This appeal concerns the constitutionality of New York's version of "Megan's Law."1 The statute requires sex offenders, after serving their sentences, to register with law enforcement officials, and provides for various degrees of public notification of the identity and address of these offenders. The specific issue is whether the statute inflicts "punishment," in which event the Ex Post Facto Clause would prohibit its application to those who committed their offenses prior to enactment of the statute.

The three plaintiffs, proceeding under pseudonyms, effectively represent all others similarly subject to the retroactive application of the registration and public notification provisions of New York's Sex Offender Registration Act ("SORA" or "the Act"), N.Y. Correct. Law §§ 168-168v (McKinney Supp.1997).2 They challenged the Act on ex post facto and other constitutional and statutory grounds. On cross-motions for summary judgment by the plaintiffs and the defendants--the Governor of New York and other state officials and agencies responsible for the implementation and operation of the SORA--the District Court for the Southern District of New York (Denny Chin, Judge) upheld retroactive application of the Act's registration requirement, but ruled that similar application of the Act's community notification provisions constituted "punishment" in violation of the Ex Post Facto Clause. See Doe v. Pataki, 940 F. Supp. 603 (S.D.N.Y.1996) ("Doe v. Pataki II "). Judge Chin permanently enjoined enforcement of the notification provisions of the Act against persons whose crimes occurred before January 21, 1996, the effective date of the Act.

Both parties appeal from the District Court's judgment, entered on September 26, 1996. Although the question is not free from doubt, we conclude that neither the registration nor the notification provisions of the Act constitute "punishment" for purposes of the Ex Post Facto Clause, and that both sets of provisions may be imposed upon offenders convicted before the Act's effective date. We therefore affirm in part, reverse in part, and remand for further consideration of plaintiffs' remaining claims.

The seriousness of the harm that sex offenders' actions cause to society and the perception, supported by some data, that such offenders have a greater probability of recidivism than other offenders have recently combined to prompt the enactment of numerous laws across the country directed specifically toward persons convicted of crimes involving sexual conduct. Studies have shown that sex crimes are widespread, see, e.g., Brief of Amicus Curiae United States at 4 (citing Bureau of Justice Statistics, U.S. Dept. of Justice, Fact Sheet, National Crime Victim Survey Redesign (Oct. 30, 1994) (table) (312,000 rapes and attempted rapes and 173,000 other sexual assaults in 1993)), and that their impact on both the victim and society as a whole is devastating, see, e.g., id. at 5-6 (citing John Briere & Marsha Runtz, Childhood Sexual Abuse: Long-Term Sequelae and Implications for Psychological Assessment, 8 J. Interpersonal Violence 312, 324 (Sept.1993) (noting that molested children are likely to develop severe psychosocial problems) and Alphie Kohn, Shattered Innocence, Psychology Today, Feb. 1987, at 54, 58 (noting that sexually abused boys are more likely than non-abused boys to become sex offenders themselves, and that sexually abused girls are more likely than non-abused girls to have children who are abused)).

Some studies have also demonstrated that, as a group, convicted sex offenders are much more likely than other offenders to commit additional sex crimes. See, e.g., id. at 8-9 (citing studies reporting that rapists repeat their offenses at rates as high as 35%). Although other studies have reported that sex offenders as a category do not have a higher rate of recidivism than other categories of offenders, see generally Abril R. Bedarf, Comment, Examining Sex Offender Community Notification Laws, 83 Cal. L.Rev. 885, 893-98 (1995), the plaintiffs did not challenge the defendants' and the amici 's recidivism statistics in the District Court, which accepted them as true for the purpose of resolving the parties' cross-motions for summary judgment. Pataki II, 940 F. Supp. at 606. We therefore also accept the validity of these figures on this appeal.

New York's version of "Megan's Law," the SORA, was passed on July 25, 1995, and became effective on January 21, 1996. It requires individuals convicted of certain listed sex offenses to register with law enforcement officials, and it authorizes those officials, in some instances, to notify the public of, or provide the public with access to, the identity, whereabouts, and background of registrants. The legislature articulated two goals served by the SORA: (1) protecting members of the community, particularly their children, by notifying them of the presence of individuals in their midst who may present a danger, and (2) enhancing law enforcement authorities' ability to investigate and prosecute future sex crimes.

The Act applies to sex offenders incarcerated or on parole or probation on its effective date, as well as to those sentenced thereafter, thereby imposing its obligations on many persons whose crimes were committed prior to the effective date. The three plaintiffs in this case committed their sex crimes prior to this date, but are nonetheless subject to its requirements.3 The plaintiffs claim that the Act, by forcing them to register and by providing for notification to the community of their background and whereabouts, increases the amount of "punishment" that was, or could have been, imposed upon them at the time of their offense, thereby violating the Ex Post Facto Clause. To resolve this claim, we begin by examining the SORA in detail.

All convicted "sex offenders," as defined under the Act, are subject to its requirements. The category of sex offenders includes any person who is convicted of a "sex offense," N.Y. Correct. Law § 168-a(2), or a "sexually violent offense," id. § 168-a(3). See id. § 168-a(1). "Sex offenses" include, for example, rape in the second or third degree, sodomy in the second or third degree, sexual abuse in the second degree, and attempts. See id. § 168-a(2). "Sexually violent offenses" include, for example, rape in the first degree, sodomy in the first degree, sexual abuse in the first degree, and attempts. See id. § 168-a(3). These two categories, including attempts, encompass 36 offenses, some of which do not require proof of sexual contact, see, e.g., id. § 168-a(2)(a) (kidnapping or unlawful imprisonment of victim under age of seventeen), and seven of which are misdemeanors. First time offenders are included within the coverage of the Act.

Any person convicted of a sex offense or a sexually violent offense must register with the Division of Criminal Justice Services ("DCJS") as a sex offender within ten days of his discharge, release, or parole. See id. § 168-f(1).4 To register, sex offenders must provide such identifying information as name, date of birth, sex, race, height, weight, eye color, driver's license number, and home address. They must also provide a description of the offense of conviction, the date of conviction, and the sentence imposed, as well as a photograph and fingerprints. Id. §§ 168-b(1), 168-i. For each sex offender subject to the registration requirement, the DCJS must establish and maintain a file containing the required information. Id. § 168-b.

The duration and frequency of the registration obligations depend upon whether the sex offender is classified as a "sexually violent predator," as defined by sections 168-a(7) and 168-l(6)(c).5 An offender who is not considered a sexually violent predator must register on a yearly basis for ten years. Id. §§ 168-f(1), 168-h. On each anniversary of the offender's initial registration date during this ten year period, the offender must mail a form verifying his current address to the DCJS. Id. § 168-f(2). Additionally, the offender must register with the DCJS within ten days prior to any change of address. Id. § 168-f(4).

Offenders classified as sexually violent predators are subject to additional requirements. First, they must register for a minimum of ten years and are potentially subject to lifetime registration, unless the sentencing court determines, upon the offender's petition for relief, see id. § 168-o, that the offender "no longer suffers from a mental abnormality that would make him likely to engage in a predatory sexually violent offense." Id. § 168-h. Second, sexually violent predators must register quarterly, instead of annually, and must do so in person, rather than by mail. Id. § 168-f(3).

As noted, any offender required to register "may be relieved of any further duty to register upon the granting of a petition for relief by the sentencing court." Id. § 168-o. When the court receives such a petition, it is required to notify the Board of Examiners of Sex Offenders ("Board"), an entity created by the Act to administer and implement several of its provisions, see id. § 168-l, and to request an updated report from the Board pertaining to the petitioning offender, see id. § 168-l (7). The court may then grant or deny the relief sought.

All offenders required to register must do so or risk a criminal penalty. An offender who fails to register is guilty of a class A misdemeanor for the first offense, and is guilty of a class D felony for each subsequent offense. Id. § 168-t.

The Act provides for three levels of notification to law enforcement agencies and/or the public based on an assessment of the offender's risk of recidivism, termed "re-offense," and the degree of harm he potentially presents. Id. § 168-l (6). The basic principle is simple: the greater "the risk of a repeat offense ... and the threat posed to the public safety," id. § 168-l (5), the greater the extent of the notification, as to both who will receive the offender's registration information and the amount of information disseminated or made publicly accessible.

The five-member Board, composed of "experts in the field of the behavior and treatment of sex offenders" from the Division of Parole and the Department of Probation, id. § 168-l (1), is charged with the responsibility of developing guidelines and procedures to assess the risk of re-offense and the threat posed, id. § 168-l (5). The Act specifically lists some of the factors upon which the risk-level guidelines should be based: "criminal history factors indicative of high risk of repeat offense"; "conditions of release that minimize risk [of] re-offense, including ... whether the sex offender is under supervision [or] receiving counseling, therapy or treatment"; "physical conditions that minimize risk of re-offense, including ... advanced age or debilitating illness"; "whether psychological or psychiatric profiles indicate a risk of recidivism"; "the sex offender's response to treatment"; "recent behavior, including behavior while confined"; "recent threats or gestures against persons or expressions of intent to commit additional offenses"; and "review of any victim impact statement." Id.

