Remote harms and non-constitutive crimes.
The field is large. A wide variety of non-constitutive offenses exists to prevent harm. It includes what criminal lawyers traditionally call the "inchoate" offenses, which prohibit actors from inciting, (4) conspiring, or attempting to commit a (substantive) crime. The inchoate crimes are generic, capable of attaching to any substantive crime, whether constitutive or non-constitutive. Other offenses operate in a more specific, preemptive manner, by identifying and criminalizing some overt step that is typically taken toward committing a particular substantive crime. These "substantive-inchoate" (5) offenses include burglary (entry as a trespasser with intent to commit theft, etc.) and, indeed, the doubly-inchoate offense of "going equipped"--merely being out and about with instruments of burglary. (6) Alternatively, many risk-based offenses safeguard against conduct that simply increases the likelihood of causing harm, in as much as the conduct creates a risk, or heightens the risk, that harm will result. Dangerous driving is an example of the latter. The driving is not itself harmful, and need not be directed toward increasing the risk of injury to others; it is criminalized primarily because it does increase that risk.
Non-constitutive offenses pose problems for criminalization theory because they seem not to fit neatly within the harm principle. Prima facie, they generate over-criminalization, in that they prohibit instances of conduct that are not themselves harmful. The scope of risk-based offenses includes harmful instances of the same type of conduct; but substantive-inchoate offenses do not even do that. Carrying the tools of burglary is never itself harmful. Either way, non-constitutive offenses ask us to forgo options that may be valuable and that are sometimes (or even always) harmless. Not only may I not culpably hurt you, I may not do various other things just in case you receive an injury. Taken to extremes, this threatens severely to undermine my liberty, since scenarios exist in which almost anything I do might ultimately affect your chances of suffering harm.
Building upon an earlier foray into the topic, (7) in this essay we explore the scope of the harm principle and its application to non-constitutive crimes. As we shall see, the governing criminalization principles differ across various types of such offenses. But we begin their analysis with one fundamental principle that unites them.
General Requirement of Wrongfulness
The foundational principle of any criminalization inquiry applies to non-constitutive offenses too. It is that any prohibited act, remote from harm or no, must be wrongful. This follows from the very nature of the criminal law as a condemnatory institution. (8) One cannot appropriately be blamed except for doing something wrong. Any formal judgment of blame, including a criminal conviction, must be predicated upon norm-violating conduct--like all blaming verdicts, it expresses moral reproof of a person for that norm-violating conduct. This may sometimes be obscured by the language of mens rea: a defendant is taken to be culpable (his mens is deemed to be rea) when he intends or foresees the actus reus, or sometimes if he is negligent with respect to its occurrence. But first the conduct must be, as it were, reus. (9) The wrong comes before the blame. I can intend a good deed; I can be responsible for doing a good deed; but I cannot be culpable for it.
Sometimes, the very act of criminalization plays a constitutive role in marking conduct out as wrongful, and this is a matter to which we shall return. But either way, legitimate criminalization requires that the prohibited conduct is in some manner reprehensible, wherefore one who perpetrates the behavior may be considered as a candidate for censure.
How far can we proceed beyond this fundamental starting point? We shall suggest some tentative steps shortly. Before that, however, it is opportune to consider Douglas Husak's proposal for four supplementary principles to limit the legitimacy of criminalizing non-constitutive offenses. The first two are relatively straightforward. First, the offense must be designed to reduce a substantial risk: "Liability to prevent an insubstantial risk would punish individuals who are even less culpable than those who are negligent, and codes should rarely if ever impose liability on persons with so little culpability." (10) Second, the prohibition "must actually decrease the likelihood that the ultimate harm will occur" (11)--a requirement that Husak derives from his more general principle that all criminal statutes must directly advance the government's interests. (12)
It is hard to disagree with these constraints. Similar limitations ought to apply even to constitutive crimes within the standard harms analysis. We pause just to note a minor point. It is unclear why there must be a formal threshold of a "substantial" risk, rather than an unreasonable risk or, more precisely, a risk sufficient within the terms of the standard harms analysis. (13) Obviously, if a risk is minimal, running that risk is likely to be permissible and unsuitable for proscription. But even a very slight risk may suffice for criminalization if the harm being risked is grave enough and the act concerned has insufficient value. Failing to observe certain precautions at a water purification plant might, perhaps, be criminalized even if the chances of harm resulting were just 1 in 10,000. In some cases, it may be appropriate to proscribe an activity altogether if the risk can only be minimized and cannot be eliminated entirely. The relevant inquiry, therefore, is not whether the risk is substantial, but whether it is sufficiently serious--taking into account the probability, the nature and gravity of the harm, and so forth--to warrant intervention by the criminal law.
Husak's third principle is his consummate harm requirement: The state may not proscribe conduct to reduce the risk of a given harm unless the state would be [morally] permitted to proscribe conduct that intentionally and directly causes that same harm." (14) That is to say, the ultimate harm should be of a kind and seriousness that its direct infliction would itself qualify for prohibition under the standard harms analysis. We concur with this, but Husak's focus on intentional consummate-harm cases is misleading. While an adequate proxy in mainstream non-constitutive offenses, such as attempt, the focus on intention threatens to reorient the criminalization analysis in terms of relative culpability. Husak argues:
The rationale for the consummate harm requirement is straightforward. It cannot be worse to risk bringing about an undesirable state of affairs than to engage in conduct that deliberately and directly brings about that same state of affairs. In any hierarchy of culpable states, recklessness is less culpable than intention (or purpose). If the act of intentionally and directly causing a result should not be criminalized, the state cannot be justified in enacting an inchoate offense to prevent persons from merely creating a risk of that result. (15)
This is problematic for at least four reasons. First, it is not necessarily the case that recklessness is less culpable than intention; the difference is arguably one of kind rather than degree. (16) Sometimes, inhuman callousness is no better than deliberate wickedness. Second, and more importantly, whether something is "worse" may depend on factors other than blameworthiness; what counts for this purpose is overall seriousness rather than culpability simpliciter. As such, it is possible that doing something recklessly or negligently may be on occasion more likely to cause harm--or more likely to cause a range of harms than intending a particular harm. (17) Third, intending a harmful act may have value (flag-burning; on-stage destruction of guitars; tattooing; boxing; perhaps duelling; or the "Easy Rider" who rejects a motorcycle helmet), whereas bringing about the same harm incidentally may not.
