Learn About the Law
Get help with your legal needs
FindLaw’s Learn About the Law features thousands of informational articles to help you understand your options. And if you’re ready to hire an attorney, find one in your area who can help.
Mary DOE; John Doe a; John Doe B; John Doe C; John Doe D; John Doe E; John Doe G; John Doe H; Mary Roe, Plaintiffs-Appellants/cross-Appellees, v. Gretchen WHITMER, Governor of the State of Michigan; James Grady, Ii, Colonel, Defendants-Appellees/cross-Appellants.
OPINION
Ten years ago, in Does #1-5 v. Snyder, 834 F.3d 696, 705–06 (6th Cir. 2016) (“Does I”), we concluded that the 2011 version of Michigan's Sex Offenders Registration Act (“SORA 2011”) imposed punishment, and consequently, we held that its retroactive application violated the Ex Post Facto Clause. In 2021, Michigan amended SORA in an effort to remove the features of the scheme that led to its invalidation. The plaintiffs and class members are convicted sex offenders required to register under the current iteration of the scheme (“SORA 2021”). They brought this suit shortly after SORA 2021's enactment, claiming that it continued to impose retroactive punishment in violation of the Ex Post Facto Clause and that it violated the Constitution on myriad other grounds.
As relevant here, the district court found that SORA 2021's retroactive application violates the Ex Post Facto Clause and that the Michigan State Police's (“MSP”) treatment of individuals with non-Michigan convictions violates their procedural-due-process rights. But the district court rejected substantive-due-process and equal-protection challenges to the scheme's reliance on offense-based tiers, an equal-protection challenge to limits on some registrants’ abilities to petition for removal from the scheme's requirements, and void-for-vagueness challenges to certain reporting obligations.
For the reasons that follow, we AFFIRM IN PART, VACATE IN PART, and REMAND for further proceedings consistent with this opinion.
I. BACKGROUND
A. Michigan's Sex Offenders Registration Act
In 1994, Congress enacted the Jacob Wetterling Crimes Against Children and Sexually Violent Offender Registration Act, Pub. L. No. 103–322, 108 Stat. 2038, 42 U.S.C. § 14071 et seq. (repealed 2006), which tied states’ eligibilities for certain federal funds to their adoption of sex-offender registries. Michigan responded that year, enacting SORA. See Pub. Act No. 295, 1994 Mich. Pub. Acts 1522. SORA created “a non-public registry maintained solely for law enforcement use.” Does I, 834 F.3d at 697. The statute required registrants to provide law enforcement with their names, social security numbers, addresses, and photographs, and assigned registrants to either twenty-five-year terms (for one qualifying offense) or lifelong registration (for two or more qualifying offenses). Pub. Act No. 295, §§ 5(1), (3)–(4), 7, 1994 Mich. Pub. Acts at 1524–26. Willful violations were subject to criminal penalties. Id. § 9(1), 1994 Mich. Pub. Acts at 1526.
Since enactment, Michigan has amended SORA over twenty times. 1 In the late 1990s, Michigan made the registry public, first allowing the public to inspect the registry in person at police stations, Act of Jan. 7, 1997, Pub. Act No. 494, § 10(2), 1996 Mich. Pub. Acts 2283, 2285, and then creating a public website, Act of June 28, 1999, Pub. Act No. 85, § 8(2), 1999 Mich. Pub. Acts 271, 279–80. Michigan, moreover, added both periodic and episodic reporting requirements. The periodic-reporting requirements compelled registrants to verify their information in person either annually or quarterly, depending on their offenses of conviction. Id. § 5a(4), 1999 Mich. Pub. Acts at 277. The episodic-reporting requirements obligated registrants to provide notice within ten days (and not necessarily in person) of changes to their residences, places of work, or schools. Id. § 5(1), 1999 Mich. Pub. Acts at 275.
Before Michigan enacted SORA 2021, two sets of amendments stood apart from the rest. First, in 2006, Michigan created “[s]tudent safety zone[s].” Mich. Comp. Laws § 28.733(f) (2006) (repealed 2021). These zones subjected most registrants to criminal penalties if they lived, worked, or loitered within 1,000 feet of a school. Id. §§ 28.734, 28.735 (2006) (repealed 2021). SORA 2011 retained the student safety zones.
Second, in 2011, Michigan overhauled SORA in response to shifting federal requirements. Five years earlier, Congress had enacted the Sex Offender Registration and Notification Act (“SORNA”), Pub. L. No. 109-248, 120 Stat. 590, 34 U.S.C. § 20901 et seq. SORNA accomplishes two objectives: it both regulates state and federal sex offenders directly and “set[s] forth comprehensive registration-system standards.” Reynolds v. United States, 565 U.S. 432, 435 (2012). States that fail to “substantially implement” SORNA's standards lose out on federal funds. 34 U.S.C. §§ 20927(a), 20912(a).
Broadly speaking, to access those funds, states must adopt public, offense-based sex-offender registries. Per SORNA, states must create public websites that allow users to search by zip code or within a set “geographic radius” for registrants. Id. § 20920(a). States must also allow “individual[s]” to receive notice when registrants update their registrations. Id. § 20923(b)(7). Registrants themselves must be sorted into three tiers based on their offenses of conviction. Id. § 20911(2)–(4). Tier assignments, in turn, set the length of time that registrants must remain on the registry (fifteen years at tier I, twenty-five years at tier II, and life at tier III), id. § 20915(a); periodic in-person reporting obligations (annual reporting at tier I, biannual reporting at tier II, and quarterly reporting at tier III), id. § 20918; eligibility for exclusion from online registries, see id. § 20920(c)(1); and eligibility to petition for early removal from registries, id. § 20915(b). States must collect a plethora of information about registrants (including their names; social security numbers; home, work, and school addresses; license plate numbers; photographs; fingerprints; DNA samples; “[r]emote communication identifiers”; and passport information). See id. § 20914(a)–(b); 28 C.F.R. § 72.6. States must oblige registrants to report episodic occurrences (such as changes to names, addresses, employment, or student status) in person and within three days. 34 U.S.C. § 20913(c); see 28 C.F.R. § 72.7(c)–(f). SORNA, moreover, requires states to criminalize registrants’ noncompliance. 34 U.S.C. § 20913(e).
