter of official discretion or rather binding on courts or prosecutors.Footnote 24 He discusses the pros and cons of these two options, but we mention this issue only to set it aside. A further question Duff addresses is whether the policy embodied in DMP should be made public or kept secret.Footnote 25 His discussion seems sensible to us, highlighting as it does the concerns about legitimacy and transparency that would arise if the DMP were kept secret. Again, however, we set this issue aside.Footnote 26 Our focus will be on the success of the overall strategy for reconciling over-inclusive offenses with SWC.

To summarize, Duff settles on a legislative structure that, though admittedly complex, adroitly resolves the challenge posed by over-inclusive offenses. This structure acknowledges and vindicates the good reasons for having some formally over-inclusive offense definitions, but at the same time ensures that non-wrongful acts that fall within these definitions will not substantively be criminalized—that is, the legal system “in action” will (at least ideally) not punish these non-wrongful acts, thanks to DMP. This resolution reflects Duff’s conclusion that while the good reasons that support over-inclusive offenses are not sufficient to justify substantively criminalizing non-wrongful conduct, they are sufficient to put people who non-wrongfully commit formal offenses at the mercy of courts’ and prosecutors’ determinations as to whether their conduct was actually non-wrongful. This allows Duff to vindicate existing criminal law systems that include over-inclusive offenses while still adhering to the absolutist view that any substantive criminalization of non-wrongful conduct is categorically morally prohibited. The result is a strong case in support of the contours of our existing criminal law that also excavates and renders explicit its previously hidden moral logic.

Before proceeding, note that Duff himself is admirably upfront about a range of possible drawbacks to the solution he defends. His strategy resembles, he says, “a familiar criminalization technique” in which the legislature enacts an over-inclusive offense and relies on prosecutors to exercise their discretion to avoid prosecuting innocent conduct.Footnote 27 Duff acknowledges that “[s]uch a technique is, of course, far from unproblematic.”Footnote 28 For one thing, it might “encourage legislatures to be sloppy, rather than trying to define offences as precisely as possible” in order to capture only conduct that is pre-criminally wrongful.Footnote 29 Second, it might make it more difficult for those who are subject to the law to understand what the law requires and foresee the legal consequences of their conduct.Footnote 30 And third, this strategy “accords considerable discretion to prosecutors and other enforcement officials: we must ask how their exercise of that discretion can be made accountable.”Footnote 31

These are all dangers or costs associated with Duff’s approach to reconciling over-inclusive offenses with SWC (and we will return to them below). However, Duff suggests that this tradeoff could, at least sometimes, be worth it. Despite the potential drawbacks of DMP, it at least shows how SWC could in principle be satisfied even if we also think there are sufficiently good reasons for the criminal law to define certain offenses in over-inclusive terms.

3 A Proposed Refinement and Two Observations

We find Duff’s proposal attractive and nuanced. Nevertheless, our task is to identify issues, costs, and limitations. We begin in this Part by suggesting a refinement that we believe strengthens and clarifies the account, and then offer two observations about the scope of what Duff’s proposal achieves, assuming it is successful. In Part 4 we raise some more searching questions about the limitations of Duff’s approach.

3.1 Proposed Refinement Concerning Formal Versus Substantive Criminalization

Some of the theorists Duff engages with (like Cornford) focus primarily on SWC as a constraint on formal criminalization—“the law in the books.”Footnote 32 By contrast, Duff thinks SWC more properly applies only to substantive criminalization—“the law in action,” or which conduct is subject to conviction and punishment at the end of the day and which isn’t.Footnote 33 As Duff observes, by relaxing SWC in this way, we can render “negative legal moralism more plausible … by seeing criminalization as something done not merely by formal legislation, but by the operation of the criminal justice system as a whole.”Footnote 34 In support of this move, Duff contends that SWC, “like any principle of criminalization, should guide all the system’s officials; to see whether it is respected we must look at the system as a whole—at enforcement as well as at legislation.”Footnote 35 In fact, this point appears to be crucial for Duff’s solution. After all, even where DMP is enacted, it will not actually alter the statutory definitions of over-inclusive crimes.