None of this is to deny the importance of a consummate harm principle, but it is to suggest that it needs reformulation. In line with the basic requirement of wrongfulness that we identified earlier, what counts is whether it would be morally permissible to prohibit conduct that wrongfully (rather than intentionally) causes the ultimate harm. (18) This way of framing the consummate harm principle helps to bring out a fourth point. Conceivably, it may sometimes be legitimate to proscribe conspiring with another to do an act that in itself is not wrongful--or, at least, is insufficiently serious--when done by one person acting alone. Indeed, this possibility underpins the common law offense of conspiracy to defraud. The wrongfulness of an act may vary with the manner of its doing; with whether it is repeated (as in harassment), or done collectively (say, by a gang (19)) rather than individually. Husak here treats intention simply as a culpability ingredient, one that conjoins actors to the ultimate harm that justifies state intervention. But even within the terms of an extended harm principle, (20) the ultimate harm need not be the sole source of the wrong.
A similar worry affects Husak's final constraint. His theory offers a presumption against overinclusive criminalization; he argues against legislation that proscribes a wider range of conduct than is necessary to prevent the relevant harm. In the context of non-constitutive offenses, Husak proposes a culpability principle:
This principle withholds liability from persons who create a risk of harm unless they have some degree of culpability for the ultimate harm risked. It is not enough that the performance of the proscribed conduct just happens to make the occurrence of the ultimate harm more likely. (21)
The culpability requirement might be satisfied provided the ultimate harm were at least foreseeable, although Husak's principle is open to a more stringent standard such as recklessness. Still, "even a culpability requirement of negligence for offenses of risk prevention would be a major innovation in the substantive criminal law." (22)
Although culpability requirements are generally welcome in the criminal law, in our view this proposal is misguided. The issue is not one of culpability, but of responsibility and, more immediately, of wrongdoing. The ingredients of such judgments overlap, but they are not quite the same things, especially in the institutional context of a legal system.
One may, for instance, be culpable (that is, morally blameworthy) for failing to prevent an event for which one is not responsible--being a stranger who owed no duty to intervene. Of course, this is not a counterexample to Husak's principle because his principle is unidirectional: it is exclusionary, not inclusionary. The point is, however, that considerations affecting the legitimacy of state coercion may subtract from ordinary moral reasons. Conversely, we claim, it can add to them.
This is, in part, a familiar point. Sometimes, the legal system may crystallize a norm as a determination of some more abstract, pre-legal wrong. (23) Driving too fast is a moral wrong; driving in excess of 70 miles per hour on a UK motorway particularizes that wrong. (24) Less obviously, a regulatory regime designed to facilitate industrial safety, or judicial processes, or (say) efficient market pricing may contain a variety of reporting and other requirements which serve the needs of that regime.
Suppose that, in order to serve such harm-preventing ends, the state enacts an offense of [phi]-ing, and that its doing so is justified by those ends (subject to limitations that we explore below). Then we have a reason not to [phi]; other things equal, [phi]-ing is now morally wrong. Further, assuming D is or should be aware of the prohibition, (25) if D then intentionally [phi]-s without justification he does so culpably. In such a case, we contend, D may legitimately be convicted of [phi]-ing notwithstanding that D was unaware of the ultimate harm lying behind the prohibition, even if that harm would not have been apparent to an ordinary or reasonable person. (Indeed, as we argue below, this may sometimes hold even in cases in which the ultimate harm is not risked at all by D's particular act of [phi]-ing.)
Culpability judgments tend to be ex post, but we must investigate criminalization questions ex ante. Thus the relevant question is not, as Husak suggests, whether D is culpable for the ultimate harm. It is not even whether D is responsible for that harm (which, if the upstream prohibition is effective, will not even occur); although, as we shall see below, considerations of responsibility are relevant. Rather, the key question for ex ante prohibition is, why is D's conduct (in [phi]-ing) wrongful? If it is, and its permission would lead to harm, the criminal law is not yet ruled out.
Not yet. In what follows, we propose further limiting principles that constrain the legitimacy of non-constitutive offenses.
Types of Non-constitutive Crimes and Their Governing Principles
At the level of theory, non-constitutive crimes occur in four broad types. Each raises different issues of principle, which we consider in the following sections.
The first type comprises what we shall call endangerment offenses, in which the prohibited activity, [rho], creates a risk of causing harm to others. Dangerous driving, driving on the wrong side of the road, and discharging a firearm in a public place would be instances of endangerment offenses.
The second and third types are what we term prophylactic offenses: cases in which the prohibited activity, [sigma], conduces to harm only if accompanied by an autonomous act of [phi]-ing, either by the offender or by another. Causally speaking, we can further distinguish between cases in which [sigma] is a cause of harm alongside [phi] (conjunctive harm cases) and those in which the harm is caused by [phi] rather than [sigma] (mediating intervention cases).
Finally, we consider a special set of cases: those in which the offense is concerned to prohibit activities that are wrongs independently of the harm they cause. We shall discuss these cases briefly in closing.
Many endangerment offenses can be accommodated straightforwardly within the harm principle. That the ultimate harm does not always occur is not by itself an objection, since the standard harms analysis is probabilistic. (26) Where conduct is less likely to cause a given harm to others, the case for prohibition is correspondingly weakened. Running down corridors is not criminalized, but discharging firearms is--the difference lies primarily in the level of dangerousness, in that the latter activity involves greater risk of causing serious injury to others.
