Anthony Kebodeaux, a federal sex offender, was convicted, under the Sex Offender Registration and Notification Act (“SORNA”), of failing to update his change of address when he moved intrastate. A panel of this court affirmed. United States v. Kebodeaux, 647, F.3d 137 (5th Cir.2011). The panel majority rejected Kebodeaux’s argument that Congress does not have the power to criminalize his failure to register because it cannot constitutionally reassert jurisdiction over his intrastate activities after his unconditional *234release from federal custody. Judge Dennis concurred in the judgment and assigned lengthy reasons, urging that SORNA is authorized by the Commerce Clause. The panel opinion was vacated by our decision to rehear the case en banc. United States v. Kebodeaux, 647 F.3d 605 (5th Cir.2011). Because we agree with Kebodeaux that, under the specific and limited facts of this case, his commission of a federal crime is an insufficient basis for Congress to assert unending criminal authority over him, we reverse and render a judgment of dismissal.
I.
While in the military, Kebodeaux had consensual sex with a fifteen-year-old when he was twenty-one and was sentenced in 1999 to three months in prison. He fully served that sentence, and the federal government severed all ties with him. He was no longer in federal custody, in the military, under any sort of supervised release or parole, or in any other special relationship with the federal government when Congress enacted a statute that, as interpreted by the Attorney General, required Kebodeaux to register as a sex offender.1 When he failed to update his state registration within three days of moving from San Antonio to El Paso, he was convicted under 18 U.S.C. § 2250(a) (also enacted in 2006) and sentenced to a year and a day in prison.
Kebodeaux argues that § 2250(a)(2)(A) and the registration requirements that it enforces are unconstitutional as applied to him, because they exceed the constitutional powers of the United States. He is correct; Absent some jurisdictional hook not present here, Congress has no Article I power to require a former federal sex offender to register an intrastate change of address after he has served his sentence and has already been unconditionally released from prison and the military.2
The federal requirement that sex offenders register their address is unconstitutional on narrow grounds. We do not call into question Congress’s ability to impose conditions on a prisoner’s release from custody, including requirements that sex offenders register intrastate changes of address after release. After the federal government has unconditionally let a person free, however, the fact that he once committed a crime is not a jurisdictional *235basis for subsequent regulation and possible criminal prosecution. Some other jurisdictional ground, such as interstate travel, is required.3
This finding of unconstitutionality therefore does not affect the registration requirements for (1) any federal sex offender who was in prison or on supervised release when the statute was enacted in 2006 or (2) any federal sex offender convicted since then. Instead, it applies only to those federal sex offenders whom the government deemed capable of being unconditionally released from its jurisdiction before SORNA’s passage in 2006.4 Moreover, even as to those sex offenders, it means only that Congress could treat them exactly as all state sex offenders already are treated under federal law. It also has no impact on state regulation of sex offenders.
II.
SORNA says, in relevant part, that “[a] sex offender shall register, and keep the registration current, in each jurisdiction where the offender resides, where the offender is an employee, and where the offender is a student.”5 Those requirements are made applicable to former federal sex offenders via 42 U.S.C. § 16913(d) and 28 C.F.R. § 72.3.6 SORNA then includes the following criminal provision:
*236Whoever—
(1) is required to register under [SORNA];
(2)(A) is a sex offender as defined for the purposes of [SORNA] by reason of a conviction under Federal law ...; or
(B) travels in interstate or foreign commerce ...; and
(3) knowingly fails to register or update a registration as required by [SORNA];
shall be fined under this title or imprisoned not more than 10 years, or both.
18 U.S.C. § 2250(a). Kebodeaux argues that Congress has no authority under Article I to subject him to conviction pursuant to § 2250(a)(2)(A). The government, on the other hand, maintains that its power to criminalize the conduct for which Kebodeaux was originally convicted includes the authority to regulate his movement even after his sentence has expired and he has been unconditionally released.
The most analogous Supreme Court decision is United States v. Comstock, — U.S. -, 130 S.Ct. 1949, 1954, 176 L.Ed.2d 878 (2010), in which the Court examined whether Congress has the Article I power to enact a civil-commitment statute that authorizes the Department of Justice to detain mentally ill, sexually dangerous federal prisoners beyond when they would otherwise be released. The Court upheld that statute on narrow grounds because of “five considerations, taken together.” Id. at 1956,1965.
Kebodeaux’s facts go beyond those in Comstock, however, because this case is not merely about whether Congress can regulate the activity of someone still in federal custody past the expiry of his sentence. Importantly, it raises the further question whether Congress can regulate his activity solely because he was once convicted of a federal crime. The “considerations” that the Court found important in Comstock are not expansive enough to subject Kebodeaux to federal criminal sanctions under the unusual circumstances that he presents.
A.
First, the Comstock Court explained, and the panel majority here stressed, that Congress has broad authority to enact legislation under the Necessary and Proper Clause. Id. at 1956. Thus, to be constitutional under that clause, a statute must constitute a means that is “rationally related”7 or “reasonably adapted”8 to an enumerated power. Congress has “a large discretion” as to the choice of means, id. at 1957 (quoting Lottery Case, 188 U.S. 321, 355, 23 S.Ct. 321, 47 L.Ed. 492 (1903)), and we apply a “presumption of constitutionality” to its enactments, id. (quoting United States v. Morrison, 529 U.S. 598, 607, 120 S.Ct. 1740, 146 L.Ed.2d 658 (2000)). This first factor is not fact-specific; it suggests that the analysis always *237starts with a heavy thumb on the scale in favor of upholding government action.9
We must take care not to misunderstand the use of the words “rationally related” as implying that the Necessary and Proper Clause test is akin to rational-basis scrutiny under the Due Process and Equal Protection Clauses.10 That would mean that federal action would be upheld so long as there is merely a conceivable rational relationship between an enumerated power and the action in question.11 But that would be inconsistent with both the Court’s Commerce Clause jurisprudence12 and Comstock, which held that 18 U.S.C. § 4248 is constitutional because of “five considerations, taken together,” only one of which involves “the sound reasons for the statute’s enactment in light of the Government’s [legitimate interest].”13 Thus, unless this court were to hold that the other “considerations” in Comstock were entirely superfluous, it follows that, although our analysis begins with great deference to constitutionality, we should not confuse it with Due Process Clause rational-basis scrutiny.
B.
The second factor in Comstock, 130 S.Ct. at 1958, is that the civil-commitment statute at issue was but “a modest addition to a set of federal prison-related mental-*238health statutes that have existed for many decades.” Although “even a longstanding history of related federal action does not demonstrate a statute’s constitutionality,” id. (citing Walz v. Tax Comm’n of N.Y., 397 U.S. 664, 678, 90 S.Ct. 1409, 25 L.Ed.2d 697 (1970)), it expands the deference afforded to a statute.14 Conversely, the absence of an historical analog reduces that deference.15
SORNA’s sex-offender-registration requirements have a short history: They have existed only since 2006, and federal law relating to sex-offender registration only since 1994.16 The government admits that federal sex-offender registration laws are of “relatively recent vintage” but urges that they should be analogized to probation or supervised-release laws, which have a longer pedigree.
There is, however, a big difference between SORNA’s sex-offender-registration requirements and probation or supervised release — a distinction that goes to the heart of this case. Unlike the situation involving probation or supervised release, SORNA’s sex-offender-registration requirements (and § 2250(a)(2)(A)’s penalties) were not a condition of Kebodeaux’s release from prison, let alone a punishment for his crime.17
The Department of Justice cannot find a single authority, from more than two hundred years of precedent, for the proposition that it can reassert jurisdiction over someone it had long ago unconditionally released from custody just because he once committed a federal crime. Thus, SORNA’s registration requirements for federal sex offenders are constitutionally novel, as the panel majority conceded. This factor weighs against the government.
C.
This brings us to the third factor. That inquiry is whether Congress reasonably extended its well-established laws by applying sex-offender-registration requirements to someone long free from federal *239custody or supervision.18
1.
The government argues, and the panel majority held, that the statute is reasonably adapted to Congress’s military powers. For that proposition, they again rely on the analogy between sex-offender-registration requirements, on the one hand, and supervised release and probation, on the other: Because the latter are constitutional, the former must be too, or so the argument goes.
