John Doe v. Gary Settle

PUBLISHED

UNITED STATES COURT OF APPEALS
FOR THE FOURTH CIRCUIT

No. 20-1951

JOHN DOE,

Plaintiff - Appellant,

v.

COLONEL GARY T. SETTLE, in his official capacity as Superintendent of the Virginia
Department of State Police,

Defendant - Appellee.

Appeal from the United States District Court for the Eastern District of Virginia, at
Norfolk. Raymond A. Jackson, District Judge. (2:20-cv-00190-RAJ-LRL)

Argued: September 23, 2021 Decided: January 28, 2022

Before RICHARDSON and QUATTLEBAUM, Circuit Judges, and KEENAN, Senior
Circuit Judge.

Affirmed by published opinion. Judge Richardson wrote the opinion, in which Judge
Quattlebaum and Senior Judge Keenan joined.

ARGUED: Kenton Craig Welkener, Jr., BOSSON LEGAL GROUP PC, Fairfax,
Virginia, for Appellant. Michelle Shane Kallen, OFFICE OF THE ATTORNEY
GENERAL OF VIRGINIA, Richmond, Virginia, for Appellee. ON BRIEF: Timothy P.
Bosson, BOSSON LEGAL GROUP PC, Fairfax, Virginia, for Appellant. Mark R.
Herring, Attorney General, Michael A. Jagels, Acting Deputy Attorney General, Holli
Reeves Wood, Assistant Attorney General, Toby J. Heytens, Solicitor General, Martine E.
Cicconi, Deputy Solicitors General, Jessica Merry Samuels, Assistant Solicitor General,
Kendall T. Burchard, John Marshall Fellow, OFFICE OF ATTORNEY GENERAL OF
VIRGINIA, Richmond, Virginia, for Appellee.

2
RICHARDSON, Circuit Judge:

Two months after he turned 18, John Doe was caught having sex with his 14-year-

old girlfriend. Given the facts of his arrest, Doe may well have been charged with “carnal

knowledge of a child,” a Class 4 felony that prohibits sex with 13- and 14-year-old children.

But instead he was charged with and pleaded to a lower-class felony, “taking indecent

liberties with children,” which only prohibits behavior like propositioning a child for sex.

Doe’s plea may have gotten him a shorter prison sentence, but due to a quirk in Virginia

law, it also led to worse treatment by Virginia’s sex-offender registry. Both crimes

generally put an offender on the highest tier of the registry for life, but there is a narrow

exception to that rule. When an offender is less than 5 years older than his victim, he may

be removed from the registry in time. But that mitigating exception only applies to carnal

knowledge, the crime with the higher sentencing range, and not to indecent liberties. So

while Doe may have felt lucky to only be charged with indecent liberties, given the

potential for a lower prison sentence, that plea ended up condemning him to worse

treatment on the registry. Because of that oddity, Doe will spend the rest of his life on

Virginia’s sex-offender registry with no hope for relief.

Doe—now in his 30s—sued Colonel Gary T. Settle, Superintendent of the Virginia

Department of State Police, hoping to persuade a court to remove him from that registry

and its burdens. Doe argues that the registry and the 5-year-gap provision violate multiple

constitutional principles. In his Fourteenth Amendment equal protection claim, Doe asks

us to consider why an offender convicted of having sex with a child, as Doe might have

been, should be treated better than an offender convicted only of propositioning a child for

3
sex, Doe’s actual charge. In his Eighth Amendment claim, Doe asks us whether a lifelong

registration requirement is an appropriate sanction for a single nonviolent crime committed

by a high-school student.

Both appeals present significant issues of fairness, but at bottom, they ask us to

question the wisdom of the Virginia legislature and its sex-offender registry. That is not

our place. When the Constitution is invoked, our place is to determine whether state laws

comply with the specific dictates of that document. And Virginia’s sex-offender registry

complies with the Eighth and Fourteenth Amendments. So we affirm the district court’s

dismissal.

I. Background

A. Facts

When John Doe was 17 years old, he began dating a girl at his high school. She

was 14. Months later, the two were caught having sex in a parked car behind the local

middle school. But by then, Doe was 18 years old—18 years and 2 months, to be exact—

and the girl was still 14—98 days from her 15th birthday. That is criminal under Virginia

law, so Doe was arrested. Doe could have been prosecuted for a violation of “carnal

knowledge of a child between thirteen and fifteen,” a Class 4 felony under Va. Code § 18.2-

63 with a minimum sentence of 2 years. But he was allowed to plead to “[t]aking indecent

liberties with children,” a lesser Class 5 felony under Va. Code § 18.2-370(A) with a

minimum sentence of only 1 year.

According to Doe, his attorney advised him to plead guilty to the charge, and Doe

did so in 2008. He was sentenced to 3 years in prison but only served 4 months; the rest

4
of the sentence was suspended. Doe alleges that no one mentioned the sex-offender

registry to him before his plea. So only later did Doe realize he would have to register as

a sex offender for the rest of his life.

B. Virginia’s Sex-Offender Registry

To assist law enforcement and to help communities protect themselves from repeat

sex offenders, Virginia created a sex-offender registry. See Va. Code § 9.1-900. The Sex

Offender and Crimes Against Minors Registry separates offenders into three tiers based on

the seriousness of their offenses, with Tier III status being reserved for the worst crimes,

including rape and murder. § 9.1-902. 1

The registry requires extensive information from all offenders: photographs,

fingerprints, DNA samples, home address, employer information, vehicle information, and

internet usage information like email addresses and other online identities. § 9.1-903.

Virginia State Police are then charged with publishing much of that information on the

internet: name and address, employment, a photograph, and “such other information as the

State Police may from time to time determine is necessary to preserve public safety.” § 9.1-

913. And this is not a one-time deal; offenders must continually verify and reverify that

information. Tier III offenders like Doe must verify their information every three months

to start, with the chance of less frequent reporting over time. § 9.1-904. If an offender

1
Tier III offenses used to be categorized as “sexually violent offenses.” See Va.
Code §§ 9.1-902, 9.1-908 (2019). The current three-tier system was introduced in 2020,
but the substance of the registry is largely unchanged. See generally 2020 Va. Acts ch.
829.
5
fails to verify his information on time, he can be charged with a felony and then required

to verify his information even more often. §§ 9.1-904, 18.2-472.1. 2

And those are only the periodic requirements: Certain other changes to a registered

sex-offender’s personal information demand almost immediate notification to the

authorities. See §§ 9.1-903(D)–(F) (requiring notification 3 days after a changed name,

residence, employment, or vehicle registration), 9.1-903(G) (requiring notification 30

minutes after a change in email or other internet identification), 9.1-903(D) (requiring

notification 10 days after a move to another state). Virginia State Police must physically

verify an offender’s registration information twice a year and can get a warrant for further

investigation when they have probable cause to believe some registration violation has

occurred. § 9.1-907(C). In Doe’s case, he reports to a sex-offender investigative officer

who has been permanently assigned to him and who performs what Doe describes as

“random home checks” every 6 months.

Beyond simply providing information, other consequences flow from an offender’s

status on the registry. Tier III offenders cannot enter a school during school hours without

court-ordered permission. § 18.2-370.5. And offenders on the registry are not eligible for

2
In 2017, Doe failed to verify his registration information on time. This led to
another arrest and another guilty plea. Doe was first charged with § 18.2-472.1(B), which
is the charge for Tier III sex offenders who fail to verify their information and which is a
Class 6 felony with a statutory range of 1 to 6 years in prison. Va. Code § 18.2-10. But
Doe pleaded to a lesser offense under § 18.2-472.1(A). For that failure, Doe had to wear
an ankle monitor for 6 months and his registration duties were increased, from once a
quarter to once a month.

