We must decide whether a Washington statute of the kind popularly referred to as "Megan's law" violates the Constitution.

* Willie Russell and Johnny Stearns are both convicted sex offenders who have been released from confinement and are now residing in Washington State. Russell was convicted in Washington in 1989 of second degree rape and attempted second degree rape; Stearns was convicted in Washington in 1989 of first degree robbery and attempted second degree rape. Both were imprisoned. In 1990, the Washington legislature passed the Community Protection Act, 1990 Wash. Laws, ch. 3 ("the Act") which included provisions requiring sex offenders to register with local law enforcement authorities and subjecting some offenders to community notification of conviction information including the offender's photograph and approximate residential location.1

After their releases, both Russell and Stearns registered as required. Each learned that he would be subject to community notification under the Act. In early 1996, each filed civil rights actions under 42 U.S.C. § 1983 against various state officials, including Washington Attorney General Christine Gregoire, claiming that the registration and notification provisions of the Act would deprive him of his constitutional rights, and requesting declaratory and injunctive relief. Specifically, each claimed that the Act violated the Ex Post Facto Clause, and abridged his rights to privacy and due process.

* The Act requires both registration and notification. The registration element provides that any person convicted of a sex offense (or found not guilty by reason of insanity of a sex offense) register with the sheriff for the county of the person's residence. Wash. Rev.Code § 9A.44.130(1). The offender must provide his or her name, address, date and place of birth, place of employment, crime for which he or she was convicted, date and place of conviction, aliases used, and social security number. Id. at § 9A.44.130(2). The sheriff must obtain a photograph and a copy of the offender's fingerprints. Id. at § 9A.44.130(5). The information is forwarded to the Washington State Patrol for inclusion in a central registry. Id. at § 43.43.540.

A sex offender released from custody must register within 24 hours of his or her release; sex offenders who change addresses must register within 10 days of the change. Id. at § 9A.44.130(3)(a). The Act also requires a sex offender convicted in another jurisdiction to register within 30 days of moving into Washington. Id. If a sex offender changes residences, he or she must notify the sheriff 14 days before moving, or, if not within 14 days, as soon as the new address is known. Id. at § 9A.44.130(4). Failure to register within the time required is punishable as a Class C felony or gross misdemeanor depending on the degree of the offender's underlying sex offense. Id. at § 9A.44.130(7).

The notification element authorizes public agencies to release "relevant and necessary information" regarding a sex offender2 to the public when "necessary for public protection." Id. at § 4.24.550. The Washington Supreme Court3 has held that "a public agency must have some evidence of an offender's future dangerousness, likelihood of reoffense, or threat to the community, to justify disclosure to the public in a given case." State v. Ward, 123 Wash.2d 488, 869 P.2d 1062, 1070 (1994). "An agency must disclose only that information relevant to and necessary for counteracting the offender's dangerousness." Id. Any notice given to the public must contain a warning against violence towards the offender. Id. Further, information may only be disseminated within a narrow geographic area. Id. 869 P.2d at 1070-71.

Prior to an offender's release, Washington's Sex Offender Oversight Committee reviews information provided by the Department of Corrections to assess the seriousness level of the offender. If an offender is classified as Level One, no public notification occurs. If an offender is classified as Level Two, standard notification forms are provided to government and law enforcement agencies, to schools within the federal census tract where the offender is living, and to Block Watch Captains in that census tract and the adjoining census tracts. If an offender is classified as Level Three, notification forms are distributed in the same way, but are also provided to local news media. Russell and Stearns were each classified as Level Three offenders.

The notification form provides the offender's picture, name, age, date of birth, and other identifying information. It also contains a summary of the offender's crime and the general vicinity of the offender's residence. It does not contain the offender's exact address, nor does it contain any information about the offender's employment. Finally, the form has a long "caveat" regarding the information contained in the notice, including a warning that threats, intimidation, or harassment of the offender "will not be tolerated."

The district court granted and renewed Temporary Restraining Orders in favor of Russell and Stearns, preventing law enforcement agencies from making the statutory notification to the community until hearing argument on their motions for preliminary injunction. After a hearing, the district court denied their motions for preliminary injunction, thus permitting notification. Russell and Stearns filed this timely joint appeal from the order denying a preliminary injunction.

A preliminary injunction may issue "if the movant has shown either a likelihood of success on the merits and the possibility of irreparable injury, or that serious questions are raised and the balance of hardships tips sharply in the movant's favor." Coalition for Economic Equity v. Wilson, 110 F.3d 1431, 1438 (9th Cir.1997) (quotation marks and citation omitted).