In accordance with the Act's mandate, the Board has developed a set of "Risk Assessment Guidelines" for determining an offender's level of notification. See Sex Offender Registration Act: Risk Assessment Guidelines and Commentary (Jan.1996) ("Guidelines"). This detailed, point-based system--somewhat resembling the federal Sentencing Guidelines--assigns numerical values to fourteen risk factors, placed into four different categories (relating to the offender's current offense, criminal history, post-offense behavior, and planned release environment).6 A presumptive "risk level"7 is calculated for an offender by adding up the points assigned to the offender in each category. The Board "may not depart from the presumptive risk level unless it concludes that there exists an aggravating or mitigating factor of a kind, or to a degree, not otherwise adequately taken into account by the [G]uidelines." Id. at 4; cf. 18 U.S.C. § 3553(b) (departure standard for federal Sentencing Guidelines).

For all sex offenders sentenced or released from a state correctional or mental institution after the effective date of the Act, the original sentencing court has the responsibility, at the time of the offender's discharge, release, or parole, of determining the appropriate risk level. N.Y. Correct. Law §§ 168-d(3), 168-n(2).8 However, the court cannot make this determination until it has received from the Board a recommendation regarding the proper risk level for a particular offender. Id. §§ 168-l (6), 168-n(2). An offender who wishes to challenge the Board's recommended risk level may "appear and be heard," and, if necessary, have appointed counsel. Id. § 168-d(3); see id. § 168-n(3). Although the Act is silent on the degree of deference a sentencing court must accord to the Board's recommendation during such a hearing, one New York court has ruled that the Board's risk-level recommendations should be accepted by the sentencing court unless arbitrary and capricious. See People v. Ross, 169 Misc. 2d 308, 312, 646 N.Y.S.2d 249, 252 (Sup.Ct.N.Y.Co.1996). Ross also ruled that an offender challenging the Board's assessment before the sentencing court bears the burden of proving that the Board has erroneously applied the Guidelines. Id., 646 N.Y.S.2d at 252.

If an offender's risk of re-offense is "low," he is assigned a "level-one" designation. N.Y. Correct. L. § 168-l (6)(a). For level-one offenders, the offender's registration information is supplied, without request, only to the law enforcement agency having jurisdiction over the offender. Id. However, the Act also permits limited information concerning level-one offenders to be obtained through a special "900" telephone number; for a small fee, anyone can call this number to inquire whether a particular individual is listed in the central registry maintained by the DCJS. Id. § 168-p. Callers cannot obtain any information through this service unless they first provide specified information that reasonably identifies the offender, such as, an exact street address, a birthdate, a driver's license number, along with additional information, such as a social security number or a physical description.9 Id. § 168-p(1). When the 900 number is called, a recorded message advises the caller that the caller's telephone number will be recorded by the DCJS, that a fee will be assessed for the call, that the caller must provide identification and a current address before the request will be processed, that the 900 number is not a crime "hotline," and that it is illegal to use any information obtained through the 900 number to commit a crime against, or to engage in illegal discrimination or harassment against, any listed person. Id. § 168-p(2). If a caller seeks information on an identified person who has been designated as a level-one offender, the defendants assert, Brief for Appellants at 24, and the plaintiffs do not dispute, that the caller will be informed only that the offender is listed in the central registry and that the offender's risk level is level one. The Act authorizes the state attorney general, a local district attorney, or "any person aggrieved by the misuse of the number" to bring a civil action for injunctive relief against any person or persons "engaged in a pattern or practice of misuse of the '900' number." N.Y. Correct. Law § 168-p(3). The Act additionally provides that persons misusing information provided through this service will be subject to a fine. Id.

If an offender's risk of re-offense is "moderate," he is assigned a "level-two" designation. Id. § 168-l (6)(b). Registration information for this category of offenders can be disseminated to the public in two ways. First, callers can obtain such information through the 900 number. Second, the offender's registration information will be forwarded by the DCJS to the law enforcement authority having jurisdiction over the offender; that authority may then, at its discretion, disseminate relevant portions of this information to "any entity with vulnerable populations related to the nature of the offense committed by such sex offender." Id. The distributed information includes the offender's approximate address based on the zip code, a photograph, background information regarding the offense of conviction and the offender's modus operandi, and a description of any special conditions of parole or probation. Although the term "entity with vulnerable populations" is not defined in the Act, the parties assume that it refers to organizations such as child-care centers, schools, or women's shelters. After receiving registration information on an offender from local law enforcement officials, such organizations may then "disclose or further disseminate such information at their discretion." Id.

If an offender's risk of re-offense is "high," he is "deemed a sexually violent predator" and is assigned a "level-three" designation. Id. § 168-l (6)(c). For this category of offenders, registration information is made available to the public in three ways. First, callers can obtain such information through the 900 number. Second, the DCJS will forward the offender's registration information to the law enforcement authority having jurisdiction over the offender; that authority may then disseminate this information--including the offender's exact street address--to any entity with vulnerable populations. Id. As noted above, such an entity may then further disseminate the information it receives at its discretion. Finally, registration information on all level-three offenders will be included in a "subdirectory of sexually violent predators," which contains, among other things, the exact street address of the offender, a photograph, his name and physical description, general background information relating to the crime of conviction, the offender's modus operandi, the type of victim targeted, and any special conditions of the offender's parole or probation. Id. § 168-q(1); see id. § 168-l (6)(c). The subdirectory will categorize sexually violent predators by county and zip code, and copies of the subdirectory will be distributed to local police departments. Anyone who "in writing expresses a purpose" will be allowed to look through the subdirectory. Id. § 168-q(1). The Act specifically provides that anyone who uses information obtained through the subdirectory "in violation of the law" will be subject to a fine and that unauthorized removal or duplication of the subdirectory is prohibited. Id. § 168-q(2).

To summarize, (1) registration information concerning level-one offenders is provided to law enforcement officials and is potentially available in a limited fashion to the public through the 900 phone number; (2) registration information concerning level-two offenders is available to the public through the above route as well as through possible dissemination to organizations with vulnerable populations at the discretion of law enforcement officials, and to other members of the community at the discretion of organizations with vulnerable populations; and (3) registration information concerning level-three offenders is available to the public through all of the above routes as well as through the sexually violent predator subdirectory found at local police stations.

In addition to the procedural safeguards to prevent misuse or abuse of registration information already mentioned, see, e.g., id. § 168-p(3) (regarding 900 number) and id. § 168-q(2) (regarding subdirectory), the Act generally provides that "[t]he unauthorized release of any information required by this article shall be a class B misdemeanor." Id. § 168-u. Moreover, as noted previously, all offenders required to register under the Act--and thus subject to some form of community notification or public access--can petition the original sentencing court to be relieved of the duty to register, and thereby freed from being the subject of further notification pursuant to the Act. Id. § 168-o.

Plaintiffs filed their complaint in March 1996, contending that the registration and notification provisions of the SORA violated various statutory and constitutional safeguards, including the Ex Post Facto Clause. Relying solely on the ex post facto argument, the District Court first issued a preliminary injunction, see Doe v. Pataki, 919 F. Supp. 691 (S.D.N.Y.1996) (Doe v. Pataki I ), and then issued a permanent injunction on the parties' cross-motions for summary judgment, see Doe v. Pataki II, enjoining the defendants from enforcing the public notification provisions of the Act against persons who committed their offenses before January 21, 1996. In the latter opinion, the Court also concluded that retrospective application of the Act's registration requirements did not violate the ex post facto prohibition. See id., 940 F.Supp. at 629-30. The Court did not reach the plaintiffs' remaining statutory and constitutional arguments in either opinion.

The parties and the District Court agreed that the central question in this case is whether, as the plaintiffs contend, the Act increases the "punishment" for sex offenses, or whether, as the defendants contend, the Act merely regulates by protecting the public. See id. at 612. In a thoughtful decision, Judge Chin canvassed numerous decisions of courts around the country presenting issues relevant to the resolution of this question, including state and federal court decisions discussing whether retroactive application of sex offender registration or notification laws would violate the Ex Post Facto Clause,10 and Supreme Court decisions discussing the more general question of whether a sanction or a proceeding was sufficiently "punitive" or "criminal" to invoke the protections of the Ex Post Facto Clause, the Double Jeopardy Clause, the Excessive Fines Clause, the Cruel and Unusual Punishments Clause, or the procedural safeguards of the Fifth and Sixth Amendments.11 Endeavoring to extract unifying principles from these decisions, the District Court concluded that to resolve the question of whether registration or notification constituted punishment for ex post facto purposes, it would analyze all pertinent circumstances by grouping them into four areas: (1) legislative intent, (2) statutory design, (3) historical analogies, and (4) effect of the statute. Doe v. Pataki II, 940 F. Supp. at 620. The Court cautioned that these factors should not be considered as a " 'rigid series of hurdles,' " but must rather be evaluated in a " 'less structured fashion.' " Id. (quoting W.P. v. Poritz, 931 F. Supp. 1199, 1209 (D.N.J.1996)).