In our view, however, this significant relaxation of SWC is both normatively worrisome and not obviously necessary for Duff’s theoretical purposes; a more modest and arguably more attractive adjustment would serve his aims just as well. Duff makes the plausible claim that SWC, “like any principle of criminalization, should guide all the system’s officials.”Footnote 36 Where legal officials have a choice between a law enforcement action that is consistent with SWC and one that is not, it seems clear that they should choose the former. For this reason, it may be that to know whether SWC is fully respected within a given system “we must look at the system as a whole—at enforcement as well as at legislation.”Footnote 37 But of course this does not entail that legislation considered on its own—the “law in the books”—is not also separately bound by SWC, irrespective of what the system as a whole is doing. The legal system might, instead, be required to comply with SWC at both levels—formally and substantively.

We are open to the idea that SWC should constrain the law in action, but we need not take a stand on the matter here. Our basic point is only that SWC should function as an independent constraint on the content of the criminal law in its own right as well. The way in which the law formally defines crimes is itself morally significant. By attaching, if only formally, the label of crime to certain acts and of criminal to those who perform them, the “law in the books” gives authoritative voice to important communal values and attaches a powerful stigma. It should only do this where the acts encompassed by its definitions are legitimately stigmatized in this way. Declining to enforce an over-inclusive criminal offense cannot absolve the formal law of unjustifiably stigmatizing innocent conduct, any more than declining to carry out a wrongful threat can absolve the threatener of having wrongfully made the threat in the first place—these are half measures, at best.

Accordingly, our view is that, even if SWC applies to the substantive law “in action,” it also provides a natural way to evaluate the formal content of the criminal law (regardless of how it is applied in practice). But we think this point can be handled in a way that is hospitable to Duff’s project. In particular, we suggest that, rather than attempting to argue that SWC is a constraint only on the criminal law as enforced (“the law in action”), it would be better to focus on SWC as a principle that at least also independently constrains the set of criminal prohibitions and principles contained in the criminal code as a whole. We thus propose that SWC should be seen as a constraint on which actions are ultimately legally punishable according to the terms of the code, once all applicable provisions are taken into account (and independently of the realities of law enforcement).

This represents something of a compromise position between viewing SWC as a constraint on the content of individual offense definitions and Duff’s suggestion that SWC should function only or primarily as a constraint on what conduct is actually prosecuted and punished (i.e., the law in action). We remain neutral on whether some additional version of SWC should also operate as a constraint on the actual operations of the substantive criminal law. Instead, we maintain there is at least one version of SWC that operates on the content of the formal criminal law viewed holistically. The latter is all Duff really needs in order to carry out his project (and so this amounts to a friendly amendment, which lessens the commitments he needs to take on to achieve his theoretical aims). After all, suppose SWC is taken to apply holistically to the content of the law, i.e. as a constraint on what ultimately is punishable once all applicable liability-determining rules contained in the criminal code are taken into account. In that case, we end up with a constraint on the formal criminal law that is nonetheless relaxed sufficiently to allow reference to be made to DMP when determining whether individual offenses (or indeed entire formal systems of criminal law) comply with SWC. The result is that DMP can serve its assigned role of bringing what would otherwise be over-inclusive crimes into conformity with SWC.

This has the added benefit of locating different kinds of injustice in the right place. Even if the “law in action” were to contain no injustice or violations of some broad practically focused version of SWC, it would still be a problem—a distinct form of injustice—if the criminal code as a whole allowed convictions to be imposed on innocent conduct. The latter would be a violation of the formal version of SWC as a constraint on the content of the criminal law. There is an intuitive distinction between what the law says (its content) and how what it says is applied, and erasing all injustices in how the law is applied would not cure remaining injustices that attach to what the law says is punishable. Recognizing that SWC also constrains the content of the criminal law (construed holistically) preserves the distinction between what the law says and how it is applied.

Now, this does not threaten Duff’s overall attempt to reconcile over-inclusive offenses with SWC. We are offering a friendly amendment, after all. But it does show that he doesn’t need to retreat all the way to SWC as a constraint on the law in action; it is enough for his purposes to agree that SWC is a holistic constraint on the content of the criminal law.

3.2 Two Observations About Applying DMP

We now turn to two observations about the scope of what Duff’s proposal would achieve from a practical standpoint, assuming it is successful. The last section was not meant to suggest that Duff’s concern with application is misplaced as a general matter. His attention to the degree of justice displayed in the law in action is laudable. But DMP creates new challenges where application of the law is concerned—in two different ways.