6
certain commercial driver’s licenses, cannot drive a tow-truck, and cannot work as

rideshare drivers for companies like Uber or Lyft. §§ 46.2-116, 46.2-341.9, 46.2-2099.49.

Not every negative consequence of having committed a sex crime is a part of the

registry though. Take § 18.2-370.2, which prohibits a sex offender from loitering within

100 feet of any place that they know or have reason to know is a school, daycare,

playground, athletic facility, or gym. That criminal statute does not refer to the registry or

its tiers. Rather, it restricts such loitering “as part of [an offender’s] sentence.” Id. And

because this restriction is imposed as a part of the sentence and not as a part of the registry,

it presumably still applies to sex offenders who eventually get off the registry. For

example, kidnapping under § 18.2-47(A) is a Tier I offense that would allow a perpetrator

to get off the registry in 15 years, but which also triggers the lifelong no-loitering

restriction. §§ 18.2-370.2, 9.1-902. Other restrictions on sex offenders are similarly tied

to the conviction and not the registry. See § 63.2-1205.1 (forbidding those who have “been

convicted of an offense requiring registration” under Chapter 9 from adopting); § 18.2-

370.4 (prohibiting certain people from working or volunteering at schools based on their

sex offenses, without reference to the registry); § 22.1-296.1 (requiring applicants for

public teaching positions to certify that they have not committed certain sexual offenses

and punishing false statements). Even with these clarifications, the registry is a serious

imposition.

For some, there is a chance to get off the registry in time. After 15 years, most Tier

I offenders can petition a court to be removed from the registry. § 9.1-910(A). The

offender must have completed all court-ordered counseling and treatment and paid

7
restitution to his victims. § 9.1-910(B). “If, after such hearing, the court is satisfied that

such person no longer poses a risk to public safety, the court shall grant the petition.” Id.

Tier II offenders must wait 25 years to petition for removal. § 9.1-910(A). But for

everyone else—Tier III offenders like Doe, repeat offenders, and murderers—the registry

is for life with no exception. See §§ 9.1-908, 9.1-910. The best Doe can hope for is to

have his reverification duty limited to once a year. § 9.1-909.

As we have said, the offense determines an offender’s tier on the registry, and two

offenses are relevant here: “[c]arnal knowledge of a child between the ages of thirteen and

fifteen years of age” under § 18.2-63(A) and “[t]aking indecent liberties with children”

under § 18.2-370(A).

Carnal knowledge prohibits sex with 13- and 14-year-old children. 3 Indecent

liberties prohibits a range of behavior with children under 15 years old that leads up to but

does not include sex acts, things like exposing your genitals to a child, proposing sex to a

child, or enticing a child to enter a house for sex. 4 Even though there is considerable

3
Section 18.2-63(A) reads in part: “If any person carnally knows, without the use
of force, a child thirteen years of age or older but under fifteen years of age, such person
shall be guilty of a Class 4 felony.” To “carnally know” means “any sexual bodily
connection, not simply sexual intercourse.” Singson v. Commonwealth, 621 S.E.2d 682,
686 (Va. Ct. App. 2005) (quoting Santillo v. Commonwealth, 517 S.E.2d 733, 740 (Va. Ct.
App. 1999)).
4
Section 18.2-370 reads in part:
A. Any person 18 years of age or over, who, with lascivious intent, knowingly
and intentionally commits any of the following acts with any child under the
age of 15 years is guilty of a Class 5 felony:
(Continued)
8
overlap between these offenses, there are significant differences. For instance, carnal

knowledge can be committed by adults or minors while indecent liberties can only be

committed by adults. And carnal knowledge only includes victims of 13 and 14 years old

while indecent liberties includes any child under 15.

The two crimes are also treated differently by the registry. Indecent liberties is an

out-and-out Tier III offense. Carnal knowledge is Tier III but only “where the perpetrator

is more than five years older than the victim.” § 9.1-902. Otherwise, it is Tier I. This so-

called Romeo-and-Juliet provision is meant to “ameliorate the sex offender registry

requirement for teenagers convicted of consensual sex crimes.” See Virginia Crime

Commission, “Romeo and Juliet” Laws (2007). So most adults who are convicted of either

crime end up on Tier III, but for some young-adult offenders, the 5-year-gap provision

makes a difference. Carnal-knowledge offenders who are less than 5 years older than their

victims—which will include some 18- and 19-year-olds—are Tier I and can petition to be

removed from the registry after 15 years. But indecent-liberties offenders—even 18- and

19-year-olds within the 5-year Romeo-and-Juliet window—will always be Tier III.

(1) Expose his or her sexual or genital parts to any child . . . or propose
that any such child expose his or her sexual or genital parts to such
person; or . . .
(3) Propose that any such child feel or fondle his own sexual or genital
parts or the sexual or genital parts of such person or propose that such
person feel or fondle the sexual or genital parts of any such child; or
(4) Propose to such child the performance of an act of sexual
intercourse . . . ; or
(5) Entice, allure, persuade, or invite any such child to enter [a place] for
any of the purposes set forth in the preceding subdivisions of this
subsection.
9
II. Discussion

After more than a decade on the sex-offender registry, Doe argues that his Tier III

classification violates the federal and Virginia constitutions. The district court dismissed

Doe’s federal claims for failure to state a claim, and once those federal claims were gone,

the district court declined to exercise supplemental jurisdiction over the state claims. 5 Doe

appealed and we have jurisdiction. 28 U.S.C. § 1291.

A. Equal Protection Claim

“No State shall . . . deny to any person within its jurisdiction the equal protection of

the laws.” U.S. Const. amend. XIV. The Equal Protection Clause embodies a simple

principle of government: “[A]ll persons similarly situated should be treated alike.” City

of Cleburne v. Cleburne Living Ctr., 473 U.S. 432, 439 (1985). Courts have distilled that

aspirational standard into a two-part test. First, a plaintiff must prove that he has been

treated differently from others who are similarly situated to him. Second, if he makes out

that initial showing, the court must consider whether the classification can be justified

5
The district court granted Doe leave to pursue this lawsuit under a pseudonym, but
it is not clear why. A litigation pseudonym is a “rare dispensation.” Doe v. Public Citizen,
749 F.3d 246, 274 (4th Cir. 2014) (citing James v. Jacobson, 6 F.3d 233, 238 (4th Cir.
1993)); cf. Eugene Volokh, The Law of Pseudonymous Litigation (Draft Jan. 13,
2021). The public has a strong interest in knowing the names of parties, and it is the district
court’s job to ensure that a party’s individual interest in anonymity truly outweighs the
public interest in open-air litigation. While a litigant’s identity may not be as important in
purely legal or facial challenges, see Doe v. Megless, 654 F.3d 404, 409 (3d Cir. 2011),
most of Doe’s arguments in favor of using a pseudonym were based on the theory that it
would be a burden to reveal his sex-offender status to the public. But Doe has already been
publicly revealed as a sex offender. That is what the sex-offender registry does. So
perhaps the district court incorrectly balanced the interests here. But since this issue has
gone unchallenged and unbriefed, we leave any reconsideration to the district court.
10
under the appropriate level of constitutional scrutiny. Morrison v. Garraghty, 239 F.3d

648, 654 (4th Cir. 2001). Both parties here agree that there is no fundamental right or

suspect classification at issue to trigger heightened scrutiny, so if we get to the second step,

we apply rational-basis review. See Nordlinger v. Hahn, 505 U.S. 1, 10 (1992).

