On July 29, 1994, Megan Kanka, a seven year old child, was abducted, raped, and murdered near her home. The man who confessed to Megan's murder lived in a house across the street from the Kanka family and had twice been convicted of sex offenses involving young girls. Megan, her parents, local police, and the members of the community were unaware of the accused murderer's history; nor did they know that he shared his house with two other men who had been convicted of sex offenses.

By October 31, 1994, New Jersey had enacted the Registration and Community Notification Laws, Pub.L.1994, Chs. 128, 133 (codified at N.J.S.A. 2C:7-1 to 7-11) as part of a ten-bill package collectively referred to as "Megan's Law." This legislation required registration by those who had committed certain designated crimes involving sexual assault and provided for the dissemination of information about those required to register. Other states followed suit with their own versions of Megan's Law and Congress passed a statute requiring a state program of registration and notification as a condition of receiving certain federal funds. By May of 1996, forty-nine states had adopted sex offender registration laws and thirty-two states maintained some form of community notification program.

We have before us challenges to the constitutionality of the notification requirements of New Jersey's Megan's Law based on the Ex Post Facto, Double Jeopardy, and Due Process Clauses of the United States Constitution. The issues before us are difficult but relatively narrow. We are not called upon to decide whether Megan's Law can constitutionally be applied to one who has committed one of the designated sex crimes after its enactment. Nor, of course, is it our responsibility to determine whether the policy judgments reflected in Megan's Law are prudent ones.

We hold that (1) the notification requirements of Megan's Law do not constitute state inflicted "punishment" on Tier 2 and Tier 3 registrants for purposes of the Ex Post Facto and Double Jeopardy Clauses; (2) the Due Process Clause of the United States Constitution forecloses New Jersey from placing the burden of persuasion on the registrant in a proceeding challenging a Tier 2 or Tier 3 classification and notification plan; and (3) the Due Process Clause requires the state at such a proceeding to shoulder the burden of justifying the classification and notification plan by clear and convincing evidence.

Public reaction to Megan's murder was intense, and New Jersey's governor and legislature responded quickly. By August 15, 1994, two weeks after the discovery of Megan's body, bills providing for registration and community notification had been introduced in the General Assembly. Two weeks later, the General Assembly declared the bills an "emergency," allowing them to bypass committee and be passed the same day.

In the Senate, no registration or notification bills had been introduced as of August 29, 1994. However, the Law and Public Safety Committee held a hearing upon pending legislation that pre-dated Megan's Law and would have required victim notification on the release of offenders. In connection with its consideration of that legislation, the Committee received testimony and/or written reports from, inter alia, the American Civil Liberties Union, municipal officials, inmates, state and federal legislators, and the Attorney General on issues related to sex offender registration and community notification. Registration and community notification bills identical to their General Assembly counterparts were introduced in the Senate on September 12, 1994. After hearing testimony from the ACLU, the New Jersey Coalition of Crime Victims, and corrections officials on September 26, 1994, the Senate Law and Public Safety Committee revised the bills by: (1) supplementing the list of crimes which require registration,1 (2) directing the Attorney General to consult with a twelve-member Advisory Council of experts to establish guidelines concerning the risk of reoffense, (3) identifying certain factors material to the determination of risk of reoffense, and (4) narrowing the scope of community notification. The Committee then favorably reported the amended versions to the Senate, see Senate Law & Pub. Safety Comm., Statement to Substitute for Senate Bill No. 14 & Assembly Bill No. 85 (N.J. Sept. 26, 1994), which approved the bills on October 3. The General Assembly followed suit by debating and approving the revised bill on October 20, 1994, and Governor Whitman signed it into law on October 31, 1994.

Megan's Law establishes both a registration requirement and a three-tiered notification program. See Artway v. Attorney General, 81 F.3d 1235, 1243 (3d Cir.1996). The registration provisions were the subject of this court's decision in Artway, where we upheld their constitutionality in the face of ex post facto, double jeopardy, bill of attainder, due process, equal protection, and vagueness challenges. We there summarized the operation of the registration provision:

The registration provision requires all persons who complete a sentence for certain designated crimes involving sexual assault after Megan's Law was enacted to register with local law enforcement. N.J.S.A. 2C:7-2b(1). Those committing these offenses and completing all incarceration, probation, and parole before the Law's enactment must register only if, at the time of sentencing, their conduct was found to be "characterized by a pattern of repetitive and compulsive behavior." Id.

The registrant must provide the following information to the chief law enforcement officer of the municipality in which he resides: name, social security number, age, race, sex, date of birth, height, weight, hair and eye color, address of legal residence, address of any current temporary legal residence, and date and place of employment. N.J.S.A. 2C:7-4b(1). He must confirm his address every ninety days, notify the municipal law enforcement agency if he moves, and re-register with the law enforcement agency of any new municipality. N.J.S.A. 2C:7-2d to e.

The registration agency then forwards the registrant's information, as well as any additional information it may have, to the prosecutor of the county that prosecuted the registrant. N.J.S.A. 2C:7-4c to d. The prosecutor, in turn, forwards the information to the Division of State Police, which incorporates it into a central registry and notifies the prosecutor of the county in which the registrant plans to reside. Id. This information is available to law enforcement agencies of New Jersey, other states, and the United States. N.J.S.A. 2C:7-5. The registration information is not open to public inspection.... Failure of the sex offender to comply with registration is a fourth-degree crime. [N.J.S.A. 2C:7-2a.]

81 F.3d at 1243. The registration requirement persists for a period of 15 years from the date of conviction or the date of release from a correctional facility, whichever is later. It is only after this 15 year period that a registrant may make application to the Superior Court to terminate the obligation to register. The obligation may be terminated only upon a persuasive showing that the registrant is not likely to pose a threat to the safety of others. N.J.S.A. 2C:7-2f.

The registration information provides a basis for the next step--notification. The prosecutor of the county where the sex offender intends to reside and the prosecutor from the county of conviction use the registration information and other data to jointly assess the risk of reoffense by the registered individual. N.J.S.A. 2C:7-8d(1). They determine whether the sex offender poses a low (Tier 1), moderate (Tier 2), or high (Tier 3) reoffense risk. N.J.S.A. 2C:7-8c. Every registrant at least qualifies for Tier 1 treatment, otherwise known as "law enforcement alert," where notification extends only to law enforcement agencies likely to encounter the registrant. N.J.S.A. 2C:7-8c(1). In the case of those registrants posing a moderate risk of reoffense, Tier 2 notification, or "law enforcement, school and community organization alert," issues to registered schools, day care centers, summer camps, and other community organizations which care for children or provide support to women and where individuals are likely to encounter the sex offender. N.J.S.A. 2C:7-8c(2). The high risk registrants merit Tier 3's "community notification," where members of the public likely to encounter the registrant are notified. N.J.S.A. 2C:7-8c(3).

In order to preserve uniformity in the tier classification and notification process, the state Attorney General, in consultation with an advisory council, is required to develop and promulgate guidelines to be consulted by prosecutors in assessing the degree of risk of reoffense. N.J.S.A. 2C:7-8a, d. By statute, the guidelines are required to include the following considerations:

(1) Conditions of release that minimize risk of reoffense, including but not limited to whether the offender is under supervision of probation or parole; receiving counseling, therapy or treatment; or residing in a home situation that provides guidance and supervision;

(2) Physical conditions that minimize risk of re-offense, including but not limited to advanced age or debilitating illness;

(3) Criminal history factors indicative of high risk of reoffense, including:

(a) Whether the offender's conduct was found to be characterized by repetitive and compulsive behavior;

(b) Whether the offender served the maximum term;

(c) Whether the offender committed the sex offense against a child;

(4) Other criminal history factors to be considered in determining risk, including:

(a) The relationship between the offender and the victim;

(b) Whether the offense involved the use of a weapon, violence, or infliction of serious bodily injury;

(c) The number, date and nature of prior offenses;

(5) Whether psychological or psychiatric profiles indicate a risk of recidivism;

(6) The offender's response to treatment;

(7) Recent behavior, including behavior while confined or while under supervision in the community as well as behavior in the community following service of sentence; and

(8) Recent threats against persons or expressions of intent to commit additional crimes.

N.J.S.A. 2C:7-8b.

Pursuant to this statutory delegation of authority, the Attorney General has developed guidelines for law enforcement for classification and notification. See Guidelines for Law Enforcement for Notification to Local Officials and/or the Community of the Entry of a Sex Offender into the Community, June 1, 1996 ("Guidelines"). The Attorney General's Guidelines require the prosecutors to use the Registrant Risk Assessment Scale (the "Scale"), a numerical scoring system designed with the assistance of mental health and law enforcement professionals, to evaluate the degree of risk of the sex offender. See Registrant Risk Assessment Scale Manual, Oct. 3, 1995 ("Manual"). The New Jersey Supreme Court has said of the creation of the Scale:

A Committee of mental health professionals and legal experts ... developed the Scale. They examined risk assessment scales being used in the United States and Canada. After reviewing the scientific literature, the Committee selected for inclusion in the Scale those factors that met two conditions. First, all of the factors selected had to be empirically supported in the risk assessment field as criteria positively related to the risk of re-offense. Second, all of the factors selected had to be fairly concrete criteria that could be gathered in a consistent and reliable manner.

In re C.A., 146 N.J. 71, 679 A.2d 1153, 1169 (1996).

The Scale itself is a matrix with thirteen factors grouped into four general categories: (1) Seriousness of Offense; (2) Offense History; (3) Characteristics of Offender; and (4) Community Support. See Artway, 81 F.3d at 1244.2 Guided by the promulgated examples and commentary, the prosecutors determine whether the registrant poses a low, moderate, or high risk to the community under each of the factors and assign zero, one, or three points, respectively, for each factor. Then the prosecutors multiply these raw scores by a coefficient, reflective of the relative weight attributed to the various general categories by the creators of the Scale; raw scores for factors under Seriousness of Offense are multiplied by five, under Offense History by three, under Characteristics of Offender by two, and under Community Support by one. Prosecutors total the resulting amounts and place the registrant in the appropriate tier: Tier 1, low risk--0 to 36 points; Tier 2, moderate risk--37 to 73 points; and Tier 3, high risk--74 to 111 points. Finally, the prosecutors consider the applicability of two exceptions:

1) If an offender has indicated that he will reoffend if released into the community and the available record reveals credible evidence to support this finding, then the offender will be deemed to be a high risk of reoffense regardless of the weighting procedure; and 2) if the offender demonstrates a physical condition that minimizes the risk of reoffense, including but not limited to advanced age or debilitating illness, then the offender will be deemed to be a low risk of reoffense regardless of the outcome of the weighting procedure.

Manual at 1; see Artway, 81 F.3d at 1244.

While the class of those who receive notification differs depending on a registrant's classification, the type of information distributed is the same regardless of the classification. The package of information provided includes the registrant's name, a recent photograph, a physical description, the offense of conviction, home address, place of employment or schooling, and a vehicle description and license plate number. "Those notified under Tier 2 are informed that the information is not to be shared with the general public, and every notification must contain a warning about the criminal consequences of vandalism, threats and assaults against the registrant or any of his associates." Artway, 81 F.3d at 1244.3D.

The New Jersey courts have played an active role in refining and developing the Megan's Law scheme. See In re G.B., 147 N.J. 62, 685 A.2d 1252 (1996); In re C.A., 146 N.J. 71, 679 A.2d 1153 (1996); Doe v. Poritz, 142 N.J. 1, 662 A.2d 367 (1995). In Doe, the New Jersey Supreme Court upheld the constitutionality of Megan's Law and read into the statute and Guidelines certain additional procedures designed to prevent any "excessiveness of community notification." 662 A.2d at 381. First, the Court added the "likely to encounter" the registrant restriction to Tier 2 notification. Id.4 As a result of the Doe decision, a prosecutor who has classified a registrant in Tier 2 must make an "individual determination" concerning the appropriate institutions and organizations to include in the notification program he creates. Id. As articulated in the Guidelines, "[t]he decision as to which groups should appropriately be notified should be made on a case-by-case basis, following careful review." Guidelines at 11. There is no "automatic inclusion of an organization simply because it is 'registered' " with the local law enforcement agencies; rather, "likely to encounter" requires "having a fair chance to encounter" the registrant. Doe, 662 A.2d at 385. The Guidelines interpret the Court's articulations to mean that the types of interactions which occur at the location and their attendant circumstances must demonstrate that contact with the offender is "reasonably certain." Guidelines at 6-7. They provide, for example, that if a registrant regularly stops at a gas station merely to refuel, there would not be a "fair chance to encounter" him there. Id. at 7.

Ordinarily, the "critical" factor for " 'likely to encounter' is geography--how close is the institution or organization, in the case of Tier Two notification, to the offender's residence or place of work or school." Doe, 662 A.2d at 385. However, the New Jersey Supreme Court explained:

In some municipalities, not every institution or organization that would otherwise qualify for notification may be close enough to warrant same, but in some cases, ... institutions or organizations in other municipalities may be close enough. The same observations can be made for Tier Three notification. We do not attempt to define the area around the offender's residence or place of work or school that may be included within the notification process, and assume it may differ from one locale to another. Depending upon the particular offender, factors other than geography may be considered if they are relevant to the offender's likely whereabouts, such as an offender's proclivity for certain locations, and geographic considerations may be affected by the nature of the offender's characteristics and the institution in question, e.g., a repetitive and compulsive pedophile and a large elementary school.

Id. 662 A.2d at 385-86.

Moreover, the Guidelines provide that notification must be appropriately tailored to reach those members of the public who are at risk from the particular offender. The tailoring must include consideration of the relationship between the registrant and his prior victims. As the Guidelines suggest, sex offenders who have only victimized members of their own households may not pose a threat to most members of the community, and those that have targeted adult women may be of little risk to children; thus, the prosecutor may appropriately limit notification as all registered community organizations are not "likely to encounter" the offenders in either example.

Doe also added to the Megan's Law scheme a requirement that the prosecutor provide the registrant with notice of a Tier 2 or Tier 3 classification and the proposed notification plan. Id. 662 A.2d at 382. The Court insisted that the written notice describe the manner and details of the notification plan and inform the registrant of his rights to retain counsel and to challenge the prosecutor's decisions. However, the Court "realize[d] that in some cases it may be impossible as a practical matter to give such notice, or to give it timely, and in those cases it may be dispensed with." Id. The Guidelines elaborate on dispensing with the notice requirement:

[I]f a Prosecutor['s] Office does not receive notification of release of a person determined to be a Tier 3 offender until after the date of release, then, in order to protect the public, notice to the offender may be dispensed with. The Prosecutor's Office may apply to the designated judge for an order allowing notification to take place without service of notice to the offender, upon receipt of the judge's order. This may occur, for example, when an offender who has been civilly committed is released on short notice by a judge.

Also, cases will arise where registrants will avoid service of the notice. In those cases, the Prosecutor's Office may apply to the designated judge for an order allowing notification to take place without service of notice to the offender, where the Prosecutor can demonstrate that every good faith effort was made within the allotted timeframe to serve the registrant. If service has not been completed within 3 days of the date that the tier decision is made, then the Prosecutor may apply to the court for the order allowing notification to occur without notice to the registrant.

Guidelines at 17-18.

Finally, Doe required the state to make available a pre-notification judicial review process for sex offenders who wish to contest their classification or the notification plan. 662 A.2d at 382. The registrant bears the burden of persuasion in these summary, in camera proceedings, where the court decides only whether to affirm or reverse the prosecutor's determination. Thus, where the state has met its burden of presenting evidence that "prima facie justifies the proposed level and manner of notification," the court will affirm the prosecutor's determination "unless it is persuaded by a preponderance of the evidence that it does not conform to the laws and Guidelines." Id. 662 A.2d at 383. The "only issue for the court on the Tier level of notification is the risk of reoffense;" review of the notification plan largely involves interpretation and application of the "mandatory" limits on notification, such as the "likely to encounter" standard, articulated in the Doe opinion. Id. 662 A.2d at 383-84. Still, the courts are to understand that "the Scale is merely a tool," In re G.B., 685 A.2d at 1261, and they are cautioned not to "blindly follow the numerical calculations" but to make a "case-by-case" determination regarding tier classification and scope of notification. In re C.A., 679 A.2d at 1171-72.

The New Jersey Supreme Court has recognized that "a registrant is entitled to lodge three distinct challenges to his tier designation":

First, a registrant may introduce evidence that the calculation that led to the Scale score was incorrectly performed either because of a factual error, because the registrant disputes a prior offense, because the variable factors were improperly determined, or for similar reasons. Second, a registrant may introduce evidence at the hearing that the Scale calculations do not properly encapsulate his specific case; or phrased differently, a registrant may maintain that his case falls outside the "heartland" of cases and, therefore, that he deserves to be placed in a tier other than that called for by the prosecutor's Scale score. Finally, a registrant may introduce evidence that the extent of notification called for by his tier categorization is excessive because of unique aspects of his case. Challenges to the Scale itself, or challenges to the weight afforded to any of the individual factors that comprise the Scale, are not permitted. Instead, all challenges must relate to the characteristics of the individual registrant and the shortcomings of the Scale in his particular case.

In re G.B., 685 A.2d at 1264.

