Corps de l’article

Introduction

It is a principle of fundamental justice that a law should not be overbroad. An overbroad law that affects the life, liberty or security of the person therefore infringes section 7 of the Canadian Charter of Rights and Freedoms.[1] While such an infringement may be justified under section 1 as a reasonable limit on the section 7 right, there appears to be only one appellate case where such a justification was accepted.[2] A successful overbreadth claim is therefore a powerful argument against the constitutionality of legislation. The section 7 norm against overbreadth was first recognized in 1994[3] and has played an important role in the Supreme Court of Canada’s approach to both high-profile policy issues[4] and lower-profile but important issues of criminal procedure.[5] The Supreme Court of Canada’s recent consideration of the norm in R v. Ndhlovu and R v. Sharma[6] shows, however, that there is a sharp division between the judges of the Supreme Court of Canada as to the content of the norm, and therefore as to the scope of its application. The majority in Ndhlovu applied what I will call a strict version of the norm against overbreadth, according to which a law is overbroad if it applies in even one (actual or hypothetical) circumstance that is not directly necessary to the achievement of its purpose. The dissent in Ndhlovu (and, arguably, the majority in Sharma)[7] rejected the strict version and applied what I will call a relaxed version of the norm. The relaxed version of the norm allows the law to apply more broadly than the strict version if that broader application is reasonably necessary to the operation of the law.

The strict version has some apparent advantages over the relaxed version. It is arguably more consistent with the canonical version of the norm articulated in Canada (Attorney General) v. Bedford; it appears to provide greater protection for the fundamental interests mentioned in section 7 (life, liberty, and security of the person); and its content is more precisely defined. Despite these advantages, I will argue that the strict version of the norm rests on two untenable assumptions about how law operates. First, it assumes both that a law is always an instrument for achieving a purpose that can be fully specified apart from the idea of legal order. And second, it assumes that a law can be drafted and applied so that it never goes beyond that purpose. These assumptions are both mistaken. Not all laws are instrumental in that sense. But those laws that are instrumental are general, and so, if they are effective at all, they inevitably apply where their application is not directly necessary. The efforts of courts to fit such laws into the model of law required by the strict version of the norm have generally required them either to disregard the distinction between means and ends required by the strict norm or to ignore applications of the law that are, according to the strict norm, obviously overbroad. Moreover, it is doubtful whether the strict version of the norm satisfies the Supreme Court of Canada’s own criteria for a principle of fundamental justice. It would be better to recognize that the strict version of the norm is untenable and, accordingly, adopt the relaxed version of the norm as a principle of fundamental justice.

I. Specifying the Norm against Overbreadth

A. Origin of the Norm against Overbreadth

Section 7 of the Charter provides, in effect, that any law that affects “life, liberty [or] security of the person” must comply with the “principles of fundamental justice.”[8] As is well known, when the Charter came into force in 1982, the predominant view among commentators was that the “principles of fundamental justice” were principles of procedural fairness, not norms of substantive justice.[9] But the Supreme Court of Canada quickly rejected that view, holding in Re BC Motor Vehicle Act that although the principles of fundamental justice certainly included procedural fairness, they also included substantive principles, as long as those principles were among “the basic tenets of our legal system.”[10] Initially, it appeared that those substantive principles might be limited to penal proceedings, perhaps more specifically the conditions of penal responsibility. BC Motor Vehicle Act itself concerned basic fault principles of penal law, as did landmark decisions such as R v. Vaillancourt, R v. Martineau, and R v. Creighton.[11] But by the early 1990s, the Supreme Court of Canada was prepared to recognize substantive principles that could be applied not only to the conditions of penal responsibility but also to the substantive definition of an offence and even outside the realm of penal law altogether. The concern in these cases is less with the conditions under which it is fundamentally just to find someone guilty of an offence and more with the plight of the person who is attempting to comply with the law. Is the law sufficiently precise that it gives fair notice to the legal subject? If not, it is unconstitutionally vague.[12] Does the law demand compliance in situations that have no connection with its rationale, thus unduly interfering with personal liberty? If so, it is unconstitutionally arbitrary.[13]

It was in this context that the Supreme Court of Canada first held that an overbroad law was contrary to the principles of fundamental justice. In R v. Heywood, the Supreme Court of Canada was concerned with a branch of the vagrancy offence that prohibited persons who had been convicted of certain sexual offences from “loitering in or near a school ground, playground, public park or bathing area.”[14] The majority, per Justice Cory, held that the purpose of this provision was “[to protect] children from becoming victims of sexual offences.”[15] Giving the word “loiter” its “ordinary meaning, namely to stand idly around, hang around, linger, tarry, saunter, delay, dawdle, etc.,”[16] the provision was overbroad in relation to this purpose because it prohibited for life, without notice, and without possibility of review, all persons to whom it applied from attending at a wide range of public places, whether or not children were likely to be present in those places and whether or not those persons posed any risk to children.[17] There was little discussion of whether, or why, a norm against overbreadth was a principle of fundamental justice. On this point, Justice Cory referred only to Justice Gonthier’s earlier remarks, which seemed to hold that, while overbreadth was relevant to justification under section 1, it was not a distinct constitutional principle in its own right.[18] Nevertheless, the norm against overbreadth is now well-established among the principles of fundamental justice.

B. The Norm against Overbreadth as a Norm of Instrumental Rationality

The section 7 norm against overbreadth is one of three principles of fundamental justice that are concerned with the relationship between the purposes or objectives of a law and its effects on the interests protected by section 7 of the Charter: the norms against arbitrary, overbroad, and grossly disproportionate laws. As the Supreme Court of Canada has said, these norms are concerned with “failures of instrumental rationality,”[19] that is, with laws that are flawed on their own terms, not because of the purposes they seek to achieve,[20] but because of the means they use to achieve those purposes.

The canonical statement of the three norms is found in Bedford. The Supreme Court of Canada, speaking unanimously through Chief Justice McLachlin, described them as follows. An arbitrary law is one “that imposes limits on these [section 7] interests in a way that bears no connection to its objective ...”;[21] such a law is instrumentally irrational because it affects the section 7 interests without advancing the law’s objective at all. An overbroad law is one that “includes some conduct that bears no relation to its purpose”;[22] such a law is instrumentally irrational because it goes further than necessary in pursuing its objectives. And a grossly disproportionate law is one that has “effects on life, liberty or security of the person [that] are so grossly disproportionate to its purposes that they cannot rationally be supported”;[23] such a law is instrumentally irrational because its costs (its effects on the section 7 interests) grossly exceed its benefits (its effectiveness in promoting its objective). The essential problem with a law that infringes any one of these three norms is that its impact on the section 7 interests is excessive in light of its effectiveness in achieving its purpose.[24]

Whether a law is overbroad is usually assessed from the point of view of a person who is trying to comply with the law; its impact on the life, liberty, or security of the Charter applicant normally arises from the way it restricts the choices of a law-abiding person. The overbreadth of the prostitution-related laws at issue in Bedford was assessed from the perspective of a sex worker who was attempting to carry on their work lawfully. The impact of the law on the sex worker’s section 7 interests arose from the way that compliance with the law rendered the (otherwise lawful) work more dangerous than it would have been without the law. The overbreadth of the prohibition on assisted suicide in Carter v. Canada (Attorney General) was assessed from the point of view of the person who wished to obtain medical assistance in dying but could not lawfully do so, and so faced a “cruel” choice between two lawful alternatives: “[S]he can take her own life prematurely, often by violent or dangerous means, or she can suffer until she dies from natural causes.”[25] While overbreadth issues can arise in relation to the consequences of a finding of guilt, including sentencing provisions,[26] even in such cases the analysis proceeds on the unstated assumption that the accused will comply with those consequences. The norm against overbreadth is centrally concerned with the effect of law-abiding behaviour on the section 7 interests in life, liberty, and security of the person.