The need to locate wrongful conduct can also be met. In general, risk-taking can become wrongful for either of two reasons. First, it may be unreasonable for an actor to run the risk; this is an assessment that requires the very kind of balancing exercise undertaken by the standard harms analysis. (Conversely, an offense that criminalized reasonable risk-taking would violate the harm principle.) Second, it can be a wrong to act in order to harm another, at least in circumstances in which the victim has a right not to be so harmed; here the wrong resides in the attack upon that right and upon the interests of the person who holds that right. (27) The three traditional inchoate crimes are typically wrongs for this second reason. Either way, however, the wrong is not sufficient; its existence does not dispense with the requirement for harm, or at least for enough risk of sufficiently grave harm to meet the conditions of the standard harms analysis. This is one reason why attempt liability is subject to a preparation threshold: (28) short of that, the ultimate harm may be too remote, too dependent on contingencies, for state intervention so intrusive as the criminal law to be justified, whether or not such preparatory conduct is directed ultimately toward harming another.
The worry, then, is that endangerment offenses appear to dispense with the harm requirement, since they prohibit conduct independently of the contingencies that make that conduct dangerous. This is the concern about overinclusiveness. When are such offenses justified?
To answer that, we should begin by distinguishing two types of contingent situations. Suppose that a statute is enacted to prohibit careless driving. The driver who cuts a comer commits this offense if he cannot see the road ahead, even when there is in fact no oncoming traffic and so, for the omniscient, no risk of harm. Rightly so. From the driver's perspective, this is a contingency about which she has neither ex ante knowledge nor control. Such cases involve genuine risk--what we might call concrete endangerment. For concrete endangerment, the constraint of overinclusion adds nothing to the standard harms analysis. It requires only that there be an appropriate balance between the magnitude and gravity of the harm risked and the value of the prohibited conduct that generates that risk.
Consider, on the other hand, a rule such as the speed limit of 70 mph on U.K. motorways. Like the ban on careless driving, the speed limit is designed to protect people from harm. But a person who drives at 80 mph commits the offense even if the road ahead is clearly empty and no genuine risk is created. Thus the 70 mph rule involves abstract endangerment. It deals with conduct that generates risk only if certain contingencies are present, but the rule applies even if those contingencies are absent and the actor knows of their absence. In such cases--and unlike concrete endangerment offenses--there is no genuine risk and, it may seem, no wrong by the actor. Why, if at all, are abstract endangerment offenses justified? (29) Why should the genuinely safe actor be required to forgo valuable options merely because others cannot be trusted?
To some extent, this is a matter of practical necessity. Proof that conduct was in fact dangerous may impose impractical difficulties and costs of law enforcement, and may require monitoring mechanisms that are themselves unduly intrusive. In that case, an abstract endangerment offense may better prevent harm than its concrete counterpart. The choice then becomes one between over- and undercriminalization. (30) Although it remains a balancing exercise, the case for a concrete endangerment offense then militates in favor of the abstract endangerment version.
But the case is stronger than that, because abstract endangerment formulations can also protect against the secondary risk of mistake (about the actual levels of concrete endangerment). In particular, they may offer improved guidance to actors through specifying non-contingently what measures are appropriate to avoid the risk of harm: the injunction not to drive too fast is more easily obeyed as an enumerated limit. If, instead, concrete endangerment were required for liability, drivers would constantly be required to exercise discretionary judgment about whether the conditions of endangerment are present (for example, what other kinds of drivers are there on the road when I drive? How much alcohol can I safely drink?). The human tendency to erroneous and self-deceptive judgments will result in drivers driving too fast, or with too much alcohol, when there actually are people on the road who would be endangered. True, the actor could then be prosecuted for crimes of actual endangerment--but by then the risk would have been incurred, and unnecessarily. Neither is that outcome in the interests of the actor, who may have been well-intentioned and just mistaken. (31) The preemptive, abstract-endangerment rule can help to preclude all this: saying, "never mind trying to figure out what other drivers may be on the road, and how fast might be safe; just do not drive over the speed limit, simpliciter."
In these cases, a preemptive, abstract-endangerment rule operates against concrete endangerment. Still, the rule is likely also to operate beyond that, and restrict the options of those who make accurate judgments about the risks. A 70 mph speed limit may prevent me from deciding, unwisely, that I can easily handle the vehicle at 100 mph; but when Juan Manuel Fangio is the driver, such an assessment may be entirely appropriate. Are there any additional reasons for compliance that we can offer to persons such as Fangio, and to those who, say, hold a current FIA Super Licence? (32)
One possible argument rests upon obligations of co-operation among participants in mutual activity. (33) We may usefully separate abstract-endangerment rules that provide "internal" protection--namely, protection of the members of the group whose conduct is being regulated--from those providing "external" protection of other persons from harm. A speed limit or a drunk-driving rule are examples of internal protection, insofar as it is the conduct of drivers that is being regulated and the rule is meant (inter alia) to protect drivers. By contrast, a prohibition relating to factory pollution is an example of externality: the factory operator is being regulated, but the beneficiaries are others, such as citizens living in the region. Of course, the factory operator might live in the area too, but then her benefit is incidental and not internal to the regulated activity. By contrast, the driver benefits qua driver. Internal endangerment rules constitute, in essence, a scheme of reciprocal protection: if each of us abides by the speed limit, this will not only protect others from our driving, but also protect us from others' driving at excessive speeds. In turn, the fact that the scheme provides reciprocal benefits to all participants suggests a prima facie reason why all should comply, even those of superior ability. All participants have an interest in the efficacy of the scheme, and that interest grounds responsibility in each of us for its compliance. The regime is a package deal.
If that is right, then Fangio's claim to be different, and to drive at 100 mph, is wrong because it is selfish. But perhaps that is too permissive. If internality supplies a reason why an actor must give up options whenever she benefits from a regulatory regime, it could be very broad indeed (and suspiciously paternalistic). It would, for example, seem to permit a prohibition against using aerosolsprays, since such a prohibition arguably would protect the user himself. (34) Moreover, it may be questioned whether selfishness is the kind of wrong that ought to attract the sanctions of the criminal law: unless bound by duties, people are generally entitled to be selfish and to act in their own interests. And, after all, Fangio can drive safely at 100 mph, so why should his doing so be subject to legitimate penal censure?
