Supreme Court of Canada

Appeal Heard: February 8, 2022 Judgment Rendered: October 28, 2022 Docket: 39360

Between: Eugene Ndhlovu — Appellant and His Majesty The King — Respondent

— and —

Attorney General of Canada, Attorney General of Ontario, Attorney General of Quebec, Attorney General of Nova Scotia, Attorney General of British Columbia, Criminal Lawyers' Association (Ontario), Canadian Civil Liberties Association, HIV & AIDS Legal Clinic Ontario, HIV Legal Network, Mobile Legal Clinic and Association québécoise des avocats et avocates de la défense — Interveners

Indexed as: R. v. Ndhlovu

2022 SCC 38

File No.: 39360.

2022: February 8; 2022: October 28.

Present: Wagner C.J. and Moldaver, Karakatsanis, Côté, Brown, Rowe, Martin, Kasirer and Jamal JJ.

On appeal from the Court of Appeal for Alberta

Constitutional law — Charter of Rights — Right to liberty — Fundamental justice — Remedy — Section 490.012 of Criminal Code requiring mandatory registration on national sex offender registry of offenders found guilty of designated sexual offences — Section 490.013(2.1) of Criminal Code requiring lifetime registration for offenders convicted of more than one designated offence — Whether provisions infringe right to liberty of offenders — If so, whether infringement justified — Appropriate remedy if right to liberty unjustifiably infringed — Constitution Act, 1982, s. 52(1) — Canadian Charter of Rights and Freedoms, ss. 1, 7 — Criminal Code, R.S.C. 1985, c. C‑46, ss. 490.012, 490.013(2.1) — Sex Offender Information Registration Act, S.C. 2004, c. 10.

In 2015, the accused pled guilty to two counts of sexual assault against two complainants with respect to assaults that had both occurred at a party in 2011. The sentencing judge imposed a global sentence of six months' imprisonment and three years' probation. After canvassing the accused's background and the evidence, the sentencing judge found that he was unlikely to reoffend. Despite this finding, due to his convictions on the two counts of sexual assault, the accused was subject, pursuant to ss. 490.012(1) and 490.013(2.1) of the Criminal Code, to mandatory lifetime registration in the national sex offender registry created by the Sex Offender Information Registration Act ("SOIRA"). Section 490.012 provides that SOIRA orders are mandatory for offenders convicted of designated offences such as sexual assault and other sexual offences. Section 490.013(2.1), added in 2011, mandates lifetime registration for individuals convicted of more than one designated offence. As well, previously existing prosecutorial discretion and judicial discretion to impose SOIRA orders were removed in 2011. Registered offenders are subject to many reporting requirements: they must report in person to a registration centre to provide extensive personal information, update their information in person yearly, and report in person at the registration centre any changes in primary or secondary address and name, or report if they receive a driver's licence or passport. Moreover, offenders must notify the registration centre within seven days of any change regarding their employment or volunteering information, and notify the registration centre if they intend to be away from their primary or secondary residence for seven or more consecutive days. Non‑compliance with any of these conditions may result in prosecution, with penalties of up to 2 years' imprisonment, up to $10,000 in fines, or both. Further, police officers conduct random compliance checks to verify the information on the registry. At a minimum, offenders are subject to at least one annual verification of their residential address.

The accused brought an application to challenge the constitutionality of both provisions. The sentencing judge concluded ss. 490.012 and 490.013(2.1) breached s. 7 of the Charter. She found that the provisions were overbroad because registering offenders with little or no recidivism risk, like the accused, did not advance SOIRA's purpose. She also found them grossly disproportionate, given the onerous cumulative effects of registering. She concluded s. 1 did not save the impugned provisions, and declared them of no force or effect. A majority at the Court of Appeal allowed the Crown's appeal, concluding that neither provision violated s. 7. It found that s. 490.012 was not overbroad since all convicted sex offenders have an increased propensity to commit sex crimes in the future. It also found that s. 490.013(2.1) was not overbroad because Parliament could infer that committing more than one sexual offence is a proxy for an increased recidivism risk, warranting a longer registration period. The majority further concluded that neither provision was grossly disproportionate. The dissenting judge disagreed that the provisions complied with s. 7. While she agreed the provisions were not grossly disproportionate, she found they were overbroad and concluded that the breach of s. 7 was not justified.

Held (Wagner C.J. and Moldaver, Côté and Brown JJ. dissenting in part): The appeal should be allowed.

Per Karakatsanis, Rowe, Martin, Kasirer and Jamal JJ.: Sections 490.012 and 490.013(2.1) of the Criminal Code infringe s. 7 of the Charter, and cannot be saved by s. 1. The provisions are therefore declared of no force or effect under s. 52(1) of the Constitution Act, 1982. The declaration in respect of s. 490.012 is suspended for one year and applies prospectively. However, the accused is exempted from the suspension of the declaration. As for s. 490.013(2.1), the declaration is immediate and applies retroactively.

In order to demonstrate a violation of s. 7 of the Charter, a claimant must first show that the law interferes with their life, liberty or security of the person. Liberty protects the right to make fundamental personal choices free from state interference. It also protects against physical restraint ranging from actual imprisonment or arrest to the use of state power to compel attendance at a particular place. The impact of a SOIRA order on an offender's liberty can only fairly be described as serious. SOIRA creates an ongoing obligation to report extensive information, subject to random checks and other compliance measures, under threat of prosecution and punishment by way of imprisonment, fines, or both. This creates continuous state monitoring that can last decades and, for some offenders, a lifetime. SOIRA also compels offenders to structure their travel and residency on an ongoing basis to remain in compliance with the legislation. There are thus burdens associated with the ongoing obligations to maintain the currency of the information on the registry, and the potential of imprisonment makes the deprivation of liberty even more severe. The impact on liberty can also be aggravated by an offender's life circumstances. Offenders whose job requires regular, prolonged travel will frequently need to take additional measures to remain in compliance. Even worse, offenders who experience homelessness, substance use issues, and cognitive or mental health challenges may find compliance extremely difficult. As a result, ss. 490.012 and 490.013(2.1) clearly interfere with offenders' liberty.

Once it is established that s. 7 is engaged, the next step is showing that the deprivation is inconsistent with the principles of fundamental justice, such as the principle of fundamental justice against overbreadth. The first step in an overbreadth analysis is to determine the purpose of the challenged provisions. The focus is on the purpose of the challenged provisions, not of the entire act in which they appear, although a correspondence between those purposes may sometimes occur. To determine an impugned law's purpose, courts may consider: statements of purpose in the legislation, if any; the text, context, and scheme of the legislation; and extrinsic evidence such as legislative history and evolution.

SOIRA's overall purpose is readily identified. As stated in s. 2(1), it is to help police services prevent and investigate crimes of a sexual nature by requiring the registration of certain information relating to sex offenders. Since SOIRA and ss. 490.012 and 490.013(2.1) of the Criminal Code form an integrated legislative scheme, SOIRA's overall purpose informs the interpretation of the challenged provisions' purpose. Further, the connection between the purpose of the provisions and SOIRA's overall purpose is reinforced by other provisions in SOIRA such as s. 2(2), which implies the provisions should be read as closely tied to the overall aim of assisting police. Taking into account legislative history and SOIRA's explicit overall purpose, the purpose of s. 490.012 is to capture information about offenders that may assist police to prevent and investigate sex offences. The means to achieve this purpose is mandatory registration. As for s. 490.013(2.1), no legislative history sheds light on its purpose. However, given the language of the provision, in the context of the existing scheme, Parliament's basis for seeking a longer period of access to information on offenders when more than one offence is committed must be that it believed that these offenders were more likely to reoffend relative to other sex offenders. Given this greater risk of harm, Parliament preferred to have the offender's information available on the registry as long as possible. This is consistent with Parliament's approach to target offenders who commit more serious offences. As such, s. 490.013(2.1) was designed to give police a longer period of access to information on offenders at a greater risk of reoffending. The means to achieve this purpose is lifetime registration for sex offenders who commit more than one designated offence.

Having identified the purpose of the challenged provisions, the next step is to determine whether they are overbroad. A law is overbroad when it is so broad in scope that it includes some conduct that bears no relation to its purpose, making it arbitrary in part. In other words, overbreadth addresses the situation where there is no rational connection between the purpose of the law and some, but not all, of its impacts. A law cannot deprive the life, liberty, or security of the person of even one individual in a way that is inconsistent with the principles of fundamental justice. As a consequence, laws that are broadly drawn to make enforcement more practical run afoul of s. 7 should they deprive the liberty of even one person in a way that does not serve the law's purpose.

Mandatory registration under s. 490.012 is overbroad as it leads to the registration of offenders who are not at an increased risk of committing a future sex offence. Registering such offenders bears no connection to the purpose of capturing information about offenders that may assist police prevent and investigate sex offences. In certain cases, an offender's personal circumstances mean they are not at an increased risk of reoffending, undermining any real possibility that their information on the registry will ever prove useful to police. A rough proxy like a prior conviction for a sexual offence does not readily account for those circumstances. Moreover, it is inaccurate to say that all sexual offenders are at an enhanced risk of reoffending. While a previous conviction for a sexual offence is a risk factor, about 10 percent of the individuals with such a conviction are not, at time of sentencing, at an enhanced risk of reoffending when compared to the general criminal population. Finally, the overbroad nature of s. 490.012 cannot be salvaged by the difficulty with risk assessments at sentencing. An argument based on enforcement practicality implicitly accepts that an individual's rights are breached but holds it is justified for the sake of a benefit to the public — making the administration or enforcement of a law more practical or convenient. Such an argument should be addressed under s. 1 of the Charter.

Lifetime registration of those convicted of more than one sexual offence is also overbroad. The purpose of the measure is to give police a longer period of access to information on offenders at a greater risk of reoffending. Yet, as the expert evidence establishes, committing more than one offence without an intervening conviction is not associated with an enhanced recidivism risk. As such, the measure captures some offenders who are not at a relatively greater risk of reoffending.

Sections 490.012 and 490.013(2.1) are not saved under s. 1 of the Charter. A breach of the Charter is justified under s. 1 when the challenged law has a pressing and substantial object and the means chosen are proportional to that object. The law is proportionate where the means adopted are rationally connected to the law's objective, minimally impairing of the right in question, and the law's salutary effects outweigh its deleterious effects.

The prevention and investigation of sex crimes is a pressing and substantial purpose, and the measures are rationally connected to their objectives. However, ss. 490.012 and 490.013(2.1) are not minimally impairing of an offender's rights. There are reliable, tailored alternatives available that would substantially achieve the challenged measures' objective. Restoring judicial discretion in the registration process would allow for a 90 percent inclusion rate of offenders in the registry. In addition, a variety of tools are available to improve the accuracy of judicial risk assessments, including expert evidence. Alternatively, Parliament can enumerate specific criteria to guide judges on when registering an offender is unlikely to advance the scheme's objective. Regarding s. 490.013(2.1), the Crown has not explained why exempting offenders who commit more than one offence without an intervening conviction would not achieve s. 490.013(2.1)'s purpose. Further, ss. 490.012 and 490.013(2.1)'s deleterious effects outweigh their salutary effects. The evidence on the provisions' benefits is sparse, whereas the deleterious impact on anyone who is subject to the reporting requirements of a SOIRA order is clear. The scope of the personal information registered, the frequency at which offenders are required to update their information and, above all, the threat of imprisonment, make the conditions onerous, especially on marginalized populations.

The appropriate remedy is a declaration of invalidity. With respect to s. 490.012, reading it down so that it would simply not apply to offenders who are not at an increased risk of reoffending or who suffer grossly disproportionate impacts would, in practice, reinstate judicial discretion and contradict Parliament's clear intention to remove all judicial discretion to exempt offenders at the time of sentencing from the registry. In addition, on balance, the circumstances justify a suspension of the declaration of invalidity for 12 months. Declarations of invalidity should be suspended when the government demonstrates that an immediately effective declaration of invalidity would endanger an interest of such great importance that, on balance, the benefits of delaying the effect of that declaration outweigh the cost of preserving an unconstitutional law that violates Charter rights. Although the rights violation that the suspension would temporarily prolong is significant and granting a suspension runs counter to the public's interest in legislation that complies with the Constitution, declaring s. 490.012 to be of no force or effect immediately would effectively preclude courts from imposing SOIRA orders on any offenders, including those at high risk of recidivism, and could therefore endanger the public interest in preventing and investigating sexual offences committed by high-risk offenders, undermining public safety. Furthermore, the declaration should apply prospectively. A retroactive application of the declaration could frustrate the compelling public interests that require a period of transition, creating uncertainty and removing the protection that justifies the suspension in the first place. A prospective declaration would not unduly prejudice offenders who have been registered since 2011 but whose rights under s. 7 are still violated. Those offenders will be able to ask for a personal remedy pursuant to s. 24(1) of the Charter in order to be removed from the registry if they can demonstrate that SOIRA's impacts on their liberty bears no relation or is grossly disproportionate to the objective of s. 490.012.

With respect to s. 490.013(2.1), an immediate declaration of invalidity is appropriate given that the offenders will remain registered and there is no gap for Parliament to fill. Furthermore, there is no compelling reason to rebut the presumption of retroactive application of the declaration. Because the declaration affects all those impacted by the enactment of the provision since 2011, offenders who are subject to a lifetime order pursuant to this provision after having been convicted of more than one sexual offence without an intervening conviction can seek a s. 24(1) remedy to change the length of their registration.

Per Wagner C.J. and Moldaver, Côté and Brown JJ. (dissenting in part): The appeal should be allowed in part. There is agreement with the majority that mandatory lifetime registration in s. 490.013(2.1) of the Criminal Code is overbroad. However, s. 490.012 is constitutional. It is appropriately tailored to its purpose of helping the police prevent and investigate sexual crimes, and does not limit an offender's s. 7 rights in a manner that bears no connection to its objective. The exercise of judicial discretion to exempt offenders from registration under SOIRA was the very problem that prompted Parliament to amend the Criminal Code in 2011 to provide for automatic registration of sex offenders. Many judges had exercised their discretion to exempt offenders in a manifestly improper manner, and the Registry's low inclusion rate undermined its efficacy. The evidence is clear that even low risk sex offenders, relative to the general criminal population, pose a heightened risk to commit another sexual offence. It is also clear that it cannot be reliably predicted at the time of sentencing which offenders will reoffend. In the face of that uncertain risk, Parliament was entitled to cast a wide net.

The s. 7 analysis first requires a determination of the impugned provision's purpose. Section 490.012 contains no explicit statement of the purpose of automatic registration, but the broader legislative scheme offers some clues. Section 2(1) of SOIRA states that the purpose of that Act is to help police services prevent and investigate crimes of a sexual nature by requiring the registration of certain information relating to sex offenders, and s. 2(2) describes the guiding principles of the legislation. Extrinsic evidence reveals that Parliament amended the statutory scheme in response to concerns that the Registry's efficacy was compromised by the exclusion of nearly half of all convicted sex offenders because the Crown prosecutor had not made an application for their registration or judges had concluded that they should not be registered. The amendments responded to those concerns by providing for automatic registration of sex offenders in s. 490.012 to make the Registry as effective as possible when police are investigating crimes of a sexual nature. The foregoing evidence supports the conclusion that the purpose of s. 490.012(1) is to help police prevent and investigate sexual crimes by requiring the registration of certain information relating to sex offenders, which conforms to the overall statutory objective.

Automatic registration is not arbitrary. Arbitrariness describes the absence of a rational connection between the law's purpose and the impugned effect on the individual, or where it can be shown that the impugned effect undermines the objective of the law. Mandatory registration of convicted sex offenders in SOIRA is not arbitrary since there is a clear connection between having accurate and up-to-date information about persons more likely to commit sexual offences and the investigation and prevention of sexual crimes.

Automatic registration is not grossly disproportionate. Gross disproportionality will be found where the law's effects on life, liberty or security of the person are so grossly disproportionate to its purposes that they cannot rationally be supported. SOIRA imposes a burden on offenders, but that burden is not totally out of sync with the objective of investigating and preventing sexual offences. There are also a number of arguments militating against gross disproportionality. Access to the Registry is strictly controlled. The reporting is confidential and use of the information is strictly limited to police for the prevention and investigation of sexual offences. Also, any stigma experienced by an offender from being labelled a sex offender flows from the convictions themselves, not registration. Anxiety caused by a law is not usually considered an infringement of liberty. Further, SOIRA's requirements do not amount to punishment since they do not significantly limit lawful activities in which the accused can engage, where the accused can go, or with whom an accused can communicate or associate. An offender who is placed on the Registry has already been tried and convicted of a designated sexual offence. A SOIRA order is thus characterized as a consequence of conviction. Finally, the duration of a SOIRA order is directly connected to the maximum term of imprisonment for that sexual offence. By linking the length of time for reporting to the severity of the offence, Parliament built proportionality into the legislative scheme. If there is truly a disproportionate impact on privacy or liberty, a termination order will be available for offenders who can meet the high standard of demonstrating it.

Section 490.012 is not overbroad. A law will be overbroad where it captures some conduct unrelated to its purpose, recognizing that a law can be rational in some cases but overreach in others. In enacting s. 490.012, Parliament deliberately chose not to distinguish between more serious and less serious sexual offences or higher risk and lower risk offenders. Instead, Parliament required registration for all sex offenders based on a shared characteristic: a heightened risk of committing a future sexual offence. The government can enact legislation that treats all individuals with a common characteristic in the same manner, without offending s. 7, provided there is a rational connection between that characteristic and the government's purpose.

There is a logical link between automatic registration on the basis of a sex conviction and the purpose of s. 490.012. The expert evidence indicates that offenders convicted of a sexual offence are five to eight times more likely to reoffend than those convicted of a non-sexual offence. As a group, sex offenders will always pose a greater risk than the rest of the population for engaging in that activity and, thus, a prior conviction for a sex offence is a reliable indication of risk and a proper method of assessing that risk. Further, the experts agreed that the recidivism risk cannot be determined with certainty at the time of sentencing. A risk assessment cannot guarantee whether any individual will reoffend, and observed recidivism rates underestimate the true rates of sexual reoffending. As such, it is dangerous to use a risk-based assessment to determine which offenders should be registered. It was within Parliament's purview to draw a line based on that known increased risk of unknown degree, rather than leaving it to prosecutors and judges to weigh if an offender poses an increased risk in each case. An important reason that Parliament drew that line was to fix a flaw in the Registry that allowed judges to exempt offenders who they deemed not to be predators, contrary to the legislative intent. Prior to the amendments, some judges granted exemptions based not on the impact of registration on the offender (as required by the statutory provision) but on whether the offender was the type of person for whom the Registry was intended — in other words, a real sex offender. The category of real sex offender has been, at times, defined so narrowly as to exclude offenders who sexually assaulted people they knew, child pornography users, opportunistic offenders, and historic offenders. Judges have granted exemptions even where the victim was a stranger and the offences were highly predatory. Exemptions have also been granted to offenders who occupied positions of trust and abused vulnerable victims. Given what appears to have been a persistent, routine failure to appreciate the seriousness of these offences, it can be confidently predicted that the rampant misuse of judicial discretion prior to the amendments will recur once automatic registration is removed.