C. Specifying the Purpose

The first step in applying the norm against overbreadth is to determine the purpose or objective of the law in question. Determining the objective or purpose of a law is essentially an exercise in statutory interpretation. The modern approach to statutory interpretation and the interpretive techniques associated with it can lead to different results in the hands of different judges.[27] I have no suggestions for making this exercise more certain than it is now; I will assume that the task of interpretation for section 7 purposes is generally no different from any other exercise in statutory interpretation.[28]

The Supreme Court of Canada has, however, given some guidance as to the methodology for identifying a law’s purpose or objective in the specific context of an overbreadth challenge. In R v. Moriarity, the Supreme Court of Canada cautioned against two related dangers in this exercise. On the one hand, the Supreme Court of Canada said, “the statement of the challenged provision’s purpose should, to the extent possible, be kept separate from the means adopted to achieve it.”[29] If the objective is identified too closely with the means used to achieve the objective, there is no room for an argument that it is overbroad in relation to its purposes. On the other hand, “If the purpose is articulated in too-general terms, it will provide no meaningful check on the means employed to achieve it.”[30] This is because any measure might be a suitable means for pursuing a very general purpose, and again there will be little or no room for an overbreadth argument. Thus, the objective of the law should be stated at “an appropriate level of generality and capture the main thrust of the law in precise and succinct terms.”[31] This methodology assumes that it is in general necessary and appropriate to distinguish the objective of a law from the means it uses to achieve that objective. I will question that assumption in Part II-A, below.

D. Two Versions of the Norm against Overbreadth

An overbroad law is one “that goes too far by sweeping conduct into its ambit that bears no relation to its objective.”[32] But just how “sweeping” does a law have to be in order to infringe the norm against overbreadth? In the Supreme Court of Canada’s case law, there appear to be two competing ways of approaching this question, which I will refer to as the “strict” and “relaxed” versions of the norm.[33] According to the strict version of the norm, a law is overbroad if it applies in even one case where its application is not directly necessary to its purpose. As Chief Justice McLachlin put it in Bedford, an “overbroad ... effect on one person is sufficient to establish a breach of s. 7.”[34] I will refer to this holding as the “one person” rule. Moreover, that one person need not be the Charter applicant themself—it could be a person affected by the law in a hypothetical but reasonably imaginable situation.[35] The strict version of the norm, with specific reference to the “one person” rule, was restated in the majority decision in Ndhlovu and the dissenting opinion in Sharma.[36]

According to the relaxed version of the norm, a law is overbroad if it “goes further than reasonably necessary”[37] to achieve its purpose.[38] In this version of the norm, the fact that a law applies to one person where that application is not directly necessary to the law’s purpose does not necessarily show that the law “goes further than reasonably necessary” because some degree of overbreadth (in the strict sense) may nevertheless be reasonably necessary for the proper operation of the law. Suppose, for example, the purpose of a regulatory or criminal offence is to prevent the creation of a certain type of risk, but the law is drafted in such a way that, in an actual or reasonably imaginable case, a person can be found guilty of the offence even if they have not created that risk.[39] On the strict version of the norm, the law would be overbroad because its application in such cases is not directly necessary to its purpose. On the relaxed version of the norm, the law would not necessarily be overbroad; it might not infringe the norm if applying it in such cases was reasonably necessary to its operation—for example, if a broad application was necessary to ensure that it applied in all cases where the relevant risk was created. In other words, if a law is overinclusive, the strict version of the norm would say that the law is overbroad, while on the relaxed version, over-inclusiveness might be reasonably necessary to avoid under-inclusiveness, and if so, the law would not be overbroad.

When overbreadth challenges are rejected, judges tend to state the norm in terms of the relaxed standard and rarely invoke the “one person” rule.[40] Relaxed versions of the norm were recently restated by the dissenting reasons in Ndhlovu; moreover, one way of reading the majority reasons in Sharma is that it adopts the relaxed version.[41]

The issue that divided the Supreme Court of Canada in Ndhlovu was whether mandatory registration of sex offenders was overbroad. Section 490.012 of the Criminal Code required a sentencing judge to make a registration order where an accused was sentenced[42] for any of the designated sexual offences. From 2011 onwards, registration was mandatory; the Criminal Code provided no exceptions or exemptions.[43] The majority held that mandatory registration was overbroad. The purpose of registration was to assist the investigation and prosecution of offences by providing the police with information about possible offenders;[44] but, since registration was mandatory, it would inevitably require registration of those who presented no risk of reoffending and would therefore limit the liberty interests of such persons without directly advancing the purpose of the legislation.[45] The majority specifically mentioned the “one person” rule and provided a specific (real, not hypothetical) example of one person who, because she posed no risk of reoffending, would be unconstitutionally affected by mandatory registration.[46] Justice Brown, dissenting on this issue,[47] accepted the majority’s articulation of the purpose of mandatory registration, but read the expert evidence tendered at the accused’s sentencing hearing as establishing that “an offender’s risk cannot be determined with certainty at the time of sentencing.”[48] In other words, although mandatory registration may well capture persons who pose no risk of reoffending, it is not possible to identify those persons at the sentencing hearing, and therefore it is necessary for the law’s proper operation to require all offenders to register. On the strict standard, capturing persons who pose no risk of reoffending would mean that the law was overbroad; thus, the dissent implicitly rejects the strict standard. Moreover, Justice Brown’s response to the specific example mentioned by the majority explicitly rejects the “one person” rule: “That my colleagues can point to only a single, extreme case where it was clear at the time of sentencing that the offender did not pose an ‘increased risk’ tends to prove my point, not theirs.”[49] On a proper application of the “one person” rule, registration in one “single, extreme case” is all that is needed to make mandatory registration overbroad. If that one case does not make the law overbroad, the “one person” rule does not apply and the standard for overbreadth is accordingly relaxed.