Doe’s equal protection claim focuses on the differential treatment created by the

beneficial application of the Romeo-and-Juliet provision to offenders convicted of carnal

knowledge while no Romeo-and-Juliet provision applies for those convicted of indecent

liberties. So to make out an equal protection claim, Doe—having been convicted of

indecent liberties—must prove that he is similarly situated to someone convicted of carnal

knowledge. Then he must show that the distinction in the sex-offender registry between

that crime and indecent liberties fails rational-basis review. While we assume without

deciding that Doe may be similarly situated to offenders convicted of carnal knowledge,

he fails to make the exceptional showing required to defeat a state law under rational-basis

review.

1. Similarly Situated

Doe argues that he is similarly situated to a hypothetical person convicted of carnal

knowledge whose victim was also within the Romeo-and-Juliet provision’s 5-year

window. To make out this first part of an equal protection claim, Doe must show that these

two groups—those convicted of indecent liberties and those convicted of carnal

knowledge—are “in all relevant respects alike.” Nordlinger, 505 U.S. at 10.

11
 Doe’s best argument to make this showing relies on Skinner v. Oklahoma ex rel.

Williamson, 316 U.S. 535, 538–39 (1942). 6 In Skinner, the Supreme Court held that

Oklahoma’s Habitual Criminal Sterilization Act violated the Equal Protection Clause. Id.

at 536; Oklahoma Habitual Criminal Sterilization Act, 1935 Okla. Sess. Laws 94–99. That

law authorized the sterilization of criminals who had been convicted of three separate

felonies “involving moral turpitude.” Oklahoma Habitual Criminal Sterilization Act, § 3.

Felonies “involving moral turpitude” included larceny but expressly excluded certain

offenses, including embezzlement. § 24A. Skinner had been convicted of three larcenies:

once for stealing chickens and twice for armed robbery. So a judgment was entered

requiring sterilization. Skinner, 316 U.S. at 537. Skinner argued that there was no

legitimate reason to sterilize him as a three-time larcenist but not a three-time embezzler.

The Supreme Court agreed: “When the law lays an unequal hand on those who have

committed intrinsically the same quality of offense . . . it has made an invidious

6
Virginia asks us to rely on cases like Waddell v. Department of Correction, 680
F.3d 384, 390 n.5 (4th Cir. 2012), which hold that there is no equal protection issue where
prisoners are given different sentences for their crimes. See also United States v. Pierce,
409 F.3d 228, 235 (4th Cir. 2005); United States v. Hughes, 632 F.3d 956, 961 (6th Cir.
2011); Hardin v. State, 587 S.E.2d 634, 636 (Ga. 2003); People v. Strean, 74 P.3d 387, 395
(Colo. App. 2002). But see People v. Hofsheier, 129 P.3d 29, 36 (Cal. 2006) (“It may well
be that in most cases . . . persons who commit different crimes are not similarly situated,
but there is not and cannot be an absolute rule to this effect, because the decision of the
Legislature to distinguish between similar criminal acts is itself a decision subject to equal
protection scrutiny.”), rev’d on other grounds, Johnson v. Dep’t of Justice, 341 P.3d 1075
(Cal. 2015). But those cases deal with direct punishment; here we are dealing with
collateral consequences applied categorically to criminal offenses, and in such cases,
Skinner controls. Cf. Rinaldi v. Yeager, 384 U.S. 305, 308–09 (1966) (“The Equal
Protection Clause requires more of a state law than nondiscriminatory application within
the class it establishes. It also imposes a requirement of some rationality in the nature of
the class singled out.”).
12
discrimination.” Id. at 542. At the next step the Court struck down the law, finding the

different treatment of similarly situated defendants unjustified. While that outcome was

reached under strict scrutiny and not the rational-basis test we use here, by the nature of

the two-step equal protection inquiry described above, the Court’s opinion necessarily

included a holding that the larcenist and the embezzler were similarly situated with respect

to the Oklahoma sterilization law.

Virginia wants to distinguish Skinner by the fact that it applies strict scrutiny, which

of course it does, but that puts the cart before the horse. The level of scrutiny is the second

step in the analysis. At this first step, we only ask whether the defendant and his

hypothetical counterpart are similarly situated. On that question, Skinner binds us. When

a law imposes collateral consequences based on criminal convictions, two impacted

offenders who are treated differently can be similarly situated if their convictions are

similar enough. The only question is how similar the crimes must be.

Consider first the two crimes at issue in Skinner. Larceny and embezzlement have

much in common. Larceny requires taking possession of another’s property with the intent

to steal and without consent. See Wm. L. Clark & Wm. L. Marshall, A Treatise on the Law

of Crimes §§ 303, 315, at 378, 399 (3d ed. 1927); William E. Mikell, Criminal Law, 15

Modern American Law § 164, at 177 (Eugene Allen Gilmore ed., 1921); see also 3 Wayne

R. LaFave, Substantive Criminal Law § 19.1(a) (3d ed., 2020 update). Embezzlement

requires the fraudulent conversion of the property of another by someone who is already

in lawful possession. Clark & Marshall, supra, § 341, at 450; Mikell, supra, § 164, at 177;

LaFave, supra, § 19.6(a). Both crimes, therefore, involve the unauthorized taking of

13
another’s property. And in Oklahoma at the time, larceny and embezzlement were

punished in the same manner when the same amount was taken. Skinner, 361 U.S. at 538

(citing 21 Okla. Stat. tit. 21, §§ 1462, 1705 (1942)). Whether $20 was stolen by larceny or

stolen by embezzlement, the punishment was the same.

In Justice Douglas’s words, the only real distinction between the two is “when the

felonious intent arose.” Id. at 539. 7 In fact, the reason we have two crimes instead of one

is largely path dependent: The creation of embezzlement was to fill a loophole in the old

common law definition of larceny, which required a “trespass in the taking” and let many

would-be embezzlers off the hook. See LaFave, supra, § 19.6(a). For this reason, the

Model Penal Code suggests consolidating larceny and embezzlement (among other crimes)

into a single offense called simply “theft.” Model Penal Code § 223.1 explanatory note for

Sections 223.1-223.9 (Am. L. Inst., Proposed Official Draft 1962). So when the Supreme

Court looked at the “highly technical” and “close distinctions” between the crimes that are

based in part on historical accident, they held that “the nature of the two crimes is

intrinsically the same.” Skinner, 316 U.S. at 539–40, 542. The larcenist and the embezzler

were sufficiently similarly situated with respect to the sterilization law.

Even with Skinner as our guide, it is not exactly clear how to sort distinctions that

alter the very nature of a crime from distinctions that merely rearrange the furniture.

7
That probably understates the difference. It assumes that every larcenist would
eventually obtain lawful possession if they were only patient enough to wait. In fact, it is
likely that many larcenists would never have the opportunity to embezzle because they
would never be given lawful possession. So there is something more going on here than a
simple difference in timing.
14
Virginia points to four distinctions that make indecent liberties and carnal knowledge

different enough to escape Skinner: the acts and mental states required to commit each

crime, the age of the defendants, and the age of the victims. We discuss each in turn.

To start, while the crimes certainly proscribe different acts, that is only true to an

extent. Carnal knowledge is precise; it covers only the act of sex with a child, without

force, and sometimes, without consent. § 18.2-63. Indecent liberties is broader; it covers

any of the following: exposing one’s genitals to a child or asking the child to expose

himself; proposing that the child feel or fondle himself or others; proposing sex, anal sex,

cunnilingus, fellatio, or anilingus to the child; or enticing or inviting the child into some

place to do anything else on the list. § 18.2-370. There is no denying that the crimes cover

different territory, but there is significant similarity. Indecent liberties may not quite be a

lesser-included offense of carnal knowledge, because there are no actual overlapping

elements, but indecent liberties includes the potential avenues leading up to carnal

knowledge: enticing, inviting, proposing, and undressing. In short, it is hard to imagine

how one might commit carnal knowledge without committing indecent liberties on the

way. Viewed from another angle, any offender convicted of carnal knowledge likely has

gone at least as far as indecent liberties; the usual difference is he has gone further. While

there are differences here, the differences mostly comprise the worse behavior of the better-

treated offender, so it is not clear that these differences separate the intrinsic nature of the

crimes.