As Michigan's deadline to comply with SORNA approached, 2 the state enacted amendments that restructured the scheme. SORA 2011 authorized a public website that allowed users to search for registrants by zip code, municipality, and geographic area. Mich. Comp. Laws § 28.728(2), (7) (2011). Via the website, Michigan allowed the public to sign up to be notified whenever a registrant, within a set geographic radius, was added to the online registry or updated their information. Id. § 28.730(3) (2011). SORA 2011 adopted three offense-based tiers, id. § 28.722(r)–(w) (2011), and imposed the same registration durations and periodic in-person reporting requirements as SORNA, id. §§ 28.725(10)–(13), 28.725a(3) (2011). Tier designation further determined registrants’ presence in the online registry, id. § 28.728(4)(c) (2011) (excluding tier I registrants), and ability to petition for removal, id. § 28.728c(1)–(3), (12)–(15) (2011) (permitting tier I registrants to petition for removal five years early but generally barring tier II and III offenders from petitioning for removal). Michigan collected a set of information from registrants that was substantially identical to that required by SORNA, id. § 28.727(1) (2011), and required registrants to report episodic occurrences, including those required by SORNA, in person and within three business days, id. §§ 28.722(g), 28.725(1)–(2), (6)–(7) (2011). SORA 2011, furthermore, criminalized registrants’ violations of its reporting requirements. Id. § 28.729 (2011).
SORA 2011 did not align perfectly with SORNA, however. Most notably, SORNA required nothing like the student safety zones that SORA 2011 retained. See id. §§ 28.734, 28.735 (2006) (repealed 2021). SORA 2011's episodic reporting requirements exceeded those set by SORNA. Compare id. § 28.725(1)–(2), (6)–(7) (2011) (requiring, for example, registrants to report changes to internet identifiers, temporary residences, and vehicle information in person) with 34 U.S.C. § 20913(c) and 28 C.F.R. § 72.7(c)–(f) (allowing jurisdictions to collect these updates by “whatever means”). Although not mandated by SORNA, Michigan charged registrants a one-time registration fee. Mich. Comp. Laws § 28.725a(6) (2011). And SORA 2011 published more information about registrants online than SORNA required. Compare id. § 28.728(2)(l) (2011) (publishing tier designations) with 34 U.S.C. § 20920(a)–(c) and National Guidelines for Sex Offender Registration and Notification, 73 Fed. Reg. 38030, 38059 (July 2, 2008) (excluding tier designations from the list of information that jurisdictions must publish on their websites).
Notwithstanding these differences, Michigan decided to apply both the 2006 and 2011 amendments retroactively, Does I, 834 F.3d at 698, as SORNA requires, see 28 C.F.R. § 72.3. Retroactive application of the 2011 amendments altered the status of thousands of registrants. Before the 2011 amendments, roughly a quarter of registrants were required to register for life; after, roughly three quarters of registrants received life terms. See R. 128-15 (Does I Joint Statement of Facts ¶¶ 289–90) (Page ID #6756). All told, SORA 2011 extended roughly 17,000 registrants’ registration terms to life. Id. ¶ 288 (Page ID #6756).
In Does I, six plaintiffs, all tier III registrants convicted of sex offenses before 2006, challenged the retroactive application of SORA 2011 pursuant to the Ex Post Facto Clause. 834 F.3d at 698. Although we determined that the legislature did not intend SORA 2011 to impose punishment, we concluded that its actual effect was punitive. Id. at 700–06. As a result, we held that “[t]he retroactive application of SORA's 2006 and 2011 amendments to Plaintiffs [was] unconstitutional,” and ordered Michigan to “cease” applying them to the plaintiffs. Id. at 706.
On the heels of Does I, six new plaintiffs filed a class-action suit seeking relief for all registrants convicted before 2011. Doe v. Snyder (“Does II”), 449 F. Supp. 3d 719 (E.D. Mich. 2020). The Does II court read Does I as invalidating the 2006 and 2011 amendments in their entirety, found that the amendments were not severable, and permanently enjoined the retroactive application of SORA 2011. Does II, 449 F. Supp. 3d at 728–29, 731–35.
Litigation concerning SORA 2011 proceeded in parallel in Michigan state court. About one month before the entry of final judgment in Does II, the Michigan Supreme Court decided People v. Betts, 968 N.W.2d 497 (Mich. 2021). Betts agreed with Does I and Does II, holding that “the retroactive imposition of the 2011 SORA increases registrants’ punishment for their committed offenses in violation of federal and state constitutional prohibitions on ex post facto laws.” Id. at 515. The Michigan Supreme Court further concluded that the 2006 and 2011 amendments were not severable. Id. at 516–21. Accordingly, the court invalidated SORA 2011's retroactive application. Id. at 521.
Michigan amended SORA in 2021. See Mich. Comp. Laws § 28.721 et seq. Altogether, SORA 2021 included three major amendments. First, SORA 2021 jettisoned the student safety zones introduced in the 2006 amendments. Id. §§ 28.733–28.735 (2006) (repealed 2021). Second, it removed registrants’ tier designations from display in the online registry. Id. § 28.728(3)(e). Third, it tinkered with the scheme's episodic reporting requirements. Under SORA 2021, the Michigan State Police (“MSP”) may eschew in-person reporting requirements when registrants update their names, addresses, places of work, schools, vehicle information, email addresses, internet identifiers, 3 or temporary residences. Id. § 28.725(1), (2).
SORA 2021 retained most other aspects of SORA 2011. SORA 2021 mandates the creation of a publicly accessible online registry with the same search and notification features as SORA 2011 required. See id. §§ 28.728(2), (7), 28.730(3). The website continues to publish large quantities of information about registrants. Id. § 28.728(2). It now, however, allows the public to “submit ․ tip[s]” about registrants. R. 129-2 (Defs.’ SOMF Resp. ¶ 250) (Page ID #7359–60). SORA 2021 still divides registrants into three tiers based on their offenses of conviction, id. § 28.722(q)–(v), 4 and for each tier, imposes the same registration durations, periodic-reporting frequencies, and removal-petition limitations as SORA 2011, id. §§ 28.725(11)–(13), 28.725a(3), 28.728c(1)–(3), (12)–(15). Although SORA 2011 excluded all tier I offenders from the online registry, SORA 2021 excludes most, but not all, tier I offenders. Id. § 28.728(4)(c). Michigan continues to collect virtually the same information from registrants as it did under SORA 2011. Id. § 28.727(1). SORA 2021 retains some in-person reporting requirements for episodic occurrences, but the MSP has utilized its authority under SORA 2021 to allow registrants to report certain episodic occurrences by mail. See id. § 28.725(1)–(2), (7)–(8); R. 126-17 (Mich. Sex Offender Registration/Verification/Update Form at 4) (Page ID #6207). SORA 2021 still charges registration fees, but now charges annual fees on top of an initial registration fee. Mich. Comp. Laws § 28.725a(6)(a)–(b). When registrants “willfully” fail to abide by SORA 2021's requirements, they are subject to criminal liability. Id. § 28.729.