The issue in Sharma was the constitutional validity of paragraph 742.1(c) of the Criminal Code, which made offenders who had committed offences punishable by 14 years or life imprisonment ineligible for a conditional sentence, even if they otherwise satisfied the relevant criteria.[50] Both the majority and the dissent understood the purpose of paragraph 742.1(c) as ensuring that offenders who committed a class of serious offences would typically receive custodial sentences.[51] The majority found that the paragraph was not overbroad in relation to this purpose, while the dissent found that it was. There are several ways to read this disagreement; one reading, I suggest, is that the majority applied the relaxed version of the norm while the dissent applied the strict version. This reading turns on their disagreement over the following question: Is the maximum punishment for the offence a suitable “proxy” for the seriousness of the offence?[52] The majority held that it was, while the dissent concluded that it was not. Without paragraph 742.1(c), the sentencing judge would determine whether the offence committed by the offender was serious enough to warrant a custodial as opposed to a conditional sentence. The proxy of maximum punishment replaces this assessment. Thus, the proxy guarantees that some offenders whose offences were not serious enough to warrant custodial sentences would nonetheless be sentenced to custody. That is why the dissent found paragraph 742.1(c) to be overbroad.[53] The majority implicitly held that this possibility did not make the paragraph overbroad. If it is constitutionally permissible to replace the sentencing judge’s determination of seriousness with “bright-line rules” or a “proxy,”[54] as the majority held, then it is inevitable that some offenders whose offences were not sufficiently serious to deprive them of a conditional sentence would receive custodial sentences, contrary to the “one person” standard.[55]

It is, unfortunately, not entirely clear which version of the norm against overbreadth enjoys majority support from the judges of the Supreme Court of Canada. In Ndhlovu, the majority accepted the strict version of the norm, while the dissent explicitly rejected it; but in Sharma, released one week later, one member of the Ndhlovu majority (Justice Rowe) joined the four members of the Ndhlovu dissent and coauthored the majority reasons. Moreover, Ndhlovu is not cited in Sharma. From a purely doctrinal point of view, the question of which version of the norm is authoritative remains to be decided.[56] For the reasons given in Parts III and IV below, despite some apparent advantages of the strict version, the relaxed version should be preferred.

II. The Trouble with Overbreadth

The norm against overbreadth rests on two related assumptions about how law works: first, that every law is instrumental to an objective that is external to that law, in the sense that its objective can be adequately specified without reference to legal order itself; and, second, that it is possible to draft and interpret a law so that it never applies in situations where its application would not directly forward that externally defined objective. The strict version of the norm requires both assumptions, while the relaxed version accepts the first but rejects the second.

Both assumptions are mistaken. The first assumption overlooks the ways in which a specific law can, and sometimes does, in the context of a legal order, constitute the very purpose it is supposed to promote. The second assumption overlooks the fact that even when laws are instrumental, as they often are, they operate as general norms.[57] It is an unavoidable characteristic of general norms that they are more general than the particular fact situations to which they are intended to apply; thus, the idea that a law can always be both effective in achieving its purposes and yet applicable only where its application directly forwards those purposes cannot be sustained. The norm against overbreadth, stated in its strict form, sets an unachievable standard. The norm against overbreadth, stated in its relaxed form and applied only to instrumental laws, is much more plausible as a principle of fundamental justice.

A. Some Laws are Not Instrumental (and Therefore Not Overbroad)

As noted above, in his discussion of the proper statement of a law’s purpose, Justice Cromwell cautioned that “the statement of the challenged provision’s purpose should, to the extent possible, be kept separate from the means adopted to achieve it.”[58] Why this caution? Because “[i]f undue weight is given to the means in articulating the legislative objective in an overbreadth analysis, there will be nothing left to consider at the rational connection stage of the analysis.”[59] Although Justice Cromwell did not expressly say so, his underlying assumption appears to be that any law that affects the section 7 interests must, at least in principle, be subject to challenge for overbreadth; thus, every such law should be construed as an instrument for achieving an objective definable independently of it. But not all laws are instrumental; indeed, I share the view (though I cannot fully elaborate it here) that the purpose of legal order in general is not instrumental at all, because its function is not to achieve a good that is definable independently of it but to constitute a system of right that governs the interactions of human beings who pursue their own purposes freely in the sense of not being subject to each others’ private power, but only to properly constituted public power. The success of a legal order of this kind cannot be judged according to the amount of some independently definable good, such as the amount of happiness or the level of atmospheric carbon dioxide, that it produces. Although oriented towards human freedom, it cannot be understood as an instrument for producing human freedom, because human freedom is not a good that can be quantified and assessed independently of the idea of legal order that is supposed to constitute it. The success of such a legal order must be judged according to how well it constitutes the relevant idea of freedom.[60] But even if you do not entirely share that view, you might think that some laws are enacted just to do what they do; for example, to create a system of rules just to make exchange possible, apart from any good that exchange might serve (contract law),[61] or to prohibit certain conduct merely because it is, in some sense, inherently wrongful (the law of theft). Such laws are intended to apply to all the behaviour that falls within their ambit, without regard to whatever other effects they might have.[62] One might say that such laws are necessarily not overbroad,[63] but it would be more accurate to say that the norm against overbreadth simply does not apply to them. The difficulty with Justice Cromwell’s formulation of the methodology for determining a law’s purpose is that it precludes, by definition, the possibility of thinking about a law in this non-instrumental way.

Let me give a rather dramatic example—the law of murder. The Criminal Code tells us that a person commits murder when they cause the death of another person by culpable homicide with one of three subjective mental states, all involving an intention to cause death or a high degree of recklessness in relation to the death.[64] What is the purpose of imposing criminal liability for murder? Considering several possible answers to this question will show the difficulty of understanding the criminal prohibition of murder as an instrument to some value that can be fully specified independently of the legal wrongness of murder. One tempting answer is that the purpose of having an offence of murder is to punish one of the most serious legal and moral wrongs that a person can commit; but this statement of the objective runs afoul of Justice Cromwell’s caution because it does not separate the purpose from the means: in order to fulfill its purpose, the law must apply in every case that it applies to, so there is “nothing left to consider” in applying the norm against overbreadth. Another tempting answer expressed more instrumentally, is that the purpose of the prohibition of murder is to protect human life. But that, too, seems too broad to be consistent with Justice Cromwell’s caution, particularly in light of the availability of the defences to murder. Indeed, that is why in Carter, the Supreme Court of Canada rejected the government’s submission that the purpose of the law against assisted suicide was to “preserve life.” As the Supreme Court of Canada put it, “If the object of the prohibition is stated broadly as ‘the preservation of life,’ it becomes difficult to say that the means used to further it are overbroad ... The outcome is to this extent foreordained.”[65] Moreover, this formulation of the objective seems quite inconsistent with the regime enacted in response to Carter, which specifically authorizes physicians and nurse practitioners to intentionally kill other people in certain circumstances (i.e., to commit what would otherwise be murder); accordingly, the regime is structured as an exception to the homicide provisions of the Criminal Code.[66]

Perhaps, then, it is better to specify the purpose of the law of murder along the lines of protecting everyone’s interest in the continuation of their own life. Now we have a purpose that is definable independently of the means (the offence definition) used to promote it. But it is not difficult to think of cases in which convicting an accused of committing a murder, as defined in the statute, would be overbroad in that it does not promote that purpose because there are some situations where there is no personal interest in the continuation of human life. Suppose, for example, that Harold has gone through all the procedures required to obtain medical assistance in dying and that the procedure is scheduled for a certain day. Owing to unforeseen circumstances, the procedure cannot be performed at the scheduled time and is postponed for a month. During that month, Harold has no interest in continuing his life—quite the contrary. And so, if George were to intentionally kill Harold during that month, George would have committed murder as defined by the Criminal Code; but convicting George of murder would be applying the law to one person more than directly necessary for the purpose of the offence; hence, the definition of murder (according to the strict statement of the norm) would be overbroad.[67]