Cases Cited

By Karakatsanis and Martin JJ.

Considered: Carter v. Canada (Attorney General), 2015 SCC 5, [2015] 1 S.C.R. 331; R. v. Safarzadeh-Markhali, 2016 SCC 14, [2016] 1 S.C.R. 180; Canada (Attorney General) v. Bedford, 2013 SCC 72, [2013] 3 S.C.R. 1101; R. v. T.L.B., 2006 ABQB 533, 403 A.R. 293, aff'd 2007 ABCA 135, 404 A.R. 283; R. v. Heywood, [1994] 3 S.C.R. 761; referred to: R. v. Long, 2018 ONCA 282, 45 C.R. (7th) 98; R. v. Debidin, 2008 ONCA 868, 94 O.R. (3d) 421; R. v. Dyck, 2008 ONCA 309, 90 O.R. (3d) 409; R. v. Cross, 2006 NSCA 30, 241 N.S.R. (2d) 349; R. v. C. (S.S.), 2008 BCCA 262, 234 C.C.C. (3d) 365; R. v. J.D.M., 2006 ABCA 294, 417 A.R. 186; R. v. Desmeules, 2006 QCCQ 16773; Blencoe v. British Columbia (Human Rights Commission), 2000 SCC 44, [2000] 2 S.C.R. 307; R. v. Clay, 2003 SCC 75, [2003] 3 S.C.R. 735; R. v. Malmo‑Levine, 2003 SCC 74, [2003] 3 S.C.R. 571; Fleming v. Ontario, 2019 SCC 45, [2019] 3 S.C.R. 519; R. v. Beare, [1988] 2 S.C.R. 387; R. v. D.T.; R. v. Callahan; R. v. Firingstoney, 2017 ABQB 343; R. v. Caruana, 2016 ONCJ 367; R. v. G.E.W., 2006 ABQB 317, 396 A.R. 149; Re B.C. Motor Vehicle Act, [1985] 2 S.C.R. 486; R. v. K.R.J., 2016 SCC 31, [2016] 1 S.C.R. 906; R. v. Moriarity, 2015 SCC 55, [2015] 3 S.C.R. 485; RJR-MacDonald Inc. v. Canada (Attorney General), [1995] 3 S.C.R. 199; Singh v. Minister of Employment and Immigration, [1985] 1 S.C.R. 177; R. v. Sharpe, 2001 SCC 2, [2001] 1 S.C.R. 45; R. v. Oakes, [1986] 1 S.C.R. 103; Ontario (Attorney General) v. G, 2020 SCC 38, [2020] 3 S.C.R. 629; Toronto Star Newspapers Ltd. v. Canada, 2010 SCC 21, [2010] 1 S.C.R. 721; R. v. Appulonappa, 2015 SCC 59, [2015] 3 S.C.R. 754; Health Services and Support — Facilities Subsector Bargaining Assn. v. British Columbia, 2007 SCC 27, [2007] 2 S.C.R. 391; Alberta v. Hutterian Brethren of Wilson Colony, 2009 SCC 37, [2009] 2 S.C.R. 567; Thomson Newspapers Co. v. Canada (Attorney General), [1998] 1 S.C.R. 877; R. v. Boudreault, 2018 SCC 58, [2018] 3 S.C.R. 599; Schachter v. Canada, [1992] 2 S.C.R. 679; R. v. Albashir, 2021 SCC 48, [2021] 3 S.C.R. 531.

By Brown J. (dissenting in part)

Canada (Attorney General) v. Bedford, 2013 SCC 72, [2013] 3 S.C.R. 1101; R. v. Michaud, 2015 ONCA 585, 127 O.R. (3d) 81; R. v. Long, 2018 ONCA 282, 45 C.R. (7th) 98; R. v. Redhead, 2006 ABCA 84, 384 A.R. 206; R. v. T.A.S., 2018 SKQB 183; R. v. Dyck, 2008 ONCA 309, 90 O.R. (3d) 409; R. v. Debidin, 2008 ONCA 868, 94 O.R. (3d) 421; R. v. Cross, 2006 NSCA 30, 241 N.S.R. (2d) 349; R. v. C. (S.S.), 2008 BCCA 262, 234 C.C.C. (3d) 365; R. v. K.R.J., 2016 SCC 31, [2016] 1 S.C.R. 906; R. v. Hooyer, 2016 ONCA 44, 129 O.R. (3d) 81; R. v. Moriarity, 2015 SCC 55, [2015] 3 S.C.R. 485; R. v. Stillman, 2019 SCC 40, [2019] 3 S.C.R. 144; R. v. F. (P.R.) (2001), 57 O.R. (3d) 475; R. v. Beare, [1988] 2 S.C.R. 387; R. v. Dyck (2005), 203 C.C.C. (3d) 365; R. v. Safarzadeh-Markhali, 2016 SCC 14, [2016] 1 S.C.R. 180; R. v. B.P.M., 2019 BCPC 156; Carter v. Canada (Attorney General), 2015 SCC 5, [2015] 1 S.C.R. 331; C. (P.S.) v. British Columbia (Attorney General), 2007 BCSC 895, 222 C.C.C. (3d) 230; R. v. Sharpe, 2001 SCC 2, [2001] 1 S.C.R. 45; R. v. Malmo-Levine, 2003 SCC 74, [2003] 3 S.C.R. 571; R. v. Have, 2005 ONCJ 27, 194 C.C.C. (3d) 151; R. v. Burke, 2005 ONCJ 422; R. v. Y. (B.T.), 2006 BCCA 331, 210 C.C.C. (3d) 484; R. v. B.S.S., 2006 BCPC 135; R. v. Worm, 2005 ABPC 92; R. v. Randall, 2006 NSPC 38, 247 N.S.R. (2d) 205; R. v. Aldea, 2005 SKQB 461, 271 Sask. R. 272; R. v. S. (M.W.), 2007 BCSC 1188, 52 C.R. (6th) 77; R. v. Friesen, 2020 SCC 9, [2020] 1 S.C.R. 424; R. v. Mabior, 2012 SCC 47, [2012] 2 S.C.R. 584.

Statutes and Regulations Cited

Bill C-34, Protecting Victims From Sex Offenders Act, 2nd Sess., 40th Parl., 2009.

Canadian Charter of Rights and Freedoms, ss. 1, 7, 11, 12, 24(1).

Christopher's Law (Sex Offender Registry), 2000, S.O. 2000, c. 1, s. 2.

Constitution Act, 1982, s. 52(1).

Criminal Code, R.S.C. 1985, c. C‑46, ss. 490.011(1) "designated offence" (a), 490.012 [repl. 2010, c. 17, s. 5], 490.013 [idem, s. 6], 490.015(1), 490.016(1), 490.031(1).

Protecting Victims From Sex Offenders Act, S.C. 2010, c. 17.

Sex Offender Information Registration Act, S.C. 2004, c. 10, ss. 2 [repl. 2010, c. 17, s. 28], 4, 4.1, 5, 5.1, 6(1), 8, 16(2)(a) [idem, s. 44].

Authors Cited

Benedet, Janine. "A Victim-Centred Evaluation of the Federal Sex Offender Registry" (2012), 37 Queen's L.J. 437.

Benedet, Janine. "Long and Ndhlovu: The Federal Sex offender Registry and Section 7 of the Charter" (2018), 45 C.R. (7th) 132.

Canada. Government Response to the Fifth Report of the Standing Committee on Public Safety and National Security Entitled Statutory Review of the Sex Offender Information Registry Act, April 12, 2010 (online: https://www.ourcommons.ca/DocumentViewer/en/40-2/SECU/report-5/response-8512-403-180; archived version: https://www.scc-csc.ca/cso-dce/2022SCC-CSC38_1_eng.pdf).

Canada. House of Commons. House of Commons Debates, vol. 144, No. 67, 2nd Sess., 40th Parl., June 3, 2009, p. 4148.

Canada. House of Commons. House of Commons Debates, vol. 144, No. 70, 2nd Sess., 40th Parl., June 8, 2009, p. 4324.

Canada. House of Commons. House of Commons Debates, vol. 145, No. 112, 3rd Sess., 40th Parl., December 7, 2010.

Canada. House of Commons. Standing Committee on Public Safety and National Security. Evidence, No. 15, 2nd Sess., 40th Parl., April 21, 2009, pp. 3‑4, 8‑9.

Canada. House of Commons. Standing Committee on Public Safety and National Security. Statutory Review of the Sex Offender Information Registry Act: Report of the Standing Committee on Public Safety and National Security, 2nd Sess., 40th Parl., December 2009.

Canada. Library of Parliament. Parliamentary Research Branch. Bill C-16: Sex Offender Information Registration Act, Legislative Summary 470E, by Robin MacKay, Law and Government Division, February 16, 2004.

Canada. Senate. Standing Senate Committee on Legal and Constitutional Affairs. Proceedings of the Standing Senate Committee on Legal and Constitutional Affairs, No. 3, 3rd Sess., 40th Parl., April 14, 2010, p. 32.

Stewart, Hamish. Fundamental Justice: Section 7 of the Canadian Charter of Rights and Freedoms, 2nd ed. Toronto: Irwin Law, 2019.

Sullivan, Ruth. The Construction of Statutes, 7th ed. Toronto: LexisNexis, 2022.

APPEAL from a judgment of the Alberta Court of Appeal (Slatter, Schutz and Khullar JJ.A.), 2020 ABCA 307, 12 Alta. L.R. (7th) 225, 392 C.C.C. (3d) 459, 66 C.R. (7th) 34, 466 C.R.R. (2d) 151, [2021] 1 W.W.R. 537, [2020] A.J. No. 914 (QL), 2020 CarswellAlta 1573 (WL), setting aside two decisions of Moen J., 2018 ABQB 277, 68 Alta. L.R. (6th) 89, 45 C.R. (7th) 137, [2018] 6 W.W.R. 590, [2018] A.J. No. 427 (QL), 2018 CarswellAlta 794 (WL), and 2016 ABQB 595, 44 Alta. L.R. (6th) 382, 32 C.R. (7th) 392, 366 C.R.R. (2d) 20, [2017] 3 W.W.R. 343, [2016] A.J. No. 1105 (QL), 2016 CarswellAlta 2054 (WL). Appeal allowed, Wagner C.J. and Moldaver, Côté and Brown JJ. dissenting in part.

Counsel

Elvis Iginla, for the appellant.

Jason R. Russell, for the respondent.

Jeffrey G. Johnston, for the intervener the Attorney General of Canada.

Michael S. Dunn, for the intervener the Attorney General of Ontario.

Sylvain Leboeuf and Julie Dassylva, for the intervener the Attorney General of Quebec.

Written submissions only by Glenn Hubbard, for the intervener the Attorney General of Nova Scotia.

Lesley A. Ruzicka, K.C., for the intervener the Attorney General of British Columbia.

Mark C. Halfyard, for the intervener the Criminal Lawyers' Association (Ontario).

Christine Mainville and Carly Peddle, for the intervener the Canadian Civil Liberties Association.

Matthew R. Gourlay, Ryan Peck and Robin Nobleman, for the interveners the HIV & AIDS Legal Clinic Ontario and the HIV Legal Network.

Donald Tremblay, for the intervener the Mobile Legal Clinic.

Stéphanie Pelletier-Quirion, for the intervener Association québécoise des avocats et avocates de la défense.

Joint Reasons for Judgment

(paras. 1 to 143)

Karakatsanis and Martin JJ. (Rowe, Kasirer and Jamal JJ. concurring) —

I. Overview

[ 1 ] Parliament and courts have increasingly recognized the grave harms which flow from the wide variety of sexual offences prohibited in the Criminal Code, R.S.C. 1985, c. C-46. Over the years, the substantive elements of some sexual offences have been modified: consent is now expressly defined in the Criminal Code; procedures were introduced to address thorny evidentiary questions; testimonial assistance is available to vulnerable witnesses; and sentencing provisions and principles reflect the seriousness of sexual offences.

[ 2 ] In 2004, concerned about the sexual abuse and abduction of children, Parliament passed the Sex Offender Information Registration Act, S.C. 2004, c. 10 (SOIRA or Act). Through this legislation, Parliament sought to help police investigate crimes of a sexual nature by creating a national sex offender registry. At the time, a sex offender would only be placed on the registry if the Crown prosecutor first chose to apply to the court for an order requiring the offender to comply with SOIRA. Moreover, the legislation gave sentencing judges the discretion to exclude offenders from the registry if the effects of the order on their privacy or liberty interests were grossly disproportionate to the public interest in protecting society.

[ 3 ] These two separate safeguards were removed in 2011 following the enactment of the Protecting Victims From Sex Offenders Act, S.C. 2010, c. 17. Instead, s. 490.012 of the Criminal Code now requires the mandatory registration of all offenders who have been found guilty of any one of the 27 different sexual offences designated in s. 490.011(1)(a). Now each and every such sexual offender is compelled to register their personal information on Canada's national sex offender registry, regardless of their individual risk of reoffending. In addition to compulsory SOIRA orders, Parliament also imposed a mandatory lifetime registration for offenders who commit more than one offence, irrespective of the nature or timing of the offences and even if they are part of the same transaction (s. 490.013(2.1)).

[ 4 ] In this case, the appellant, Eugene Ndhlovu, pled guilty in 2015 to two counts of sexual assault against two complainants at a party in 2011. He was 19 years old at the time. At sentencing, the judge was tasked with tailoring a proportionate sentence that was fit in relation to both Mr. Ndhlovu and the sexual assaults he committed. After canvassing his background and the evidence, the judge found that Mr. Ndhlovu was unlikely to reoffend. However, due to Parliament's amendments in 2011, the Criminal Code obliged the judge to issue an order requiring Mr. Ndhlovu to comply with SOIRA, and for the rest of his life.

[ 5 ] As a result, like all other such offenders, he would be required to report to a police station and forced to supply extensive personal information which would be placed on Canada's national sex offender registry. SOIRA also imposes ongoing reporting requirements which are numerous, invasive and extensive; including that offenders must keep their information up to date, report their plans for any travel lasting seven or more consecutive days and report any change to their home or employment address. He would have to report annually to the police and be subject to random police checks. Non-compliance with any of the reporting obligations associated with registration carries the threat of prosecution, a maximum of two years' imprisonment, a fine, or both (Criminal Code, s. 490.031(1)). His presence in the database would mean he would be among the list of persons police may consider to be of interest in their investigations, which may generate further interactions with the police. The impact on Mr. Ndhlovu and anyone subject to these provisions is considerable. The scope of the personal information registered, the frequency at which offenders are required to update their information, the ongoing monitoring by the state, and the threat of prosecution and imprisonment all interfere with what it means to be free in Canada.

[ 6 ] This appeal requires this Court to determine whether Parliament complied with the Canadian Charter of Rights and Freedoms when it chose to remove prosecutorial and judicial discretion from s. 490.012 and introduced, under s. 490.013(2.1), lifetime registration for offenders convicted of more than one designated sexual offence. Through s. 490.012, Parliament sought to capture information about offenders that may assist police prevent and investigate sexual offences. Similarly, s. 490.013(2.1) is designed to give police a longer period of access to information on offenders at a greater risk of reoffending.

[ 7 ] Even when Parliament acts with a laudable purpose, it must still legislate in a constitutional manner and comply with the Charter. It failed to do so when it enacted ss. 490.012 and 490.013(2.1). These measures infringe the liberty interest under s. 7 of the Charter because registration has a serious impact on the freedom of movement and on the freedom to make fundamental choices of people who are not at an increased risk of reoffending over their lifetime.

[ 8 ] Because the mandatory registration of those offenders who are not at an increased risk of reoffending does not assist police, it is inconsistent with the principle of fundamental justice against overbreadth. Mandatory and lifetime registration overshoot the mark: subjecting sex offenders who do not have an increased risk of reoffending to obligatory reporting requirements is not connected to Parliament's purpose of capturing information that assists police prevent and investigate sex offences. Requiring lifetime registration also goes too far and denies the rights of some individuals in a way that bears no relation to Parliament's objective.

[ 9 ] There are offenders who, because of their individual characteristics, are at a negligible risk of reoffending. Further, the reality is that 75 to 80 percent never reoffend. Based on the Crown's statistical evidence, there are also a significant number of sex offenders who are at no greater risk of reoffending than members of the general criminal population. As a result, s. 490.012 applies to offenders for whom there is no real possibility that their information may ever assist police — and there is no discretion to exclude such persons from the wide reach of SOIRA's onerous and ongoing obligations. In addition, the Crown's expert evidence established that committing more than one sexual offence without an intervening conviction is not associated with a greater risk of reoffending.

[ 10 ] The two challenged provisions, therefore, suffer from the same constitutional defect. They both use categorical and unyielding proxies that are too broad, resulting in the measures casting too wide a net. To the extent they require the registration, sometimes for life, of offenders who demonstrate no increased risk of reoffending, they threaten the liberty interests of offenders in a manner which is overbroad and violates s. 7 of the Charter.

[ 11 ] Nor are they justified under s. 1 because they are not minimally impairing of Charter rights and the deleterious effects of the provisions outweigh their salutary ones. The blanket and blunt requirement that all designated sex offenders must be registered, and those convicted of more than one offence must be registered for life, restricts the liberty of offenders who are not at an increased risk of reoffending without any evidence that doing so enhances the ability of police to prevent and investigate sex crimes. While the Crown has asserted that it believes it is necessary to include all offenders for the registry to be as effective as the Crown wants it to be, any such avowal is insufficient to meet its burden of proof under which it is required to justify, not merely explain, the infringement on liberty. Critically, the Crown has adduced no evidence that demonstrates how these provisions are effective in helping police prevent and investigate sex crimes. Indeed, the sparse information in the record points in the opposite direction.

[ 12 ] We would allow the appeal and declare ss. 490.012 and 490.013(2.1) of no force or effect under s. 52(1) of the Constitution Act, 1982. A one-year suspension of the declaration is appropriate for mandatory registration, given concerns about public safety and the many ways Parliament could remedy the provision's overbreadth. An immediate declaration, however, is warranted for lifetime registration for offenders convicted of more than one offence.