The registrant's hearing "is civil, not criminal, and remedial, not adversarial." In re C.A., 679 A.2d at 1164. It follows the "format ... for probation violation hearings" in New Jersey. Id. 679 A.2d at 1166. The court possesses broad discretion over whether and to what extent witnesses and cross examination will be allowed. Doe, 662 A.2d at 382-83. Rules of evidence do not apply, and the court may rely on documentary evidence, such as expert opinions, for all issues. Id. 662 A.2d at 383. Reliable hearsay is admissible. In re C.A., 679 A.2d at 1165. Moreover, "non-conviction offenses [i.e., criminal activities that have not been the subject of a conviction] are to be considered in evaluating a registrant's risk of re-offense, provided there is sufficient evidence that the offense occurred." Id. 679 A.2d at 1162.

Where the proof, whether in the form of reliable hearsay, affidavits, or offers of live testimony, creates a genuine issue of material fact that the tier designation or manner of notification is inappropriate, "then the trial court should convene a fact-finding hearing and permit live testimony." Id. 679 A.2d at 1166. Both sides may use expert testimony, but the proceedings are not to be converted into "long drawn-out contests between experts." Doe, 662 A.2d at 384. Thus, courts must permit registrants to introduce expert testimony which tends to establish that the Scale does not properly account for aspects of the registrant's character or prior offense, where those aspects are relevant and material to the tier classification, and, in the court's opinion, would assist in the disposition of the case. In re G.B., 685 A.2d at 1265-66.

In Artway, we sustained the constitutionality of the provisions of Megan's Law requiring registration and Tier 1 notification. We declined, however, to address the accompanying constitutional challenge to the provisions requiring the broader notification authorized for Tier 2 and Tier 3 classifications. We found that challenge unripe in large part because the plaintiff there had not been classified and had not received a notification plan. We also noted that the record there lacked evidence of the effects of notification on the community. 81 F.3d at 1250. For purposes of the ensuing discussion, we will follow the convention established in our Artway opinion, whereby "registration" includes Tier 1 notification and "notification" refers to Tier 2 and Tier 3 notification. Id., 81 F.3d at 1244.

We have two actions before us: E.B. v. Verniero, and W.P. v. Verniero. They involve identical challenges to Megan's Law; each alleges that notification violates ex post facto, double jeopardy, and procedural due process protections conferred by the United States Constitution. The plaintiffs in both actions are sex offenders who were convicted of their offenses prior to the enactment of Megan's Law. The plaintiff in the individual action, E.B., comes within the broad language defining the class certified in W.P., constituting:

All persons required to register as a sex offender[sic ] pursuant to N.J.S.A. 2C:7-1 et seq. and whose offenses were committed prior to October 31, 1994, the effective date of the New Jersey Registration and Community Notification Laws, and who have been or will be classified as a tier II or tier III offender.

W.P. v. Poritz, 931 F.Supp. 1187, 1192 (D.N.J.1996). The defendants in E.B. are the Attorney General, the local county prosecutor, and the police chief, while in W.P. they are the Attorney General and various county prosecutors.

In 1974, E.B. pled guilty in New Jersey Superior Court to three offenses of sexual abuse against young boys and received a thirty-three-year sentence. Two years later, he pled guilty in the Circuit Court in Petersburg, Virginia, to two separate murders and was sentenced to concurrent terms of twenty years of incarceration in that state to run consecutive to the New Jersey sentence. In 1979, after serving less than six years of his thirty-three-year New Jersey sentence, E.B. was paroled and extradited to Virginia to serve the murder sentences. On June 15, 1989, E.B. was paroled by Virginia. He is now free, subject to supervised release by the New Jersey Bureau of Parole until July 23, 2006.

Pursuant to Megan's law, E.B. registered with the authorities in Englewood, New Jersey. On October 24, 1995, the Bergen County Prosecutor's Office notified E.B. that he was classified as a Tier 3 sex offender and proposed to issue notification to "all public and private educational institutions and organizations within a one-half mile radius of the Plaintiff's home, and all parties who resided or worked within a one block radius of the Plaintiff's home." E.B. Complaint at p 13. Upon E.B.'s objection to the classification and notification, a hearing was held in New Jersey Superior Court, Law Division. On December 18, 1995, the court ruled that the classification was appropriate and permitted notification to: (1) 82 public and private educational institutions, licensed day care centers and summer camps in Englewood, Teaneck, Bergenfield, Tenafly, Englewood Cliffs, Leonia and Fort Lee, and (2) all residences within a one block radius of E.B.'s house. E.B.'s appeals to the Appellate Division and the State Supreme Court were unsuccessful, but notification remained stayed by court order during the pendency of the proceedings.

E.B. then filed his federal action. The district court entered a preliminary injunction, enjoining the defendants from implementing notification. E.B. v. Poritz, 914 F.Supp. 85 (D.N.J.1996). Defendants appeal from that order and a subsequent order denying their application for a stay of the preliminary injunction.

Seven plaintiffs filed the initial complaint in W.P. in January 1996. Two months later, when the court certified the class, there were 22 representative plaintiffs, all classified as either Tier 2 or Tier 3 and facing prosecutors' notification plans ranging in scope from notification of three schools to notification of all schools, day care centers, and registered community organizations in the city of Trenton, as well as all residents within a certain area of the city. Some of the representative plaintiffs had sought relief from a state court and were subject to the resulting state court orders. The district court promptly entered a preliminary injunction preventing notification for any of the class members. W.P. v. Poritz, 931 F.Supp. 1187 (D.N.J.1996).5

Thereafter, the court entered summary judgment for the defendants. W.P. v. Poritz, 931 F.Supp. 1199 (D.N.J.1996). Plaintiffs filed this appeal from the entry of summary judgment.

The record in these cases contains affidavits from registrants and state authorities, the Attorney General's publications concerning the Scale, registration and notification data, newspaper articles, and reports from other jurisdictions maintaining notification programs. The district court held the plaintiffs' constitutional claims were ripe for review, and no one has challenged that determination on appeal.

New Jersey's Administrative Office of the Courts reports that, as of May 6, 1996, there were 528 registrants designated as Tier 1; 585 as Tier 2; and 59 as Tier 3; or 45 percent, 50 percent, and 5 percent, respectively, of all classified registrants. According to the county prosecutors, as of May 16, 1996, notification was completed for 135 out of the 644 individuals classified to Tier 2 or Tier 3. Administrative Office information also indicates that of the 117 registrants who pursued their notification challenges to a resolution, 62 had their tier levels affirmed. Fifty-two challenges resulted in changed tier classifications and 13 resulted in modification of the scope of notification.

The record contains anecdotal evidence concerning the experiences of a total of at least nineteen sex offenders in New Jersey.6 In only six of these cases had state-compelled notification under Megan's Law been carried out. In the remaining cases, members of the community had received information about the sex offenders from sources other than a Megan's Law notification.7 In all the cases, the sentenced offender had experienced adverse repercussions. Loss of employment, eviction, and verbal abuse were not uncommon. Vandalism and threats were experienced but considerably less frequently. Two registrant affidavits speak of physical assaults following notification. One registrant reported being physically attacked on three separate occasions. In another case, a father and son broke into the registrant's residence and assaulted a house guest whom they mistook for the sex offender. Police arrived on the scene and arrested the assailants, who were later prosecuted and convicted for criminal trespass.

According to law enforcement records, the 135 cases in which Tier 2 and Tier 3 notifications have been completed have produced only a single instance of a physical assault being reported to the authorities--the father and son attack on the person mistaken for a registrant. In addition, there was a total of four reports to law enforcement personnel of threats, harassment, or other offensive actions. In Bergen County, one Tier 3 registrant contacted the local police department and reported that his mother's car had been vandalized. In Somerset County, a juvenile who registered under Megan's Law reported to police that a harassing note had been left on his car at school. In Sussex County, the prosecutor's office received a call from the wife of a Tier 3 registrant who reported that a threatening note had been mailed to the registrant's home. In Atlantic County, a Tier 2 registrant's employer reported that the local school had disclosed the employment of the registrant and a boycott was planned for the employer's restaurant. The police defused the situation by contacting the potential picketers and the school's principal, who agreed to speak to his staff concerning the confidentiality of information received through notification.

The record also includes information from the registration and notification experiences of other jurisdictions. A review of community notification in Washington state found that of the 176 notifications completed between March 1990 and March 1993, 14 incidents of harassment were reported, ranging in severity from multiple incidents of verbal abuse to a death threat and one assault.8

In addition, the record contains a January 1995 study by the Oregon Department of Corrections, undertaken to investigate the impact of the first 14 months of the state's 1993 community notification statute. Even before this statute, probation and parole officers with a sex offender under supervision had provided notification to "local police; immediate and extended family members in contact with the offender; victims; other residents in the offender's home; regular visitors to the home; employers; therapists; Children's Services Division; landlords and apartment managers; ministers, pastors, and other officials where the offender attends church; select neighbors; specific business[es] frequented by the offender; and close associates to the offender." Oregon Dep't of Corrections, Sex Offender Community Notification in Oregon at 7 (Jan.1995). This practice continued after enactment of the statute requiring notification to "a broader public." As of the time of the study, there had been 237 notification plans submitted under the new law. In this context, the Oregon Department of Corrections reported as follows:

In January 1995, forty-five parole/probation sex offender specialists from thirty-five counties responded to a survey of their experience with Community Notification. These officers were responsible for a total caseload of 2,160 sex offenders. The following information was gained from the surveys and [Sex Offender Supervision] Network discussions:

* * * * * *

Less than 10% of offenders experienced some form of harassment. Incidents reported included name calling, graffiti, toilet papering and minor property vandalism, monitoring of a home by video camera, repeated reports of unfounded violations to parole/probation officers, and picketing of residences.

There were two extreme cases of retaliation. One sex offender had a gun pointed at him and was threatened. In another case, a victim had tires slashed and the offender was blamed. Although the offender passed a polygraph and was accountable for the time, there were threats made that the offender's home would be burned down.

* * * * * *

Other circumstances reported by parole/probation officers included:

Community notification has made it more difficult to find residences for some sex offenders released from prison.

* * * * * *

Notification has [affected] employment opportunities for sex offenders.

* * * * * *

Businesses who were initially willing quietly to employ a sex offender sometimes do not provide jobs when the hiring will clearly become public.

Id. at 12-14.

There is a threshold jurisdictional issue for decision. The appellants in E.B. contend that the district court was without subject matter jurisdiction under the doctrine articulated by the Supreme Court in Rooker v. Fidelity Trust Co., 263 U.S. 413, 44 S.Ct. 149, 68 L.Ed. 362 (1923), and District of Columbia Court of Appeals v. Feldman, 460 U.S. 462, 103 S.Ct. 1303, 75 L.Ed.2d 206 (1983). Section 1257 of Title 28 of the United States Code bestows upon the Supreme Court of the United States appellate jurisdiction to review final judgments of the highest courts of the respective states. The so-called Rooker-Feldman doctrine teaches that, by negative implication, the inferior federal courts lack subject matter jurisdiction to review judgments of those courts. We have interpreted the doctrine to encompass final decisions of lower state courts as well. See Port Auth. Police Benevolent Ass'n, Inc. v. Port Auth. of N.Y. & N.J., 973 F.2d 169, 177-78 (3d Cir.1992).

Appellants point out that E.B. demanded and received judicial review of the prosecutor's Tier 3 classification and notification plan and that he advanced federal constitutional arguments in that proceeding for preventing the classification and notification plan from being put into effect. See Tr. Megan's Law Hearing (N.J.Super.Ct. Law Div. Dec. 7, 1995) at 6-9. The Superior Court, Law Division, after a hearing, rejected E.B.'s challenge and ordered that notification be given. E.B. appealed to the Appellate Division, which affirmed. The Supreme Court of New Jersey thereafter denied E.B.'s petition for certification of appeal. As appellants stress, the relief E.B. seeks in this proceeding is an injunction directing that the notification ordered by the New Jersey Superior Court, Law Division, not be carried out.

We agree with appellants that this is a paradigm situation in which Rooker-Feldman precludes a federal district court from proceeding. To grant E.B. relief would require an inferior federal court to determine that the New Jersey court's judgment was erroneous and would foreclose implementation of that judgment. See FOCUS v. Allegheny County Court of Common Pleas, 75 F.3d 834, 840 (3d Cir.1996).

The district court reached a contrary conclusion because it believed that although E.B. raised constitutional issues, he "was denied an opportunity to meaningfully raise constitutional challenges to Megan's Law." 914 F.Supp. at 89 (emphasis supplied). Its belief was based primarily on the fact that the Supreme Court of New Jersey in Doe had described a Megan's Law proceeding in the trial court as a "summary proceeding" and had stated that "the only issue for the court on the Tier level of notification is the risk of reoffense." Id. 914 F.Supp. at 89-90; Doe, 662 A.2d at 382-83. This suggested to the district court that the New Jersey courts do not consider constitutional challenges in a Megan's Law proceeding. 914 F.Supp. at 90.

If we shared the belief of the district court that E.B.'s constitutional challenges were not considered by the New Jersey courts--and, under Doe, could not be considered by them--we would also conclude that Rooker-Feldman did not deprive the district court of jurisdiction. However, we do not read the Doe opinion as instructing New Jersey courts to ignore properly raised claims based on the federal Constitution,9 and it is clear that the New Jersey courts do not so read that opinion. In In re G.B., 286 N.J.Super. 396, 669 A.2d 303, 306 (N.J.Super.Ct.App.Div.1996), aff'd, 685 A.2d 1252, the Appellate Division considered constitutional challenges and rejected them on the merits because these same issues had been previously considered and rejected in Doe. Shortly thereafter, the Superior and Supreme Courts of New Jersey, in appeals from a denial of relief in a Megan's Law proceeding, addressed constitutional challenges to Megan's Law for which there was no binding precedent. See In re C.A., 679 A.2d at 1153. Even if there were not this clear evidence, however, we would have to "assume that state procedures will afford an adequate remedy, in the absence of unambiguous authority to the contrary." Pennzoil Co. v. Texaco Inc., 481 U.S. 1, 15, 107 S.Ct. 1519, 1528, 95 L.Ed.2d 1 (1987).

The only remaining issue with respect to E.B. and the Rooker-Feldman doctrine is whether a litigant can be said to have a meaningful opportunity to raise an issue in a state proceeding when the highest court of that state has rejected, in another litigant's case, the same argument the litigant wishes to raise. Our answer is in the affirmative.

Rooker-Feldman abstention is necessary to preserve the United States Supreme Court's appellate jurisdiction--as well as to limit federal court review of state court decisions to the avenue provided for such by Congress. See Ernst v. Child and Youth Services of Chester County, 108 F.3d 486, 491 (3d Cir.1997). The federal court structure established by Congress intends that only the Supreme Court have the opportunity to decide that a state court has reached an erroneous conclusion on a federal constitutional claim. Nothing suggests that this structure should be altered where the state court's decision is based upon what is already settled precedent in that state.

As we have previously observed, the interests served by Rooker-Feldman are quite similar to those served by giving a state court judgment res judicata effect in a subsequent federal proceeding. Marks v. Stinson, 19 F.3d 873, 885-86 n. 11 (3d Cir.1994); Valenti v. Mitchell, 962 F.2d 288, 297 (3d Cir.1992). If a litigant resorts to a state court and suffers an adverse judgment, a lower federal court must respect that judgment unless and until it is overturned. The litigant's only remedy is by way of appeal through the state court system and by way of petition to the Supreme Court of the United States thereafter.10

We will, accordingly, reverse the judgment of the district court in E.B.'s case11 and remand with instructions to dismiss for want of subject matter jurisdiction.

This does not mean, however, that the district court lacked jurisdiction over the class claims in W.P. As we concluded in Valenti, 962 F.2d at 298, " Rooker-Feldman does not bar individual constitutional claims by persons not parties to earlier state court litigation." In W.P., at least some of the representative plaintiffs were not the subject of any kind of judicial order when they filed this suit to secure injunctive relief against enforcement of Megan's Law. Indeed, neither they nor the state had petitioned any state court for any relief. The claims of these class plaintiffs were sufficient to confer subject matter jurisdiction12 on the district court.13

V. THE EX POST FACTO AND DOUBLE JEOPARDY ISSUES

The Ex Post Facto Clause forecloses retroactive application of a law that "inflicts a greater punishment, than the law annexed to the crime, when committed." Calder v. Bull, 3 U.S. (3 Dall.) 386, 390, 1 L.Ed. 648 (1798). The Double Jeopardy Clause forbids "multiple punishments for the same offense." United States v. Halper, 490 U.S. 435, 440, 109 S.Ct. 1892, 1897, 104 L.Ed.2d 487 (1989). Accordingly, neither clause is implicated unless the state has inflicted "punishment." Since no one here suggests that "punishment" has a different meaning under one of these clauses than under the other, the critical issue to which we now turn is whether the notification called for in situations involving Tier 2 and Tier 3 registrants is "punishment" for purposes of the Ex Post Facto and Double Jeopardy Clauses.A. The Artway Standard

In Artway, when we addressed the issue of whether registration under Megan's Law constituted "punishment," we found no Supreme Court precedent addressing a similar statutory provision. In order to "divine" a "test for punishment," we reviewed the Supreme Court case law and looked for common considerations. 81 F.3d at 1254-63. Recognizing "that the appropriate 'punishment' analysis depends on the context," we derived an "analytical framework for this case." Id. 81 F.3d at 1261, 1263. Specifically, we concluded that a "measure must pass a three-prong analysis--(1) actual purpose, (2) objective purpose, and (3) effect--to constitute non-punishment." Id. 81 F.3d at 1263.