There are good reasons for rejecting the claim that this example is sufficient to show that the law of murder is overbroad; but taking those reasons seriously involves rejecting either the picture of the law of murder as an instrument for the protection of the interest in the continuation of life or the “one person” rule (or both). If George’s act is wrong and punishable regardless of the nature of the interests in the continuation of Harold’s life, then the instrumental picture of the law of murder as a device for protecting the interest in life must be rejected. If we retain that instrumental picture but ask about the effects of not prohibiting George’s conduct on that interest, then we must reject the “one person” rule. There are (at least) two possibilities. First, if George is aware of the postponement of Harold’s procedure and is motivated solely by compassion for Harold’s situation, he nevertheless acts outside the boundaries of the medically assisted dying regime, so finding the law of murder overbroad in this application would arguably undermine the integrity of that scheme. Second, if George is not aware of Harold’s situation—if Harold, for example, is a victim of George’s random act of violence or is murdered during George’s commission of an offence, such as a home invasion—then it might be said that finding the law of murder overbroad in this application would tend to undermine whatever effect prosecutions and convictions for this offence are supposed to have in deterring the commission of the offence. Those are very plausible arguments, but they require abandoning the “one person” rule or adopting a relaxed version of the norm because they amount to the claim that it is necessary to the purpose of the law to apply it to Harold’s act of killing George even though that application is not directly necessary to achieve the purpose of the law.

My point in considering these different ways of thinking about the purpose of the offence of murder is not to suggest that the law of murder is, or is not, overbroad. My suggestion, rather, is that the norm against overbreadth does not apply at all to the Criminal Code’s definition of murder because the purpose of having an offence of murder in a legal order designed to constitute a system of freedom for all is not to achieve any goal that can be specified independently of that system but to prohibit a particular kind of behaviour that is inherently incompatible with the operation of such a system.[68] This purpose cannot be understood independently of the place of the law of homicide in that kind of legal order. The point of the definition of murder in the Criminal Code is to identify a particularly serious class of homicide offences and to make all offences that fall within that class punishable by life imprisonment because of their seriousness. The offence definition constitutes the conduct to which the offence definition applies.[69] Now, it may be that there is a general sense in which all criminal law is instrumental because it is intended, at least in part and by various mechanisms, to protect the legal order as a whole by discouraging the conduct that it defines as wrongful.[70] But that does not imply that every offence must target specific objectives that can be defined independently of the conduct that the offence definition covers. There are some offences whose purpose is best understood as simply prohibiting what they prohibit because the prohibited conduct is inherently incompatible with a rights-based order.[71] Murder is one of them. The terrorism offences, the assault offences, the theft offences likely also fall into this category. The norm against overbreadth is not applicable to the definition of such offences.

The majority decision in Sharma is arguably an illustration of this type of reasoning. I suggested above that the majority in Sharma applied a relaxed version of the norm against overbreadth; but another, and perhaps preferable, way of understanding their reasoning is that they were construing the law at issue non-instrumentally so that the norm against overbreadth would not apply to it. The accused, Cheyenne Sharma, pleaded guilty to importing almost two kilograms of cocaine. She was a young first offender of Indigenous ancestry. Her personal circumstances were overwhelmingly sympathetic and her culpability for the offence was low. The sentencing judge found that a fit sentence was eighteen months’ imprisonment and that the accused would have been an ideal candidate for a conditional sentence.[72] But at that time paragraph 742.1(c) of the Criminal Code specified that a conditional sentence was not available where the maximum sentence was imprisonment for 14 years or life; the offence the accused committed is punishable by life imprisonment; accordingly, a conditional sentence was unavailable.[73] The accused argued that paragraph 742.1(c) was unconstitutional. Among her arguments was the claim that the exclusion of certain classes of offences from the conditional sentence regime was overbroad. A majority of the Supreme Court of Canada rejected this argument. They characterized the purpose of paragraph 742.1(c) as “to enhance consistency in the conditional sentencing regime by making imprisonment the typical punishment for certain serious offences and categories of offences.”[74] The means used to achieve this purpose was “to remove the availability of a conditional sentence for certain offences and categories of offences.”[75] Taking the maximum sentence for an offence as the definitive indicator of the seriousness of that type of offence (regardless of the circumstances of the offender or the particular offence), the majority held that these means were not overbroad.[76] In reaching this conclusion, the majority emphasized that “the definition of a serious offence is a normative assessment in respect of which Parliament must be granted significant leeway.”[77]

Although framed as a standard overbreadth analysis, including a repetition of the cautions from Moriarity about clearly distinguishing means and ends,[78] the majority’s approach in Sharma leaves so little room for any distinction between the purpose and the operation of paragraph 742.1(c) that it may be better understood as entirely insulating certain legislative decisions from overbreadth analysis. If Parliament’s objective is to ensure imprisonment is the typical sentence for certain serious offences, and it does so by enacting a statute that ensures imprisonment for those offences,[79] then it seems as though the purpose of the statute is to do what it does, rather than to achieve some objective external to it. It seems easy to think of a situation in which the removal of a conditional sentence would have an overbroad effect on one person. Cheyenne Sharma herself would seem to be exactly that person, but since the majority takes the legislative judgment of seriousness to relate to the type of offence committed, rather than to the circumstances of the offender and the specific offence committed, in their view she is not. And so, it is unsurprising that the majority does not mention the “one person” rule in its articulation of the norm against overbreadth.

This collapse of overbreadth analysis can also be observed in two earlier decisions of the Court of Appeal for Ontario, namely R v. Forcillo and R v. NS.[80] In Forcillo, the accused, a police officer, was found guilty of attempted murder. Because the offence was committed with a firearm (his service weapon), he was subject to a minimum term of imprisonment of four years.[81] He argued that this provision was overbroad.[82] Its purpose, he said, “was ‘to deter people from choosing to carry a firearm to carry out an unlawful purpose.’”[83] This statutory purpose was obviously inapplicable to police officers who carry firearms while on duty; accordingly, he argued, the provision was overbroad as applied to police officers. In light of the jurisprudential background concerning minimum sentences for the use of firearms in the commission of offences, this submission was very plausible. These minimum sentences were originally a response to a series of Supreme Court of Canada decisions in the late 1980s and early 1990s striking down the “constructive murder” provision of the Criminal Code.[84] That provision, in brief, stated that culpable homicide was murder, whether or not the Crown could prove an intent to kill (as would otherwise be required by section 229) if death was caused while committing several specified offences and in various circumstances, notably where the accused used a weapon or had it on his person while committing the offence.[85] For reasons that need not be detailed here, the Supreme Court of Canada eventually held that section 7 of the Charter required proof of subjective foresight of death for a murder conviction.[86] Since the constructive murder provision did not require proof of subjective foresight of death, it violated section 7. In considering whether this infringement of section 7 could be justified under section 1, the Supreme Court of Canada noted that “punishing for murder all those who cause a death by using or carrying a weapon, whether the death was intentional or accidental, might well be thought to discourage the use and the carrying of weapons.” But the Supreme Court of Canada held that this thought could not justify the infringement, explaining that “[i]f Parliament wishes to deter the use or carrying of weapons, it should punish the use or carrying of weapons.”[87] The minimum sentences enacted in the early 1990s, including the one at issue in Forcillo, were legislative responses to this type of reasoning.