But that theory obscures two crucial distinctions: First, as we have mentioned, SORNA’s registration requirements, unlike probation and supervised release, are not a means to punish a sex offender for committing his crime19 but instead are merely civil regulations.20 Indeed, they cannot serve any punitive purpose in the case of Kebodeaux, because SORNA was enacted long after he committed his crime. If SORNA’s registration requirements were — like probation and supervised release — criminal punishments, they would violate the Ex Post Facto Clause.21 But because SORNA’s registration requirements are civil and were enacted after Kebodeaux committed his crime, the government cannot justify their constitutionality on the ground that they merely punish Kebodeaux for the crime he committed while in the military.22
Secondly, unlike SORNA’s registration requirements, probation and supervised release are conditions of release from (or instead of) custody.23 Like the civil con*240finement statute at issue in Comstock, they are thus “reasonably adapted ... to Congress’ power to act as a responsible federal custodian” of its prisoners, because they “avert the public danger likely to ensue from the release of ... detainees.” Com-stock, 130 S.Ct. at 1961 (internal quotation marks and citations omitted). By contrast, although § 2250(a) is surely meant to “avert ... public danger,” it is not, at least in eases such as Kebodeaux’s, from “the release of ... detainees,”24 because it applies even to those who have long severed all ties with the criminal justice system. It therefore makes no sense to say that SORNA’s registration requirements are — like probation, supervised release, or the civil commitment of mentally ill prisoners — “reasonably adapted” to the government’s role as “custodian ... of its prison system.”25
The tenuousness of the government’s position can be shown just by listing the chain of causation from Congress’s military power to its criminalization of Kebodeaux’s failure to register a change of address: Congress can supervise military personnel, so it can establish crimes for them, so it can prosecute and convict them, so it can supervise them for the duration of their sentence and while they are in federal custody, so it can pass a law to protect society from someone who was once in prison but seven years ago had fully served his sentence and has not since been in contact with the federal government. That last power is not reasonably adapted to Congress’s ability to regulate the military.
2.
The government, like the panel majority, responds by seizing on language in Com-stock that says that the power to imprison violators of federal law includes “the additional power to regulate the prisoners’ behavior even after their release.” Id. at 1964 (emphasis added). But the government and the majority quote the Court too selectively by omitting the beginning of the sentence. What Comstock actually says is, “Indeed even the dissent acknowledges that Congress has ... the additional power to regulate the prisoners’ behavior even after their release.” Id. The Court was merely enumerating those government actions that even the Comstock dissent conceded were constitutional.26 And the portions of the dissent cited by the *241majority assert only that Congress has the power to regulate a prisoner’s behavior post-release as part of his sentence; the dissent specifically rejects the notion that the government has open-ended authority to regulate him after his punishment has ended merely by virtue of some sort of vague “special relationship” between the federal government and one who once committed a federal crime.27
The Comstock majority distanced itself from the notion that the panel majority endorsed here. The Court cabined its holding by noting that the Solicitor General had conceded that the government could not commit a person who had already been released from federal custody or sent to state custody;28 only if he was still in federal custody could the government commit him.29 But if the power to regulate a person stems merely from the fact that he was once convicted of a federal crime, then whether he is presently in federal prison or subject to federal supervision would make no difference: Once he has been convicted of a federal crime, the government’s authority over him to protect society would continue as long as he lives.
Thus, in the instant case the government is reneging on precisely those concessions that caused the Court to reason that the civil commitment statute at issue in Com-stock was “narrowly tailored ... [to] pursuing the Government’s legitimate interest as a federal custodian in the responsible administration of its prison system.” Id. at 1965. And the panel majority endorsed the government’s about-face.
3.
The other case on which the panel majority relied is Carr, which it cited for the startling proposition that § 2250(a)(2)(A) is constitutional because the federal government has a “direct supervisory interest” over anyone who once committed a federal sex offense. It is true that Carr stated, 130 S.Ct. at 2239, that “the Federal Government has a direct supervisory interest” over federal sex offenders. But, as the panel majority acknowledged, Carr did not address the extent of Congress’s Article I power at all — it involved a statutory-interpretation issue and an Ex-Post-FactoClause question that the Court avoided.30 *242Moreover, the briefs in that ease show that no one — neither parties nor amici curiae — raised the argument that Kebodeaux brings here.31 Thus, the panel took an isolated statement from Carr out of context to make it a constitutional principle with far-reaching implications about the scope of federal power.
The panel majority was correct that § 2250(a)(2)(A) applies to individuals over whom the federal government has a “direct supervisory interest” because they are in custody or have been released from custody on the condition that they comply with SORNA.32 But that section also applies, as relevant here, to those who have long been free of federal custody and supervision after fully serving their sentences. To say that Congress continues to have a “direct supervisory interest” over such persons — like Kebodeaux — is to announce that it has an eternal supervisory interest over anyone who ever committed a federal sex crime. And that is no different from saying that Congress has such an interest over anyone who ever committed any federal crime, because there is nothing that is constitutionally special about sex crimes.33
4.
In sum, as applied to Kebodeaux, SORNA’s registration requirements are not, and cannot be, an attempt to punish the initial crime or to act as a responsible custodian of prisoners; they are merely an effort to protect the public from those who may be dangerous because they once were convicted of a sex offense. By that logic, Congress would have never-ending jurisdiction to regulate anyone who was ever convicted of a federal crime of any sort, no matter how long ago he served his sen*243tence, because he may pose a risk of re-offending.
Indeed, that logic could easily be extended beyond federal crimes: Congress could regulate a person who once engaged in interstate commerce (and was thereby subject to federal jurisdiction) on the ground that he now poses a risk of engaging in interstate commerce again. In short, the only “rational relation” between § 2250(a)(2)(A)’s application to Kebodeaux and an enumerated federal power is that Kebodeaux was once subject to federal jurisdiction — reasoning that is so expansive that it would put an end to meaningful limits on federal power. The third Com-stock “consideration” thus favors Kebodeaux.
D.
The fourth “consideration” is whether “the statute properly accounts for state interests.” Comstock, 130 S.Ct. at 1962. “[T]he ‘States possess primary authority for defining and enforcing the criminal law.’ ” Lopez, 514 U.S. at 561 n. 3, 115 S.Ct. 1624 (quoting Brecht v. Abrahamson, 507 U.S. 619, 635, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993)). Thus, “[w]hen Congress criminalizes conduct already denounced as criminal by the States, it effects a ‘change in the sensitive relation between federal and state criminal jurisdiction.’ ” Id. (quoting United States v. Enmons, 410 U.S. 396, 411-12, 93 S.Ct. 1007, 35 L.Ed.2d 379 (1973)). Alternatively, it “displace^] state policy choices ... [when] its prohibitions apply even in States that have chosen not to outlaw the conduct in question.” Id. (citation omitted).
As the government points out, some aspects of SORNA do accommodate state interests. A state forgoes only ten percent of its federal funding by failing substantially to comply with SORNA (for example, by failing to maintain a registry). See 42 U.S.C. § 16925(a). And § 2250 itself allows an affirmative defense if “uncontrollable circumstances” — which, according to the government, would include a state’s failure to collect registration data— prevent an individual from complying with its registration requirements. 18 U.S.C. § 2250(b). Indeed, as the panel pointed out, this court recently upheld SORNA against a Tenth-Amendment challenge on the ground that the statute does not require the states to comply with it. United States v. Johnson, 632 F.3d 912, 920 (5th Cir.), cert. denied, — U.S. -, 132 S.Ct. 135, 181 L.Ed.2d 55 (2011).
Nevertheless, the degree of state accommodation with respect to § 2250(a)(2)(A) is substantially less than that present in Comstock, 130 S.Ct. at 1962-63, in which the Court found that Congress’s statutory scheme for civilly confining mentally ill and sexually dangerous prisoners accommodated state interests because the Attorney General was required to notify interested states about the confinement and to release prisoners if a state wished to assert authority over them. Thus, continued federal confinement was, in essence, continually subject to the states’ veto.