Under this Artway analysis, we first look to whether the adverse effect on individuals results from a desire on the part of the legislature to punish past conduct or is a by-product of a bona fide legislative effort to remedy a perceived societal problem. "If the legislature intended Megan's Law to be 'punishment,' i.e. retribution was one of its actual purposes, then it must fail constitutional scrutiny. If, on the other hand, 'the restriction of the individual comes about as a relevant incident to a regulation,' the measure will pass this first prong." Id. (quoting De Veau v. Braisted, 363 U.S. 144, 160, 80 S.Ct. 1146, 1155, 4 L.Ed.2d 1109 (1960)).

The second inquiry--into "objective purpose"--focuses on the operation of the legislative measure and on whether analogous measures have traditionally been regarded in our society as punishment. In Artway, we suggested that there were three aspects of "objective purpose" that should be considered by a court before deciding whether the party challenging the statute has carried its burden of showing that an objective observer in our society would perceive the measure as punitive. Id. It is important to consider the measure's proportionality--whether the remedial purpose of a legislative measure purporting to be non-punitive can explain all the adverse effects on those involved. While it is true that "even remedial sanctions carry the sting of punishment," id. 81 F.3d at 1260 (internal quotation marks omitted), only if the sting is not "reasonably related" to the remedial goal would an objective observer be justified in perceiving a punitive purpose, id. 81 F.3d at 1265. It is also important to consider history. If analogous measures have traditionally been regarded by our society as "serv[ing] punitive purposes" and the text and the legislative history do "not make [the legislature's] plausible remedial purposes clear," id. 81 F.3d at 1257, there is an objective basis for regarding the measure as punishment. Finally, we noted in Artway that some measures are intended to have a mixed salutary and deterrent effect. The examples we gave were taxes on illegal activities (like possession of drugs) and on activities that the state concededly wished to discourage. See id. 81 F.3d at 1259. Such mixed measures will not be deemed to have an objectively punitive purpose despite their deterrent purpose unless that deterrent purpose is an unnecessary complement to the measure's salutary operation, the measure is operating in an unusual manner inconsistent with its historically mixed purposes, or the deterrent purpose overwhelms the salutary purpose. See id. 81 F.3d at 1263.

"The final prong [of the Artway analysis] examines whether the effects--or 'sting'--of a measure is so harsh 'as a matter of degree' that it constitutes 'punishment.' " Id. 81 F.3d at 1266 (citing California Dep't of Corrections v. Morales, 514 U.S. 499, 509, 115 S.Ct. 1597, 1603, 131 L.Ed.2d 588 (1995)). This prong necessarily involves difficult line-drawing. Unfortunately, the Supreme Court case law provides only a few fixed points. We know that, under certain circumstances, the "sting" of incarceration or forfeiture of one's citizenship is sufficiently extraordinary to require a finding of punishment, see Miller v. Florida, 482 U.S. 423, 107 S.Ct. 2446, 96 L.Ed.2d 351 (1987); Trop v. Dulles, 356 U.S. 86, 78 S.Ct. 590, 2 L.Ed.2d 630 (1958), and we have recently been told that civil commitment of violent sex offenders does not, see Kansas v. Hendricks, --- U.S. ----, 117 S.Ct. 2072, 138 L.Ed.2d 501 (1997).

There are two recent Supreme Court cases which potentially bear upon our decision: United States v. Ursery, --- U.S. ----, 116 S.Ct. 2135, 135 L.Ed.2d 549 (1996), and Kansas v. Hendricks, at ----, 117 S.Ct. at 2072. Appellees insist that after Ursery and Hendricks, Artway does not provide an appropriate standard for determining whether Megan's Law notification constitutes "punishment" for purposes of the Ex Post Facto and Double Jeopardy Clauses. We disagree.

In Ursery, the Supreme Court held that "civil forfeitures ... do not constitute 'punishment' for purposes of the Double Jeopardy Clause" even when the value of the property forfeited is arguably excessive when compared to the harm suffered by the government from the conduct giving rise to the forfeiture. --- U.S. at ----, 116 S.Ct. at 2138. The Court first emphasized that its case law had sharply distinguished between in rem forfeiture proceedings and in personam civil fine proceedings. It explained that in the latter "it is the wrongdoer in person who is proceeded against ... and punished" while in the former "it is the property which is proceeded against, and by resort to a legal fiction, held guilty and condemned." Id. at ----, 116 S.Ct. at 2145 (quoting from Various Items of Personal Property v. United States, 282 U.S. 577, 580-81, 51 S.Ct. 282, 284, 75 L.Ed. 558 (1931)). Thus, civil forfeitures are not "criminal punishments because they [do] not impose a second in personam penalty for the criminal defendant's wrongdoing." Id. at ----, 116 S.Ct. at 2141. Second, the Court noted, "[c]ivil forfeitures, in contrast to civil penalties, are designed to do more than simply compensate the Government. Forfeitures ... are designed primarily to confiscate property used in violation of the law, and to require disgorgement of the fruits of illegal conduct. [For this reason,] it is virtually impossible to quantify, even approximately, the nonpunitive purposes served by a particular civil forfeiture." Id. at ----, 116 S.Ct. at 2145. Accordingly, while a court can determine whether a civil fine has a punitive component by comparing its size to the harm experienced by the government, a court is not in a position "to determine whether a particular forfeiture bears no rational relationship to the nonpunitive purposes of that forfeiture." Id.

The holding of Ursery is a narrow one limited to civil forfeitures. Neither of the principal rationales supporting its conclusion is pertinent here and we find nothing in the Court's reasoning that is inconsistent with the Artway standard.14 It necessarily follows that Ursery provides no justification for abandoning that standard. See Third Circuit Internal Operating Procedures 9.1.

After the district court's decision in these cases, the Supreme Court decided Kansas v. Hendricks, at ----, 117 S.Ct. at 2072. The Court there upheld a Kansas statute that provides for the civil commitment of "sexually violent predators." See Kan. Stat. Ann. § 59-29a01 et seq. Under the statute, a person convicted or charged with a violent sexual offense and suffering from a "mental abnormality or personality disorder which makes the person likely to engage in the predatory acts of sexual violence," § 59-29a02(a), may be confined to state custody for "control, care and treatment until such time as the person's mental abnormality or personality disorder has so changed that the person is safe to be at large," § 59-29a07(a). Prior to Leroy Hendricks' scheduled release from prison, the state invoked the statute to have him confined as a sexual predator. Hendricks, who had an extensive history of molesting children, challenged the act on substantive due process, ex post facto, and double jeopardy grounds. The Supreme Court rejected all three claims and held that the state's involuntary commitment program did not constitute punishment for the purpose of ex post facto or double jeopardy.

Like Ursery, Hendricks does not establish "a single 'formula' " for identifying which legislative measures constitute punishment and which do not. Morales, 514 U.S. at 509, 115 S.Ct. at 1603. However, the context involved in Hendricks--civil commitment of sex offenders--is, obviously, more closely related to the context involved here than was the context of Ursery. In determining the continuing viability of Artway, therefore, we must give careful consideration to how Hendricks addressed the question of whether civil commitment is punishment. We find substantial overlap between the factors relied on in Hendricks and those that comprise the Artway test and we discern no need to abandon (or overhaul) Artway.

The Court's analysis in Hendricks begins by inquiring into "the legislature's stated intent," --- U.S. at ----, 117 S.Ct. at 2082, just as Artway directs that we begin with the legislature's actual purpose. The Court found Kansas' placement of the challenged provision in the probate code instead of the criminal code, and the legislature's description of its creation as a "civil commitment procedure," to be evidence of the legislature's "disavow[ing] any punitive intent." Id. at ----, ----, 117 S.Ct. at 2082, 2085. "Nothing on the face of the statute suggest[ed] that the legislature sought to create anything other than a civil commitment scheme designed to protect the public from harm." Id. at ----, 117 S.Ct. at 2082.

Hendricks then goes beyond the legislature's stated intent to consider additional factors, including those factors Artway incorporates into its objective purpose prong. Like Artway 's inquiry into proportionality, Hendricks repeatedly describes how the Kansas statute is tailored to achieve its remedial purpose of protecting the public. The Court observes that prior criminal conduct is appropriately examined for the narrow evidentiary purpose of predicting dangerousness. See id. The Court also notes that Kansas "limited confinement to a small segment of particularly dangerous individuals," id. at ----, 117 S.Ct. at 2085, and that those affected individuals do not "remain confined any longer than [they] suffer[ ] from a mental abnormality rendering [them] unable to control [their] dangerousness," id. at ----, 117 S.Ct. at 2083. As the Court recognizes, "[f]ar from any punitive objective, the confinement's duration is instead linked to the stated purposes of the commitment, namely, to hold the person until his mental abnormality no longer causes him to be a threat to others." Id. Finally, the Court observes that the individuals are subject only to the conditions placed on any involuntarily committed person in a state mental institution and not to the "more restrictive conditions" placed on state prisoners. Id. at ----, 117 S.Ct. at 2082.

Hendricks, like Artway, relied heavily on history. In the Court's view, the confinement involved is "one classic example" in a long history of measures restricting the freedom of the dangerously mentally ill--legislative initiatives which have been consistently held to be nonpunitive. Id. at ----, 117 S.Ct. at 2083. The Court specifically analogized the Kansas confinement to the quarantines of those afflicted with highly contagious diseases, and recognized that it has "never held that the Constitution prevents a State from civilly detaining those for whom no treatment is available, but who nevertheless pose a danger to others." Id. at ----, 117 S.Ct. at 2084.

There is also support in Hendricks for Artway 's inquiry into the relationship between a "mixed" measure's salutary and deterrent purposes. Hendricks discusses the multiple purposes of the Kansas statute, including incapacitation of dangerous sex offenders as well as their treatment, and concludes that the statute would not constitute punishment even if providing treatment were merely an "ancillary purpose"--and not the "primary" purpose--for passing the statute. Id. This is consistent with Artway 's allowance that a measure can be non-punitive even when it does not have solely "salutary" purposes such as treatment.

Though Hendricks does not explicitly discuss what Artway calls the "effects prong," we find nothing in Hendricks inconsistent with Artway 's direction to examine what the challenged measure actually does to the affected individuals. This is not to say, of course, that Hendricks lacks implications for the application of the effects prong. The Court held that potentially indefinite civil commitment of dangerous sex predators is not punishment. This provides a new and important "fixed point" that is of great utility in determining on which side of the punitive/nonpunitive line to place community notification.

Although Hendricks thus does not suggest to us that any of the considerations identified as relevant in Artway are no longer relevant to a challenge based on the Ex Post Facto and Double Jeopardy Clauses, we do discern a teaching in Hendricks that we do not discern in the Supreme Court case law preceding Artway. In the course of holding that Kansas' Sexually Violent Predator Act "does not impose punishment," id. at ----, 117 S.Ct. at 2086, the Hendricks Court made the following cogent observation regarding the deference that must be accorded to the legislature's judgment as to whether its action is remedial:

Although we recognize that a "civil label is not always dispositive," Allen [v. Illinois, 478 U.S. 364, 369, 106 S.Ct. 2988, 2992, 92 L.Ed.2d 296 (1986) ], 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." United States v. Ward, 448 U.S. 242, 248-249, 100 S.Ct. 2636, 2641, 65 L.Ed.2d 742 (1980).

Id. at ----, 117 S.Ct. at 2082.

As we pointed out in Artway, the Supreme Court had previously required this degree of deference only in cases where the issue before it was "whether a proceeding is effectively criminal so that the procedural protections of the Fifth and Sixth Amendments must apply" in that proceeding. Artway, 81 F.3d at 1262 n. 26. After Hendricks, however, it seems clear that similar deference to the legislative judgment is required whenever legislative measures are challenged on the basis of the Ex Post Facto and Double Jeopardy Clauses.15 While the Hendricks Court did characterize Hendricks' claim at one point as an "argument ... that the Act establishes criminal proceedings," --- U.S. at ----, 117 S.Ct. at 2081, the issue before the Court was whether the Act imposed "punishment" for purposes of the Ex Post Facto and Double Jeopardy Clauses, and the Court's holding was that the Act did not.

Accordingly, in Artway terms, if we determine that the actual legislative purpose was remedial, we must sustain Megan's Law against the current challenges unless its objective purpose or its effect are sufficiently punitive to overcome a presumption favoring the legislative judgment.

As we have indicated, in Artway we addressed only whether Tier 1 registrants under Megan's Law are subjected to punishment--that is, whether being required to register, and having the resulting disclosures available to law enforcement personnel, constitute punishment. In that context, we determined "whether the legislature's actual purpose [when enacting Megan's Law] was to punish." Artway, 81 F.3d at 1264. Looking to the statute's own statement of purpose16 and the scant legislative history,17 we found that the legislative purpose of Megan's Law was to identify potential recidivists and alert the public when necessary for the public safety, and to help prevent and promptly resolve incidents involving sexual abuse and missing persons. We then noted that "[p]rotecting the public and preventing crimes are the types of purposes [the Supreme Court has] found 'regulatory' and not punitive." Id.; see also De Veau, 363 U.S. at 160, 80 S.Ct. at 1154-55. We therefore concluded that the restrictive provisions of Megan's Law passed the "actual purpose" test.

Since in Artway we were only dealing with a challenge to registration, we were not required to definitively resolve the legal question of the actual purpose of notification.18 However, the record evidence of legislative intent is exactly the same for both registration and notification. Nothing has been called to our attention that causes us to change the conclusion we reached in Artway regarding this evidence. While the appellants view the context in which Megan's Law was enacted as indicative of a punitive intent, wefind it entirely consistent with its declared remedial purpose. Accordingly, we have no basis for questioning the legislature's declared purpose, which is remedial and devoid of any indication of an intent to punish. We must give substantial deference to that judgment.

In Artway, we concluded that registration and Tier 1 notification of law enforcement personnel was fully explained by the nonpunitive, legislative purpose. We explained:

Here, the solely remedial purpose of helping law enforcement agencies keep tabs on these offenders fully explains requiring certain sex offenders to register. Registration may allow officers to prevent future crimes by intervening in dangerous situations.... [T]he registrant may face some unpleasantness from having to register and update his registration[, b]ut the remedial purpose of knowing the whereabouts of sex offenders fully explains the registration provision.... And the means chosen--registration and law enforcement notification only--is not excessive in any way. Registration, therefore, is certainly "reasonably related" to a legitimate goal: allowing law enforcement to stay vigilant against possible re-abuse.

81 F.3d at 1265.

The issue now before us is whether the provisions of Megan's Law that call for dissemination of information about registrants beyond law enforcement personnel are also fully explained by the nonpunitive, legislative purpose. In addressing this issue, there is a lesson in the above quoted portion of Artway that we must keep in mind. The relevant issue is whether these provisions are " 'reasonably related' to a legitimate goal." Nothing in Artway or the Supreme Court cases upon which it relies requires a perfect fit between end and means. Nor does anything in Ursery or Hendricks. An absence of remedial, objective purpose is not demonstrated by pointing out that the legislature did not address what might be perceived as another aspect of the same problem or that there may be a means of serving the legislative end that would be more effective than the means chosen. If a reasonable legislator motivated solely by the declared remedial goals could have believed the means chosen were justified by those goals, then an objective observer would have no basis for perceiving a punitive purpose in the adoption of those means.

We conclude that the Tier 2 and 3 dissemination of information beyond law enforcement personnel is reasonably related to the nonpunitive goals of Megan's Law. As we have already indicated, these goals include identifying potential recidivists, notifying those who are likely to interact with such recidivists to the extent necessary to protect public safety, and helping prevent future incidents of sexual abuse. The fundamental premise of Megan's Law is that registration and carefully tailored notification can enable law enforcement and those likely to encounter a sex offender to be aware of a potential danger and "to stay vigilant against possible re-abuse." Id. This is not an unreasonable premise.

Moreover, these goals have not been pursued in a way that has imposed a burden on registrants that clearly exceeds the burden inherent in accomplishment of the goals. The statutory scheme is a measured response to the identified problem that does not subject all registrants to dissemination of information beyond law enforcement personnel. The Guidelines call for a risk assessment based on objective criteria, all of which might reasonably be perceived as relevant to the degree of risk presented by each registrant. This risk assessment is utilized to determine the maximum scope of the notification concerning the registrant. In the case of Tier 1 registrants, who comprise over 45% of those required to register, dissemination is limited to law enforcement personnel. In the case of the moderate risk registrants in Tier 2, who comprise 50% of those evaluated, dissemination is limited to those in the community who have responsibility for, or provide support to, those who are most likely to be victimized if the registrant recidivates. Even with respect to the 5% of registrants determined to pose higher risk, there is no unlimited public dissemination. Under the Guidelines, information is disseminated only to those who are "reasonably certain" to encounter the registrant.