Most significantly, SORA 2021 applies retroactively, extending the registration terms of thousands of registrants convicted before 2011—often to lifetime registration. R. 129-2 (Defs.’ SOMF Resp. ¶ 140) (Page ID #7287). Thus, as was the case under SORA 2011, the vast majority—seventy-three percent—of registrants under SORA 2021 are in tier III. Id. ¶ 76 (Page ID #7260). Tier II, by contrast, comprises twenty percent of SORA 2021 registrants, and tier I, just seven percent. Id.
Most recently, and after oral argument in this case, the Michigan Supreme Court determined that SORA 2021, like SORA 2011, violates the ex post facto clauses of the state and federal constitutions. People v. Smith, 38 N.W.3d 736 (Mich. 2026) (mem.). SORA 2021, the court held, cannot “be retroactively applied to registrants whose criminal acts subjecting them to registration occurred before enactment of the 2011 SORA amendments.” Id. Further, as in Betts, the court determined that “severance or revival of prior versions of the SORA is not appropriate.” Id.; see Betts, 968 N.W.2d at 516–21.
B. Procedural History
In response to SORA 2021, the plaintiffs brought a class-action suit in the U.S. District Court for the Eastern District of Michigan in February 2022. R. 1 (Compl. ¶¶ 1–764) (Page ID #1–198). The operative complaint names nine plaintiffs relevant here—John Does A, B, C, D, E, G, H, Mary Doe, and Mary Roe. 5 R. 108 (Am. Compl. ¶¶ 23–153) (Page ID #2780–2802). All committed registerable offenses before July 1, 2011 (save for John Doe H, who committed his in 2015), and none have been convicted of another registerable offense since committing their original offenses. R. 129-2 (Defs.’ SOMF Resp. ¶¶ 3–71) (Page ID #7234–59). Nearly all saw their registration terms extended under SORA 2021 to life. See id. All were assigned to their tiers without receiving individualized risk assessments, and nearly all can never petition for removal from the registry. See id.
The district court certified a class of all “people who are or will be subject to registration under” SORA 2021. R. 35 (5/18/2022 Order Grant'g Class Certification ¶ 2) (Page ID #1116–17). In addition to this primary class, the district court certified seven subclasses, including subclasses of individuals who committed their registerable offenses before SORA 2011's enactment (the “pre-2011 ex post facto subclass”), individuals required to register for life as a result of SORA amendments (the “retroactive extension of registration subclass”), individuals mandated to register because they pleaded guilty to a registerable offense before SORA 2011 (the “plea bargain subclass”), individuals unable to petition for removal from the registry (the “barred from petitioning subclass”), and individuals covered by SORA 2021 as a result of non-Michigan convictions (the “non-Michigan offense subclass”). R. 35 (5/18/2022 Order Grant'g Class Certification ¶¶ 3–8) (Page ID #1117–18); R. 109 (5/9/2023 Order Grant'g Class Certification ¶ 1) (Page ID #3128).
The operative complaint challenges SORA 2021's constitutionality in eleven counts brought pursuant to 42 U.S.C. § 1983. R. 108 (Am. Compl. ¶¶ 20, 727–840) (Page ID #2780, 2956–80). Seven counts are relevant here: Counts I and II, ex post facto challenges to SORA 2021's retroactive application, id. ¶¶ 727–45 (Page ID #2956–59); Count III, due-process and equal-protection challenges to SORA 2021's reliance on offense-based tiers, id. ¶¶ 746–56 (Page ID #2959–61); Count IV, an equal-protection challenge to tier II and III registrants’ general inability to petition for removal, id. ¶¶ 757–72 (Page ID #2962–66); Count VI, a due-process challenge to SORA 2021's retroactive alteration of plea agreements, id. ¶¶ 789–97 (Page ID #2969–71); Count VIII, void-for-vagueness challenges to certain of SORA 2021's reporting requirements, id. ¶¶ 809–12 (Page ID #2974); and Count XI, challenges to SORA 2021's registration process and requirements for those convicted of non-Michigan offenses under the Due Process, Equal Protection, and Privileges and Immunities Clauses, id. ¶¶ 833–40 (Page ID #2978–80).
The plaintiffs moved for a preliminary injunction, R. 7 (Pls.’ Prelim. Inj. Mot. at 1–67) (Page ID #857–946), and the defendants moved to dismiss, R. 41 (Defs.’ Mot. to Dismiss at 1–57) (Page ID #1327–83). The district court denied both motions. R. 54 (D. Ct. Order Deny'g Pls.’ Prelim. Inj. Mot. & Defs.’ Mot. to Dismiss at 1–6) (Page ID #1847–52). Following extensive discovery, the parties cross-moved for summary judgment. R. 123 (Pls.’ MSJ at 1–73) (Page ID #3621–93); R. 129 (Defs.’ MSJ at 1–77) (Page ID #7128–7204). The district court granted summary judgment to the plaintiffs on their ex post facto claims (Counts I and II) and their challenges to SORA 2021's non-Michigan-convictions provisions (Count XI), partial summary judgment to the plaintiffs and the defendants on the void-for-vagueness claims (Count VIII), and summary judgment to the defendants on the challenges to SORA 2021's offense-based tiers (Count III) and petitioning restrictions (Count IV). Does v. Whitmer (“Does III”), 751 F. Supp. 3d 761, 784–815, 818–21 (E.D. Mich. 2024); Does v. Whitmer (“Does IV”), 773 F. Supp. 3d 380, 394–400 (E.D. Mich. 2025). The district court did not address the merits of plaintiffs’ due-process claim regarding the retroactive alteration of plea agreements (Count VI), finding that claim moot in light of the plaintiffs’ success on their ex post facto counts. Does III, 751 F. Supp. 3d at 814.