Next, the mental states. There are two possible differences in the mental states of

these crimes: a “lascivious intent” element and a knowledge-of-age element. But again, it

15
is not clear that either definitively splits these crimes. First, there is some unpublished

Virginia caselaw that suggests that, unlike carnal knowledge, indecent liberties may require

knowledge of the victim’s age. See Detzler v. Commonwealth, No. 1779-08-4, 2010 WL

1286350, at *3 (Va. Ct. App. Apr. 6, 2010). But the better reading of how Virginia

interprets its own law is that neither crime requires knowledge of the age of the victim:

“[W]here sexual crimes involve actual or attempted physical contact with a minor, a

defendant’s knowledge of the victim's age need not be shown in order to prove the

defendant’s guilt.” Kilpatrick v. Commonwealth, 857 S.E.2d 163, 174 (Va. Ct. App. 2021);

see also Rainey v. Commonwealth, 193 S.E. 501, 501–02 (Va. 1937); Va. Prac. Jury

Instruction § 88:15 (“Even if the defendant did not know her actual age, for any reason at

all, that is not a defense to [Va. Code § 18.2-63].”).

For the second possible mental-state difference, the indecent liberties statute

requires “lascivious intent,” but the carnal knowledge statute has no such language. That

phrase “describes a state of mind that is eager for sexual indulgence, desirous of inciting

to lust or of inciting sexual desire and appetite.” McKeon v. Commonwealth, 175 S.E.2d

282, 284 (Va. 1970). On a first pass, it is hard to see how there is any meaningful difference

between the mental states of these two crimes where one requires a desire to indulge in sex

and the other requires the actual indulgence in sex. While it might gesture at a difference

between normal sexual desire and something “shameful or morbid,” cf. Roth v. United

States, 354 U.S. 476, 487 & n.20 (1957), carnal knowledge requires its own “shameful or

morbid” act—the act of sex with a child under the age of qualified consent. Instead,

“lascivious intent” might make an exception for less morally culpable cases, like the loving

16
boyfriend. But the facts of this very case disprove that reading, since by all accounts, Doe

was in a romantic relationship with his victim and was convicted of indecent liberties

anyway. So again, it is hard to say that any difference in the mental states is more than

“highly technical.”

Finally, the crimes involve different age requirements for both defendants and

victims. Indecent liberties can only be committed by adults. Va. Code § 18.2-370 (“Any

person 18 years of age or over . . . ”). Carnal knowledge has no age limit, so it will include

some child-offenders. § 18.2-63. Carnal knowledge can only be committed against 13- or

14-year-olds. Id. 8 Indecent liberties can be committed against “any child under the age of

15 years.” § 18.2-370. Virginia argues that these differences alone make the crimes

dissimilar enough to avoid any scrutiny under the Equal Protection Clause. That only one

statute contemplates child-offenders and the other statute contemplates much younger

victims, see, e.g., Bass v. Commonwealth, 829 S.E.2d 554, 556 (Va. Ct. App. 2019) (8-

year-old victim), may well separate these crimes enough to avoid Skinner.

In response, Doe argues that if you view the crimes through the lens of the Romeo-

and-Juliet provision, the age differences mostly disappear. For example, because indecent

liberties can only be committed by those 18 or older, the 5-year-gap provision would, if

extended to indecent liberties, only ever apply to 13-year-old victims at the youngest, just

the same as carnal knowledge. Something similar happens with the age of the defendants.

8
This may seem like a curiously narrow law, but it makes sense in context. Under
Virginia law, sex with a child under 13 is the more serious crime of rape, and on the other
side, it is only a misdemeanor for an adult to have sex with a 15-, 16-, or 17-year-old. Va.
Code § 18.2-371. Carnal knowledge fits between these two.
17
If the provision was applied to both crimes, it is a mathematical certainty that some adults

would get the benefit under both, notwithstanding that carnal knowledge can be committed

by children while indecent liberties cannot. So Doe correctly points out that, once you

accept the framing of the 5-year gap, we know that even if the provision was applied to

both crimes, it would never apply to child-victims younger than 13 and would always

benefit some adult offenders. In other words, Doe argues that because the mitigating

provision narrowly focuses on victims and defendants who are close in age, the broader

age differences between the crimes are no longer a “relevant respect” in which the crimes

are different. See Nordlinger, 505 U.S. at 10 (emphasis added).

In total then, Doe contends that under Skinner and viewed properly through the

Romeo-and-Juliet provision, indecent liberties and carnal knowledge are “intrinsically the

same quality of offense”—both are Tier III sex crimes with child victims and young-adult

defendants, that are nonviolent and that do not include other aggravating factors. To ratify

that argument, we would need agree that none of the offered distinctions—neither

separately nor in the aggregate—make the crimes intrinsically different. Further, we would

need to use the Romeo-and-Juliet provision as Doe suggests, as a lens to simplify away the

age distinctions between the crimes. It is not clear whether we would be justified in doing

so, and even that step would not simplify away the fact that only carnal knowledge

contemplates child-offenders. But in any event, we need not take that step. We can assume

without deciding that Doe is similarly situated to the offender convicted of carnal

knowledge because Doe’s claim will fail at the next step.

18
 2. Rational-Basis Review

Now we move to constitutional scrutiny, and with no suspect class or fundamental

right, we apply rational-basis review. This test represents a powerful presumption of

validity. Lyng v. Auto. Workers, 485 U.S. 360, 370 (1988). The showing required to

overturn that presumption is steep. A challenger must show there is no “rational

relationship between the disparity of treatment and some legitimate governmental

purpose.” Heller v. Doe, 509 U.S. 312, 320 (1993). The state need not make any showing;

no evidence of any kind is required; reasonable speculation is enough. FCC v. Beach

Commc’ns, Inc., 508 U.S. 307, 315 (1993). As for the justification, any conceivable reason

will do. It does not matter what motivated the classification. Id. at 315. Nor is there any

place in rational-basis review to question the wisdom or logic of a state’s legislation; rough

line-drawing, even “illogical” or “unscientific” line drawing, is often necessary to

governing. Heller, 509 U.S. at 321 (citing Metropolis Theater Co. v. City of Chicago, 228

U.S. 61, 69–70 (1913)). And unlike higher levels of scrutiny, there is no tailoring

requirement under rational-basis review. Id. All that is needed is an imperfect fit between

a plausible reason and some legitimate end. In total, this test requires an extraordinary

showing by a plaintiff like Doe: He must “‘negate every conceivable basis which might

support’ the legislation.” Giarratano v. Johnson, 521 F.3d 298, 303 (4th Cir. 2008)

(quoting Lehnhausen v. Lake Shore Auto Parts Co., 410 U.S. 356, 364 (1973)).

Before applying this test, a clarification about the level of generality at which we

should apply it. For an equal protection challenge, we must justify the classification itself,

not the broad statutory scheme in which the classification resides. See Beach Commc’ns,

19
508 U.S. at 313; Sylvia Dev. Corp. v. Calvert Cnty., 48 F.3d 810, 819 (4th Cir. 1995) (“If

the classification utilized is explicitly stated on the face of a statute . . . , then the equal

protection analysis requires us to determine whether an appropriate relationship exists

between the legislative purpose and the classification adopted to achieve that purpose.”).