Nevertheless, the Court of Appeal for Ontario ultimately rejected Forcillo’s argument that the minimum sentence was overbroad. It did so partly by construing the purpose of the provision more broadly than in the accused’s submission,[88] but also by attributing to it another and different kind of purpose, namely denunciation: “Denunciation applies to all cases of attempted murder with a firearm, whether or not the person is a criminal, gang member, ordinary citizen or a police officer. Indeed, given the serious trust reposed in police officers, the need to denounce the criminal misuse of a firearm may be even more compelling in these circumstances.”[89]

If denunciation is the purpose of the provision, it would be overbroad only if some offences falling under it were not worthy of being denounced. The Court of Appeal for Ontario did not consider this possibility. Once the sentence is interpreted as having “a denunciatory element represent[ing] a symbolic, collective statement that the offender’s conduct should be punished for encroaching on our society’s basic code of values as enshrined within our substantive criminal law,”[90] it is no longer instrumental to any good that can be specified apart from the sentencing process itself.[91] If denunciation by imposition of a minimum sentence applies to every instance of attempted murder with a firearm, then a law imposing a minimum sentence for the purpose of denunciation is incapable of being overbroad because there is no distinction between the law and the purpose it serves.

The Court of Appeal for Ontario’s approach to the overbreadth challenge in NS was similar. The accused challenged some of the new sex work offences in the Criminal Code—section 286.2 (material benefit), section 286.3 (procuring), and section 286.4 (advertising).[92] These offences were enacted in 2014 in response to the Supreme Court of Canada’s decision in Bedford.[93] In that case, the Supreme Court found that the offences of keeping a common bawdy house, living on the avails of the prostitution of another, and communicating for the purpose of engaging in prostitution violated section 7 of the Charter.[94] However, under the pre-Bedford regime, sex work itself was not illegal; and that fact was central to the Supreme Court’s holding that the purposes of the invalidated offences were to prevent exploitation and to control nuisances, rather than to condemn or discourge sex work as such. In stark contrast, the centrepiece of the new regime enacted in response to Bedford is section 286.1, which makes it an offence to obtain, or to communicate for the purpose of obtaining, the sexual services of another person for consideration. For the first time since the enactment of the Criminal Code in 1892, the exhange of sex for money between adults is criminalized, although the Criminal Code purports to immunize sex workers themselves from criminal liability in respect of such transactions.[95] In NS, the accused did not challenge section 286.1 itself, but instead challenged the material benefit, procuring, and advertising offences. I will focus on the procuring offence. The trial judge had found this provision overbroad on the basis that one of its purposes was to protect the health and safety of sex workers and that, in some reasonably imaginable hypothetical situations, it would prevent them from doing so. The Court of Appeal for Ontario disagreed as to the purpose of the legislation in general and the procuring provision in particular. The Court of Appeal for Ontario held that the legislation in general had three purposes: first, “to reduce the demand for prostitution with a view to discouraging entry into it, deterring participation in it, and ultimately abolishing it to the greatest extent possible in order to protect communities, human dignity and equality”; second, “to prohibit the prostitution of others” and the commercialization of sex work; and, third, “to mitigate some of the dangers associated with the continued, unlawful provision of sexual services for consideration.”[96] Against that background, the purpose of section 286.3 in particular was “to denounce and prohibit the promotion of the prostitution of others in order to protect communities, human dignity and equality.”[97] This purpose did not encompass the health and safety of those who were already involved in sex work.[98] Section 286.3 was not overbroad because “[t]he prohibited conduct—a wide range of conduct intended to procure a person to offer or provide sexual services for consideration or engaged in for the purpose of facilitating an offence under s. 286.1—is directly and rationally related to the purpose of the provision.”[99] Once the purpose of the offence is construed as discouraging the behaviour that it prohibits, particularly against a background understanding of the overall purpose of the legislation as protecting vital but unquantifiable values such as human dignity, a finding that it is not overbroad is inevitable.

B. All Instrumental Laws are General (and Therefore Overbroad)

My claim is not that all criminal law, much less all law in general, is non-instrumental in the way described in the previous section. As demonstrated in Bedford, the former offence of keeping a common bawdy house was instrumental to nuisance control, not only because of the legislative and jurisprudential history that the Supreme Court of Canada recounts but also because the objective (controlling nuisances) was extrinsic to and readily definable independently of the offence itself (keeping a common bawdy house). There is no intrinsic relationship between bawdy houses and nuisances; that is, there are other ways that nuisances can be caused and controlled, and it is possible to keep a bawdy house without creating a nuisance. Similarly, it is entirely plausible to read offences such as impaired operation of a conveyance and possession of prohibited substances as instrumental to objectives that are independent of the means used to achieve them. For instance, the criminalization of dangerous driving is instrumental to public safety. The current British Columbia experiment in decriminalizing possession of small quantities of drugs indicates that the federal and British Columbia governments also take this view of drug offences. Whether a harm reduction strategy, criminalization, or, as seems more likely, some combination of the two, is most effective in controlling the harms of drug use is a complicated question. But it is a question that is eminently suited to empirical investigation that could inform a court’s decision on whether drug offences are overbroad in relation to their objectives.[100]

There is, nevertheless, a serious difficulty in applying the norm against overbreadth even to laws that can be understood instrumentally. Laws are, by their nature, general norms; they are therefore not perfectly aligned with the specific fact situations to which they apply. This is so whether a law is more plausibly characterized as a rule (e.g., “do not drive at more than 100 km/h”) or a standard (“exercise reasonable care in driving”).[101] All instrumental laws are likely overbroad in the sense that there will inevitably be at least one actual or reasonably hypothetical fact situation in which the application of the law is not directly necessary to its purpose. If the strict version of the norm, with the accompanying “one person” rule, applies to all such laws, then it is very likely the case that every law that can be analyzed instrumentally—every law that has an objective that is well-defined independently of the law understood as a means for promoting that objective—is overbroad. That is because laws, even instrumental ones, are stated in general terms and must be applied to a vast and unpredictable range of individual cases. It is therefore likely that every instrumental law that affects section 7 interests also infringes section 7 of the Charter by virtue of its overbreadth.