The first concerns the discretion DMP affords legal officials. Duff acknowledges that his proposal “accords considerable discretion to prosecutors and other enforcement officials: we must ask how their exercise of that discretion can be made accountable.”Footnote 38 But to be clear, this is not merely an incidental cost of his proposal; it identifies a condition that must be met in order for the proposal to accomplish its central aim of demonstrating how a system that contains over-inclusive offenses can be rendered compliant with SWC. A legal system that has DMP on the books, but in which DMP is not applied appropriately by the relevant officials, might conform to SWC formally but yet repeatedly commit serious injustices at the level of its operations.Footnote 39 This means that in order fully to resolve the tension between over-inclusive offenses and SWC, the legal system must ensure that its officials are both accountable and, crucially, capable of applying DMP appropriately and consistently.

This is no trivial point, for DMP asks a lot of officials. It effectively requires judges (and perhaps other officials)Footnote 40 to ascertain, with respect to each crime on the books, what harm or evil the crime seeks to prevent, what kinds of conduct “actually” cause or threaten such harm or evil, and what degree of harm or evil is required for conviction. Never again could a judge, faced with a defendant protesting her conviction on the ground that her conduct was morally innocent, provide an adequate answer by saying “I’m sorry, but that is the speed limit and you were traveling well in excess of it,” or, “I’m sorry, but you knowingly sold that traffic signal preemption transmitter to a nonqualifying user.”Footnote 41 Any such application of a criminal offense definition would need to be backed up by a sufficiently detailed and complete justificatory theory of the crime in question for the court to be confident, not merely that the defendant’s conduct satisfied the terms of the offense definition (often challenging enough!), but that the defendant’s conduct (i) actually caused or threatened (ii) the target harm or evil (iii) to a sufficient degree to warrant the condemnation of conviction. Particularly in light of the vast (and ever growing) scope and complexity of our criminal codes, this would appear to require judges (and perhaps other officials) regularly to accomplish formidable feats of reverse legislative engineering and normative theorizing.Footnote 42 There are at least reasonable grounds for skepticism about the prospects for officials to perform this task with sufficient accuracy, consistency, and accountability to avoid leading the system as a whole into serious conflict with SWC.

This problem of implementation is, of course, not merely a puzzle for theorists; it represents a concrete practical difficulty that may give pause to legislatures considering whether to implement Duff’s proposal. In light of this problem, conscientious lawmakers might very well conclude, either with respect to particular crimes or quite generally, that the only reliable and effective way to ensure that the legal system remains compliant with SWC is simply to eschew over-inclusive offense definitions, thereby avoiding the pitfalls of relying on courts and other officials to perform the unusual and demanding role assigned to them by DMP. In other words, it might turn out, once the facts on the ground are taken into account, that Duff’s in-principle solution to the problem of over-inclusive crimes is not viable in practice.

None of this will be news to Duff,Footnote 43 but it is worth spelling out because it indicates the considerable distance that remains between Duff’s proposal and a complete and operational solution to the problem of over-inclusive offenses. (There may, moreover, be a bit of a sense here of moving around a bump in the carpet, which we will return to in Part 4.)

Our second, related observation is that Duff’s proposal assumes a great deal about the structure and culture of the legal system, which means that, even bracketing the previous observation, his solution may not be available in a range of systems that do not satisfy these assumptions. As the foregoing discussion indicates, DMP can only be acceptable in systems that are willing to delegate significant substantive discretion to officials. This assumes, among other things, a highly trained and principled professional class of legal officials who are considered competent and trustworthy to exercise such discretion. If we are dealing with a more rule-bound system, one that aims to define reasonably precise instructions for how to act in every scenario, this strategy of using DMP to satisfy SWC will not be a good fit. That is, it will not be a feasible legislation strategy in “low discretion” systems.Footnote 44 Similarly, it may not be feasible in systems that depend heavily on, or find value in, the participation of lay persons in official roles (such as the system of lay magistrates in England and Wales or lay judges in Norway). (It is also interesting here to compare the phenomenon of jury nullification, which could be seen as an alternative way of implementing a DMP-like solution, but has always provoked controversy and never been fully embraced for this purpose.Footnote 45) And other legal systems and legal cultures may prize clarity, transparency, predictability, or other values in a way that makes the DMP solution a poor fit. In any of these cases, again, it might turn out that the only reliable or effective way to ensure meaningful compliance with SWC is simply to eschew over-inclusive offense definitions—or, more problematically from the point of view of Duff’s approach, it might turn out that violations of SWC simply have to be tolerated for the sake of other values or for the good functioning of the system.

