ORDER AND AMENDED OPINION
ORDER
The clerk is directed to hold the mandate pending further order of the court.
The opinion filed September 10, 2009, slip op. 13109, and appearing at 581 F.3d 977 (9th Cir.2009), is hereby amended as follows:
1. Slip op. at 13124, line 21: after and before, insert the following footnote:
Our reasons for distinguishing Doe apply as well to
Russell v. Gregoire, 124 F.3d 1079 (9th Cir.1997) and Hatton v. Bonner; 356 F.3d 955 (9th Cir.2004). Russell and Hatton rejected Ex Post Facto challenges to sex offender registration requirements for those convicted of sex offenses as adults. Neither discussed whether the registration requirements would be punitive if imposed on those adjudicated delinquent in the juvenile justice system, nor did they address the contrasting approaches to privacy/publicity in the juvenile and adult systems. In fact, the Russell court specifically recognized that “[t]he information collected and disseminated by the Washington statute is already fully available to the public.... ” Russell, 124 F.3d at 1094 (rejecting offenders’ contention that the registration and notification requirements violated their right to privacy).
2. Slip op. at 13140-41, replace the text of footnote 16 with the following:
*926United States v. George, 579 F.3d 962 (9th Cir.2009), addressed an ex post facto challenge to SORNA’s criminal provisions from an adult defendant who was convicted of a sex offense prior to SORNA and then convicted under SORNA for failure to register. Smith v. Doe had already established that under SORNA adults may be constitutionally required to register as sex offenders based on pre-SORNA convictions, 538 U.S. 84, 123 S.Ct. 1140, 155 L.Ed.2d 164 (2003), and George did not consider the separate issue we decide here, whether juvenile offenders may be constitutionally required to register based on pre-SORNA adjudications. In any event, George was lawfully required to register as a sex offender as a condition of his preSORNA plea agreement. George argued that his failure to register was a one-time event that took place before SORNA took effect, and therefore his conviction for violating SORNA amounted to an unconstitutional retrospective application of a criminal law. We held otherwise, stating, inter alia, that George, whose initial requirement to register was lawful, violated the law not only when he failed to comply but as long thereafter as he continued to fail to do so. In short, we held that when there is a lawful obligation to register, that obligation is a continuing one. George’s offense of not registering continued from SORNA’s passage on, and SORNA’s imposition of criminal liability for the post-SORNA conduct raised no ex post facto issue. George does not affect our decision here. Because Juvenile Male (S.E.) could not lawfully be required to register on the basis of his pre-SORNA conduct, and that was the only improper sexual conduct with which he was charged, he did not, under our decision here, violate any lawful requirement of SORNA. As there was no obligation on S.E.’s part to register, there was, of course, no continuing obligation to do so.
OPINION
As a society, we generally refuse to punish our nation’s youth as harshly as we do our fellow adults, or to hold them to the same level of culpability as people who are older, wiser, and more mature. The avowed priority of our juvenile justice system (in theory if not always in practice) has, historically, been rehabilitation rather than retribution. Juvenile proceedings by and large take place away from the public eye, and delinquency adjudications do not become part of a young person’s permanent criminal record. Rather, young offenders, except those whose conduct a court deems deserving of treatment as adults, are classified as juvenile delinquents and placed in juvenile detention centers. Historically, an essential aspect of the juvenile justice system has been to maintain the privacy of the young offender and, contrary to our criminal law system, to shield him from the “dissemination of truthful information” and “[transparency” that characterizes the punitive system in which we try adults. Compare 18 U.S.C. § 5038(e) (“[Njeither the name nor picture of any juvenile shall be made public in connection with a juvenile delinquency proceeding.”) with Smith v. Doe, 538 U.S. 84, 99, 123 S.Ct. 1140, 155 L.Ed.2d 164 (2003) (“[0]ur criminal law tradition insists on public indictment, public trial, and public imposition of sentence.”).
In a surge of national concern, however, over the commission of sex offenses, particularly those against children, Congress in 2006 enacted the Sex Offender Registration and Notification Act (“SORNA” or “the Act”) and applied its registration and reporting requirements not only to adults but also to juveniles who commit certain serious sex offenses at the *927age of fourteen years or older. The Attorney General, exercising authority delegated by Congress, determined that SORNA would apply retroactively to all sex offenders convicted of qualifying offenses before its enactment, including juvenile delinquents. 28 C.F.R. § 72.3 (2007).
The retroactive application of SORNA’s juvenile registration provision affects people of all ages — not only juveniles. As we are still close in time to SORNA’s passage, some, like S.E., were adjudicated delinquent relatively recently and are still minors or young adults. The vast majority of persons affected, however, were adjudicated delinquent years or even decades before SORNA’s enactment and quite obviously are no longer juveniles. Indeed, the brunt of SORNA’s retroactive application to juvenile offenders is felt mainly by adults who committed offenses long ago as teenagers — many of whom have built families, homes, and careers notwithstanding their history of juvenile delinquency, which before SORNA’s enactment was not a matter of public record. For these adults, sex offender registration and reporting threatens to disrupt the stability of their lives and to ostracize them from their communities by drawing attention to decades-old sex offenses committed as juveniles that have, until now, remained sealed. Although from this point forward no new individuals will be affected by the retroactivity provision, its effects will be felt by numerous individuals for the rest of their adult lives.1
We must decide as a matter of first impression — in our court and in any other circuit court — whether the retroactive application of SORNA’s provision covering individuals who were adjudicated juvenile delinquents because of the commission of certain sex offenses before SORNA’s passage violates the Ex Post Facto Clause of the United States Constitution. In light of the pervasive and severe new and additional disadvantages that result from the mandatory registration of former juvenile offenders and from the requirement that such former offenders report in person to law enforcement authorities every 90 days for 25 years, and in light of the confidentiality that has historically attached to juvenile proceedings, we conclude that the retroactive application of SORNA’s provisions to former juvenile offenders is punitive and, therefore, unconstitutional.2
I.