The district court then entered judgment, R. 177 (Am. Judgment at 1–16) (Page ID #9377–92), and the parties cross-appealed, R. 180 (Defs.’ Notice of Appeal) (Page ID #9404); R. 181 (Pls.’ Notice of Appeal at 1–2) (Page ID #9405–06). The defendants moved for a stay pending appeal, R. 173 (Defs.’ D. Ct. Stay Mot. at 1–23) (Page ID #9206–28), which the district court denied, Does v. Whitmer, No. 22-CV-10209, 2025 WL 1428243, at *7 (E.D. Mich. May 19, 2025). The defendants then moved this court for a stay, D. 21 (Defs.’ Emergency Stay Mot. at 1–32), which we granted, Doe v. Whitmer, No. 25-1413, 2025 LX 164483, at *9 (6th Cir. June 20, 2025) (order).
II. ANALYSIS
On appeal, the parties contest the district court's judgment on six counts. The plaintiffs appeal the district court's judgment with respect to their challenges to SORA 2021's offense-based tiers (Count III), petitioning provisions (Count IV), and episodic reporting requirements pertaining to education, employment, and vehicles (Count VII). D. 67 (Pls.’ Corr. First Br. at 28–62). The defendants’ appeal concerns three different counts: the ex post facto challenges (Counts I and II) and the non-Michigan-offense challenges (Count XI). D. 70 (Defs.’ Second Br. at 13–50). We address the parties’ arguments in turn.
A. Standard of Review
We review de novo a district court's grant of summary judgment. Nelson v. MillerKnoll, Inc., 181 F.4th 651, 660 (6th Cir. 2026). Summary judgment is proper if “the movant shows that there is no genuine dispute as to any material fact and the movant is entitled to judgment as a matter of law.” Fed. R. Civ. P. 56(a). We view the evidence in the light most favorable to the nonmovants and draw all reasonable inferences in their favor. Nelson, 181 F.4th at 660. A genuine dispute of material fact exists “if the evidence is such that a reasonable jury could return a verdict for the nonmoving party.” Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 248 (1986).
B. Ex Post Facto Clause Challenges
We begin with the plaintiffs’ ex post facto claims. The Ex Post Facto Clause provides that “No State shall ․ pass any ․ ex post facto Law.” U.S. Const. Art. I, § 10. Among other types of retroactive laws, the Clause has long been understood to prohibit “[e]very law that changes the punishment, and inflicts a greater punishment, than the law annexed to the crime, when committed.” Calder v. Bull, 3 U.S. 386, 390 (1798) (opinion of Chase, J.). In Counts I and II, plaintiffs argue that SORA 2021 does just that, punishing registrants for offenses they were convicted of before 2011. See, e.g., D. 74 (Pls.’ Corr. Third Br. at 4–5).
Recently, in Smith, the Michigan Supreme Court determined that SORA 2021 constitutes “an ex post facto punishment, in violation of” the federal and state constitutions, and therefore “may not be retroactively applied to registrants whose criminal acts subjecting them to registration occurred before enactment of the 2011 SORA amendments.” 38 N.W.3d at 736. Following this decision, the plaintiffs moved to remand Counts I and II for the district court to consider whether Smith mooted those claims. D. 92-1 (Pls.’ Mot. to Remand Counts I & II at 1–3). The plaintiffs argue that the MSP has already begun to remove registrants “who committed their relevant crime prior to” the enactment of SORA 2011 to comply with the Michigan Supreme Court's order. Id. at 1–2. The defendants oppose this motion, arguing that Counts I and II are not moot because Smith may still be reconsidered or reviewed through a petition for certiorari and therefore is not final, and because the district court below required different relief on Counts I and II than was granted in Smith. 6 See D. 94 (Defs.’ Resp. to Mot. to Remand at 2–6). The plaintiffs respond that “neither [c]lass [c]ounsel nor this [c]ourt currently has the facts necessary to ․ outright” dismiss Counts I and II as moot. D. 95-1 (Pls.’ Reply Br. Supp. Mot. to Remand at 2). The plaintiffs also note that the district court's relief has been stayed pending this appeal and would not bind the defendants if Counts I and II were mooted. Id. at 7 (citing Doe v. Whitmer, 2025 LX 164483, at *8–9; D. 36 (Order Clarifying Stay)). If we determine that Counts I and II are mooted ourselves, the defendants suggest in the alternative that we remand those counts to the district court with instructions to vacate and dismiss them. D. 94 (Defs.’ Resp. to Mot. to Remand at 7–8). On balance, we agree with the plaintiffs.
While we are not bound by the Michigan Supreme Court's interpretation of the federal Constitution, see WCI, Inc. v. Ohio Dep't of Pub. Safety, 774 F. App'x 959, 967 (6th Cir. 2019), we generally defer to a state supreme court's interpretation of that state's own statutes, Bush v. Palm Beach Cnty. Canvassing Bd., 531 U.S. 70, 76 (2000); see also Coombes v. Getz, 285 U.S. 434, 441 (1932) (a state supreme court opinion “construing and applying [the state's] own [c]onstitution and laws generally is binding” on federal courts); Mathis v. Eli Lilly & Co., 719 F.2d 134, 141 (6th Cir. 1983) (similar). Although Counts I and II do not raise claims under the Michigan Constitution, see R. 108 (Am. Compl. ¶¶ 727, 741) (Page ID #2956, 2958), mootness is a jurisdictional question, and “we must assure ourselves of our own jurisdiction to hear” the plaintiffs’ claims. Fouts v. Warren City Council, 97 F.4th 459, 464 (6th Cir. 2024); see also Ford v. Wilder, 469 F.3d 500, 504–05 (6th Cir. 2006) (“We have] no authority to render a decision upon moot questions” (quoting United States v. City of Detroit, 401 F.3d 448, 450 (6th Cir. 2005)). Accordingly, as the plaintiffs suggest, we must determine whether Smith moots any of the plaintiffs’ claims in this action before opining on their merits.