Yet there is another reason why Fangio's conduct is wrong, and not merely selfish. Once the standard rule is in place, it is typically no longer fully safe, even for him, to drive at 100 mph. Coordinating regimes such as road-traffic rules not only address the individual actor; they set the terms of interaction with other drivers (and pedestrians). Here, especially, no man is an island: others predictably and permissibly rely on those terms when themselves maneuvering on the road. They cannot be expected to accommodate the chance that an egregiously skilled driver will not behave like the rest of us, so that what would ordinarily be a safe driving maneuver may suddenly be dangerous when a distant driver closes in at exceptional speed. Conversely, though Fangio can control his own car, he cannot anticipate or control the driving of others. When he drives in a manner that ordinary drivers do not expect, he therefore generates a concrete and not merely abstract risk of harm to others.
What we have termed prophylactic crimes resemble endangerment crimes, in that they too proscribe conduct that increases the risk of some ultimate harm--harm, moreover, the wrongful infliction of which might legitimately be proscribed. (35) Unlike endangerment offenses, however, the risk of that harm does not arise straightforwardly from the prohibited act. It arises only after further human interventions, either by the original actor or by others. Indeed, apart from completed attempts, (36) even the traditional Inchoate offenses are prophylactic: they do not lead directly to harm, and are merely steps along that path.
For clarity, we shall call the original actor S In such cases, and S's act [sigma]; the later intervention, [phi], may be perpetrated by S or another actor, P. Many of these cases present distinctive problems for criminalization because, at least in cases in which the intervening actor is a free, autonomous agent, the intervention threatens to disrupt S's responsibility for the ultimate harm. If S's conduct is not itself harmful, and [phi] does not occur or is perpetrated autonomously by P, where is S's wrong?
Sometimes the state deals with these problems through the law of complicity, using doctrines that impose accomplice liability upon those who aid and abet criminal behavior by others. But complicity doctrines are ad hoc. They do not prohibit designated activities (such as [sigma]-ing) in general. Rather, they attach liability to S for [sigma]-ing only under particular conditions. Enacting a prophylactic crime can therefore be a simpler alternative, one that better informs S of her rights and duties. Given their potential to facilitate harmful conduct by purchasers, and subject to what is said in the next section, it might be appropriate to prohibit shopkeepers from selling certain kinds of automatic firearms or bazookas altogether, rather than to use complicity law to regulate individual transactions. Prophylactic offenses of this sort can offer more definite guidance to citizens, clarifying their legal position and leaving them better in control of their own criminal liability. (37) Such offenses can also facilitate pre-emptive enforcement measures, and conviction for the prophylactic offense may on occasion better describe the nature of S's conduct than would complicity-based liability for the crime by P. (38)
The problem with this technique is that [sigma]-ing is not inherently wrong. Prophylactic offenses thus criminalize conduct that should not, in and of itself, be criminal. S's mental state with respect to the ultimate harm might help turn [sigma]-ing into wrongful conduct--and this insight underpins the rationale of complicity liability (39)--but the elements of many prophylactic offenses require no such connection. They stand by themselves.
In essence, the problem is that even supposing [sigma] is correlated with and predictably followed by [phi], that by itself does not make [sigma]-ing wrong. Further reasons must be offered why the prospect of subsequent [phi]-ing is S's responsibility, so that S ought to refrain from her per se harmless act of [sigma]-ing. In any liberal conception of the state, people have a fundamental right to be treated as separate individuals, as autonomous moral agents who are distinctively responsible for, and only for, the consequences of their own actions. The legal system should, in turn, judge its citizens according to their own actions and not according to the conduct of others. S's act does not become wrong merely because of P's freely chosen act--let alone by reason of the mere risk of P's act.
That, at least, is the starting point. Yet we retain the sense that prophylactic offenses are sometimes legitimate. In what follows, we explore certain additional ingredients that may justify extending responsibility for [phi]-ing to S, such that her original act of [sigma]-ing can be treated as wrong. As foreshadowed earlier, the cases fall into two main categories.
Many prophylactic offenses prohibit [sigma]-ing because of its tendency to lead to eventual harm where that harm is mediated by someone else's independent intervening choice. Indeed, this is the paradigm of autonomous intervening actor cases. The prohibited conduct by S is in itself harmless, but may in some manner lead P to decide to [phi], which causes or risks the ultimate harm. An intervening choice is likewise involved, however, where it is made by the same actor: although [sigma]-ing is harmless in itself, its performance may make it easier or more tempting for the S herself to decide to [phi].
Why is Imputation in Intervening-choice Situations Problematic?
As we have indicated, the prima facie objection to mediating-intervention liability is that (unless and until done by S herself) [phi]-ing is not S's responsibility and, consequently, it is unsuitable for the censuring response of the criminal law. S's own behavior is harmless. It becomes a candidate for criminalization only because an autonomous intervening actor, whose actions S does not cause or control, may be led to engage in further behavior that does or risks harm. To impose criminal liability on S in such situations ignores the separability of persons as choosing agents. Analogous objections hold where it is S herself who makes a separate, subsequent choice. When harmless conduct is proscribed merely because someone who engages in it is then more likely to choose (freely) to commit further acts that are harmful, subjects are being treated as one might a child: as persons who lack the insight or self-control to resist the later temptation. Assuming S is a competent actor, such treatment would fail to respect her as a moral agent, capable of deliberation and self-control.