Applying its four factor test to the notification provisions of the Act, the Court concluded that they increased the plaintiff's punishment in violation of the Ex Post Facto Clause. First, although acknowledging that the legislature's stated intent and the Act's apparent purposes were to protect the public from potentially dangerous sex offenders and to facilitate law enforcement efforts concerning future sex crimes, the Court concluded that the state legislature also intended to punish sex offenders, citing as support several statements made by legislators during floor debates purportedly evidencing their hostility toward, and desire to punish, sex offenders. Id. at 604-05. Second, the Court found that the Act's design "contains classic indicia of a punitive scheme": the notification provisions are triggered by the commission of a crime, provide for the sentencing judge to determine the level of notification, permit the judge to consider victim-impact statements, and broadly cover 36 offenses and first-time offenders. Id. at 605. Third, the Court ruled that historical considerations revealed the punitive nature of public notification, which Judge Chin called the "modern-day equivalent of branding and banishment." Id. Finally, the Court concluded that "[p]erhaps more clearly than anything else, the effects of community notification show that these provisions are punitive." Id. at 626. Notification, he observed, results in an affirmative disability with nearly " 'unlimited' " consequences, interferes with offenders' ability to rehabilitate, and serves the traditional goals of punishment--deterrence, retribution, and incapacitation. Id. at 626-29.

Applying the same test to the registration requirements, the Court concluded that they did not impose additional punishment on previously convicted sex offenders. Registration, Judge Chin stated, "plainly serves the valid regulatory goals of the Act to protect the public and aid law enforcement," id. at 629; "can be accomplished privately and [ ] does not suffer from the same excesses in design that exist with respect to notification," id. at 630; "has not historically been considered punishment," id.; and, "[m]ost significantly, [ ] does not impose a substantial affirmative disability or restraint" upon offenders, id.

Both parties appeal from the District Court's judgment. The defendants challenge the invalidation of retroactive application of the SORA's notification provisions, and the plaintiffs challenge upholding retroactive application of the Act's registration provisions.

Although the "presumption against retroactive legislation is deeply rooted in our jurisprudence," Landgraf v. USI Film Products, 511 U.S. 244, 265, 114 S. Ct. 1483, 1497, 128 L. Ed. 2d 229 (1994), the Ex Post Facto Clause of the Constitution "applies only to penal statutes which disadvantage the offender affected by them," Collins v. Youngblood, 497 U.S. 37, 41, 110 S. Ct. 2715, 2718, 111 L. Ed. 2d 30 (1990) (emphasis added); see U.S. Const. Art. I, § 10, cl. 1 ("No State shall ... pass any ... ex post facto Law....").12 "Ex post facto" is a term of art applicable only to "punishment"--legislative action that retroactively "punishes as a crime an act previously committed, which was innocent when done," "makes more burdensome the punishment for a crime, after its commission," or "deprives one charged with crime of any defense available according to law at the time when the act was committed." Beazell v. Ohio, 269 U.S. 167, 169-70, 46 S. Ct. 68, 68-69, 70 L. Ed. 216 (1925); see Collins, 497 U.S. at 43, 110 S.Ct. at 2719 ("The Beazell formulation is faithful to our best knowledge of the original understanding of the Ex Post Facto Clause."). The question before us concerns the second Beazell category: whether notification and registration under the SORA "increase the punishment for criminal acts." Id.; see California Department of Corrections v. Morales, 514 U.S. 499, 506 n. 3, 115 S. Ct. 1597, 1602 n. 3, 131 L. Ed. 2d 588 (1995) ("[T]he focus of the ex post facto inquiry is not on whether a legislative change produces some ambiguous sort of 'disadvantage,'... but on whether any such change ... increases the penalty by which a crime is punishable.").

I. "Punishment" in the Ex Post Facto Analysis

"Punishment"--along with its many cognates and synonyms, for instance, "penal," "punitive," "penalty," and "criminal"--is an imprecise concept with meanings that vary depending on the purpose for which the concept is defined. A standard dictionary definition states that to "punish" is to "impose a penalty (as of pain, suffering, shame, strict restraint, or loss) upon for some fault, offense, or violation." Webster's Third New International Dictionary 1843 (1993). Somewhat less circular but perhaps no more illuminating definitions can be found in the legal literature, see, e.g., Note, Toward a Constitutional Definition of Punishment, 80 Colum. L.Rev. 1667, 1678-81 (1980) ("Punishment is the deprivation of legal rights in response to a prior offense for the purpose of deterrence and social condemnation."), and the philosophical literature, see, e.g., Joel Feinberg, The Expressive Function of Punishment, reprinted in The Philosophy of Law 636 (4th ed.1991) ("Punishment is a conventional device for the expression of attitudes of resentment and indignation, and of judgments of disapproval and reprobation, either on the part of the punishing authority himself or of those in whose name the punishment is imposed.") (quotations omitted).

For the precise purpose of determining whether a retroactively imposed burden constitutes "punishment" within the meaning of the Ex Post Facto Clause, the Supreme Court has instructed that we should focus our attention on the twin purposes to be served by this constitutional prohibition. First, the Clause ensures that legislative acts "give fair warning of their effect and permit individuals to rely on their meaning until explicitly changed." Weaver v. Graham, 450 U.S. 24, 28-29, 101 S. Ct. 960, 964, 67 L. Ed. 2d 17 (1981). Second, the Clause erects a barrier to legislative abuses in the form of arbitrary or vindictive legislation directed against disfavored groups. See Miller v. Florida, 482 U.S. 423, 429, 107 S. Ct. 2446, 2451, 96 L. Ed. 2d 351 (1987). The legislature's "responsivity to political pressures poses a risk that it may be tempted to use retroactive legislation as a means of retribution against unpopular groups or individuals." Landgraf, 511 U.S. at 266, 114 S.Ct. at 1497; see Cummings v. Missouri, 71 U.S. (4 Wall.) 277, 322, 18 L. Ed. 356 (1866) (Ex Post Facto Clause adopted because " 'the framers of the Constitution viewed with some apprehension the violent acts which might grow out of the feelings of the moment' ") (quoting Fletcher v. Peck, 10 U.S. (6 Cranch) 87, 137, 3 L. Ed. 162 (1810)).

Since at least 1898, the Supreme Court has focused upon the intent underlying the enactment of, or the end served by, the challenged sanction as the touchstone of the ex post facto analysis. In Hawker v. New York, 170 U.S. 189, 18 S. Ct. 573, 42 L. Ed. 1002 (1898), a physician, convicted of the then-felony crime of abortion in 1878, contended that a law, passed by the legislature in 1893, that prohibited persons who had been convicted of a felony from practicing medicine violated the Ex Post Facto Clause. The Court rejected this challenge since "[t]he state [was] not seeking to further punish a criminal, but only to protect its citizens from physicians of bad character." Id. at 196, 18 S.Ct. at 576. Although, as the Court acknowledged, the "form in which this legislation is cast suggests the idea of the imposition of an additional punishment for past offenses," id., the Court concluded that the law did not intrude upon the ex post facto prohibition because it is more properly "regarded as ... prescribing the qualifications for the duties to be discharged and the position to be filled, and naming what is deemed to be ... appropriate evidence of such qualifications," id. at 200, 18 S.Ct. at 577.

The approach in Hawker, focusing upon the intended or apparent purpose served by the retroactively imposed disability, has been repeatedly affirmed since 1898. As Chief Justice Warren summarized for the four-Justice plurality in Trop v. Dulles, 356 U.S. 86, 96, 78 S. Ct. 590, 595, 2 L. Ed. 2d 630 (1958), "[i]n deciding whether or not a law is penal [for the purpose of ex post facto analysis], this Court has generally based its determination upon the purpose of the statute." Thus, if a legislative burden is imposed "for the purposes of punishment--that is, to reprimand the wrongdoer, to deter others, etc., it has been considered penal." Id. However, if a disability is imposed "not to punish, but to accomplish some other legitimate governmental purpose," then it has been considered "nonpenal." Id.

The Court continued this approach in DeVeau v. Braisted, 363 U.S. 144, 80 S. Ct. 1146, 4 L. Ed. 2d 1109 (1960), which upheld a statute that prohibited unions from soliciting or collecting dues from workers on the New York waterfront if any officer or agent of the union had been previously convicted of a felony. The law thus effectively barred convicted felons from working for the unions. Writing for the four-Justice plurality, Justice Frankfurter reaffirmed the intent- and purpose-centered approach of Hawker and Trop:

The mark of an ex post facto law is the imposition of what can fairly be designated punishment for past acts. The question in each case where unpleasant consequences are brought to bear upon an individual for prior conduct, is whether the legislative aim was to punish that individual for past activity, or whether the restriction of the individual comes about as a relevant incident to a regulation of a present situation, such as the proper qualifications for a profession.