A question is mooted “when the issues presented are no longer ‘live’ or the parties lack a legally cognizable interest in the outcome.” Already, LLC v. Nike, Inc., 568 U.S. 85, 91 (2013) (quoting Murphy v. Hunt, 455 U.S. 478, 481 (1982) (per curiam)). We have recognized “that an intervening state supreme court decision on a particular issue may render an appeal to this court on the same issue moot.” Johnson v. City of Cincinnati, 310 F.3d 484, 490 (6th Cir. 2002) (citing WJW-TV, Inc. v. City of Cleveland, 878 F.2d 906, 910 (6th Cir. 1989) (per curiam)). In this instance, the intervening decision “arose after appellate briefing was completed and implicates unsettled facts and law that were not addressed by the district court below.” FemHealth USA, Inc. v. Williams, 83 F.4th 551, 557 (6th Cir. 2023). As MSP continues to adjust its practices in light of Smith, the facts underlying Counts I and II present “a moving target.” Maryville Baptist Church, Inc. v. Beshear, 977 F.3d 561, 566 (6th Cir. 2020) (per curiam) (remanding to the district court to consider mootness in the first instance); see D. 92-2 (Official Statement of MSP) (indicating that, in September, MSP “removed more than 20,000 offenders” from the registry but noting that “these numbers could change.”); D. 95-2 (MSP Legal Update No. 172 at 1) (indicating that registrants covered by Smith “will be permanently removed from Michigan's Sex Offender Registry”). Under these circumstances, “[t]he district court is best positioned to resolve ․ factual disputes,” and any later review by our court “will benefit from a more complete factual record.” FemHealth USA, 83 F.4th at 557–58. Accordingly, we vacate and remand Counts I and II for the district court to determine in the first instance whether Smith moots those claims.
We also note that the district court declined to address the merits of Count VI, the plaintiffs’ retroactive-alteration-of-plea-agreements due-process claim, finding it moot in light of the court's ruling on Counts I and II that SORA 2021 violates the Ex Post Facto Clause. Does III, 751 F. Supp. 3d at 814. The ex post facto ruling, the district court determined, would also prohibit the retroactive modification or nullification of plea agreements, the basis for Count VI. Id. Because we vacate and remand Counts I and II for the district court's consideration in light of Smith, we also vacate and remand Count VI.
In Smith, the Michigan Supreme Court determined only that SORA 2021 constituted an ex post facto violation under the federal and state constitutions when applied retroactively. 38 N.W.3d at 736. Smith, therefore, does not moot or otherwise impact the other four counts at issue on appeal.
C. Substantive-Due-Process and Equal-Protection Challenges to SORA 2021's Reliance on Offense-based Tiers
The plaintiffs next raise broad, amorphous Due Process and Equal Protection Clause challenges to SORA 2021. D. 67 (Pls.’ Corr. First Br. at 28–50). In essence, the plaintiffs argue that SORA 2021's tiering, reporting, and notice provisions are insufficiently related to recidivism reduction because the scheme relies on offense-based (not individualized) risk assessments. See id. “To qualify as having a rational basis, the challenged law need only be rationally related to legitimate government interests.” Fowler v. Benson, 924 F.3d 247, 262 (6th Cir. 2019) (citation modified). Rational-basis review is a “highly deferential” standard that statutes fail “only in rare or exceptional circumstances.” Doe v. Mich. Dep't of State Police, 490 F.3d 491, 501 (6th Cir. 2007).
The plaintiffs raise three buckets of empirical evidence to support their irrationality argument: (1) data showing that registrants with the longest reporting periods, tier II and III offenders, present a lower recidivism risk than those with the shortest, tier I offenders, R. 123-6 (Rep. on Class Data ¶¶ 99–103) (Page ID #3980–81); (2) data calling into question the wisdom of decades-long or lifelong registration periods and prohibiting tier II and III registrants from petitioning for removal, id. ¶¶ 7–8, 61–62 (Page ID #3952–53, 3967–68); R. 129-2 (Defs.’ SOMF Resp. ¶¶ 189–94) (Page ID #7321–24); and (3) evidence that offense-based public registration schemes like SORA 2021 not only fail to reduce recidivism, but may actually increase it, R. 123-9 (Letourneau Rep. ¶ 23) (Page ID #4240); R. 123-10 (Prescott Rep. ¶ 13) (Page ID #4282); R. 129-2 (Defs.’ SOMF Resp. ¶ 158) (Page ID #7302).
Even so, the district court found that SORA 2021 was rationally related to recidivism reduction. Does III, 751 F. Supp. 3d at 792–95. We agree. To start, the plaintiffs’ evidence to the contrary rests on the rates at which registrants are rearrested or reconvicted of sexual offenses—not the rates at which they reoffend. See R. 129-2 (Defs.’ SOMF Resp. ¶ 574) (Page ID #7553). Thus, the plaintiffs rely on data that “necessarily understates the actual number of offenses committed, both by registrants (who have been convicted of sexual offenses in the past) and by non-registrants (who have not).” Id. ¶ 577 (Page ID #7554–55). This is no concern, the plaintiffs claim, because although “[t]he extent of unreported sexual crime is unknown,” “the rate of detected and undetected sexual offending should be roughly the same for registrants and non-registrants.” Id. ¶¶ 578, 583 (Page ID #7555, 7557). To be sure, the plaintiffs present evidence suggesting that this might be true. See R. 123-8 (Hanson Rebuttal Rep. ¶¶ 38–41) (Page ID #4193–94). At the same time, however, “there remains considerable uncertainty concerning the rate of undetected offending.” Id. ¶ 42 (Page ID #4194). In light of this uncertainty, Michigan could have “rational[ly] speculat[ed]” that registrants, given their backgrounds, commit more unreported offenses than nonregistrants. Midkiff v. Adams Cnty. Reg'l Water Dist., 409 F.3d 758, 770 (6th Cir. 2005) (quoting Heller v. Doe ex rel. Doe, 509 U.S. 312, 320 (1993)). Michigan thus “rationally could have believed that [SORA 2021] would promote its objective” of protecting the public. W. & S. Life Ins. Co. v. State Bd. of Equalization of Cal., 451 U.S. 648, 672 (1981).
Moreover, the district court identified some evidence that offense-based public registration schemes, like SORA 2021, reduce recidivism. The district court noted that one of the plaintiffs’ experts “acknowledg[ed] that some studies do show some public benefit through reduction of recidivism.” Does III, 751 F. Supp. 3d at 793. That expert described a study analyzing Washington's registry, which found that “the policy was associated with a significant reduction in recidivism,” as well as a study concerning Minnesota's registry which “found lower recidivism among the high risk offenders subjected to registration, relative to what was predicted based on their risk scores.” R. 123-9 (Letourneau Rep. ¶ 7) (Page ID #4220, 4223). The plaintiffs’ expert's report therefore provides some support for the notion that SORA 2021 reduces recidivism. This is sufficient to surmount the low bar set by rational-basis review. See Williamson v. Lee Optical of Okla. Inc., 348 U.S. 483, 488 (1955) (“It is enough that there is an evil at hand for correction, and that it might be thought that the particular legislative measure was a rational way to correct it.”).
