On January 20, 2026, the Supreme Court handed down a unanimous decision in Ellingburg v. United States that could reshape how courts analyze the "civil" consequences of criminal convictions.
The case involved a narrow question about federal restitution. But Justice Thomas's concurrence—joined by Justice Gorsuch—fired a shot across the bow of Smith v. Doe, the 2003 decision that has shielded retroactive sex offender registration laws from constitutional challenge for over two decades.
The Ellingburg Decision: A Unanimous Court, A Divided Rationale
Holsey Ellingburg robbed a bank in 1995. He was sentenced in 1996 under the newly enacted Mandatory Victims Restitution Act, which required him to pay $7,567.25 to his victims.
Nearly thirty years later, after paying down the balance while incarcerated, Ellingburg still owed over $13,000 because of interest. The government claimed it could keep collecting until 2042, based on a retroactive amendment that extended the collection period.
The constitutional question: Is restitution "criminal punishment" for purposes of the Ex Post Facto Clause?
All nine justices said yes. Justice Kavanaugh, writing for the Court, focused on statutory text and structure. The MVRA labels restitution a "penalty." It's imposed at sentencing. It appears in Title 18, "Crimes and Criminal Procedure." Failure to pay can result in imprisonment.
"When viewed as a whole," Kavanaugh wrote, "the MVRA makes abundantly clear that restitution is criminal punishment."
So far – unremarkable. The Court applied its existing two-step framework from Smith v. Doe: first, ask whether the legislature intended the measure to be civil or criminal; second, if civil, ask whether the scheme is "so punitive either in purpose or effect" as to negate that intent.
But then Justice Thomas wrote separately—and what he wrote should concern every prosecutor defending retroactive sex offender registration.
Thomas's Concurrence: Dismantling the Smith v. Doe Framework
Justice Thomas agreed with the result but attacked the analytical framework. His eighteen-page concurrence, joined by Justice Gorsuch, argues that modern Ex Post Facto jurisprudence has drifted far from its constitutional moorings. The drift, he wrote, has given legislatures a roadmap to "manipulate when the protection will apply."
The problem, according to Thomas, is that the Smith v. Doe framework turns on "legislative labeling and semantics." Under that framework, a state can impose crushing burdens on people after they commit crimes—restrictions on where they can live, work, and travel; requirements to report in person; lifetime internet monitoring—and call it all "civil regulation."
As long as the legislature expresses a nonpunitive intent, puts the law in the civil code, and avoids certain procedural safeguards, courts will generally defer.
Thomas called this "incongruous with the historical purpose" of the Ex Post Facto Clause. The Founders, he noted, viewed ex post facto laws as "the most essential" protection in the Constitution, worth "ten thousand declarations of rights."
James Madison called them "contrary to the first principles of the social compact." Alexander Hamilton ranked the prohibition "among the greatest securities to liberty."
These protections, Thomas argued, should not depend on clever legislative drafting.
A Return to Calder v. Bull
Thomas's alternative framework returns to the Constitution's original meaning. In Calder v. Bull (1798), the Court established that the Ex Post Facto Clauses apply to laws imposing "punishment" for "crimes." Thomas argues that, properly understood, this means any "coercive penalty" for a "public wrong" regardless of the label.
The key question under this approach: Does the law impose a penalty enforced by the government to redress an injury to the sovereign? If yes, it's subject to Ex Post Facto protection. If the law merely resolves private disputes between citizens, it's not.
This test, Thomas wrote, "simplifies the law" and "harmonizes the Ex Post Facto Clauses with their historical purpose." Instead of weighing twelve factors across two tests with "indeterminate weights" that invite "arbitrary discretion" courts would ask a straightforward question: Who enforces the law, and to redress what kind of wrong?
Thomas explicitly called out nominally civil penalties that should receive Ex Post Facto protection: "civil proceedings seeking fines for public offenses," "enforcement proceedings brought by administrative agencies," and even "municipal sanctions like speeding tickets."
If the state imposes retroactive burdens as coercive penalties for public wrongs, Thomas wrote, the Constitution protects against them.
At the end of his concurrence, Thomas invited a future challenge: "In a future case, the Court should consider returning to Calder's understanding."
Florida's Sex Offender Registry: A Prime Target
Few statutory schemes illustrate Thomas's concerns better than Florida's sex offender registration laws.
Since 1993, Florida has repeatedly expanded the burdens on people required to register. What began as a straightforward reporting requirement has grown into what the Sixth Circuit—striking down Michigan's similar scheme in Doe v. Snyder (2016)—called a "byzantine code governing in minute detail the lives of the state's sex offenders."
Consider what Florida now requires:
Registration requirements: Sexual predators and offenders must report in person to the sheriff's office within 48 hours of establishing any residence—permanent, temporary, or transient.
Transient registrants must report every 30 days. Registrants must provide their name, date of birth, social security number, address, employer information, vehicle information, internet identifiers, email addresses, and phone numbers.
Residency restrictions: Since 2005, Florida has prohibited sexual predators from living within 1,000 feet of schools, day care centers, parks, playgrounds, and school bus stops. Many municipalities have added their own restrictions, some extending to 2,500 feet. In urban areas, these zones overlap so extensively that registrants face de facto banishment.
Employment and travel restrictions: Registrants must report changes in employment within 48 hours. They must notify the sheriff before leaving the state. They cannot work or volunteer in any capacity that involves contact with minors at schools, day care facilities, parks, playgrounds, or other places where children congregate.
Internet identifier requirements: Florida requires registrants to report all email addresses, instant messaging names, chat room identities, and social networking profiles—requirements that didn't exist when most people on the registry committed their offenses.
Driver's license marking: Sexual predators must carry licenses marked with a designation that identifies their status to anyone who sees it.