Here, by contrast, there is no provision by which someone federally prosecuted under SORNA can be subjected to state penalties or transferred to state custody instead. Unless a former federal sex offender proves that a state has made it impossible for him to register,34 he is subject to federal prosecution and up to ten years of imprisonment for failing to update his state registration within three days of a change of address, employment, name, or student status, even if the state believes a *244more moderate response would be appropriate35 (which Texas and many other states apparently do36). The state is thus forced into the binary choice of keeping a former federal sex offender off its own registry entirely or subjecting him to § 2250(a)(2)(A)’s harsh penalties; it cannot control the punishment given to those who fail to update their registration.
Thus, because SORNA mandates federal penalties for the failure of a state resident to update his state sex offender registration solely because of an intrastate change of address without giving states a veto of the sort present in Comstock, it is a much more substantial imposition on the states’ traditional police-power authority over the criminal law within their own borders than what was at issue in Comstock. It is true that § 2250(a)(2)(A) applies only to federal sex offenders; but, as we have discussed, in the case of persons such as Kebodeaux those are individuals with whom the federal government had previously severed all ties. Accordingly, the fourth Comstock “consideration” ultimately cuts in Kebodeaux’s favor.
E.
The final factor is whether the “links between [the statute] and an enumerated Article I power are not too attenuated” and the “statutory provision [is not] too sweeping in its scope.” Comstock, 130 S.Ct. at 1963. The panel majority’s position was that the statute is narrow because it applies only to sex offenders. But even assuming that a statute that applies to all sex offenders were considered narrow, its logic is expansive, because the only jurisdictional basis for § 2250(a)(2)(A) is the fact that a person once committed a federal sex crime. That reasoning opens the door, as discussed in part II.C, to congressional power over anyone who was ever convicted of a federal crime of any sort. That is anything but narrow. Accordingly, the fifth Comstock factor also cuts in Kebodeaux’s favor.
F.
In summary, even taking into account “the breadth of the Necessary and Proper Clause,” Comstock, 130 S.Ct. at 1965, SORNA’s registration requirements and criminal penalty for failure to register as a sex offender, as applied to those, like Kebodeaux, who had already been unconditionally released from federal custody or supervision at the time Congress sought to regulate them, are not “rationally related” or “reasonably adapted” to Congress’s power to criminalize federal sex offenses to *245begin with. The statute’s regulation of an individual, after he has served his sentence and is no longer subject to federal custody or supervision, solely because he once committed a federal crime, (1) is novel and unprecedented despite over 200 years of federal criminal law, (2) is not “reasonably adapted” to the government’s custodial interest in its prisoners or its interest in punishing federal criminals, (3) is unprotective of states’ sovereign interest over what intrastate conduct to criminalize within their own borders, and (4) is sweeping in the scope of its reasoning. For those reasons, and with high respect for its careful reasoning, the panel majority wrongly decided this case.37
III.
Finally, the government, like the panel concurrence, offers an alternative argument for upholding the statute: that SORNA’s registration requirements for federal sex offenders, and the criminal penalties for failing to comply, are necessary and proper to effect Congress’s Commerce Clause power. Under its Commerce-Clause and Necessary-and-Proper-Clause authority, Congress may (1) “regulate the use of the channels of interstate commerce,” (2) “regulate and protect the instrumentalities of ... or persons or things in interstate commerce, even though the threat may come only from intrastate activities,” and (3) “regulate those activities having a substantial relation to interstate commerce, ie., those activities that substantially affect interstate commerce.”38
The panel concurrence maintains that this case fits into the first two categories of Commerce Clause authority. According to that view, SORNA’s regulation of federal sex offenders can be seen as necessary and proper regulation of “the channels of’ or “persons ... in interstate commerce” because it reduces the risk of unmonitored interstate travel by sex offenders. The argument in the concurrence runs as follows: Because a federal sex offender would face no federal sanction for failing to register until he travels interstate, he could hide from authorities before he does so. Thus, to prevent the purported risk that he evades detection before traveling interstate, no requirement of interstate travel ought to be necessary; Congress should be able to criminalize the mere act of failing to register, even if a sex offender never travels interstate, because it reduces the risk that he will someday travel interstate undetected.
Thus, the concurring judge on the panel would subtly but significantly expand Congress’s power under the first two catego*246ries of Commerce Clause authority beyond the regulation of “the use of the channels of interstate commerce” or “persons or things in interstate commerce,” Lopez, 514 U.S. at 558, 115 S.Ct. 1624 (emphasis added), to the regulation of the possible use of the channels of interstate commerce and persons or things because they will potentially be in interstate commerce. With due respect for the concurrence’s well-stated position, its contention is both contrary to precedent and so expansive that it would confer on the federal government plenary power to regulate all criminal activity — precisely what the Court sought to avoid in Lopez and Morrison.
A.
1.
Under the first category of its Commerce Clause authority, Congress may regulate the use of the channels of interstate commerce: “the use of the interstate transportation routes through which persons and goods move.” Morrison, 529 U.S. at 613 n. 5, 120 S.Ct. 1740 (internal quotation marks omitted). “Congress may impose relevant conditions and requirements on those who use the channels of interstate commerce in order that those channels will not become the means of promoting or spreading evil.... ”39 Because the federal government “exercis[es] [a] police power ... within the field of interstate commerce,” Brooks, 267 U.S. at 436-37, 45 S.Ct. 345, i.e., with respect to the channels, instrumentalities, persons, and goods involved in interstate commerce, Congress may regulate those who use the channels of interstate commerce even if their activity is non-economic in nature. Thus, for example, Congress may prohibit “enticing a woman from one state to another for immoral ends, whether for commercial purposes or otherwise,” id. at 437, 45 S.Ct. 345, transporting kidnaped persons across state lines, United States v. Darby, 312 U.S. 100, 113, 61 S.Ct. 451, 85 L.Ed. 609 (1941), traveling across state lines to commit domestic violence, United States v. Lankford, 196 F.3d 563, 572 (5th Cir.1999), or traveling interstate as a state sex offender without having first registered as such.40
But just as this category of Commerce-Clause authority gives the federal government a “police power” over those who use the channels of interstate commerce, even if their activity is non-commercial, Brooks, 267 U.S. at 437, 45 S.Ct. 345, the corollary is that that police power must also be limited to the “field of interstate commerce,” see id. at 436, 45 S.Ct. 345. For example,, although Congress may regulate those who use the channels of interstate commerce for any reason, “[t]he regulation ... of intrastate violence that is not directed at the instrumentalities, channels, or goods involved in interstate commerce has always been the province of the States.” Morrison, 529 U.S. at 618, 120 S.Ct. 1740 (emphasis added).
In Whaley, 577 F.3d at 259-60, in which this court upheld SORNA’s requirement that state sex offenders register their address — as distinguished from the federal sex-offender-registration requirement at *247issue here — -we were careful to limit our holding by explaining that the statute at issue there neither targets nor sanctions anyone who did not in fact use the channels of interstate commerce. We explained that, with respect to state sex offenders, SORNA punishes a person only if he travels interstate without having first registered or updated his registration. Id. at 261. Thus, the registration requirement’s “focus” with regard to state offenders is solely on enforcing the criminal prohibition on traveling interstate without having registered — “rather than on requiring sex offender registration generally.” Id. at 259.41
2.
As the Court explained in Carr, 130 S.Ct. at 2238, however, Congress “chose to handle federal and state sex offenders differently.”42 In contrast to SORNA’s regulatory scheme with regard to state sex offenders, Congress, for federal offenders, “requir[es] sex offender registration generally.” Whaley, 577 F.3d at 259. The statutes regulating the movement of all federal sex offenders, 42 U.S.C. § 16913 and 18 U.S.C. § 2250(a)(2)(A), apply to all intrastate as well as interstate movement without regard to whether a sex offender ever uses the channels of interstate commerce. Those statutes therefore do not regulate only activity “directed” at the channels of interstate commerce. Morrison, 529 U.S. at 617, 120 S.Ct. 1740. Federal sex offenders are subject to criminal sanctions if they fail to register or update their registration even if they never step foot outside their state. In short, federal sex offenders are regulated merely by virtue of the fact that they are federal sex offenders. The view expressed in the panel concurrence would thus do away with precisely the limits we considered crucial to our holding in Whaley.