Id. at 160, 80 S.Ct. at 1155 (emphasis added). Because the proof was "overwhelming that New York sought not to punish ex-felons, but to devise what was felt to be a much-needed scheme of regulation of the waterfront," the law was upheld. Id.

In more recent decisions, the Court has articulated a two-part approach for determining whether legislation should be classified as punitive. In United States v. Ward, 448 U.S. 242, 100 S. Ct. 2636, 65 L. Ed. 2d 742 (1980), the Court considered whether a "civil penalty" imposed for discharging oil into navigable waters should be considered "penal" for the purpose of triggering the safeguards of the Fifth and Sixth Amendments. In concluding that the penalty was not "penal," the Court said:

Our inquiry in this regard has traditionally proceeded on two levels. First, we have set out to determine whether Congress, in establishing the penalizing mechanism, indicated either expressly or impliedly a preference for one label or the other. Second, where Congress has indicated an intention to establish a civil penalty, we have inquired further whether the statutory scheme was so punitive either in purpose or effect as to negate that intention. In regard to this latter inquiry, we have noted that "only the clearest proof could suffice to establish the unconstitutionality of a statute on such a ground."

Id. at 248-49, 100 S.Ct. at 2641 (citations omitted) (quoting Flemming v. Nestor, 363 U.S. 603, 617, 80 S. Ct. 1367, 1376, 4 L. Ed. 2d 1435 (1960)). Thus, in accordance with Hawker, Trop, and DeVeau, the Court in Ward considered the threshold and primary issue in the ex post facto analysis to be the legislature's intent in enacting the challenged sanction. Ward, however, made explicit what was only implicit in those earlier decisions--that even a sanction intended to serve nonpunitive, regulatory goals can be "transfor[med]" into a criminal penalty if the sanction is "so punitive either in purpose or effect." Id. at 249, 100 S.Ct. at 2641 (quotations omitted). See Montana v. Kurth Ranch, 511 U.S. 767, 777, 114 S. Ct. 1937, 1945, 128 L. Ed. 2d 767 (1994) ("[T]he legislature's description of a statute as civil does not foreclose the possibility that it has a punitive character.")

Subsequent decisions in the double jeopardy and the ex post facto contexts have adhered to the two stage inquiry set out by Ward. In United States v. Ursery, --- U.S. ----, 116 S. Ct. 2135, 135 L. Ed. 2d 549 (1996), the Court considered the question of whether the purportedly "civil" forfeiture of property used to facilitate marijuana violations constituted a second punishment within the meaning of the Double Jeopardy Clause. The Court explained its approach as follows:

First, we ask whether Congress intended proceedings under [the challenged forfeiture statutes] to be criminal or civil. Second, we turn to consider whether the proceedings are so punitive in fact as to "persuade us that the forfeiture proceeding[s] may not legitimately be viewed as civil in nature," despite Congress' intent.

Id. at ----, 116 S.Ct. at 2147 (emphasis added) (citations omitted) (quoting United States v. One Assortment of 89 Firearms, 465 U.S. 354, 366, 104 S. Ct. 1099, 1107, 79 L. Ed. 2d 361 (1984)). Echoing Ward, the Court emphasized that, at the second stage, the party challenging the law must show by "the clearest proof" that the sanctions imposed "are so punitive in form and effect as to render them criminal despite Congress' intent to the contrary." Id. at ----, 116 S.Ct. at 2148 (emphasis added).

This Circuit has recently ruled that the two-part inquiry used by Ursery in the double jeopardy context is equally appropriate for determining whether a sanction constitutes punishment for purposes of ex post facto analysis. See United States v. Certain Funds, 96 F.3d 20, 26 (2d Cir.1996). That equivalence has been implicitly approved by the Supreme Court. In the recent decision of Kansas v. Hendricks, --- U.S. ----, 117 S. Ct. 2072, 138 L. Ed. 2d 501 (1997), the Court ruled that a statute requiring the involuntary commitment of "mentally abnormal" persons previously convicted of sex crimes did not impose punishment in violation of either the Double Jeopardy Clause or the Ex Post Facto Clause. The Court undertook only a single analysis to answer both constitutional challenges:

The categorization of a particular proceeding as civil or criminal "is first of all a question of statutory construction." We must initially ascertain whether the legislature meant the statute to establish "civil" proceedings. If so, we ordinarily defer to the legislature's stated intent....

Although we recognize that a "civil label is not always dispositive," we will reject the legislature's manifest intent only where a party challenging the statute provides "the clearest proof" that "the statutory scheme [is] so punitive either in purpose or effect as to negate [the State's] intention" to deem it "civil."

Id. at ----, 117 S.Ct. at 2082 (citations omitted). Finding that (1) "[n]othing on the face of the statute suggests that the legislature sought to create anything other than a civil commitment scheme designed to protect the public from harm," id., and that (2) the challenger "has failed to satisfy th[e] heavy burden" of demonstrating that the statutory scheme was so punitive in form and effect as to render it punitive or criminal despite the legislature's apparent nonpunitive intent, id., the Court concluded that involuntary confinement pursuant to Kansas's civil commitment statute does not amount to punishment. Id. at ---, 117 S.Ct. at 2086.

The Supreme Court has not spelled out the precise nature of the second-stage inquiry, in which the challenger must overcome an initial presumption of nonpunitiveness by demonstrating that the sanction, in form and operation, has an essentially punitive "character." Cf. Nixon v. Administrator of General Services, 433 U.S. 425, 475-76, 97 S. Ct. 2777, 2806-07, 53 L. Ed. 2d 867 (1977) (describing similar inquiry in bill of attainder context as "analyzing whether the law under challenge, viewed in terms of the type and severity of burdens imposed, reasonably can be said to further nonpunitive legislative purposes"). The Court suggested that in determining whether a purportedly civil sanction has been transformed into a criminal or punitive one, the list of considerations in Kennedy v. Mendoza-Martinez, 372 U.S. 144, 168-69, 83 S. Ct. 554, 567-68, 9 L. Ed. 2d 644 (1963),13 though "certainly neither exhaustive nor dispositive, has proved helpful in our own consideration of similar questions and provides some guidance in the present case," Ward, 448 U.S. at 249, 100 S. Ct. at 2641-42 (citation omitted); see 89 Firearms, 465 U.S. at 365, 104 S.Ct. at 1106.

An examination of the case law confirms that whether a sanction intended as regulatory or nonpunitive is "so punitive in fact" as to violate the ex post facto prohibition is a highly context specific matter. See Nestor, 363 U.S. at 616, 80 S.Ct. at 1375 ("[E]ach case has turned on its own highly particularized context."). Sometimes one factor will be considered nearly dispositive of punitiveness "in fact," while sometimes another factor will be crucial to a finding of nonpunitiveness. Cf. 89 Firearms, 465 U.S. at 365-66, 104 S.Ct. at 1106-07 (discussing only one commonly considered factor--"whether or not the proscribed behavior is already a crime"--as arguably supporting party challenging law); Ward, 448 U.S. at 249-50, 100 S.Ct. at 2641-42 (same).14 Other recent cases have cited some of the factors listed in Mendoza-Martinez to support a characterization of the challenged burden as civil, and either ignored or discounted the presence of other factors generally considered inconsistent with such a characterization. See, e.g., Ursery, --- U.S. at ---- - ----, 116 S. Ct. at 2148-49 (supporting conclusion that forfeiture is civil by noting that it is traditionally considered as civil, and that scienter is not required, although acknowledging that forfeiture serves purpose of deterrence and is "tied to criminal activity"); Hendricks, --- U.S. at ---- - ----, 117 S. Ct. at 2082-83 (supporting conclusion that commitment is civil by noting that criminal conviction is not prerequisite for commitment, and that no finding of scienter is required, although acknowledging that statute uses procedural safeguards of criminal proceedings and that civil commitment involves "affirmative restraint" and incapacitation of the offender).

With these principles in mind, we proceed to examine first the notification and then the registration provisions of the SORA.

All persons convicted of sex offenses listed under the Act are subject to its community notification provisions, in one form or another. As described earlier, the DCJS and local law enforcement officials are required under the SORA either to provide the public with access to, or to affirmatively disseminate to the public, the information contained in the sex offender's registration file, including the offender's identity, and, depending on the offender's risk classification, his description, address, crime of conviction, modus operandi, and conditions of release. To determine whether such notification is punishment, we will examine the notification provisions under the two part framework discussed above.

There is ample evidence that the New York legislature intended the SORA to further nonpunitive goals. The Act's preamble contains the following statement of the legislature's findings and intent:

The legislature finds that the danger of recidivism posed by sex offenders, especially those sexually violent offenders who commit predatory acts characterized by repetitive and compulsive behavior, and that the protection of the public from these offenders is of paramount concern or interest to government. The legislature further finds that law enforcement agencies' efforts to protect their communities, conduct investigations and quickly apprehend sex offenders are impaired by the lack of information about sex offenders who live within their jurisdiction and that the lack of information shared with the public may result in the failure of the criminal justice system to identify, investigate, apprehend and prosecute sex offenders.