We review the district court's order denying a preliminary injunction for an abuse of discretion, which occurs if the district court bases its decision on an erroneous legal standard or on clearly erroneous findings of fact. Does 1-5 v. Chandler, 83 F.3d 1150, 1152, (9th Cir.1996). We review de novo the legal issues underlying the decision to deny an injunction, as well as the conclusion that plaintiffs are likely to fail on the merits of those issues. International Molders' and Allied Workers' Local Union No. 164 v. Nelson, 799 F.2d 547, 551 (9th Cir.1986).4

Article I, § 10 of the Constitution provides: "No State shall ... pass any ... ex post facto Law...." It prohibits the states from enacting any law "which imposes a punishment for an act which was not punishable at the time it was committed; or imposes additional punishment to that then prescribed." Weaver v. Graham, 450 U.S. 24, 28, 101 S. Ct. 960, 964, 67 L. Ed. 2d 17 (1981) (quoting Cummings v. Missouri, 71 U.S. (4 Wall.) 277, 325-26, 18 L. Ed. 356 (1866)). "[T]he focus of the ex post facto inquiry is not on whether a legislative change produces some sort of 'disadvantage,' ... but on whether any such change alters the definition of criminal conduct or increases the penalty by which a crime is punishable." California Dep't of Corrections v. Morales, 514 U.S. 499, 506 n. 3, 115 S. Ct. 1597, 1602 n. 3, 131 L. Ed. 2d 588 (1995); see Collins v. Youngblood, 497 U.S. 37, 50-52, 110 S. Ct. 2715, 2723-24, 111 L. Ed. 2d 30 (1990). Since Russell and Stearns argue that the Act is an ex post facto law, the principal question presented in this case is whether the registration and notification provisions impose "punishment."A

Our court has not previously established a clear test to determine what constitutes punishment under the Ex Post Facto Clause. Likewise, the Supreme Court has not articulated a "formula" for identifying the legislative changes that fall within the constitutional prohibition. Morales, 514 U.S. at 508-10, 115 S.Ct. at 1603.

In Kennedy v. Mendoza-Martinez, 372 U.S. 144, 83 S. Ct. 554, 9 L. Ed. 2d 644 (1963), a case dealing with the procedural safeguards of the Fifth and Sixth Amendments, the Court enumerated factors to be considered:

Whether the sanction involves an affirmative disability or restraint, whether it has historically been regarded as a punishment, whether it comes into play only on a finding of scienter, whether its operation will promote the traditional aims of punishment-retribution and deterrence, whether the behavior to which it applies is already a crime, whether alternative purposes to which it may rationally be connected is assignable for it, and whether it appears excessive in relation to the alternative purpose assigned are all relevant to the inquiry, and may often point in differing directions.

Id. at 168-69, 83 S.Ct. at 567-68. As the Court later noted, however, Mendoza-Martinez 's list of considerations is often "helpful" but is "certainly neither exhaustive nor dispositive." United States v. Ward, 448 U.S. 242, 249, 100 S. Ct. 2636, 2641, 65 L. Ed. 2d 742 (1980).

The Supreme Court has decided a series of punishment cases recently, including United States v. Ursery, --- U.S. ----, 116 S. Ct. 2135, 135 L. Ed. 2d 549 (1996).5 Ursery held that in rem civil forfeitures were not punishments under the Double Jeopardy Clause. In reaching that conclusion, the Court returned to the two-part test for punishment it announced in United States v. One Assortment of 89 Firearms, 465 U.S. 354, 104 S. Ct. 1099, 79 L. Ed. 2d 361 (1984), which we shall call the "intent-effects" test. "First, we ask whether Congress intended proceedings ... 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." Ursery, --- U.S. at ----, 116 S. Ct. at 2147 (quoting 89 Firearms, 465 U.S. at 366, 104 S.Ct. at 1107). The Court in Ursery applied a broad view of nonpunitive purposes for the provision, and allowed only "the 'clearest proof' " of a punitive effect to overwhelm a nonpunitive purpose. Id. at ---- - ----, 116 S. Ct. at 2147-48; Sardone, 94 F.3d at 1235-36. And although the Court did not explicitly cite Mendoza-Martinez, its analysis under the "effects" prong examined many of the Mendoza-Martinez factors: whether the statute has nonpunitive goals or effects, whether the sanction has historically been regarded as punishment, whether scienter is required, and whether it is tied to criminal activity. Ursery, --- U.S. at ---- - ----, 116 S. Ct. at 2148-49.

On the surface, it would appear that Ursery 's version of the "intent-effects" test should govern this case. After we had oral argument in this case, however, the Supreme Court handed down two decisions construing the Ex Post Facto Clause: Lynce v. Mathis, --- U.S. ----, 117 S. Ct. 891, 137 L. Ed. 2d 63 (1997), and Kansas v. Hendricks, --- U.S. ----, 117 S. Ct. 2072, 138 L. Ed. 2d 501. We must first determine the impact of Lynce and Hendricks on the Ursery test for punishment.

* In Lynce, Florida had granted early release credits to state prisoners to alleviate prison overcrowding. It later revoked those credits for certain violent offenders, including Lynce, who had already been released. As a result, Lynce was rearrested and incarcerated. The Court ruled that the application of the revocation to Lynce violated the Ex Post Facto Clause.

Lynce is the latest installment in the line of cases dealing with changes in the law that decrease the chances that a prisoner will be released from prison early or will receive a shorter prison term. This line began with Lindsey v. Washington, 301 U.S. 397, 57 S. Ct. 797, 81 L. Ed. 1182 (1937), which held that a retrospective law setting a minimum prison sentence was ex post facto "since the measure of punishment prescribed by the later statute is more severe than that of the earlier." Id. at 401, 57 S.Ct. at 799. Later, the Court decided Weaver v. Graham, 450 U.S. 24, 101 S. Ct. 960, 67 L. Ed. 2d 17 (1981), ruling that a law reducing the availability of "gain time" credits for early release for prisoners violated the Ex Post Facto Clause because it made "more onerous the punishment for crimes committed before its enactment." Id. at 36, 101 S.Ct. at 968. Then, in Miller v. Florida, 482 U.S. 423, 107 S. Ct. 2446, 96 L. Ed. 2d 351 (1987), the Court, quoting language from both Lindsey and Weaver, invalidated the retrospective application of a state sentencing guidelines statute which increased the petitioner's presumptive prison sentence. Id. at 432-35, 107 S.Ct. at 2452-54.