In response, the plaintiffs assert that the district court “misconstrued [their] expert's analysis after misreading a few divergent studies.” D. 67 (Pls.’ Corr. First Br. at 44). At bottom, the plaintiffs (and amici) take issue with the district court's reliance on studies evaluating Minnesota's and Washington's registries because those states utilize risk-based tiers and limit the scope of their online registries. Id. at 44–45; D. 62 (Ten Scholars’ Amici Curiae Br. at 27). Thus, they assert, those studies provide little insight into an offense-based public registry like SORA 2021. D. 67 (Pls.’ Corr. First Br. at 44–45); D. 62 (Ten Scholars’ Amici Curiae Br. at 27). But different states’ registration schemes inevitably vary from one another. The plaintiffs cannot show that these specific differences make studies of Minnesota's and Washington's registries inapplicable to Michigan's registry. The modicum of support from these studies for SORA 2021's recidivism-reducing effects supplies a rational link to a nonpunitive purpose. To be certain, this link may well be “tenuous”—but that satisfies rational basis review. Romer v. Evans, 517 U.S. 620, 632 (1996). The scholarly consensus that the plaintiffs identify may well establish that an offense-based public registry like SORA 2021's is unwise, even foolish. Yet rational-basis review “is not a license for courts to judge the wisdom, fairness or logic of legislative choices.” TriHealth, Inc. v. Bd. of Comm'rs, Hamilton Cnty., 430 F.3d 783, 791 (6th Cir. 2005) (quoting FCC v. Beach Commc'ns, Inc., 508 U.S. 307, 313–14 (1993)).
We therefore conclude that SORA 2021 is rationally related to a legitimate nonpunitive purpose. Consequently, the plaintiffs’ due-process and equal-protection challenges fail unless the plaintiffs identify some basis for us to apply a heightened standard of review. The plaintiffs, however, argue neither that SORA 2021 implicates a fundamental right nor that it engages in suspect classifications. We therefore apply rational-basis review and hold that SORA 2021 rationally relates to recidivism reduction. See Ohio Ass'n of Indep. Schs. v. Goff, 92 F.3d 419, 422 (6th Cir. 1996) (“If a fundamental right is not at stake, the statute is generally reviewed for a rational basis”); Herschfus v. City of Oak Park, 150 F.4th 489, 501 (6th Cir. 2025) (explaining that non-suspect “classifications must pass only rational-basis review”).
Attempting to stave off defeat, the plaintiffs and amici argue that even if rational-basis review governs, a more exacting form of rational-basis review applies here because SORA 2021 was motivated by animus. D. 67 (Pls.’ Corr. First. Br. at 29–34); D. 58 (Araiza, Janus, & Mayson Amici Br. at 4–28). The plaintiffs and amici claim that Bannum, Inc. v. City of Louisville, 958 F.2d 1354 (6th Cir. 1992), exemplifies our application of this elevated variant of rational-basis review. D. 67 (Pls.’ Corr. First. Br. at 29–33); D. 58 (Araiza, Janus, & Mayson Amici Br. at 4, 12–13).
But even assuming that our caselaw establishes a heightened form of rational-basis review, SORA 2021 meets it. This is not a case, as was Bannum, where “no evidence” connects the statute at issue to a legitimate aim. See 958 F.2d at 1360 (emphasis added). As discussed above, there is some evidence tying schemes like SORA 2021 to recidivism reduction. That is sufficient to establish that SORA 2021 was not motivated by “mere negative attitudes[ ] or fear.” Cf. City of Cleburne v. Cleburne Living Ctr., 473 U.S. 432, 448 (1985).
We accordingly affirm the district court's grant of summary judgment to the defendants on Count III.
D. Equal-Protection Challenge to SORA's Restrictions on Petitioning for Removal
The plaintiffs next assert that permitting tier I registrants—but not tier II or III registrants who remain offense-free—to petition for removal from the registry violates the Equal Protection Clause. D. 67 (Pls.’ Corr. First Br. at 51–59). SORA 2021 allows tier I registrants to petition for removal from the registry in state court, provided that they have been registered for at least ten years, abstained from reoffending, completed any required term of probation or parole, and finished a sex-offender treatment program. Mich. Comp. Laws § 28.728c(1), (12). Assuming other procedural prerequisites are met, the court must consider certain enumerated factors and deny the petition if it “determines that the individual is a continuing threat to the public.” Id. § 28.728c(11). Similarly, a tier III registrant who committed their registerable offense as a juvenile may petition for removal under the same conditions, except that they must wait at least twenty-five years to do so. Id. § 28.728c(2), (13). Finally, all registrants may petition for removal if they committed a limited subset of offenses against a victim between the ages of thirteen and sixteen and they were “not more than 4 years older than the victim.” See id. § 28.728c(3), (14).
The plaintiffs argue that members of the barred-from-petitioning subclass, which includes all “registrants who meet the eligibility criteria but cannot petition after ten years” as well as tier III registrants who committed their offenses as juveniles and must wait twenty-five years to petition, are indistinguishable from petition-eligible tier I registrants. D. 67 (Pls.’ Corr. First Br. at 53–54 & n.18). In support, they point to their evidence regarding the inverse relationship between tiers and registrants’ recidivism risks, plus their evidence showing that registrants’ recidivism risks approach the general public's offense risk within ten to twenty years. See id. at 54–56. This demonstrates, the argument goes, that generally barring the subclass from petitioning is irrational. Id. at 56–59.
This argument fails. The plaintiffs’ evidence focuses on the rates that registrants will be convicted of committing any offense. The plaintiffs, therefore, never address whether tier I, II, and III offenders typically reoffend by committing similarly severe offenses or whether, as compared to tier I offenders, tier II and III offenders tend to reoffend by committing more serious offenses. 7 Michigan could have rationally concluded that, even if tier I offenders’ and the subclass's probabilities of reoffending converge over time, the subclass nevertheless poses a greater risk to the public. As the district court put it, “the legislature could [have] reasonably conclude[d] that [those] who committed more serious crimes should be monitored for longer periods of time without the opportunity for removal” because “[m]aking a mistaken judgment about who is not dangerous carries potentially more severe consequences when made about a more serious offender.” Does III, 751 F. Supp. 3d at 813.
We further disagree with the plaintiffs’ assertion that Michigan treats juvenile registrants “worse than petition-eligible adults.” D. 67 (Pls.’ Corr. First Br. at 54 n.18). SORA 2021 requires juveniles to register only if they committed a tier III offense and were at least fourteen at the time. Mich. Comp. Laws § 28.722(a)(iii), (iv). Thus, Michigan treats juvenile registrants better than other tier III registrants by allowing them to petition for removal from the registry after twenty-five years while generally barring adult tier III registrants from ever doing so.