But we must also avoid focusing too precisely on how “general lines . . . may affect

particular individuals.” Wilson v. Lyng, 856 F.2d 630, 633 (4th Cir. 1988). Our focus must

be on the provision, not the person and not the whole scheme. At this point, we need not

decide whether the sex-offender registry as a whole is justifiable—though it surely is. Here

we must justify the Romeo-and-Juliet provision that provides mitigation only for some;

that is the classification that Doe challenges.

One argument that Virginia makes in defense of its law is that the two crimes are

different in nature, as if that justifies the worse treatment for indecent liberties. This is

where Doe’s common-sense argument is at its best. We do not see how worse treatment

for people in Doe’s shoes is rationally related to the different natures of the acts. Carnal

knowledge is by all accounts the worse crime. Sex with a child is worse than asking the

child for sex. And a rational response would be to treat the worse crime worse. Cf. People

in Int. of Z.B., 757 N.W.2d 595, 600 (S.D. 2008) (finding an equal protection violation

where juvenile sex offenders were treated worse than adult sex offenders). Virginia does

just that in sentencing. Compare Va. Code § 18.2-63 (carnal knowledge is a Class 4

felony), with § 18.2-370 (indecent liberties is a Class 5 felony), by way of § 18.2-10

(describing the more severe punishments for Class 4 felonies). So if this were the only

justification that could uphold this classification, it may well fail.

20
 But there is a better justification. In fact, we are tasked with imagining any

conceivable justification for this classification, see Beach Commc’ns, 508 U.S. at 315, and

there is at least one that will do: ensuring that children do not become Tier III sex

offenders. Above we mentioned how carnal knowledge can involve 15-, 16-, and 17-year-

old offenders and how indecent liberties only ever involves offenders over 18. While that

minor distinction may not definitively separate the two crimes in our similarly situated

analysis, it is decisive here. We do not doubt that the government has a legitimate interest

in not imposing its harshest collateral consequences on children, even children who commit

serious felonies. Cf. Graham v. Florida, 560 U.S. 48, 68 (2010). And this 5-year-gap

provision is at least rationally related to that purpose. It ensures that, even though children

21
can be charged and convicted of carnal knowledge, they will not become Tier III offenders

on the sex-offender registry. 9 That is enough to uphold this distinction. 10

The differential treatment between carnal knowledge and indecent liberties

offenders satisfies rational-basis scrutiny. So Doe’s equal protection claim must fail, and

we affirm the district court.

9
More detail is needed to fully bring the point home. Carnal knowledge is split into
two subsections, which are treated separately by the registry, 18.2-63(A) and 18.2-63(B).
Subsection (A) provides: “If any person carnally knows, without the use of force, a
child . . . , such person shall be guilty.” That subsection is Tier III as a default. Subsection
(B) reads: “If any person carnally knows, without the use of force, a child . . . who consents
to sexual intercourse and the accused is a minor . . . , the accused shall be guilty.”
(emphasis added). That subsection is Tier I as a default. The Romeo-and-Juliet provision
only applies to subsection (A) and not to subsection (B). At first blush, it might appear
that all child-perpetrators would fall into subsection (B), which looks to see whether the
“accused is a minor.” That might defeat this protect-the-kids justification for the provision
granting mercy to subsection (A) offenders. But there is yet another distinction in
subsection (B) that is missing from (A): “consent.” So the language of the statute appears
to allow some children to be prosecuted under (A)—those who carnally know a child,
without force, but also without consent. The category of no-force, no-consent is with all
likelihood a small one, but it is conceivable that the 5-year-gap provision was meant to
ensure that any child who fell into that category would be spared Tier III treatment.
10
This may not have been the real motivation for this provision. Va. Code § 9.1-
902 (using a more natural phrase to protect children, “where the perpetrator is 18 years of
age or older,” in the same statutory sentence as our Romeo-and-Juliet provision); S.B. 590,
2008 Sess. (Va. 2008) (offering an early draft of the Romeo-and-Juliet provision that
includes both an under-21 clause and the 5-year-gap clause, suggesting that the 5-year-gap
was not meant solely to protect child perpetrators); Virginia Crime Commission, “Romeo
and Juliet” Laws (2007) (describing how best to implement Romeo-and-Juliet laws in
Virginia, published to the legislature a year before the provision here was added); see also
Report of the Va. Crime Comm’n, Sex Offenders in Virginia 21 (2006) (discussing how
carnal knowledge involved the highest recidivism rates among covered sex crimes (34%)
and indecent liberties was among the lowest (14–18%)). But all that is irrelevant to the
rational-basis analysis where we operate in pure hypothetical. See Beach Commc’ns, 508
U.S. at 315.
22
 B. Eighth Amendment Claim

Doe also claims that his placement on the Virginia sex-offender registry violates the

Eighth Amendment’s prohibition on “cruel and unusual punishments.” U.S. Const. amend.

VIII. Because the Clause only regulates “punishments,” we must begin our analysis by

determining whether Virginia’s sex-offender registry is “punishment” before moving on to

consider whether it is cruel and unusual. A cruel-and-unusual regulation may violate other

constitutional protections, but unless it is a punishment, the Eighth Amendment does not

apply.

The Supreme Court has created a two-part test for determining whether a statute

imposes punishment. First, we must ask if the legislature intended to inflict punishment,

which is a question of statutory interpretation. Smith v. Doe, 538 U.S. 84, 92 (2003). 11 If

we find the intent was punitive, that is end of the inquiry. If not, we then must look to the

effects of the law. Id. If the effects are punitive, they may override the legislature’s intent,

but we must give deference to the legislature on this point, and we will require “the clearest

proof” to overturn those intentions. Id. at 105. The Court has devised a multi-factor test

to determine whether the punitive effect of the law is so overwhelming that it negates the

State’s nonpunitive intentions. See Kennedy v. Mendoza-Martinez, 372 U.S. 144, 168–69

Smith is an Ex Post Facto Clause case, but the punishment analysis is the same in
11

both contexts. See United States v. Under Seal, 709 F.3d 257, 263 (4th Cir. 2013); Does
1-7 v. Abbott, 945 F.3d 307, 313 n.9 (5th Cir. 2019).

23
(1963); Smith, 538 U.S. at 96. 12 The Virginia sex-offender registry is not a punishment at

either step.

1. Punitive Intent

We start by looking to the Virginia legislature’s intent, which is largely a question

of statutory interpretation, Smith, 538 U.S. at 92, and helpfully, Virginia’s intention is

12
Other legal commentators and linguists have defined punishment more
straightforwardly. William Blackstone, for instance, defined punishment in his famous
Commentaries as “evils or inconveniences consequent upon crimes and misdemeanors;
being devised, denounced, and inflicted by human laws, in consequence of disobedience
or misbehavior in those to regulate whose conduct such laws were respectively made.” 4
William Blackstone, Commentaries on the Laws of England *7. He thought that
punishment was “chiefly intended for the prevention of future crime,” id. at *12, 16, and
that prevention was accomplished in three primary ways: by amending the offender’s
behavior, by depriving the offender of the ability to do evil, or by making “a terror of his
example,” id. at *12.
The legal philosopher H.L.A. Hart used a similar definition. For a legal obligation
to be characterized a legal punishment: “(i) It must involve pain or other consequences
normally considered unpleasant. (ii) It must be for an offence against legal rules. (iii) It
must be of an actual or supposed offender for his offence. (iv) It must be intentionally
administered by human beings other than the offender. (v) It must be imposed and
administered by an authority constituted by a legal system against which the offence is
committed.” See H.L.A. Hart, Punishment and Responsibility 4–5 (1968).
Justice Stephen Field offered this definition in the 19th century: “The deprivation
of any rights, civil or political, previously enjoyed, may be punishment, the circumstances
attending and the causes of the deprivation determining this fact.” Cummings v. Missouri,
71 U.S. 277, 286 (1866).
Founding era legal dictionaries echo this focus on the cause-and-effect interaction
between a legal violation and the imposition of some negative consequence by the state.
See, e.g., Samuel Johnson, A Dictionary of the English Language, (10th ed. 1792) (“Any
infliction imposed in vengeance of a crime”); Noah Webster, An American Dictionary of
the English Language (1828) (“Any pain or suffering inflicted on a person for a crime or
offense, by the authority to which the offender is subject”); Giles Jacob, A New Law
Dictionary (1782) (“Is the penalty for transgressing the law: And as debts are discharged
to private persons by payment; so obligations to the public, for disturbing society, are
discharged when the offender undergoes the punishment inflicted for his offense.”).
But we follow the Supreme Court’s two-part test.
24
written right into the code. Section 9.1-900 is called “Purposes of the Sex Offender and