Indeed, notably, the Solicitor General has expressly denied that § 2250(a)(2)(A) is constitutional as a regulation of the channels of interstate commerce, asserting instead that it applies because the federal government has a “direct supervisory interest” over those who committed federal offenses, see Carr, 130 S.Ct. at 2238-39, irrespective of whether they have a connection to interstate commerce.43 Here the government makes an about-face only now that its original justification for the statute’s constitutionality — that of the panel majority — is in question in light of the fact that the panel opinion has been vacated for rehearing en banc..
3.
The panel concurrence nevertheless urges that SORNA’s registration scheme *248for federal sex offenders is constitutional as well, because it allows the federal government better to monitor sex offenders in case they someday travel interstate. The concurrence therefore would expand the federal police power over individuals who “use ... the channels of interstate commerce,” see Lopez, 514 U.S. at 558, 115 S.Ct. 1624; Brooks, 267 U.S. at 437, 45 S.Ct. 345, to those who might someday do so.
Neither this court nor the Supreme Court, however, has ever extended Congress’s “police power” over those who use the channels of interstate commerce to punish those who are not presently using them but might do so. The theory expressed in the panel concurrence is unprecedented,44 and for good reason: Because every person is mobile, anyone might someday travel interstate. Thus, by the reasoning of the concurrence, the federal government could regulate anyone on that ground who might someday travel interstate. Myriad, longstanding federal statutes, both economic and non-economic, that have as a jurisdictional nexus the movement of a person across state lines would suddenly no longer need that nexus.45
For example, it is a federal crime to travel across state lines to evade child-support obligations. 18 U.S.C. § 228(a)(2). As with former federal sex offenders, deadbeat parents might move around within a state to evade state authorities, and as with former federal sex offenders, that might increase the risk that they go undetected before they travel across state lines. Therefore, by the logic of the panel concurrence, the federal government should be able to regulate the intrastate movement of deadbeat parents as well.
Thus, Congress could require anyone who owes child support obligations under state law to report their changes of address to the federal government, and if they do not, the Attorney General could criminally prosecute them; the government would no longer need to wait until deadbeat parents cross state lines: The crime would be complete when they move *249intrastate without notifying federal authorities, because of the likelihood that they might otherwise someday cross state lines undetected. The federal government could, as here, use the mere risk of travel across state lines to justify far-reaching intrastate regulation in an area of traditional and exclusive state concern.
Indeed, there is nothing about the panel concurrence’s reasoning that limits its application to reporting requirements and criminal punishments for failing to comply with them. For example, it is a federal crime to transport a kidnaped person across state lines. 18 U.S.C. § 1201(a)(1). As with former federal sex offenders, someone who is transporting a kidnaped person is capable of moving around and thereby potentially evading state authorities. And as with former federal sex offenders, were the federal government to have no jurisdiction over kidnappers until they cross state lines, the likelihood that they would evade authorities before traveling interstate would be greater. Thus, according to the concurrence, the federal government should have the power to criminalize the intrastate transportation of kidnaped persons, just as it should have the power to proscribe the intrastate movement of sex offenders who did not register, because, in both cases, it would reduce the risk that the criminals evade detection before crossing state lines.
More generally still, every crime (indeed every act) brings with it the risk that the perpetrator will flee across state lines before being detected. Although the panel concurrence is stated in the context of former sex offenders, there is nothing limiting its logic to past, rather than present, criminals. Accepting his logic — that the mere risk that a dangerous person will cross state lines undetected gives the federal government authority to police his intrastate movements preemptively— would mean that the federal government would have the power to arrest someone who committed a murder, rape, or any other crime traditionally subject to state authority on the ground that he might otherwise evade state authorities and escape across state lines undetected after doing so. In short, the concurrence offers no limiting principle that would allow the federal government to track and arrest former sex offenders because they might someday travel interstate, but not allow it to do the same to anyone else for that same reason.
4.
The basic flaw in the panel concurrence is that it overlooks the role of the states in policing within their own borders, relying on the implicit premise that the federal government must regulate sex offenders’ intrastate movements because the states will not do so. Every state has its own sex offender registry and has every incentive to track and arrest sex offenders as long as they remain intrastate. For example, it was state, not federal, authorities — specifically, El Paso Police Department officers — who both registered Kebodeaux and discovered that he had failed to update his registration. Indeed, the federal sex-offender registry consists of nothing more than the amalgamation of state registry (along with tribal and territorial registry) data obtained from local officials.46
*250Only if a sex offender travels out-of-state — i.e., uses the channels of interstate commerce — does a state’s jurisdiction end, making it inadequate to the task of tracking and arresting a sex offender — and the federal government’s role there begins. To give, instead, to the federal government the overlapping power to do exactly what a state could already do itself, in an area completely unrelated to commerce, just because criminals, like all human beings, can potentially cross state lines, would violate basic tenets of federalism.47 In effect, the panel concurrence asserts that the federal government should be able to police individuals within state borders just because states might not do so and those individuals might thus pose a risk to inhabitants of other states. But the federal government’s jurisdiction does not expand or contract based on a state’s criminal-policy choices.48
B.
The panel concurrence fares no better under the second category of Congress’s Commerce-Clause authority: Congress may regulate the instrumentalities of, and, as most relevant here, persons or things in, interstate commerce, as well as intrastate activities threatening them. Lopez, 514 U.S. at 558, 115 S.Ct. 1624. For example, the Court has upheld the regulation of vehicles used in interstate commerce,49 the destruction of aircraft,50 and thefts from interstate shipments51 on those grounds.
The panel concurrence took this category of authority to mean that Congress may police any person or thing that might cross state lines. That misunderstands the precedent. First, crossing state lines does not mean a person is engaging “in interstate commerce,” because that mere fact does not constitute engaging in “commerce” by any definition of the term. Rather, it constitutes a “use of the channels of interstate commerce,” which the first category of Commerce-Clause authority is meant to regulate. See part III.A. With all due respect, the concurrence thus confuses the first category of regulable activity with the second.
Second, a person who only might cross state lines is not engaging “in interstate commerce,” because he has not yet engaged in interstate activity. Thus, SORNA’s sex-offender-registration requirements do not regulate persons in interstate commerce, because sex offenders do not engage in activity that is either “interstate” or “commerce” just by virtue of being sex offenders. That a person might someday engage in inter*251state commerce is very different from saying that he is a “person[ ] ... in interstate commerce.” Lopez, 514 U.S. at 558, 115 S.Ct. 1624. Under this category of authority, Congress may regulate and protect the latter, not the former. See id.
Lastly, though Congress may protect the instrumentalities of, and persons or things in, interstate commerce from intrastate threats, those threats must be “directed at” the instrumentalities of, or perr sons or things in, interstate commerce; they cannot just be a general threat to society of the sort that sex offenders pose.52 For example, Congress may regulate the destruction of an “aircraft used, operated, or employed in interstate, overseas, or foreign air commerce,” 18 U.S.C. § 32(a)(1), even though the destructive activity occurs within a single state,- because aircraft are themselves “instrumentalities of interstate commerce,” Perez, 402 U.S. at 150, 91 S.Ct. 1357. Analogously, Congress may regulate thefts from interstate shipments, even though the thefts occur within a single state, because the shipments themselves are “things in [interstate] commerce.” Id. (citing 18 U.S.C. § 659). Those regulations are permissible because Congress limited itself to regulating threats “directed at” interstate commerce. See Morrison, 529 U.S. at 618, 120 S.Ct. 1740.
In short, none of the Court’s cases under the second Commerce Clause category even hints, let alone turns on the fact, that Congress could regulate someone because he might someday threaten interstate commerce. And for good reason: By that flawed logic, Congress could regulate ordinary thieves on the ground that they pose a “threat” to interstate commerce by virtue of the fact that, someday, they might steal an instrumentality of interstate commerce. Accordingly, the panel concurrence’s reliance on the second Commerce Clause category is unpersuasive.
C.
Indeed, it is telling that the panel concurrence’s main source of authority is Gonzales v. Raich, 545 U.S. 1, 125 S.Ct. 2195, 162 L.Ed.2d 1 (2005), which held that a congressional statute prohibiting marihuana possession was constitutional under the third category of Commerce-Clause authority, Congress’s “power to regulate activities that substantially affect interstate commerce,” id. at 17, 125 S.Ct. 2195. Indeed, the Court stated that “[o]nly the third category” of Congress’s Commerce-Clause authority was “implicated in the case at hand.” Id. It logically follows that the Court believed that the case did not “implicate” the two other “categories” of Commerce-Clause power— those at issue here: Congress’s powers to “regulate the channels of interstate commerce” and to “regulate and protect ... persons or things in interstate commerce.” See id. at 16-17, 125 S.Ct. 2195. That is unsurprising, given that the statute at issue criminalized purely intrastate marihuana possession, which is not a part of “the channels of’ or a “thing[] in interstate commerce” or a “threat” to “things in interstate commerce.”