The central question, seen also by Antony Duff, (40) is whether we may legitimately coerce S to give up options merely because others might later misbehave. Our answer is, no. Something more than predictability or correlation is needed to make the prospect of [phi]-ing S's responsibility, such that her [sigma]-ing becomes wrongful conduct. How should we characterize that "something more"? The underlying idea, loosely expressed, is that S needs some form of normative involvement in P's subsequent choice, an involvement that makes P's act of [sigma]-ing the business of S. In what follows, we offer some admittedly preliminary thoughts on how this idea of normative involvement may be fleshed out. (41) Broadly, we are speaking of situations in which S, through her conduct, in some sense affirms or underwrites the intervening actor's subsequent choice. But notice the "in some sense" clause here: it allows for actions that are like affirmation, and which in various ways assist, encourage, or otherwise endorse wrongdoing by others. One may therefore doubt whether a general criterion exists that will capture all cases well. Still, it seems to us that a liberal conception of the criminal law requires more than a probabilistic connection between [sigma] and [phi]: to pursue risk-prevention through conduct-profiling techniques is, essentially, illiberal.
Some Types of Normative Involvement:
What types of normative involvement might there be? Perhaps the simplest kind of case is advocacy: S urges P to [phi]. A standard example is the inchoate crime of incitement, as when I try to persuade you that the fastest route to your inheritance would be to kill Granny. Although such prophylactic offenses might sometimes incur free-speech objections (for example, when I write a treatise advocating the use of force in the course of civil disobedience or a political protest), the element of normative involvement seems to be straightforwardly present: if I tell you that you ought to kill Granny, I certainly affirm such behavior. Here, the moral character of my behavior is apparent: in urging P to [phi], I ally myself with P's wrong. In turn, my conduct is wrongful in virtue both of my intention and of the objective nature of the behavior itself.
(ii) Encouragement and imitations
It is the element of communicated endorsement that distinguishes advocacy from activities which merely beget imitation. Famously, in R. v. Brown, (42) Lord Jauncey suggested that consensual sadomasochistic behavior among adults might set a bad example, leading others to engage in such conduct with youngsters unable to give proper consent. Here, there seems no basis for holding the defendants responsible for those subsequent, hypothesized choices. If one engages in otherwise permissible self-regarding conduct, this in itself involves no affirmation that other free agents should use one's example in order to engage in harmful behavior. Having masochistic sex with adults does not implicitly endorse or communicate that others should do the same with young adolescents.
Similarly, if less spectacularly, it is often claimed that certain fictional depictions (say, of sadistic sex) inspire some persons actually to engage in such conduct. Perhaps they might--although the onus of showing empirical likelihood is rarely discharged and, of course, the protection of fictional portrayals is strengthened by norms governing freedom of expression. Even so, two further factors militate against holding the depictor responsible for possible imitations so that his creative liberty may legitimately be considered for restriction. The first is, crucially, that such depictions are like bad examples: the behavior is simply presented, without suggesting that it be imitated. This objection is buttressed by the fictional nature of the depiction. In writing Crime and Punishment, Dostoievski invites the reader to suspend judgment imaginatively; he does not suggest that anyone in real life should actually kill tiresome landladies.
Certain everyday cases escape this exclusion. The bad example may be a matter of concern when, say, children are involved: perhaps Dad should not drink or swear in front of Junior, because Junior might imitate such behavior. Here, it seems intuitively plausible that avoiding that development can be Dad's responsibility. But this conclusion is based on two crucial differences. First, Junior is not yet an autonomous intervening actor. Second, Dad has an obligation of care toward Junior, his dependant. These features are not present in imitation scenarios such as Brown.
(iii) Assistance: Supply of products used to do harm
This is the most difficult case. Prophylactic offenses quite commonly prohibit activities that facilitate subsequent harmful conduct, such as bans on the possession or supply of certain armaments and firearms. Taking product supply as a standard variety, it is clear that these offenses present very significant challenges for criminalization theory. Prima facie, the very fact of facilitation constitutes a form of normative involvement. Yet virtually all products can be misused: if S sells a car to P, it is always possible that P may drive the car in such a manner as to injure others, and may even do so deliberately. Yet no one seriously suggests that car sales should therefore be prohibited. Neither should we ban the sale of computer disk drives, notwithstanding that many owners use them illegally to make pirated copies of music and film. On the other hand, almost everyone agrees with the prohibition of Main Street sales of military flamethrowers. On handguns, views diverge more sharply. What, if anything, is the difference?
What follows is admittedly tentative. Assistance is not, we propose, just a matter of probabilities. As Husak argues, only a tiny percentage of handguns are put to misuse (especially contrasted with disk drives). Of course, the probabilities still count: even were we to concede that computer manufacturers have some responsibility for the misconduct of buyers, the scale and likelihood of eventual harm must still be weighed against the costs--in terms of valuable legitimate uses--of banning disk drives. But we should not concede that responsibility; at least, not merely on the basis that disk drives happen to be useful to pirates.
Part of the difference is that cars, and disk drives, have standard legitimate uses. Cars are meant to be driven. Since driving is the widely understood function or telos of the automobile, by selling the car S implicitly affirms that it is being sold to be driven. But here, P has no independent legal or moral reason for desisting from use of the car: driving per se is neither wrongful nor harmful.
Contrast the sale of military flamethrowers to civilians, where the very design and function of the device is as an anti-personnel weapon. In such a case, one cannot regard the harmful use as incidental or ancillary, and unconnected to its supply. Our argument is, in effect, that to supply a tool is to condone the use of that tool for its core function. This is a sufficient normative involvement to make that use the seller's concern.
Obviously, the supply by S has the effect of helping P to inflict harm. But the suggestion here is that there is also a communicative element to S's conduct, such that the assistance is not merely happenstance but something S endorses and for which S too is responsible. The core function of a thing is ordinarily a constitutive part of our social understanding of that thing. As such, the act of supplying the thing has expressive meaning as a facilitation of its use. (Indeed, this is why, in contract law, buyers normally benefit from an implied warranty that goods are "fit for purpose," and can rescind purchase when they are not--something buyers cannot do when the defect is incidental.) In these cases, it is not open for S to deny that the product's deleterious use has anything to do with her. Similarly, sales of police radar detectors may legitimately be considered for prohibition within the terms of an extended harm principle, in as much as their core function or telos is illegitimate.