Crimes Against Minors Registry,” and it aims “to assist the efforts of law-enforcement

agencies and others to protect their communities and families from repeat offenders and to

protect children from becoming victims of criminal offenders by helping to prevent such

individuals from being allowed to work directly with children.” Though community

protection and deterrence are goals of both criminal and civil legislation, the Supreme

Court in Smith tells us that a state can pursue objectives like these through a regulatory

scheme without making the statute a “punishment.” 538 U.S. at 93–94 (holding that a

statement of purpose to “protect[] the public from sex offenders” showed nonpunitive

legislative intent); see also United States v. One Assortment of 89 Firearms, 465 U.S. 354,

364 (1984) (“Keeping potentially dangerous weapons out of the hands of unlicensed

dealers is a goal plainly more remedial than punitive.”). Considering Smith, the Virginia

sex-offender registry and its statement of purpose cannot be read as expressing a punitive

intent. See also Prynne v. Settle, 848 F. App’x 93, 100 (4th Cir. 2021).

Doe makes several contrary arguments, but none can shift our reading of the

statutory purpose. He points out that some of the registry’s requirements are set out in the

Virginia criminal code. See, e.g., Va. Code § 18.2-370.5 (prohibiting entering schools and

daycares). The Supreme Court has rejected an argument just like this one. Smith, 538 U.S.

at 94 (“The location and labels of a statutory provision do not by themselves transform a

civil remedy into a criminal one.”). Doe also argues that punitive intent is shown by the

fact that the registry only applies to convicts and is administered by law enforcement. The

Supreme Court has rejected that argument too. See id. at 95–96.

25
 Doe’s best argument has not yet been resolved by the Supreme Court. He notes that

the sex-offender registry was originally set up within the criminal code and lacked any

statement of purpose. See Va. Code §§ 19.2-298.1–298.4, repealed by 2003 Va. Acts ch.

584. It was only after Smith that the Virginia legislature moved the registry and added the

statement of purpose. See generally 2003 Va. Acts ch. 584. A cynical mind might read

this history as gamesmanship, reconfiguring the state law to avoid federal constitutional

review. But we read this as more evidence that Virginia has no punitive intent and simply

wanted to be clear about it.

Even if we were tempted to question the legislature’s sincerity, a Virginia

intermediate appellate court has interpreted this statute and held that it is not penal. Kitze

v. Commonwealth, 475 S.E.2d 830, 832 (Va. Ct. App. 1996). And “we generally treat

intermediate appellate-court decisions as good evidence of state law,” absent good reason

to doubt their conclusions. United States v. Smith, 939 F.3d 612, 617 (4th Cir. 2019). As

explained above, we have every reason to agree with the Virginia court’s statutory

interpretation that the Virginia legislature’s intent was not penal. So we will not second-

guess both the Virginia legislature and the Virginia judiciary on this point.

Because we find that the Virginia sex-offender registry was intended to be a civil

regulation and not a punishment, we move on to Smith’s next step and ask if the law’s

punitive effect is so overwhelming that it negates the State’s intentions.

2. Punitive Effect

To assess punitive effect, we look to the list of seven factors first compiled in

Kennedy v. Mendoza-Martinez, 372 U.S. at 168–69, and then adopted in Smith. 538 U.S.

26
at 97. These factors have been used in a handful of constitutional contexts—Ex Post Facto

Clause, Sixth Amendment, and Eighth Amendment—and they create a framework for a

general, constitutional theory of a “punishment.” See id. In the sex-registry context, the

Supreme Court has focused on 5 of these factors, whether the scheme:

(1) has been regarded in our history and traditions as punishment;
(2) imposes an affirmative disability or restraint;
(3) promotes the traditional aims of punishment;
(4) has a rational connection to a nonpunitive purpose; [and]
(5) is excessive with respect to this purpose.

United States v. Wass, 954 F.3d 184, 193 (4th Cir. 2020) (quoting Smith, 538 U.S. at 105)

(cleaned up and reformatted). 13 We must consider each of these factors in turn and

determine whether the sum provides the “clearest proof” that the registry is punitive in

effect. Smith, 538 U.S. at 105.

But we do not operate in the open field; both the Supreme Court and this Court have

applied this test to sex-offender-registration laws before—the Supreme Court considered

the Alaska Sex Offender Registration Act in Smith and this Court considered the federal

Sex Offender Registration and Notification Act in Under Seal. 14 Each court found the

13
In Smith, Justice Kennedy declines to consider a sixth Mendoza-Martinez factor,
whether the behavior was already a crime before the law was passed, because, he says,
“recidivism was the statutory concern” of the registration scheme. 538 U.S. at 105; see
also Under Seal, 709 F.3d at 264 n.5. While common-sense definitions of punishment
often look to whether the “evils [are] consequent upon crimes,” 4 Blackstone, at *7; see
also Hart, at 4–5, we are bound by the Supreme Court’s reasoning on this point and will
not consider this factor.
14
We have also considered the punitive effects of Virginia’s registry, but only in
unpublished opinions. Prynne, 848 F. App’x 93; Ballard v. F.B.I., Chief, 102 F. App’x
828, 829 (4th Cir. 2004).

27
registry to lack the overwhelming punitive effects needed to overturn legislative intent. In

Smith, Alaska’s sex-offender registry and notification law included many of the same

regulations as the Virginia registry here: prompt registration, fingerprinting, large amounts

of information, notification of movement, the threat of criminal prosecution for

noncompliance, and the posting of a convict’s photo, information, and status on the

internet. See 538 U.S. at 89–91. In Under Seal, the federal registry included a statement

of purpose much like the one in Virginia’s registry and most of the same requirements:

registration, regular in-person verification, and the provision and publication of personal

information. See 709 F.3d at 260–61. While we examine Virginia’s registry under the

Mendoza-Martinez factors and find some differences from the schemes considered in Smith

and Under Seal, we cannot find that Virginia’s registry is so different from these other

registries that the result is different.

Rational connection to a nonpunitive purpose. Because Smith tells us that this is

the most significant factor in assessing punitive effect, we begin here, and Virginia’s sex-

offender registry is rationally connected to the legitimate goal of public safety. 538 U.S.

at 102 (holding that the statute need not have “a close or perfect fit with the nonpunitive

aims it seeks to advance”); Doe v. Miller, 405 F.3d 700, 721 (8th Cir. 2005) (suggesting

that this factor “is not demanding”). Both Smith and Under Seal found that the registries

there related to the purpose of “alerting the public to the risk of sex offenders in their

communit[y].” See Smith, 538 U.S. at 103; Under Seal, 709 F.3d at 265. The same is true

here. The statute explicitly adopts this purpose. Va. Code § 9.1-900. Public safety is a

28
quintessentially legitimate justification, and these registries are no doubt connected to that

goal. See Shaw v. Patton, 823 F.3d 556, 573–74 (10th Cir. 2016).