II. Facts and Judicial History

A. Facts

[ 13 ] Mr. Ndhlovu pled guilty to two counts of sexual assault in June 2015. The convictions resulted from sexual assaults charged on a single indictment against two complainants at a house party in 2011. Mr. Ndhlovu touched both of the complainants' buttocks and one of the complainant's thighs. Later in the evening, one of the complainants awoke to Mr. Ndhlovu inserting his fingers inside her vagina. After she motioned to him to stop, he tried to reinsert his fingers. The complainant pushed him and told him to stop, but Mr. Ndhlovu instead tried to remove her bra. The complainant once again told him to stop and he fled. Mr. Ndhlovu was 19 years old at the time of the offences.

B. Judicial History

(1) First Instance Proceedings

[ 14 ] Three proceedings before Moen J. are pertinent to this appeal.

[ 15 ] The first was a sentencing hearing in which Moen J. was tasked with tailoring a fit and proportional punishment. She found that six months' imprisonment and three years' probation were warranted in all the circumstances. In reaching this conclusion, she took into account the characteristics of the offender, as well as the gravity of the offences he committed. She received evidence concerning Mr. Ndhlovu, including a presentence report. She noted that Mr. Ndhlovu took responsibility for his actions and was remorseful. He had no criminal history. His offences related to excessive alcohol consumption, but he had stopped drinking to excess. He had the support of his family and community.

[ 16 ] Importantly, based on the evidence placed before her by the Crown and defence, the sentencing judge found that Mr. Ndhlovu was "unlikely to offend again" (A.R., vol. II, at p. 38). Moen J. stated that he "will be safe to release into the community. I have absolutely no concerns that [he] will re-offend. Nor does the Crown suggest that [he] will" (p. 38).

[ 17 ] Despite this finding, due to his conviction for two designated offences, Mr. Ndhlovu was subject to mandatory lifetime registration in Canada's national sex offender registry pursuant to ss. 490.012(1) and 490.013(2.1) of the Criminal Code. Following sentencing, Mr. Ndhlovu brought an application to challenge both provisions as contrary to ss. 7 and 12 of the Charter.

[ 18 ] The second hearing addressed whether this mandatory lifetime registration breached those Charter rights. The Crown called the evidence of Det. Arlene May Hove, a police investigator with the Edmonton Police Service (EPS), who was responsible for administering access to the database for police in the Edmonton area.

[ 19 ] Moen J. concluded ss. 490.012 and 490.013(2.1) breached s. 7 of the Charter (2016 ABQB 595, 44 Alta. L.R. (6th) 382 (ABQB reasons (2016))). The purpose of the provisions was "to protect vulnerable people including children in society, by allowing police quick access to current information on convicted sex offenders" (para. 87). The provisions deprived an offender's liberty and the deprivation was "quite onerous" given the depth of information, the continuing obligation to report changes, the annual in-person reporting requirements, and the consequences of breaching the order, along with random checks and registration for life (para. 52). She also noted the stigma of being on the registry, the fear that this information may not be kept confidential and how random compliance checks at home and at work risked divulging their registration status.

[ 20 ] The provisions offended the principles of fundamental justice. While not arbitrary, the provisions were overbroad: registering offenders with little or no recidivism risk, like the appellant, did not advance SOIRA's purpose. The measures were therefore "broader than necessary" (para. 116). They were also grossly disproportionate, given the onerous cumulative effects of registering. Having found the provisions breached s. 7, Moen J. declined to address the defence's arguments on s. 12 of the Charter.

[ 21 ] Following Moen J.'s s. 7 ruling, the Crown sought to justify the provisions under s. 1 of the Charter. A third hearing was held to address whether the provisions were justified under s. 1 and, at this time, the Crown submitted expert evidence from Dr. Robert Karl Hanson about recidivism rates and the risk associated with sexual offenders subsequently committing further offences. The defence called Dr. Kristen Marie Zgoba who gave evidence on recidivism rates and also addressed the efficacy and impact of sex offender registries.

[ 22 ] Moen J. concluded s. 1 did not save the impugned provisions (2018 ABQB 277, 68 Alta. L.R. (6th) 89). She agreed that the public interest addressed by Parliament was the protection of society through the effective and quick investigation of crimes of a sexual nature by providing police with rapid access to information about known sex offenders. While this was a sufficiently pressing and substantial objective, the means chosen to achieve it was not proportional. Removing judicial discretion and requiring mandatory and sometimes lifetime registration was not rationally connected to this objective, in part because there was no evidence before either her, or the Parliamentary Committee and Senate Committee studying the 2011 amendments, that "there would be more arrests made more quickly as a result of the 2011 amendments" (para. 44). The removal of judicial discretion was not minimally impairing because the Crown "produced no evidence that suggested that judicial discretion had caused any difficulty for the police in their investigations of sexual offences" (para. 110). In addition, the evidence established that the mandatory imposition of lifetime registration does not minimally impair the rights of persons convicted of at least two sexual offences. When comparing registration's proven costs with the claimed benefits of having everyone on the registry, Moen J. relied heavily on the evidence of the Crown expert to conclude that the deleterious effects of the impugned provisions outweighed any salutary effects. Thus, she declared ss. 490.012 and 490.013(2.1) of no force or effect, and declined to order Mr. Ndhlovu to register.

(2) Court of Appeal

[ 23 ] A majority at the Court of Appeal of Alberta (Slatter and Schutz JJ.A.) allowed the Crown's appeal, concluding that neither provision violated s. 7 (2020 ABCA 307, 12 Alta. L.R. (7th) 225). The majority held the purpose of s. 490.012 was to "require mandatory registration of all sex offenders convicted of designated offences" (para. 74). The fact that all convicted sex offenders have an increased propensity to commit sex crimes in the future provides the necessary connection between mandatory registration under s. 490.012 and the purpose of the legislation. Thus, the measure was not overbroad, since a certain degree of recidivism risk was not a "necessary prerequisite" (para. 95).

[ 24 ] Lifetime registration was also not overbroad. The majority held the purpose of s. 490.013(2.1) was "to further public safety by subjecting sex offenders who are at enhanced risk of re-offending to a longer period of registration" (para. 103, quoting R. v. Long, 2018 ONCA 282, 45 C.R. (7th) 98, at para. 102). Parliament could infer that committing more than one sexual offence is a proxy for an increased recidivism risk, warranting a longer registration period. Nor were ss. 490.012 and 490.013(2.1) grossly disproportionate. If not strictly modest, the measures were "reasonable and not dissimilar to many other reporting obligations that routinely occur as part of the everyday life of all Canadians, who are state-compelled to provide information from time to time" (para. 130).

[ 25 ] Khullar J.A., in dissent, disagreed that the provisions complied with s. 7. While she agreed the provisions were not grossly disproportionate, they were overbroad. The provisions had the same purpose as the overall Act: to assist law enforcement in preventing and investigating sexual crimes by centralizing information about convicted sex offenders. It was unnecessary to register offenders at negligible risk of reoffending, like the appellant, to further the provisions' objective. The Crown's argument that the difficulties of predicting risk necessitated the registration of all offenders should be addressed under s. 1, as Moen J. concluded. Yet, there was no evidence to show the measures' salutary effects outweighed their deleterious effects. The breach of the Charter right was not justified.

III. Issues

[ 26 ] Mr. Ndhlovu appeals on the ground that the Court of Appeal erred in its determination that ss. 490.012 and 490.013(2.1) of the Criminal Code comply with s. 7 of the Charter. As such, the following issues arise on this appeal:

(a) Do ss. 490.012 and 490.013(2.1) of the Criminal Code breach s. 7 of the Charter?

(b) If so, are the breaches justified under s. 1 of the Charter?

(c) If they are not justified, what is the appropriate remedy?

IV. Analysis

[ 27 ] We begin our analysis with a survey of SOIRA, its history and the obligations it imposes on offenders, concluding that the provisions interfere with the liberty interests of those required to register. We turn next to identifying Parliament's objective in enacting these provisions, ultimately determining that the provisions are overbroad and therefore the limit on liberty is not in accordance with the principles of fundamental justice. Finding a breach of s. 7, we proceed to examine whether the breach is justified under s. 1 of the Charter. We conclude that it is not: ss. 490.012 and 490.013(2.1) are not minimally impairing of Charter rights and the deleterious effects of the provisions outweigh their salutary ones. Finally, we conclude by determining the appropriate remedy.

A. Sex Offender Information Registration Act

(1) The Challenged Provisions

[ 28 ] Two provisions in the Criminal Code which govern the registration of offenders under SOIRA are at issue in this appeal. The first is s. 490.012, which provides that a court "shall" make an order requiring sex offenders convicted of designated offences to comply with SOIRA for the applicable period specified in s. 490.013. Subsection (1) provides: "When a court imposes a sentence on a person for [a] . . . designated offence . . . , it shall make an order in Form 52 requiring the person to comply with the Sex Offender Information Registration Act for the applicable period specified in section 490.013". Similar language is used in subss. (2) and (3) of the same provision.

[ 29 ] The second is s. 490.013(2.1), one of several measures that sets out the applicable registration period, or the length of time that an offender is required to comply with SOIRA. The orders have 3 different durations — 10 years, 20 years or life — depending on the maximum punishment for the offence or if the offence was prosecuted summarily. Section 490.013(2.1) provides that "[a]n order made under subsection 490.012(1) applies for life if the person is convicted of . . . more than one . . . designated offence".

[ 30 ] Having set out the challenged provisions, we survey SOIRA's history and canvass the overall scheme and the obligations it imposes on offenders in greater detail.

(2) The History of SOIRA

[ 31 ] SOIRA came into force in 2004. SOIRA established a national registry of sex offenders. It was modelled on Ontario's registry, created by Christopher's Law (Sex Offender Registry), 2000, S.O. 2000, c. 1 (Christopher's Law), which was the first registry in Canada. Ontario's registry was enacted in 2001 following the recommendation of a provincial inquest into the abduction and murder of an 11‑year‑old boy by a convicted sex offender (Library of Parliament, Bill C-16: Sex Offender Information Registration Act, Legislative Summary 470E, February 16, 2004, at p. 8). Parliament followed suit with a national registry after calls mounted for an interprovincial database of sex offenders. Unlike Christopher's Law, SOIRA is maintained by the Royal Canadian Mounted Police, although in conjunction with local and provincial police services who oversee the administration of and access to the registry in designated regions.

[ 32 ] When SOIRA came into force, it was largely seen as a tool to help police stop the sexual abuse and abduction of children, a type of investigation where "time is of the essence" (House of Commons, Standing Committee on Public Safety and National Security, Statutory Review of the Sex Offender Information Registry Act: Report of the Standing Committee on Public Safety and National Security, 2nd Sess., 40th Parl., December 2009, at pp. 3-4). The Act stated that the registry's purpose was to help police investigate crimes of a sexual nature (s. 2(1), as it read in 2004). To manage the enrollment of offenders in the registry, Parliament added several measures to the Criminal Code and under these provisions, prosecutors exercised discretion over whether to bring an application to register an offender. Moreover, if prosecutors brought an application, judges could still exempt the offender from the registry if they were "satisfied that the person ha[d] established that, if the order were made, the impact on them, including on their privacy or liberty, would be grossly disproportionate to the public interest in protecting society through the effective investigation of crimes of a sexual nature" (see Criminal Code, s. 490.012(4), as it read in 2004).

[ 33 ] In 2011, both SOIRA and the Criminal Code were amended by way of the Protecting Victims From Sex Offenders Act. The objective of helping police prevent sex offences was added to SOIRA's purpose statement in s. 2(1). Further, prosecutorial discretion and judicial discretion to impose SOIRA orders were removed from the provisions of the Criminal Code. From that point forward, s. 490.012 provided that SOIRA orders were mandatory for offenders convicted of designated offences. Parliament also added s. 490.013(2.1) to the Criminal Code, which mandated lifetime registration for individuals convicted of more than one designated offence. The constitutionality of both these amendments is at issue in this case.

[ 34 ] As well, before the 2011 amendments, police needed to show they were investigating a crime and had reasonable grounds to suspect the crime was of a sexual nature before they could consult the registry (SOIRA, s. 16(2)(a), as it read in 2004). In 2011, however, the reasonable grounds requirement was removed. Thus, since 2011, police are now able to consult the registry to either prevent or investigate sex offences, whether or not they reasonably suspect an offence was or will be committed (SOIRA, s. 16(2)(a)).

[ 35 ] Since SOIRA's inception, many offences have led to registration under the Act. Section 490.011(1)(a) of the Criminal Code presently designates 27 offences that lead to automatic registration under s. 490.012(1) when a sentence is imposed or a verdict of not criminally responsible on account of mental disorder is rendered for those offences. The list of designated offences spans a broad range of criminal activity, including sexual assault, exposure, and withholding or destroying travel documents to facilitate the trafficking of persons under 18 years of age. SOIRA's ambit is amplified by designated offences like sexual assault, which can be committed in innumerable ways. The conduct captured by sexual assault includes everything from touching a complainant's buttocks over clothing to prolonged, violent assaults. SOIRA, as a result, captures a highly varied and diverse group of offenders. The number and breadth of designated sexual offences means that the net of sexual offenders subject to SOIRA is itself cast widely — many are subject to its provisions.

[ 36 ] SOIRA is nearly 20 years old. Despite its long existence, there is little or no concrete evidence of the extent to which it assists police in the prevention and investigation of sex offences. Det. Hove, an investigator with the EPS who administers the registry in the Edmonton region, testified before the sentencing judge; she could only speculate how SOIRA could be used to help the EPS prevent sexual offences. Moreover, she noted she received only about 15 requests from police to access the registry for an investigative purpose in her 2 years administering the registry. The experts at trial were unaware of any study of SOIRA's efficacy.

(3) The Legislative Scheme and Reporting Requirements Under SOIRA

[ 37 ] SOIRA imposes many obligations on offenders to report to a registration centre in person (ss. 4 and 4.1) and provide personal information for the purpose of the registry (s. 5), on an ongoing basis. SOIRA also imposes corresponding obligations on the police officers administering the registry to collect certain information and keep it confidential (s. 8). Access to the information contained in SOIRA is restricted to certain persons, including police who, since 2011, require access for "the purpose of preventing or investigating a crime of a sexual nature" (s. 16(2)(a)).

[ 38 ] Offenders must comply with SOIRA for the applicable registration period specified under s. 490.013 of the Criminal Code. Offenders, however, can seek a termination order, which (if granted) releases the offender from SOIRA obligations before the end of the applicable registration period (s. 490.016(1)). The order is only granted if a judge is satisfied that continuing the offender's registration is grossly disproportionate "to the public interest in protecting society through the effective prevention or investigation of crimes of a sexual nature, to be achieved by the registration of information relating to sex offenders under . . . [SOIRA]" (s. 490.016(1)). Moreover, an offender can only bring an application for a termination order after several years have elapsed (s. 490.015(1)). In the case of lifetime registration following a conviction for more than one offence under s. 490.013(2.1), a termination order can be sought 20 years after the order to comply with SOIRA was made (s. 490.015(1)(c)).

[ 39 ] Registered offenders are subject to many reporting requirements. Following an order made pursuant to s. 490.012(1), offenders must report in person to a registration centre, which are police stations designated with administering the registry in a geographic region. At the centre, the offender must provide extensive personal information, including their name, date of birth, gender, the address of their principal and secondary residences, the address of every place of employment or volunteer location, the name of their employer or volunteer supervisor and a description of the work done, the address of every educational institution at which they are enrolled, their height and weight, a description of every physical distinguishing mark that they have, and the licence plate number, make, model, body type, year of manufacture and colour of every vehicle registered in their name or that they use regularly (SOIRA, s. 5(1)). They must also report a contact phone number for each location where they can be reached and every mobile phone and pager in their possession (s. 5(1)(f)). They must supply information relating to all driver's licenses and passports they may hold. The registration centre may take their photograph and record their eye colour and hair colour.

[ 40 ] Offenders must update their information in person yearly (ss. 4(3) and 4.1(1)). They must also report in person at the registration centre any changes in primary or secondary address and name. They must also report in person if they receive a driver's licence or passport (s. 4.1(1)).

[ 41 ] Moreover, offenders must notify the registration centre within seven days of any change regarding their employment or volunteering information (ss. 5(1)(d) and 5.1). They must also notify the registration centre if they intend to be away from their primary or secondary residence for seven or more consecutive days (s. 6(1)). Specifically, offenders must notify the registration centre, before departure, of their departure and return dates and of every address or location at which they expect to stay, whether the addresses or locations are in or outside Canada (s. 6(1)(a)). Similar reporting requirements are imposed on offenders who decide, after departure, not to be at their primary or secondary residence for seven or more consecutive days (s. 6(1)(b)).

[ 42 ] Failing to comply with SOIRA brings serious consequences for offenders. The Criminal Code makes it an offence for an offender to fail to comply with SOIRA's reporting obligations without "reasonable excuse" (s. 490.031(1)). Non-compliance with any of these conditions may result in prosecution, with penalties of up to 2 years' imprisonment, up to $10,000 in fines, or both (s. 490.031(1)). The risks are clearly high for offenders if they fail to adhere to SOIRA's numerous requirements.

[ 43 ] Further, police officers conduct random compliance checks to verify the information on the registry. At a minimum, offenders are subject to at least one annual verification of their residential address. Det. Hove of the EPS also testified the current policy in Edmonton was to restrict compliance checks to the offender's primary residence, although SOIRA does not restrict where these checks are carried out. Nothing, as a result, prevents officers from showing up at an offender's place of employment. Thus, as the sentencing judge found, "offenders on the registry will be subject to further police interference due to the investigatory, and now preventative steps taken by police officers in relation to sex crimes" (ABQB reasons (2016), at para. 59).

[ 44 ] A number of appellate courts have concluded SOIRA's reporting requirements have a "minimal" or "modest" impact on registered offenders (see *Long*, at para. 147; R. v. Debidin, 2008 ONCA 868, 94 O.R. (3d) 421, at para. 82; R. v. Dyck, 2008 ONCA 309, 90 O.R. (3d) 409, at paras. 104-6; R. v. Cross, 2006 NSCA 30, 241 N.S.R. (2d) 349, at paras. 50 and 66; R. v. C. (S.S.), 2008 BCCA 262, 234 C.C.C. (3d) 365, at para. 55). The Court of Appeal in this case, while finding that the measures were "not strictly modest", equated registration to everyday reporting obligations.

[ 45 ] With respect, we cannot agree. Rather, the impact on anyone subject to SOIRA's reporting requirements is considerable. The requirements impact privacy and liberty, personal interests that are fundamental to society: liberty of movement and choice, mobility, and freedom from state monitoring or intrusion in our personal lives. The scope of the personal information registered, the frequency at which offenders are required to update their information, the ongoing monitoring by the state, and, of course, the threat of imprisonment make the conditions onerous. They simply cannot be compared to reporting requirements that "routinely occur as part of the everyday life" such as those associated with filing income tax forms, obtaining a driver's licence or a passport, or registering with banks or telephone companies (see *Dyck*, at para. 110).