4 Two Challenges

We said at the outset that the tension between over-inclusive offense definitions and SWC raises a significant challenge for lawmakers, legal officials and theorists, as well as citizens who must navigate the demands on their conduct of both law and morality. This challenge, however, does not look the same from all of these perspectives, and Duff’s DMP-based solution has a very different significance depending on whose point of view is adopted. The theorist’s way is made much easier by Duff’s proposal, as we are furnished with a tidy theoretical picture in which SWC is preserved as a categorical constraint, sparing us from having to answer difficult normative questions about how to balance SWC against other values. Similarly, the proposal to a significant extent spares lawmakers from having to agonize over whether and exactly to what degree each offense definition they enact runs afoul of SWC. After all, from lawmakers’ point of view, these concerns are neatly swept under the carpet of DMP. But these genuine and important theoretical gains come, we believe, in an important sense at the expense of legal officials and (as we will see) private citizens. We have already discussed (in section 3.2) the demands that Duff’s solution places on legal officials, but there we were considering what difficulties we would face if we were to accept and seek to implement Duff’s proposal. Here we wish instead to ask whether we as a self-governing society should be willing to accept the proposal in the first place, given the way in which it tends to redistribute the difficulties associated with over-inclusive offenses.

The way in which Duff’s proposal shifts rather than resolves some central difficulties can be illustrated by considering its significance in terms of two familiar concepts: separation of powers and vagueness. The following are not meant as knock-down objections, but we believe in each case that the proposal has weighty costs that may be too easily underestimated.

4.1 Separation of Powers

Duff, as we have seen, is alive to worries about giving too much discretion to courts and prosecutors—including concerns about how his proposal could produce legislative “sloppiness,” a loss of legal transparency and predictability, as well as insufficient accountability.Footnote 46 To this list we have added (in section 3.2) a concern about how well legal officials will be able to perform the unusual and demanding task that DMP assigns to them. But the problem we mean to raise now is different from all of these.

It is a worry about separation of powers, and whether Duff’s proposal sufficiently respects the wisdom in having different branches of government perform different tasks. Adopting DMP would amount to the legislature offloading an important part of its distinctive task—the part that involves delivering answers to difficult normative and policy questions through a deliberative democratic process—onto courts and prosecutors. It would force courts and prosecutors directly to confront the question of what conduct is sufficiently wrong to merit “substantive criminalization.” We think there are good reasons for our system not to operate in this way, relating to the proper exercise of legal officials’ and lawmakers’ distinctive powers, competences, and roles. We consider the role of legal officials first, before turning to lawmakers.

We cannot give an exhaustive defense of separation of powers here. Instead, consider one possible way to draw out an important aspect of this ideal—namely, the idea that the legislature is appropriately tasked with answering difficult policy and normative questions in virtue of its distinctively open, democratic, and deliberative nature. This approach to answering potentially divisive questions is meant to give all citizens a voice through public deliberation and exchange of views among their elected and accountable representatives. Even if individuals end up disagreeing with the conclusion reached by the legislature on a given question, the law’s answer retains some measure of legitimacy because of the transparent way in which all citizens were (in theory) able to have input, via their representatives, into the decision-making process.

By contrast, the same is not true when the courts or law enforcement officials answer difficult normative questions for themselves. The reason is that these institutions largely lack the open, democratic, and deliberative structure that the legislature (at least ideally) exemplifies. Courts, prosecutors, and police are not equipped to provide all citizens with a voice in settling normative or policy-related questions. When they answer such questions, the answers are not reached by the people’s elected representatives via a public process of deliberation.Footnote 47

Now, it might be thought that the issue here is no more serious than for common law judging or statutory interpretation generally. But the separation of powers problem we mean to point out seems to us different in kind, for at least two reasons.