Some courts have found that sex-offender registries may increase recidivism and

therefore harm public safety, but the Virginia legislature is free to disagree with that

empirical prediction or pursue other goals like investigatory efficiency. See Kitze, 475

S.E.2d at 832–34 (offering similar justifications for the Virginia law); Am. C.L. Union of

Nevada v. Masto, 670 F.3d 1046, 1057 (9th Cir. 2012). But see Does #1-5 v. Snyder, 834

F.3d 696, 704–05 (6th Cir. 2016) (finding that studies showing a limited effect on

recidivism suggest there is not a reasonable fit); In Int. of T.H., 913 N.W.2d 578, 595 (Iowa

2018) (same for juveniles); People ex rel. T.B., 489 P.3d 752, 768 (Colo. 2021) (same).

Because this factor favors Virginia and because it is the most important factor, this

is strong evidence that the law is nonpunitive in effect.

Excessive with respect to its purposes. This excessiveness inquiry is not about

second-guessing the legislature’s choice of solution; we only ask whether the chosen means

are “reasonable in light of the nonpunitive objective.” Smith, 538 U.S. at 105. This inquiry

is reminiscent of the rational basis test we have already considered; the Virginia legislature

is due our deference, and we may only question their work in rare circumstances. We need

only confirm that fit between the purpose and its execution is reasonable. The registry

furthers the nonpunitive goal of public safety.

Doe argues that it is excessive in doing so. We disagree. The information required

of the offenders is useful and relevant to the purposes of the law and helps ensure that

police and the public can make informed decisions. The limited work restrictions and

29
limits on entering schools without permission are reasonably related to protecting the

vulnerable and preventing recidivism.

That Doe himself may not pose a danger is beside the point. The Eighth Amendment

“does not preclude a State from making reasonable categorical judgments that conviction

of specified crimes should entail particular regulatory consequences.” Smith, 538 U.S. at

106; Miller, 405 F.3d at 721–22.

Virginia’s registry differs in some respects from those in Under Seal and Smith but

those differences do not make it unreasonably overbroad. The federal sex registry also

uses a three-tier system, but the third tier of Virginia’s system is arguably broader.

Compare 34 U.S.C. § 20911, with Va. Code § 9.1-903. The Alaska registry considered in

Smith had a default registration requirement of only 15 years, without any need for a

petition to be removed, and lifetime registration was only imposed for an aggravated

offense or a second offense. 538 U.S. at 90; cf. id. at 104 (suggesting the duration of the

requirement is not excessive given some recidivism may happen as late as 20 years later).

Virginia is more severe on both counts. With that said, we still echo the Supreme Court’s

assessment in Smith: This registry is “consistent with grave concerns over the high rate of

recidivism among convicted sex offenders and their dangerousness as a class.” Id. at 103.

All of Doe’s other arguments would have us apply a more stringent fit analysis than is

required by Smith. While the law may be overbroad in some sense—e.g., the lascivious

flasher may not require lifelong registration and tracking by the State—we do not find it

unreasonably excessive as a matter of law.

30
 Historical punishments. Next, we look to a survey of historical punishments

“because a State that decides to punish an individual is likely to select a means deemed

punitive in our tradition.” Smith, 538 U.S. at 97. Because offender registries are a modern

invention, we proceed by analogy, comparing the registration scheme to historical

punishments to consider whether the effects are the same. See id. at 97–99.

The first potential comparison is public shaming. There is a whole suite of strange

old punishments that fit into this category. In his Commentaries, Blackstone includes in

his list of punishments that “consist principally in their ignominy” things like “the pillory,

the stocks, and the ducking-stool.” 4 Blackstone, at *370. These punishments often

involved standing in public with a sign describing your crime. See Second Trial of Titus

Oates, 10 How. St. Tr. 1227, 1315 (K.B. 1685) (“Thirdly, The Court does award, That you

do stand upon the Pillory, and in the Pillory, here before Westminster-hall gate, upon

Monday next, for an hour’s time, between the hours of 10 and 12; with a paper over your

head (which you must first walk with round about to all the Courts in Westminster-hall)

declaring your crime.”). Sometimes a murderer might be branded with the letter “M” to

shame him. See Smith, 538 U.S. at 98 (citing R. Semmes, Crime and Punishment in Early

Maryland 35 (1938)). There is an intuitive comparison between these kinds of

punishments and the public dissemination of sex-crimes information, but that is not enough

to make them analogs.

The courts in Smith and Under Seal rejected this comparison. One important

distinction is that those old punishments involved physical pain and direct confrontation

with the community. Smith, 538 U.S. at 98; see also 4 Blackstone, at *370 (“[M]ost of

31
them are mixed with some degree of corporal pain.”). But that was not always true; for a

famous example, take the real Massachusetts law that inspired Nathaniel Hawthorne’s The

Scarlet Letter. As punishment for adultery, the guilty were required to “for ever after wear

a capital A, of two inches long and proportionable bigness, cut out in cloth of a contrary

colour to their cloaths, and sewed upon their upper garments, on the outside of their arm,

or on their back, in open view.” An Act, Against Adultery and Polygamie, Acts and Laws,

Passed by the Great and General Court or Assembly of their Majesties Province of the

Massachusetts-Bay in New-England 72 (1694).

That Massachusetts law proves that some shaming punishments did not necessarily

involve physical pain or mass public confrontation. But this is the closest to an analog Doe

might point to, and even this punishment was more degrading than a photo on a website.

More importantly, “[o]ur system does not treat dissemination of truthful information in

furtherance of a legitimate governmental objective as punishment.” Under Seal, 709 F.3d

at 265 (quoting Smith, 538 U.S. at 98).

Other details do not change things. Designating the crime “Tier III” might add some

extra shame, but convictions are already public, and it is hard to see how this label would

add much embarrassment. Compare Prynne, 848 F. App’x at 101–03, with id. at 110

(Agee, J., dissenting). The “geographic reach of the Internet” brings in other

complications, considering that a sex offender can now be identified by anyone in the world

rather than the few people who fit in the town square. But the Supreme Court has held that

this does not make a registry “punitive”; rather, it is “necessary for the efficacy of the

32
scheme.” 538 U.S. at 99. Virginia’s sex-offender registry is not like historical shaming

punishments.

Another candidate is banishment. The comparison seems strained and metaphorical

at first glance, but the Sixth Circuit has found the analogy useful. See Snyder, 834 F.3d at

701–02. That court found that a limitation on where sex offenders can live and work is

“very burdensome, especially in densely populated areas.” Id. (including a map of the

Grand Rapids area illustrating this point). But that scheme imposed restrictions of 1,000

feet. Id. Virginia law prohibits some sex offenders from loitering within 100 feet of a

school or daycare. Va. Code § 18.2-370.2. That restriction is an order of magnitude less

severe than the limitation in Snyder, which also involved significant allegations about

hardship that this case does not. 834 F.3d at 701–02. More to the point, the 100-foot

limitation is imposed “as a part of [an offender’s] sentence,” so that limit is not part of the

registry. Va. Code § 18.2-370.2. So even if this geographic restriction resembles

banishment, it is not relevant to Doe’s complaints about the registry. There is no argument

that the registry’s requirements are anything like banishment; it does not make offenders

“dead in law [and] entirely cut off from society.” 1 Blackstone, at *132.