Each case in the Lindsey-Weaver-Miller trio contained language suggesting that enhancements to the measure of punishment are prohibited by the Ex Post Facto Clause because they operate to the "disadvantage" of covered offenders. See Lindsey, 301 U.S. at 401, 57 S. Ct. at 799; Weaver, 450 U.S. at 29, 101 S.Ct. at 964-65; Miller, 482 U.S. at 433, 107 S.Ct. at 2452-53. In Collins v. Youngblood, 497 U.S. 37, 41, 110 S. Ct. 2715, 2718-19, 111 L. Ed. 2d 30 (1990), however, the Supreme Court conducted an extensive examination of the original understanding of the Ex Post Facto Clause and disapproved of similar "disadvantage" language. The Court went a step further in California Dep't of Corrections v. Morales, 514 U.S. 499, 115 S. Ct. 1597, 131 L. Ed. 2d 588 (1995), which upheld California's retrospective increase in the time between parole hearings for serious offenders. Morales expressly repudiated the Lindsey-Weaver-Miller trio's "disadvantage" language and stated that it "was unnecessary to the results in those cases and is inconsistent with the framework developed in Collins." Id. at 506 n. 3, 115 S. Ct. at 1602 n. 3 (citation omitted). Instead, "the focus of the ex post facto inquiry is ... on whether a legislative change ... alters the definition of criminal conduct or increases the penalty by which a crime is punishable." Id.

What is the common thread of these cases? When seen in the light of the Collins-Morales framework, Lindsey, Weaver, Miller, and Morales each turned on whether the change increased the quantum of punishment attached to an already-committed crime. The Court in those cases had no need to determine the threshold question of whether the sanction was punishment since a criminal sentence of imprisonment is plainly punishment. Properly viewed, these are not punishment-defining cases at all.

Lynce is no different. It focuses on whether the Florida statute increased Lynce's punishment-not on whether Lynce's reimprisonment was punishment in the first place. See, e.g., Lynce, --- U.S. at ----, 117 S. Ct. at 895 ("The narrow issue that we must decide is ... whether [the] consequences [of the law] disadvantaged petitioner by increasing his punishment."); id. at ----, 117 S.Ct. at 897 ("[Morales concluded that] the change at issue had neither the purpose nor the effect of increasing the quantum of punishment." ); id. at ---- n. 16, 117 S. Ct. at 898 n. 16 (distinguishing Morales because statutory scheme there was not "more 'onerous' "). Most significantly, Lynce does not even mention Ursery or the other cases addressing whether a particular sanction is punishment, but confined itself to a discussion of Lindsey, Weaver, Miller, and Morales.

Understanding Lynce 's place in the Court's jurisprudence is important for our decision today. Russell and Stearns point out in their supplemental brief that Lynce contains a discussion of the limited role of legislative intent in examining whether a sanction increases the quantum of punishment. See id. at ----, 117 S.Ct. at 897. They argue that Lynce thereby discards, sub silentio, Ursery 's "intent-effects" test.

We reject this argument for two reasons. First, Lynce teaches only that legislative intent is not generally relevant to whether the amount of a given punishment has increased. Lynce did not question whether the prison sentence was punishment, but only whether a decrease in early release credits increased that punishment. In contrast, intent is very important in determining whether a particular sanction is punishment in the first place. Ursery, --- U.S. at ----, 116 S. Ct. at 2147. These are distinct questions, and legislative intent is crucial in answering the latter, but not necessarily so in answering the former.

Second, the Court has just reiterated the primacy of legislative intent in Hendricks, to which we now turn.

Hendricks involved Kansas' Sexually Violent Predator Act, which establishes procedures for the civil commitment of persons who, due to a "mental abnormality" or "personality disorder," are likely to engage in "predatory acts of sexual violence." Hendricks, --- U.S. at ----, 117 S. Ct. at 2076. The Court ruled that, even though sexual predators are confined in state institutions, the Kansas Act does not constitute punishment and therefore does not violate the Ex Post Facto and the Double Jeopardy Clauses. In reaching its conclusion, the Court applied essentially the same test for punishment as in Ursery:

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 (citation and brackets omitted). To prove that a civil proceeding imposes punishment is a "heavy burden." Id. Indeed, even if the legislature's nonpunitive purpose was merely ancillary to another overriding or primary purpose of the statute, a court may conclude that the statute is not punitive. Id. at ----, 117 S.Ct. at 2082.

The Court went on to note that the Kansas statute does not implicate either of the two primary objectives of criminal punishment, retribution and deterrence, and does not turn on a finding of scienter. Id. at ----, 117 S.Ct. at 2082. That the statute imposes an affirmative restraint and imposes a sanction traditionally regarded as punishment does not override its nonpunitive nature. Id. at ----, 117 S.Ct. at 2079.