Appellants nevertheless insist that the remedial goal of Megan's Law does not fully justify the means selected. First, they point to the fact that risk assessment under the Guidelines is based primarily on the registrant's past behavior. Past criminal conduct is the basis for 90 of the possible 111 points in the Registrant Risk Assessment Scale. Id. at 1266 n. 30. According to appellants, this Scale fails to take sufficient account of treatment or other positive changes in a registrant's life. They conclude that "the reach of this law will necessarily be excessive, encompassing those who do not actually pose a genuine risk of reoffense." Appellants' Br. at 41. However, the non-existence of a perfect predictor of recidivism should not preclude legislative resort to a rationally based instrument of risk assessment, developed and validated by mental health professionals. The most appellants have done is to suggest that a more effective predictor might be devised; that is not enough to make the objective purpose of the predictor adopted a punitive one.19

Appellants further suggest that the information disseminated is often excessive in light of the stated remedial aims. The information disseminated with respect to a Tier 2 or Tier 3 registrant includes his or her name, description, recent photograph, address, place of employment or schooling, and a description of any vehicle used by him or her along with its license number. Appellants point out that some of this information will sometimes be unnecessary. "[F]or example, if the registrant works 20 or 30 miles from his home, the registrant's neighbor who receives notification is not 'likely to encounter' the registrant at his place of employment. Likewise, those who live near the same registrant's place of employment are not 'likely to encounter' the registrant at his home. Yet in both instances, notification includes the same information...." Id.

We are not persuaded. First, information that an offender does not spend all of his time in the vicinity, but does have a residence or a place of employment/school elsewhere, may indeed serve a remedial purpose in helping individuals know when it is that they are "likely to encounter" the offender. Moreover, even if this were not so, a decision not to expend the resources necessary to tailor each notice to the circumstances of the person receiving notice is hardly inconsistent with good faith pursuit of the declared remedial purposes.

Having found a reasonable "fit" between end and means, we turn to historical precedent. To appellants, the dissemination of information beyond law enforcement personnel is closely analogous to the well-recognized historical punishments of public shaming, humiliation and banishment as those practices were employed in colonial times. We rejected a very similar argument in United States v. Criden, 648 F.2d 814 (3d Cir.1981). There, the district court had denied the media the right to copy, for rebroadcast, video and audio tapes admitted into evidence and played to the jury during a criminal trial. In support of its decision to foreclose post-trial dissemination of public record information to the public, the district court made the following observation:

The greater and more widespread the publicity about a particular criminal case, the more likely it is that penalties not prescribed by the law will be visited upon the accused and, more importantly, upon innocent relatives and friends....

Given the nature of our society these side effects are inevitable; indeed, it can be argued that they form an important, if unofficial, part of the sanctions imposed by society upon lawbreakers. The unfortunate fact is, however, that these side effects are not uniformly visited upon persons accused of violating the law. And, since they are not an official part of the criminal justice process, and are beyond the reach of that process, there is probably no acceptable way of ensuring uniformity of application.

Id. 648 F.2d at 824 (quoting United States v. Criden, 501 F.Supp. 854, 860 (E.D.Pa.1980)). In pursuing this theme, the district court likened the proposed rebroadcast to placing the defendant in public stocks.

Nor can we accept the [district] court's strained analogy of rebroadcast to "parading a convicted defendant through the streets, or holding him up to public ridicule by exhibiting him in a cage or in the stocks." 501 F.Supp. at 860.

Id. 648 F.2d at 825. Nor can we accept the suggested analogy between notification's re-publication of information publicly available at the time of a sex offender's trial and the holding of a convicted defendant up to public ridicule. Public shaming, humiliation and banishment all involve more than the dissemination of information. State dissemination of information about a crime and its perpetrators was unnecessary in colonial times because all in the colonial settlement would have knowledge of these matters. Rather, these 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.

The "sting" of Megan's Law for Tier 2 and 3 registrants results not from their being publicly displayed for ridicule and shaming but rather from the dissemination of accurate public record information about their past criminal activities and a risk assessment by responsible public agencies based on that information. This distinction makes a substantial difference when one looks for the relevant historical understanding of our society. Dissemination of information about criminal activity has always held the potential for substantial negative consequences for those involved in that activity. Dissemination of such information in and of itself, however, has never been regarded as punishment when done in furtherance of a legitimate governmental interest.

When there is probable cause to believe that someone has committed a crime, our law has always insisted on public indictment, public trial, and public imposition of sentence, all of which necessarily entail public dissemination of information about the alleged activities of the accused. As this court has explained, we insist upon this public dissemination for a number of reasons: It "heightens public respect for the judicial process," it "permits the public to ... serve as a check upon the judicial process," and it "plays an important role in the ... free discussion of governmental affairs." Publicker Industries, Inc. v. Cohen, 733 F.2d 1059, 1070 (3d Cir.1984) (internal quotation marks omitted). Wholly independent of the criminal sanctions that conviction may entail, the consequences of our law requiring this public dissemination of information can be severe. In every case, a conviction becomes a matter of public record, and in many cases that conviction may receive widespread media attention. Depending upon the crime and the circumstances, information disseminated as a result of our insistence on public prosecution may be the source of a wide range of adverse consequences for the convicted defendant, running from mild personal embarrassment to social ostracism and/or vigilante retribution. Employment may be lost, and the opportunity for future employment may be dramatically reduced. It may take a lifetime of effort on the part of a convicted defendant to restore previously existing relationships with those with whom he deals personally, and restoration of his reputation among others may never occur. Nevertheless, our laws' insistence that information regarding criminal proceedings be publicly disseminated is not intended as punishment and has never been regarded as such.

We believe the required dissemination of information generated by our criminal justice system and the subsequent dissemination of "rap sheet" information to regulatory agencies, bar associations, prospective employers and interested members of the public20 constitute far more compelling analogies than the stocks, cages, and scarlet letters referenced by appellants.21

We also agree with appellees that various forms of state warnings about threats to public safety provide more apt analogies to Tier 2 and Tier 3 notification than the referenced colonial practices. In order to provide members of the public with an opportunity to take steps to protect themselves, the government has traditionally published appropriate warnings about a range of public hazards. Posters warning that a pictured individual is abroad in the community and to be regarded as armed and dangerous come most readily to mind. But there are others as well. The state has traditionally, for example, posted quarantine notices when public health is endangered by individuals with infectious diseases. Cf. Hendricks, at ----, 117 S.Ct. at 2084 ("A State could hardly be seen as furthering a'punitive' purpose by [isolating] persons inflicted with a[ ] highly contagious disease.").22 Significantly, these warnings communicate not only facts about past events but also the fact that a public agency has found a significant future risk based on those events.

Whenever these state notices are directed to a risk posed by individuals in the community, those individuals can expect to experience embarrassment and isolation. Nevertheless, it is generally recognized that the state has a right to issue such warnings and the negative effects are not regarded as punishment. Because the closest analogies have not historically been regarded as punishment, we conclude that historical precedent does not demonstrate an objective punitive purpose.

Finally, we turn to the third consideration involved in assessing objective purpose. That consideration, as we understand it, is a savings provision--that is, even if the remedial purpose of a measure cannot fairly be said to justify all of its aspects, it will nevertheless be found nonpunitive if measures of this type, like taxes, have traditionally served both remedial and deterrent purposes and the particular measure before the court serves such purposes in a manner consistent with its analogous antecedents. Having concluded that the remedial purpose of Megan's Law justifies all of its aspects, it necessarily follows in this case, as it did in Artway, 81 F.3d at 1266, that this third consideration does not counsel in favor of a finding that it is punitive.

As we have indicated, we hypothesized in Artway that "a law [could] constitute unconstitutional 'punishment' because of its effects" even where no actual or objective punitive purpose is shown. 81 F.3d at 1260. We explained:

[An] examination of effects, like the Austin [v. United States, 509 U.S. 602, 113 S.Ct. 2801, 125 L.Ed.2d 488 (1993),] inquiry into history, is necessary to limit what would otherwise be the untenable results of the De Veau subjective purpose inquiry and the Halper means-end calculus. While even a substantial "sting" will not render a measure "punishment," ... at some level the "sting" will be so sharp that it can only be considered punishment regardless of the legislators' subjective thoughts.

Id. 81 F.3d at 1261.

It is clear from Artway, however, that for the effects of a measure to render it "punishment," those effects must be extremely onerous. Even deprivation of one's livelihood is not sufficiently onerous. Flemming v. Nestor, 363 U.S. 603, 80 S.Ct. 1367, 4 L.Ed.2d 1435 (1960) (termination of social security benefits); Hawker v. People of State of New York, 170 U.S. 189, 18 S.Ct. 573, 42 L.Ed. 1002 (1898) (revocation of license to practice one's profession). Moreover, while Artway 's third prong serves as an independent hurdle that a legislative measure must surmount, when it is applied, the burden imposed must still be evaluated in the light of the importance of any legitimate governmental interest served. The only examples the case law suggests of effects sufficiently onerous are deprivation of one's United States citizenship that leaves one a "stateless person" and a complete deprivation of personal freedom (i.e., incarceration). Even these deprivations are not per se punishment, however. While in some circumstances making one a "stateless person" is punishment, denaturalization as a remedy for citizenship fraudulently obtained is regarded not as punishment but as a necessary part of regulating naturalization of aliens. See Trop, 356 U.S. at 98, 78 S.Ct. at 596-97. Even incarceration is not always punishment. Pre-trial detention and post-sentence civil commitment of dangerous offenders have both been expressly found to be nonpunitive measures when justified by important state interests. See United States v. Salerno, 481 U.S. 739, 107 S.Ct. 2095, 95 L.Ed.2d 697 (1987); Hendricks, at ----, 117 S.Ct. at 2072.

The direct effects of Megan's Law clearly do not rise to the level of extremely onerous burdens that sting so severely as to compel a conclusion of punishment. All Megan's Law mandates is registration and notification. Under Megan's Law, New Jersey has not deprived appellants of their freedom or their citizenship. The state has imposed no restrictions on a registrant's ability to live and work in a community, to move from place to place, to obtain a professional license or to secure governmental benefits.

What concerns registrants, however, are the indirect effects: Actions that members of the community may take as a result of learning of the registrant's past, his potential danger, and his presence in the community. People interact with others based on the information they have about them. Knowing that someone is a convicted sex offender and has been evaluated as a continuing risk is likely to affect how most people treat that person.

There can be no doubt that the indirect effects of Tier 2 and Tier 3 notification on the registrants involved and their families are harsh. The record documents that registrants and their families have experienced profound humiliation and isolation as a result of the reaction of those notified. Employment and employment opportunities have been jeopardized or lost. Housing and housing opportunities have suffered a similar fate. Family and other personal relationships have been destroyed or severely strained. Retribution has been visited by private, unlawful violence and threats and, while such incidents of "vigilante justice" are not common, they happen with sufficient frequency and publicity that registrants justifiably live in fear of them. It also must be noted that these indirect effects are not short lived. While there are suggestions in the record that the circumstances of a registrant may stabilize as time passes after notification, the statute permits repeat notification over a period of many years.

The primary sting from Megan's Law notification comes by way of injury to what is denoted in constitutional parlance as reputational interests. This includes the burdens of isolation, harassment, loss of opportunities, and the myriad of more subtle ways in which one is treated differently by virtue of being known as a potentially dangerous sex offender. The other type of indirect effect is exposure to an increased risk of private violence that can result in damage to one's property or injury to one's person. We will focus on each class of indirect effects in turn.

Injury to reputation has traditionally been regarded in our society as a serious matter. Our law of defamation has from our earliest days protected reputation and provided compensation for wrongful injury to reputational interests. It has provided recourse, for example, for those whose reputations are injured by false allegations of criminal activity. At the same time, however, reputational interests have not been accorded the same level of protection in our society as interests that have been found "implicit in the concept of ordered liberty." Paul v. Davis, 424 U.S. 693, 713, 96 S.Ct. 1155, 1166, 47 L.Ed.2d 405 (1976).

In Paul v. Davis, law enforcement officials decided to alert local area merchants to possible shoplifters who might be operating during the Christmas season. They distributed a "flyer" to 800 merchants which contained the name and "mug shot" photo of individuals described as "Active Shoplifters." Davis, who had previously been arrested for--but never convicted of--shoplifting was included.

Davis brought a civil rights action against the law enforcement officials arguing that, by destroying his reputation in the community, they had violated his "right to privacy guaranteed by the First, Fourth, Fifth, Ninth, and Fourteenth Amendments." Id. at 712, 96 S.Ct. at 1166. Though acknowledging that the Constitution secures a right to personal privacy, the Supreme Court rejected the notion that Davis' interest in his reputation was sufficiently fundamental to come within that constitutional right. The Court observed:

In Roe [v. Wade, 410 U.S. 113, 93 S.Ct. 705, 35 L.Ed.2d 147 (1973) ], the Court pointed out that the personal rights found in this guarantee of personal privacy must be limited to those which are "fundamental" or "implicit in the concept of ordered liberty" as described in Palko v. Connecticut, 302 U.S. 319, 325, 58 S.Ct. 149, 152, 82 L.Ed. 288 (1937). The activities detailed as being within this definition were ones very different from that for which respondent claims constitutional protection--matters relating to marriage, procreation, contraception, family relationships, and child rearing and education. In these areas it has been held that there are limitations on the States' power to substantively regulate conduct.

Respondent's claim is far afield from this line of decisions. He claims constitutional protection against the disclosure of the fact of his arrest on a shoplifting charge. His claim is based, not upon any challenge to the State's ability to restrict his freedom of action in a sphere contended to be "private," but instead on a claim that the State may not publicize a record of an official act such as an arrest. None of our substantive privacy decisions hold this or anything like this, and we decline to enlarge them in this manner.

Id. at 713, 96 S.Ct. at 1166.

The indirect effects experienced by Tier 2 and Tier 3 registrants, while quite likely more profound than those complained of by Davis, are clearly of a similar nature. Just as Davis sought constitutional protection from the consequences of state disclosure of the fact of his shoplifting arrest and law enforcement's assessment that he was a continuing risk, so registrants seek protection from what may follow disclosure of facts related to their sex offense convictions and the resulting judgment of the state that they are a continuing risk. It follows that, just as the officers' publication of the official act of Davis' arrest did not violate any fundamental privacy right of Davis', neither does New Jersey's publication (through notification) of registrants' convictions and findings of dangerousness implicate any interest of fundamental constitutional magnitude. The reputational interests asserted by appellants are "very different" from matters relating to marriage, procreation, and child rearing, and are therefore "far afield" from what has been deemed "fundamental" by the Constitution.23

Hendricks, and the long line of cases on which it relies, counsels that bona fide remedial legislation may inflict very substantial individual hardship without implicating the Ex Post Facto and Double Jeopardy Clauses. It necessarily follows that some limit must be placed on the situations in which a measure's sting alone, despite its remedial purpose and effect, will constitute punishment under those clauses and that classification as punishment on the basis of sting alone must be reserved for cases involving deprivation of the interests most highly valued in our constitutional republic. "[F]reedom from physical restraint 'has always been at the core of the liberty protected' " by the Constitution. Hendricks, at ----, 117 S.Ct. at 2079 (quoting Foucha v. Louisiana, 504 U.S. 71, 80, 112 S.Ct. 1780, 1785, 118 L.Ed.2d 437 (1992)). Freedom of thought and expression and freedom from state interference with the privacy interests identified in Davis are similarly "implicit in our concept of ordered liberty." Davis, 424 U.S. at 713, 96 S.Ct. at 1166. Interests such as these are sufficiently fundamental to our constitutionally secured liberty that state interference with them can be justified only by the most important of state interests.24 Davis establishes that reputational interests are not among these fundamental liberty interests.

We believe the state's interest in protecting the public here is similar to, and as compelling as, the state interest served by the civil commitment statute in Hendricks. Accordingly, based on Hendricks, we believe that the state's interest here would suffice to justify the deprivation even if a fundamental right of the registrant's were implicated. Given that something less than a fundamental interest is implicated, the impact of Megan's Law on the registrants' reputational interests is necessarily insufficient alone to constitute "punishment."

We now turn to the second type of indirect effects arising from notification. As we earlier observed, the record bears evidence of retributive assaults on registrants by private individuals. There is also evidence of vandalism and other damage to property of registrants and their associates. As we have also noted, however, each notification is accompanied by a warning against misuse of the information conveyed and an assurance that any private violence will be prosecuted. This is thus not a situation in which the state has encouraged private violence. Nor is it a situation in which the state has in some way incapacitated a person from taking steps to protect him-or herself against private violence or has deprived a citizen of the law enforcement protection accorded to others in the population generally. On the contrary, the state has taken affirmative steps to discourage private violence in response to notification, and is providing registrants with the law enforcement protection available to others.

We agree with the district court that the risk of private violence stems primarily from a registrant's past criminal activity. The most that can be said about notification is that the state, by disseminating accurate information about a registrant's crime and its assessment of future risk, may materially extend the period during which the increased risk of private violence may exist. While the extension of that increased risk is understandably of concern to plaintiffs, they have not persuaded us that the magnitude of the risk is such as to require classification of its extension as punishment. Although the record reflects that personal injury and property damage from private violence has occurred, it also reflects that these occurrences are relatively rare. Of the 135 notifications completed in New Jersey for which there is record data, only two occasioned assaults or property damage deemed serious enough by the victim-registrant to warrant a report to law enforcement authorities. Even if we were prepared to broaden our consideration to include examples of physical harm to registrants not reported to police, this would increase the total number of record cases to just three. Our record with respect to Washington and Oregon also evidences that reported instances of personal injury or property damage are rare.25

As we view this matter, there is unfortunately a background risk of private violence that is necessarily assumed by everyone in our society. When one commits a reprehensible crime and is publicly prosecuted, that risk is undoubtedly augmented to a limited degree. The duration of that degree of augmented risk is likely to be extended by notification pursuant to Megan's Law and this is understandably a concern for registrants. Nevertheless, we believe the Supreme Court would not regard this indirect effect of Megan's Law as sufficiently burdensome to require classification of the law as punitive. Certainly, in terms of the impact on the everyday lives of registrants, the burden of this aspect of Megan's Law pales by comparison to the civil commitment of sex offenders sanctioned in Hendricks.