[ 46 ] Additionally, the cost of compliance varies from offender to offender based on their life circumstances. While SOIRA's reporting requirements are always serious, offenders whose job requires regular, prolonged travel will frequently need to take additional measures to remain in compliance. Even worse, offenders who experience homelessness, substance use issues, and cognitive or mental health challenges may find compliance extremely difficult (see, e.g., R. v. J.D.M., 2006 ABCA 294, 417 A.R. 186, at para. 9; R. v. Desmeules, 2006 QCCQ 16773, at paras. 25-27). Quite simply, we must recognize the full scope of the restrictions that are imposed by SOIRA orders — both physical and informational — to properly assess the constitutionality of ss. 490.012 and 490.013(2.1).

[ 47 ] The Act thus imposes numerous exacting obligations concerning initial registration, ongoing reporting, state monitoring and possible prosecution and imprisonment. We turn now to consider whether the challenged provisions breach s. 7 of the Charter.

B. Sections 490.012 and 490.013(2.1) Infringe Section 7 of the Charter

[ 48 ] Section 7 of the Charter provides that "[e]veryone has the right to life, liberty and security of the person and the right not to be deprived thereof except in accordance with the principles of fundamental justice."

[ 49 ] In order to demonstrate a violation of s. 7, the claimant must first show that the law interferes with their life, liberty or security of the person. Once they have established that s. 7 is engaged, they must then show that the deprivation in question is inconsistent with the principles of fundamental justice (Carter v. Canada (Attorney General), 2015 SCC 5, [2015] 1 S.C.R. 331, at para. 55).

[ 50 ] In the first section we conclude that the provisions interfere with the liberty interests of those required to register. We then identify Parliament's objective in enacting these provisions, ultimately determining that the provisions are overbroad. Finding a breach of s. 7, we next examine whether the breach is justified under s. 1 of the Charter and explain why it is not.

(1) Sections 490.012 and 490.013(2.1) Interfere With the Offender's Liberty

[ 51 ] Underlying the rights in s. 7 is a concern for the protection of individual autonomy and dignity (Carter, at para. 64). Liberty protects "the right to make fundamental personal choices free from state interference" (Blencoe v. British Columbia (Human Rights Commission), 2000 SCC 44, [2000] 2 S.C.R. 307, at para. 54; see R. v. Clay, 2003 SCC 75, [2003] 3 S.C.R. 735, at paras. 31-32). Liberty also protects against physical restraint ranging from actual imprisonment or arrest (R. v. Malmo‑Levine, 2003 SCC 74, [2003] 3 S.C.R. 571, at para. 89; Fleming v. Ontario, 2019 SCC 45, [2019] 3 S.C.R. 519, at para. 65) to the use of state power to compel attendance at a particular place (R. v. Beare, [1988] 2 S.C.R. 387, at p. 402).

[ 52 ] The Crown properly concedes SOIRA engages the liberty interest of the accused but argues that the infringement is limited, analogous to fingerprinting, and exists only to the extent SOIRA compels attendance at a particular time and place. The similarities between a mandatory SOIRA order and the fingerprinting order considered in Beare certainly support a finding that s. 7 is engaged here. In Beare, there was a deprivation of liberty because the order obliged "a person to appear at a specific time and place and oblige[d] that person to go through an identification process on pain of imprisonment for failure to comply" (p. 402).

[ 53 ] However, the nature and extent of the deprivations at issue are much greater here than in Beare. SOIRA does not merely oblige offenders to appear once at a specific time and place and provide one type of personal information. Rather, it creates an ongoing obligation to report extensive information, subject to random checks and other compliance measures, under threat of prosecution and punishment by way of imprisonment, fines, or both. This creates continuous state monitoring that can last decades and for some offenders, like Mr. Ndhlovu, a lifetime.

[ 54 ] The impact of a SOIRA order on an offender's liberty can only fairly be described as serious. The most obvious impact on liberty is the risk of prosecution and imprisonment for failure to meet the reporting requirements without "reasonable excuse". Indeed, there are numerous reported cases involving offenders who have received terms of imprisonment for failing to comply with SOIRA orders (e.g., R. v. D.T., at para. 56; R. v. Callahan, at para. 62; R. v. Firingstoney, 2017 ABQB 343, at paras. 178‑79; R. v. Caruana, 2016 ONCJ 367, at para. 7).

[ 55 ] But the mandatory measures also involve constraints on liberty that are insidious and pervasive for all those who must comply. That offenders must report to a registration centre within tight timelines to provide the information of any changes in primary or secondary address, or name, as well as if they receive a driver's licence or passport, under threat of penalty, is a clear deprivation of liberty. It compels offenders to structure their travel and residency on an ongoing basis to remain in compliance with SOIRA (see, e.g., R. v. G.E.W., 2006 ABQB 317, 396 A.R. 149, at paras. 19 and 25, where the court considered the impact on an offender who worked in the trucking industry). There are burdens associated with the ongoing obligations to maintain the currency of the information on the registry. The potential of imprisonment makes the deprivation even more severe (Re B.C. Motor Vehicle Act, [1985] 2 S.C.R. 486, at p. 515).

[ 56 ] As noted above, the impact on liberty can also be aggravated by an offender's life circumstances. For persons experiencing homelessness or frequent changes in housing, complying with SOIRA's requirement in s. 4.1(1)(a) to report in person any changes to the location of their "main" residence (regardless of whether they have a formal address) is an extremely onerous obligation, that can be virtually impossible to respect, even more so since it can last for their lifetime.

[ 57 ] It is clear to us that ss. 490.012 and 490.013(2.1) interfere with offenders' liberty in serious ways. Liberty is obviously undermined when personal information is collected, under threat of imprisonment, for the very purpose of monitoring a person in the community and promptly identifying the person's whereabouts in the course of a criminal investigation.

[ 58 ] Finally, to be clear, we make no finding as to whether SOIRA orders constitute punishment under the K.R.J. test (R. v. K.R.J., 2016 SCC 31, [2016] 1 S.C.R. 906). This question was neither raised nor argued by either party before us, and we would not speculate, without the benefit of submissions, on whether SOIRA orders engage s. 11 of the Charter and, if they do, whether they would survive a Charter challenge.

(2) The Purpose of Sections 490.012 and 490.013(2.1)

[ 59 ] The first step in an overbreadth analysis is to determine the purpose of the challenged provisions (R. v. Safarzadeh-Markhali, 2016 SCC 14, [2016] 1 S.C.R. 180, at para. 24; R. v. Moriarity, 2015 SCC 55, [2015] 3 S.C.R. 485, at para. 24). It is to that preliminary question that we now turn.

[ 60 ] Several principles have emerged to assist a court in properly characterizing a law's purpose.

[ 61 ] The focus is on the purpose of the challenged provisions, not of the entire act in which they appear, although a correspondence between those purposes may sometimes occur (Moriarity, at paras. 29 and 48; RJR-MacDonald Inc. v. Canada (Attorney General), [1995] 3 S.C.R. 199, at para. 144).

[ 62 ] The law's purpose should be succinct, precise, and characterized at the appropriate level of generality, which "resides between the statement of an 'animating social value' — which is too general — and a narrow articulation" that amounts to a virtual repetition of the challenged provision, divorced from its context (Safarzadeh‑Markhali, at para. 27, quoting Moriarity, at para. 28).

[ 63 ] A law's purpose is distinct from the means used to achieve that purpose (Safarzadeh‑Markhali, at para. 26; Moriarity, at para. 27).

[ 64 ] To determine an impugned law's purpose, courts may consider: statements of purpose in the legislation, if any; the text, context, and scheme of the legislation; and extrinsic evidence such as legislative history and evolution (Safarzadeh-Markhali, at para. 31; Moriarity, at para. 31).

[ 65 ] SOIRA's overall purpose is readily identified. The statement of purpose in s. 2(1) of SOIRA states the Act aims "to help police services prevent and investigate crimes of a sexual nature by requiring the registration of certain information relating to sex offenders". In this case, there is no question that the statement of purpose fully reflects Parliament's aims in enacting SOIRA. Both at the time it was enacted and when it was later amended, Parliament emphasised the Act was designed to assist police. Moreover, it has clearly indicated that SOIRA is intended to assist police in the prevention and investigation of sex offences. However, the challenge in this case is not to the Act as a whole, but is confined to two particular sections of the Criminal Code: one that provides no judicial discretion to exempt offenders from the registry and another that requires lifetime registration for those convicted of more than one designated sexual offence.

[ 66 ] When assessing the purpose of these challenged provisions, several sources of legislative interpretation closely tie these two provisions to SOIRA's overall purpose.

[ 67 ] To begin, SOIRA and ss. 490.012 and 490.013(2.1) of the Criminal Code form an integrated legislative scheme. The provisions subject offenders to the reporting requirements listed in SOIRA. When Parliament enacts related legislation dealing with the same subject, the legislation is presumed to offer a coherent and consistent treatment of the subject (R. Sullivan, The Construction of Statutes (7th ed. 2022), at § 13.04). As a result, the provisions must be interpreted in conjunction with the scheme established under the Act. SOIRA's overall purpose informs the interpretation of the challenged provisions' purpose.

[ 68 ] The connection between the purpose of the provisions and SOIRA's overall purpose is reinforced by other provisions in SOIRA. Section 2(2) of SOIRA lists three principles that must guide the Act's interpretation. Two emphasize that information collected and disclosed under the Act is intended to assist police prevent and investigate sex offences (s. 2(2)(a) and (c)(i)). Since the Criminal Code provisions similarly affect the scope of information collected in the database, s. 2(2) of SOIRA implies the provisions should be read as closely tied to the overall aim of assisting police.

[ 69 ] Turning to legislative history, statements made during the amendment process indicate that the specific purpose of mandatory registration is to capture information about offenders that may assist police prevent and investigate sex offences. No extrinsic aids, however, shed light on the purpose of lifetime registration.

[ 70 ] Mandatory registration was introduced following a report by the Standing Committee on Public Safety and National Security on SOIRA's operation and efficacy (Statutory Review of the Sex Offender Information Registry Act: Report of the Standing Committee on Public Safety and National Security). The Standing Committee's mandate was to identify amendments to ensure the registry is "best able to fulfill the purpose for which it was enacted" (p. 2). Among its recommendations, the report expressed concerns that SOIRA's inclusion rate was too low. Only around 50 percent of sex offenders were registered in the database (p. 8). The Standing Committee recommended eliminating prosecutorial discretion. Importantly, the Standing Committee concluded that judicial discretion to exempt offenders where the measures were grossly disproportionate should be maintained (p. 9).

[ 71 ] Parliament, however, opted to eliminate all discretion. Before Parliament, the Minister of Public Safety and his representatives reiterated the Standing Committee's concerns on SOIRA's low inclusion rate. The Minister remarked before the Standing Senate Committee on Legal and Constitutional Affairs that some recidivistic offenders were "falling through the cracks" because prosecutors were failing to bring applications and judges were excluding offenders when applications were brought (Proceedings of the Standing Senate Committee on Legal and Constitutional Affairs, No. 3, 3rd Sess., 40th Parl., April 14, 2010, at p. 32). Similarly, in the House of Commons, the Minister's parliamentary secretary echoed the concerns that SOIRA was under-inclusive and its ability to assist police was undermined as a result (House of Commons Debates, vol. 145, No. 112, 3rd Sess., 40th Parl., December 7, 2010).

[ 72 ] The majority at the Court of Appeal concluded that the purpose of mandatory registration was to register all sex offenders. While such a purpose may find some support in parts of the Hansard, we cannot agree with this characterization of the provision's purpose. First, it replicates the error warned against in Moriarity, at paras. 27‑28: it fails to adequately distinguish between ends and means, which forecloses any separate inquiry into the connection between them. In this way, it immunizes mandatory registration from review for overbreadth. Second, it fails to consider mandatory registration in light of the Act's overall purpose. Parliament did not enact SOIRA just for the sake of creating a database of all sex offenders. Rather, Parliament enacted SOIRA to create a repository of information that assists police prevent and investigate sex offences.

[ 73 ] Legislative provisions often encompass a range of objectives: from broad societal values to the specific means enacted to advance those goals. The legislative statements may well include the aspirational goals that animate the legislation, the specific objectives, and the anticipated effect of the provisions. As noted above, the court must articulate the objective of the provision at the appropriate level of generality, in light of the provision itself, the statutory regime and any external evidence. Here, mandatory registration must be interpreted in light of SOIRA's explicit overall purpose to help police prevent and investigate sex offences. Taking this broader context into account, we conclude the purpose of mandatory registration is to capture information about offenders that may assist police prevent and investigate sex offences.

[ 74 ] Turning to lifetime registration for more than one offence, we conclude that its purpose is to give police a longer period of access to information on offenders at a greater risk of reoffending. As noted, no legislative history sheds light on the purpose of s. 490.013(2.1). It must be discerned from its text and the overall legislative scheme. To begin, SOIRA imposes lifetime registration periods in several scenarios. Before s. 490.013(2.1) was added in 2011, lifetime registration only applied when an offender committed relatively more serious offences that warrant longer sentences (s. 490.013(2)(c)) or if they were or had been subject to a previous SOIRA order (s. 490.013(3) and (4)). The challenged provision also targets scenarios where more than one offence is committed, but goes further than the former s. 490.013(3) and (4), as it applies where more than one offence was committed whether or not the offender was already subject or was subject at any time to a SOIRA order.

[ 75 ] Given the language of the provision, in the context of the existing scheme, we agree with the Court of Appeal that Parliament's basis for seeking a longer period of access when more than one offence is committed must be that it believed that these offenders were more likely to reoffend relative to other sex offenders. Given this greater risk of harm, Parliament preferred to have the offender's information available on the registry as long as possible. This is consistent with Parliament's approach to target offenders who commit more serious offences. When s. 490.013(2.1) was introduced in 2011, the Criminal Code already imposed lifetime registration for more serious offences. Parliament appears to have provided for longer registration in the face of a greater risk of harm, either because the offence was more serious (as in s. 490.013(2)(c)), or because the risk of recidivism was higher (s. 490.013(2.1)). Thus, we conclude that the purpose of the measure is to give police a longer period of access to information on offenders at a greater risk of reoffending.

[ 76 ] In sum, SOIRA was not enacted with complete or total registration as an end in itself. It was enacted to help police prevent and investigate sex offences. The purpose of both challenged measures in the Criminal Code is closely tied to this overall purpose. The specific purpose of s. 490.012 is to capture information about offenders that may assist police prevent and investigate sex offences. The means to achieve this purpose is mandatory registration. Section 490.013(2.1) was similarly designed to give police a longer period of access to information on offenders at a greater risk of reoffending. The means to achieve this purpose is lifetime registration for sex offenders who commit more than one designated offence.

(3) The Challenged Measures Are Overbroad

[ 77 ] Having identified the purpose of the measures, the next step is to determine whether they are overbroad. A law is overbroad when it is so broad in scope that it includes some conduct that bears no relation to its purpose, making it arbitrary in part (Canada (Attorney General) v. Bedford, 2013 SCC 72, [2013] 3 S.C.R. 1101, at para. 112). In other words, overbreadth addresses the situation where there is no rational connection between the purpose of the law and some, but not all, of its impacts (para. 112).

[ 78 ] The Court in Bedford clarified that a law is overbroad even if it overreaches in only a single case (paras. 113 and 123). A law cannot deprive the life, liberty, or security of the person of even one individual in a way that is inconsistent with the principles of fundamental justice. As a consequence, laws that are broadly drawn to make enforcement more practical run afoul of s. 7 should they deprive the liberty of even one person in a way that does not serve the law's purpose (para. 113). The Court in Bedford concluded that enforcement practicality may justify a broad law under s. 1 of the Charter (para. 144), but it "is no answer to a charge of overbreadth under s. 7" (Safarzadeh‑Markhali, at para. 53, citing Bedford, at para. 113).

(a) Mandatory Registration

[ 79 ] The sentencing judge concluded that mandatory registration (s. 490.012) is overbroad as it leads to the registration of offenders who are not at an increased risk of committing a future sex offence. We agree. As we explained, the purpose of mandatory registration is to capture information about offenders that may assist police prevent and investigate sex offences. Registering offenders who are not at an increased risk of reoffending bears no connection to this purpose. The provision is overbroad.

[ 80 ] At the s. 1 hearing, the Crown adduced expert evidence on statistical sexual recidivism rates by the testimony of Dr. Hanson, a clinical psychologist. Most of Dr. Hanson's testimony was focused on structured approaches to conducting risk assessments of sexual offending. According to this expert evidence, however, the majority of sexual offenders — about 75 to 80 percent — never actually reoffend. Before the sentencing judge, the Crown conceded that the registry captures people who will never reoffend.

[ 81 ] To overcome that hurdle, the Crown relies on two interrelated arguments. First, it argues the connection between the effects of s. 490.012 on all offenders and its objective is predicated on the fact that, although not all sexual offenders reoffend, they are, as a group, all at an enhanced risk of reoffending based on statistical evidence. A conviction is a logical, practical and sufficient proxy for an enhanced risk of committing a sexual offence. Second, there are no reliable means to identify those who are not at an enhanced risk to reoffend at sentencing.

[ 82 ] As we will demonstrate, however, the Crown's submissions are flawed. We proceed in three parts. First, the Crown's reasoning ignores that some individual circumstances may allow a court to identify offenders that are not at an increased recidivism risk. Second, the Crown's position is not supported by the expert statistical evidence presented at trial: it is inaccurate to say that all sexual offenders are at an enhanced risk of reoffending. Third, the overbroad nature of s. 490.012 cannot be salvaged by the difficulty with risk assessments at sentencing: such arguments should be dealt with under s. 1.

[ 83 ] To summarize our analysis, we see no error in the sentencing judge's or Khullar J.A.'s conclusion that including offenders who are not at an increased risk of reoffending on the registry bears no connection to mandatory registration's purpose. Subjecting all offenders, regardless of their future risk, to continuing and onerous reporting requirements, random compliance checks, possible prosecution, and imprisonment does not bear a connection to the purpose of the measure. In this case, Moen J. found at first instance, based on evidence at the sentencing hearing, that Mr. Ndhlovu was such an offender: she had "absolutely no concerns that [he] will re‑offend".

[ 84 ] The extent to which requiring all offenders to register is necessary due to the difficulties in assessing risk is an argument grounded in enforcement practicality or administrative convenience. It relates to justification rather than overbreadth. While mandatory registration has the attraction of simplicity and ease, the convenience of requiring every sex offender to register does not make it constitutional.