Clearly, any argument for normative involvement based on core functions of a product will admit of hard cases. Sometimes, these will arise because the thing has multiple standard uses, some innocent and some not (crowbars are an example). In such cases, S can be regarded as being prima facie normatively involved in any of the standard uses to which the thing is then put; although the existence of standard legitimate usages should count, in the normal way, as a weighting factor against criminalization. One option in such cases might be to include an ulterior mens rea element in the prophylactic offense, such that S offends only if actually aware that the buyer's likely use is improper.
Other cases will be borderline because of uncertainty about the core nature of a thing's functions. Handguns, perhaps, are in this category. To the extent that their core use is uncertain, the case for prohibiting their supply is also borderline, and may well differ across societies. Effectively by constitutional flat, the possession and use of handguns is deemed prima facie legitimate within the U.S. It seems hard to claim, therefore, that their core use is in mounting wrongful attacks on people or their property. (43) By contrast, the social understanding of handguns in the U.K. associates their use much more closely with violence and wrongdoing; perhaps reflecting the traditionally more crowded nature of domestic housing, which permits few opportunities for recreational use. (44) Insofar as this is true, their supply can be more clearly understood as associating suppliers with the wrongdoing they facilitate.
(iv) Assistance: Supply of advice
Clearer cases are assistance by supplying advice: for example, by publishing a "recipe book" for murder. Do recipes, manuals, and other forms of advice constitute advocacy? Not quite. Recipes usually do not urge the reader to employ them, but simply describe how to do something if the reader wishes to. Nevertheless, the telos of a recipe is its use: a cookbook is made for helping people to cook. Thus the requisite normative involvement is present--making it appropriate, subject to free speech limitations, to consider them for prohibition within the terms of the harm principle.
What if the manual disclaims any encouragement? Consider a recipe on "How to kill your rich granny" which contains a statement that killing grandmothers is illegal in certain jurisdictions and not advocated, and that any user of the recipe is proceeding under his own responsibility. However, conventional advocacy is neither necessary nor sufficient here, since there may be other ways of becoming normatively involved in another's conduct. Thus the case qualifies despite the disclaimer: the telos of the recipe still lies in its use. Its publication can, therefore, still be wrong.
In these cases, the conduct to be prohibited does the feared injury only when combined with similar acts of others. Dumping waste in the river is proscribed as a health hazard, but the conduct actually endangers health only when a sufficient number of other persons do likewise. In such situations, the proscribed act is a token of the type of conduct that cumulatively does the harm; ordinarily, therefore, the actor cannot draw a moral distinction between her behavior and that of others who also contribute to the injury. By contrast, in mediating intervention situations S's conduct is harmless even when accumulated.
Conjunctive harms present fewer difficulties for criminalization than do mediating interventions. Suppose that overboard sewage discharge poisons the river, but that the discharge from S's own facility is insufficient by itself to produce that outcome. In this case, S is a participant in bringing about the harm, since her conduct makes a causal contribution to the resulting pollution. Yet, if her contribution is neither necessary nor sufficient for the ultimate harm, S might then contend, "if others comply, why need I?" Here, the type/token argument helps explain why compliance is also her responsibility. Granted, the harm would not occur if others desisted from the conduct and S persisted. But since her token conduct is no different in character from that of others, being an instance of the type of conduct that causes the ultimate harm, she has no reason to be exempted from responsibility for that harm. Moreover, the claim of causal insufficiency supplies no answer to the question, why should S be treated any differently? (45)
Neither is there the impediment posed by mediating interventions to finding a wrong. Indeed, our pollution example involves two kinds of wrong. Since the ultimate harm is collectively caused, the immediate wrong that causes it is a collective one. In turn, S's own contribution is individually wrong in virtue of its participating in the collective wrong.
Some Mediating Considerations
We may safely conclude that responsibility for conjunctive harm extends also to S. However, that conclusion does not dispose of questions about the allocation of compliance obligations, and it is worth adding something about those questions here. If my conduct alone is harmless and becomes harmful only when we do it, the demand that we all desist or restrict our behavior enlists us all in a scheme of cooperation requiring joint action (or, more precisely, joint desistance). But such demands are potentially oppressive: whereas much that I might do alone is harmless, a great deal of otherwise innocent conduct may become harmful, and a candidate for proscription, should enough others do it too.
The issue is one of allocating burdens of desistance: how much need S restrict her conduct, along with others engaging in the same conduct, within a scheme for preventing accumulative harms? The answer may depend on reasons affecting particular classes of actor.
Consider specialists, such as those engaged in the relevant activity as a livelihood. (46) Their responsibility for the ultimate harm may seem obvious from a causal perspective, inasmuch as the specialist is likely to contribute more to the harm--the factory dumps more sludge in the river than any individual householder. But that may not be so. There need be nothing causally distinctive about the specialist: the harm could occur without him, if others continue. Thus pollution may depend on multiple factories (and multiple individual dumpers) with none essential to the result. So, at most, one could say that factories as a class produce more river pollution than individual householders as a class. Even that may not be true, however, depending on the numbers involved. If there are just one or two factories and many householders, the latter as a class may contribute more. Ultimately, moreover, it is not obvious why a "stronger" causal link furnishes a stronger ground for responsibility.
What seems more significant than mere cause is the fact that the specialist puts more of his interest, time, and resources into the harm-causing activity. With such greater involvement should come greater concern. The casual user can argue that there are so many other potential harms from her other possible activities, all of which compete for her attention, that this kind of harm should not be her particular concern. The specialist cannot readily make this claim: the harmful results of this activity are distinctively his business because this activity is one upon which he concentrates a greater portion of his time and attention. (47)
No doubt a counter-consideration is concern about extra burdens. Because the specialist does the activity as his livelihood, stringent regulation will be more burdensome on him than on the casual user. In many cases, users can readily deal with the burden of regulation as a cost of doing business. But in cases where the activity has non-commercial value, and/or the specialist has fewer resources, the case might be otherwise: hence Maine's rule that lobster fishermen may fish for lobsters, but sports fishermen face greater restrictions. We cannot pursue these cases in detail here. The point is that, in conjunctive-harm cases, it does not follow from shared responsibility for the ultimate harm that the criminalization burden should fall equally on all. The matter becomes one of degree: of how great is the burden of compliance, and of how that burden is fairly allocated.