First, it is one thing to resolve normative questions around the edges in developing the law—a bit here and a bit there in the piecemeal fashion of common law judging; but it is another thing to throw into the courts’ lap the core question of what conduct causes or threatens to a sufficient degree the “target harm or evil” that a criminal statute seeks to prevent. That is the fundamental question of offense definition that sets the agenda for the whole of criminal law, and it seems a mark of progress that this question has generally been taken over from the courts by legislatures. In U.S. federal law, for example, the definition of criminal offenses has long since ceased to be a matter of common law development.Footnote 48 One worries that DMP would represent a step backwards in the direction of a common law criminal system, with all the due process concerns that entails.Footnote 49

Second, DMP represents the delegation to officials of a question that is not only fundamental in character but unusually broad in scope of application. By its explicit terms, DMP is an invitation to exempt any defendant from prosecution or conviction of any crime, based on the relevant official’s best understanding of fraught and contested categories like “target harm or evil” and what “warrant[s] the condemnation of conviction.” The difficulty is thrown into sharp relief by the hypothetical situation of a court or prosecutor who in good faith concludes that very little of the conduct covered by a particular offense definition—or perhaps even none of it—actually causes or threatens what is best understood as the offense’s “target harm or evil.” This is a problem that would not be possible even under a full-blooded common law system, but DMP leaves this open as a conceptual possibility. What is an official in this situation supposed to do? Do we really want to put officials in the position of having to wrestle with and resolve such sweeping, open-ended legislative-type questions? Doing so seems to exact an unusually steep cost in terms of the values promoted by separation of powers.Footnote 50

Turn the focus now from legal officials to lawmakers. Duff, recall, acknowledges that DMP “might encourage legislatures to be sloppy, rather than trying to define offences as precisely as possible, so that they capture, as far as is practicable, all and only the types of conduct that ‘cause or threaten the harm or evil sought to be prevented.’”Footnote 51 But we see another ground for concern here that is distinct from the general fear of sloppiness due to over-reliance on prosecutorial discretion. It is a concern less about the importance of accuracy in drafting and more about the moral complacency DMP may enable among lawmakers.

Suppose the legislature is persuaded by Duff’s proposal to adopt DMP; and further suppose that the legislature continues to approach its task conscientiously, determined to avoid sloppiness and to define offenses as precisely as possible. Still, in order to do its job of offense definition well, the legislature needs to be able reliably and accurately to weigh up the benefits (in terms of clarity, determinacy, etc.) of broad offense definitions against the costs—including the difficulty of officials’ task in carrying out DMP, the possibility of errors in practice that would result in violations of SWC, and the significance and seriousness of those violations for the individual citizens on whom they will fall. It seems to us that there is reason to doubt whether the legislature can be in a good position to identify and give appropriate weight to these costs.

To see this, notice that after the enactment of DMP the legislature’s task is subtly transformed. Offense definitions that would previously have run afoul of SWC are now acceptable, in light of the pressure valve provided by DMP. The legislature is in a position to enact most any overbroad definition secure in the belief that any apparent conflict with SWC will be resolved through the operation of DMP as implemented through the discretion of courts and other officials. Regardless of how carefully it has done its job in crafting offense definitions, the legislature is put in a position to feel that it has paid no moral cost, at least with regard to SWC.

Three features of this situation are worthy of note. First, the legislature can—and indeed, according to the logic of this approach, should—devolve determinations about difficult or finely balanced cases to the courts and other officials. This is not a matter of sloppiness; the very logic of the DMP proposal is that these sorts of questions are not best dealt with at the more general level of offense definition, and should instead be addressed on a case-by-case basis by enforcement officials. As a result, the direct difficulty of actually engaging with and resolving these issues is effectively removed from the legislature’s consideration. It would be a kind of moral dumping that can create worrisome externalities.

Second, the reality, of course, is that no bureaucratic system is perfect, and especially in light of the difficult task that DMP assigns to courts and other officials there are bound to be errors in practice. But these errors will by their nature be invisible to the legislature (this is in contrast to the breadth of enacted offense definitions, which is of course going to be visible to the legislature). This greatly complicates the legislature’s task in assessing the extent to which each offense definition it enacts may give rise to violations of SWC.

Third, even if the legislature does its job carefully and conscientiously, it is effectively insulated against a wide range of SWC-based complaints. In answer to a citizen who protests that she has been convicted under an over-inclusive offense definition despite having done nothing wrong, the legislature can respond: “Even if your complaint is legitimate, it is not appropriately addressed to us. We are responsible for the formal criminal law, whose compliance with SWC we have ensured. Your complaint can only concern the substantive criminal law, and should thus be directed to the legal officials who wrongfully enforced the law against you.” This kind of immunity to complaint may be desirable to individual elected legislators, but it is patently not desirable for a well-functioning polity.