(i) An Offender's Personal Circumstances May Show There Is No Increased Risk of Reoffending

[ 85 ] The Crown's focus on the recidivism risk of sex offenders faces an obvious difficulty. In certain cases, an offender's personal circumstances mean they are not at an increased risk of reoffending, undermining any real possibility that their information on the registry will ever prove useful to police. But a rough proxy like a prior conviction for a sex offence does not readily account for those circumstances. This common sense argument is consistent with the Crown expert's testimony.

[ 86 ] Trials based on historical sexual offences are commonplace in our courts. An offender who committed a sexual offence in the past is sometimes only convicted and sentenced decades later, when they are at an advanced age and have highly limited mobility. Subjecting such offenders to SOIRA would obviously bear no connection to the purpose of capturing information about offenders that may assist police prevent and investigate sex offences. The conclusion that these personal circumstances may have a bearing on recidivism risk, yet are not captured in statistical models, is consistent with the Crown expert's testimony.

[ 87 ] The Crown's expert, Dr. Hanson, rightly admitted the limitations on the statistics that formed the basis of his testimony. He recognized statistical models have inherent limits: even the best studies or models cannot account for all personal characteristics that influence one's likelihood of reoffending. Bearing in mind that an effect on one person's liberty that bears no connection to s. 490.012's purpose is sufficient to make it overbroad (Bedford, at para. 123), the breach here becomes obvious. We are satisfied that the personal circumstances of some offenders mean they are at no increased risk of reoffending.

[ 88 ] Consider, for instance, the offender in R. v. T.L.B., 2006 ABQB 533, 403 A.R. 293, aff'd 2007 ABCA 135, 404 A.R. 283. The offender was wheelchair‑bound due to cerebral palsy, unable to work, and required daily assistance from a caregiver to assist with her personal needs (paras. 8-9). She was convicted of child pornography and sexual interference offences involving her six-year-old child. She was pressured to make images of her child and to commit a sexual act with her son after entering into an online relationship with a man who was a pedophile (paras. 2-7). After the incident was disclosed, her son was apprehended from her care (para. 6). At sentencing, T.L.B. was exempted from complying with SOIRA — under the previous provisions — on the grounds it would be "grossly disproportionate" (para. 80). Her risk of reoffending was very low based on the assessment of a clinical and forensic psychologist (para. 65). Her offences were directed against her child, who was no longer in her care, and she did not have contact with any other children (para. 65).

[ 89 ] T.L.B.'s personal circumstances make it highly improbable that she would reoffend. There is no increased risk that an offender like T.L.B. would ever commit another sex offence. Nor did anything in her pattern of offending suggest that her location or identity would not be readily ascertainable without SOIRA if police investigated her for a future offence. As a result, her registration would not be connected to the objective of capturing information about offenders that may assist police prevent and investigate sex offences. Focusing only on T.L.B.'s convictions neglects all of the unique circumstances in her case that make it highly improbable that police would ever benefit from her registration.

[ 90 ] These examples illustrate the constitutional infirmity of s. 490.012 under s. 7. The question is not whether sexual offenders are generally at an increased risk of reoffending, but rather whether there are some offenders who are not. This is because a law that has an overbroad impact on even one individual is inconsistent with the principles of fundamental justice (Bedford, at para. 123). As T.L.B. illustrates, in some cases, the personal circumstances of an offender will undermine any increased risk of reoffending. There is, as a result, no connection between subjecting an offender like T.L.B. to SOIRA and the purpose of capturing information that may assist police prevent and investigate sex offences.

(ii) Mandatory Registration Is Overbroad Since Some Sex Offenders Are Not at an Increased Statistical Risk of Reoffending

[ 91 ] The expert evidence, which the sentencing judge accepted, made clear that there is no perceptible difference in sexual recidivism risk at the time of sentencing between the lowest-risk sexual offenders — the bottom 10 percent — and the population of offenders with convictions for non-sexual criminal offences. In both instances, about two percent of individuals — whether they be the lowest-risk sexual offenders or the people with a criminal record unrelated to a sexual offence — commit a sexual offence over the next five years.

[ 92 ] Mandatory registration is overbroad to the extent it sweeps in these lowest‑risk sex offenders. As a result of their risk profile, there is no connection between subjecting them to a SOIRA order and the objective of capturing information that may assist police prevent and investigate sex offences because they are not at an increased risk of reoffending. The purpose of the provision is not advanced by including these offenders.

[ 93 ] The Crown nevertheless contends that the measures are not overbroad since sex offenders are, as a group, at an enhanced risk of recidivism. It adds that even the lowest-risk sex offender is more dangerous than members of the general public. It further argues that, in any event, all sexual offenders present a more than de minimis risk of committing a further sexual offence and that Parliament is entitled to legislate in order to mitigate this risk of harm. In our view, none of these arguments is sufficient to cure the provision's overbreadth.

[ 94 ] First, we agree that the commission of a sexual offence is one of many empirically validated predictors of increased sexual recidivism. But so are other factors, such as age, unusual or atypical sexual interests, sexual preoccupation, lifestyle instability or poor cognitive problem solving (to name a few). Recidivism risk also varies depending on the pattern of offences: for instance, whether the offence is a non‑contact sexual offence, or whether it is committed against a child, a stranger, an acquaintance or a family member. Yet the expert evidence makes clear that valid risk assessments must consider a range of risk relevant variables — there is no single factor that, on its own, yields an offender's recidivism risk. In short, many factors affect a sex offender's recidivism risk.

[ 95 ] As a result, although an offender may be part of a group that is generally — in other words, on average — at an enhanced risk of reoffending, this does not hold true for every individual in the group. To be clear, an overbreadth analysis does not focus on the group, but on the individuals within that group. As explained above, while a previous conviction for a sexual offence is a risk factor, about 10 percent of the individuals with a prior conviction for a sex offence are not, at time of sentencing, at an enhanced risk of reoffending when compared to the general criminal population. Focusing only on the commission of a past sexual offence inevitably detracts the focus from the other variables affecting an offender's recidivism risk. This is hardly surprising, considering that sexual offences cover a broad range of conduct and that sex offenders are not a uniform group.

[ 96 ] Second, there is no evidence to support the Crown's reliance on an enhanced rate of offending relative to the general public. The Crown's own expert could only speculate and testified that the rate of sexual offending by those without any prior conviction for any offence had not been established.

[ 97 ] Third, a de minimis risk of harm or risk of reoffending is insufficient to avoid overbreadth. The Crown relies on Malmo-Levine for the proposition that Parliament is entitled to enact measures that mitigate a more than de minimis risk of harm, a point accepted by the majority at the Court of Appeal (para. 89). Since all sex offenders are allegedly at a more than de minimis risk of reoffending, the Crown says Parliament complied with s. 7 when it elected to include relatively low-risk offenders in the registry.

[ 98 ] As we note above, only about 20 to 25 percent of the sex offender group actually reoffend, and 10 percent of the group are at about a 2 percent chance of reoffending after 5 years. And within the rates of recidivism, the range of conduct captured — and the degree of harm — varies dramatically. Nonetheless, we agree that "the precise weighing and calculation of the nature and extent of the harm is Parliament's job" (Malmo-Levine, at para. 133). However, the question in an overbreadth analysis is not whether Parliament perceived a more than de minimis risk of harm when it enacted a law. Rather, the question is whether the law is so broad in scope that it has some impact on a s. 7 interest that bears no connection to the law's purpose (Bedford, at para. 112).

[ 99 ] The difference between these two questions is illustrated by an early decision in this Court's overbreadth jurisprudence, R. v. Heywood, [1994] 3 S.C.R. 761. In Heywood, the Court considered a provision that made it an offence for certain offenders to loiter in or near school grounds, playgrounds, public parks, or bathing areas (then s. 179(1)(b) of the Criminal Code). The purpose of the measure was to protect children from becoming victims of sexual offences (p. 794). The Court struck down the measure for overbreadth because it applied to locations where children were not likely to be present, such as every part of all public parks in Canada, including vast and remote wilderness parks (p. 795). Clearly, however, the measure applied to sex offenders who would, according to the Crown's logic, pose a more than de minimis risk of harm. Nevertheless, the provision breached s. 7 of the Charter because it impacted the liberty of offenders in a manner that bore no connection to its purpose. While Heywood predates Bedford and Carter, it remains a paradigmatic application of the overbreadth principle (H. Stewart, Fundamental Justice: Section 7 of the Canadian Charter of Rights and Freedoms (2nd ed. 2019), at p. 154).

[ 100 ] Quite simply, the Crown cannot rely on the fact that any offender has a risk — albeit not an increased risk — to commit a sexual offence in the future. If a mere risk of committing a future criminal offence could provide the requisite connection to meet the Bedford threshold, such a threshold would be met, by reason of logic, in every case where forward-looking measures are implemented by Parliament. If one accepts the basic premise that it is impossible to predict the future with absolute certainty, then, by definition, anyone has a risk of committing a criminal offence in the future. If risk is all it takes to make a measure depriving individuals from their liberty compliant with s. 7, the law on overbreadth would be deprived of its normative value.

[ 101 ] In the end, the evidence — accepted by the sentencing judge — demonstrates that s. 490.012 catches offenders who, at the time of their release, are no longer at significant risk to reoffend. The commission of a past sexual offence, therefore, is an inexact proxy for those offenders whose information may assist police. A proxy of such a broad nature, which applies to a large number of people and a broad range of conduct, inevitably captures individuals who are not at an increased risk of committing the criminal conduct that Parliament sought to prevent and investigate with the registry.

(iii) Uncertainty at Sentencing Cannot Cure Section 490.012's Overbreadth

[ 102 ] The Crown submits s. 490.012 is not overbroad, since the risk of recidivism is difficult to assess and even expert assessments are error-prone. Thus, it says, there is no other alternative that would capture the information to assist police. There are two difficulties with the Crown's argument.

[ 103 ] First, as discussed above, the Crown cannot save a law from overbreadth for reasons of "administrative convenience" or to make enforcement more practical (Singh v. Minister of Employment and Immigration, [1985] 1 S.C.R. 177, at p. 219; see Bedford, at para. 113; Carter, at para. 88; Safarzadeh-Markhali, at para. 53). As the Crown relies on the same concern here, it does not save mandatory registration under s. 7, but must be addressed under s. 1.

[ 104 ] It is worth elaborating on why questions of enforcement practicality are generally not addressed under s. 7 of the Charter. This type of argument implicitly accepts that an individual's rights are breached but holds it is justified for the sake of a benefit to the public — making the administration or enforcement of a law more practical or convenient. As a result, the argument goes to the heart of the s. 1 analysis: the justification of a breach of an individual's rights in pursuit of a greater public good (i.e., laws that are easier to enforce or administer), rather than the "narrower" question under s. 7, which concerns "whether the impugned law infringes individual rights" (Bedford, at para. 125). Considerations of enforcement practicality balance harm to an individual's rights against a benefit to the public and so fall at the core of s. 1. They are appropriately addressed as part of the Crown's burden under s. 1 as a result.

[ 105 ] Since Bedford, the Court has consistently rejected arguments premised on enforcement practicality and administrative convenience under s. 7 of the Charter. In Bedford, the Court concluded that the criminal offence of living on the avails of prostitution was overbroad. The purpose of the offence was "to target pimps and the parasitic, exploitative conduct in which they engage" (para. 137). The offence was overbroad to the extent it captured persons who were not in exploitative business relationships with sex workers, including accountants and receptionists (para. 142). The Attorneys General of Canada and Ontario argued the law needed to be broadly drawn due to the blurry line between exploitative and non-exploitative relationships (para. 143). The Court concluded this argument, premised on enforcement practicality, was better addressed under s. 1 of the Charter (para. 144).

[ 106 ] Later, in Carter, the Court considered a challenge to a criminal prohibition on assisted dying. The purpose of the prohibition was to protect vulnerable individuals from committing suicide in a moment of weakness (para. 78). The Court concluded the prohibition was overbroad to the extent it applied to individuals who were not vulnerable (para. 86). In reaching this conclusion, the Court rejected the Crown's argument that the law needed to be broadly drawn due to the difficulty of conclusively identifying vulnerable individuals (paras. 87-88). The Court again held this argument was properly addressed under s. 1 (para. 88).

[ 107 ] A year later, in Safarzadeh‑Markhali, the Court considered a challenge to a provision that barred offenders who were denied bail due to a prior conviction from receiving enhanced credit for presentence custody. The provision's purpose was "to enhance public safety and security by increasing violent and chronic offenders' access to rehabilitation programs" (para. 47 (emphasis deleted)). The Court concluded it was overbroad since it applied to offenders who did not pose a threat to public safety or security, including those denied bail for non-violent prior offences, such as missing a court date (paras. 52-53). The Crown argued that the provision cast a broad net since that was more practical than identifying only those offenders that posed a public safety risk (para. 53). The Court, citing Bedford, held this concern for enforcement practicality did not save the provision from overbreadth under s. 7 (para. 53).

[ 108 ] Bedford, Carter, and Safarzadeh-Markhali all affirm that Parliament cannot rely on enforcement practicality or administrative convenience to immunize a law from overbreadth under s. 7 of the Charter. The proper approach is to address such considerations under s. 1. The Crown's attempts to rely on enforcement practicality under s. 7 similarly fail in this case.

[ 109 ] Moreover, judges make risk assessments routinely, including those informed by expert assessments. Notwithstanding these assessments may not be certain, they are capable of being well informed by an individual's personal circumstances and the best expert evidence. Clearly, there are instances where a sentencing judge can reasonably conclude that it is remote or implausible that an offender's information will ever prove useful to police.

[ 110 ] Finally, in the past, judges may have improperly exempted offenders on the basis of gross disproportionality by relying on myths and stereotypes about sexual assaults (see J. Benedet, "A Victim-Centred Evaluation of the Federal Sex Offender Registry" (2012), 37 Queen's L.J. 437). To the extent some trial judges may have interpreted the former exemption too broadly, those trial decisions are always subject to appellate review and guidance. This cannot render an unconstitutional law constitutional.

[ 111 ] Thus, mandatory registration is overbroad. Since it captures offenders who are not at an increased risk to reoffend, s. 490.012 breaches s. 7 of the Charter.

(b) Lifetime Registration

[ 112 ] Lifetime registration of those convicted of more than one sexual offence (s. 490.013(2.1)) is also overbroad.

[ 113 ] The Crown's expert distinguished between two categories of offenders who commit more than one offence. The first is an offender, like Mr. Ndhlovu, who commits more than one sex offence without an intervening conviction. The second is an offender who, after being convicted of a sex offence, goes on to commit another offence. Dr. Hanson explained that committing more than one offence without an intervening conviction is not associated with an enhanced recidivism risk. As he put it, "individuals who are convicted of . . . two or three offences at the same sentencing occasion are the same risk as an individual who is convicted of one" (A.R., vol. II, at p. 196). He noted, however, that committing another offence after a conviction did substantially increase recidivism risk.

[ 114 ] Dr. Hanson's evidence establishes that lifetime registration for more than one offence without an intervening conviction is overbroad. The purpose of the measure is to give police a longer period of access to information on offenders at a greater risk of reoffending. Yet, as the expert evidence establishes, the measure captures some offenders who are not at a relatively greater risk of reoffending because their two or more offences were committed, for example, in a single transaction. Section 490.013(2.1), however, provides no discretion to exempt offenders in this circumstance.

[ 115 ] The Crown submits Parliament was entitled to rely on a common sense inference that committing more than one offence on a single occasion increases recidivism risk. We would not give effect to this argument. The expert evidence clearly undermines the plausibility of any such inference. For this reason, we agree with the defence that s. 490.013(2.1) is overbroad.

(4) Gross Disproportionality

[ 116 ] Having concluded the measures are overbroad, we need not decide whether they are grossly disproportionate. That said, we would not foreclose the possibility that the effects of the provisions may be grossly disproportionate to their purposes in some cases. As discussed above, the impact on anyone who is subject to the reporting requirements of a SOIRA order is considerable. The personal information registered, the frequency at which offenders are required to update their information and the threat of imprisonment make the conditions onerous. They are not routine reporting requirements. Section 490.012 previously allowed for an exemption to ensure no individual would be subject to a grossly disproportionate order. That option, however, is no longer available.

C. Sections 490.012 and 490.013(2.1) Are Not Justified Under Section 1 of the Charter

[ 117 ] Having concluded that ss. 490.012 and 490.013(2.1) breach s. 7 of the Charter, it remains to determine whether the measures can be justified under s. 1. The Crown submits the measures are saved by s. 1. We do not accept this submission.

[ 118 ] The Crown bears the burden of establishing that the challenged measures' infringement of s. 7 is justified under s. 1 of the Charter. To meet its burden under s. 1, the Crown must show the infringement is "demonstrably justified", which means the infringing measures must be justified based on a "rational inference from evidence or established truths" (RJR-MacDonald, at para. 128). Bare assertions will not suffice: evidence, supplemented by common sense and inference, is needed (R. v. Sharpe, 2001 SCC 2, [2001] 1 S.C.R. 45, at para. 78).

[ 119 ] A breach of the Charter is justified under s. 1 when the challenged law has a "pressing and substantial object and . . . the means chosen are proportional to that object" (Carter, at para. 94). The law is proportionate where the means adopted are rationally connected to the law's objective, minimally impairing of the right in question, and the law's salutary effects outweigh its deleterious effects (R. v. Oakes, [1986] 1 S.C.R. 103, at pp. 136-40). The focus of the analysis is on the infringing measures, not on the overall legislative scheme. Thus, the Crown's burden requires it to show the challenged measures, and not SOIRA as a whole, impose a reasonable limit on s. 7 (Ontario (Attorney General) v. G, 2020 SCC 38, [2020] 3 S.C.R. 629, at para. 72, citing RJR‑MacDonald, at para. 144; Toronto Star Newspapers Ltd. v. Canada, 2010 SCC 21, [2010] 1 S.C.R. 721, at para. 20).

[ 120 ] The defence concedes that the prevention and investigation of sex crimes is a pressing and substantial purpose. We agree. Parliament's goals in enacting SOIRA are laudable. In our view, this holds true for the specific objectives of the provisions at issue. Parliament's efforts to provide tools to police that make it easier to prevent and investigate sex offences are clearly aligned with the public's interest in preventing sex crimes and bringing sex offenders to justice.

[ 121 ] Furthermore, the measures are rationally connected to their objectives. The standard is not onerous; Oakes requires a rational connection, not a complete rational correspondence (R. v. Appulonappa, 2015 SCC 59, [2015] 3 S.C.R. 754, at para. 80; Health Services and Support — Facilities Subsector Bargaining Assn. v. British Columbia, 2007 SCC 27, [2007] 2 S.C.R. 391, at para. 148). This test is met here. Since a conviction for a sexual offence is a reliable indicator of an increased risk of reoffending and committing another sex offence after a conviction can increase recidivism risk, it is reasonable to suppose the provisions may further their respective objectives (Alberta v. Hutterian Brethren of Wilson Colony, 2009 SCC 37, [2009] 2 S.C.R. 567, at para. 48). Yet, as we explain, the measures fail the other branches of the Oakes test.