Where does the Ursery punishment test stand now? We are persuaded that the Court's approach to the punishment question is essentially the same in both Ursery and Hendricks. Nothing in Lynce is to the contrary. Consequently, we will apply the Ursery-Hendricks "intent-effects" test to determine whether the Act imposes punishment.6 When examining whether a law violates the Ex Post Facto Clause, we inquire whether (1) the legislature intended the sanction to be punitive, and (2) the sanction is "so punitive" in effect as to prevent the court from legitimately viewing it as regulatory or civil in nature, despite the legislature's intent. Ursery, --- U.S. at ----, 116 S. Ct. at 2147; see Hendricks, --- U.S. at ----, 117 S. Ct. at 2080. The first part of the test ("intent") looks solely to the declared purpose of the legislature as well as the structure and design of the statute.7 Ursery, --- U.S. at ----, 116 S. Ct. at 2147 (examining terms used by Congress and structure of forfeiture statute under first part of test); see United States v. Huss, 7 F.3d 1444, 1447 (9th Cir.1993) (deciding pre-Ursery that court should look to intent and design of statute as well as effects). The second part of the test ("effects") requires the party challenging the statute to provide "the clearest proof" that the statutory scheme is so punitive either in purpose or effect as to negate the State's nonpunitive intent. Hendricks, --- U.S. at ----, 117 S. Ct. at 2082; Ursery, --- U.S. at ----, 116 S. Ct. at 2148. See Flemming v. Nestor, 363 U.S. 603, 617, 80 S. Ct. 1367, 1376, 4 L. Ed. 2d 1435 (1960) ("[O]nly the clearest proof could suffice to establish the unconstitutionality of a statute on [the ground that the 'history and scope' of the statute reveal a punitive purpose notwithstanding the legislative intent]."). In assessing the Act's effects, we shall refer to the appropriate Mendoza-Martinez factors. See Hendricks, --- U.S. at ---- - ----, ----, 117 S. Ct. at 2079-80, 2082 (considering some Mendoza-Martinez factors); Ursery, --- U.S. at ---- - ----, 116 S. Ct. at 2148-49 (same).

We now turn our attention to the registration portion of the Act. Applying the "intent-effects" test, we must first look at the language of the statute to see if we may discern the legislature's intent.

* Section 401 of the Act declares:

The legislature finds that sex offenders often pose a high risk of reoffense, and that law enforcement's efforts to protect their communities, conduct investigations, and quickly apprehend offenders who commit sex offenses, are impaired by the lack of information available to law enforcement agencies about convicted sex offenders who live within the law enforcement agency's jurisdiction. Therefore, this state's policy is to assist local law enforcement agencies' efforts to protect their communities by regulating sex offenders by requiring sex offenders to register with local law enforcement agencies as provided in section 402 of this act.

1990 Wash. Laws, ch. 3, § 401. The Supreme Court of Washington looked no further than this recital in determining the intent of the Legislature, noting that the statement of purpose evidences an unequivocal regulatory motivation, State v. Ward, 869 P.2d at 1068, and we agree.

The overall design of the statute's registration provisions indicates a regulatory, not punitive, intent. Registration does no more than apprise law enforcement officials of certain basic information about an offender living in the area. It places no restraint on the offender's movements; an offender need only notify the sheriff of any change of address 14 days before moving, or, if not within 14 days, as soon as the new address is known. Wash. Rev.Code § 9A.44.130(4). He or she may do so by simply mailing a written notice to the sheriff. Id. The information required to be divulged in registering is not burdensome-an offender must provide his or her name, address, date and place of birth, place of employment, crime of conviction, date and place of conviction, aliases used, and social security number. Id. at § 9A.44.130(2). A less-serious offender's duty to register terminates after 10 or 15 years, depending on the class of the underlying offense. Id. at § 9A.44.140(1). These provisions evidence a clear intent to monitor the whereabouts of the offender; we are satisfied that they do not manifest any intent to punish.

The second part of the "intent-effects" test asks whether Russell and Stearns have provided "the clearest proof" that the sanction is "so punitive" in effect that it overcomes the nonpunitive legislative intent. Their only argument on this point is that the registration requirement imposes an affirmative duty to register and a criminal penalty for failure to do so-a situation they claim is analogous to Weems v. United States, 217 U.S. 349, 30 S. Ct. 544, 54 L. Ed. 793 (1910), and United States v. Paskow, 11 F.3d 873 (9th Cir.1993). Those cases are properly distinguished, however.

In Weems, the Supreme Court examined whether a sentence imposed under Philippine law for falsifying a public document-a crime committed by the mere misstatement of facts in a public document even if there was no intent to defraud or actual injury to another-was cruel and unusual punishment. Weems was sentenced to twelve years "hard and painful" labor while shackled at wrist and ankle, "civil interdiction" (loss of property rights, marital rights, and rights of parental authority), perpetual loss of political rights, and perpetual "surveillance" which included notifying authorities of his domicile and obtaining written permission to change residences. The Court concluded that the sentence was cruel and unusual punishment because it was disproportionately harsh when compared to the crime committed. Weems, 217 U.S. at 380-81, 30 S.Ct. at 554-55. In describing the harsh nature of the sentence, the Court discussed the burdens of "surveillance".8 Russell and Stearns latch on to this dictum as demonstrating the severe effects of registration. With clever ellipses, however, they omit from their quotation the fact that Weems not only had to inform the authorities of his domicile, but also had to obtain written permission from them to change residences. Obtaining permission to move is a much greater burden than simple registration, and the Court's language should be read with this sanction in mind. Even the Third Circuit, which has formulated a much broader test for punishment, concluded that, especially in light of the other harsh penalties imposed on Weems, "the Court's dictum about the harshness of 'surveillance' hardly establishes that registration is 'punishment.' " Artway v. Attorney General of New Jersey, 81 F.3d 1235, 1266 (3rd Cir.1996) (addressing New Jersey's sex offender registration requirement).