Because Megan's Law satisfies each of the three elements of the Artway test, we hold that the notification required by Megan's Law does not constitute punishment for purposes of the Ex Post Facto and Double Jeopardy Clauses.

The Fourteenth Amendment of the United States Constitution provides that "no person shall be deprived of life, liberty, or property without due process of law." U.S. Const. Amend. XIV. Appellants insist that they have a liberty interest that entitles them to the protection of procedural due process under this provision. Appellees insist that there is no such interest.

Liberty interests that trigger procedural due process may be created by state law or by the federal constitution itself. See Sandin v. Conner, 515 U.S. 472, 115 S.Ct. 2293, 132 L.Ed.2d 418 (1995). We need not reach the issue of whether appellants have a liberty interest recognized by the federal constitution because we are satisfied that appellants have a liberty interest created by the New Jersey Constitution of which they cannot be deprived without being accorded the process due under the Fourteenth Amendment.

If a state law requires that the freedom of a person on parole or probation cannot be taken away without cause, the state has created a liberty interest that cannot be taken away without the process due under the Fourteenth Amendment. See Gagnon v. Scarpelli, 411 U.S. 778, 93 S.Ct. 1756, 36 L.Ed.2d 656 (1973); Morrissey v. Brewer, 408 U.S. 471, 92 S.Ct. 2593, 33 L.Ed.2d 484 (1972). Similarly here, we know from Doe that the New Jersey Constitution gives Tier 2 and Tier 3 registrants the right to be free from Tier 2 and Tier 3 notification absent a showing of an overriding state interest. The New Jersey Supreme Court there held not only that Tier 2 and Tier 3 registrants had a right to the procedural due process guaranteed by the New Jersey Constitution, but also that they had a substantive right under that Constitution to be free of the disclosures required by Megan's Law, absent a demonstration that such disclosures are required by a legitimate and substantial state interest.26 As the court explained:

With its declaration of the right to life, liberty, and the pursuit of happiness, Article I, § 1 of the New Jersey Constitution encompasses the right of privacy.... We have found a constitutional right of privacy in many contexts, including the disclosure of confidential or personal information. Hennessey v. Coastal Eagle Point Oil Co., 129 N.J. 81, 96, 609 A.2d 11 (1992) (citing In re Martin, 90 N.J. 295, 447 A.2d 1290 (1982)).

In resolving conflicts between the government's need for information and the individual's right of confidentiality, this Court has adopted a balancing test similar to that adopted by the federal courts. Martin, supra, 90 N.J. at 318, 447 A.2d 1290. We concluded, in Martin, that " 'even if the governmental purpose is legitimate and substantial ... the invasion of the fundamental right of privacy must be minimized by utilizing the narrowest means which can be designed to achieve the public purpose.' " Ibid. (quoting Lehrhaupt v. Flynn, 140 N.J.Super. 250, 262, 264, 356 A.2d 35 (App.Div.1976), aff'd o.b., 75 N.J. 459, 383 A.2d 428 (1978))....

B. Standards For Determining The Process Due

Having concluded that Tier 2 and Tier 3 registrants are entitled to due process under the Fourteenth Amendment of the federal Constitution, we turn to the issue of what process is due them. Appellants contend that two procedural protections are due that are absent from the Megan's Law scheme. They insist that due process requires both that the burden of persuasion at a Megan's Law hearing be on the state rather than the registrant, and that the state's burden at such a hearing be to demonstrate the propriety of the tier classification and the notification plan by clear and convincing evidence.28

Mathews v. Eldridge, 424 U.S. 319, 96 S.Ct. 893, 47 L.Ed.2d 18 (1976), provides the framework we must apply to analyze both the burden of persuasion claim--whether it is the state or the registrant who must persuade the court on the material points--and the standard of proof claim--whether, if the burden of persuasion is on the state, the state must prove its case by a preponderance or by clear and convincing evidence. As Mathews teaches:

[D]ue process is flexible and calls for such procedural protections as the particular situation demands. Morrissey v. Brewer, 408 U.S. 471, 481, 92 S.Ct. 2593, 2600, 33 L.Ed.2d 484 (1972)....

More precisely, our prior decisions indicate that identification of the specific dictates of due process generally requires consideration of three distinct factors: First, the private interest that will be affected by the official action; second, the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards; and finally, the Government's interest, including the function involved and the fiscal and administrative burdens that the additional or substitute procedural requirement would entail.

Mathews, 424 U.S. at 334-35, 96 S.Ct. at 903.

The Supreme Court has twice applied the Mathews test in the specific context of a challenge to the preponderance of evidence standard of proof. Santosky v. Kramer, 455 U.S. 745, 102 S.Ct. 1388, 71 L.Ed.2d 599 (1982), posed the issue of whether due process requires the state to prove its case in a termination of parental rights proceeding by clear and convincing evidence, rather than merely by a preponderance of evidence. Addington v. Texas, 441 U.S. 418, 99 S.Ct. 1804, 60 L.Ed.2d 323 (1979), presented the issue of the state's burden in a civil commitment proceeding. In each instance, the Court, in addition to identifying the private and public interests at stake and evaluating the relative risk of error in the particular kinds of proceedings involved, addressed whether the standard employed "fairly allocates the risk of an erroneous factfinding between the[ ] parties." Santosky, 455 U.S. at 761, 102 S.Ct. at 1399. As the Santosky Court explained:

Addington teaches that, in any given proceeding, the minimum standard of proof tolerated by the due process requirement reflects not only the weight of the private and public interests affected, but also a societal judgment about how the risk of error should be distributed between the litigants.

Id. at 755, 102 S.Ct. at 1395.

In both Santosky and Addington, the Court held that due process required the state to carry the burden of persuasion by more than a preponderance of the evidence, since the preponderance standard requires litigants to "share the risk of error in roughly equal fashion." Addington, 441 U.S. at 423, 99 S.Ct. at 1808. Neither a person threatened with a termination of parental rights nor one standing in jeopardy of a civil commitment "should ... be asked to share equally with society the risk of error when the possible injury to the individual is significantly greater than any possible harm to the state." Id. at 427, 99 S.Ct. at 1810.

We first address whether the Due Process Clause permits New Jersey to allocate the burden of persuasion in a Megan's Law proceeding to the registrant. We begin, as Mathews directs, by identifying the private and public interests involved. The private interests that will be affected by the state's notification in Tier 2 and Tier 3 cases if the outcome of the hearing is in the state's favor are very substantial. Notification puts the registrant's livelihood, domestic tranquility, and personal relationships with all around him in grave jeopardy. This jeopardy will not only extend to virtually every aspect of the registrant's everyday life, it will also last at least 15 years. As the New Jersey Supreme Court recognized in Doe, a registrant thus has a compelling interest in an accurate and reasonable disposition of the issues before the court in a Megan's Law hearing.

The state, on the other hand, has a compelling interest in protecting its citizens by giving prompt notification to potential victims and relevant caregivers with respect to registrants who are accurately determined to be Tier 2 or Tier 3 risks. New Jersey thus has a compelling interest in having an expedited, summary process. However, the state also has an interest in ensuring that its classification and notification system is both fair and accurate. Put conversely, the state has no substantial interest in notifying persons who will not come into contact with the registrant; nor has it any interest in notifying those who will come into contact with a registrant who has erroneously been identified as a moderate or high risk. Cf. Santosky, 455 U.S. at 766, 102 S.Ct. at 1401-02 (noting that even when the state seeks to terminate parental rights, it shares an interest with the parents in an accurate factfinding as to their fitness).

Given the respective private and public interests at stake in a Megan's Law hearing, we conclude that the burden of persuasion must be placed on the state if, compared to proceedings in which that burden is on the registrant, the risk of error will be materially reduced without materially impairing the state's ability to secure a prompt determination and without imposing substantial new administrative burdens on the state.

Our next step, then, is to identify the issues that are before the court in a Megan's Law hearing--a necessary predicate for assessing any potential reduction in the risk of error. When a challenge is mounted to a tier classification, the issues for resolution by the court are limited to: (1) whether the Scale has been accurately applied to the facts of the case in accordance with its terms; and (2) whether there is something extraordinary about the particular registrant's case that takes it out of the "heartland" of the cases within the scope of the tier that would otherwise be indicated. In re G.B., 685 A.2d at 1264. If the registrant challenges not just the tier classification but also the reasonableness of the prosecutor's notification plan, the court must also exercise a judgment about whether the scope of the proposed notification is appropriate to the risk presented by the particular registrant.

Since the validity of the Scale as a risk assessor must be accepted by the court, resolution of the first issue primarily requires factfinding--albeit factfinding in a context that poses more than a normal risk of error, as we explain hereafter. Resolution of a registrant's contention that his case is outside the "heartland" or that the notification proposed is excessive, on the other hand, necessarily involves a subjective judgment by the court regarding the degree and nature of the risk posed by the particular registrant. Thus, in resolving these issues, the court is necessarily required to assess future dangerousness. While a state is clearly entitled to require a court to undertake such an assessment, it is an undertaking involving substantial uncertainty. See Randy K. Otto, On the Ability of Mental Health Professionals to "Predict Dangerousness": A Commentary on Interpretation of the "Dangerousness" Literature, 18 Law & Psychol. Rev. 43, 45, 62-63 (1994) (noting that researchers in the 1970s began compiling data showing that the presumption that professionals could predict violent behavior was incorrect, but also noting that recent data showed "some" predictive ability); American Psychiatric Association, Report of the American Psychiatric Association Task Force on Clinical Aspects of the Violent Individual 20 (1974) (concluding that "[n]either psychiatrists nor anyone else have demonstrated an ability to predict future violence or dangerousness").

Resolution of factual issues made relevant by the Scale is of critical importance to the outcome at a Megan's Law hearing. These issues include the circumstances of the crime that has required registration as well as other criminal conduct in which the registrant has allegedly engaged. See, e.g., In re C.A., 679 A.2d at 1153. In every hearing, the court will be called upon to find facts relating to the circumstances of a sex offense of which the registrant has been convicted. Many of these facts will not have been determined by the trier of fact in the criminal proceeding. Since a prosecutor may also rely on conduct for which the registrant has not been convicted, the court will be called upon in some proceedings to determine the circumstances of sex offenses that have never been the subject of a criminal proceeding. Sex offenses are almost always committed in private. This means that potential witnesses with relevant knowledge of whether, and if so how, an alleged sex offense occurred are generally limited to the victim and the alleged offender. One can therefore confidently predict that there will frequently be issues of importance in Megan's Law hearings where the information available to the court will be limited to the victim's word against the word of the alleged offender.

These issues must be resolved in a proceeding in which the rules of evidence do not apply. The prosecutor may base her case entirely on hearsay, if it shows indicia of reliability. Most importantly, these issues must be resolved in a proceeding in which the registrant cannot compel testimony from the victim without the approval of the court, and the court must follow the following admonition of the New Jersey Supreme Court:

The trial courts should only seek to compel such testimony when there is a real need for the testimony that cannot be met in an alternative manner. We expect that only in the rarest of cases will a court compel the testimony of a victim. In those cases, we suggest that, when possible, the trial court itself conduct all questioning of the victim.

Id., at 1166.

Accurate factfinding is also made more difficult by the timetable on which the proceedings must be conducted. Because of the public interest in a prompt resolution of the issues posed in any such proceeding, the Supreme Court of New Jersey has ordered that the time from the date of notice to a registrant until the time of trial court decision on the tier classification should not exceed 40 to 45 days. See Supreme Court of New Jersey, Outline of Procedure for Hearings on Objections to Megan's Law Tier 2 and Tier 3 Classification and Manner of Notification Determinations p I. While clearly justified, this requirement does substantially constrict both sides in their preparations for the truthseeking hearing process.

Courts are human institutions and there is, of course, risk of error in every judicial proceeding. Given the nature of the issues typically presented in a Megan's Law hearing and the process established by New Jersey for resolving them, however, we believe the risk of error in such a hearing is substantially greater than that in a typical civil damage suit. We further conclude that, in this context, the allocation of the burden of persuasion is of critical importance and the assignment of that burden to the prosecutor will substantially reduce the risk of an erroneous outcome.

When the court in a Megan's Law hearing simply cannot tell which of two conflicting accounts (regarding the use of force, perhaps) represents the historical truth, allocation of the burden of persuasion is likely to be outcome determinative. The same is true in those cases in which the trier of fact finds inconsistencies or implausible elements in the victim's account, but, at the same time, is inclined to discount the registrant's account because of his criminal history or the enormity of his stake in the outcome. In these cases, as well as others in which the trier of fact discounts the account of the registrant for similar reasons, requiring the prosecutor to affirmatively convince the court of the important facts can be expected to materially reduce the risk of error.

Finally, we must consider whether an allocation of the burden of persuasion to the state would materially impair the state's ability to receive a prompt determination or would impose new administrative burdens on it. Under the current procedural scheme, the state has the burden of presenting a prima facie case. This means, of course, that the prosecutor is already required to marshal and tender evidence that, if believed, will establish the facts she relied upon. While allocation of the burden of persuasion to the state may motivate the prosecutor to utilize live testimony rather than affidavits where substantial credibility issues are anticipated, this would not appear to impose a substantial administrative burden. Moreover, we perceive no reason to predict that any possible increase in the utilization of live witnesses would materially impair the prosecutor's ability to meet her responsibility under New Jersey's 45 day timetable for Megan's Law determinations.

Given (1) the interest of the registrant and the state in an accurate determination of the relevant issues of fact in a Megan's Law hearing, (2) the absence of a substantial economic or other burden to the state from allocating the burden of persuasion to it, and (3) our conclusion that such an allocation will materially reduce the risk of error in those cases in which the allocation of that burden plays a role, we hold that due process requires that the prosecutor shoulder the burden of persuading the court of the truth of the facts upon which she has relied.D. Extent Of The State's Evidentiary Burden

The remaining issue is whether due process requires that the prosecutor prove her facts by clear and convincing evidence, as appellants claim, rather than merely by a preponderance of evidence. Santosky and Addington require that we address whether, in light of the relative importance of the private and public interests at stake and the impact on those interests of an erroneous determination, the preponderance of evidence standard fairly allocates the risk of error between the parties. This requires "a very fundamental assessment of the comparative social costs of erroneous factual determinations" in the context of Megan's Law proceedings. In re Winship, 397 U.S. 358, 370, 90 S.Ct. 1068, 1075, 25 L.Ed.2d 368 (1970) (Harlan, J., concurring). Based on such an assessment, we must determine whether registrants can fairly "be asked to share equally with society the risk of error." Addington, 441 U.S. at 427, 99 S.Ct. at 1809.

We have previously identified the private and public interests at stake in a Megan's Law proceeding. For present purposes, it is important to add that the impact of an erroneous determination on those interests is significantly dissimilar. An erroneous underestimation of an individual's dangerousness will not necessarily result in harm to protected groups. Registration alone, which Megan's Law mandates regardless of an offender's classification, allows law enforcement officials to monitor offenders and provides considerable disincentive to offenders to commit criminal acts because of the high likelihood of being apprehended. On the other hand, an overestimation of an individual's dangerousness will lead to immediate and irreparable harm to the offender: his conviction becomes public, he is officially recorded as being a danger to the community, and the veil of relative anonymity behind which he might have existed disappears.

In this context, we find Addington to be the most helpful authority. The civil commitment statute involved there required the court to determine, inter alia, whether "hospitalization [of the individual who was the subject of the proceeding] in a mental hospital [was required] for his own welfare and protection or the protection of others." Id. at 420, 99 S.Ct. at 1806. Thus, there, as here, the trier of fact was required to predict future dangerousness. Moreover, the interests which the state had in civil commitment proceedings under the statute were similar to the interests of the state here--protecting the public from violence--and the risks attending an erroneous finding against the state are, therefore, similar. Additionally, registrants share with individuals facing civil commitment an important interest that was stressed in Addington. As the Court explained:

[I]t is indisputable that involuntary commitment to a mental hospital after a finding of probable dangerousness to self or others can engender adverse social consequences to the individual. Whether we label this phenomena "stigma" or choose to call it something else is less important than that we recognize that it can occur and that it can have a very significant impact on the individual.

Id. at 425-26, 99 S.Ct. at 1809.

It is true, as the state points out, that a registrant in a Megan's Law proceeding does not face the same restrictions on his physical freedom that a potential committee faces in civil commitment proceedings. It is clear from Santosky, however, that due process requires a clear and convincing standard even in the absence of a threat of physical restraint when the "loss threatened by [the] particular type of proceeding is sufficiently grave to warrant more than average certainty on the part of the factfinder." Santosky, 455 U.S. at 758, 102 S.Ct. at 1397. As the Court observed in Santosky:

This Court has mandated an intermediate standard of proof--"clear and convincing evidence"--when the individual interests at stake in a state proceeding are both "particularly important" and "more substantial than mere loss of money." Addington v. Texas, 441 U.S. at 424, 99 S.Ct. at 1808. Notwithstanding "the state's 'civil labels and good intentions,' " id. at 427, 99 S.Ct. at 1810, quoting In re Winship, 397 U.S. at 365-66, 90 S.Ct. at 1073-74, the Court has deemed this level of certainty necessary to preserve fundamental fairness in a variety of government initiated proceedings that threaten the individual involved with "a significant deprivation of liberty" or "stigma." 441 U.S. at 426, 99 S.Ct. at 1809. See, e.g., Addington v. Texas, supra (civil commitment); Woodby v. INS, 385 U.S. at 285, 87 S.Ct. at 487-88 (deportation); Chaunt v. United States, 364 U.S. 350, 353, 81 S.Ct. 147, 149-50, 5 L.Ed.2d 120 (1960) (denaturalization); Schneiderman v. United States, 320 U.S. 118, 125, 159, 63 S.Ct. 1333, 1336-37, 1353, 87 L.Ed. 1796 (1943) (denaturalization).