At the age of thirteen, defendant-appellant S.E. engaged in non-consensual sexual acts with a ten-year-old child of the same sex. The sexual activity continued until S.E. was fifteen years old and the younger child was twelve. S.E. pled “true” to the commission of acts that, had they been committed by an adult, would constitute aggravated sexual abuse under 18 U.S.C. § 1153 and § 2241(c), because the younger child was, during the period of the charges, under twelve. As a result, S.E. was adjudicated delinquent under 18 U.S.C. § 5031, et seq.3
*928In 2005, a year before SORNA was adopted, the district court sentenced S.E. to two years of. detention at a juvenile facility followed by supervised release until his twenty-first birthday. He was not at this point, of course, ordered to register as a sex offender. S.E. completed his two-year confinement and moved to a prerelease center where, pursuant to the terms of his sentence, he was to reside for six months. When S.E. failed to engage in a required job search, center officials deemed him a program failure and requested his removal. In 2007, a year after the enactment of SORNA, the district court revoked S.E.’s supervised release due to his failure to reside at the center as required by his conditions of supervision, and ordered an additional six months of confinement and continued supervision until S.E.’s twenty-first birthday. The judge also imposed a “special condition” mandating that S.E. register as a sex offender. S.E. objected to the imposition of the registration requirement and timely filed a notice of appeal. The government argues that the special condition is valid because S.E. is required to register by SORNA. S.E. responds that the Ex Post Facto Clause of the United States Constitution bars the retroactive application of the registration provision of SORNA to persons who prior to its passage were designated as juvenile offenders.
Reviewing all questions at issue here de novo, see Beeman v. TDI Managed, Care Services, 449 F.3d 1035, 1038 (9th Cir.2006) (questions of statutory interpretation); Hunter v. Ayers, 336 F.3d 1007, 1011 (9th Cir.2003) (violations of the Ex Post Facto Clause), we hold that SORNA’s juvenile registration provision may not be applied retroactively to individuals adjudicated delinquent under the Federal Juvenile Delinquency Act, and we reverse the directive that S.E. must register under the Act.4
II.
A. Federal Juvenile Delinquency Act (“FJDA”)
The Federal Juvenile Delinquency Act (“FJDA”) sets forth the procedures governing federal juvenile adjudications. 18 U.S.C. § 5031 et seq. “The purpose of the FJDA is to ‘remove juveniles from the ordinary criminal process in order to avoid the stigma of a prior criminal conviction and to encourage treatment and rehabilitation.’ ” United States v. Doe, 94 F.3d 532, 536 (9th Cir.1996) (internal citation omitted); see also In re Sealed Case (Juvenile Transfer), 893 F.2d 363, 367 (D.C.Cir.1990) (noting that the FJDA’s “underlying purpose is to rehabilitate, not to punish, so as ‘to assist youths in becoming productive members of our society’ ”) (quoting S.Rep. No. 1011, 93d Cong., 2d Sess. 22 U.S.Code Cong. & Admin. News 1974 p. 1267 (1974)). The FJDA, accordingly, provides that information about juvenile delinquency pro*929ceedings “shall be safeguarded from disclosure to unauthorized persons.” 18 U.S.C. § 5038(a). This ensures that “a juvenile delinquent for whom there is some hope of rehabilitation [does] not receive the stigma of a criminal record that would attach to him throughout his life.” United States v. Three Juveniles, 61 F.3d 86, 88 (1st Cir.1995) (quoting S.Rep. No.1989, 75th Cong., 3d Sess. 1 (1938)). “The confidentiality provisions of the Act are therefore quite essential to the Act’s statutory scheme and overarching rehabilitative purpose.” Three Juveniles, 61 F.3d at 88. Such provisions include, among other things, the mandate that “neither the name nor picture of any juvenile shall be made public in connection with a juvenile delinquency proceeding.” 18 U.S.C. § 5038. Although the FJDA specifies limited circumstances in which records about juvenile delinquency proceedings may be released to certain officials for law enforcement, judicial, or treatment purposes, it mandates that information from delinquency proceedings “may not be released when the request for information is related to an application for employment ... or any civil right or privilege.” Id.
B. Sex Offender Registration and Notification Act (“SORNA”)
On July 27, 2006, Congress enacted the Adam Walsh Child Protection and Safety Act, 42 U.S.C. § 16901 et seq., which includes the Sex Offender Registration and Notification Act (“SORNA”). SORNA was enacted “[i]n order to protect the public from sex offenders and offenders against children, and in response to the vicious attacks by violent predators” against seventeen named victims of sex crimes. 42 U.S.C. § 16901. SORNA “establishes a comprehensive national system for the registration of [sex] offenders,” id., and requires anyone convicted of specified crimes, including aggravated sexual abuse, to register with the national sex offender registry. 42 U.S.C. § 16911(4)(A)(i). SORNA defines convictions to include juvenile delinquency adjudications of aggravated sexual abuse if the offender is fourteen years of age or older at the time of the offense. 42 U.S.C. § 16911(8).