Paskow involved an offender whose supervised release was revoked after he violated the terms of that release. A statute passed after he committed the underlying offense-but before he violated the terms of his supervised release-imposed a mandatory minimum period of revocation, which we held violated the Ex Post Facto Clause. Paskow, 11 F.3d at 875-76. Paskow is distinguishable, however, because anyone punished for failing to register under the Act is being punished for a new offense, and hence no ex post facto problem exists. In Paskow, we noted that "[r]evocation of parole is not a punishment for a new offense, although the conduct on which revocation is based may be punished separately. For revocation purposes, the conduct simply triggers the execution of the conditions of the original sentence." Id. at 881.

We emphasize that the crime of failing to register under the Act constitutes a separate offense. The fact that a prior conviction for sexual misconduct is an element of the "failure to register" offense is of no consequence. It is hornbook law that no ex post facto problem occurs when the legislature creates a new offense that includes a prior conviction as an element of the offense, as long as the other relevant conduct took place after the law was passed. The Supreme Court has recently suggested as much. See United States v. Watts, --- U.S. ----, 117 S. Ct. 633, 136 L. Ed. 2d 554 (1997).

A consideration of the other Mendoza-Martinez factors does not support a finding that registration has a punitive effect: no affirmative restraint or disability is imposed; registration is typically and historically a regulatory measure; it does not have a retributive purpose but does have legitimate nonpunitive purposes; and it is not excessive given the state interest at stake. Although registration arguably has a deterrent effect, Ursery declared that deterrence can serve both civil and criminal goals. Ursery, --- U.S. at ----, 116 S. Ct. at 2149 (citing Bennis v. Michigan, --- U.S. ----, ----, 116 S. Ct. 994, 1000, 134 L. Ed. 2d 68 (1996)). Ursery also noted that the fact that a sanction may be tied to criminal activity alone is insufficient to render the sanction punitive. Id.; see Hendricks, --- U.S. at ----, 117 S. Ct. at 2082.

Finally, registration provisions have overwhelmingly been sustained as constitutional by other courts. See, e.g., Doe v. Pataki, 120 F.3d 1263, 1265-66 (2d Cir.1997); Artway, 81 F.3d at 1267; Doe v. Kelley, 961 F. Supp. 1105 (W.D.Mich.1997); Doe v. Weld, 954 F. Supp. 425 (D.Mass.1996); People v. Afrika, 168 Misc. 2d 618, 648 N.Y.S.2d 235 (N.Y.Sup.Ct.1996); Doe v. Poritz, 142 N.J. 1, 662 A.2d 367 (1995); State v. Costello, 138 N.H. 587, 643 A.2d 531, 533 (1994); State v. Noble, 171 Ariz. 171, 829 P.2d 1217, 1224 (1992); Kitze v. Commonwealth, 23 Va.App. 213, 475 S.E.2d 830 (1996); State v. Manning, 532 N.W.2d 244 (Minn.Ct.App.1995). But cf. Rowe v. Burton, 884 F. Supp. 1372 (D.Alaska 1994) (holding registration is punitive where dissemination of information will result). In another case addressing the ex post facto implications of the Act, a different district judge in the Western District of Washington has concluded that the registration provision of the Act is constitutional, and we agree. Doe v. Gregoire, 960 F. Supp. 1478, 1484 (W.D.Wash.1997).

Thus, we are satisfied that, as a matter of law, the registration provisions of the Act do not amount to punishment subject to the Ex Post Facto Clause.

We next turn to the notification portion of the Act. The statute authorizes "[p]ublic agencies ... to release relevant and necessary information regarding sex offenders to the public when the release of the information is necessary for public protection." Wash. Rev.Code § 4.24.550(1). The Washington Supreme Court construed this provision as requiring a public agency to "have some evidence of an offender's future dangerousness, likelihood of reoffense, or threat to the community, to justify disclosure to the public in a given case. This statutory limit ensures that disclosure occurs to prevent future harm, not to punish past offenses." State v. Ward, 869 P.2d at 1070. The court also found an implied geographic limitation on the dissemination of the information. Id. at 1070-71. Notification provides the public with the offender's name, picture, age, date of birth, facts regarding the offender's convictions, and the general vicinity of the offender's domicile. The notification form also contains a caveat and a statement that harassment "will not be tolerated."

The district court considered the overall design of the notification regime, including the additional requirements imposed by the Washington Supreme Court, and concluded that notification is not punitive.

* As we did in the analysis of the registration provision, in applying the "intent-effects" test, we look first to the language of the notification provision of the statute.