The point is most obvious when the law seeks to promote its objective in a rule-like way, by drawing a so-called bright-line between permitted and forbidden conduct. Consider, for example, provincial “stunt driving” offences. Despite their evocative labels, these offences can be committed simply by exceeding the posted speed limit by a stated amount, without any additional element that might be described as a “stunt.”[102] The objective of these offences is to promote road safety; the means is to discourage a particular kind of risky driving. But the possibility that a driver might commit this offence without creating any risk to the safety of the roads cannot be excluded a priori. That a driver might exceed the speed limit under conditions where it was safe to do so—imagine a skilful and alert driver driving a vehicle in good mechanical condition during the daytime on a straight, dry, empty highway in good repair—is a reasonable hypothetical. It may be tempting to respond to this example by arguing that the prohibited conduct is per se dangerous—but that is just another way of saying that the legislature can authoritatively deem it to be dangerous, which, as in Sharma, collapses the distinction between means and ends that is required for the norm against overbreadth to have any purchase.[103] Bright-line rules are almost always going to be overbroad, according to the strict version of the norm against overbreadth.

A law that is more standard-like than rule-like may seem less likely to be overbroad because the standard established by such a law is supposed to enable a decision-maker to apply the policy behind the law directly and sensitively on a case-by-case basis, rather than indirectly through a rule. The offence of dangerous operation, for example, is defined as “operat[ing] a conveyance in a manner that, having regard to all the circumstances, is dangerous to the public.”[104] Its purpose is to promote safety, and the means it uses to promote that purpose is to hold all operators of conveyances to a general and objectively defined standard of behaviour.[105] But it is precisely because the offence operates by means of a general standard that it will inevitably be applied in cases where it is not necessary to its purpose to do so. That is because there will inevitably be errors of judgment in its application; even where the facts concerning the accused’s behaviour are not in dispute, there will inevitably be a difference of opinion, and therefore the possibility of errors in judgment, in the application of the standard.[106]

It might be objected that my claim that all instrumental laws are overbroad cannot be correct, for the simple reason that Canadian courts, most notably the Supreme Court of Canada itself, have found that some instrumental laws are not overbroad, even when applying the strict version of the norm. But in finding that some laws are not overbroad, Canadian courts have generally used one of three strategies, each of which is inconsistent with the strict version of the norm. The first strategy is to state the objective of the law in such terms that the law could not be overbroad in any application. As indicated above, I do not object to this strategy as such, but it is contrary to the Moriarity methodology for identifying the objective of a law, which requires that “the statement of the challenged provision’s purpose should, to the extent possible, be kept separate from the means adopted to achieve it,”[107] so that there is at least a possibility of overbreadth. The second strategy is to disregard or overlook obvious cases where the application of the law would be unnecessary to its purpose. And the third strategy is to assume that the application of standards by the relevant legal officials will necessarily be without error.

The Court of Appeal for Ontario’s decision in R v. AB[108] provides a good example of the first and second strategies. The accused, who was 21 years old at the time, was charged with two sexual offences arising out of his relationship with a 15-year-old girl. The age of consent to sexual activity in the Criminal Code is 16 years, but the Code provides for a number of “close-in-age” exceptions under which the consent of a person under 16 will be legally valid.[109] The exception relevant to the complainant is found in subsection 150.1(2.1), which provides that if the complainant is between 14 and 16 years of age and the accused is not more than five years older than the complainant, consent is a defence. The trial judge found as a fact that all sexual activity between the accused and the complainant was consensual and non-exploitative; however, the accused was aware of her age and, because he was more than five years older than the complainant, subsection 150.1(2.1) was inapplicable. Accordingly, the Crown had proved both mens rea and actus reus, and the accused was found guilty. But, the trial judge held that section 150.1 was overbroad in the Bedford sense. The purpose of the offences in question, together with the constellation of exceptions to those offences, was “to protect young people under 16 from exploitation.”[110] Since, on her factual findings, the relationship between the accused and the complainant was consensual and involved no exploitation, section 150.1 was unconstitutionally overbroad as applied to this accused. Accordingly, she stayed the proceedings.[111]

The trial judge’s approach appears to be a paradigmatic application of the strict version of the norm against overbreadth. She identified the purpose of the legislation in a way that was sufficiently distinct from the law’s operation to permit overbreadth analysis,[112] and she identified a case in which the application of the law was not directly necessary for its purpose. No resort to reasonable hypotheticals was required—on the trial judge’s factual findings, the “one person” to whom the application of the law was overbroad was the accused before her. Nevertheless, when the case eventually reached the Court of Appeal for Ontario, the law was found to be consistent with the norm against overbreadth.[113] The Court of Appeal for Ontario, per Justice Feldman, after considering the accused’s and the Crown’s submissions, said that “[t]he purpose of the law, stated succinctly, is to protect children from sexual contact with adults or the invitation to have sexual contact by adults.”[114] The Court of Appeal for Ontario added that the policy motivation for this purpose was “Parliament’s view that the inherent power imbalance between adults and children vitiates consensual sexual relations between them.”[115] The means of pursuing this objective was “to draw a bright-line age of protection of 16 years and to carve out a five-year close-in-age exception for non-exploitative conduct, where the defence of consent would be available.”[116] Justice Feldman observed that there was abundant precedent for pursuing a legislative objective by drawing a bright-line rule based on age, as long as the line drawn was “reasonable.”[117] Accordingly, the law was not overbroad.

There are, I think, two ways to understand this reasoning. On either understanding, if the strict version of the norm applies, it is erroneous; but if the relaxed version of the norm applies, it is completely defensible. First, we might read AB as sharply distinguishing between the objective (protecting children from sexual activity) from the background policy objective (prohibiting sexual activity in situations where power imbalances vitiate consent). But on this reading, the Court of Appeal for Ontario essentially construed the law as having the purpose of prohibiting the conduct that it prohibits and thus made an overbreadth analysis impossible. This reading is contrary to the caution in Moriarity that the objective should be stated in such a way that, at least in principle, permits overbreadth analysis. Second, we might read it as not separating the objective and the background policy so neatly, as saying that the purpose of the law is to prohibit sexual activity between children and adults in circumstances where a power imbalance vitiates consent. But on that reading, the accused’s overbreadth argument was strong. In a case where the accused’s conduct is non-exploitative (as on the facts of the case itself), there would be no power imbalance to protect children from and the law would be overbroad in relation to its protective purpose as applied to such a case. It cannot be the case that the passage of one day—the complainant’s 16th birthday—can instantly transform a sexual relationship from one that per se involves a power imbalance to one that, depending on the facts, may or may not. If it was not the day after the complainant’s 16th birthday, it might well not have been the day before. Ignoring that possibility is contrary to the “one person” rule, particularly where that one person was the accused before the Court of Appeal for Ontario.[118]

My point is not that the result in AB is wrong. I think it is correct. It is a good thing that the Court of Appeal for Ontario held that the bright-line rules established by section 150.1 of the Criminal Code are consistent with the principles of fundamental justice and, by extension, that legislative line-drawing exercises in general do not inevitably offend section 7 of the Charter. In the particular context of the offences against children, abandoning bright-line age-based rules in favour of a standard, such as lack of exploitation or abuse of power, would require a case-by-case inquiry into whether the purposes of the legislation were served by conviction in the particular case; that is, in every case where section 150.1 applies, there would have to be a case-by-case investigation into the factual circumstances of the sexual interaction between a child and an adult, including the maturity of the complainant. That is not a welcome prospect, not because courts are incapable of making those types of determinations, but because the increased complexity of those determinations would be undesirable, not only because it would make trials more complex, but also because it would significantly detract from the law’s function of guiding its subjects. My point is that the Court of Appeal for Ontario’s reasoning is inconsistent with the strict version of the norm against overbreadth stated in Bedford and reiterated in the majority decision in Ndhlovu. On the other hand, the reasoning is perfectly consistent with the relaxed version of the norm. Does section 150.1 go further than reasonably necessary to achieve its objectives (regardless of how precisely we state them) of protecting children from sexual activity with adults? When such objectives are pursued by means of a bright-line rule, it is inevitable that the rule will sometimes be applied where it is not directly necessary to do so; but that is only a problem if we retain the “one person” rule as the standard for overbreadth. The cost, in terms of the section 7 interests, is that some persons whose sexual relationships would be unlawful only because of their ages must wait for the passage of time until their ages align with the law before engaging in sexual activity. The relaxed version of the norm accepts that some degree of overbreadth of this kind may well be “reasonably necessary” to the law’s objectives, by avoiding both complex factual determinations and under-inclusiveness.