Congress delegated to the Attorney General the decision whether SORNA should apply retroactively to sex offenders who were convicted before the statute’s effective date. 42 U.S.C. § 16913(d). Congress gave the Attorney General no instruction regarding whether SORNA should apply retroactively or not, and certainly gave no indication to the Attorney General that if applied retroactively to adults, it should be so applied to juveniles as well. Exercising the delegated authority, the Attorney General promulgated a regulation that renders SORNA applicable to “all sex offenders, including sex offenders convicted of the offense for which registration is required prior to the enactment of that Act.” 28 C.F.R. § 72.3 (2007). The regulation went into effect immediately as an interim rule, without providing for a notice and comment period in advance of SORNA’s retroactive application. Office of the Attorney General, Applicability of the Sex Offender Registration and Notification Act, 72 Fed.Reg. 8894-01, 8896-97 (Feb. 28, 2007). The regulation contains no exception for persons adjudged juvenile delinquents, and it does not appear that the Attorney General considered any such exception. Indeed, there is no indication that the Attorney General, in determining the scope of SORNA’s retroactivity, gave any consideration at all to the special circumstances of juveniles who had been adjudicated delinquent under a different— and largely confidential — judicial system, or to the societal costs versus benefits of applying SORNA’s juvenile registration requirement retroactively. See generally 72 Fed.Reg. 8894-01.
*930III.
A statute or regulation that imposes retroactive punishment violates the constitutional prohibition on the passage of ex post facto laws. U.S. Const. Art. I § 9, cl. 3; Doe, 538 U.S. at 92, 123 S.Ct. 1140. The application of SORNA, enacted in 2006, to S.E., who was found delinquent in 2005, is clearly retroactive. See 28 C.F.R. § 72.3 (applying SORNA “to all sex offenders, including sex offenders convicted of the offense for which registration is required prior to the enactment of that Act”).
The question we must answer then is whether the application of SORNA’s juvenile registration provision is punitive.
If the intention of the legislature was to impose punishment, that ends the inquiry. If, however, the intention was to enact a regulatory scheme that is civil and nonpunitive, we must further examine whether the statutory scheme is so punitive either in purpose or effect as to negate [Congress’s] intention to deem it “civil.”
Doe, 538 U.S. at 92, 123 S.Ct. 1140 (internal citations and quotation marks omitted). S.E. has properly not disputed that in enacting SORNA, Congress intended to establish a civil regulatory scheme rather than a criminal one. We must then inquire whether SORNA’s juvenile registration provision is nevertheless punitive because (a) its purpose is to punish or (b) its effect is clearly shown to be punitive. Doe, 538 U.S. at 92-93, 123 S.Ct. 1140. Whether SORNA was passed with a punitive purpose, or whether the Attorney General applied SORNA retroactively in order to punish past conduct, has not been answered in our circuit.5 Because, however, we need not answer that question and because S.E. conceded at oral argument that Congress’s intent was not punitive, despite arguing to the contrary in his briefs, we will assume for the purposes of this case, without deciding the issue, that the answer is no.
Congressional intent, and even the Attorney General’s, notwithstanding, we will find an ex post facto violation if the effect of SORNA’s juvenile registration provision is punitive. See Kansas v. Hendricks, 521 U.S. 346, 361, 117 S.Ct. 2072, 138 L.Ed.2d 501 (1997). The Supreme Court has explained that, “[b]ecause we ordinarily defer to the legislature’s stated intent, only the clearest proof will suffice to override legislative intent and transform what has been denominated a civil remedy into a criminal penalty .... ” Doe, 538 U.S. at 92, 123 S.Ct. 1140 (citation and quotation marks omitted). The requirement of “clearest proof’ is not, however, a requirement that the petitioner present evidence in the record regarding the effects of the statute as applied to him. “Instead, courts must evaluate the question [whether a statute is punitive] by reference to a variety of factors considered in relation to the statute on its face.” Seling v. Young, 531 U.S. 250, 262, 121 S.Ct. 727, 148 L.Ed.2d 734 (2001) (quotation and quotation marks omitted) (emphasis added).
Indeed, when an individual challenges a new law, such as SORNA was at the time this case began, it would appear to be impossible for him to develop a record which contains the “clearest proof’ of the punitive effects that the law will have upon him or indeed upon others. Certainly, we would not require S.E. to suffer and *931then document the ill effects of SORNA’s juvenile registration provision before permitting a challenge to its retroactive application. We interpret the “clearest proof’ requirement in the only way that is sensible: that the terms of the statute, the legal obligations it imposes, the practical and predictable consequences of those obligations, our societal experience in general, and the application of our own reason and logic, establish conclusively that the statute has a punitive effect.
In considering whether the statute has a punitive effect, we refer to the factors first set forth in Kennedy v. Mendoza-Martinez:
[wjhether the sanction involves an affirmative disability or restraint, whether it has historically been regarded as a punishment, whether it comes into play only on a finding of a scienter, whether its operation will promote the traditional aims of punishment — retribution and deterrence, whether the behavior to which it applies is already a crime, whether an alternative purpose to which it may rationally be connected is assignable for it, and whether it appears excessive in relation to the alternative purpose assigned.
Kennedy v. Mendoza-Martinez, 372 U.S. 144, 168-69, 83 S.Ct. 554, 9 L.Ed.2d 644 (1963) (internal citations omitted). These factors, while helpful, are “neither exhaustive nor dispositive, but are useful guideposts.” Doe, 538 U.S. at 97,123 S.Ct. 1140 (internal citations and quotation marks omitted). Here, as in Doe,
[t]he factors most relevant to our analysis are whether, in its necessary operation, the regulatory scheme: has been regarded in our history and traditions as a punishment; imposes an affirmative disability or restraint; promotes the traditional aims of punishment; has a rational connection to a nonpunitive purpose; or is excessive with respect to this purpose.