Together, these features tend towards a similar effect, which is to distance the legislature from, and hide from its view, many important SWC-related costs—the very costs that must be balanced against the benefits of broad offense definitions in order for the legislature to do its job of defining offenses accurately and well by the lights of SWC. DMP to a significant extent shifts the responsibility for safeguarding against violations of SWC from the legislature to legal officials, and it is in the very nature of this shift that relevant difficulties will be made less visible, vivid, and salient to the legislature. Accordingly, we worry that DMP will affirmatively obscure the moral costs of over-inclusive offenses and thereby lead to their proliferation—not as mere legislative sloppiness (which is surely all too common already) but as a mechanism that predictably causes more injustice. The fear, in other words, is that DMP may lead to obliviousness and a false sense of moral innocence on the part of the legislature, which will predictably cause more of the very moral tragedy that DMP is designed to avoid.

This therefore suggests another reason to preserve the familiar division of responsibilities between legislature and law enforcement officials that prevails under traditional separation of powers: in order to do its job well the legislature needs as much as possible to assess, face up to, and take responsibility for the moral tradeoffs and dangers that are associated with the legislation it enacts. In sum, then, we worry that Duff’s DMP-based solution would both ask legal officials to perform a function for which they are ill suited and put the legislature in a position where it cannot effectively take account of the SWC-related costs of broad offense definitions.

To this, Duff might reply by accepting our concern about separation of powers and moral complacency by the legislature, but nonetheless insist that the proper response is for the legislature to impose greater accountability on the courts for correctly applying DMP. Perhaps legislative committee oversight of the application of DMP might mitigate the separation of powers concern.

However, we are not entirely satisfied by this response. Even if this is a step in the right direction, it would face difficult hurdles in implementation. First, how would this mechanism of added accountability function? To impose such accountability, someone would need to be tasked with deciding when particular applications of DMP are correct and when they’re not. This, in turn, will not only further deplete already scarce resources, but it will require staking out a position on what the evil to be prevented by the relevant criminal law is, and whether the defendant’s conduct imposed a sufficiently small risk thereof for him or her to escape liability. However, if the legislature had sufficiently clear views on this question—or could obtain them through consultation—it could have just taken this into account when drafting the relevant offense definition. Thus, it is likely to have to be an appellate court (or some other judicially competent body) that will impose such accountability for correct implementation of the DMP, and this recreates the separation of powers problem we are seeking to resolve. It would still require an answer on these legislative type questions from within the judiciary, and thereby still create the political cover and moral complacency among the legislature that we were concerned about in the first place. Thus, we doubt that general accountability mechanisms imposed to ensure the correct application of DMP are likely to be a sufficient solution. Maybe they could be developed further, but we would need to see what they look like to know if they are workable and would justify their considerable cost to run.

4.2 Vagueness

We take it to be uncontroversial that Duff’s DMP-based solution would introduce a significant new source of vagueness or uncertainty into the criminal law.Footnote 52 And just as Duff is alive to general concerns about giving too much discretion to courts and prosecutors, he is alive to the concern that DMP may “make it harder for those who are affected by the law to understand its scope and foresee the consequences of their actions so that they can regulate their conduct without breaking the law.”Footnote 53 He would thus no doubt concede that the uncertainty, from the point of view of those subject to the law, that DMP introduces will to some extent undermine the benefits of adopting overbroad offenses in the first place, namely clarity and determinacy. But here again we worry that the depth and extent of the difficulty may be too easily underestimated.

The inherent vagueness of DMP is manifest enough. In particular, DMP’s phrases “the harm or evil sought to be prevented by the law defining the offense” and “to an extent too trivial to warrant the condemnation of conviction” are highly indeterminate, resulting in a significant degree of uncertainty as to what is and is not legally permissible conduct. Uncertainty of this kind carries familiar costs: reduced deterrence value,Footnote 54 less certainty and guidance for citizens keen to stay within the law, less certainty and security for potential victims,Footnote 55 and opportunities for bias and abuse by prosecutors and other officials. Moreover, the general nature of DMP means that it will tend to have these effects for the full range of criminal offenses, not just one offense here or there.