A good example of the third strategy—the assumption that the application of a standard will perfectly align with the purpose of applying the standard in every case—is furnished by R v. Boutilier.[119] The accused pleaded guilty to a number of (non-sexual) offences. The Crown brought a dangerous offender application under subsection 753(1) of the Criminal Code. To succeed, the Crown had to show that the accused posed the degree of dangerousness defined by paragraph 753(1)(a) on the basis of an assessment report and other evidence led at the hearing. The accused argued that subsection 753(1) was overbroad on its own terms because it permitted a judge to declare an offender dangerous who did not meet the criterion of dangerousness. Specifically, he argued that the statutory scheme limited the judge, at the stage of determining dangerousness, to considering evidence of the accused’s past patterns of behaviour;[120] that is, the sentencing judge could not consider whether any treatment or therapy the accused would receive in the future might mitigate his dangerousness until after declaring the offender to be dangerous; therefore, evidence on that issue could therefore not affect the finding of dangerousness.[121] The Supreme Court of Canada rejected this argument on grounds of statutory interpretation, holding that subsection 753(1) does and always has permitted a sentencing judge to consider both past behaviour and future dangerousness in deciding whether to declare an offender dangerous: “This, in turn, means that the designation provision is not overbroad as it does not capture offenders who, though currently a threat to others, may cease to be in the future, notably after successful treatment.”[122] The Supreme Court of Canada did not consider the possibility (though it does not appear to have been argued) that the judicial application of such a standard may nevertheless be erroneous—that is, that a judge may well find an offender to be dangerous when he is not—not because of any error in fact-finding, but because of the inherent imprecision in the application of standards to particular factual situations.[123] Errors of judgment in the application of a standard to given facts are unavoidable, but this implies that standards are, on the strict standard, inevitably overbroad. The point is not that a standard may be improperly administered; there are legal remedies for that, such as appeals and applications for judicial review.[124] The point is rather that any decision process, even when properly applied, will occasionally produce the wrong result. If some of those erroneous results affect the section 7 interests, then, according to the “one person” rule, the procedure is overbroad.

C. The Role of Section 1

Any infringement of a Charter right, even an infringement of section 7, can potentially be justified under section 1 as a reasonable limit if it satisfies the test established in R v. Oakes. The infringement will be justified if it has a pressing and substantial objective and if it is proportional in the sense that (i) it is rationally connected to its objective, (ii) it minimally impairs the Charter right in question, and (iii) its salutary effects on its objectives are proportionate to its deleterious effects on the Charter right.[125] There is an obvious similarity between the norm against overbreadth and the minimal impairment step of the Oakes test. Accordingly, there is also a similarity between the strict and relaxed versions of the norm and the more or less rigorous versions of the minimal impairment test that the Supreme Court of Canada has invoked over time.[126] Some have thought that the similarity is so strong that an overbroad law is necessarily not minimally impairing and so must necessarily fail the Oakes test.[127] But there is an important difference between the two legal doctrines. The norm against overbreadth is concerned with the relationship between the objectives of a law and its effects on the section 7 interests. The minimal impairment step of the Oakes test is concerned with the relationship between the objective of the infringement of a Charter right and the various ways (both infringing and non-infringing) in which that objective might be pursued.[128] While these two objectives may be related, they are not the same. Where a law is found to infringe section 7 because of an overbroad effect on the section 7 interests, the minimal impairment branch of the Oakes test must be put in terms of whether the section 7 violation—the fact that the law is overbroad—is the minimally impairing way of achieving whatever pressing and substantial purpose the legislature chose an overbroad law to achieve. At the section 1 stage, we know the law is overbroad; the question now is whether that overbreadth can be justified.

Thus, one way to deal with the very demanding nature of the strict version of the norm against overbreadth would be to accept that, because of the “one person” rule, it is indeed fairly easy to show that a law is overbroad and to look to the possibility of justification under section 1 of the Charter. For example, if, as I have argued, both bright-line rules and standards, when enacted for instrumental reasons, are almost certain to violate the strict version of the norm against overbreadth, then perhaps the place for arguments in favour of their constitutional validity is not in section 7 but in section 1. Perhaps the offence of stunt driving is overbroad and so violates section 7 of the Charter; but perhaps, in the context of highway traffic regulation, that overbreadth can be justified as a reasonable limit on the section 7 right not to be subjected to an overbroad law.

This is just what was suggested in Bedford. The applicants argued that the former offence of living on the avails of the prostitution of another was overbroad. While intended to target parasitic and exploitative relationships between pimps and prostitutes, it was drafted so broadly that it also penalized individuals who benefited from prostitution without being parasitic or exploitative, such as dependent relatives.[129] The government argued that, while that might be so, the law was necessarily overbroad because it was difficult to prove the element of exploitation that the offence targeted; thus, if limited to such circumstances, the offence would be underinclusive.[130] The Supreme Court of Canada held that the law was indeed overbroad and that the evidentiary issues raised by the government should be dealt with under section 1;[131] but when it got to the section 1 stage, the Supreme Court of Canada rejected the government’s claim that this consideration could justify an overbroad law because its very overbreadth meant that it was not minimally impairing of the section 7 right.[132]

Similarly, in Carter, the purpose of the prohibition on assisted suicide was found to be protecting the vulnerable. Under section 7, the question was whether the prohibition was overbroad in relation to that purpose. The Supreme Court of Canada found that it was. The questions under section 1 were related but different: What is the purpose of having an overbroad offence and is that overbroad offence a minimal impairment of the section 7 right? The government argued that the law needed to be overboard because of the evidentiary difficulty of distinguishing vulnerable and non-vulnerable candidates for medically assisted death: an objective that is related to, but distinct from, the objective of the prohibition itself. The Supreme Court of Canada recognized this difference and therefore, even though it had already found the law to be overbroad under section 7, carefully considered the government’s section 1 argument before rejecting it.[133]

Thus, despite the difference between the section 7 norm against overbreadth and the minimal impairment element of the Oakes test, rejecting a section 1 justification for an overbroad law has been the typical result because the Supreme Court of Canada is always inclined to say that the legislature could craft a law that was not overbroad and that would therefore necessarily be less infringing of the Charter right (because it would not infringe the section 7 right at all).[134] Indeed, there appears to be only one case where an overbroad law has been justified under section 1.[135] If, as I have suggested, the strict version of the norm is too rigorous, it appears that section 1 does not mitigate its rigours.