Id.6
Before applying this legal framework, we consider the extent to which the Supreme Court’s decision in Doe controls the outcome of the present case. The Supreme Court in Doe applied the Mendoza-Martinez factors and concluded that the retroactive application of Alaska’s Sex Offender Registration Act (“the Alaska statute”) to adult sex offenders did not have a punitive effect, and therefore did not violate the Ex Post Facto Clause. It would be tempting to conclude, without looking carefully at the special circumstances of former juvenile offenders, that in light of Doe, sex offender registration by its nature does not constitute punishment. Doe does not, however, mandate that result, and the case before us presents substantially different facts and issues that significantly affect our analysis, and which govern our understanding of how the Constitution must be applied. For both similar and different reasons, Doe is not dispositive of the reporting provisions.
Historically, our country has had two separate systems of justice, one for adults and the other for juveniles. The criminal justice system that applies to adults is fundamentally a public one. We view its public nature as an essential protection for the rights of both the defendant and soci-
*932ety at large. As Doe explains, “[t]ransparency is essential to maintaining public respect for the criminal justice system, ensuring its integrity, and protecting the rights of the accused.” Id. at 99,123 S.Ct. 1140. Our requirement of “public indictment, public trial, and public imposition of sentence,” id. (emphasis added), is central to our vision of a punitive system that is fair and just.
Juvenile adjudications, by contrast, by and large take place outside the public domain. We have historically made the decision to shield juvenile offenders from the public eye — both from the protections that public scrutiny provides against government oppression, and from the burdens that public scrutiny imposes through the stigmatization of those convicted of crimes. As Chief Justice, then Justice, Rehnquist explained, “[i]t is a hallmark of our juvenile justice system in the United States that virtually from its inception at the end of the [nineteenth] century its proceedings have been conducted outside of the public’s full gaze and the youths brought before our juvenile courts have been shielded from publicity.” Smith v. Daily Mail Pub. Co., 443 U.S. 97, 107, 99 S.Ct. 2667, 61 L.Ed.2d 399 (1979) (Rehnquist, J., concurring). Juveniles are denied certain procedural rights afforded to adult criminal defendants, including a public trial by jury,7 but they are, in turn, beneficiaries of an adjudicatory system designed, though not always successfully, to rehabilitate rather than punish — a system ill-suited to public exposure. There are some exceptions to confidentiality in juvenile proceedings, which we will describe further below. However, our juvenile justice system from its origins was established in order to make the child “feel that he is the object of [the state’s] care and solicitude,” and that he would “be treated and rehabilitated” through “clinical” procedures “rather than punitive” ones. Gault, 387 U.S. at 15-16, 87 S.Ct. 1428. The FJDA, which governs juvenile proceedings, is for that reason designed “to ‘remove juveniles from the ordinary criminal process in order to avoid the stigma of a prior criminal conviction and to encourage treatment and rehabilitation.’ ” United States v. Doe, 94 F.3d at 536 (quoting United States v. Brian N., 900 F.2d 218, 220 (10th Cir.1990) (citations omitted)). Our punitive system is public; our rehabilitative system for juveniles, quite deliberately, is not.
In light of these two different systems of justice — one public and punitive, the other largely confidential and rehabilitative — the impact of sex offender registration and reporting upon former juvenile offenders and upon convicted adults differs in ways that we cannot ignore. According Doe its full precedential weight, we are nonetheless compelled to conclude here that the effect of the retroactive application of SORNA’s juvenile registration and reporting requirements is different both in nature and degree than the retroactive application of Alaska’s statute to adult offenders.8 We are also compelled to con-*933elude, for what it’s worth, that it would be a breach of faith to those young persons, some of whom are now elderly, who voluntarily accepted status as a juvenile delinquent believing that their juvenile offense would not later be made known to the world at large.
A. Affirmative disability or restraint
We begin by considering whether the retroactive application of SORNA’s juvenile registration provision “imposes an affirmative disability or restraint.” , Doe, 538 U.S. at 97, 123 S.Ct. 1140. We look to “how the effects of [SORNA’s juvenile registration provision] are felt by those subject to it. If the disability or restraint is minor and indirect, its effects are unlikely to be punitive.” Id. at 99-100, 123 S.Ct. 1140. Because we conclude that the retroactive application of SORNA’s juvenile registration provision imposes a disability that is neither “minor” nor “indirect,” but rather severely damaging to former juvenile offenders’ economic, social, psychological, and physical well-being, this factor strongly supports a determination that the statute’s effect is punitive.9 In fact, given the degree of damage former juvenile offenders may suffer in their adult lives by the retroactive application of the statutory requirement, we conclude that this factor is by far the most compelling in our analysis.
We recognize, of course, that the Supreme Court in Doe concluded that the retroactive application of the Alaska statute did not impose an “affirmative disability or restraint” upon adult sex offenders sufficient to constitute punishment. The Court reasoned that “[t]he Act imposes no physical restraint, and so does not resemble the punishment of imprisonment, which is the paradigmatic affirmative disability or restraint,” Doe, 538 U.S. at 100, 123 S.Ct. 1140; that “[although the public availability of the information may have a lasting and painful impact on the convicted sex offender, these consequences flow not from the Act’s registration and dissemination provisions, but from the fact of conviction, already a matter of public record,” id. at 100-01, 123 S.Ct. 1140; that the statute did not impose an in-person registration requirement, id. at 101, 123 S.Ct. 1140; and that the requirements of registration are less onerous than the conditions of probation and supervised release, id. The Supreme Court’s conclusion must be understood, however, in the context of the public criminal justice system. The burden of sex offender registration upon a former juvenile offender is substantially, and decisively, different.