Santosky, 455 U.S. at 756, 102 S.Ct. at 1396.

We must, therefore, ask whether the preponderance of evidence standard, which "allocates the risk of error nearly equally" between an erroneous overestimation or underestimation of a registrant's future dangerousness, "reflect[s] properly the [ ] relative severity" of these erroneous outcomes. Id. at 766, 102 S.Ct. at 1401. Addington supplies the answer. Because "the possible injury to the individual [registrant] is significantly greater than any possible harm to the state," the registrant, consistent with due process, cannot "be asked to share equally with society the risk of error." 441 U.S. at 427, 99 S.Ct. at 1809. It necessarily follows that the Due Process Clause requires that the state prove its case by clear and convincing evidence in a Megan's Law proceeding.

In reaching this conclusion, we have not been unmindful of the Supreme Court's decision in McMillan v. Pennsylvania, 477 U.S. 79, 106 S.Ct. 2411, 91 L.Ed.2d 67 (1986). In McMillan, a Pennsylvania sentencing statute set a mandatory minimum for those convicted of a crime while using a visible firearm. The issue there was whether a "sentencing consideration," such as the presence of a firearm, need be proven by more than a preponderance of evidence. The Court held, affirming a long tradition, that proof by a preponderance of evidence is sufficient in the sentencing context. Here, however, we are not dealing with sentencing. Sentencing occurs during and is part of the criminal proceeding; its purpose is to specify the sanction to be imposed as a result of one's conviction by proof beyond a reasonable doubt. A Megan's Law hearing, by contrast, is a civil proceeding that stands apart from the criminal proceeding in which one was convicted and sentenced. See C.A., 679 A.2d at 1164.29 Moreover, as we have discussed, the factual determinations required in a Megan's Law hearing are of greater complexity than those typically involved in sentencing. Accordingly, we conclude that it is entirely consistent with McMillan to require a higher standard of proof in a Megan's Law proceeding.

Application of Megan's Law to the class certified by the district court will not violate the Ex Post Facto or Double Jeopardy Clauses of the Constitution. The Due Process Clause, however, would be violated by any Tier 2 or Tier 3 notification that occurred without a prior opportunity to challenge the registrant's classification and notification plan in a hearing at which the prosecutor has the burden of persuasion and must prove her case by clear and convincing evidence. Accordingly, the judgment of the district court will be reversed and this matter will be remanded with instructions (1) to enter an injunction foreclosing notification in Tier 2 and Tier 3 cases without compliance with these requirements of procedural due process, and (2) to deny any further relief.

BECKER, Circuit Judge, concurring and dissenting with respect to Parts I-V of the majority's opinion, which primarily discuss the question whether the notification provisions of the challenged statute violate the Ex Post Facto Clause or the Double Jeopardy Clause; and concurring in Part VI of the majority's opinion regarding the process due a registrant at a tier classification hearing.

                               TABLE OF CONTENTS
  I. DO THE NOTIFICATION PROVISIONS OF MEGAN'S LAW CONSTITUTE PUNISHMENT?  1113
     A.   Introduction ................................................... 1113
     B.   The History Subpart of the Artway Test; Overview ............... 1114
     C.   Historical Analogues to Notification Provisions ................ 1115
          1.   The Applicability of Criden ............................... 1115
          2.   Analogy to Shaming Punishments ............................ 1115
          3.   Warning Posters, Wanted Posters, and Quarantine Notices
                 Compared ................................................ 1116
          4.   The Mechanism of Notification; Its Relation to the Choice
                 of Historical Analogues ................................. 1117
          5.   Summary: Shaming Punishments as the Best Analogy .......... 1119
     D.   Does the Text, Legislative History, or Design of the
            Notification Provisions Demonstrate That They are not
            Punitive? .................................................... 1119
          1.   Introduction; The Role of Law Enforcement ................. 1119
          2.   Promoting the Aims of Punishment .......................... 1120
          3.   Excessiveness ............................................. 1121
          4.   Summary of "Design" ....................................... 1122
     E.   Notification Fails the History Subpart of Artway ............... 1122
 II.  EFFECTS ............................................................. 1122
     A.   Introduction ................................................... 1122
     B.   Methodology: The Proper Standard for Evaluating Effects ........ 1123
     C.   Actual Effects ................................................. 1125
     D.   Summary ........................................................ 1126
III.  THE "CLEAREST PROOF" DOCTRINE ....................................... 1126
 IV.  CONCLUSION .......................................................... 1128

The societal pressure for legislation designed to prevent terrible tragedies such as befell Megan Kanka and her parents is hydraulic. The pressure is understandable, for Americans are a fundamentally decent people, and legislation such as Megan's Law is thus the product of good intentions. Unfortunately, however, earthly life is fraught with so much uncertainty that we cannot legislate against the vagaries of chance. In their desire to make everything right, legislators sometimes overlook this basic fact, and enact laws that not only fail to achieve their laudable ends, but also cause serious harm. This appeal involves a textbook example of that phenomenon.

I do not quarrel with much of what the majority has said in that portion of its fine opinion dealing with the definition of punishment. Importantly, I agree with its conclusion that our decision in Artway v. Attorney General of New Jersey, 81 F.3d 1235 (1996) remains viable even in the wake of Kansas v. Hendricks, --- U.S. ----, 117 S.Ct. 2072, 138 L.Ed.2d 501 (1997), and United States v. Ursery, --- U.S. ----, 116 S.Ct. 2135, 135 L.Ed.2d 549 (1996), and that we must apply the Artway test for what constitutes punishment under the Double Jeopardy Clause and the Ex Post Facto Clause to determine the validity of the challenged statute. Where we part company is over the second prong of the Artway test--objective (legislative) intent, a consideration that is heavily freighted with history. History, a consideration deemed by the Supreme Court to be the barometer of legitimacy in so many constitutional cases, is telling in the resolution of this appeal.

Tacitly recognizing the deep, historic roots of the plaintiff's double jeopardy and ex post facto claims, the majority deftly tiptoes over the "stocks, cages, and scarlet letters referenced by appellants," and asserts that other forms of the dissemination of information generated by our criminal justice system "constitute far more compelling analogies" to Megan's Law. But the majority's analogies are misplaced. Warning or wanted posters and quarantine notices provide very different information from that provided by the community notification provisions of Megan's Law. Instead, the more apt historical antecedents to notification can be found in the shaming punishments of colonial America, which were indubitably and unabashedly punitive. Moreover, nothing in the design or operation of the notification provisions of Megan's Law contradicts this historical understanding. Because the history of notification evidences an objective punitive intent, and because the design or operation of notification does not negate this objective intent, the notification provisions of Megan's Law must be considered punishment under Artway 's second prong. I therefore dissent from the majority's conclusion in Part V that Megan's Law passes constitutional muster.

Failure to meet the second prong of the Artway test is fatal to the statute, and hence I do not ground my dissent on Artway 's third prong dealing with the "effects" of notification. However, because of the relevance of effects to application of the presumption in favor of subjective legislative intent over objective manifestations of that intent, and because of the general importance of the issue, which I think is much closer than the majority describes it, I report my conclusion that its treatment of the "effects" prong of Artway is quite problematic. More specifically, its novel holding that nothing short of the deprivation of a sufficiently fundamental interest can give rise to an effect that would constitute punishment is, I believe, incorrect. Further, the majority improperly narrows the Artway effects test both procedurally and substantively.

I join in those parts of the majority's opinion finding a Rooker-Feldman bar to our review of E.B.'s challenge, and declaring unripe the challenge to the state's authority to dispense with notice prior to tier classification hearings in emergency situations. I also join in Part VI of the majority's opinion holding that the due process clause forbids the imposition of the burden of persuasion at a Megan's Law tier classification hearing on the offender and that that burden should be by clear and convincing evidence.

Finally, while I recognize that there is arguably a strong presumption favoring the subjective intent of a legislature in determining whether a measure is punitive such that only the clearest proof of objective intent will undermine that subjective intent, I believe that application of this standard in the present context is misplaced. The purpose of the standard is to determine legislative intent. There is thus no need to apply the standard here because the historical antecedents to notification provisions make that intent patent. Put slightly differently, assuming that such a standard does apply, I believe that the history of notification, the design of notification provisions in Megan's Law, and the effects of notification provide sufficient proof to show an objective punitive intent, notwithstanding the subjective intent to the contrary. In other words, the objective manifestations of the legislative intent evidence a punitive purpose.

I. DO THE NOTIFICATION PROVISIONS OF MEGAN'S LAW CONSTITUTE

PUNISHMENT?

The central issue in the case, as the majority's opinion makes clear, is whether the notification provisions of Megan's Law constitute punishment.1 If not, then neither the Ex Post Facto Clause nor the Double Jeopardy Clause is implicated.

Like the majority, I believe that Artway provides the proper legal standard to govern whether notification is to be considered punishment. I join in its lucid explanation as to why the Artway test survives the recent Supreme Court cases in Ursery, and in Hendricks. I therefore turn to how the Artway test applies to the notification provisions of Megan's Law. Because, as the majority explains, our conclusion in Artway that the actual purpose of the registration provisions of Megan's Law is non-punitive effectively requires us to conclude that the actual purpose of the notification provisions is similarly non-punitive (satisfying Artway 's first prong), I proceed to Artway 's second prong.

B. The History Subpart of the Artway Test; Overview

I begin with the so-called history subpart of the objective purpose prong of the Artway test. Pursuant to this subpart, if historical analysis shows that the measure in question has been regarded as punishment, and if the text or legislative history of the measure does not negate this traditional understanding, we must consider the measure punitive. See Artway, 81 F.3d at 1263.

It is here that I most disagree with the majority's opinion. In particular, I believe that it incorrectly frames the historical analysis, first by relying on immaterial precedent, and then by applying insufficiently comparable historical analogues to the notification provisions of Megan's Law. A better reasoned analysis likens the notification provisions of Megan's Law to the shaming punishments of colonial America--the scarlet letters of literary fame--leaving no doubt that the objective purpose of these provisions is punitive. The discussion may be put in proper perspective by looking at Supreme Court jurisprudence, and we do not have to look beyond a few of the Supreme Court cases decided in the last month of the October 1996 term. These cases are suffused by references to and reliance on historical analysis.

Perhaps the most apposite case is Hendricks itself, where the Court made use of the historical understanding of a measure to determine whether it was punitive. Holding that a Kansas civil commitment statute was not punitive, the Court noted that restricting the freedom of the dangerously mentally ill "is a legitimate non-punitive governmental objective and has been historically so regarded." Hendricks, at ----, 117 S.Ct. at 2083.

In Richardson v. McKnight, --- U.S. ----, 117 S.Ct. 2100, 138 L.Ed.2d 540 (1997), holding that prison guards employed by private prison management firms do not enjoy qualified immunity, the Court looked to the historical traditions of immunity applicable to privately employed prison guards, going so far as to examine the operation of jails in medieval England. Similarly, in Washington v. Glucksberg, --- U.S. ----, 117 S.Ct. 2258, 138 L.Ed.2d 772 (1997), canvassing the historical treatment of suicide and assisted suicide, the Court noted that "for over 700 years, the Anglo-American common-law tradition has punished or otherwise disapproved of both suicide and assisting suicide." Id. at ----, 117 S.Ct. at 2263.

In Printz v. United States, --- U.S. ----, 117 S.Ct. 2365, 138 L.Ed.2d 914 (1997), the Court relied heavily on history in analyzing the Brady Handgun Violence Prevention Act, surveying statutes and executive actions from the earliest days of the Republic to determine whether the federal government can constitutionally require states to execute federal regulatory laws. And, in Reno v. ACLU, --- U.S. ----, 117 S.Ct. 2329, 138 L.Ed.2d 874 (1997), in striking down provisions of the Communications Decency Act of 1996, the Court looked to more modern history and distinguished the Internet from radio broadcasts, noting that the Internet has no history of limited First Amendment protections.

Although I have cited only a few examples, these recently decided cases make patent that history plays a vital role in constitutional adjudication. See John Paul Stevens, A Judge's Use of History, 1989 Wis. L.Rev. 223, passim (suggesting that the interaction of history and the law is ripe for study and discussing the use of historical analysis in three cases).2C. Historical Analogues to Notification Provisions

The majority's treatment of history focuses first on United States v. Criden, 648 F.2d 814 (3d Cir.1981). There, we rejected a suggested analogy between the media rebroadcast of material placed into evidence at a criminal trial and the shaming punishments of colonial America. See id. 648 F.2d at 825. The majority essentially argues that the notification provisions at issue here are like the challenged re-broadcast in Criden. It reasons that, because the challenged re-broadcast is unlike the shaming punishments, the notification provisions must also be unlike them, and hence the purpose of the notification provisions does not correspond with the purpose of the shaming punishments. I believe that the majority's reliance on Criden is misplaced. The notification provisions in Megan's Law are different from the re-broadcast in Criden in a number of respects. The most striking is that the rebroadcast itself is carried out by the private media (who obtained the information from the state), whereas the notification provisions are carried out by the state.

This distinction makes all the difference. Here, New Jersey not only made the criminal history of convicted sex offenders publicly available, it also instituted an affirmative, state-run program to disseminate that information. In one imaginable scenario, New Jersey could simply have allowed private entities to disseminate this public information about the offender's record. The state, however, chose to take the additional step of disseminating the information itself. The purpose of that deliberate decision is at issue here. To determine the objective purpose of the notification provisions, we must look to measures in which the dissemination of criminal history information is state-run, not to measures in which the dissemination occurs independently from state action. For the same reasons that Criden is inapposite, the majority's discussion of public indictments, public trials, the public imposition of sentence, and "rap sheets" is unhelpful. In none of these cases is the state itself actively disseminating information.

In contrast, the state-run dissemination of criminal history information is central to the operation of shaming punishments.3 As a result, shaming punishments are, as a group, measures that should be considered in our historical analysis. By shaming punishments, I mean a variety of punishments, common in the American colonies, ranging from the admonition to branding or maiming and banishment.

In an admonition, a magistrate or clergyman would lecture the offender privately about his misdeeds and seek his repentance. See Adam J. Hirsch, From Pillory to Penitentiary: The Rise of Criminal Incarceration in Early Massachusetts, 80 Mich. L.Rev. 1179, 1224 (1982). The offender would then offer a public apology before the community either in court, see id., or in church, see Toni M. Massaro, Shame, Culture, and American Criminal Law, 89 Mich. L.Rev. 1880, 1912-13 (1991). Once forgiven by the community, the offender was drawn back into its fold, thereby restoring a moral order upset by the offense. See id. at 1913. In a similar vein, offenders were often forced to stand in a public place for a time displaying their offense written on their clothing or on their bodies. See Lawrence M. Friedman, Crime and Punishment in American History 38 (1993). This public display took many forms in the colonies, and might include "[s]entences to the pillory, to the stocks, to lashes at the whipping-post, and to hours on the gallows with a rope around the neck." Hirsch, supra, at 1225.

A court might also require that the offender permanently display a label representing his offense--for instance, a scarlet letter "A" for the crime of adultery. See Massaro, supra, at 1913. Such labels were generally cut from cloth and sewn on the offender's outer garments. See Friedman, supra, at 40. Another form of permanent labeling was the brand, in which the authorities burned a label directly onto the offender's body. See id. A murderer might be branded with an "M", a thief with a "T". See Jon A. Brilliant, Note, The Modern Day Scarlet Letter: A Critical Analysis of Modern Probation Conditions, 1989 Duke L.J. 1357, 1361. Colonial New Jersey, for example, punished burglary by branding the offender's hand for a first offense, and his forehead for subsequent offenses. See id.

Mutilation or maiming was similar to branding except that it did not necessarily signal the precise offense the offender had committed. See Friedman, supra, at 40. A common form of mutilation or maiming was the detachment of an ear. See id. The effect of branding, mutilation, or maiming was often to cast the offender out of society once and for all. See Hirsch, supra at 1228. Unlike the aftermath of an admonition, the community did not welcome the offender back into society, but shunned him. See id. Just a short step away from branding, mutilation, and maiming, was banishment, a forced exclusion from the community. Banishment was reserved for those who presented "a permanent danger" or who engaged in "repeated criminality." See Friedman, supra, at 40. "Those who would not repent, those who could not be regathered into the bosom of society, had to be driven out." Id.