Moreover, in holding that the marihuana-possession statute was constitutional under the third Commerce-Clause category, the Raich Court explicitly based its decision on the fact that the statute was part of a comprehensive regulation of *252“quintessentially economic” activity.53 That the statute regulated economic activity was what distinguished the case from Lopez and Morrison, which struck down statutes regulating intrastate conduct because of the “noneconomic, criminal nature of the conduct at issue.”54 Raich thus merely followed the line drawn in Lopez and Morrison between economic and non-economic activity under the third category.
In contrast to the statute in Raich, and like the statutes in Lopez and Morrison, the statute here regulates non-economic, intrastate conduct that is not “an essential part of a larger regulation of economic activity.” Lopez, 514 U.S. at 561, 115 S.Ct. 1624. It is a criminal statute that “by its terms has nothing to do with ‘commerce’ or any sort of economic enterprise, however broadly one might define those terms.” Morrison, 529 U.S. at 610, 120 S.Ct. 1740 (quoting Lopez, 514 U.S. at 561, 115 S.Ct. 1624). It would thus fail the Lopez/Morrison/Raich test under the third Commerce Clause category, as it should. To hold a non-commercial statute regulating purely intrastate conduct constitutional would read the word “commerce” out of the Commerce Clause.55
But by the logic urged in the panel concurrence, Raich should not have turned on the economic/non-economic distinction or on the third category of Commerce Clause authority at all. Because marihuana possessed intrastate surely poses a risk of subsequently moving interstate, the Court instead should have found the statute constitutional as a regulation of “the channels of’ or “things in interstate commerce” without any need to resort to the catchall category of intrastate “activities that substantially affect interstate commerce.” But that was not what the Court did or said in Raich.
The panel concurrence’s reliance on the first two “categories” of Congress’s Commerce-Clause authority instead of the third amounts to an avoidance of Lopez, Morrison, and Raich. That reasoning, far from faithfully applying Raich, expands the first two “categories” to cover non-economic, intrastate activities that could not be regulated under the third. The fatal flaw with that argument is that it fails to come to terms with the role of the economic/non-economic distinction in the Court’s Commerce-Clause jurisprudence: To be constitutional, regulations of intrastate activity affecting interstate commerce must, logically, have something to do with commerce. The statute at issue here does not.
D.
Finally, the panel concurrence contends that § 2250(a)(2)(A), although a regulation of mimstate activity, is constitutional as a necessary and proper means of enforcing § 2250(a)(2)(B)’s regulation of mierstate *253travel under Raich.56 But it is questionable how subsection (A), which criminalizes federal sex offenders’ failure to update registration, helps effect subsection (B), which criminalizes state sex offenders’ failure to update. Subsection (B) makes it a crime for a state sex offender to fail to update his registration if he travels in interstate commerce without having registered. Subsection (A) mirrors subsection (B) for federal sex offenders, except that there is no interstate-travel requirement. Not having an interstate travel requirement for federal sex offenders in no way helps to protect society from the interstate travel of state sex offenders.
E.
Therefore, as we have explained, the approach reflected in the panel concurrence fails, because it is an attempt to place under the Commerce Clause a regulation that is neither “interstate” nor “commercial.” SORNA’s regulation of federal sex offenders does not fit into any of the three categories of regulations that the Supreme Court has upheld under the Commerce Clause, so it cannot be justified under the commerce power.
Upholding § 2250(a)(2)(A) would go a big step further than has the applicable caselaw, because, unlike § 2250(a)(2)(B), this statute regulates federal sex offenders “generally,” Whaley, 577 F.3d at 259, regardless of whether they engage in interstate activity.57 The activity criminalized by § 2250(a)(2)(A) is thus not “directed” at interstate commerce in the way that all previously upheld provisions regulating the use of the channels of interstate commerce have been.58
IV.
In summary, and for the reasons discussed in parts II and III, 42 U.S.C. § 16913’s registration requirements and § 2250(a)(2)(A)’s criminal penalties for failing to register after intrastate relocation are unconstitutional solely as they apply to former federal sex offenders who had been unconditionally released from federal custody before SORNA’s passage in 2006. Every federal sex offender subject to federal custody or supervision when SORNA was enacted, or who was convicted since then, is unaffected. Moreover, those who had been unconditionally released before SORNA’s passage need not go unmonitored; they could still be regulated just as state sex offenders currently are under federal law, and they remain subject to state authority.
The statute is an unlawful expansion of federal power at the expense of the tradi*254tional and well-recognized police power of the state.59 The conviction is REVERSED, and a judgment of dismissal is RENDERED.
. See 42 U.S.C. § 16913 (2006) (requiring a sex offender to register in each jurisdiction in which he resides and to update that registration); 28 C.F.R. § 72.3 (2007) (specifying that § 16913’s requirements apply to all sex offenders, "including sex offenders convicted of the offense for which registration is required prior to the enactment of [§ 16913]”). Because Kebodeaux committed his offense before SORNA’s passage, his duty to register comes from the Attorney General's regulation rather than the statute itself. Reynolds v. United States, - U.S. -, 132 S.Ct. 975, 984, 181 L.Ed.2d 935 (2012). Despite the fact that the Attorney General did not follow the procedures laid out in the Administrative Procedure Act when issuing the regulation, we found that to be harmless error as applied to a defendant who had moved interstate but was otherwise in substantially the same situation as is Kebodeaux. United States v. Johnson, 632 F.3d 912, 931-32 (5th Cir.), cert. denied, - U.S. -, 132 S.Ct. 135, 181 L.Ed.2d 55 (2011). Although the rule may be valid as applied to a sex offender who moves interstate, the portion of the statute that gives the Attorney General the authority to apply SORNA to pre-act offenders who move intrastate would not be valid if Congress does not have the power under Article I to apply the statute to pre-act sex offenders. Therefore, our analysis focuses on whether Congress had that authority that it attempted to grant to the Attorney General.
. Cf. 18 U.S.C. § 2250(a)(2)(B) (criminalizing state sex offenders' failure to register or update registration if they travel in interstate commerce).
. Thus, even with respect to past federal sex offenders such as Kebodeaux, Congress presumably could remedy the constitutional problem merely by adding an element of interstate travel to the crime of failing to register. Because it is not before us, however, we make no ruling on that speculative issue.
. In her well-written dissent, Judge Haynes disputes that the federal government unconditionally released Kebodeaux from its jurisdiction upon his release from custody. Citing the Wetterling Act of 1994, as amended by the Lychner Act of 1996, 42 U.S.C. §§ 14071-14073, repealed by SORNA Pub.L. No. 109— 248, § 129, 120 Stat. 587, 600 (2006), the dissent argues that Kebodeaux has been subject to federal registration ever since his 1999 conviction. But that notion overlooks a fundamental difference between SORNA and its predecessors.
Although SORNA directly imposes a registration requirement on covered sex offenders, see § 16913(a), pre-SORNA federal law merely conditioned federal funding on states' maintaining their own sex-offender registries that were compliant with federal guidelines, see § 14071(g) (2000) (repealed by SORNA). Only sex offenders residing in non-compliant states were subject to federal registration for intrastate changes in residence. See § 14072(g)(1) — (3), (i) (2000) (repealed by SORNA). •
Because his state of residence, Texas, was compliant with federal guidelines at the time of his offense, Kebodeaux was not subject to federal registration requirements. See Creekmore v. Attorney Gen. of Tex., 341 F.Supp.2d 648, 654 (E.D.Tex.2004) (observing that Texas enacted its registry in 1991 and amended it "four times: in 1993, 1995, 1997, and 1999 to ensure that the program met minimum federal requirements” (citations omitted)); Creekmore v. Attorney Gen. of Tex., 116 F.Supp.2d 767, 773 (E.D.Tex.2000) (noting that Texas's registration program was “federally-approved”); Wayne A. Logan, Sex Offender Registration and Community Notification: Past, Present, and Future, 34 New Eng. J. on Crim. & Civ. Confinement 3, 6 (2008) (observing that all fifty states and the District of Columbia had complied with the Wetterling Act by the end of 1996). Thus, before the passage of SORNA, Kebodeaux was subject only to state, not federal, registration obligations.