These are the kinds of concerns that underlie the “void for vagueness” doctrine in U.S. constitutional law. Whether or not a provision like DMP that is directed in the first instance at courts (and perhaps other officials) would be subject to challenge by individual defendants as void for vagueness, the void for vagueness doctrine provides a useful framework for evaluating DMP in terms of the impact that it would have on individuals’ ability to understand and predict the operations of the criminal law. The U.S. Supreme Court in Skilling v. United States put the test as follows: “To satisfy due process, ‘a penal statute [must] define the criminal offense [1] with sufficient definiteness that ordinary people can understand what conduct is prohibited and [2] in a manner that does not encourage arbitrary and discriminatory enforcement.’”Footnote 56 If this is the test to be applied to determine whether DMP introduces too much vagueness to be acceptable, we think Duff’s proposal may be in trouble.Footnote 57 As he says, DMP effectively puts it to citizens that “if they are found to have broken [the law], even for what they take to be good reasons, they will avoid prosecution, conviction, and punishment (substantive criminalization) only if they can persuade the relevant official that those reasons were indeed good enough.”Footnote 58 Imagine that you were in Jack’s position in the scenario previously described, in a committed romantic relationship with a younger teenager like Jill. In light of this standard, how confident would you be in your judgment of what the law requires of you?

To these concerns, Duff could seek to reply that the exculpation on non-wrongfulness grounds that DMP makes possible still represents a substantial improvement in terms of avoiding violations of SWC. Granted, DMP does sacrifice a bit of determinacy, notice, deterrence, and prosecutorial predictability. But, Duff might respond, couldn’t this still be the best compromise of all the relevant considerations in play, and one that at least opens up a space for “a more informal and more productive examination of doubtful cases”?Footnote 59 This is not an easy question to answer. In general, it seems difficult to determine whether the improvement offered by DMP is worth it, or whether the setback to determinacy, deterrence, and guidance to citizens is too great to bear. But we think there is a crucially important dynamic in place here that significantly weakens this line of response.

The dynamic that troubles us is this: The more effective DMP is in bringing broad offense definitions into compliance with SWC, and the more it is put to use for this purpose, the stronger the concerns about vagueness become. This is because the vagueness effects of DMP are directly related to how often it is invoked. In a system where DMP is treated as a measure to be used only in extremis to address exceptional and unexpected instances of injustice, the vagueness introduced by DMP will be minimal. But the more DMP is treated as a measure to be used liberally, as a resource that officials are expected, encouraged, and indeed considered to be obligated to apply routinely for the sake of maintaining the moral acceptability of the system as a whole, the more substantial the vagueness introduced by DMP will be. And Duff’s proposal would appear to create significant incentives (and indeed even normative pressure) for legislators and the legal system to move in the direction of the latter scenario. These inducements include not only avoiding the pangs of a guilty conscience but providing the legislature with a convenient way to avoid criticism.

We conclude, then, that the vagueness concerns confronting Duff’s proposal are both substantial and unusually sticky, in that the more the proposal succeeds in the task that it sets for itself, the graver the vagueness concerns become.

5 Conclusion

We have been concerned to highlight some troubling aspects of Duff’s proposal: the way in which it relies on legal officials to perform what is fundamentally a legislative function; its tendency to obscure from the legislature’s view the difficulties surrounding the implementation of over-inclusive crimes; and the danger that reliance on DMP would become sufficiently routine to introduce significant vagueness problems. At bottom, the worry is that offering DMP as a solution to the problem of over-inclusive crimes serves to deliver a neat theoretical picture in which moral tradeoffs and frictions are as much as possible sequestered and removed from view (at least, the view of theorists and lawmakers). That serves the aesthetics of theory building very well (especially if one of the desiderata for a good theory of the criminal law is that it more or less vindicates our actual laws, which include quite a few malum prohibitum crimes). But in designing a criminal code, theoretical aesthetics should not be the only guiding principle; instead, it is plausible that we should want any moral tradeoffs to stick out and be in a sense as ugly and visible as possible, so that they are as salient as they can be to the relevant officials, including lawmakers. After all, we want the legislature to have to reckon with the risks and costs of their preferred policies, especially in as morally charged an area as the criminal law. Thus, we wonder if Duff’s suggestion that DMP may be seen as a solution to the puzzle of over-inclusive offenses actually minimizes the moral difficulties we want the legislature to be forced to grapple with.

While we have focused on trying to develop doubts and concerns, we admire and have great sympathy for Duff’s subtle and plausible treatment of these issues. Indeed, we are open to ultimately concluding—after all the worries and costs (and benefits!) are totted up—that Duff’s response to the problem of over-inclusive offenses proves to be the best one on offer. However, our main point is that these costs must be clearly reckoned with—not only by the theorist but also by the legislature in its search for a responsibly crafted and just criminal law.