Finally, parole and probation. The Court in Smith said this comparison had “some

force” but rejected it because “[p]robation and supervised release entail a series of

mandatory conditions and allow the supervising officer to seek the revocation of probation

or release in case of infraction.” 538 U.S. at 101–02. The Alaska registration scheme there

did not involve any real supervision. Id. But this Virginia scheme creates criminal

violations for disregarding the registry’s rules. See, e.g., Va. Code § 18.2-370.5. The

33
Virginia scheme also requires the State Police to physically verify an offender’s

registration information twice a year, § 9.1-902, and Doe alleges that this requirement

involves random home visits to verify the required information. “Historically, a probation

officer took a far more active role in a probationer’s life than simply collecting information

for a database.” Shaw, 823 F.3d at 564 (in-person reporting to officers is not like the many

requirements of parole). Taken as a whole, the Virginia registry may have shades of

probation, but probation often involves a greater degree of intrusion and regularity.

We also note two founding era laws that suggest that—far from being considered

punishments—registries and other publication of personal information would have been

considered common regulatory tools. First, consider the census. In 1790, the United States

held its first census. United States Census Bureau, 1790 Overview, History, Census.gov

(last visited Jan. 20, 2022). The Census Act of 1790 required not only that States count

their inhabitants but also that a copy of the official schedule of the census for each district

be “set up at two of the most public places” in the district, “there to remain for the

inspection of all concerned.” An Act Providing for the enumeration of the Inhabitants of

the United States, ch. 2, § 6, 1 Stat. 101, 101–03 (1790). That schedule included the names

of the heads of all the counted families. § 1, 1 Stat. at 101–02; see also United States

Census Bureau, 1790 Census: Heads of Families at the First Census, Census.gov (last

visited Jan. 20, 2022). Of course, there is nothing inherently damaging in having your

name publicly listed as the head of a household, so in that way, it is distinguishable from a

sex-offender registry. With that said, this example provides at least one data point that,

historically, public registries were used as regulatory tools and not punishments. The next

34
example shows that even registries publishing negative information were not considered

punishments.

Less than a decade later, Congress passed the infamous Alien and Sedition Acts,

which included a registry for all aliens. See An Act supplementary to and to amend the

act, intituled “An act to establish an uniform rule of naturalization; and to repeal the act

heretofore passed on that subject,” Ch. 54, § 4, 1 Stat. 566, 567 (1798). Aliens had to report

to the local district clerk and provide their name, age, residence, and occupation; that

registration information was then sent to the Secretary of State; and anyone who was caught

avoiding this requirement would be fined upwards of $2 and sometimes even “committed

to the common gaol.” § 5, 1 Stat. at 568. Not only were new aliens required to register,

aliens who were already here had to register within 6 months of the passage of the law—

which might have raised the specter of the Ex Post Facto Clause if the founding generation

thought of mere registration as a punishment.

In sum, while there are similarities between historical punishments and this modern-

day registry, we find that the Virginia sex-offender registry is more like the common

regulatory devices used at the founding than it is like historical forms of punishment.

Affirmative disability or restraint. The Virginia registry imposes some affirmative

disabilities and restraints on offenders like Doe. On this factor, “minor and indirect”

restraints will not be punitive; we look for something more substantial. Smith, 538 U.S. at

100. Imprisonment is the “paradigmatic” example of an affirmative restraint, see id. at 97–

100; Kansas v. Hendricks, 521 U.S. 346, 356 (1997), and Doe is not subject to anything

like prison. That is why Smith and Under Seal found this factor to cut against a finding of

35
punitive effect, because neither registry imposed anything like prison and neither sought to

“restrain activities sex offenders may pursue” or require them to “seek permission” before

acting. Under Seal, 709 F.3d at 265 (quoting Smith, 538 U.S. at 100). The same is true

here, as one can move or change jobs without permission. Id.

Virginia’s scheme includes some restrictions and impositions, but they do not

approach the level of restraint imposed by a prison sentence. To start, offenders cannot

hold certain jobs. But job restrictions cannot alone make a punishment. See Smith, 538

U.S. at 97–100. 15 Tier III offenders must also ask permission to enter a school during

school hours, even to visit their own child. See Va. Code § 18.2-370.5. And offenders are

required to register and re-register in person, § 9.1-903, which the Court in Smith implied

may amount to a restraint or disability, see 538 U.S. at 101. The Virginia registry imposes

some disabilities and restraints on people like Doe, but at worst, they are “minor and

indirect.” See id. at 100.

Promotes the traditional aims of punishment. Finally, while Doe can find aspects

of the traditional justifications for punishment in Virginia’s registry, the law is better

understood as promoting public safety. This factor considers whether the government

action promotes the traditional goals of punishment, mainly retribution and deterrence.

Smith, 538 U.S. at 102; see also Under Seal, 709 F.3d at 265. In working through this

factor, the Court in Smith did not find it sufficient to simply state how the law might further

15
The Supreme Court has found various categorical restrictions on occupations to
be nonpunitive. See, e.g., Hawker v. New York, 170 U.S. 189, 197 (1898) (prohibiting
felons from practicing medicine); De Veau v. Braisted, 363 U.S. 144, 160 (1960) (plurality
opinion) (prohibiting felons from being union agents).
36
the goals of retribution and deterrence; to show punitive effect, the Court required

retribution and deterrence to be primary factors of the law. 538 U.S. at 102.

That is simply not the case here. There are wisps of deterrence and retribution, but

not enough to show punitive intent. The threat of the registry does provide some deterrence

to would-be sex offenders, but that is true of many nonpunitive government programs. Id.

While it might look like retribution to place offenders into Tiers based on the seriousness

of their crimes instead of their risk of reoffending, using broad categories and

generalizations is a reasonable way of assessing risk. See id. So all these policies can be

explained as “reasonably related to [the] regulatory objective” of community safety. Id. at

102. The traditional punitive justifications are not the driving force here.

On balance, we find that these factors do not demonstrate punitive effect, especially

considering the deference we must give to the legislature’s intent. Even if we were inclined

to disagree with the legislature, Virginia’s registry does not have punitive effects so far

beyond the registries in Smith and Under Seal that they would make this case

distinguishable. So we hold that the effect of the Virginia sex-offender registry is not so

clearly punitive that it overcomes the intent of the legislature. Therefore, because this

registry is not a punishment, the Eighth Amendment claim was properly dismissed.

3. Remaining Claims

Doe also brings claims under a federal substantive due process theory and under the

Virginia Constitution. Both can be dealt with in short order. A substantive due process

challenge is considered under rational-basis review unless some fundamental right is

implicated. Herndon v. Chapel Hill-Carrboro City Bd. of Ed., 89 F.3d 174, 177 (4th Cir.

37
1996). We hold that Virginia’s sex-offender registry is rationally related to the legitimate

public interest in public safety because it “alert[s] the public to the risk of sex offenders in

their communit[y].” See Smith, 538 U.S. at 102–03.

As to the state claims, the district court declined to exercise supplemental

jurisdiction over them after dismissing all of Doe’s federal claims. Because we agree with

the district court’s dismissal of his federal claims, we also agree with the district court’s

dismissal of his state claims.

* * *

If an 18-year-old man in Virginia has “consensual” sex with his 14-year-old

girlfriend, and the next day, sends her a text message asking her to do it again, he will have

committed two crimes. But under the letter of the law in Virginia, only one of those crimes

will place him on the worst tier of sex offenders on the registry with the rapists and the

murderers: the text message. That may not make much sense.

But our Constitution “presumes that even improvident decisions will eventually be

rectified by the democratic process.” See Cleburne, 473 U.S. at 440. The judiciary is not

meant to revise laws because they are clumsy, unwise, or—even in some cosmic sense—

unfair. In cases like this, courts are asked to make judgments about what is inside and what

is outside the precise lines drawn by the Constitution. And whatever else they may be,

Virginia’s sex-offender registry and its narrow Romeo-and-Juliet provision are

constitutional. Accordingly, the district court’s judgment is

AFFIRMED.

38