(1) Sections 490.012 and 490.013(2.1) Are Not Minimally Impairing of an Offender's Rights

[ 122 ] A key issue on this appeal is whether the measures are minimally impairing of an offender's rights. To be minimally impairing, the challenged provisions must interfere with s. 7 "as little as reasonably possible in order to achieve the[ir] legislative objective" (RJR-MacDonald, at para. 160). The analysis turns on whether there are alternative, less drastic means of achieving the objective in a real and substantial manner (K.R.J., at para. 70). The Crown bears the burden of showing no less drastic means are available (Safarzadeh-Markhali, at para. 63). A court need not find that the alternative measures "satisfy the objective to exactly the same extent or degree as the impugned measure[s]" (Hutterian Brethren, at para. 55 (emphasis in original)). Instead, it suffices that the alternative measures "substantially" achieve the challenged measures' objective (paras. 55 and 60). For this reason, the Court in G rejected the Crown's argument that since recidivism risk could not be perfectly predicted, the mandatory and permanent registration of offenders who were found not criminally responsible on account of mental disorder in Ontario's sex offender registry was minimally impairing. The Court concluded that absolute certainty in risk assessments cannot be expected (para. 75). For similar reasons, we conclude the Crown has not discharged its burden on this step.

[ 123 ] The Crown argues that mandatory SOIRA orders (s. 490.012) are necessary to achieve "Parliament's objective of providing police with a comprehensive source of information on convicted sexual offenders, and particularly those who reoffend" and "to achieve an offender inclusion rate similar to that of the Ontario registry" (R.F., at paras. 162-63). This argument, however, is based on a flawed premise, since it mischaracterizes s. 490.012's objective. As stated above, the purpose is to capture information about offenders that may assist police prevent and investigate sex offences. The means chosen is the registration of all sex offenders. The Crown's argument under minimal impairment inappropriately shifts the characterization of mandatory registration's purpose.

[ 124 ] When the purpose of mandatory registration is properly characterized, it is apparent that the Crown has not met its burden under minimal impairment. To begin, the Crown concedes that restoring judicial discretion in the registration process would allow for a 90 percent inclusion rate of offenders in the registry. Yet the Crown did not adduce any evidence to explain why an inclusion rate of 90 percent would not substantially achieve s. 490.012's purpose of capturing information that assists police prevent and investigate sex offences. Specifically, no evidence or plausible argument was provided to explain why a discretion to exclude offenders from SOIRA where the impacts are likely too onerous or unrelated to s. 490.012's purpose would not substantially achieve Parliament's aims. In fact, the Crown did not adduce any evidence about the difficulties faced by police in preventing or investigating sexual offences before 2011, when a form of judicial discretion was in place. Indeed, the Crown's expert, Dr. Hanson, was unable to find any study of SOIRA's efficacy or consequences before the 2011 amendments.

[ 125 ] There is also no persuasive evidence that the difficulties in assessing recidivism risk renders judicial discretion incompatible with Parliament's goals. Judges frequently assess risk. As noted, this Court said in G at the minimal impairment stage, "[i]ndividual assessment does not need to perfectly predict risk — certainty cannot be the standard" (para. 75). A variety of tools are available to improve the accuracy of judicial risk assessments, including expert evidence. Alternatively, Parliament can enumerate specific criteria to guide judges on when registering an offender is unlikely to advance the scheme's objective. We do not accept the Crown's submission that reliable, tailored alternatives are simply not available. There is, it would seem, a variety of measures that Parliament may introduce, keeping in mind the measures must neither be overbroad nor grossly disproportionate.

[ 126 ] In essence, the Crown's argument rests on the proposition that all offenders must be registered unless the defence can demonstrate that judicial discretion will not hamper the police's ability to prevent and investigate sex offences. The unproven premise is that police can only effectively prevent and investigate sex offences if all designated offenders are registered. The assumption appears to be that if some are good, more is better, and all is best. The Court in Carter rejected a similar argument at the minimal impairment stage because it "effectively reverses the onus under s. 1, requiring the claimant whose rights are infringed to prove less invasive ways of achieving the prohibition's object" (para. 118). The same concern in Carter arises in this case. It is the Crown, not Mr. Ndhlovu, who bears the burden under s. 1. Rather than call evidence that shows less infringing measures would fail to substantially achieve the measure's objective, the Crown relies on assertion and conjecture. That is not enough to meet its burden under s. 1. As the Court stated in Carter, at para. 119, "[j]ustification under s. 1 is a process of demonstration, not intuition or automatic deference to the government's assertion of risk" (citing RJR-MacDonald, at para. 128).

[ 127 ] Nor has the Crown met its burden on lifetime registration (s. 490.013(2.1)). As noted, the expert evidence indicates that offenders who commit more than one offence without an intervening conviction have no increased recidivism rate relative to offenders who commit a single offence. The Crown has not explained why exempting this category of offenders would not achieve s. 490.013(2.1)'s purpose — that is, to give police a longer period of access to information on offenders at a greater risk of reoffending — in a real and substantial manner.

[ 128 ] Consequently, as in G, the Crown has failed to tender the evidence needed to discharge its burden at the minimal impairment stage. There is no basis on which to conclude that alternative measures do not achieve the purpose of ss. 490.012 and 490.013(2.1) in a real and substantial manner.

(2) Sections 490.012 and 490.013(2.1)'s Deleterious Effects Outweigh Their Salutary Effects

[ 129 ] At the heart of the s. 1 analysis is the determination of whether the salutary effects outweigh the negative impacts of the challenged measures. This is the final stage of the proportionality inquiry in Oakes. We conclude that the Crown has also not met its burden at this stage.

[ 130 ] The final stage requires a court to weigh the harm to the claimant's rights against the public benefits conferred by the challenged measure, by asking whether "the benefits which accrue from the limitation [of the claimant's rights] are proportional to its deleterious effects" (K.R.J., at para. 77, quoting Thomson Newspapers Co. v. Canada (Attorney General), [1998] 1 S.C.R. 877, at para. 125). The final stage of the Oakes test, as a result, involves a broader assessment than in the prior stages of "whether the benefits of the impugned law are worth the cost of the rights limitation" (Hutterian Brethren, at para. 77). Benefits that are speculative and marginal in nature carry less weight when balanced against a measure's significant and tangible deleterious effects (K.R.J., at para. 92; Thomson Newspapers Co., at paras. 129-30).

[ 131 ] The Crown's evidence on the challenged measures' benefits is insufficient to meet its burden under s. 1. The goal of capturing information that assists police to prevent and investigate sex offences is obviously an important consideration in whether the provisions' overbreadth is justified under s. 1. So too is the aim of giving police longer access to the information of offenders at greater risk of reoffending. To the extent the provisions succeed, they are strongly in the public interest.

[ 132 ] The Crown's arguments on the benefits of ss. 490.012 and 490.013(2.1) are premised on the alleged benefits flowing from the sex offender registry. The Crown's evidence on its efficacy was, however, limited, both on the investigation and prevention of sexual offences. The Crown cites evidence in its factum regarding the general benefits of sex offender registries. Specifically, it relies on Dr. Zgoba's testimony that some studies in the United States found registries may assist police in detaining offenders more quickly if they reoffend. Yet the experts also testified that the United States' registries were public and therefore not comparable to SOIRA. The Crown, moreover, identified no cases where SOIRA helped police solve or prevent a sex offence, either before or after the challenged measures were introduced. In fact, at the s. 1 hearing, the Crown called the registry's national database manager with the RCMP and, on cross-examination, he testified that he was unaware of any offences that were solved using SOIRA.

[ 133 ] It is, moreover, unclear how SOIRA could even prevent a sex offence. Det. Hove provided only a hypothetical example where the police used the database to intercept an offender after witnessing some suspicious behaviour. In theory, of course, the registry might prevent offences if a serial offender is apprehended using it, preventing future assaults by that person, but no evidence was adduced to support this hypothesis.

[ 134 ] More significantly, the Crown has adduced no evidence that demonstrates the salutary effects of the challenged measures. Under s. 1, the onus is on the Crown to justify the specific infringing measures, not the overall scheme (G, at para. 72). The Crown did not adduce any evidence on the difficulties that police faced in investigating sexual offences with SOIRA before the 2011 amendments and how the amendments mitigated these difficulties. No evidence was adduced to demonstrate the benefit of registering every sex offender, without regard for their risk of reoffending. To the contrary, we note that Dr. Hanson, the Crown's own expert, testified that "[b]lanket policies that treat all sex offenders as 'high risk' waste resources by over-supervising lower risk offenders and risk diverting resources from the truly high-risk offenders who could benefit from increased supervision and human service" (A.R., vol. II, at p. 236). No more evidence was presented on the benefits that flow from police having longer access to the information of offenders who have committed multiple offences without an intervening conviction.

[ 135 ] In this case, we must weigh those potential and theoretical benefits against the impact on registrants. The impact on anyone who is subject to the reporting requirements of a SOIRA order is considerable. To reiterate, SOIRA's reporting requirements are not routine: the scope of the personal information registered, the frequency at which offenders are required to update their information and, above all, the threat of imprisonment make the conditions onerous. Additionally, these effects are more acute when considering their effects on marginalized populations, such as people experiencing homelessness. Considering these deleterious impacts, the sparse evidence on the provisions' benefits and the fact that the registration of approximately 10 percent of offenders who have the lowest recidivism risk does not serve the provisions' purpose, we conclude the Crown did not meet its burden at this stage either. As result, the Crown has not shown that ss. 490.012 and 490.013(2.1) are saved under s. 1 of the Charter.

V. Result and Remedy

[ 136 ] We would allow the appeal and declare ss. 490.012 and 490.013(2.1) of no force or effect under s. 52(1) of the Constitution Act, 1982. On mandatory registration, we find a one-year suspension of the declaration is appropriate given concerns about public safety and since there are many ways Parliament could address the legislative gap for individualized assessment (G, at para. 165). An immediate declaration, however, is appropriate for lifetime registration.

[ 137 ] The framework governing Charter remedies was recently revisited in G. Once the court has determined the extent of the law's inconsistency with the Charter (para. 160), the next step is to determine whether a tailored remedy would be appropriate (such as reading down, reading in, or severance), rather than a declaration of invalidity applying to the whole of the challenged law (para. 163).

A. Section 490.012

[ 138 ] The Crown submits that "[a] tailored remedy is not appropriate in this case" (R.F., at para. 175). Since the issue is the mandatory registration of all sex offenders, its unconstitutionality does not lend itself to such a remedy. As the Crown notes, reading down s. 490.012 so that it would simply not apply to offenders who are not at an increased risk of reoffending or who suffer grossly disproportionate impacts would, in practice, reinstate judicial discretion and contradict Parliament's clear intention to remove all judicial discretion to exempt offenders at the time of sentencing from the registry (R. v. Boudreault, 2018 SCC 58, [2018] 3 S.C.R. 599, at para. 100; Schachter v. Canada, [1992] 2 S.C.R. 679, at p. 718). We agree with the Crown that the appropriate remedy is a declaration of invalidity.

[ 139 ] Declarations should be suspended when the government demonstrates that "an immediately effective declaration of invalidity would endanger an interest of such great importance that, on balance, the benefits of delaying the effect of that declaration outweigh the cost of preserving an unconstitutional law that violates Charter rights" (G, at para. 117; see also paras. 133, 139 and 156). Declaring s. 490.012 to be of no force or effect immediately would effectively preclude courts from imposing SOIRA orders on any offenders, including those at high risk of recidivism. Granting an immediate declaration could therefore endanger the public interest in preventing and investigating sexual offences committed by high-risk offenders, undermining public safety. Balanced against this consideration is the significance of the rights violation that the suspension would temporarily prolong. Granting a suspension also runs counter to the public's interest in legislation that complies with the Constitution. On balance, however, the circumstances justify a suspension of the declaration of invalidity for 12 months.

[ 140 ] A declaration of invalidity is presumed to operate retroactively (R. v. Albashir, 2021 SCC 48, [2021] 3 S.C.R. 531, at paras. 34 and 38). However, in this case, a retroactive application of the declaration at the conclusion of the suspension could frustrate the compelling public interests that require a period of transition, creating uncertainty and removing the protection that justifies the suspension in the first place (paras. 46, 52 and 72). Specifically, a retroactive declaration would undermine the purpose of the suspension (i.e., ensuring high-risk offenders are registered on SOIRA for public safety). Moreover, a prospective declaration of invalidity would not unduly prejudice offenders who have been registered since 2011 but whose rights under s. 7 are still violated. Those offenders will be able to ask for a personal remedy pursuant to s. 24(1) of the Charter in order to be removed from the registry if they can demonstrate that SOIRA's impacts on their liberty bears no relation or is grossly disproportionate to the objective of s. 490.012.

[ 141 ] Finally, we would grant Mr. Ndhlovu a remedy under s. 24(1) and exempt him from the suspension of the declaration. It is generally desirable that a claimant who brings a successful constitutional challenge benefit from their efforts in litigating the issue (G, at paras. 148 and 182). Further, the Crown only adduced expert evidence on recidivism risk at the s. 1 hearing, after the sentencing judge had made a determination that Mr. Ndhlovu was at little risk to reoffend. While the Crown's expert, Dr. Hanson, opined someone with some of the same characteristics as Mr. Ndhlovu would have an enhanced risk of reoffending, one that was about average for sex offenders, he indicated risk assessment is an individualized exercise involving many variables. That is precisely what the sentencing judge did in this case in finding that Mr. Ndhlovu was unlikely to reoffend. Moen J. found that he "will be safe to release into the community. I have absolutely no concerns that [he] will re-offend. Nor does the Crown suggest that [he] will." Nor does the Crown argue on appeal that the sentencing judge made a palpable and overriding error in making this finding of fact. Based on this finding, there is no connection between subjecting Mr. Ndhlovu to a SOIRA order and the objective of capturing information about offenders that may assist police prevent and investigate sex offences.

B. Section 490.013(2.1)

[ 142 ] With respect to lifetime registration, the Crown conceded a suspension would not be appropriate. We agree: an immediate declaration is appropriate given those offenders will remain registered and there is no "gap" for Parliament to fill. As a result, the existing provisions that dictate a length of registration will operate, pending any new constitutional provision that would target offenders who commit more than one offence. For instance, those convicted of offences with a maximum term of imprisonment of 2 to 5 years will receive a 10-year registration order, while those convicted of an offence with a maximum term of imprisonment of 10 to 14 years would receive a 20-year registration order (s. 490.013(2)). Here, there is no compelling reason to rebut the presumption of retroactive application of the declaration of invalidity. Section 490.013(2.1) is therefore declared invalid. Because the declaration affects all those impacted by the enactment of the provision since 2011, offenders who are subject to a lifetime order pursuant to this provision after having been convicted of more than one sexual offence without an intervening conviction can seek a s. 24(1) remedy to change the length of their registration.

VI. Conclusion

[ 143 ] We would allow the appeal. The judgment of the Court of Appeal of Alberta is set aside. Sections 490.012 and 490.013(2.1) of the Criminal Code infringe s. 7 of the Charter, and the Crown has not demonstrated the infringement is justified under s. 1. The provisions are therefore declared of no force or effect under s. 52(1) of the Constitution Act, 1982. The declaration in respect of s. 490.012 is suspended for one year and applies prospectively. Mr. Ndhlovu is exempted from the suspension of the declaration. An immediate declaration is granted for s. 490.013(2.1) and applies retroactively.

Reasons Dissenting in Part

(paras. 144 to 196)

Brown J. (Wagner C.J. and Moldaver and Côté JJ. concurring) —

I. Introduction

[ 144 ] I agree with the majority that mandatory lifetime registration in s. 490.013(2.1) of the Criminal Code, R.S.C. 1985, c. C-46, is overbroad. The Court of Appeal erred in saying there was no evidence to displace the "common sense inference" that individuals convicted of multiple offences at one time pose a greater risk of reoffending than those convicted of a single offence. The expert evidence was clear and uncontested: they do not. It therefore cannot reasonably be said that mandatory lifetime registration accords with the principles of fundamental justice. Nor can this be justified under s. 1 of the Charter, since Parliament could have crafted a narrower regime that distinguishes multiple sequential offences from prior convictions.

[ 145 ] I depart from the majority, however, on the constitutionality of s. 490.012. In finding it unconstitutional, my colleagues fixate on the removal of judicial discretion to exempt offenders who do not pose an "increased risk" to reoffend. But the exercise of discretion was the very problem that prompted Parliament to amend the Criminal Code to provide for automatic registration of sex offenders under the Sex Offender Information Registration Act, S.C. 2004, c. 10 ("SOIRA"). Specifically, many judges had exercised their discretion to exempt offenders in a manifestly improper manner, and the Registry's low inclusion rate undermined its efficacy. The evidence is clear that even low risk sex offenders, relative to the general criminal population, pose a heightened risk to commit another sexual offence. It is also clear that it cannot be reliably predicted at the time of sentencing which offenders will reoffend. In the face of that uncertain risk, Parliament was entitled to cast a wide net.

II. Background

[ 146 ] While the majority has summarized the background to this matter, I would stress a few points.

[ 147 ] The appellant pled guilty to sexual offences committed against two complainants, R.D. and C.B., at a party hosted by R.D. A pre‑sentence report led the sentencing judge to find he was a low risk to reoffend, that he would "be safe to release into the community", and that she had "absolutely no concerns that [he] will re-offend" (A.R., vol. II, at p. 38). She sentenced the appellant to six months' imprisonment followed by three years' probation.

[ 148 ] The sentencing judge granted the mandatory DNA order, but adjourned the imposition of a SOIRA order, instead expressly inviting counsel to challenge the constitutional validity of the amended provisions. She also volunteered her own view that it is "completely disproportionate in circumstances of a case like this to -- to put someone on a sex offender registry for the balance of their life" (p. 49). After seeking instructions, defence counsel confirmed that she would seek an adjournment to bring a constitutional challenge. The sentencing judge allowed the adjournment and explained why she had raised it: ". . . I consider this to be much harsher than any term imprisonment. So I thought it was time" (p. 50).

[ 149 ] After the hearing, the sentencing judge, unsurprisingly given her earlier statement, declared the provisions to be overbroad and grossly disproportionate to the extent they remove judicial discretion to refuse to register offenders who present no risk of reoffending (2016 ABQB 595, 44 Alta. L.R. (6th) 382 ("ABQB reasons (2016)")). She was not satisfied that including people like the appellant on the Registry would help police prevent and investigate sexual crimes. Subjecting all offenders, regardless of their future risk, to onerous reporting requirements, random compliance checks, and internal stigma goes further than is necessary to accomplish the goal of protecting the public. She further held that the limits on s. 7 could not be saved under s. 1 (2018 ABQB 277, 68 Alta. L.R. (6th) 89).