These various punishments were effective--they had "sting"--because each punishment publicized accurate information about the offender's misdeeds, and because of the cultural milieu of a colonial settlement. A settlement in the American colonies was "small-town life at its most communal." Id. at 37. Thus, once an offense had been made public, the entire community became aware of it, and such publicity was the cause of shame. In addition, colonists have been described as ultra sensitive to criticism from their fellows, see Massaro, supra, at 1912, especially from those whom they knew and respected, see Hirsch, supra, at 1233-34. Therefore, the public dissemination of the fact that a community member had committed some offense was especially calculated to provoke shame in the wrongdoer.

As the majority recognizes, these shaming punishments often were imposed in addition to physical punishments, such as whippings. See Friedman, supra, at 40. However, the physical punishment did not lessen the "sting" associated with the public dissemination of the offense. See Hirsch, supra, at 1232 ("A sentence to whipping or the pillory had worked primarily through the media of psychic pain and shame."). In fact, the physical punishment was seen as being effective only insofar as it resulted in the offender being shamed by the publicity of his offense. See id. at 1233-34 ("As the aura of shame and psychic trauma surrounding the penalty evaporated, there was left behind only a small core of physical pain quite insufficient to prevent offenses."). Further, the authorities often dispensed with physical punishments altogether because the "sting" associated with the publicity was more sharp. See id. at 1226. Finally, some shaming punishments had no physical components at all; for instance, those condemned to display a label representing their offense did not necessarily suffer physical punishment.

3. Warning Posters, Wanted Posters, and Quarantine Notices Compared

As the foregoing discussion makes clear, shaming punishments are analogous to the notification provisions contained in Megan's Law. But that does not end our inquiry. Other measures also rely on the state dissemination of information. The majority mentions two such measures: warning or wanted posters, and quarantine notices.4 The question, then, is whether the notification provisions of Megan's Law are more like the shaming punishments or more like warning or wanted posters, or quarantine notices. The difficulty in pinpointing the proper historical analogy to notification provisions lies in identifying the characteristics of such provisions (in addition to the state dissemination of information) that will help us to determine their objective purpose. This analysis is further complicated, at least in part, by the fact that "[u]nquestionably punitive statutes share traits with laws that are universally accepted as [remedial]." Note, Prevention Versus Punishment: Toward a Principled Distinction in the Restraint of Released Sex Offenders, 109 Harv. L.Rev. 1711, 1725 (1996).5

The majority ignores an important component of the shaming punishments when it reasons that warning or wanted posters and quarantine notices are more apt analogies to notification provisions than are the shaming punishments. The majority essentially believes that the characteristic that notification provisions share with warning or wanted posters and with quarantine notices (and the characteristic that is sufficient warrant to justify its analogy) is the general type of information that each measure disseminates. In other words, it reasons that, because notification provisions alert the community to a risk, they must be akin to warning or wanted posters and to quarantine notices which similarly alert the community to a risk.

The problem with this reasoning is that--like the warning or wanted posters, and quarantine notices--the shaming punishments also alerted the community to a risk, the risk that the offender would re-offend. See, e.g., Hirsch, supra, at 1228 ("Less common, but equally effective, were branding and mutilation, punishments that fixed upon the offender an indelible 'mark of infamy,' to warn community members to keep their distance."); Massaro, supra, at 1913 ("Branding and maiming also were designed in part to prevent the offender from committing future similar acts, either by warning future victims of their criminal propensities or by disabling the offender." (footnote omitted)).

Moreover, notification is a judicially endorsed pronouncement that the registrant presents a danger to the community. In that sense, notification is closer to the shaming punishments than to warning or wanted posters, or quarantine notices. There is no judicial involvement in the issuance of warning or wanted posters, or of quarantine notices. Judicial endorsement, by a disinterested magistrate, is different in kind from a determination by other public agencies. Therefore, one cannot contend that warning or wanted posters and quarantine notices are better analogies to the notification provisions than are the shaming punishments simply because warning or wanted posters and quarantine notices alert the community to a risk.

New Jersey argues that, because the purpose of the notification provisions and that of measures such as warning or wanted posters, or quarantine notices is to alert the community to some danger (a remedial purpose), we must consider notification provisions historically analogous to these other measures. As the foregoing discussion suggests, this argument is flawed. The goal of the history subpart of the Artway test was to determine the objective purpose of a particular measure by examining the historical understandings of analogues to the measure in question. It does not comport with this methodology to choose as data points historical analogues based solely on the claim that they served the same purpose as the measure in question. To do so would assume the very issue that is in dispute, and then reach a conclusion by way of circular reasoning.

4. The Mechanism of Notification; Its Relation to the

Because we are concerned with the objective purpose of the state dissemination of information, we must examine the particular information the state chooses to disseminate in order to determine that objective purpose. As the majority describes the process of notification, New Jersey provides recipients of notification with the following information: the identity of the convicted sex offender, his physical description, the location of his dwelling, the place of his employment, a description of his automobile, his license plate number, and the offense(s) for which he was convicted and by which the notification was triggered.6

This is the same type of information the state disseminated in carrying out the shaming punishments. Because the offender would have been well known to those who witnessed the shaming punishment, simply by placing the offender on display before the community was enough to disseminate his identity, his physical description, the location of his dwelling, and the place of his employment.7 In fact, the shaming punishments became less frequent when such information could not be conveyed by public display alone. See Hirsch, supra, at 1228-34 (describing the decreased use of shaming punishments as colonial communities grew in size thereby increasing the likelihood that the offender was a stranger to the witnesses of his punishment); see also Dan M. Kahan, What do Alternative Sanctions Mean?, 63 U. Chi. L.Rev. 591, 631 (1996) ("Early Americans turned to imprisonment in large part because they believed that existing criminal penalties had lost the power to shame.").8 Moreover, as noted above, central to many of the shaming punishments was some notice--e.g., a sign, a label, or a brand--of the offense(s) for which the offender was being punished.

In contrast, warning or wanted posters and quarantine notices do not disseminate the same type of information disseminated by notification provisions. A warning or wanted poster, displayed in an effort to catch escaped prisoners or to arrest alleged criminals, obviously does not include information about the location of the offender's current dwelling, nor of his current employment. If the authorities had this information, they would know how to apprehend the offender. Such posters also typically include information about the facts of the individual's escape in the case of a warning poster, and the facts of the individual's alleged crime in the case of a wanted poster. Quarantine notices, too, include information different from that included in notification provisions. The most prominent difference is that quarantine notices include health-related information; such notices make no mention of criminal or alleged criminal activity. Information provided pursuant to notification, then, links the registrant to some act for which he is blameworthy. Health related information is normally not related to culpability.

The state attempts to distinguish the notification provisions from the shaming punishments in terms of the scope of the notification. New Jersey makes much of the fact that the notification provisions, unlike the shaming punishments, do not involve the dissemination of information to the entire community. I believe that the state overstates the significance of this difference. Though notification under both Tier 2 and Tier 3 is intended to be limited, the design of the provisions seems to encourage more widespread dissemination. Tier 3 recipients are not warned that the information is confidential. Tier 2 recipients are so warned, but I fail to see how that warning is to be taken seriously. Under Tier 2, notification is given to the staff of organizations charged with the care or supervision of children and/or women. Such notification would effect the remedial purpose of the statute--the protection of the children and women under the care of the organizations--only if the organizations pass the notification information to the children and women under their care.

New Jersey also emphasizes that notification is tailored to the specific offender and may not occur at all. In emphasizing this aspect of notification, the state fails to appreciate fully the textured nuances of the shaming punishments. Shaming punishments were also tailored to the specific offender and often did not occur at all. For instance, permanent labeling and branding were reserved for offenders whose likelihood of re-offense was high. See Friedman, supra at 40. Only the "deep-dyed sinner" would suffer such a fate. Id. Further, shaming punishments were by no means automatic; not all offenders would be so punished. Fines or bonds for good behavior (payments made to the authorities that were forfeited should the surety commit a misdeed within a certain time period) were common punishments for lesser offenses. See Hirsch, supra at 1224. And, even for more serious offenses, an offender could often simply pay a fine and avoid a shaming punishment altogether. See Friedman, supra at 38 (describing the punishment for a woman who struck her husband as either half an hour at a town meeting with her offense written on her forehead or the payment of a fine to the county).

5. Summary: Shaming Punishments as the Best Analogy

In sum, the foregoing analysis demonstrates that the closest historical analogues to the notification provisions of Megan's Law are the shaming punishments, which were traditionally considered punitive.9 Like the shaming punishments, notification is carried out by the state. In that sense, notification is unlike measures in which the state merely allows private individuals or entities to access information and then allows those individuals to release that information more broadly. Moreover, like the shaming punishments, notification provides the community with information about the registrant's identity and physical description, place of residence, place of employment, and criminal history. Such information is judicially endorsed. The information provided by notification is different from that provided by warning or wanted posters, which do not provide information about residence and employment, and quarantine notices, which do not provide information about criminal history; none of this information is judicially endorsed. Above all notification is the functional equivalent of shaming punishments; notification publishes information about the registrant calculated to reach the entire community and likely to lead to public opprobrium.

D. Does the Text, Legislative History, or Design of the

Notification Provisions Demonstrate That They are

not Punitive?

1. Introduction; The Role of Law Enforcement

Under Artway, the notification provisions must be considered punishment provided the text or legislative history does not demonstrate that they are not punitive. I therefore turn to the question whether the text or legislative history so demonstrates. This part of the analysis requires an examination of the actual operation or design of the measure at issue. See Hendricks, at 2080, 117 S.Ct. at 2080-85 (examining the design of the Kansas civil commitment statute). It is an inquiry focused on the question whether the legislature designed the statutory scheme in such a manner so as "to contradict the historical understanding of [the measure] as punishment." Austin v. United States, 509 U.S. 602, 619, 113 S.Ct. 2801, 2810, 125 L.Ed.2d 488 (1993).

Perhaps the most striking feature of the statutory design is its placement of the tier classification determination and of the notification process squarely within the criminal justice system. The chapter that contains the registration and notification provisions is contained in the state's Code of Criminal Justice. Cf. Hendricks, at ----, 117 S.Ct. at 2080-82 (relying in part on the decision by the state of Kansas to place its Sexually Violent Predator Act within the probate code, instead of the criminal code, to conclude that the challenged measure was not a criminal proceeding). It is the Attorney General of New Jersey, a law enforcement officer, who is charged with "promulgat[ing] guidelines and procedures for the notification required" by Megan's Law. N.J. Stat. Ann. § 2C:7-8(a) (1995).

The guidelines are to be formulated with the advice of a "notification advisory council" comprised, at least in part, of professionals from various fields outside of official law enforcement, but the professionals are all involved, at least to some degree, in the criminal justice system, broadly defined, and this council provides, as its name suggests, mere recommendations. See id. § 2C:7-11. Once in place, the guidelines are to be implemented by the county prosecutors: they determine the risk that a particular offender poses for re-offending, thereby setting the tier classification, and they determine the means of providing notification. See id. § 2C:7-8(d).

As the guidelines are currently written, the county prosecutors have significant leeway both in determining the appropriate tier classification and in fashioning the proper notification plan. Application of the Registrant Risk Assessment Scale is by no means ministerial; the county prosecutors must determine whether the particular offender poses a low, moderate, or high risk to the community for each factor in the Scale. Although the Scale provides guidance to the prosecutors making this determination, it does not eliminate from the process prosecutorial evaluation. The guidelines allow prosecutors to enlist the assistance of persons outside the prosecutor's office, such as social workers or psychologists. However, the guidelines leave formulation of the notification to the considered judgment of the county prosecutors. It is up to those law enforcement officials to ensure that the notification is properly tailored to reach those at risk of being victimized by the particular offender.

Finally, law enforcement officers, whether of the municipality in which the offender intends to reside or of the state police force, provide the actual notification. See id. §§ 2C:7-6, 2C:7-7.

The operation of the statute will, moreover, promote "the traditional aims of punishment--retribution and deterrence." Kennedy v. Mendoza-Martinez, 372 U.S. 144, 168, 83 S.Ct. 554, 567, 9 L.Ed.2d 644 (1963); see Hendricks, at ----, 117 S.Ct. at 2082 ("As a threshold matter, commitment under the Act does not implicate either of the two primary objectives of criminal punishment--retribution or deterrence."). Of course, simply because a measure has the effect of promoting retribution and deterrence does not necessarily mean that its purpose was to do so. See Artway, 81 F.3d at 1255. Still, such an effect suggests that the particular measure was not designed in a way that contradicts the historical understanding of its analogues as punitive. That the notification provisions of Megan's Law promote retribution and deterrence is demonstrated as follows.

By publicizing an offender's crime to the community, notification realizes justice, see id. (explaining that retribution "does not seek to affect future conduct or solve any problem except realizing 'justice' "), in that it inflicts suffering on the offender. It is undisputed that notification results in shaming the offender, thereby effecting some amount of retribution. This suffering "serves as a threat of negative repercussions [thereby] discourag[ing] people from engaging in certain behavior." Id. It is, therefore, also a deterrent. There is no disputing this deterrent signal; the notification provisions are triggered by behavior that is already a crime, suggesting that those who consider engaging in such behavior should beware. See Doe v. Pataki, 940 F.Supp. 603, 623 (S.D.N.Y.1996) ("The Act is designed in such a fashion as to suggest that it is punitive. It contains classic indicia of a punitive scheme. Its provisions are triggered by behavior that is 'already a crime.' ").

The design inquiry is also furthered by an analysis of whether the notification provisions are excessive in relation to their stated remedial purpose. In a several important respects, they are. First, the criminal acts that, pursuant to Megan's Law, trigger registration and potentially subject an offender to notification, are over-broad. For example, kidnapping, even without a concomitant sexual offense, triggers notification, see N.J. Stat. Ann. § 2C:13-1(c)(2)(c); so, too, does consensual sexual contact that is criminalized merely because of the age of one of the participants, see, e.g., id. § 2C:14-2(a)(1), (b), (c)(5). See Doe v. Pataki, 940 F.Supp. at 623-24 (describing New York's Megan's Law as excessive because it covers individuals such as a "21-year old who engages in sexual intercourse with a 16-year old (who is not a spouse)," a person who engages in incest, and a person who restrains another under the age of 17); Kansas v. Myers, 260 Kan. 669, 923 P.2d 1024, 1042-43 (1996) (describing Kansas's Megan's Law as excessive because "[s]everal of the listed felonies [triggering registration and notification] include what otherwise might be viewed as voluntary sexual contact between two persons that is considered criminal because of the minority status of the victim and the fact that the victim is not married to the accused").

Next, notification under Tier 3 is often provided to those who simply do not need to know that there is a released sex offender nearby. Tier 3 notification is to be provided to "members of the public likely to encounter the person registered." N.J. Stat. Ann. § 2C:7-8(c)(3) (1995). But the "likely to encounter" standard does not limit notification to vulnerable populations. It is a standard based largely on geographic proximity, see Doe v. Poritz, 142 N.J. 1, 662 A.2d 367, 385 (1995), rather than whether the recipient of notification needs protection (e.g., a child) or can protect others (e.g., a parent). Under the statute, a move by a registrant into a retirement community will trigger notification of his neighbors.10

Similarly, the type of information required to be provided by the guidelines is excessive; it is information individual recipients often simply do not need to know. Individuals who receive notification learn of an offender's place of residence and his place of employment, regardless of their relative locations. If an offender does not work at a location near to his place of residence, which I suspect is not uncommon, then such information is only in part useful for protection. A recipient of notification who lives, attends school, works, or is otherwise located near to an offender's place of residence should be little concerned about the location of the offender's place of employment (and vice versa). Knowing the offender's place of residence might lessen the risk that the recipient will become a victim of the released offender; he or she can avoid the offender's house, for example. But, knowing the offender's distant place of employment offers no protective assistance to the recipient. If the person is not likely to encounter the offender at the offender's place of employment (or place of residence), why would he or she need or want to know such information?

4. Summary of "Design"

In sum, the design of the notification provisions does not contradict the historical understanding of analogues to such provisions as punitive. Notification is placed in New Jersey's criminal code and is structured and carried out by state law enforcement officials. Further, notification promotes the aims of retribution and deterrence. Finally, in important respects, notification is excessive. The particular recipients who receive notification and the type of information they receive are not carefully tailored to the remedial goals notification is intended to serve.

E. Notification Fails the History Subpart of Artway

As the foregoing discussion makes clear, the proper historical analogues to the notification provisions of Megan's Law are the shaming punishments of colonial America. Clearly punitive, such punishments evidence an objective punitive purpose for the notification provisions. The design of the notification provisions--especially the placement of the provisions in the state criminal code and the placement of the responsibility of enforcing them with law enforcement officials, the excessiveness of their operation, and their promotion of retribution and deterrence--does not negate this objective punitive purpose. Therefore, I believe Megan's Law fails the history subpart of the second prong of the Artway test and should be considered punishment. As a result, the judgment of the district court should be reversed. This conclusion is buttressed by my discussion infra at Part II.C. of the extent to which, by reason of the network of Megan's Laws throughout the nation, notification is akin to banishment, another traditional colonial measure in the nature of punishment. See supra, at Part I.C.2.11

The final prong of the Artway test concerns the actual effects of the challenged measure. According to Artway, "[i]f the negative repercussions--regardless of how they are justified--are great enough, the measure must be considered punishment." Artway, 81 F.3d at 1263.12 The analysis required under this part of the test is one of degree, and is guided by the signposts of already decided cases. See id.