The key word in our analysis is “impose.” To impose a disability is to place a disability on an individual where none previously existed. In the Alaska case, the statute did not impose disadvantages “that would not have otherwise occurred.” Id. at 100, 123 S.Ct. 1140. There, the stigma and disadvantages derived “not from the Act’s registration and dissemination provi*934sions, but from the fact of conviction, already a matter of public record.” Id. at 101, 123 S.Ct. 1140. Doe emphasized that it would be mere “conjecture” to conclude that the publication of sex offenders’ information on the internet would impose a new barrier to their ability to find employment or housing, as landlords and employers could already conduct background checks and discover adult offenders’ criminal history, which is public information. Id. at 100, 123 S.Ct. 1140. The Court likened the registration and notification provisions of the Alaska statute to “a visit to an official archive of criminal records.” Id. at 99, 123 S.Ct. 1140. The Court did not dispute that “substantial occupational or housing disadvantages” would result from the public’s awareness of a defendant’s status as a sex offender; indeed, it noted that “the public availability of the information may have a lasting and painful impact on the convicted sex offender....” Id. at 101, 123 S.Ct. 1140. It held, however, that there was no proof that these damages “would not have otherwise occurred” as a result of the previous availability of the same information. Id. at 100, 123 S.Ct. 1140.
Here, the precise opposite is true. SORNA’s juvenile registration provision imposes all the conditions that the Supreme Court found the Alaska statute did not impose. None of the consequences that former juvenile offenders suffer as a result of the retroactive application to them of SORNA were imposed on adult offenders by virtue of the Alaska statute. Although the information in the registry at issue in Doe was already public knowledge, information about federal juvenile delinquency adjudications was not. Such information is, in the case of juveniles, ordinarily confidential and may not under most circumstances be disclosed to employers, landlords, or the general public. 18 U.S.C. § 5038 (“Unless otherwise authorized by this section, information about the juvenile record may not be released when the request for information is related to an application for employment ... or any civil right or privilege.”).
Confidentiality in juvenile proceedings is not absolute, but it is generally carefully protected: The norm in juvenile delinquency adjudications is closed proceedings and sealed records.10 Such confidentiality has historically been one of the most significant factors differentiating juvenile adjudications, which are designed to be rehabilitative, from adult criminal proceedings, which are designed to be punitive. District judges do have discretion to open juvenile proceedings and unseal portions of the record of juvenile adjudications under the FJDA, and disclosure to certain authorized persons for certain enumerated purposes is permitted. 18 U.S.C. § 5038(a). However, judges may not expose all juvenile proceedings to public scrutiny as a general practice. They are charged, rather, with “the delicate task of weighing the interests of the juvenile and the public ... in each case.” United States v. A.D., 28 F.3d 1353, 1361 (3rd Cir.1994).11 Moreover, the identity and the image of the juvenile may not be publicly disclosed even in cases in which the proceedings are opened or some of the documents from the case are released: The FJDA mandates that “neither the name nor picture of any juvenile shall be made public in connection with a juvenile *935delinquency proceeding.” 18 U.S.C. § 5038(e). Thus, even in those cases in which the court decides to open the juvenile proceedings to those who wish to attend the trial, the juvenile defendant is not generally exposed to much more public awareness of his identity and criminal conduct than in the ordinary instance when his trial is closed. It is clear that a large-scale release of juvenile records of the magnitude authorized by SORNA, and the ensuing public display of those records on the internet, was prohibited under federal law as it existed prior to the passage of SORNA, and will have a significant and adverse life-long impact upon the individuals affected.
SORNA’s juvenile registration provision, therefore, does not merely provide for further public access to information already available; it makes public information about sex offenders that would otherwise permanently remain confidential and exposes persons who were adjudicated delinquent years before to public humiliation and ignominy for the first time. It also seriously jeopardizes the ability of such individuals to obtain employment, housing, and education.12 The registration and notification system here cannot be compared to a visit to a criminal archive, as such a visit would yield no information about juvenile adjudications. The disadvantages that flow to former juvenile offenders on account of having a public record as sex offenders must be attributed to SORNA alone.
Under SORNA, moreover, individuals who twenty or thirty years ago pled true to acts of juvenile delinquency — and who did so with the expectation that their adjudication would remain confidential — may, decades later, be required to publicly expose that information to friends, family, colleagues, and neighbors. Indeed, most of those affected by the retroactive application of SORNA’s juvenile registration provision are not juveniles but adults. Many of these individuals have for many years led entirely law-abiding and productive lives that may be dramatically disrupted by the registration requirements. Some, had they known that they would years later be subject to registration requirements, might not have pled true to the charges at all.