Being a specialist is not necessarily the only connecting role that might make the ultimate harm a person's distinctive concern. A variety of other roles is possible: if I live in the neighborhood, I should arrange my trash so it can be conveniently collected; if I use the library, I should be quiet; and so forth. The roles involved may be varied, but what they have in common is that they are capable of attracting particular norms that apply to persons in virtue of that role--as dwellers, users, and so forth. The content of any duty of cooperation will, in turn, partly depend on the norms those roles attract.
Finally, it is worth noting the important class of offenses that are concerned (primarily) with wrongs rather than harms. Crimes such as rape and battery do not, conceptually speaking, specify a harm: they are conduct- rather than result-crimes. Rape directly criminalizes the wrong, a wrong that exists independently of the harm it undoubtedly causes. Nonetheless, and contra Husak, (48) that rape is (also) harmful is an important element of the case for legislative prohibition: the wrong of rape is not criminalized merely because it is offensive, or immoral. (The difference from core result-crimes such as murder is that, in rape, the harm is not the source of the wrong. Rape is a profound wrong whatever harm it inflicts.) However, the harmfulness of rape takes various forms and is not distinctively, let alone conceptually or definitionally, characteristic of rape. So, although harm is part of the justification of the criminalization of rape, the crime of rape can be committed without causing that harm.
The same is true of inchoate offenses. Although an attempted murder is wrong in virtue of attacking V's right not to be killed, the grounds for its criminalization rest, in the background, also on the fact that it conduces to harm. This is part of the worry about impossible attempts, especially about radically impossible attempts (such as attempted murder by voodoo doll): that no harm is threatened and, if the legal specification of such cases can be adequately distinguished, they should therefore not be criminalized. However, there is nothing generally wrong with criminalizing actions that are wrongs independently of the harm they cause.
Neither is there any general objection to remote-wrong prophylactic offenses. Although the point is incidental here, it is worth noting that Duff's definition of an attack (49) is restricted to actions directed toward harming a value or interest: but one may also do wrong in attempting to wrong another, and such attempts may constitute attacks. Attempted rape is just as legitimate an offense as attempted murder; the same holds for prohibitions of selling books that advise how to perpetrate such conduct.
Nowadays, the vast majority of the criminal law comprises offenses mala prohibita, enjoining conduct that is not pre-legally wrongful. Any modern theory of criminalization needs to account for such offenses, spelling out when and why they are legitimate. The task is not easy. No doubt others will disagree with the analysis offered here. But what is not questionable is that one integral component of any successful theory is the explanation of non-constitutive offenses. The challenge is to justify why individuals may, sometimes, be denied liberties for the sake of preventing remote harms.
(1) The discussion in this essay is restricted to offenses designed to prevent harm to others; that is, within the scope of the harm principle.
(2) Indeed, possession of a handgun is prohibited altogether within England and Wales: Firearms (Amendments) Act 1997 (UK), chap. 5.
(3) We adopt the usage from D. Husak, Overcriminalization: The Limits of Criminal Law (New York: Oxford University Press, 2007), 160 and, more generally, 159-77.
(4) Or, in some jurisdictions, "soliciting."
(5) Cf. P. Roberts, "The Presumption of Innocence Brought Home? Kebilene Deconstructed," Law Quarterly Review 44 (2002): 48, 55-56. In our own analysis below, we shall use slightly different definitions.
(6) Cf. Theft Act, 1968, s. 25(1): "A person shall be guilty of an offence if, when not at his place of abode, he has with him any article for use in the course of or in connection with any burglary, theft or cheat."
(7) A. von Hirsch, "Extending the Harm Principle: 'Remote' Harms and Fair Imputation," in Harm and Culpability, ed. A. P. Simester and A. T. H. Smith (Oxford, UK: Clarendon, 1996), 259.
(8) See, e.g., A. von Hirsch and A. Ashworth, Proportionate Sentencing (Oxford, UK: Oxford University Press, 2005), chap. 2; A. P. Simester and G. R. Sullivan, Criminal Law: Doctrine and Theory, 3rd ed. (Oxford, UK: Hart, 2007), chaps. 1, 16.
(9) Admittedly, the conclusion that particular conduct is wrongful may itself be dependent on the actor's mens rea (for elaboration of this familiar point, see, e.g., A. P. Simester, "Intoxication is Never a Defence," Criminal Law Review 14 : 3, 10-11). In such cases, the role of mens rea in establishing wrongfulness is none the less distinct from its role in establishing culpability, although in practice the two will overlap.
(10) Husak, Overcriminalization, 162. Cf. Model Penal Code, [section] 2.02(b) (substantial risk a prerequisite of recklessness or negligence).
(11) Husak, Overcriminalization, 162.
(12) It is worth observing of the second constraint that many effective substantive-inchoate offenses tend by their nature to advance our interest in harm-prevention indirectly rather than directly, suggesting that Husak's more general principle may need qualification.
(13) The standard analysis is outlined in von Hirsch, "Extending the Harm Principle," 261-62.
(14) Husak, Overcriminalization, 165-66. The proposal has some pedigree: see J. Feinberg, Harm to Others (New York: Oxford University Press, 1984), 109-114; D. Husak, "The Nature and Justifiability of Nonconsummate Offenses," Arizona Law Review 37 (1995):151; von Hirsch, "Extending the Harm Principle," 262.
(15) Husak, Overcriminalization, 166.
(16) For development of this line of thought, see A. P. Simester, "Why Distinguish Intention From Foresight?" in Simester and Smith, Harm and Culpability, 71.
(18) We omit here Husak's additional criterion, "directly" (Overcriminalization, 162), which is intended to exclude consequences brought about by intervening wrongdoers. The work done by that criterion in excluding autonomous intervening actors is adequately reflected within more sophisticated legal tests of causation: see Simester and Sullivan, Criminal Law: Doctrine and Theory, sec. 4.2(iii). Indeed, better reflected, since not all wrongdoers are autonomous, and not all autonomous interveners are wrongdoers.