We thus affirm the district court's grant of summary judgment to the defendants on Count IV.
E. Procedural-Due-Process Challenge to non-Michigan Offense Registration
The plaintiffs argue that the MSP violates the procedural-due-process rights of individuals convicted of non-Michigan offenses because the MSP requires them to enroll in the registry and designates their tier without providing notice or an opportunity to be heard. 8 D. 74 (Pls.’ Corr. Third Br. at 46–52). This creates a risk, the plaintiffs claim, that individuals with non-Michigan convictions could be required to register as sex-offenders erroneously or be placed in the incorrect tier. See id.
SORA 2021 requires persons with non-Michigan sex-offense convictions who move to, work in, or study in Michigan to register if: (1) their convictions are “substantially similar” to a registerable offense under Michigan law, Mich. Comp. Laws § 28.722(r)(xi), (t)(xiii), (v)(ix); or (2) the state of conviction requires them to register “under a comparable statute,” id. § 28.723(1)(d). SORA 2021 does not designate who determines whether an individual's out-of-state conviction is “substantially similar” to a Michigan offense or whether other states’ registration schemes are “comparable” to Michigan's—nor does it define either term. The MSP, however, has assumed the responsibility of deciding whether individuals with non-Michigan convictions must register and, if so, at which tier. R. 129-2 (Defs.’ SOMF Resp. ¶ 536) (Page ID #7532–33).
The MSP has not promulgated any regulations governing this process. Id. Even so, it appears that the MSP follows an internal set of procedures. The MSP first decides whether an out-of-state conviction is substantially similar to a registerable offense. See id. ¶ 537 (Page ID #7533–34). To make that determination, the MSP's staff rely on a spreadsheet that lists over 2,200 out-of-state offenses and assigns roughly 1,400 of them tier designations. 9 Id. ¶ 540 (Page ID #7535). Of the 1,400 offenses deemed substantially similar to a Michigan offense, however, “[o]nly 43 ․ list a comparable Michigan offense and only 18 offense comparisons show as ‘approved by MSP legal.’ ” Id. ¶ 541 (Page ID #7535–36).
According to the defendants, the MSP does not list comparable registerable offenses for many out-of-state offenses “because some out-of-state offenses have language that requires review on a case-by-case basis.” Id. In particular, some out-of-state offenses are defined in statutes with “alternative elements,” so the MSP must often “examine the facts of the non-Michigan offense to determine the basis for the conviction.” Id. ¶ 545 (Page ID #7537–38); see also id. ¶ 556 (Page ID #7543–44) (explaining that the MSP “look[s] at pleadings, police reports, transcripts, and communications from the other state[’s] registering authority”). And even then, “a non-Michigan offense may be similar to several different Michigan offenses.” Id. ¶ 545 (Page ID #7537–38). For these reasons, the MSP's legal advisor admitted that “different people can reach different conclusions about whether a given non-Michigan offense is ‘substantially similar’ to a given Michigan offense.” Id.
Once the MSP decides whether an out-of-state offense is substantially similar, it then considers whether the state of conviction requires registration. See id. ¶¶ 537, 552 (Page ID #7533, 7540–41). The MSP also appears to rely on a database to make this determination, one listing other states’ registration requirements. See id. ¶ 555 (Page ID #7542–43). Yet the “MSP does not monitor other states’ registry laws, does not know when other states change their laws, and thus does not remove people with non-Michigan convictions who are no longer subject to registration in their convicting jurisdiction.” Id. If the MSP finds either that an out-of-state conviction is substantially similar to a registerable offense or that the convicting state requires registration, it then makes a tier designation. See id. ¶ 537 (Page ID #7533).
Throughout this process, the MSP does not provide non-Michiganders notice or an opportunity to be heard. When an out-of-state offense both is substantially similar to a registerable offense and requires registration in the state of conviction, the MSP imposes the registration period required by either SORA 2021 or the state of conviction, “whichever is longer.” Id. ¶ 552 (Page ID #7540–41). And even when there is no substantially similar registerable offense, the MSP requires that a person register in Michigan for the period imposed by the convicting state's registration laws. Id. ¶ 553 (Page ID #7541–42). At the end of this process, the MSP does not inform those with non-Michigan offenses how it determined their reporting obligations or which, if any, registerable offense it deemed substantially similar. Id. ¶ 561 (Page ID #7546–47). Such offenders do, however, “receive a registration verification update form with an explanation of duties to register as a sex offender, which informs [them] of their tier.” Id. Moreover, there exists no formal appeal process; non-Michiganders can ask the MSP to review their designations, but SORA 2021 does not specify how, if at all, the MSP conducts reviews. Id. ¶¶ 562–63 (Page ID #7547).
Mary Doe and John Doe G represent the non-Michigan offense subclass. R. 109 (5/9/2023 Order Grant'g Class Certification ¶ 2) (Page ID #3128). Both assert that their registration terms have run under the laws of their states of conviction. R. 129-2 (Defs.’ SOMF Resp. ¶¶ 564–70 (Page ID #7548–7551). Their registration terms under SORA 2021 persist, they assert, because the MSP seemingly misidentified the Michigan offenses most similar to their out-of-state convictions and incorrectly assigned them to tier III. Id.
We evaluate procedural-due-process claims using a two-step inquiry. We first consider whether the state has interfered with the plaintiffs’ liberty or property interests. Buchanan v. Apfel, 249 F.3d 485, 490 (6th Cir. 2001). If the state has done so, we evaluate whether the plaintiffs received the process that they were due. Id. “Due process is flexible and calls for such procedural protections as the particular situation demands.” Kaplan v. Univ. of Louisville, 10 F.4th 569, 578 (6th Cir. 2021) (quoting Gilbert v. Homar, 520 U.S. 924, 930 (1997)). To determine how much pre-deprivation process is required, we apply the balancing test established by Mathews v. Eldridge, 424 U.S. 319 (1976). Mathews requires courts to “balance three factors”: (1) “the private interest that will be affected by the official action”; (2) “the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards”; and (3) “the Government's interest.” Kaplan, 10 F.4th at 578 (quoting Gilbert, 520 U.S. at 931–32).