The legislature finds that sex offenders pose a high risk of engaging in sex offenses even after being released from incarceration or commitment and that protection of the public from sex offenders is a paramount governmental interest. The legislature further finds that the penal and mental health components of our justice system are largely hidden from public view and that lack of information from either may result in failure of both systems to meet this paramount concern of public safety. Overly restrictive confidentiality and liability laws governing the release of information about sexual predators have reduced willingness to release information that could be appropriately released under the public disclosure laws, and have increased risks to public safety. Persons found to have committed a sex offense have a reduced expectation of privacy because of the public's interest in public safety and in the effective operation of government. Release of information about sexual predators to public agencies and under limited circumstances, the general public, will further the governmental interests of public safety and public scrutiny of the criminal and mental health systems so long as the information released is rationally related to the furtherance of those goals.

Therefore, this state's policy as expressed in section 117 of this act is to require the exchange of relevant information about sexual predators among public agencies and officials and to authorize the release of necessary and relevant information about sexual predators to members of the general public.

1990 Wash. Laws, ch. 3, § 116.

The Washington Supreme Court read this pronouncement as "a clear regulatory intent to limit the exchange of relevant information to the general public to those circumstances which present a threat to public safety." State v. Ward, 869 P.2d at 1070. Russell and Stearns do not contend that the legislature intended the notification provisions to be punitive; instead, they argue that the law was enacted in an "hasty and impassioned response to public outcry" after a young boy was attacked by a sex offender. The Ex Post Facto Clause was designed to prevent exactly this sort of legislative action, they say.

The language of section 116 makes clear that the legislature intended the notification provision to prevent future attacks by recidivist sex offenders, and that the law may have a deterrent purpose as well as a remedial one. Neither of these purposes would result in an ex post facto violation, however. There is no indication that the legislature intended to punish already-convicted offenders (rather than merely deterring them or preventing future crimes). However quickly a law was passed, and however heated the public sentiment around it, we look to the legislature's manifest intent-which is found in the text and structure of the law. See Hendricks, --- U.S. at ----, 117 S. Ct. at 2082.

The text and structure of the notification provisions reveal no intent to punish, but rather a regulatory purpose. Notification occurs only "when the release of the information is necessary for public protection." Wash. Rev.Code § 4.24.550(1). The notification generally must occur two weeks in advance of the offender's release from prison in order to allow "communities to meet with law enforcement to discuss and prepare for the release, to establish block watches, to obtain information about the rights and responsibilities of the community and the offender, and to provide education and counseling to their children." 1994 Wash. Laws, ch. 129, § 1. Notification is thus designed to avoid a hasty or retaliatory response from the community. The law is tailored to help the community protect itself from sexual predators under the guidance of law enforcement, not to punish sex offenders.9

Any remaining doubt about the purpose of the law is dispelled by the construction given it by the Washington Supreme Court, which we must regard as authoritative. The law contains careful safeguards to prevent notification in cases where it is not warranted and to avoid dissemination of the information beyond the area where it is likely to have the intended remedial effect. State v. Ward, 123 Wash.2d 488, 869 P.2d 1062, 1070-71 (1994). Only information "relevant to and necessary for counteracting the offender's dangerousness" is disclosed, and always accompanied by a warning against violence toward the offender. Id. A law designed to punish an offender would not contain these strict limitations on notification.

As to the second part of the "intent-effects" test, the question again is whether there is "the clearest proof" that the notification provision is so punitive in effect as to overcome the nonpunitive legislative intent. We consider the relevant Mendoza-Martinez factors, including the historical use of notification provisions, in this part of the test.

As Ursery noted, the most significant question under this stage of the analysis is whether the law, "while perhaps having certain punitive aspects, serve[s] important nonpunitive goals." Ursery, --- U.S. at ----, 116 S. Ct. at 2148. Russell and Stearns argue that notification serves the goal of deterrence, which is a traditional goal of punishment. This overlooks the statements in Ursery and Bennis v. Michigan that a deterrent purpose can serve both civil as well as criminal goals. Id. at ----, 116 S.Ct. at 2149; Bennis v. Michigan, --- U.S. ----, ----, 116 S. Ct. 994, 1000, 134 L. Ed. 2d 68 (1996). That a sanction has a deterrent purpose does not make it punitive. Tort law, for example, attempts to deter certain conduct but neither imposes punishment nor is criminal in nature. See Bennis, --- U.S. at ----, 116 S. Ct. at 1000.

Although the Act may implicate deterrence, it does not implicate the other primary objective of criminal punishment, retribution, because it neither labels the offender as more culpable than before (though his or her culpability may be more widely publicized), nor does it turn on a finding of scienter. Some persons who have not been convicted of a sex offense may be subject to notification-those incompetent to stand trial, or committed as sexual psychopaths or sexually violent predators, for example. See Wash. Rev.Code § 4.24.550(3). These are three of the hallmarks that distinguish retributive sanctions from other sanctions. See Hendricks, --- U.S. at ----, 117 S. Ct. at 2082 (discussing retributive sanction as that which affixes culpability for prior criminal conduct, turns on a finding of scienter, and is triggered by a criminal conviction). Instead, the Washington notification provision has both strong remedial aspects and serves important nonpunitive goals: alerting the community to the presence of sexual predators adjudged likely to offend again, and giving guidance to the community to allow it to avert new and tragic sexual offenses.

Next, Russell and Stearns strenuously argue that many historical punishments depended on the punitive effect of public notification-the "sting" of shame and humiliation. They claim that public notification is punishment because the punishments of the past-pillorying, branding, or the sanction described in Hawthorne's The Scarlet Letter-notified the community of the offender's misdeeds.