(19) This is part of the idea behind joint enterprise liability: see A. P. Simester, "The Mental Element in Complicity," Law Quarterly Review 122 (2006): 578, 59-600.
(20) That is, a principle that makes specific allowance for remote-harm offenses, thus going beyond the standard harms analysis.
(21) Husak, Overcriminalization, 174.
(22) Ibid., 175.
(23) See, e.g., J. Finnis, Natural Law and Natural Rights (Oxford, UK: Clarendon, 1980), 284; A. M. Honore, "The Dependence of Morality on Law" Oxford Journal of Legal Studies 13 (1993): 1.
(24) See further the discussion of endangerment offenses, below.
(25) For discussion of this constraint, see D. Husak and A. von Hirsch, "Culpability and Mistake of Law," in Action and Value in Criminal Law, ed. S. Shute, J. Gardner, and J. Horder (New York: Oxford University Press, 1993), 157.
(26) Cf. von Hirsch, "Extending the Harm Principle," 261-62.
(27) Cf. R. A. Duff, "Criminalizing Endangerment," in Defining Crimes: Essays on the Special Part of the Criminal Law, ed. R. A. Duff and S. P. Green (New York: Oxford University Press, 2005), 43, sec. 2. Our usage of "endangerment" differs from Duff's, in that it includes hostile "attacks" on an interest or value.
(28) Although not the only reason: for example, evidential uncertainty about an actor's intentions is typically greater at the point of preliminary conduct, which is likely to be ambiguous in character.
(29) Interestingly, Husak has moved his own position on this issue, now favoring a rebuttable presumption against overinclusive prohibitions rather than an outright ban. See Husak, Overcriminalization, 168, and compare his "Reasonable Risk Creation and Overinclusive Legislation," Buffalo Criminal Law Review 1 (1998): 599.
(30) We assume here that the abstract-endangerment offense being considered is the least overinclusive formulation available. To this extent, we agree with Husak that offenses should be no more extensive than is necessary to achieve their (justified) objectives. See Husak, Overcriminalization, 168.
(31) Thus abstract endangerment formulations can also advance rule-of-law values see Simester and Sullivan, Criminal Law: Doctrine and Theory, sec. 2.3.
(32) Selecting exceptions on the basis of identified qualifications helps circumvent the objection that they are ad hoc or a matter of self-identification--the obverse case, in effect, of restrictions imposed on those holding a provisional licence.
(33) We are grateful for discussions with Nils Jareborg that prompted this line of thought.
(34) Such a prohibition would be an instance of conjunctive harm, discussed below.
(35) Cf. the section on "Husak's Constraints" above.
(36) Completed attempts might also require intervening actors; but in such cases the interventions are typically not autonomous and free (as when D lays a trap, but V must still walk into it). Even if some completed attempts still require autonomous interventions, that possibility can generally be neglected since the remaining completed-attempt cases are likely to justify criminalization on probabilistic, endangerment-based grounds.
(37) By contrast with complicity liability, where S's liability lies partly in P's hands: see Simester, "The Mental Element in Complicity," 578.
(38) A fair labelling concern: see Simester and Sullivan, Criminal Law: Doctrine and Theory, sec. 2.4.
(39) Simester, "The Mental Element in Complicity," 578.
(40) Duff, "Criminalizing Endangerment," in Duff and Green, Defining Crimes, 43, 64; Duff cites von Hirsch, "Extending the Harm Principle," 259.
(41) Perhaps the most obvious example is the incomplete attempt, in which S's act of [sigma]-ing is performed in order to facilitate, or as part of, S's own act of [phi]-ing. As we shall see, however, such a close connection is not always required.
(42) R. v. Brown  1 AC 212, 246.
(43) Although Saturday Night Specials may lie closer to that status than do many other weapons.
(44) Once in place, handgun laws may themselves help to shape differences in cultural traditions. However, though this may be a legitimate bootstrap consideration, it cannot be relied upon to justify the initial proscription.
(45) This is not to deny that there may be further reasons why S should be treated differently, and we note this possibility in the discussion of mediating considerations below.
(46) Cf. von Hirsch, "Extending the Harm Principle," 271-73. Note that this earlier paper treats these considerations differently, analyzing them as an imputation issue. We now think the problem is not one of responsibility but rather of freedom, in that burdens of compliance may interfere too greatly with S's freedom of choice.
(47) This argument is distinct from a rule-of-law claim about greater notice: that the specialist is better able to inform himself of the specific rules applicable to this activity. The notice problem could, for non-specialists, be overcome by taking measures to publicize the rule, etc.
(48) Husak requires only that "statutes are designed to prohibit a nontrivial harm or evil" and defines "consummate" offenses as those in which "each act-token of an act-type proscribed ... produces a harm or evil" (Husak, Overcriminalization, 66, 160). In this essay, we have preferred the term "constitutive" to "consummate," principally in order to preserve the distinction outlined here in the text.
(49) Duff, "Criminalizing Endangerment," in Duff and Green, Defining Crimes, 43, sec. 2.
A. P. SIMESTER AND ANDREW VON HIRSCH *
* A. P. Simester is Professor of Law, National University of Singapore, and Fellow of Wolfson College, University of Cambridge. Andrew von Hirsch is Emeritus Honorary Professor of Penal Theory and Penal Law, University of Cambridge, where he is also Director of the Centre for Penal Theory and Penal Ethics, Institute of Criminology, and Honorary Fellow of Wolfson College. Email: firstname.lastname@example.org
|Printer friendly Cite/link Email Feedback|
|Author:||Simester, A.P.; Von Hirsch, Andrew|
|Publication:||Criminal Justice Ethics|
|Date:||May 1, 2009|
|Previous Article:||The architecture of criminalization.|
|Next Article:||Douglas Husak on dispensing with the malum prohibitum offense of money laundering.|