Beyond these societal consequences of public registration as a sex offender, SORNA imposes additional administrative burdens in the form of “periodic in person verification.” 42 U.S.C. § 16916 (“[A] sex offender shall appear in person, allow the jurisdiction to take a current photograph, and verify the information in each registry *936in which that offender is required to be registered.”). In upholding the retroactive application of Alaska’s statute, the Court in Doe explicitly noted that the statute did not mandate in-person registration. Id. at 101, 123 S.Ct. 1140. Every former juvenile offender subject to SORNA, by contrast, must register in person four times a year for at least 25 years.13 This requirement for appearances every three months before law enforcement officials is neither “minor” nor “indirect.” Doe, 538 U.S. at 99-100, 123 S.Ct. 1140. Every three months, the former juvenile offenders will be required to be absent from work, appear before public officials, and publicly reaffirm that they are guilty of misdeeds that were previously protected from disclosure.14
Because SORNA’s juvenile registration provision, retroactively applied to former juvenile offenders, imposes a serious disability by making public otherwise confidential delinquency records relating to sexual offenses, and because the in-person registration requirement is substantially burdensome, SORNA’s juvenile registration provision imposes an onerous “affirmative disability or restraint” on former juvenile offenders. Mendoza-Marbinez, 372 U.S. at 168, 83 S.Ct. 554. As we have already stated, this factor weighs heavily in support of a finding that SORNA’s juvenile registration requirement has a punitive effect. Given the severity of its burdens, it would be difficult to reach any other conclusion.
B. Historical treatment
We next consider whether requiring former juvenile sex offenders to register and report to law enforcement regularly is an historical means of punishment. The fact that sex offender registration and notification statutes “are of fairly recent origin,” Doe, 538 U.S. at 97, 123 S.Ct. 1140, suggests, initially, that it is not. Doe held that the apparent similarity between sex offender registration and early forms of shaming punishments is “misleading,” Id. at 98, 123 S.Ct. 1140, and explained that *937Id. at 98-99, 123 S.Ct. 1140 (emphasis added). We are struck, once again, by the vastly different situation of adult criminal defendants from that of juvenile offenders, and of the corresponding difference in the effect of registration upon the two groups. As Doe recognizes, adult criminal proceedings have long been a matter of public record, and indeed the right to a public trial in all criminal prosecutions is fundamental under our Constitution. U.S. Const. Amend. VI. Full disclosure of the offense and the offender is an integral part of our punitive system. The public availability of information is not, however, a traditional part of the rehabilitative juvenile justice system. In fact, quite the opposite is true. A core distinguishing feature of the juvenile justice system has historically been that juveniles are, with certain exceptions, permanently shielded from the public eye. The federal juvenile justice system was designed precisely in order to “remove juveniles from the ordinary criminal process in order to avoid the stigma of a prior criminal conviction and to encourage treatment and rehabilitation.” Doe, 94 F.3d at 536 (quotation marks omitted).
*936the stigma of Alaska’s Megan’s Law results not from public display for ridicule and shaming but from the dissemination of accurate information about a criminal record, most of which is already public. Our system does not treat dissemination of truthful information in furtherance of a legitimate governmental objective as punishment. On the contrary, our criminal law tradition insists on public indictment, public trial, and public imposition of sentence. Transparency is essential to maintaining public respect for the criminal justice system, ensuring its integrity, and protecting the rights of the accused. The publicity may cause adverse consequences for the convicted defendant, running from mild personal embarrassment to social ostracism. In contrast to the colonial shaming punishments, however, the State does not make the publicity and the resulting stigma an integral part of the objective of the regulatory scheme.
*937Historically, information from juvenile adjudications has been made public only when a juvenile’s case is transferred to adult criminal court for punitive purposes. Beyond the exceptions to confidentiality in juvenile proceedings detailed supra in Part III.A, there are a select number of juveniles who are prosecuted publicly. In certain circumstances, courts transfer such cases to adult court, shifting the juvenile out of a rehabilitative system entirely and into a punitive one. Under federal law, courts evaluate several factors when determining whether a juvenile will be treated as an adult, including the prior record of the offender, his response to past treatment, and the nature of the offense. 18 U.S.C. § 5032. Courts may give special weight to “the heinous nature of the crime,” United States v. Doe, 94 F.3d at 536-37, and must strike a balance “between providing a rehabilitative environment for young offenders as well as protecting society from violent and dangerous individuals and providing sanctions for anti-social acts.” Id. (quoting United States v. E.K., 471 F.Supp. 924, 932 (D.Or.1979)). A court’s decision to send a juvenile to adult court is thus based in part on a prediction that rehabilitation is improbable. See United States v. Alexander, 695 F.2d 398, 401 (9th Cir.1982). A decision that a juvenile is beyond rehabilitation is a decision to expose him to the punitive elements of adult court, including the publication of his criminal record.
Although SORNA does not transfer a juvenile to adult court, it does make public the record of an otherwise confidential juvenile adjudication. Creating a public record of a federal juvenile offense is something that, historically, has been done only after the court’s determination that the juvenile’s case merits punishment, rather than rehabilitation. In short, the public disclosure mandated by SORNA’s juvenile registration provision is historically a central feature of a punitive rather than a rehabilitative system of justice. Still, in the end, we cannot say that this factor weighs in favor of holding the juvenile registration and notification provisions to be punitive in nature.
C. Traditional aims of punishment
We turn next to whether SORNA promotes the traditional aims of punishment — in particular, the aim of retribution. See Mendoza-Martinez, 372 U.S. at 168-69, 83 S.Ct. 554. As stated previously, we decline to rest our holding in this case on whether SORNA was enacted with a punitive intent. Nevertheless, whether SORNA “will promote ... retribution,” id., is tied to the question whether Congress en*938acted SORNA with the goal of retribution in mind. In this light, we will consider whether SORNA’s text and history suggest that the disadvantages imposed are purely regulatory, or were designed, at least in part, in order to promote the traditional aims of punishment, and thus whether SORNA serves that purpose.