. 42 U.S.C. § 16913(a). In addition, "[f]or initial registration purposes only, a sex offender shall also register in the jurisdiction in which convicted if such jurisdiction is different from the jurisdiction of residence." Id. A registration must be updated within three days of any change. § 16913(c).
. See § 16913(d) ("The Attorney General shall have the authority to specify the applicability of the requirements of this subchapter to sex offenders convicted before the enactment of this chapter ... and to prescribe rules for the registration of any such sex offenders....”); 28 C.F.R. § 72.3 (specifying that § 16913's requirements apply to all sex offenders, "in-*236eluding sex offenders convicted of the offense for which registration is required prior to the enactment of that Act").
. Comstock, 130 S.Ct. at 1956-57 (citing Gonzales v. Raich, 545 U.S. 1, 22, 125 S.Ct. 2195, 162 L.Ed.2d 1 (2005); Sabri v. United States, 541 U.S. 600, 605, 124 S.Ct. 1941, 158 L.Ed.2d 891 (2004); United States v. Lopez, 514 U.S. 549, 557, 115 S.Ct. 1624, 131 L.Ed.2d 626 (1995); Hodel v. Va. Surface Mining & Reclamation Ass’n, Inc., 452 U.S. 264, 276, 101 S.Ct. 2352, 69 L.Ed.2d 1 (1981)).
. Comstock, 130 S.Ct. at 1957 (quoting Raich, 545 U.S. at 37, 125 S.Ct. 2195) (Scalia, J„ concurring in the judgment); United States v. Darby, 312 U.S. 100, 121, 61 S.Ct. 451, 85 L.Ed. 609 (1941).
. Although the panel majority was correct that there is a presumption of constitutionality, it is troubling that it engaged in an extended discussion of all the different constitutional challenges against which SORNA has been upheld, as though those instances somehow make it more likely that Kebodeaux’s constitutional challenge fails. That courts have upheld the five-year-old statute against an ex post facto challenge, a due process challenge, a non-delegation challenge, and a Commerce Clause challenge to a clause that explicitly is limited to persons traveling in interstate commerce does not suggest that we must uphold this SORNA provision against this challenge.
. See Comstock, 130 S.Ct. at 1966 (Kennedy, J., concurring in the judgment) ("This Court has not held that the [Williamson v. Lee Optical of Okla., Inc., 348 U.S. 483, 487-88, 75 S.Ct. 461, 99 L.Ed. 563 (1955),] test, asking if 'it might be thought that the particular legislative measure was a rational way to correct’ an evil, is the proper test in this context.... Indeed, the cases the Court cites in the portion of its opinion referring to 'rational basis’ are predominantly Commerce Clause cases, and none are due process cases.”).
. See id. (Kennedy, J., concurring in the judgment) (explaining that rational-basis scrutiny under the Due Process Clause requires asking whether " ‘it might be thought that the particular legislative measure was a rational way to correct’ an evil” (quoting Lee Optical, 348 U.S. at 487-88, 75 S.Ct. 461)).
. See id. at 1967 (Kennedy, J., concurring in the judgment) ("[The Court’s Commerce Clause] precedents require a tangible link to commerce, not a mere conceivable rational relation, as in Lee Optical.”). For example, in Morrison the Court struck down a civil remedy for violence against women under the Commerce Clause despite copious evidence that such violence had a substantial effect on (and thus was conceivably rationally related to) interstate commerce. See Morrison, 529 U.S. at 615, 120 S.Ct. 1740 (finding statute unconstitutional because, "[i]f accepted, petitioners’ reasoning would allow Congress to regulate any crime as long as the nationwide, aggregated impact of that crime has substantial effects on employment, production, transit, or consumption”); id. at 628-29, 120 S.Ct. 1740 (Souter, J., dissenting) (discussing the “mountain of data assembled by Congress ... showing the effects of violence against women on interstate commerce”). So, plainly, more is required.
. Comstock, 130 S.Ct. at 1965. For example, the Comstock Court also relied on the fact that the statute was "narrowly tailored” or "narrow [in] scope,” id., an analysis that is not necessary to uphold a law under rational-basis scrutiny under the Due Process or Equal Protection Clause, see, e.g., Lee Optical, 348 U.S. at 487-88, 75 S.Ct. 461.
. Cf. Walz, 397 U.S. at 678, 90 S.Ct. 1409 (" 'If a thing has been practised for two hundred years by common consent, it will need a strong case for the Fourteenth Amendment to affect it....'” (quoting Jackman v. Rosenbaum Co., 260 U.S. 22, 31, 43 S.Ct. 9, 67 L.Ed. 107 (1922))).
. Va. Office for Prot. & Advocacy v. Stewart, - U.S. -, 131 S.Ct. 1632, 1641, 179 L.Ed.2d 675 (2011) ("Respondents rightly observe that federal courts have not often encountered lawsuits brought by state agencies against other state officials. That does give us pause. Lack of historical precedent can indicate a constitutional infirmity ...” (citing Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., - U.S. -, 130 S.Ct. 3138, 3159, 177 L.Ed.2d 706 (2010))); Free Enter. Fund, 130 S.Ct. at 3159 ("Perhaps the most telling indication of the severe constitutional problem with the PCAOB is the lack of historical precedent for this entity” (quoting Free Enter. Fund v. Pub. Co. Accounting Oversight Bd., 537 F.3d 667, 699 (D.C.Cir.2008) (Kavanaugh, J., dissenting))).
. See Carr v. United States, - U.S. -, 130 S.Ct. 2229, 2232, 176 L.Ed.2d 1152 (2010); Richard G. Wright, Sex Offender Post-Incarceration Sanctions: Are There Any Limits?, 34 New Eng. J. on Crim. & Civ. Confinement 17, 29-36 (2008) (discussing history of federal sex-offender-registration laws).
. Every circuit, including ours, has held that, unlike probation or supervised release, SORNA’s registration requirements are civil regulations whose purpose is not to punish for crimes. See United States v. Young, 585 F.3d 199, 204 (5th Cir.2009) (per curiam); cf. Smith v. Doe, 538 U.S. 84, 101-02, 123 S.Ct. 1140, 155 L.Ed.2d 164 (2003) (upholding Alaska's sex-offender-registration statute against ex post facto challenge and distinguishing it from probation and supervised release because it is not a punishment).
. See Comstock, 130 S.Ct. at 1961 (explaining that the third factor is that "Congress reasonably extended its longstanding civil-commitment system to cover mentally ill and sexually dangerous persons who are already in federal custody, even if doing so detains them beyond the termination of their criminal sentence").
. See id. at 1979 n. 12 (Thomas, J., dissenting) (referring to supervised release as a "form of punishment”); United States v. Knights, 534 U.S. 112, 119, 122 S.Ct. 587, 151 L.Ed.2d 497 (2001) ("Probation, like incarceration, is a ‘form of criminal sanction. ..” (quoting Griffin v. Wisconsin, 483 U.S. 868, 874, 107 S.Ct. 3164, 97 L.Ed.2d 709 (1987))).
. See Young, 585 F.3d at 204; see also Smith, 538 U.S. at 101, 123 S.Ct. 1140 (explaining why Alaska's sex-offender-registration requirements are not, like probation and supervised release, forms of punishment).
. Young, 585 F.3d at 204; see also United States v. Caulfield, 634 F.3d 281, 283 (5th Cir.2011) ("The. heart of the Ex Post Facto Clause bars application of a law that changes the punishment, and inflicts a greater punishment, than the law annexed to the crime, when committed.” (quoting Johnson v. United States, 529 U.S. 694, 699, 120 S.Ct. 1795, 146 L.Ed.2d 111 (2000))).