Although historical punishments did notify the community and humiliate the offender, an adequate historical analysis is not that simple. Unlike the civil forfeiture provisions at issue in Ursery and Austin v. United States, 509 U.S. 602, 113 S. Ct. 2801, 125 L. Ed. 2d 488 (1993), the notification provisions of the Act do not have identical historical antecedents. See W.P. v. Poritz, 931 F. Supp. 1199, 1215 (D.N.J.1996). History does not tell us whether this sort of notification ought to be regarded as punishment. At best, we can draw an analogy between the Act and the punishments of yesteryear. That analogy is not unassailable, however. Historical shaming punishments like whipping, pillory, and branding generally required the physical participation of the offender, and typically required a direct confrontation between the offender and members of the public. As the Third Circuit recently stated: "Public shaming, humiliation and banishment all involve more than the dissemination of information.... [T]hese colonial practices inflicted punishment because they either physically held the person up before his or her fellow citizens for shaming or physically removed him or her from the community." E.B. v. Verniero, 119 F.3d 1077, 1099-1100 (3rd Cir.1997). Put another way, "the potential ostracism and opprobrium that may result from [notification] is not inevitable, as it was with the person whipped, pilloried or branded in public." Poritz, 931 F. Supp. at 1217. More importantly, the Washington law is not intended to be punitive-it has protective purposes-while shaming punishments "were intended to and did visit society's wrath directly upon the offender." Id.

Other historical analogies are also instructive. It is at least as appropriate to compare the notification law to "wanted" posters and warnings about escaped prisoners or other dangerous persons-practices that have not been regarded as punishment, though they disclose essentially the same information, may rouse public excitement, and may carry a greater risk of vigilantism.

This discussion reveals the perils of using historical parallels to determine whether a sanction is punishment. True, punishment in the past often relied on humiliation. But humiliation alone does not constitute punishment. A law imposing punishment has other ingredients-most importantly, an intent to punish. See Ursery, --- U.S. at ----, 116 S. Ct. at 2147. As the Supreme Court has repeatedly stated, a statute does not punish merely because it works a detriment. Morales, 514 U.S. at 506 n. 3, 115 S. Ct. at 1602 n. 3; see Flemming v. Nestor, 363 U.S. 603, 616, 80 S. Ct. 1367, 1375-76, 4 L. Ed. 2d 1435 (1960) (noting that "often-severe effects" of a regulation do not make it punishment). Plastering "wanted" posters all over town works a detriment to the person named in them, but even if he or she is innocent, there has been no punishment. See Paul v. Davis, 424 U.S. 693, 96 S. Ct. 1155, 47 L. Ed. 2d 405 (1976) (holding that man wrongly identified in police flier as active shoplifter not deprived of liberty or property interest).

We are not persuaded that the analogy to historical shaming punishments is strong enough to overcome the law's nonpunitive intent. Without an identical historical antecedent, and with other persuasive analogies to nonpunitive traditional practices, we cannot conclude that there is "clear proof" of an overwhelming punitive effect.

Next, Russell, Stearns and amicus curiae American Civil Liberties Union object that public notification imposes an affirmative disability or restraint because notification has a devastating effect on the offenders' personal and professional lives. Similarly, they claim that notification leads to excessively harsh results, including threats, ostracism, harassment, and vigilantism.

Notification may well subject offenders to humiliation, public opprobrium, ostracism, and the loss of job opportunities. We actively weigh these detriments to offenders. But our inquiry into the law's effects cannot consider the possible "vigilante" or illegal responses of citizens to notification. Such responses are expressly discouraged in the notification itself and will be prosecuted by the state. See Poritz, 931 F. Supp. at 1212. Indeed, courts must presume that law enforcement will obey the law and will protect offenders from vigilantism. See Artway, 81 F.3d at 1267.

We conclude that, considering the entire range of possible community responses not prohibited by Washington law, the Act's effect is not so egregious as to prevent us from viewing the Act as regulatory or remedial. In doing so, we are sensitive to the fact that the Act may have a lasting and painful impact on a sex offender's life, which ought not be lightly disregarded. Yet, we cannot say that the Act violates the Constitution. Other regulatory sanctions have had harsh effects similar to the community's likely response and have been upheld. See De Veau v. Braisted, 363 U.S. 144, 80 S. Ct. 1146, 4 L. Ed. 2d 1109 (1960) (forbidding work as a union official); Hawker v. People of New York, 170 U.S. 189, 18 S. Ct. 573, 42 L. Ed. 1002 (1898) (revocation of a medical license); Mahler v. Eby, 264 U.S. 32, 44 S. Ct. 283, 68 L. Ed. 549 (1924) (deportation); Flemming v. Nestor, 363 U.S. 603, 80 S. Ct. 1367, 4 L. Ed. 2d 1435 (1960) (termination of Social Security benefits). Moreover, "whether a sanction constitutes punishment is not determined from the defendant's perspective, as even remedial sanctions carry the 'sting of punishment.' " Department of Revenue v. Kurth Ranch, 511 U.S. 767, 777 n. 14, 114 S. Ct. 1937, 1945 n. 14, 128 L. Ed. 2d 767.10