....

[T]he legislature finds that releasing information about sex offenders to law enforcement agencies and, under certain circumstances, providing access to limited information about certain sex offenders to the general public, will further the primary government interest of protecting vulnerable populations and in some instance the public, from potential harm.

N.Y. Correct. Law § 168 (Historical and Statutory Notes).15

The legislative history of the Act supports the preamble's characterization of the twin purposes served by the SORA--protecting communities by notifying them of the presence of individuals who may present a danger and enhancing law enforcement authorities' ability to fight sex crimes. As Assemblyman Daniel L. Feldman stated in the bill jacket for the SORA:

Community notification promotes a state interest in advancing the protection of the public.... The public is notified so that they can be [the] "eyes and ears" [of law enforcement agencies]. The public can notify the appropriate authorities if the sex offender violates any condition of the offender's parole or probation which would enable the authorities to intervene when a releasee's behavior begins to pose a threat to community safety. This is extremely important given the high recidivism rate many of these offenders have and the lack of scientific evidence ... that proves treatment programs reduce sex offender recidivism. A notified community may prevent crimes with greater attention and caution.

Bill Jacket for L.1995, ch. 192, at 3. A memorandum issued by the SORA's sponsor in the State Senate, Senator Dean G. Skelos, and his counterpart in the State Assembly, Assemblyman Feldman, further confirms the regulatory aims of the Act. See Introducer's Memorandum in Support of Sen. Bill S. 11-B, at 2-3 (1996) ("Protecting the public, especially children, from sex offenders is a primary governmental interest.... The New York statutory plan provided under this bill is predominantly regulatory.... This balanced approach ... affords concerned citizens with the ability to access information which may ultimately provide significant protection to their family.").

Similarly, the decision to make the SORA retroactive was guided by the legislature's desire to protect the public from potentially dangerous persons. As Assemblyman Feldman observed, without retroactive reach, the Act would "leave[ ] the majority of sexual offenders cloaked in anonymity," despite the fact that offenders who have already committed their crimes are no less dangerous than those who will commit crimes after the enactment of the Act. Bill Jacket, supra, at 5. The views of a law's sponsors, though not conclusive, are entitled to considerable weight. See, e.g., North Haven Board of Education v. Bell, 456 U.S. 512, 526-27, 102 S. Ct. 1912, 1920-21, 72 L. Ed. 2d 299 (1982); see also United States v. Jackson, 805 F.2d 457, 462-63 (2d Cir.1986).

Review of the transcripts of the Senate and Assembly floor debates largely confirms our conclusion that the state legislature enacted the SORA in order to "provide parents and communities with information as to potential risks from convicted sex offenders." See Ross, 169 Misc. 2d at 311, 646 N.Y.S.2d at 251. Although we agree with the District Court that the comments of some legislators reveal their animosity toward, and even a desire to punish, sex offenders, see Doe v. Pataki II, 940 F. Supp. at 621-22, we decline to rely on these isolated statements to characterize the legislature's intent as punitive. We have observed that floor debates are of "particularly limited assistance in resolving highly controversial issues of [legislative] intent." Butts v. City of New York Department of Housing Preservation and Development, 990 F.2d 1397, 1405 (2d Cir.1993). Because "any member can make a floor speech, there is scant utility in totaling up the number of speeches on each side of an issue and attempting to divine [legislative] intent from the quantity of [minute] pages generated on either side." Id.; see also Murphy v. Empire of America, FSA, 746 F.2d 931, 935 (2d Cir.1984) (isolated remarks, particularly when unclear or conflicting, are entitled to little or no weight unless no official legislative history exists).

More specifically, courts have cautioned that in determining whether a sanction was imposed for punitive or nonpunitive reasons, "[j]udicial inquiries into Congressional motives are at best a hazardous matter, and when that inquiry seeks to go behind objective manifestations it becomes a dubious affair indeed." Nestor, 363 U.S. at 617, 80 S.Ct. at 1376; see Bae v. Shalala, 44 F.3d 489, 494 (7th Cir.1995) ("[R]eliance on statements made by politicians in their efforts to persuade colleagues to enact a law is a wholly unreliable method for determining the nature of a sanction."); Wiley v. Bowen, 824 F.2d 1120, 1122 (D.C.Cir.1987) ("[W]e cannot lightly attribute to Congress as a whole the impermissible motives of a few of its members."); cf. Selective Service System v. Minnesota Public Interest Research Group, 468 U.S. 841, 855 n. 15, 104 S. Ct. 3348, 3357 n. 15, 82 L. Ed. 2d 632 (1984) (bill of attainder) (noting that "several isolated statements [by legislators] expressing understandable indignation over the decision of some nonregistrants to show their defiance of the law" were insufficient to establish punitive intent). We therefore decline to "construe the subjective intent expressed by one or more legislators to reflect the objective intent" of the legislature, see Bae, 44 F.3d at 494, and we conclude that the plaintiffs have failed to produced the "unmistakable evidence of punitive intent" required to demonstrate punitive motivation, see Nestor, 363 U.S. at 619, 80 S.Ct. at 1375.

Moreover, the Act's text and core structural features reasonably bear out its stated nonpunitive goals of protecting the public and facilitating future law enforcement efforts. Although the Act is codified in the "Corrections Law" volume of the New York statutes, nothing in its text suggests that the legislature sought to punish sex offenders for their past offenses rather than to prevent any future harms that they might cause, and several of its features manifest a nonpunitive purpose. We examine these features in detail.

1. Extent of notification. First and foremost, the extent of notification is carefully calibrated to, and depends solely upon, the offender's perceived risk of re-offense: the greater the likelihood of re-offense, the broader and more detailed the notification to the public. The criteria set out in the Act for determining an offender's risk level fully bear out this prospective, regulatory aim. See N.Y. Correct. Law § 168-l (5) (basing assessment of risk level on, among other things, "criminal history factors indicative of high risk of repeat offense," "conditions of release that minimize risk [of] re-offense," "physical conditions that minimize risk of re-offense," and "whether psychological or psychiatric profiles indicate a risk of recidivism"). Moreover, although the Act primarily assigns the ultimate responsibility for determining the offender's risk level to the sentencing court, the court must generally abide by the recommendations of the Board, which comprises "experts in the field of the behavior and treatment of sex offenders" from the parole and probation departments. Id. §§ 168-l (1), 168-n(2); see Ross, 169 Misc. 2d at 312, 646 N.Y.S.2d at 252. Additionally, the timing of the risk-level determination confirms the forward looking nature of the SORA's notification provisions. The assessment is not made at the time of the original sentencing, but immediately prior to the offender's scheduled release or discharge. See N.Y. Correct. Law § 168-n(2).

2. Control of notification. Second, the extent of notification is carefully controlled. For offenders posing a "low" risk of re-offense, the sole method of public access to registration information is through the special 900 phone number. See id. §§ 168-l (6)(a), 168-p. Even for offenders posing a "moderate" risk of re-offense, the scope of notification is limited. Law enforcement officials are permitted to disseminate the offender's registration only to "entit[ies] with vulnerable populations," and may not reveal the offender's exact address. See id. § 168-l (6)(b). Moreover, even for those "sexually violent predators" deemed to pose a "high" risk of re-offense, active notification to the community is not required. See id. § 168-l (6)(c). Although the public will be allowed access to the offender's registration information through the subdirectory, provided that they "express a purpose" for doing so, id. § 168-q, the only affirmative dissemination that can be conducted by the state is to entities with vulnerable populations, and not to neighbors, employers, landlords, or news agencies.

3. Protection against misuse of information. Third, the notification provisions contain many safeguards to prevent the misuse of registration information. For instance, callers to the 900 number must provide their name and phone number before they will be given any information, and they are warned that they cannot use the information obtained through this service to engage in illegal discrimination and harassment against a listed offender. Id. § 168-p(2). Procedural safeguards also protect against the misuse of information obtained through the subdirectory. See id. § 168-q(2) (imposing fines on unauthorized use of subdirectory information and authorizing attorney general, district attorney, or "any person aggrieved" to bring injunctive action to prevent misuse of subdirectory). Additionally, the Act provides that "[t]he unauthorized release of any information required by this article shall be a class B misdemeanor," id. § 168-u, and that offenders subject to its registration and notification requirements may petition to be relieved from these burdens, see id. § 168-o.

In sum, the SORA's text and structure convincingly support the legislature's stated regulatory and prospective intent. We therefore proceed to examine whether the plaintiffs have demonstrated by "the clearest proof" that the burdens accompanying public notification are nonetheless "so punitive in form and effect" as to negate the legislature's nonpunitive intent. See Ursery, --- U.S. at ----, 116 S. Ct. at 2148.