. The panel majority inaccurately asserted that Kebodeaux conflates his Article I argument with an Ex-Post-Facto-Clause argument. In fact, his Article I contention works only because § 2250(a)(2)(A) is not an ex post facto criminal punishment. Because SORNA's registration requirements are not criminal punishments, but a civil regulatory scheme, they do not pose an ex post facto problem. But for that very reason — that SORNA registration is a civil regulatory scheme and not a punishment imposed on Kebodeaux for his federal crime — Congress needs some other jurisdictional hook to apply the requirement to persons such as him.
. Compare 18 U.S.C. § 2250(a)(2)(A) (criminalizing the failure to register or update registration as a sex offender regardless of the date of the crime) and 28 C.F.R. § 72.3 (specifying that "[t]he requirements of [SORNA] apply to all sex offenders, including sex offenders convicted of the offense for which registration is required prior to the enactment of that Act”) with 18 U.S.C. § 3603(1) (tying the duties of the probation officer to "the conditions specified by the sentencing court”), § 3601 (same), § 3563(a) (explaining the "condition[s] of a sentence of probation”), § 3583(d) (same for supervised release), and United States v. John*240son, 529 U.S. 53, 56, 120 S.Ct. 1114, 146 L.Ed.2d 39 (2000) ("A prisoner whose sentence includes a term of supervised release after imprisonment shall be released by the Bureau of Prisons to the supervision of a probation officer who shall, during the term imposed, supervise the person released to the degree warranted by the conditions specified by the sentencing court.” (quoting 18 U.S.C. § 3624(e))).
. Comstock, 130 S.Ct. at 1961 (citation omitted) (emphasis added).
. Id. at 1965 (emphasis added) (holding that "§ 4248 is a reasonably adapted and narrowly tailored means of pursuing the Government's legitimate interest as a federal custodian in the responsible administration of its prison system”).
.See id. at 1964 ("Indeed even the dissent acknowledges that Congress has the implied power to criminalize any conduct that might interfere with the exercise of an enumerated power, and also the additional power to imprison people who violate those (inferentially authorized) laws, and the additional power to provide for the safe and reasonable management of those prisons, and the additional power to regulate the prisoners’ behavior even after their release” (citing id. at 1976-77, 1978 n. 11 (Thomas, J., dissenting))). The majority opinion cites op. p. 1979, n. 11 of the dissent, but it must have meant note 12, because note 11 does not appear on page 17 (although note 12 does), and note 11 has nothing to do with regulation after release (e.g. in the form of supervised release), whereas that is precisely what is discussed in note 12.
.See id. at 1979 n. 12 (Thomas, J., dissenting) ("Contrary to the Government's suggestion, federal authority to exercise control over individuals serving terms of ‘supervised release' does not derive from the Government's 'relationship' with the prisoner, ... but from the original criminal sentence itself.” (citations omitted) (emphasis added)); id. at 1976— 77 (Thomas, J., dissenting) (concluding that “[fjederal laws that criminalize conduct that interferes with enumerated powers, establish prisons for those who engage in that conduct, and set rules for the care and treatment of prisoners awaiting trial or serving a criminal sentence ” are constitutional (emphasis added)); id. at 1979 (Thomas, J., dissenting) ("Once the Federal Government’s criminal jurisdiction over a prisoner ends, so does any 'special relationship’ between the government and the former prisoner.” (alteration omitted)).
. See id. at 1963 ("|T]he Solicitor General acknowledges that 'the Federal Government would have no appropriate role' with respect to an individual covered by the statute once ‘the transfer to State responsibility and State control has occurred.’ ” (citation omitted)); id. at 1965 (noting that the Solicitor General conceded that "the Federal Government would not have ... the power to commit a person who ... has been released from prison and whose period of supervised release is also completed”).
. See id. at 1964-65 (quoting the Solicitor General for the proposition that “[federal authority for § 4248] has always depended on the fact of Federal custody, on the fact that this person has entered the criminal justice system ...").
. See Carr, 130 S.Ct. at 2232-33 ("At issue in this case is whether § 2250 applies to sex *242offenders whose interstate travel occurred prior to SORNA's effective date and, if so, whether the statute runs afoul of the Constitu1 tion's prohibition on ex post facto laws.'').
. A Commerce Clause argument related to applying the statute to pre-SORNA travel (i.e., not the issue Kebodeaux raises) was made by amicus but not addressed by the Court in light of its holding. See id. at 2248 (Alito, J., joined by Thomas, J., and Ginsburg, J., dissenting) (noting that "[i]t can also be argued that a broader construction would mean that Congress exceeded its authority under the Commerce Clause,” but not addressing that argument (citing Brief for the National Association of Criminal Defense Lawyers as Amicus Curiae 16-17)).
. See 18 U.S.C. § 3583(d) (making compliance with SORNA "an explicit condition” of a sex offender's supervised release).
. Similarly, the law concerning Congress's military powers suggests that Congress does not have continuing military jurisdiction over Kebodeaux after he was discharged from the military. Except in very limited situations, a discharged person is no longer subject to the Uniform Code of Military Justice. See 10 U.S.C. § 803. In United States ex rel. Toth v. Quarles, 350 U.S. 11, 13, 22-23, 76 S.Ct. 1, 100 L.Ed. 8 (1955), the Court held that the Necessary and Proper Clause does not give the federal government power to try an ex-military serviceman by court-martial five months after he left the military for a crime committed while in the military. Because he had left the military, he had the same Article III protections as did any ordinary civilian.
If anything, the link between the military power and the federal government’s action is even more attenuated in this case than in Toth, because the court-martial in Toth served the purpose of punishing someone for his illegal conduct while in the military, see id. at 13, 76 S.Ct. 1, whereas here the sex-offender-registration requirements serve no such purpose. As discussed, SORNA's purpose is merely to reduce the risk to society posed by one who has committed certain crimes. See 42 U.S.C. § 16901 (stating that SORNA’s purpose is to "protect the public from sex offenders and offenders against children”). Indeed, the government does not argue that it still has military jurisdiction over Kebodeaux, but only that its power to criminalize his predicate crime includes the power to regulate his present-day conduct.
. See 18 U.S.C. § 2250(b); Resp. to Pet. for Reh'g En Banc at 12.
. See 42 U.S.C. § 16913(a)-(c) (requiring a sex offender to register in each jurisdiction in which he resides and to update that registration); 18 U.S.C. § 2250(a) (criminalizing the failure to update registration upon any change of address if one has been convicted of a federal sex offense); U.S. Dep’t of Justice, The National Guidelines for Sex Offender Registration and Notification 6 (2008), available at http://www.ojp.usdoj.gov/smart/pdfs/finaL sornaguidelines.pdf (“[SORNA] generally constitutes a set of minimum national standards and sets a floor, not a ceiling, for jurisdictions’ programs.”).
. Texas and forty-six other states do not substantially comply with SORNA. Tex. Senate Criminal Justice Comm., Interim Report to the 82nd Legislature 14 (2011), available at http:// www.senate.state.tx.us/75r/senate/commit/c 590/c590.InterimReport81.pdf. One of the problems with SORNA is that it "relies solely on [the] offense” of conviction to determine whether a former sex offender is a threat to public safety, not "risk assessments” of a sex offender’s likelihood to reoffend. Id.; see also id. at 19 (recommending risk assessments). In addition, it does so without any apparent increase in effectiveness, because "[t]he recidivism rate of those on the registry is not lower than that of the individuals not on the registry.” Id. at 16.
. The panel majority also urged that it would be unwise to decide in favor of Kebodeaux because that would require disagreeing with United States v. George, 625 F.3d 1124, 1130 (9th Cir.2010), vacated on other grounds, 672 F.3d 1126 (9th Cir.2012). That case, however, is easily distinguishable.
Because the defendant in George was convicted in 2008, compliance with SORNA was an explicit condition of his sentence. 18 TJ.S.C. § 3583(d). He therefore fell into the category of offenders to whom SORNA is perfectly constitutional. But because Kebodeaux was long free from federal custody before SORNA even existed, he is in a different category that George had no occasion to consider. To the extent George implies that the federal government has Article I power to regulate anyone who ever committed a federal sex crime — and by implication anyone who ever committed any federal crime, because it has a "direct supervisory interest” over them — its reasoning stretches far beyond the issue before that court and is unpersuasive.
. See Lopez, 514 U.S. at 558-59, 567, 115 S.Ct. 1624 (citation omitted) (holding that because the Gun-Free School Zones Act does not fall within any of the three categories, it is an unconstitutional exercise of federal power).