As we do so, we are aware both of the Supreme Court’s decision in Doe, and of the inflamed public sentiment against sex offenders that served as the historical backdrop for SORNA’s passage. Justice Souter, concurring in Doe, saw the question of Alaska’s legislative intent as a close one. He explained:
It would be naive to look no further [than to the regulatory goal of public safety], given pervasive attitudes toward sex offenders. The fact that the Act uses past crime as the touchstone, probably sweeping in a significant number of people who pose no real threat to the community, serves to feed suspicion that something more than regulation of safety is going on; when a legislature uses prior convictions to impose burdens that outpace the law’s stated civil aims, there is room for serious argument that the ulterior purpose is to revisit past crimes, not prevent future ones.
Doe, 538 U.S. at 109, 123 S.Ct. 1140 (Souter, J., concurring in the judgment) (internal citations omitted). SORNA’s legislative history suggests that precisely such a retributive aim contributed to its passage — and more overtly than in the “close case” of Doe. See id. at 107, 123 S.Ct. 1140. Unlike Alaska’s statute, which contained no legislative purpose statement and was passed pursuant to legislative findings that focused solely on public safety, SORNA’s legislative purpose statement reveals an additional goal: to respond to the heinous crimes committed by sex offenders. SORNA was enacted “[i]n order to protect the public from sex offenders and offenders against children, and in response to the vicious attacks by violent predators against the victims listed below .... ” 42 U.S.C. § 16901 (emphasis added). The statute subsequently lists seventeen individual victims and details the crimes that were committed against them, strongly suggesting that the motivation behind SORNA’s passage was not only to protect public safety in the future but also to “revisit past crimes.” Doe, 538 U.S. at 109, 123 S.Ct. 1140 (Souter, J., concurring in the judgment). Senator Grassley’s floor statement similarly reflects a retributive sentiment that colored the legislative proceedings: “Child sex offenders are the most heinous of all criminals. I can honestly tell you that I would just as soon lock up all the child molesters and child pornography makers and murderers in this country and throw away the key.” 152 Cong. Rec. S8012, S8021 (daily ed. July 20, 2006) (statement of Sen. Grassley).
The purpose of the Ex Post Facto Clause is to prevent the passage of “potentially vindictive legislation.” Doe, 538 U.S. at 109, 123 S.Ct. 1140 (Souter, J., concurring in the judgment) (quoting Weaver v. Graham, 450 U.S. 24, 29, 101 S.Ct. 960, 67 L.Ed.2d 17 (1981)) (internal quotation marks omitted). SORNA’s legislative text and history contain substantial warning signs that its aim, while principally regulatory, to be sure, is also in some measure punitive.
D. Non-punitive purpose and excessiveness
We next consider whether SORNA’s juvenile registration provision has a non-punitive purpose and, if it does, whether the requirement is excessive in relation to that goal. We must determine “whether the disability is imposed for the purpose of punishment or whether it is but an incident of some other legitimate governmental purpose.” Bell v. Wolfish, 441 *939U.S. 520, 538, 99 S.Ct. 1861, 60 L.Ed.2d 447 (1979). If the statute is reasonably-related to a non-punitive purpose then the statute is not usually considered punitive. Id. at 539, 99 S.Ct. 1861. However, it is more likely to be punitive if it “appears excessive in relation to the alternative purpose assigned.” Mendoza-Martinez, 372 U.S. at 169, 83 S.Ct. 554.
In Doe, the Court held that the Alaska statute had a nonpunitive purpose: improving public safety. 538 U.S. at 102,123 S.Ct. 1140. Undeniably, SORNA, too, was enacted in order to achieve that regulatory aim. Whether the means that it employs in order to achieve that purpose are excessive, however, is a distinct — and a close-question. Although Doe sets a stringent standard for excessiveness, there are valid and serious concerns both with the utility and with the extreme consequences of SORNA’s juvenile offender registration requirement that warrant review.
On the one hand, Congress has extended SORNA’s juvenile registration requirement to only a portion of those who were adjudicated delinquents: those who were “14 years of age or older at the time of the offense and the offense adjudicated was comparable to or more severe than aggravated sexual abuse.... ” 42 U.S.C. § 16911(8). There was a debate in the Senate regarding whether SORNA should apply to juveniles at all, and if so to what extent. The result appears to have been a compromise.15
On the other hand, even compromises may be excessively harsh. Given the low risk that former juvenile sex offenders pose to public safety and the lifetime confidentiality that most former juveniles would otherwise enjoy, retroactively applying SORNA’s juvenile registration provision is an exceptionally severe means of achieving the statute’s nonpunitive goal.
In Doe, the Supreme Court held that the sex offender registration requirement was not excessive in light of its regulatory purpose, in part because sex offenders have a “high rate of recidivism,” are dangerous “as a class,” pose a danger to the public that is “frightening and high,” and “are much more likely than any other type of offender to be rearrested for a new rape or sexual assault.” 538 U.S. at 103, 123 S.Ct. 1140 (citations omitted).16
*940There is no evidence, however, that the “high rate of recidivism” at issue in Doe is shared by juvenile offenders. Studies cited in the legislative history of this bill indicate that the recidivism rates for juvenile offenders are significantly lower than for adult offenders. See 152 Cong. Rec. S8012-02, S8023 (daily ed. July 20, 2006) (statement of Sen. Kennedy) (“For juveniles, the public notification provision in this bill is harsh given their low rate of recidivism, which is less than 8 percent according to the most recent studies.”); Coalition for Juvenile Justice, Comments in Opposition to Interim Rule RIN 1.105— AB22, 3 (2007) (citing study showing 5 to 14% recidivism rate for juveniles, as compared to 40% rate of recidivism for adults); Human Rights Watch, supra, at 69-70 (listing studies finding low recidivism rates among juvenile sex offenders). Research suggests, moreover, that only a small portion of adult sex offenders previously committed sex offenses as juveniles. See Human Rights Watch at 70.