B. Is Notification Punishment "in Fact"?

The plaintiffs point to several features of the notification requirement that they contend reveal that, despite its avowed purpose, it is punitive in fact. They cite its damaging effects on the lives of offenders, its close ties to criminal activity, its excessiveness, its advancement of traditional criminal law goals, and its similarity to historical punitive measures. We discuss each in turn.

1. Effects of notification. The plaintiffs rely primarily on what they regard as the Act's penal effects. The District Court agreed with the plaintiffs and concluded that "[p]erhaps more clearly than anything else, the effects of community notification show that these provisions are punitive." Doe v. Pataki II, 940 F. Supp. at 626. The parties have stipulated to a number of anecdotes concerning the effects that notification has had upon registered offenders in New York and in three other states with similar notification laws--New Jersey, Washington, and California. See Stipulation Concerning Classification of Offenders pp 46-86 (Aug. 30, 1996). The stipulation describes numerous instances in which sex offenders have suffered harm in the aftermath of notification--ranging from public shunning, picketing, press vigils, ostracism, loss of employment, and eviction, to threats of violence, physical attacks, and arson. Relying in part on these episodes, the District Court, quoting Roe, 938 F. Supp. at 1091, concluded that "the consequences of [community notification] are unlimited" and that the stigma created by the Act "pervades into every aspect of an offender's life." Doe v. Pataki II, 940 F. Supp. at 627 (quotations omitted). Although we do not doubt that the Act has had unfortunate consequences for many subject to its operation, we do not agree that these detrimental consequences suffice to transform the regulatory measure of community notification into punishment.

First, the Supreme Court has noted that "whether a sanction constitutes punishment is not determined from the defendant's perspective, as even remedial sanctions carry the 'sting of punishment.' " Kurth Ranch, 511 U.S. at 777 n. 14, 114 S. Ct. at 1945 n. 14 (quoting Halper, 490 U.S. at 447 n. 7, 109 S. Ct. at 1901 n. 7). Generally, a statutory scheme that serves a regulatory purpose "is not punishment even though it may bear harshly on one affected." Nestor, 363 U.S. at 614, 80 S.Ct. at 1374. Numerous statutes imposing exceedingly harsh disabilities have been upheld against ex post facto challenges. See Hendricks, --- U.S. at ----, 117 S. Ct. at 2086 (commitment of sex offenders with "mental abnormality"); Nestor, 363 U.S. at 612-21, 80 S.Ct. at 1373-78 (termination of vested old age social security benefits of eligible persons deported for participating in communist activities); DeVeau, 363 U.S. at 160, 80 S. Ct. at 1155 (prohibition of felons from working for waterfront unions); Galvan v. Press, 347 U.S. 522, 531, 74 S. Ct. 737, 743, 98 L. Ed. 911 (1954) (deportation for prior membership in the Communist Party); Hawker, 170 U.S. at 196, 18 S. Ct. at 576 (prohibition of physicians, convicted of felony, from practicing medicine).

Courts of Appeals have similarly rejected ex post facto challenges to legislation imposing harsh consequences. See, e.g., Rise v. Oregon, 59 F.3d 1556 (9th Cir.1995) (requirement that previously convicted murderers and felony sex offenders submit blood for DNA data bank); Bae, 44 F.3d 489 (7th Cir.1995) (prohibition of felons from participating in FDA drug approval process); Manocchio v. Kusserow, 961 F.2d 1539 (11th Cir.1992) (prohibition of physicians, convicted of filing false Medicare claims, from participation in program for minimum of five years); Wiley, 824 F.2d 1120 (D.C.Cir.1987) (termination of old age social security benefits for incarcerated felons); Postma v. International Brotherhood of Teamsters, 337 F.2d 609, 611 (2d Cir.1964) (prohibition of convicted extortionists from holding union office). Thus, consequences as drastic as deportation, deprivation of one's livelihood, and termination of financial support have not been considered sufficient to transform an avowedly regulatory measure into a punitive one.

Second, although some "effects," for instance, prolonged incarceration in a correctional facility, may in and of themselves be sufficient to render a sanction punitive for purposes of the ex post facto inquiry regardless of the legislature's intent,16 the effects fairly attributable to community notification do not fall into this extremely narrow category. We focus on the attributable effects because we disagree with the assumption, apparently made by the plaintiffs, that all of the unfortunate incidents that have occurred in the aftermath of notification must be considered effects of the SORA for purposes of overcoming the Act's regulatory purpose. Though the Act is doubtless the "but for" cause of some of these incidents--those perpetrated by persons who gained knowledge of the offender's past crime and current location only because of notification, these incidents are not consequences imposed by the Act. The incidents (1) are wholly dependent on acts by private third parties, (2) result from information most of which was publicly available prior to the SORA, and (3) flow essentially from the fact of the underlying conviction.

Several factors bear on whether hostile private conduct, facilitated by the enhanced dissemination of information about an offender's prior conviction, are the sort of "effects" that render a regulatory statute penal. First, although it is foreseeable, indeed intended, that persons notified of the presence of a convicted sex offender in their midst will take some action to protect themselves and their families, such as warning their children not to socialize with the offender, illegal actions by members of the public, such as physical attacks against the offender, are not consequences that the operation of the SORA contemplates or condones. The Act explicitly provides that anyone misusing information obtained through its notification provisions will be criminally prosecuted and subject to fines beyond those already available in the criminal code. See, e.g., N.Y. Correct. Law §§ 168-p(3), 168-q(2), 168-u.

Second, although the SORA makes information regarding an offender's identity, whereabouts, and background easier for the public to obtain, much of this information was publicly accessible before the enactment of the SORA. See, e.g., Werfel v. Fitzgerald, 23 A.D.2d 306, 310, 260 N.Y.S.2d 791, 796-97 (2d Dep't 1965) (noting general New York state policy of allowing unrestricted access to public records, including criminal records); N.Y. Correct. Law §§ 149 & 149-a (requiring Corrections Department to notify local police and victim of the release of certain convicted felons); Division of Parole, Guidelines for the Supervision of Sex Offenders 3-4 (Nov.1994) (authorizing release of conviction and other background information on paroled sex offenders to schools, neighbors, and others at risk). Though the "vast difference between the public records that might be found after a diligent search of courthouse files, county archives, and local police stations ... and a computerized summary located in a single clearinghouse of information," United States Department of Justice v. Reporters Committee for Freedom of the Press, 489 U.S. 749, 764, 109 S. Ct. 1468, 1477, 103 L. Ed. 2d 774 (1989) ("Reporters Committee "), is relevant to some legal issues, such as the scope of the privacy interest assessed in construing the privacy exemption of the Freedom of Information Act, 5 U.S.C. § 552(b)(7)(C) (1994), see Reporters Committee, 489 U.S. at 762-71, 109 S.Ct. at 1476-81, that difference does not add a punitive consequence to an otherwise regulatory measure.

Finally, we note that although notification conveys to the public the information that prompts some people to take unlawful action against the convicted sex offender, it is the offender's prior conviction--or, more precisely, the offender's criminal act itself--that motivates such hostile action. As the Ninth Circuit noted in another context,

[T]he societal consequences that flow from a criminal conviction are virtually unlimited. Individuals may lose their jobs or be foreclosed from serving in future professions; their marriages are destroyed; they may be plunged into poverty. Some individuals may be deported, ... [and others] may lose their homes.... Virtually all individuals who are convicted of serious crimes suffer humiliation and shame, and many may be ostracized by their communities.

United States v. Koon, 34 F.3d 1416, 1454 (9th Cir.1994), aff'd in part, rev'd in part on other grounds, --- U.S. ----, 116 S. Ct. 2035, 135 L. Ed. 2d 392 (1996).

Whatever incremental burdens upon convicted sex offenders arise from public notification are not "so disproportionately severe and so inappropriate to nonpunitive ends" as to constitute punishment. See Nixon, 433 U.S. at 473, 97 S.Ct. at 2805 (classifying burden for bill of attainder purposes).

2. Relation of notification to criminal activity. The plaintiffs also point to the somewhat close relation between the notification provisions and criminal activity as indicative of the Act's punitive nature. As the District Court stated, in agreement with the plaintiffs, these provisions are "triggered by behavior that is already a crime," "provide[ ] for the sentencing judge to determine the level of notification," and "provide[ ] for consideration of victim impact statements in determining" the offender's risk level. Doe v. Pataki II, 940 F. Supp. at 623 (quotations omitted). We do not believe that these aspects of the SORA suffice to transform it into punishment.

First, the fact that the Act's requirements are triggered by a criminal conviction is common to all regulatory disabilities that result from a prior conviction, for instance, the loss of the right to vote.17 The disabilities mandated by the laws challenged and upheld in several Supreme Court decisions have also been triggered solely by the existence of a prior conviction. See, e.g., Hawker, 170 U.S. at 196-97, 18 S.Ct. at 576-77 (prior felony conviction conclusive evidence of lack of fitness to practice medicine). As with the laws upheld in Hawker, Galvan, DeVeau, and Flemming, the offender's prior conviction is used by the SORA "solely for evidentiary purposes," i.e., as a presumption that the offender is likely to re-offend in the future. See Hendricks, --- U.S. at ----, 117 S. Ct. at 2082.