[ 150 ] A majority at the Court of Appeal of Alberta overturned her ruling (2020 ABCA 307, 12 Alta. L.R. (7th) 225). As to overbreadth, the sentencing judge defined the purpose of the legislation too narrowly. Sections 490.012 and 490.013(2.1) must be assessed in light of SOIRA's broader purpose: to protect society from recidivist sexual offenders. Low risk does not equate to no risk. According to the majority, the commission of a sex offence is a reasonable proxy for the risk of recidivism and the appellant is an offender who, by virtue of being convicted of more than one offence, poses enhanced risk. In dissent, Khullar J.A. held that ss. 490.012 and 490.013(2.1) limit the appellant's privacy and liberty rights under s. 7. The appellant was unlikely to reoffend and the provisions requiring him to register for life overreach in their effect. She concluded that the Crown failed to justify the limits as minimally impairing and proportionate.

III. Analysis

A. Overview of Section 7 Analysis

[ 151 ] The appellant says the removal of discretion deprives offenders of liberty in a manner that is not in accordance with the principles of fundamental justice. He raises three such principles: arbitrariness, overbreadth, and gross disproportionality. All three are concerned with "failures of instrumental rationality" (Canada (Attorney General) v. Bedford, 2013 SCC 72, [2013] 3 S.C.R. 1101, at para. 107, quoting H. Stewart, Fundamental Justice: Section 7 of the Canadian Charter of Rights and Freedoms (2012), at p. 151). First, arbitrariness describes the absence of a rational connection between the law's purpose and the impugned effect on the individual, or where "it can be shown that the impugned effect undermines the objective of the law" (R. v. Michaud, 2015 ONCA 585, 127 O.R. (3d) 81, at para. 69, citing Bedford, at para. 111). Next, a law will be overbroad where it captures "some conduct" unrelated to its purpose, recognizing that a law can be rational in some cases but overreach in others (Bedford, at paras. 112 and 117 (emphasis deleted)). Finally, gross disproportionality will be found where "the law's effects on life, liberty or security of the person are so grossly disproportionate to its purposes that they cannot rationally be supported" (Michaud, at para. 71, quoting Bedford, at para. 120). As I explain below, s. 490.012 accords with all these principles.

[ 152 ] The s. 7 analysis proceeds in four stages. First, I consider the purpose of s. 490.012(1), which is to help police prevent and investigate sexual crimes by requiring the registration of certain information relating to sex offenders. Next, I address why automatic registration is neither arbitrary nor grossly disproportionate in light of that objective. In the final section, I turn to overbreadth. As I will explain, automatic registration does not go further than necessary to achieve its purpose, because all sex offenders pose a higher risk to reoffend and the risk of recidivism cannot be reliably predicted at the time of sentencing. The sentencing judge erred in concluding otherwise, and my colleagues in the majority perpetuate this error.

B. Identifying the Purpose of Section 490.012

[ 153 ] Section 490.012 contains no explicit statement of the purpose of automatic registration. The broader legislative scheme offers some clues, though. Section 2(1) of SOIRA states that "[t]he purpose of this Act is to help police services prevent and investigate crimes of a sexual nature by requiring the registration of certain information relating to sex offenders." The guiding principles of the legislation include: (a) police having rapid access to certain information relating to sex offenders; (b) ensuring such information is current and reliable by collecting it on an ongoing basis; and (c) protecting the privacy and rehabilitation interests of sex offenders by requiring that the information be collected only to enable police services to prevent or investigate crimes of a sexual nature and restricting access to the information (s. 2(2)).

[ 154 ] Extrinsic evidence reveals that Parliament amended the statutory scheme in response to concerns that the Registry's efficacy was compromised by the exclusion of nearly half of all convicted sex offenders. Parliament's reason for creating the Registry and the data on the number of offenders being registered themselves substantiate these concerns.

[ 155 ] The creation of sex offender registries in Canada was prompted by a growing public concern about "protecting children and other vulnerable victims from sex offenders through a system of careful monitoring of convicted sex offenders" (R. v. Long, 2018 ONCA 282, 45 C.R. (7th) 98, at para. 86). That concern was exemplified by the enactment of Ontario's registry in 2001, which applied automatically to everyone in the province who was convicted of, or was already serving a sentence for, a listed sexual offence (Christopher's Law (Sex Offender Registry), 2000, S.O. 2000, c. 1, s. 2). The national Registry soon followed in 2004.

[ 156 ] But after several years, lawmakers queried whether the Registry was fulfilling its purposes. The government struck a standing committee in 2009 to review SOIRA. The committee gathered evidence and recommended specific amendments "to help police departments prevent crimes of a sexual nature, solve them more quickly, and more effectively supervise sex offenders in the community" (House of Commons, Standing Committee on Public Safety and National Security, Statutory Review of the Sex Offender Information Registry Act: Report of the Standing Committee on Public Safety and National Security, 2nd Sess., 40th Parl., December 2009, at p. 2).

[ 157 ] When the committee compared the two registries, it found a significant divergence in the inclusion rate: Ontario's registry included the names of 96.84% of sex offenders, while only approximately 50% of offenders were included on the national Registry (House of Commons, Standing Committee on Public Safety and National Security, Evidence, No. 15, 2nd Sess., 40th Parl., April 21, 2009, at p. 4; Standing Committee on Public Safety and National Security, December 2009, at p. 8). And while the Ontario registry was consulted 475 times per day, the national Registry was only consulted an average of 165 times per year (Standing Committee on Public Safety and National Security, December 2009, at p. 5). Accordingly, the committee recommended, among other things, that registration should be automatic for offenders found guilty of a designated sexual offence, as is the case in Ontario, except in rare circumstances where the impact on the offender would be grossly disproportionate to the public interest.

[ 158 ] The government provided a written response, explaining that it would adopt most of the committee's recommendations by asking Parliament to enact the Protecting Victims From Sex Offenders Act, S.C. 2010, c. 17. With respect to automatic registration, the government agreed that the current model "presented concerns as some convicted sex offenders have not been included on the Registry in cases where the Crown Prosecutor has not made an application or judges have concluded that the offender met the 'grossly disproportionate' test" (Government Response to the Fifth Report of the Standing Committee on Public Safety and National Security Entitled Statutory Review of the Sex Offender Information Registry Act, April 12, 2010 (online)). The amendments responded to those concerns by creating "a fully automatic inclusion model", to make the Registry "as effective as possible when police are investigating crimes of a sexual nature" (Government Response).

[ 159 ] The same themes emerge from the legislative committees and debates. During second reading of Bill C-34, Protecting Victims From Sex Offenders Act, 2nd Sess., 40th Parl., 2009, the Minister of Public Safety emphasized that mandatory inclusion would eliminate the "flaw" in the current legislation that allows convicted sex offenders to avoid being added to the Registry, "which hampers future police investigations and exposes Canadians to greater risk" (House of Commons Debates, vol. 144, No. 67, 2nd Sess., 40th Parl., June 3, 2009, at p. 4148). Put another way, the amendment aimed to remove a "loophole" that meant the police had "no access to information on some convicted sex offenders during the investigation of a crime, either because a crown attorney has not sought an order for them to register or the presiding judge has declined to grant one" (House of Commons Debates, vol. 144, No. 70, 2nd Sess., 40th Parl., June 8, 2009, at p. 4324). As RCMP Inspector Nezan explained to the Standing Committee on Public Safety and National Security, police were concerned about recidivists "falling through the cracks":

The absence of an automatic inclusion on the registry of all offenders convicted of sexual crimes has led to the inconsistent application of the law across the country. Someone convicted of molesting a child in one province may be ordered to the registry, while in another province they may not. Given the difficulty of determining which sex offender will reoffend and which will not, this means that some of the recidivists are falling through the cracks. [Emphasis added.]

(April 21, 2009, at p. 3)

[ 160 ] In light of this evidence of legislative intent, the statements of purpose put forward by both the sentencing judge and the majority of the Court of Appeal should be rejected. The sentencing judge held that the purpose of SOIRA was "to protect vulnerable people including children in society, by allowing police quick access to current information on convicted sex offenders" (ABQB reasons (2016), at para. 87; see para. 85). Nothing in the legislative text or proceedings reveals such a narrow purpose. Public safety for vulnerable persons and children, though likely a desirable effect of the legislation, was not a cited purpose (SOIRA, s. 2(1)). The legislation was intended to increase safety for all, not just vulnerable persons (R. v. Redhead, 2006 ABCA 84, 384 A.R. 206, at paras. 37-38; R. v. T.A.S., 2018 SKQB 183, at para. 79; Long, at para. 89). For the majority at the Court of Appeal, the legislative purpose of s. 490.012 was broader: ". . . to require mandatory registration of all sex offenders convicted of designated offences" (para. 74). But that statement erroneously conflates the purpose of the law with the means for achieving the purpose. The Court of Appeal's reasoning is circular: if mandatory registration is the purpose, then requiring mandatory registration will always be rationally connected to that purpose.

[ 161 ] The foregoing evidence supports Khullar J.A.'s understanding of the purpose of the automatic registration requirement imposed by s. 490.012(1) as conforming to the overall statutory objective: ". . . to help police services prevent and investigate crimes of a sexual nature by requiring the registration of certain information relating to sex offenders" (para. 198; SOIRA, s. 2(1)). This statement of purpose is sufficiently precise and succinct to frame the s. 7 analysis. It is similar to the majority's statement of purpose, being "to capture information about offenders that may assist police prevent and investigate sex offences" (para. 73). I depart from the majority, however, in its conclusion that automatic registration of sex offenders is not rationally connected to the objective of helping police prevent and investigate sexual offences. As I explain below, it clearly is.

C. Arbitrariness

[ 162 ] The sentencing judge did not find s. 490.012 to be arbitrary, saying it was "not the case that there is no connection between providing police with up-to-date information on previous sex offenders and the goal of investigating and preventing sexual crimes" (ABQB reasons (2016), at para. 92). The majority below agreed (para. 42).

[ 163 ] The appellant has not demonstrated any error in those conclusions. There is a clear connection between having accurate and up-to-date information about persons more likely to commit sexual offences and the investigation and prevention of sexual crimes. Mandatory registration of convicted sex offenders in SOIRA is not arbitrary.

D. Gross Disproportionality

[ 164 ] The sentencing judge concluded that mandatory registration is grossly disproportionate. She found that the cumulative effects of registration were significant, taking into account not only the scope of the reporting requirements, but also the other deleterious effects on offenders, such as the impact of random compliance checks, the risk of information being divulged during those checks, and the potentially far-reaching effects on an offender's privacy (ABQB reasons (2016), at para. 124). The majority, while not reaching a conclusion on gross disproportionality, observes that SOIRA conditions are "onerous and ongoing" (para. 9; see also paras. 45, 56, 83, 116 and 135).

[ 165 ] With respect, to describe SOIRA conditions as "onerous" strips that term of all meaning. The Court of Appeal for Ontario has characterized the effects of registration on liberty as "modest", relative to the important state interest (R. v. Dyck, 2008 ONCA 309, 90 O.R. (3d) 409 ("Dyck (C.A.)"), at paras. 104, 106 and 109; Long, at para. 147). Applying Dyck (C.A.) and Long, the Court of Appeal in this case concluded the SOIRA requirements were, "if not strictly modest, reasonable and not dissimilar to many other reporting obligations that routinely occur as part of the everyday life of all Canadians, who are state-compelled to provide information from time to time" (para. 130). Moreover, these conditions that my colleagues find so "onerous" flow from the very fact that it is a registry. A registry is effective only when it has complete and up-to-date information; absent those conditions, there would be no point to even having it. My colleagues, I note, do not describe conditions of a registry that would be less "onerous" while still allowing it to be effective.

[ 166 ] I agree that the sentencing judge erred in characterizing the impact of registration. SOIRA conditions do not prevent offenders from going anywhere or doing anything (C.A. reasons, at para. 131, citing Dyck (C.A.), at paras. 106 and 111; R. v. Debidin, 2008 ONCA 868, 94 O.R. (3d) 421, at para. 82; transcript, at p. 145). The appellant raised concerns that the information might be leaked, but these concerns are entirely speculative. Access to the Registry is strictly controlled. The reporting is confidential and use of the information is strictly limited to police for the prevention and investigation of sexual offences (Dyck (C.A.), at para. 119; R. v. Cross, 2006 NSCA 30, 241 N.S.R. (2d) 349, at para. 84 (6.)). While the appellant also testified to the "stigma" he experiences from being constantly labelled a sex offender, any stigma flows from the convictions themselves, not registration (C.A. reasons, at paras. 135‑37, citing R. v. C. (S.S.), 2008 BCCA 262, 234 C.C.C. (3d) 365, at paras. 47-48; Dyck (C.A.), at para. 118; Cross, at para. 55). Further, as Khullar J.A. observed, "[a]nxiety caused by a law is not usually considered an infringement of liberty" (para. 224) and the sentencing judge did not make a finding that the appellant's concerns deprived him of his security of the person (paras. 225-26).

[ 167 ] Not only does my conclusion align with appellate jurisprudence, but it also avoids the serious doctrinal difficulties posed by the majority's reasons. If the majority is correct that SOIRA orders have a "serious" and "considerable" impact on an offender's liberty (see paras. 7, 45, 54, 57 and 135), the test for punishment as set out in R. v. K.R.J., 2016 SCC 31, [2016] 1 S.C.R. 906, at para. 41, per Karakatsanis J., would likely be met. And yet, this would be inconsistent with appellate jurisprudence consistently holding that sex offender registry requirements do not amount to punishment (see Cross; Dyck (C.A.); R. v. Hooyer, 2016 ONCA 44, 129 O.R. (3d) 81, at para. 45).

[ 168 ] Further, the majority's conclusion would likely engage s. 11(i) of the Charter, such that no one convicted of a sexual offence prior to 2004 could be required to register on the SOIRA registry, and that Parliament could not enact a new SOIRA law that expressly applies retroactively. Equally alarming are the long-term implications of the majority's conclusion. Unlike other types of punishment, SOIRA requirements do not significantly limit lawful activities in which the accused can engage, where the accused can go, or with whom an accused can communicate or associate. The "most obvious impact on liberty", according to the majority, is the risk of imprisonment for failure to meet the reporting requirements without "reasonable excuse" (para. 54). On that logic, other measures that appellate courts have previously concluded were not punishment may now constitute punishment. While my majoritarian colleagues "make no finding as to whether SOIRA orders constitute punishment under the K.R.J. test" (para. 58), that is entirely beside the point. The point is that it is implicit in their analysis that these orders likely constitute punishment. This is no mere theoretical concern. It is hardly unprecedented, after all, for a court to expressly decline to decide an issue while, at the same time, effectively deciding it by adopting a particular line of reasoning (see, e.g., and taken together, R. v. Moriarity, 2015 SCC 55, [2015] 3 S.C.R. 485, at para. 30; R. v. Stillman, 2019 SCC 40, [2019] 3 S.C.R. 144, at paras. 97-109 and 113).

[ 169 ] Further, and even where the impacts on an offender's liberty and privacy interests are more than modest, those impacts must be considered in context. An offender who is placed on the Registry has already been tried and convicted of a designated sexual offence (Long, at para. 147). Until now, a SOIRA order has not been characterized as a sentence or a punishment, but a consequence of conviction (Redhead, at para. 12). A person convicted of a serious crime should expect a significant loss of personal privacy (R. v. F. (P.R.) (2001), 57 O.R. (3d) 475 (C.A.), at para. 18, per Rosenberg J.A.; R. v. Beare, [1988] 2 S.C.R. 387, at p. 413). And the impacts must be measured against the inherent severity of the offending conduct (C.A. reasons, at paras. 79-81 and 232). The duration of a SOIRA order is directly connected to the maximum term of imprisonment for that sexual offence (s. 490.013(2)). By linking the length of time for reporting to the severity of the offence, Parliament built proportionality into the legislative scheme (I.F., Attorney General of Nova Scotia, at para. 64).

[ 170 ] Another consideration militating against the appellant's arguments on gross disproportionality is the availability of a termination order. The regime permits offenders to apply to terminate SOIRA orders after (a) 5 years, if they are subject to an order for 10 years; (b) 10 years if they are subject to a 20-year order; and (c) 20 years if they are subject to a lifetime order (s. 490.015(1)). The termination provision allows judges to assess the impacts of SOIRA at a later date when the offender can demonstrate they have been in the community without reoffending, rather than making "speculative predictions of future dangerousness" based on the limited evidence available at sentencing (J. Benedet, "A Victim-Centred Evaluation of the Federal Sex Offender Registry" (2012), 37 Queen's L.J. 437, at p. 468). Termination orders are undoubtedly rare, since an offender must meet the "high standard" of showing the impact on the offender's privacy or liberty is "grossly disproportionate to the public interest" (s. 490.016(1); Redhead, at para. 43). But if there is truly a disproportionate impact, a termination order will be available (R. v. Dyck (2005), 203 C.C.C. (3d) 365 (Ont. S.C.J.), at para. 125).

[ 171 ] Finally, the sentencing judge failed to consider the impacts of SOIRA in light of its aim: to investigate and prevent sexual offences. As Khullar J.A. held, "[t]his is not the kind of case where the objective is simply not important enough to warrant the level of infringement of liberty that it imposes" (para. 232). SOIRA imposes a burden on offenders, but that burden is not "totally out of sync" with the objectives (Bedford, at para. 120). I agree with the Attorney General of Nova Scotia that characterizing reporting obligations as "grossly disproportionate" in this context risks trivializing that term (I.F., at para. 61).

E. Overbreadth

[ 172 ] A law is overbroad, the Court explained in R. v. Safarzadeh-Markhali, 2016 SCC 14, [2016] 1 S.C.R. 180, if it goes "further than reasonably necessary to achieve its legislative goals" (para. 50). There is a heavy onus on the party alleging that legislation is overbroad:

As noted above, the root question is whether the law is inherently bad because there is no connection, in whole or in part, between its effects and its purpose. This standard is not easily met. The evidence may, as in Morgentaler, show that the effect actually undermines the objective and is therefore "inconsistent" with the objective. Or the evidence may, as in Chaoulli, show that there is simply no connection on the facts between the effect and the objective, and the effect is therefore "unnecessary". Regardless of how the judge describes this lack of connection, the ultimate question remains whether the evidence establishes that the law violates basic norms because there is no connection between its effect and its purpose. This is a matter to be determined on a case-by-case basis, in light of the evidence. [Emphasis added.]

(Bedford, at para. 119, per McLachlin C.J.)