. N. Am. Co. v. SEC, 327 U.S. 686, 705, 66 S.Ct. 785, 90 L.Ed. 945 (1946) (citing Brooks v. United States, 267 U.S. 432, 436-37, 45 S.Ct. 345, 69 L.Ed. 699 (1925)); accord Lopez, 514 U.S. at 558, 115 S.Ct. 1624 ("Congress may regulate the use of the channels of interstate commerce.”).
. See United States v. Whaley, 577 F.3d 254, 258 (5th Cir.2009) ("Because § 2250[ (a)(2)(B) ] applies only to those failing to register or update a registration after traveling in interstate commerce — in this case, Whaley traveled from Kansas to Texas — it falls squarely under the first Lopez prong.”).
. See also id. at 260 ("And perhaps most significantly ... a [state] sex offender who does not travel in interstate commerce may ignore SORNA's registration requirements without fear of federal criminal consequences.").
. Compare 18 U.S.C. § 2250(a)(2)(A) with § 2250(a)(2)(B). The structure of § 2250(a) is such that all federal sex offenders are covered under § 2250(a)(2)(A), but all remaining sex offenders, i.e., state sex offenders, are under § 2250(a)(2)(B).
. In Carr, the Solicitor General expressly asserted that § 2250(a) "reaches two categoríes of sex offenders: those whose underlying sex offenses were criminalized by virtue of federal or tribal authority ..., and all other sex offenders whose actions directly implicated Congress’s Commerce Clause authority as a result of 'travelling] in interstate or foreign commerce....’” Brief for United States at 21-22, Carr, 130 S.Ct. 2229 (No. 08-1301), 2010 WL 181570, at *21-22; see Carr, 130 S.Ct. at 2238 ("According to the Government, these categories correspond to two alternate sources of power to achieve Congress’s aim of broadly registering sex offenders." (internal quotation marks omitted)).
. The recent Tenth Circuit case that the panel concurrence cited is inapposite; it addresses only whether § 2250(a)(2)(A) is constitutional under the Commerce Clause on the assumption that requiring intrastate sex offender registration is constitutional, an assumption that trivializes the whole question. See United States v. Yelloweagle, 643 F.3d 1275, 1289 (10th Cir.2011) (holding that Congress has the power to criminalize a federal sex offender’s intrastate failure to register under § 2250(a)(2)(A) on the conceded assumption that it has the power to require a federal sex offender to register purely intrastate activity), cert. denied, - U.S. -, 132 S.Ct. 1969, 182 L.Ed.2d 821 (2012). If anything, that the panel majority made sure to consider the issue only on those exceptionally narrow grounds suggests that it attempted to avoid precisely the weightier question that we face here.
. See, e.g., 18 U.S.C. § 228(a)(2) (criminalizing interstate travel to evade child support obligations); § 1073 (interstate flight to avoid prosecution, giving testimony, service of process, or contempt proceedings under state or federal law); § 1201(a)(1) (interstate transportation of a kidnaped person); § 1231 (interstate transportation of strikebreakers); § 1369 (interstate travel with intent to injure or destroy a public monument); § 2101 (interstate travel with intent to cause riots); § 2261(a)(1) (interstate travel with intent to commit domestic violence); § 2421 (interstate transportation of prostitutes); § 2423 (interstate transportation of minors for illicit purposes); Morrison, 529 U.S. at 613 n. 5, 120 S.Ct. 1740 (noting 18 U.S.C. § 2261(a)(1), which criminalizes interstate spousal abuse). Most obviously, 18 U.S.C. § 2250(a)(2)(B), which criminalizes a state sex offender’s travel across state lines without having registered, would no longer need interstate travel as a jurisdictional hook; Congress could require registration of all sex offenders generally.
. See 42 U.S.C. §§ 16920-16921 (stating that the National Sex Offender Registry's web site shall include “relevant information ... listed on a jurisdiction’s Internet site” and that the Attorney General shall include information in the Registry obtained from “an appropriate official in the jurisdiction" of registration); Sex Offender Registry Websites, FBI.Gov, http://www.fbi.gov/scams-safety/registry (last visited June 6, 2012) (linking to every state sex offender registry and explaining that "the *250national registry simply enables a search across multiple jurisdictions”).
.See Morrison, 529 U.S. at 611, 120 S.Ct. 1740 ("Were the Federal Government to take over the regulation of entire areas of traditional state concern, areas having nothing to do with the regulation of commercial activities, the boundaries between the spheres of federal and state authority would blur” (quoting Lopez, 514 U.S. at 577, 115 S.Ct. 1624 (Kennedy, J., concurring)), "and political responsibility would become illusory,” Lopez, 514 U.S. at 577, 115 S.Ct. 1624 (Kennedy, J„ concurring)).
. See Darby, 312 U.S. at 114, 61 S.Ct. 451 ("Th[e power of Congress over interstate commerce] can neither be enlarged nor diminished by the exercise or non-exercise of state power.”).
. Lopez, 514 U.S. at 558, 115 S.Ct. 1624 (citing S. Ry. Co. v. United States, 222 U.S. 20, 32 S.Ct. 2, 56 L.Ed. 72 (1911)).
. Id. (citing Perez v. United States, 402 U.S. 146, 150, 91 S.Ct. 1357, 28 L.Ed.2d 686 (1971)).
. Id. (citing Perez, 402 U.S. at 150, 91 S.Ct. 1357).
. See Morrison, 529 U.S. at 618, 120 S.Ct. 1740 ("The regulation ... of intrastate violence that is not directed at the instrumentalities, channels, or goods involved in interstate commerce has- always been the province of the States.” (citing Cohens v. Virginia, 19 U.S. (6 Wheat.) 264, 426, 428, 5 L.Ed. 257 (1821) (Marshall, C.J.))).
. See Raich, 545 U.S. at 25, 125 S.Ct. 2195 ("Unlike those at issue in Lopez and Morrison, the activities regulated by the CSA are quintessentially economic.”); id. at 25-26, 125 S.Ct. 2195 (defining "economic” activity as "the production, distribution, and consumption of commodities”).
. See Morrison, 529 U.S. at 610-11, 120 S.Ct. 1740 ("[A] fair reading of Lopez shows that the noneconomic, criminal nature of the conduct at issue was central to our decision in that case.... Lopez's review of Commerce Clause case law demonstrates that in those cases where we have sustained federal regulation of intrastate activity based on the activity’s substantial effects on interstate commerce, the activity in question has been some sort of economic endeavor.” (citations omitted)).
.See id. at 613, 120 S.Ct. 1740 ("[T]hus far in our Nation’s history our cases have upheld Commerce Clause regulation of intrastate activity only where that activity is economic in nature.”).
. See Raich, 545 U.S. at 22, 125 S.Ct. 2195 (holding that Congress has the authority to enact "comprehensive legislation to regulate the interstate market” even where that "regulation ensnares some purely intrastate activity”); see Whaley, 577 F.3d at 259 (upholding 42 U.S.C. § 16913 — which requires sex offenders to register changes of address — even though it applies to intrastate activity, because, without it, "§ 2250 [which criminalizes the failure to register] has no substance”).
. Cf. Carr, 130 S.Ct. at 2248 (Alito, J., joined by Thomas and Ginsburg, JJ., dissenting) (noting that it “can also be argued” that interpreting § 2250(a)(2)(B) — the state sex offender provision — to apply to interstate travel that occurred before SORNA’s enactment "would mean that Congress exceeded its authority under the Commerce Clause.”). That is a fortiori the case here, with the government arguing that an analogous statute requiring no interstate travel at all is constitutional.
.See Morrison, 529 U.S. at 618, 120 S.Ct. 1740 ("The regulation ... of intrastate violence that is not directed at the instrumentalities, channels, or goods involved in interstate commerce has always been the province of the States.” (citing Cohens, 19 U.S. (6 Wheat.) at 428) (Marshall, C J.)).
. The unconstitutionality applies only as to those in the narrow and specific circumstance faced by Kebodeaux, and we make no holding as to others.
. 42 U.S.C. § 14072(i)(4) (Supp. IV 1999), repealed by Sex Offender Registration and Notification Act, Pub.L. No. 109-248, 120 Stat. 587 (2006).