Second, as we have discussed, although the sentencing judge is assigned the ultimate authority to determine an offender's risk level, two features of this process indicate its prospective and nonpunitive nature. First, the risk-level determination is not made at the time of the original sentencing, but just prior to the offender's scheduled release. Second, the court acts only upon the recommendations of the Board, a panel of experts on sex offenders relying upon a guideline that focuses exclusively on whether the offender is likely to re-offend.18

Third, the fact that the Act requires the sentencing court and the Board to consider the victim-impact statement in assessing the risk level does not indicate a desire to punish the offender. There is every reason to believe that such statements are relevant to a determination of the amount of harm likely to result from the offender's actions if he should re-offend. Under the SORA, the offender's "risk" level is a composite of both his likelihood to re-offend and the degree of harm likely to result from the re-offense. See N.Y. Correct. Law § 168-l (5). The victim's account of the effect of the offender's act is relevant to such an inquiry.

As the Supreme Court stated in Hendricks, the fact that a challenged statute uses " 'procedural safeguards usually found in criminal trials' ... does not transform a civil commitment proceeding into a criminal prosecution." --- U.S. at ----, 117 S. Ct. at 2083 (quoting Allen v. Illinois, 478 U.S. 364, 371, 106 S. Ct. 2988, 2993, 92 L. Ed. 2d 296 (1986)). Similarly, the fact that the SORA is triggered by the existence of a prior conviction for a sex offense and involves the sentencing judge and the victim's perspective in its procedures does not render the Act punitive, especially since these features, on a closer look, facilitate nonpunitive goals. See Ursery, --- U.S. at ----, 116 S. Ct. at 2149 (that a law is "tied to criminal activity ... is insufficient to render [it] punitive").

3. Excessiveness of notification. The plaintiffs also contend that the broad coverage of the SORA and the wide extent of the notification that it authorizes indicate its punitive character. They point out that the Act's notification provisions (1) cover 36 offenses, including seven misdemeanors and several crimes not involving sexual contact, (2) encompass first time offenders, (3) allow public notification even for low risk offenders, and (4) permit entities with vulnerable populations to disseminate information to the public with unfettered discretion. We believe, however, that the SORA's notification provisions are reasonably related to the nonpunitive, prospective goals of protecting the public and facilitating law enforcement efforts, and that the Act's occasionally imprecise targeting does not suffice to render it punitive.

First, although the Act technically covers a large number of offenses,19 including some, like statutory rape, that appear to present a far less compelling need for community notification than offenses like child molestation, the Act also contains a number of moderating provisions capable of greatly limiting the extent of notification or even of relieving the offender from notification altogether. For instance, the process for assigning risk levels weighs numerous factors designed to determine the offender's likelihood of re-offense and the potential harm resulting from such a relapse, including the relative seriousness of the crime of conviction, the relationship between the offender and the victim, and the offender's psychological profile. See N.Y. Correct. Law § 168-l (5). For example, an offender whose crime did not involve sexual contact with a stranger will likely be deemed to have a "low" risk of re-offense and assigned a level-one designation. Under the Act, only the most limited form of notification--restricted public access via the fee-based 900 number--is permitted for such offenders. See id. §§ 168-l (6)(a), 168-p. Moreover, the Act allows any offender within its coverage to petition the court for relief from its requirements. See id. § 168-o. If the court determines, after reviewing an updated report from the Board, that notification is no longer appropriate because the offender does not present a danger to the community, the offender will be free from the Act's operation.

Second, although the Act technically subjects even low risk offenders to public notification, the extent of such notification is extremely limited. For level-one offenders, only public access, not affirmative dissemination of information to the community, is available under the Act. Id. § 168-l (6)(a). Moreover, a member of the public may obtain information through the 900 number system only after providing specific information identifying an individual suspected of being on the DCJS's central registry. Id. § 168-p. Additionally, many safeguards exist to prevent the misuse of this service, id. § 168-p(2)-(3), and the only information that can be obtained on a level-one offender is that he is in the registry and that he presents a low risk of re-offense.

Third, although the Act applies to first time offenders, the limitations on notification are also relevant to such coverage. Notification is limited for persons unlikely to re-offend. Among the numerous factors relevant to the determination of an offender's notification level, for instance, are "criminal history factors indicative of high risk of repeat offense," id. § 168-l (5)(a), including "the number, date and nature of prior offenses," id. § 168-l (5)(b)(iii). Similarly, a first time offender who can demonstrate that he does not present a danger to the community may petition the court to be relieved from the Act's requirements. Id. § 168-o. Moreover, the legislature's refusal to exclude first time offenders from the SORA is supported by the possibility that, in many cases, a sex offender's first encounter with the criminal justice system might have occurred after numerous unreported offenses.

Finally, the fact that the Act permits entities with vulnerable populations to disseminate, at their discretion, information they receive from local law enforcement authorities does not weigh heavily toward rendering it punitive. The legislature's decision to permit a day-care center, for example, to determine whether additional notice to parents of children who attend the center would be appropriate is justified on the grounds that such "secondary" notification may be necessary in some instances to promote public safety and facilitate the prosecution of sex crimes, and that such entities know better than either the legislature or the police when additional notification is needed.

Although the SORA might be somewhat imprecise and occasionally overbroad in its coverage, the legislature acted well within its authority in determining the kind of offenses triggering notification, the category of persons subject to the Act's requirements, and the extent of notification appropriate to promote its nonpunitive goals. The legislature is not required to act with perfect precision, and its decision to cast a net wider than what might be absolutely necessary does not transform an otherwise regulatory measure into a punitive sanction.

4. Notification and the goals of the criminal law. Plaintiffs point out that the Act serves to deter future misconduct, constitutes retribution against sex offenders, and interferes with offenders' ability to rehabilitate. They contend that these criminal law-related features of the SORA render it punitive. We disagree.

First, the fact that notification serves to deter sex offenders, both specifically and generally, from committing crimes in the future is not indicative of punitiveness. As the Supreme Court said in Ursery, deterrence "may serve civil as well as criminal goals." --- U.S. at ----, 116 S. Ct. at 2149; see also 89 Firearms, 465 U.S. at 364, 104 S.Ct. at 1106. The civil forfeiture determined to be nonpunitive in Ursery, for instance, both specifically deters the person whose property is forfeited because it was used to facilitate marijuana offenses, and generally deters others who might wish to engage in similar activities. Second, though some members of the public might applaud the notification provisions for, in Judge Chin's words, "giving the sex offenders what ... they deserve," Doe v. Pataki II, 940 F. Supp. at 629, such retributive views may not be ascribed, without evidence, to the legislature that enacted the SORA. As in Hendricks, notification pursuant to the Act "is not retributive because it does not affix culpability for prior conduct"; rather, the Act relies on such conduct "solely for [the] evidentiary purpose[ ]" of determining whether the offender is likely to present a danger to the community once he is released. --- U.S. at ----, 117 S. Ct. at 2082. Finally, the plaintiffs' and the District Court's conclusion that notification interferes with an offender's ability to rehabilitate is speculative: for some there might be interference but for others it is likely that rehabilitation will be furthered by the knowledge that neighbors and the local police are aware of the offender's compulsions, thereby reducing the chances of committing further crimes without detection.

Even if the SORA advances some goals traditionally associated with the criminal law, it primarily "serve[s] important nonpunitive goals" of protecting the public from potential dangers and facilitating future law enforcement efforts. See Ursery, --- U.S. at ----, 116 S. Ct. at 2148.

5. Historical analogues. Finally, plaintiffs contend that notification is the modern day incarnation of traditional "stigmatization" penalties such as stocks, pillories, and branding. They also contend that because notification occasionally results in the offender's being driven from his residence, it is akin to the sanction of banishment. The similarity of a challenged sanction to measures that have historically been considered punitive is sometimes an important consideration in the ex post facto analysis, but we do not believe that public notification is analogous to either shaming penalties or banishment.

Stigmatization penalties of an earlier era primarily served distinctively punitive goals and operated through significantly different mechanisms than community notification pursuant to the SORA. First, such penalties were traditionally employed in small, homogeneous, and tightly knit social environments in which the "invisible whip of public opinion" upon the psyche of the offender, Lawrence M. Friedman, Notes Toward a History of American Justice, reprinted in American Law and the Constitutional Order 14 (enlarged ed.1988), was often considered sufficient to serve the traditional goals of the criminal law--deterrence, retribution, and rehabilitation. For many offenses, especially "victimless" moral or religious crimes, no other sanction apart from public humiliation was imposed by colonial communities. See, e.g., Adam J. Hirsch, From Pillory to Penitentiary: The Rise of Criminal Incarceration in Early Massachusetts, 80 Mich. L.Rev. 1179, 1226 (1982). The stigmatization penalties of that era were intended to visit the community's wrath upon the transgressor and to shame him into rejoining the flock. See H. Barnes, The Story of Punishment 39 (1930).