The conclusions I have already reached--that Megan's Law fails the objective purpose prong of the Artway test and must, therefore, be considered punitive--might make it unnecessary for me to reach the "effects" issue. However, because of the relevance of the effects to application of the clearest proof standard on which the majority relies, see infra Part III, because I believe that the majority's effects analysis is seriously flawed, and also because the enormous importance of the case counsels that I explain why, I discuss the effects of the notification provisions. As I will demonstrate, the majority, in undertaking its own analysis, narrows the test fashioned in Artway. It does so without support, and, given the tenor of the analysis, unnecessarily. I also identify problems with its substantive discussion.

B. Methodology: The Proper Standard for Evaluating Effects

To begin, I quote from the majority's opinion: "It necessarily follows that some limit must be placed on the situations in which a measure's sting alone, despite its remedial purpose and effect, will constitute punishment under those clauses and that classification as punishment on the basis of sting alone must be reserved for cases involving deprivation of the interests most highly valued in our constitutional republic.... Interests such as these are sufficiently fundamental to our constitutionally secured liberty that state interference with them can be justified only by the most important of state interests." With the second sentence, the majority states that the line marking the boundary between a non-punitive and a punitive measure varies according to the remedial interest sought to be served by the measure. In other words, it appears that the majority is holding that the more important the remedial interest served by a particular measure the more harsh the sting of the measure's effects may be before the measure is classified as punitive. Nothing in Artway (or, for that matter, in the Supreme Court jurisprudence on which it draws) suggests such a formulation of the effects prong. To the contrary, Artway posits that a particular sting either falls on the punishment side of the line or it does not. At issue here is the particular sting, not the particular remedial interest.

The majority has thus introduced a difficult-to-apply sliding scale into an already complex test. This needless complication would render it nearly impossible to determine whether a particular sting is punishment. For example, as we know from Hawker v. People of State of New York, 170 U.S. 189, 18 S.Ct. 573, 42 L.Ed. 1002 (1898), the revocation of a license to practice one's profession is not considered punishment. However, could such a revocation be punishment if the remedial interest served by the challenged measure is relatively unimportant? If so, at what point does the importance of the remedial interests render such a revocation non-punitive? Under the majority's reading of Artway, an analysis of the effects prong requires a two-track inquiry, guided only by a few fixed points. I fear that this amorphous inquiry might lead to an elusive or protean jurisprudence, something to be avoided.

Moreover, because the other prongs of the Artway test adequately stir into the mix the remedial interests served by the particular measure, we need not examine those interests under the effects prong. The actual purpose prong examines whether the legislature subjectively intended the measure to advance remedial interests. All three subparts of the objective purpose prong require the reviewing court, to some degree, to consider the remedial interests the legislature subjectively believed it was advancing by enacting the challenged measure. Considering the stated remedial purpose under the effects prong might overemphasize that stated purpose, thereby potentially allowing diversion of attention from the actual operation of the measure.

The majority also narrows the Artway test by requiring that, at a minimum, a challenged measure act to deprive affected persons of a sufficiently fundamental interest before that measure is considered to cause punitive effects. The majority offers no support for this proposition in either logic or precedent, and I am unaware of any. Nothing in Artway (or, for that matter, in the Supreme Court jurisprudence on which it draws) suggests such a formulation of the effects prong. In addition, at least as I read the majority's opinion, defining the effects prong in this manner is unnecessary to the result. The majority apparently believes that the effects caused by notification simply are not harsh enough to classify Megan's Law as punitive. Under my reading of Artway, satisfaction of the effects prong does not require overcoming such a difficult hurdle.

I am especially concerned in this regard because of the indefiniteness of the majority's formulation. It is not apparent to me what would constitute a "sufficiently fundamental interest." Furthermore, without a clear understanding of those interests the deprivation of which might constitute punishment, I am also unsure as to whether the majority adequately defines the universe of interests that it, or I, would deem worthy of protection. In short, I fear that the majority might have left too little room to deal with unforeseen cases in this difficult area of jurisprudence.

In addition to re-formulating the Artway test, the majority also treats the effects of notification in such a manner as to minimize the impact of those effects. First, it emphasizes that the effects of which the offenders complain--e.g., isolation, public humiliation, loss of employment opportunities, and physical violence--are indirect. Although I agree that such is the case, I remonstrate against what seems to be overemphasis upon that aspect of notification for, in itself, indirectness of effects is not dispositive.

The Supreme Court addressed the question of directness in California Department of Corrections v. Morales, 514 U.S. 499, 115 S.Ct. 1597, 131 L.Ed.2d 588 (1995), the very case on which Artway bases the effects prong of its test. The Court struggled with the question whether a change in the procedures governing parole suitability hearings would effect an impact on a prisoner's expected term of confinement. See id. at 506-14, 115 S.Ct. at 1602-05. In concluding that the measure did not constitute punishment, the Court determined that the changes in the relevant procedures "create[d] only the most speculative and attenuated possibility of producing the prohibited effect of increasing the measure of punishment for covered crimes." Id. at 509, 115 S.Ct. at 1603. The Court made plain, however, that even the indirect effects of a measure could render it punitive.

Here, the indirect effects of notification are neither "speculative" nor "attenuated." In fact, notification advances the stated remedial purposes of Megan's Law only insofar as it induces many of these indirect effects. For example, public safety is enhanced if potential victims of an offender are warned to avoid him, thereby isolating him from the larger community. If the legislature were not aware that at least partial isolation would necessarily result from notification, I doubt that it would have believed that notification would serve the remedial purposes it sought to advance. And, although not necessarily vital in ensuring the efficacy of Megan's Law, other indirect effects--e.g., harassment, loss of employment opportunities, and physical violence--surely were anticipated as also being inevitable. New Jersey was not the first state to adopt notification provisions, and the experiences of other states must have informed the New Jersey legislature as it considered Megan's Law.

In other states, notification has caused harassment, loss of job opportunities, and the like. A study by the Washington State Institute for Public Policy, released in December 1993 (approximately ten months prior to the enactment of Megan's Law), reported numerous instances of harassment following notification in Washington, some quite severe, under its 1990 Community Protection Act. See Sheila Donnelly & Roxanne Lieb, Community Notification: A Survey of Law Enforcement 7 (1993). In short, most of the indirect effects of notification are expected and foreseeable.

The second manner in which the majority minimizes the impact of the effects of notification is by separating the analysis into two distinct parts. It first examines the effect of notification on the reputational interests of the offender; then it examines the effect of notification on the increased risk of physical violence. The majority concludes that each of these effects, by itself, does not produce a sting harsh enough to classify notification as punishment. It fails, however, to determine whether these effects, if examined together, are sufficiently harsh. The difference between these two approaches is manifest. Individual effects each might produce only a moderate sting; adding together these little stings might, however, produce a great big sting. In the real world, it is the total sting that the recipient feels. It is not clear why the majority chose not to add these stings together. And, at least from my reading of Artway, there is no justification for choosing not to do so. Rather, I believe that Artway (and Morales ) require an analysis of all the effects of a measure, provided they are not too speculative or attenuated, and here they are not.

Turning from methodology to substance, I first note my agreement with the majority's identification of the effects caused by notification as including isolation, harassment, loss of employment and housing opportunities, damage to property, and physical violence.13

As is clear from the majority's description of the effects of notification, the burden imposed by the collective weight of all of these effects is borne by the offender in all aspects of his life. At worst, the offender is literally cut off from any interaction with the wider community. He is unable to find work or a home, cannot socialize, and is subject to violence or at least the constant threat of violence. At best, he must labor within significant confinements. Although perhaps some people will hire him or rent him a home, his social intercourse with others is all but non-existent. The effects of notification permeate his entire existence. See Doe v. Gregoire, 960 F.Supp. 1478, 1486 (W.D.Wash.1997) ("[H]ere the punitive effects are dominant and inescapable."); Roe v. Office of Adult Probation, 938 F.Supp. 1080, 1092 (D.Conn.1996) ("Notification is an affirmative placement by the State of a form of public stigma on Roe, and this stigma by its very nature pervades into every aspect of an offender's life."). And, although the majority's opinion is eminently fair, I think that it understates the effects of notification provisions. Throughout the nation, there are continual reports of harassment, threats, isolation, and violence. In the margin, I mention some of the most recent occurrences.14

Although the question is very close, I believe that there is a strong argument that the harshness of the effects of notification are closer to imprisonment and revocation of citizenship than to a loss of a profession or of benefits. Like imprisonment and the revocation of citizenship, notification is all-pervasive. In that sense, the offender has almost no refuge from the sometimes severe effects of notification. He may seek to move to another state, but the majority of states has some form of community notification. He could, perhaps, move out of the country to avoid this network of domestic Megan's Laws. At the extreme, then, notification has become, at least for that offender, akin to banishment. See Doe v. Pataki, 940 F.Supp. at 626 ("Notification statutes have resulted in the banishment of sex offenders both literally and psychologically."). This pervasive aspect of notification differentiates it from the loss of employment opportunities and the loss of benefits.15

Perhaps the most difficult question in this context is whether notification is fairly considered punishment when civil commitment--a form of involuntary confinement--is not. In Hendricks the Supreme Court held that a state statute allowing the confinement of convicted sex offenders after the expiration of their prison term did not constitute punishment. Important to the Court was the traditional understanding of civil commitment as non-punitive. But beyond that distinction, I note two respects in which notification under Megan's Law may be considered more harsh than the civil commitment statute at issue in Hendricks.

First, anyone confined under the Kansas statute was afforded some form of treatment if such was possible. See Hendricks, at ----, 117 S.Ct. at 2082-85. No such treatment is available to those subject to notification under Megan's Law, and there is at least some evidence in the record that the isolation engendered by notification may in fact cause some offenders to recidivate. See Prentky Aff. p 4, Appellants' App. at 189; see also Doe v. Pataki, 940 F.Supp. at 628. Thus, the effects of civil confinement might be rehabilitative, while those of notification are exactly the contrary. Second, the Kansas statute required a yearly reevaluation of the confined offender. See Hendricks, at ----, 117 S.Ct. at 2082-84. The registration and notification provisions in Megan's Law are applicable for at least fifteen years. See N.J. Stat. Ann. § 2C:7-2(f) (1995). It is possible, then, that the sting of notification will last far longer than that of civil commitment.

In sum, although I do not rely on my analysis of the effects prong of the Artway test to support my ultimate conclusion, I note that the majority's discussion of effects is seriously flawed in terms of both procedure and substance, casting further doubt upon the judgment and shoring up still further my dissenting posture. The majority improperly and unnecessarily narrows the effects prong of Artway by requiring that a measure deprive an individual of a constitutionally secured fundamental right and by examining the effects in isolated groupings. Finally, its substantive discussion of actual effects is, in important respects, flawed.

The majority's most serious challenge to my position inheres in its argument, citing Hendricks and referring to Ursery, that only the "clearest proof" will negate legislative intent to deem a measure non-punitive. In terms of the Artway test, then, the majority effectively holds that should a measure be considered non-punitive under the test's first (actual purpose) prong, then there is a strong presumption that the measure is non-punitive, and only the clearest proof as to the second (objective purpose) and third (effects) prongs of the test will overcome that presumption. I am unpersuaded. First, the etiology of the "clearest proof" doctrine is such that I doubt that the Supreme Court would apply it in this context with such clear and direct historical antecedents, so plainly punitive in character, to the community notification provisions of Megan's Law. Second, even if the standard were applied here, I believe that the historical context of notification, the design of Megan's Law, and the effects resulting therefrom, provide sufficiently clear proof of objective intent to negate remedial purpose.

The clearest proof standard was first articulated in Flemming v. Nestor, 363 U.S. 603, 80 S.Ct. 1367, 4 L.Ed.2d 1435 (1960). In Flemming, the Supreme Court addressed a contention that the legislative history and design of a statute that allowed the Secretary of Health, Education, and Welfare to terminate Social Security benefits payable to aliens deported due to their political affiliations evidenced a punitive congressional intent that negated a stated remedial intent. The Court stated:

We observe initially that only the clearest proof could suffice to establish the unconstitutionality of a statute on such a ground. Judicial 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. Moreover, the presumption of constitutionality with which this enactment, like any other, comes to us forbids us lightly to choose that reading of the statute's setting which will invalidate it over that which will save it.

Id. at 617, 80 S.Ct. at 1376.

The Court has since employed the clearest proof standard in at least six cases. In Communist Party of the United States v. Subversive Activities Control Board, 367 U.S. 1, 81 S.Ct. 1357, 6 L.Ed.2d 625 (1961), the Court considered whether, despite manifest congressional intent to the contrary, a measure was actually intended to outlaw the Communist Party. The Court stated that only the clearest proof would negate that congressional intent. In United States v. Ward, 448 U.S. 242, 100 S.Ct. 2636, 65 L.Ed.2d 742 (1980), the Court required the clearest proof that, despite the manifest intent to create a civil proceeding, a fine under the Federal Water Pollution Control Act was nevertheless a criminal proceeding. In United States v. One Assortment of 89 Firearms, 465 U.S. 354, 104 S.Ct. 1099, 79 L.Ed.2d 361 (1984), and in Ursery, the Court applied the clearest proof standard to determine whether civil forfeiture statutes were punitive. Examining the Illinois Sexually Dangerous Persons Act, the Court in Allen v. Illinois, 478 U.S. 364, 106 S.Ct. 2988, 92 L.Ed.2d 296 (1986), stated that only the clearest proof would negate the legislative intent that proceedings determining whether an individual should be committed to psychiatric care were civil in nature. Finally, and most recently, in Hendricks, the Court used the clearest proof standard in the context of a challenge to a civil commitment statute.

Although the Supreme Court has repeatedly applied the clearest proof standard in the context of challenges alleging that subjective legislative intent is different from objective legislative intent, I am unwilling to apply the clearest proof standard in this context, at least until the Supreme Court makes it clear that doing so is appropriate. The clearest proof standard creates a nearly irrebuttable presumption that favors subjective legislative intent over objective manifestations of that intent. In an excess of caution, I eschew exploration of the extent to which such a presumption can create incentives for legislatures to obscure their actual intent with subjective intent, rendering it unwise to employ it in certain circumstance. The purpose of the "clearest proof" exercise is to provide a technique to determine legislative intent. This technique is unnecessary here, where, as I have explained, notification measures are so plainly the direct descendants of historical punitive schemes. It seems to me, moreover, that something more than subjective intent alone must be shown to abrogate the historical understanding that notification measures are punitive. In other words, a legislature's simply denying that it is operating outside of a shared cultural tradition does not make it so.

This argument may be illuminated by flipping the coin over, as it were, and looking at the issue by assuming that the clearest proof standard applies in this case. In such event, I believe that such proof exists. At the threshold, I warn against placing too much emphasis on the meaning of "clearest proof." As Flemming and its progeny make patent, the standard is intended as a kind of warning to the federal courts to give legislatures the benefit of the doubt. It is thus consistent with familiar canons of statutory interpretation and constitutional adjudication stating that legislatures are rational bodies that intend to function within their powers to enact lawful measures. In cases in which there is little doubt, however, there is no benefit to give.

Here, there is little doubt. As Part I.C. makes clear, notification measures have historically been considered punitive. As Part I.D. makes clear, the particular design of notification under Megan's Law in no way contradicts this history. And, as Part II makes clear, the effects of notification measures suggest strongly their punitive nature; the majority's efforts to dilute the Artway effects prong, see supra Part II, are unavailing. Taking the foregoing factors together, then, I conclude that sufficient proof of an objective punitive intent motivating the notification provisions of Megan's Law exists to negate the subjective remedial intent.

We should and do endeavor mightily to protect our children from the dangers of the modern world. There is, however, a background risk of violence from which we simply cannot shield them. I believe that the New Jersey legislature desperately wanted to do all that it could to prevent the murder of any child at the hands of a released sex offender. But, if a released sex offender is intent on repeating his offense, there is no reason to believe he will necessarily limit himself to his surrounding community (or, for that matter, limit himself to his state).

Unfortunate though it may be, dangers to our children can come from anywhere. People in the community, especially parents, therefore justifiably warn children more sternly about interacting with strangers, wandering too far from home, staying out past dark, etc. There is no way to determine how many crimes will be prevented by all of the Megan's Laws throughout the country. I suspect, however, that the change in protection secured by notification will be marginal at best. Query whether this marginal change is worth tampering with "an essential thread in the mantle of protection that the law affords the individual citizen." Lynce v. Mathis, --- U.S. ----, ----, 117 S.Ct. 891, 895, 137 L.Ed.2d 63 (1997) (discussing that group of constitutional provisions protecting against the retroactive application of new laws).

It is instructive to note that this issue bears a similarity to the challenge the Supreme Court recently faced in Reno v. ACLU, --- U.S. ----, 117 S.Ct. 2329, 138 L.Ed.2d 874 (1997). There, underlying the Court's decision to strike down key provisions of a statute purporting to rid the Internet of obscenity is the notion that vital constitutional protections must not be swept away in the understandable fervor to protect our children. Basic constitutional rights fundamental to ordered liberty, like the freedom of speech and the right to be free from the retroactive application of the laws, impose on each of us certain burdens. We will remain a free people only so long as we accept those burdens, even in the face of the very safety of our children. Recognizing the rights of released sex offenders, unpalatable though that may be, is one of them.

Although I am outvoted on the double jeopardy/ex post facto issue, I am at least comforted by our holding that the notification machinery, with all of its attendant consequences, will not be triggered without the significant safeguard of requiring the state to establish the case for notification by clear and convincing evidence.