In sum, we conclude that the notification provisions were intended to be regulatory and not punitive. Weighing all of the considerations discussed above, and with an eye to the remaining Mendoza-Martinez factors, we hold that, as a matter of law, the possible effects of the notification provision are not so punitive in fact as to prevent us from legitimately viewing the Act as regulatory in nature. Even less do the possible effects amount to "the clearest proof" of a punitive effect sufficient to overcome the legislature's nonpunitive intent. This is especially so given the strong remedial goals of the notification provision. The notification provisions of the Act do not amount to punishment subject to the Ex Post Facto Clause. See Doe v. Pataki, 120 F.3d 1263, 1265-66 (2d Cir.1997); E.B., 119 F.3d at 1080-81; Doe v. Kelley, 961 F. Supp. 1105 (W.D.Mich.1997); Doe v. Weld, 954 F. Supp. 425 (D.Mass.1996); W.P. v. Poritz, 931 F. Supp. 1199 (D.N.J.1996); People v. Afrika, 168 Misc. 2d 618, 648 N.Y.S.2d 235 (N.Y.Sup.Ct.1996); Opinion of the Justices to the Senate, 423 Mass. 1201, 668 N.E.2d 738 (1996). But see Roe v. Office of Adult Probation, 938 F. Supp. 1080 (D.Conn.1996); State v. Myers, 260 Kan. 669, 923 P.2d 1024 (1996).11

Russell and Stearns contend that the Act is constitutionally infirm because the accumulation and dissemination of information about them violates their right to privacy. They do not pinpoint the source of the right or identify its contours, however, and they fail to explain precisely how the Act violates it beyond collating and releasing information. For support, they cite only two cases, Whalen v. Roe, 429 U.S. 589, 97 S. Ct. 869, 51 L. Ed. 2d 64 (1977), and Nixon v. Adm'r of Gen. Servs., 433 U.S. 425, 97 S. Ct. 2777, 53 L. Ed. 2d 867 (1977), which they claim stand for the proposition that the mere collection of private information may constitute a violation of a constitutional right to privacy.

Russell and Stearns are mistaken about the import of Whalen and Nixon. Neither established a general constitutional right to privacy in information collected in a database; instead, both cases noted that, if there was such a right, it was not violated by the provisions at issue.

In Whalen, the Supreme Court upheld a New York statute that created a centralized computer file of the names and addresses of all persons who obtained certain prescription drugs. In doing so, the Court stated that, at most, the collection and use of such information by the government is accompanied only in some circumstances by an arguable constitutional duty to avoid unwarranted disclosures. Whalen, 429 U.S. at 598-604, 605-606, 97 S. Ct. at 875-79, 879-80. This is a far cry from the supposed right Russell and Stearns assert.

Similarly, Nixon upheld the Presidential Recordings and Materials Preservation Act, which provided for temporary public custody of Presidential tapes and papers, against a right to privacy challenge by former President Nixon. In that case, the Court restated Whalen 's dictum that "[o]ne element of privacy has been characterized as 'the individual interest in avoiding disclosure of personal matters,' " Nixon, 433 U.S. at 457, 97 S.Ct. at 2797 (quoting Whalen, 429 U.S. at 599, 97 S.Ct. at 876), but concluded that, to the extent there is such a right, it is not violated if the statute contains safeguards against "undue dissemination". Id. at 458, 97 S.Ct. at 2797-98.

In this case, the collection and dissemination of information is carefully designed and narrowly limited. Even if Whalen and Nixon had established a broad right to privacy in data compilations, the Act does not unduly disseminate private information about Russell and Stearns.

Moreover, any such right to privacy, to the extent it exists at all, would protect only personal information. Whalen v. Roe, 429 U.S. 589, 599, 97 S. Ct. 869, 876, 51 L. Ed. 2d 64 (1977). The information collected and disseminated by the Washington statute is already fully available to the public and is not constitutionally protected, see Doe v. New York, 15 F.3d 264, 268 (2d Cir.1994), with the exception of the general vicinity of the offender's residence (which is published) and the offender's employer (which is collected but not released to the public). Neither of these two items are generally considered "private." Johnson v. Sawyer, 47 F.3d 716, 732-33 (5th Cir.1995) (en banc) (discussing common law invasion of privacy).

Likewise, the Supreme Court in Paul v. Davis, 424 U.S. 693, 96 S. Ct. 1155, 47 L. Ed. 2d 405 (1976), held that damage to one's reputation by a state actor does not violate a liberty or property interest "apart from some more tangible interests such as employment." Id. at 701, 96 S.Ct. at 1161. The collection and dissemination of information under the Washington law does not violate any protected privacy interest, and does not amount to a deprivation of liberty or property.

In sum, Russell's and Stearns's privacy claims are fatally defective as a matter of law, and must fail.

Russell and Stearns make a final argument based on the Due Process Clause. They claim that because the Washington statute violates their privacy rights, it deprives them of a liberty interest without giving them notice and an opportunity to be heard. Since we have already rejected their privacy claims, we conclude that they have no liberty interest at stake, and hence we reject their due process claims.

As a matter of law, the Act does not violate the Ex Post Facto Clause, the right to privacy, or the Due Process Clause. Russell and Stearns therefore have no likelihood of success on the merits of their ex post facto, privacy, and due process claims, and they are not entitled to a preliminary injunction. The decision of the district court denying the motion for a preliminary injunction was not an abuse of discretion.

AFFIRMED.