By contrast, public notification pursuant to the SORA is not imposed either in lieu of incarceration or fines--our society's preferred modes of punishment--or as another component of the offender's criminal sentence. Rather, the "burden" of notification is imposed only after the designated punishment has been served, and is intended to serve the goals of protecting the public and facilitating future law enforcement efforts. Second, traditional shaming penalties such as branding or the stocks enlisted the offender's physical participation in his own degradation. Notification pursuant to the SORA, however, requires no participation by the offender. Notification itself imposes no physical pain, mark, or restraint on the offender.

Notification is even less like the traditional penalty of banishment, the purpose of which is to remove someone considered to present a "permanent danger" to the community. Lawrence M. Friedman, Crime and Punishment in American History 40 (1993). With banishment, the state acts to remove the offender. By contrast, if a sex offender is evicted by his landlord or pressured to move from his residence by his neighbors after his criminal history becomes known, those private actions, however unfortunate, are not intended consequences of the SORA, and the historical analogy fails.

Our analysis, of necessity, has given separate consideration to each of the factors relied on by the plaintiffs to establish the penal nature of the notification provisions, but we acknowledge that their aggregate effect must ultimately be assessed. We also recognize that whenever statutes impose burdens on those convicted of crime, those burdens will appear to be punishment from the perspective of the offender, even though the proponents of the legislation will normally be able to articulate a valid regulatory purpose. It will usually be an easy task for a court to determine whether the legislature intends a regulatory purpose but often a more difficult task to determine whether an avowedly regulatory purpose has been achieved by the imposition of burdens fairly classified as punishment for purposes of the Ex Post Facto Clause.

The notification provisions of the SORA present the problem in a troublesome context, framed by the strong public interest in alerting relevant segments of the community to the presence of released sex offenders and by the understandable interest of the released offender in beginning life after confinement with minimal disabilities. We would be less than candid if we did not acknowledge that the effects of notification provisions, like those of New York and other states, will be more severe for some offenders than for others. The released sex offender living in a small community might well become a pariah, pressured to move away, unless the magnanimous spirit of neighborliness prevalent in much of small town America overcomes initial community fears. Other offenders, returning or moving to large cities, are more likely to take refuge in the anonymity of urban life.

We fully understand how Judge Chin reached his conclusion that the notification provisions are punitive, and we have given his opinion careful consideration. Nevertheless, assessing all aspects and consequences of the notification provisions, and applying the constitutional standards as we believe the Supreme Court has enunciated them to the aggregate of these consequences, we conclude that the plaintiffs have not provided "the clearest proof" that the burdens attendant to these provisions are "so punitive in form and effect," Ursery, --- U.S. at ----, 116 S. Ct. at 2148, as to transform them into punitive sanctions. Because we have previously concluded that the legislature's intent in enacting these provisions was nonpunitive and that the text and structure of the Act bear out its prospective, regulatory goals, we hold that the notification requirements of the SORA do not constitute punishment for purposes of the Ex Post Facto Clause.

We agree with the District Court that the registration requirements of the SORA do not impose punishment upon the plaintiffs. Although the plaintiffs contended in the District Court that all of these requirements violated the ex post facto prohibition, they limit their challenge on this appeal to those registration provisions applicable only to certain high risk offenders. Under the Act, offenders classified as "sexually violent predators" must register (1) personally, (2) every 90 days, and (3) for a minimum of ten years and potentially for life. See N.Y. Correct. Law §§ 168-f(3), 168-h. Following the same two part inquiry applied to the notification provisions, we conclude that these registration requirements are not punitive.

First, the legislature enacted the registration provisions primarily to serve the nonpunitive purpose of enhancing future law enforcement efforts. As the Act's preamble states:

The system of registering sex offenders is a proper exercise of the state's police power regulating present and ongoing conduct. Registration will provide law enforcement with additional information critical to preventing sexual victimization and to resolving incidents involving sexual abuse and exploitation promptly. It will allow them to alert the public when necessary for the continued protection of the community.

Id. § 168 (Historical and Statutory Notes). Moreover, because registration is the necessary prerequisite for community notification, registration also serves the general goal of protecting members of the public from the potential dangers posed by convicted sex offenders. The text and structure of the registration provisions of the Act further bear out the stated nonpunitive goals. For instance, the duration, form, and frequency of registration are determined solely by whether the offender qualifies as a sexually violent predator, see id. §§ 168-f(3), 168-h, the offender can petition the court to be relieved from the duty to register, see id. § 168-o, and the Act prohibits the unauthorized release or use of registration information, see id. § 168-u.

Second, the burdens of registration are not "so punitive in form or effect" as to transform this regulatory measure into a punitive one. For instance, registration itself, which can usually be accomplished without public notice, (1) does not ordinarily result even in societal opprobrium or harassment, (2) does not serve the goals of criminal punishments, and (3) does not resemble any measures traditionally considered punitive. Moreover, although the Act deems the failure to register a crime, see id. § 168-t, this provision is fully prospective. Additionally, we note that an offender required to register can petition the court to be relieved of that duty. See id. § 168-o.

Although we recognize that the duty to register in person every 90 days for a minimum of ten years is onerous,20 we do not believe that this burden is sufficiently severe to transform an otherwise nonpunitive measure into a punitive one. As discussed previously, the Supreme Court has consistently upheld far heavier burdens against ex post facto challenges, including deportation, termination of financial support, and loss of livelihood. Moreover, the legislature's decision to subject persons with a greater risk of re-offense to more onerous registration requirements is reasonable. Because the plaintiffs have not demonstrated by the clearest proof that the registration provisions of the SORA are punitive in form and effect, we hold that these provisions of the Act do not constitute punishment for purposes of the Ex Post Facto Clause

We conclude that application of the registration and notification provisions of the SORA to persons who committed their offenses prior to the January 21, 1996, effective date of the Act does not violate the Ex Post Facto Clause. We do not resolve the plaintiffs' remaining statutory and constitutional arguments, as the District Court did not reach them in its decision and as the parties did not present them on this appeal. We therefore affirm the portion of the District Court's judgment dismissing plaintiffs' claim that retroactive application of the Act's registration provisions violates the Ex Post Facto Clause, reverse the portion permanently enjoining the retroactive application of the Act's public notification provisions, and remand for further proceedings on plaintiffs' remaining claims.

Appellees' petition for rehearing raises three matters that warrant brief discussion.

First, appellees point out that the opportunity in the SORA to petition for relief is limited to relief from the requirement of "any further duty to register," SORA § 168-o. As appellants note, once a person has registered, relief from a duty to comply with subsequent periodic registration requirements does not afford relief from the consequences of the notification that occurred as a result of the prior registration. On the other hand, any sex offender who obtains relief from registration and thereafter moves has obtained significant relief from the burden of notification because the new address will not be subject to notification. In any event, the limited scope of the relief provision does not cause us to alter our ultimate conclusion with respect to the ex post facto challenge to the SORA. However, we will amend our opinion in two respects: (a) on slip op. p. 5887, third line from the bottom, we delete "and notification"; (b) on slip op. p. 5896, line 3 [120 F.3d at 1282], we insert "registration" before "requirements."

Second, appellees dispute that, prior to the SORA, New York had a policy of general access to public records, including criminal records. They point out that criminal history records compiled by the Division of Criminal Justice Services have been held to be exempt from disclosure pursuant to New York's Freedom of Information Law. See N.Y.Pub.Off.Law §§ 84-90 (McKinney 1988); Woods v. Kings County District Attorney's Office, 234 A.D.2d 554, 651 N.Y.S.2d 595, 596 (2d Dep't 1996). However, a court record of a defendant's conviction remains a document available for public inspection, see N.Y.Jud.Law § 255-b (McKinney 1983) (docket book, kept by clerk of court, must be kept open during business hours for examination by any person), and has been held not to be exempt from disclosure under the State's Freedom of Information Law, see Thompson v. Weinstein, 150 A.D.2d 782, 542 N.Y.S.2d 33 (1989).

Third, appellees call our attention to the statement in People v. Letterlough, 86 N.Y.2d 259, 266, 631 N.Y.S.2d 105, 108 (1995), that public disclosure of a person's crime has been regarded as a form of punishment. That statement was made with reference to a trial judge's condition of probation that required a person convicted of driving under the influence of alcohol to display on his license plate the legend "convicted dwi." The Court's decision was primarily concerned with the trial judge's invasion of the legislature's prerogatives. See id. at 266-67, 631 N.Y.S.2d at 109 ("More importantly, the detailed condition imposed here is objectionable because in fashioning it the trial court invaded the legislative domain."). In any event, the Court's observation as to how public disclosure of crime has been viewed historically was not concerned with whether such disclosure is punishment for purposes of the Constitution's Ex Post Facto Clause.

Accordingly, the opinion is amended as indicated herein. In all other respects, the petition for rehearing is denied.