Duration: For many offenders, registration is permanent. The few paths to removal were narrowed in 2007, when the legislature extended eligibility from 20 years to 25 years and eliminated relief for juvenile offenders.
Enforcement: Violations are felonies. A first offense for failing to register is a third-degree felony. Subsequent violations are second-degree felonies, punishable by up to fifteen years in prison.
Florida has applied these requirements retroactively to people whose offenses predated every amendment. Someone convicted in 1994—before Florida even created a public registry—now faces the same lifetime obligations, residency restrictions, and felony penalties as someone convicted this year.
Why Smith v. Doe Shouldn't Control Florida
The Supreme Court's 2003 decision in Smith v. Doe upheld Alaska's sex offender registration law against an Ex Post Facto challenge. But the Alaska law was modest by today's standards. It required annual or quarterly registration with law enforcement and publication of registrant information online.
It imposed no residency restrictions, no employment limitations, no requirements to report internet identifiers, and no in-person reporting requirements.
Justice Kennedy, writing for the majority, emphasized what Alaska's law did not do:
"[T]he Act does not restrain activities sexual offenders may pursue but leaves them free to change jobs or residences. . . . [O]ffenders subject to the Act are free to move where they wish and to live and work as other citizens, with no supervision."
Florida's scheme imposes exactly the kinds of restraints Smith v. Doe said would weigh toward punitiveness. Registrants are not free to move where they wish—they're excluded from vast swaths of residential areas.
They're not free to work as other citizens—they face categorical employment bans. They are subject to intensive supervision through mandatory in-person reporting, address verification, and internet monitoring.
The Sixth Circuit recognized this in Doe v. Snyder. Applying Smith v. Doe's own factors, the court found Michigan's registration scheme punitive in effect. The scheme "resembles, in some respects at least, parsing the outrageous and outdated colonial practice of branding." It imposed "affirmative disabilities" far exceeding what Alaska required in 2003.
And critically, the evidence showed that registration requirements had no "rational connection to a nonpunitive purpose"—studies indicated that sex offender registration does not reduce recidivism and may actually increase it.
Florida's scheme is at least as burdensome as Michigan's.
The Thomas Framework: A Path Forward
Justice Thomas's concurrence offers criminal defense attorneys a more direct path to challenge retroactive registration.
Under the Thomas framework, the critical question is not what the legislature called the law or where it codified it. The question is whether the law imposes a "coercive penalty" for a "public wrong."
Florida's registration scheme easily meets this test:
It penalizes a public wrong. Registration requirements apply only to people convicted of specific criminal offenses. The purpose is not to resolve private disputes but to address what the legislature views as an ongoing threat to public safety arising from the original crime.
The penalties are coercive. Failure to comply is a felony. The state can—and does—imprison people for registration violations. That's the definition of coercive.
The state is the enforcer. Registrants are not sued by victims. They're prosecuted by the State of Florida. The state initiates enforcement, determines whether violations occurred, and seeks imprisonment.
Under Thomas's framework, these features make registration subject to Ex Post Facto protection regardless of what the legislature said about its civil or regulatory nature. Retroactive application of any expansion—from the original 1993 requirements through today's comprehensive scheme—would violate the Constitution.
Implications for Criminal Defense Practice
What does this mean for practitioners?
Preserve the argument. In every case involving retroactive sex offender registration requirements—whether a failure-to-register prosecution or a direct challenge to registration—raise the Ex Post Facto argument and cite Ellingburg. The Thomas concurrence isn't binding law, but it signals where two justices would go if given the opportunity.
Build the record. The Sixth Circuit in Doe v. Snyder succeeded partly because counsel developed an extensive factual record showing that registration requirements impose substantial burdens and don't achieve their stated purpose. Future challenges should include evidence about:
- The practical effects of residency restrictions (homelessness rates, inability to live with family)
- Employment barriers created by registration
- The scope of in-person reporting and verification requirements
- Scientific literature on sex offender recidivism rates
- Studies on the effectiveness (or ineffectiveness) of registration schemes
Identify the best plaintiffs. As-applied challenges require plaintiffs whose cases illustrate the punitive nature of the scheme. Look for clients whose offenses predated major amendments, who face severe residency restrictions, who have been separated from family members, or who have experienced homelessness or employment loss due to registration.
Frame the constitutional question correctly. Don't just argue that Florida's scheme is "punitive in effect" under Smith v. Doe's second step. Argue, following Thomas, that the entire framework is flawed. The Constitution protects against retroactive coercive penalties for public wrongs—period. Legislative labeling cannot override that protection.
The Road Ahead
Justice Thomas closed his concurrence by noting that in Collins v. Youngblood (1990), the Court said that "any departure from Calder's explanation of the original understanding of the Ex Post Facto Clause was . . . unjustified." He invited the Court to follow that principle to its conclusion.
Two justices are ready to reconsider Smith v. Doe. Others may join them. The Sixth Circuit has already found a sex offender registration scheme punitive under the existing framework.
Several state supreme courts—Alaska, Indiana, Maine, Maryland, New Hampshire, Ohio, Oklahoma, and Pennsylvania among them—have reached similar conclusions under state constitutions.
Florida's registration scheme has never been comprehensively challenged in federal court. After Ellingburg, the time may be right.
The Ex Post Facto Clause exists because, as Thomas wrote, "it is far more dangerous to permit the government under guise of civil regulation to punish people without prior notice" than to leave some wrongdoing unpunished.
That danger is exactly what Florida's retroactive registration scheme presents. Calling punishment "regulation" doesn't make it constitutional.
For the thousands of Floridians subject to lifetime registration for offenses committed before most of today's requirements existed, Ellingburg offers a new reason for hope—and for their attorneys, a new argument to make.