[ 173 ] Having identified the purpose of s. 490.012(1), the question here is whether automatic registration of all offenders goes further than reasonably necessary to achieve the objective of helping police prevent and investigate sexual crimes by requiring the registration of certain information. The sentencing judge acknowledged there was a statistical probability that a sex offender will reoffend, but concluded the law was overbroad because it captured individuals who have little or no chance of reoffending (ABQB reasons (2016), at para. 119). The Crown does not dispute that the Registry may impose obligations on offenders who never end up reoffending, but emphasizes that all convicted sex offenders pose a higher risk and it is impossible to say at the time of sentencing that a particular offender poses no risk of reoffending.

[ 174 ] The Crown's argument is supported by the expert evidence. Offenders convicted of a sexual offence are five to eight times more likely to reoffend than those convicted of a non-sexual offence. Both experts agreed that the recidivism risk cannot be determined with certainty at the time of sentencing. Dr. Zgoba explained that a risk assessment cannot guarantee whether any individual will reoffend. Both experts also testified that observed recidivism rates underestimate the true rates of sexual reoffending. Professor Benedet makes a similar point, saying recidivism studies do not typically measure whether offenders commit further offences (or have already committed other offences in the past) for which they were never apprehended or charged ("Long and Ndhlovu: The Federal Sex offender Registry and Section 7 of the Charter" (2018), 45 C.R. (7th) 132; see also Standing Committee on Public Safety and National Security, April 21, 2009, at p. 9).

[ 175 ] The Crown's argument is also supported by logic and experience. Assessing future risk is "inherently imprecise" (Long, at para. 125). The Court of Appeal for Ontario has confirmed that, as a group, sex offenders will always pose "a greater risk than the rest of the population for engaging in that activity" and, thus, a prior conviction for a sex offence is "a reliable indication of risk and a proper method of assessing that risk" (Dyck (C.A.), at para. 100). More recently, the court in R. v. B.P.M., 2019 BCPC 156, dismissed a similar overbreadth argument. It concluded that all sex offenders should be registered because, "as a matter of common sense and experience, individuals who are convicted of sex crimes have an increased propensity to commit sex crimes in the future" (para. 48, quoting Long, at para. 119). Inspector Nezan explained in his Standing Committee on Public Safety and National Security testimony that predicting recidivism is "very difficult" and that, from a police perspective, it is "dangerous" to use a risk-based assessment to determine which offenders should be registered (April 21, 2009, at p. 8).

[ 176 ] The majority ignores this evidence, instead hinging its analysis on two premises: (1) that certain offenders do not pose an "increased risk" of reoffending; and (2) that registering such offenders bears no connection to the purpose. Indeed, the phrase "increased risk" is repeated 20 times in the majority's reasons (see paras. 7‑8, 10‑11, 79, 82‑83, 85, 87, 89‑90, 92, 100‑101, 111, 121 and 138).

[ 177 ] My colleagues do not explain their basis for rejecting the expert evidence that an offender's risk cannot be determined with certainty at the time of sentencing. They accept that "the commission of a sexual offence is one of many empirically validated predictors of increased sexual recidivism" (para. 94). In the same breath, however, they assert that, "[i]n certain cases, an offender's personal circumstances mean they are not at an increased risk of reoffending, undermining any real possibility that their information on the registry will ever prove useful to police" (para. 85). My colleagues boldly promenade where experts fear to tread, leaning on the term "increased" to avoid particularizing what they rely on to characterize this risk as such. In support, they claim to cite multiple "examples" (para. 90) of offenders whose likelihood of reoffending is "highly improbable" (para. 89). But in substance they cherry pick just one such example: an exceptional case involving an offender who was wheelchair bound. 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.

[ 178 ] The majority's analysis suffers from a further flaw. My colleagues assert that uncertainty in predicting the risk of reoffending is irrelevant in overbreadth, since enforcement practicality can only be considered under s. 1. They rely on Carter v. Canada (Attorney General), 2015 SCC 5, [2015] 1 S.C.R. 331, Bedford and Safarzadeh‑Markhali, where this Court held that difficulties in defining the scope of the law to adequately serve the legislative purpose are not relevant in overbreadth (majority reasons, at paras. 103‑7).

[ 179 ] But unlike those instances, here Parliament did not struggle to define the scope of the law. In enacting s. 490.012, Parliament deliberately chose not to distinguish between more serious and less serious sexual offences or higher risk and lower risk offenders. Instead, Parliament required registration for all sex offenders based on a shared characteristic: a heightened risk of committing a future sexual offence. The government can enact legislation that treats all individuals with a common characteristic in the same manner, without offending s. 7, provided there is a rational connection between that characteristic and the government's purpose. And here, there is a "logical link" between automatic registration on the basis of a sex conviction and the purpose of s. 490.012. On the information available at the time of sentencing, every person convicted of a designated offence poses a heightened risk of committing another sexual offence as compared to persons convicted of non-sexual offences.

[ 180 ] The enhanced risk may be marginal in some cases, but "Parliament is not required to precisely calibrate recidivism risk" (C.A. reasons, at para. 89). Nor is "[p]erfect symmetry between the social science measuring rates of recidivism and the reach of the registry" required (C. (P.S.) v. British Columbia (Attorney General), 2007 BCSC 895, 222 C.C.C. (3d) 230, at para. 118; see also R. v. Sharpe, 2001 SCC 2, [2001] 1 S.C.R. 45, at para. 89). Parliament is entitled to act on reasoned apprehension of harm even if on some points "the jury is still out" (R. v. Malmo-Levine, 2003 SCC 74, [2003] 3 S.C.R. 571, at para. 78). As the majority at the Court of Appeal explained, any exclusion of convicted sex offenders based on perceived seriousness of an offence or likelihood of recidivism necessarily results in the police not having information on some offenders who do, in fact, reoffend (para. 93).

[ 181 ] It was within Parliament's purview to draw a line based on that known increased risk of unknown degree, rather than leaving it to prosecutors and judges to weigh if an offender poses an "increased risk" in each case. Once it is demonstrated that the harm sought to be avoided is more than trivial, "the precise weighing and calculation of the nature and extent of the harm is Parliament's job" (Malmo-Levine, at para. 133). After that threshold is reached, the legislature is "entitled to make policy choices within a reasonable range of options" (Dyck (C.A.), at para. 124). A small risk of serious harm is sufficient.

[ 182 ] My colleagues overlook another important reason that Parliament drew that line: to fix a flaw in the Registry that allowed judges to exempt offenders who they deemed not to be "predators", contrary to the legislative intent. Professor Benedet observes that, prior to the amendments, some judges granted exemptions based not on the impact of registration on the offender (as required by the statutory provision) but on whether the offender was the type of person for whom the Registry was intended — in other words, a "real" sex offender (Benedet (2012); transcript, at p. 85). The category of "real" sex offender has been, at times, defined so narrowly as to exclude offenders who sexually assaulted people they knew, child pornography users, opportunistic offenders, and historic offenders. A few examples are telling.

[ 183 ] Beginning with R. v. Have, 2005 ONCJ 27, 194 C.C.C. (3d) 151, a line of decisions found that the aim of the Registry was to assist police in the investigation of crimes by "predatory strangers" and offenders who do not fit this model should be exempted. In Have, the offender pled guilty to two counts of possession of child pornography after police seized a large quantity of files from his computer. Duncan J. characterised the circumstances as "exceptional", because possession of child pornography is one of the few sexual offences that does not directly harm a victim and the accused had established that he had no propensity to act on his sexual interests. According to Duncan J., the "fundamental assumption" that the accused, because of his past conduct, had a propensity to commit a sexual crime in the future was "considerably weaker" than in most cases. The value of registration in those circumstances was "negligible", the impact on the accused was "substantial", and the balance, in his view, was grossly disproportionate.

[ 184 ] Following Have, Caldwell J. in R. v. Burke, 2005 ONCJ 422, found that the impact of SOIRA registration on the accused was grossly disproportionate and declined to make an order. Mr. Burke was convicted of sexual assault after trial. The judge found that he grabbed the complainant, got on top of her, and began kissing her, touching her breasts, and touching her under her shorts in the area of her vagina (paras. 2 and 4). At sentencing, he did not present any evidence of his risk of reoffending, but the judge inferred it was low based on his lack of criminal history (paras. 59 and 63). Caldwell J. found it "extremely significant that [SOIRA] obligations are on-going for a minimum of ten years and that the individual is potentially subject to conviction, fine and/or imprisonment if they provide inaccurate information" (para. 17). Mr. Burke's offence was "on the lower end of the spectrum" (para. 58), and therefore, the public interest in his registration was "substantially lower than the interest in registering the type of offender upon whom the registry was modelled" (para. 64).

[ 185 ] Notably, both judges found that registration was a significant intrusion for offenders who were not predatory strangers and remarked that an overly inclusive registry could be counter-productive because it could dilute police resources (Have, at p. 157; Burke, at para. 39). Those propositions have been discredited by appellate courts. The Court of Appeal of Alberta in Redhead rejected the conclusion from Have that the Registry's purpose was limited to investigating predatory offenders with a propensity to commit similar offences in the future. Specifically, it addressed the argument that the accused was not the type of offender covered by SOIRA because he did not have a criminal history of the same nature, the offence was less serious, and the risk of reoffending was low. The court held:

We do not agree. The language of s. 490.012 does not suggest its application is so limited. Rather, the absence of such limiting language reflects Parliament's recognition of predictable repetitive behaviour of sexual offenders, and the inordinate consequences of sexual offences for victims of any age. [para. 38]

[ 186 ] The British Columbia Court of Appeal followed Redhead in R. v. Y. (B.T.), 2006 BCCA 331, 210 C.C.C. (3d) 484, overturning a decision granting an exemption to an offender who had sexually abused his daughter over a six-year period. The trial judge found that he was not a "predatory stranger" and thus not caught within the purpose of the legislation. Rowles J.A. wrote that, by focusing the purpose of SOIRA on a particular class of offender, the trial judge "created a classification scheme that Parliament did not enact" (para. 39). The Court of Appeal for Ontario also rejected the predatory stranger model in Debidin, observing that SOIRA "does not distinguish between the predatory stranger and the opportunistic friend, relative or custodian" (para. 77).

[ 187 ] Yet, judges did not only exempt offenders deemed to be non-predatory or unlikely to reoffend. Even violent offenders who were likely recidivists have been exempted from SOIRA. In R. v. B.S.S., 2006 BCPC 135, the accused pled guilty to sexual assault and assault against his wife. He had a serious drinking problem and had a history of violence. On the day of the first offence, he raped his wife. He was assessed as posing a high risk of future violence against her. Nevertheless, Baird Ellan Prov. Ct. J. exempted the accused from registration, following Have and Burke which held that the purpose of the Registry was to deal with predatory offenders and it was not "designed to address this kind of offence" (para. 82), whatever that means.

[ 188 ] Judges have granted exemptions even where the victim was a stranger and the offences were highly predatory. In R. v. Worm, 2005 ABPC 92, the accused pled guilty to sexual assault and assault with a weapon. He grabbed the complainant's buttocks as she was jogging and threatened bystanders with a knife before running away. He was assessed at a low to moderate risk of recidivism, with the risk increasing significantly during periods of intoxication. Though the offender was a drunk predatory stranger armed with a weapon, the sentencing judge nevertheless found that the impact of registration would be grossly disproportionate "having regard to the circumstances and severity of the offence" (para. 53).

[ 189 ] Perhaps most concerning is R. v. Randall, 2006 NSPC 38, 247 N.S.R. (2d) 205, where the accused was convicted of internet luring after inviting an undercover police officer, believed to be a 13-year-old girl, to meet him for sexual activity. He insisted on meeting for sex, rejecting her suggestion that they go shopping instead. The accused arrived with packets of condoms and was arrested. The sentencing judge found that the accused was in denial about his behaviour, but still granted an exemption because he was assessed at a low risk of reoffending, "his conduct was not predatory" but rather "poor judgment", he was not considered a "hunter," and he had no prior sexual offences (para. 16). The judge was also concerned that registration would severely affect the accused because he would "carry a stigma" and his privacy would be impacted (para. 15). It is difficult to imagine a more predatory activity than online child luring (Benedet (2012), at p. 455) — yet that was not enough for the offender to be caught by SOIRA.

[ 190 ] And finally, exemptions have been granted to offenders who occupied positions of trust and abused vulnerable victims. In R. v. Aldea, 2005 SKQB 461, 271 Sask. R. 272, a priest prostituted underage girls in the church rectory and took pornographic photographs of each of them. The court held that an exemption was justified, "having regard to the nature of his vocation, his current living circumstances", and his low risk to reoffend (para. 40). In R. v. S. (M.W.), 2007 BCSC 1188, 52 C.R. (6th) 77, a doctor was convicted of sexually assaulting female patients under the guise of medical examinations over a 20-year period, including having sexual intercourse with one patient. Though the offender had little insight into his crimes and insisted the assaults were "misunderstandings" on the part of the victims, the psychologists assessed him as a low risk to reoffend because he was no longer practicing medicine (paras. 17 and 67). Bruce J. granted an exemption, noting that the public interest served by registration was "minimal to non-existent" (para. 72), SOIRA's impact on him would be "substantial" (para. 74), and he was "not the type of person Parliament had in mind when it created the SOIRA registry" (para. 76).

[ 191 ] Those cases ⸺ and many others decided prior to the amendments in 2011 ⸺ demonstrate that judges too often found "reasons" to exempt offenders who committed offences they viewed as less serious, who had no prior history of sexual offending, or who did not appear to pose a high risk (or, to use the majority's phrase, an "increased risk") to strangers. They also "provide a window into judicial thinking about the relative seriousness of types of sexual offending, and show that the exercise of judicial discretion in the area of sexual assault is fraught with the persistence of problematic assumptions about what a 'real' sex offender looks like" (Benedet (2012), at p. 440).

[ 192 ] As if to remove all doubt on this point, similar assumptions about what a "real" sex offender looks like were also relied upon by the sentencing judge in this case. She found that "low risk" offenders, such as the appellant, fall in the category of offenders who should not be captured by SOIRA, and therefore the law is overbroad. But this reasoning is flawed for two reasons. First, it ignores that "the purpose of the regime is not limited to tracking predators" and the appellant nonetheless poses a heightened risk of reoffending (T.A.S., at para. 79; see Long, at para. 89). Secondly, it assumes that offenders who commit "minor" sex crimes are less likely to reoffend. There is no evidence supporting this assumption. Both experts testified that the seriousness of a sexual offence is a poor predictor of recidivism on its own. The Registry aims to help police prevent and investigate all sexual crimes. Under s. 7, the court's task is not to assess the law's efficacy but simply to compare the objective with its effects (Bedford, at para. 123). Here, the objective (preventing and investigating sexual offences) is clearly furthered by its effect (registering all sex offenders) because we do not know who will reoffend, but we know that many will, with serious effects.

[ 193 ] On that point, I would endorse the comments of the Court of Appeal that digital penetration of a sleeping victim is never "minor" or "de minimis" conduct (paras. 77-84). All forms of sexual violence are morally blameworthy, precisely because they involve wrongful exploitation and denial of the victim's dignity as a human being (R. v. Friesen, 2020 SCC 9, [2020] 1 S.C.R. 424, at para. 89, citing R. v. Mabior, 2012 SCC 47, [2012] 2 S.C.R. 584, at paras. 45 and 48). This Court has recently described the serious harms that flow from sexual offences, particularly against children, which "are notorious for their devastating impact, often ruining the lives of their victims, and of those whose lives intersect with those victims as they move into adulthood" (K.R.J., at para. 131). Sentencing judges must "properly understand the wrongfulness of sexual offences" and "the profound harm that they cause" (Friesen, at para. 50).

[ 194 ] Given what appears to have been a persistent, routine failure to appreciate the seriousness of these offences, it can be confidently predicted that the rampant misuse of judicial discretion prior to the amendments will recur once automatic registration is removed. There is a remarkable consistency of views between the judgment in Have, which viewed the possession of child pornography as essentially a victimless crime, and the sentencing judge's reasons in this case, which refused to characterize the penetration of a sleeping woman as a major sexual assault. Indeed, and as Schutz J.A. wrote, the sentencing judge's own conclusions prove Parliament's point in making all orders mandatory: ". . . different judges may make different assessments about the seriousness of criminal conduct and an offender's risk to reoffend" (C.A. reasons, at para. 79). This explains Parliament's decision to remove judicial discretion to include all offenders convicted of a designated offence.

[ 195 ] Parliament sought to avoid those chronic and improper exercises of judicial discretion by drawing a clear line: all sexual offences are serious and all sex offenders pose a heightened risk. Both premises are backed by expert evidence, judicial precedents, and logic. The majority's reasons suggest that judicial discretion cannot be fettered. But following the majority's analysis to its inevitable conclusion, Parliament could never remove judicial discretion from a criminal law scheme. And yet, is that not what all legislation does to some extent? This Court's role is limited to examining legislation for Charter compliance, not second-guessing policy decisions. Operating within that confine, I am constrained to conclude that s. 490.012 is appropriately tailored to its purpose of helping the police prevent and investigate sexual crimes, and does not limit an offender's s. 7 rights in a manner that bears no connection to its objective.

IV. Conclusion

[ 196 ] I would allow the appeal in part. Section 490.012 is constitutional.

Appeal allowed, Wagner C.J. and Moldaver, Côté and Brown JJ. dissenting in part.

Solicitors

Solicitors for the appellant: Iginla & Company, Edmonton.

Solicitor for the respondent: Alberta Crown Prosecution Service, Appeals and Specialized Prosecutions Office, Edmonton.

Solicitor for the intervener the Attorney General of Canada: Attorney General of Canada, Vancouver.

Solicitor for the intervener the Attorney General of Ontario: Attorney General of Ontario, Toronto.

Solicitor for the intervener the Attorney General of Quebec: Attorney General of Quebec, Québec.

Solicitor for the intervener the Attorney General of Nova Scotia: Attorney General of Nova Scotia, Halifax.

Solicitor for the intervener the Attorney General of British Columbia: Attorney General of British Columbia, Victoria.

Solicitors for the intervener the Criminal Lawyers' Association (Ontario): Daniel Brown Law, Toronto.

Solicitors for the intervener the Canadian Civil Liberties Association: Henein Hutchison, Toronto.

Solicitors for the interveners the HIV & AIDS Legal Clinic Ontario and the HIV Legal Network: Henein Hutchison, Toronto; HIV & AIDS Legal Clinic Ontario, Toronto.

Solicitor for the intervener the Mobile Legal Clinic: Mobile Legal Clinic, Montréal.

Solicitors for the intervener Association québécoise des avocats et avocates de la défense: Caissy et Pelletier-Quirion, avocats, Amqui, Que.