These statistics are not surprising. Juveniles are as a general matter less mature, more impulsive, and more confused about sexually appropriate behavior than adults. They do not understand their sexual drives as well or know how to deal with them. We do not, of course, excuse such conduct as mere juvenile exuberance. We simply recognize that the predictive value of an individual’s conduct, especially sexual conduct, at the age of fourteen or fifteen is, under most circumstances, limited. For that reason, requiring former juvenile sex offenders to register as such many decades thereafter will often not only be unnecessary to secure the safety of the community but may even be counterproductive.
It is worth noting that, in practice, those who are primarily affected by SORNA’s retroactive application to former juvenile offenders are not the most likely to recidivate — they are, rather, those adults who are forced to register solely because they committed an offense as a juvenile, but who have lived the rest of their adult lives without committing another such crime. Adults who re-offended in the past after their initial juvenile offense are required to register in any event by SORNA’s retroactive application to adult sex offenders, and adults who re-offend in the future are required to register by SORNA’s basic provisions.
In addition to the personal toll on those who are labeled as sex offenders, registration of former juvenile offenders undermines the rehabilitative goals of our juvenile justice system as a whole, and derails the historical effort to avoid permanently or publicly stigmatizing juveniles as criminals. It may also seriously affect the lives of the spouses and children of these former juvenile sexual offenders. Sacrificing confidentiality and lessening the chance of rehabilitation for former juvenile sex offenders who have not re-offended are severe measures to aid in achieving public safety, in light of the importance of the contravening interests and the relatively low risk that such offenders pose to the community. Indeed, the severity of the measures may well increase the risk of recidivism within a population that other*941wise has the greatest potential for rehabilitation.
Notwithstanding all this, we recognize that “[t]he excessiveness inquiry of our ex post facto jurisprudence is not an exercise in determining whether the legislature has made the best choice possible to address the problem it seeks to remedy. The question is whether the regulatory means chosen are reasonable in light of the non-punitive objective.” Doe, 538 U.S. at 105, 123 S.Ct. 1140. Whether that test has been met here is a close and difficult question. To us, SORNA’s effect on former juvenile offenders does “appear[] excessive in relation to the [non-punitive] purpose assigned.” Mendoza-Martinez, 372 U.S. at 169, 83 S.Ct. 554. Recognizing the limited nature of our inquiry, however, as well as the Supreme Court’s decision in Doe, we will not give much weight either way to this factor in making our ultimate determination.
IV.
The retroactive application of SORNA’s provision requiring registration and reporting by former juvenile offenders imposes immense burdens, not only through onerous in-person registration and reporting requirements, but, more important, through the publication and dissemination of highly prejudicial juvenile adjudication records of individuals who have committed no offenses since their adolescence — records that would otherwise remain sealed. The juvenile registration requirement, for the first time under federal law, exposes thousands of former juvenile offenders to public notoriety and subjects them to lifetime condemnation and ostracism by their community. The effects of this exposure are wide-ranging, and likely include serious housing, employment, and educational disadvantages. Unlike in Doe, for former juvenile offenders generally these effects are solely attributable to SORNA. The publicity that once juvenile offenders, now law abiding adults, face is, moreover, something that has traditionally attached to juvenile offenders only if they are transferred to an adult — and punitive — system. Historically, public exposure constitutes an integral part of a punitive and not a juvenile or rehabilitative regime. Additionally, while it is indisputable that SORNA was enacted as a regulatory measure in order to promote public safety, there is evidence that this non-punitive aim was mixed to some degree with a less evident desire for retribution as well. Finally, imposing the burdens of registration upon former juvenile offenders is a harsh measure in view of the low rate of recidivism for juvenile offenders and the importance of the countervailing goal of rehabilitation; it is a close question, however, whether it is excessive in light of Congress’s non-punitive objectives. All this, of course, is in addition to the onerous requirement that these former juvenile offenders report in person every three months to law enforcement authorities for a large portion of their adult lives.
Although we are not bound by the Mendoza-Martinez factors, they prove useful to our analysis in this case. Taken together, they provide “the clearest proof,” Kansas v. Hendricks, 521 U.S. 346, 361, 117 S.Ct. 2072, 138 L.Ed.2d 501 (1997), that the effect of the Attorney General’s regulation that retroactively imposes SORNA’s juvenile registration and reporting requirement upon former juvenile offenders who were found delinquent prior to the passage of the statute, is punitive. Of this, we are fully persuaded. The requirement serves to convert a rehabilitative judicial proceeding, sheltered from the public eye, into a punitive one, exposed for all to see, and with long-lasting substantially adverse and harsh effects. In some instances, the retroactive implementation of SORNA’s provisions will most certainly wreak havoc upon the lives of those whose conduct as *942juveniles offended the fundamental values of our society but who, we hope, have been rehabilitated. For these reasons, we conclude that the retroactive application of SORNA’s juvenile registration and reporting requirement violates the Ex Post Facto Clause of the United States Constitution.17 We therefore VACATE the part of the judgment order that pertains to registration and reporting as a sex offender, and hold that S.E. may not constitutionally be obligated to register as a sex offender under SORNA.
VACATED in part and REMANDED in part.