D. The Norm against Overbreadth as a Principle of Fundamental Justice

The Supreme Court of Canada has, over time, articulated a three-part test for recognizing a proposed principle as sufficiently basic to be a principle of fundamental justice. The proposed principle must be a legal principle that is “capable of being identified with some precision and applied to situations in a manner which yields an understandable result” and is “vital or fundamental to our societal norms of justice.”[136] Applying these three criteria, the strict version of the norm against overbreadth is not a principle of fundamental justice. It satisfies only one criterion: it can be and has been stated and applied with precision (indeed, its precision is the source of its flaws). But it is not a legal principle; that is, it is not a plausible legal principle because it is fundamentally contrary to two important features of legal order: the fact that not all laws are instrumental, and the fact that even instrumental laws are expressed as general norms. And it is hard to see how it can be vital to our social norms of justice when it says in effect that legal regulation by means of a bright-line rule—a pervasive and generally accepted means of legal regulation—is always constitutionally suspect.[137]

In contrast, the relaxed version of the norm recognizes that none of these aspects of regulation by law are inherently suspect; it does, however, insist that an instrumental law should not go further than reasonably necessary to achieve its purposes. The viability of the relaxed version of the norm as a constitutional principle is well illustrated both by cases where laws with an extraordinarily broad scope have been invalidated (such as Heywood and Carter) and by cases where laws that look overbroad from the perspective of the strict version (such as Sharma and AB) have been upheld.

III. A Suggested Approach to Overbreadth

The strict version of the norm against overbreadth is untenable. But it is possible to draw on the case law supporting the relaxed version of the norm that could control serious legislative overreach without the kind of strained reasoning we have seen in some of the cases discussed above. The first step would involve revisiting Moriarity. Where a law is challenged for overbreadth, a court should first ask whether it has the instrumental structure that Moriarity assumes; if not, the law would not be subject to challenge for overbreadth at all. That is, after all, essentially the conclusion reached in Sharma and NS; it would be much more straightforward to say so directly rather than go through an exercise that supposedly distinguishes means from ends but ultimately fails to do so.

On the other hand, if the law does have an instrumental structure—a well-defined purpose that is independent of the means chosen to achieve it—the question should be whether the law goes further than reasonably necessary to achieve its objectives. The laws that were at issue in Bedford and Ndhlovu and in the great bulk of environmental, occupational health and safety, and highway traffic legislation are of this kind. An effect on one person would not be sufficient to demonstrate overbreadth in this sense; bright-line rules that inevitably apply to more persons than strictly necessary for their purposes might nevertheless comply with this version of the norm if that degree of overbreadth was necessary for their proper operation—as would be typical for many regulatory and even criminal offences where it is important to provide clear boundaries between permitted and prohibited behaviour. Some laws might nevertheless be overbroad even under the relaxed standard. Without the relevant evidence and argument, it is impossible to anticipate precisely what the result of any given challenge would be. In general, the narrower the objective and the more sweeping the prohibition, the more likely the law would be to run afoul of even the relaxed version of the norm. If, as the Supreme Court of Canada found in Heywood, the purpose of former paragraph 179(1)(b) was to protect children from sexual offenders, then a sweeping prohibition on such persons being present in many different types of public spaces, regardless of the nature of the risk they posed or the nature of the space in question, would likely be overbroad for the reasons that Justice Cory gave. Similarly, if the Supreme Court of Canada correctly identified the purpose of the former prohibition on assisted suicide at issue in Carter, then the fact that the prohibition was exceptionless might well make it overbroad even on the relaxed version of the norm.

These two proposed modifications of the norm against overbreadth—that it not be applicable at all in some cases, and that it be construed less stringently in all others—undoubtedly means that the proper identification of the purpose is critical.[138] Governments may attempt to avoid the norm entirely by arguing that the laws they defend merely do what they do and that they do not have any instrumental purpose. And even laws that are undoubtedly instrumental will be easier to defend if their purposes can be defined in a sufficiently broad manner. But proper identification of purpose is already critical; moreover, both of these strategies are already available, and sometimes successful, even when courts supposedly apply the strict version of the norm. As noted above, the reasoning in Forcillo, AB, and NS (not to mention Sharma) essentially amounts to construing the purpose of the law as doing just what it does, rather than aiming at an independently definable purpose that is practically immune from overbreadth review. Furthermore, the reasoning in JJ construes the purpose of the law at issue in such a multi-faceted and general way that the Supreme Court of Canada has no difficulty deflecting the overbreadth challenge.[139] Adopting the relaxed version of the norm would have the virtue of making these strategies much more visible than they are now.

Laws that were found overbroad on the relaxed standard would remain very difficult to justify under section 1 because the kind of factors that have been mentioned as possible justifications would already have been considered in the application of the relaxed version of the norm. Consider, for example, the holding in Bedford that evidentiary difficulties in proving the offence of living on the avails should be considered under section 1 rather than section 7. On the relaxed version of the norm, these considerations would indeed be relevant under section 7 because they would go to the issue of whether the law was broader than reasonably necessary to achieve its purpose. If, as the Supreme Court of Canada found, the purpose was to criminalize exploitative relationships between pimps and sex workers, then the possibility that the law might, because of difficulties of proof, catch some relationships that were not exploitative would be relevant to whether the law was overbroad. That one (possibly hypothetical) relationship might be criminalized would not be enough. But depending on the evidence and argument, a court might still find the law to be overbroad, particularly where the consequences of compliance with the law were, as the Supreme Court of Canada found, so dangerous for sex workers attempting to carry on what was, at the time, a lawful activity. Such a violation of section 7 would be very hard to justify under section 1. The pressing and substantial purpose of the infringement would be a presumably more effective prosecution of the offence;[140] however, infringements with that kind of purpose have been very hard to justify, either because the Supreme Court of Canada can imagine a less impairing alternative (usually a straightforward exercise where the infringement is a form of overbreadth), or because, at the third stage of the Oakes test, the cost in terms of convicting the innocent is disproportionate to the benefit.[141]

Conclusion

The section 7 norm against overbreadth has played an important role in restraining the scope of certain extraordinarily sweeping laws and in provoking debate and reform around certain controversial issues of social policy. However, in its strict version, as articulated in the majority decision in Ndhlovu, the norm relies on untenable assumptions about how law works—the assumption that every law is instrumental to an objective external to legal order, and the assumption that a law can be drafted so that it applies directly only to cases that align with its purpose. The first is true of some laws but not others. The second assumption (though it may be trivially true of non-instrumental laws) is rarely if ever true because legal regulation operates through general norms, which, whether drafted in a rule-like or a standard-like way, will almost always be over- or underinclusive in their application. The relaxed version of the norm recognizes these features of legal order and so is more suitable as a constitutional principle. The strict version of the norm against overbreadth is itself overbroad. The relaxed version has an important role to play in constitutional jurisprudence.