Although the defendants argued in their briefing that erroneously requiring an individual to register pursuant to SORA 2021 does not infringe a liberty or property interest, D. 70 (Defs.’ Second Br. at 42–46), they conceded at oral argument that the plaintiffs have “an interest in not being misclassified,” Oral Arg. at 10:27–10:35. We agree, and conclude that misclassification implicates liberty interests under the stigma-plus test. “[T]he stigma-plus test is used to analyze a due process claim where the action taken by the state injures the plaintiff's reputation.” Mich. Dep't of State Police, 490 F.3d at 501. “A successful plaintiff must therefore show that the state's action both damaged [their] reputation (the stigma) and that it ‘deprived [them] of a right previously held under state law’ (the plus).” Id. at 501–02 (quoting Paul v. Davis, 424 U.S. 693, 708 (1976)). Public sex-offender registration entails stigmatization and the imposition of tangible impairments, so we have held that it satisfies the stigma-plus test. See Hart v. Hillsdale County, 973 F.3d 627, 644–45 (6th Cir. 2020). Our sibling circuits agree. See, e.g., Meza v. Livingston, 607 F.3d 392, 403 (5th Cir. 2010) (recognizing a “liberty interest in being free from the stigma of registering as a sex offender”); Brown v. Montoya, 662 F.3d 1152, 1168 (10th Cir. 2011) (“[R]equiring a person to register as a sex offender triggers the protections of procedural due process.”); Kirby v. Siegelman, 195 F.3d 1285, 1291 (11th Cir. 1999) (per curiam) (holding that a prisoner has “a liberty interest in not being branded a sex offender”). Misclassification as a tier II or III offender (or a tier I offender included in the online registry) accordingly satisfies the stigma-plus test.
It is not obvious, however, whether tier I registrants excluded from the public registry suffer stigmatization. Nevertheless, an individual erroneously classified as a tier I registrant still suffers a deprivation of liberty and property. As described above, tier I registrants must comply with periodic and episodic in-person reporting requirements. Mich. Comp. Laws §§ 28.725a(3)(a), 28.725(1), (2); R. 126-17 (Mich. Sex Offender Registration /Verification/Update Form at 4) (Page ID #6207). As we recognized in Does I, these in-person reporting obligations constitute “direct restraints on personal conduct.” 834 F.3d at 703. This is sufficient to constitute the impairment of a liberty interest. Registration also requires the payment of initial and recurring fees, Mich. Comp. Laws § 28.725a(6), and money is a quintessential property interest, Bd. of Regents of State Colls. v. Roth, 408 U.S. 564, 571–72 (1972).
Next, the defendants do not argue that SORA 2021 furnishes the plaintiffs with notice or an opportunity to be heard. Instead, they first assert that the plaintiffs received all the process they are due when they were convicted, pointing to Connecticut Department of Public Safety v. Doe, 538 U.S. 1 (2003). D. 70 (Defs.’ Second Br. at 46). True, in Connecticut Department of Public Safety, the Court held that when registration “requirements turn on ․ offender[s’] conviction[s] alone,” offenders are not entitled to any further process before registration—their criminal proceedings already afforded them an adequate “opportunity to contest” their conviction. 538 U.S. at 7. But Connecticut Department of Public Safety is distinguishable. As described above, for individuals with non-Michigan convictions, their convictions alone do not determine their reporting requirements under SORA 2021.
The defendants next assert that the plaintiffs’ procedural-due-process rights are satisfied because after receiving an adverse determination from the MSP, they can file lawsuits challenging their registration obligations. D. 70 (Defs.’ Second Br. at 47–48). Granted, “[i]n some situations, ‘the failure to provide a pre-deprivation hearing does not violate due process’ if ‘the government provided adequate post-deprivation process.’ ” Johnson v. City of Saginaw, 980 F.3d 497, 510 (6th Cir. 2020) (quoting United Pet Supply, Inc. v. City of Chattanooga, 768 F.3d 464, 486 (6th Cir. 2014)). But this is true when “the deprivation at issue resulted from ‘a random and unauthorized act’ and not an ‘established state procedure,’ ” Daily Servs., LLC v. Valentino, 756 F.3d 893, 905 (6th Cir. 2014) (quoting Parratt v. Taylor, 451 U.S. 527, 541, 543 (1981)), or “in extraordinary situations where some valid governmental interest is at stake that justifies postponing the hearing until after the event,” United States v. James Daniel Good Real Prop., 510 U.S. 43, 53 (1993) (citation modified).
Here, nothing about the MSP's process is random or unauthorized, and the defendants do not identify any interest arising from an extraordinary situation that renders post-deprivation process adequate. Thus, it is far from clear why the availability of a post-deprivation lawsuit relieves Michigan from “its duty to provide some notice and some opportunity to be heard prior to final deprivation of a [liberty or] property interest.” Hicks v. Comm'r of Soc. Sec., 909 F.3d 786, 799 (6th Cir. 2018) (quoting Propert v. District of Columbia, 948 F.2d 1327, 1332 (D.C. Cir. 1991)). 10
Because the MSP's process runs up “against a constitutional bottom,” id., we agree with the district court that additional pre-deprivation procedures are required. We thus affirm the district court's grant of summary judgment to the plaintiffs on Count XI.
F. Void-for-Vagueness Challenges
Finally, the district court did not err in finding that SORA 2021's reporting requirements pertaining to education, employment, and vehicles are not facially vague. Does IV, 773 F. Supp. 3d at 394–400. “The due process clause[ ] of the ․ Fourteenth Amendment[ ] require[s] criminal statutes to provide notice to the accused of the nature and specific elements of the crime charged.” United States v. Salisbury, 983 F.2d 1369, 1377 (6th Cir. 1993). To succeed on a vagueness claim, a plaintiff “must show either that the ․ provisions (1) fail to provide people of ordinary intelligence a reasonable opportunity to understand what conduct it prohibits or (2) authorize or even encourage arbitrary and discriminatory enforcement.” Platt v. Bd. of Comm'rs on Grievances & Discipline of Ohio Sup. Ct., 894 F.3d 235, 246 (6th Cir. 2018) (citation modified).
The plaintiffs first attack Michigan Compiled Laws § 28.724a(1)(b). D. 67 (Pls.’ Corr. First Br. at 60–61). This provision requires in-person reporting within three days if:
As part of his or her course of studies at an institution of higher education in this state, [a registrant] is present at any other location in this state, another state, a territory or possession of the United States, or the [registrant] discontinues his or her studies at that location.
Mich. Comp. Laws § 28.724a(1)(b). According to the plaintiffs, the terms “part of [a registrant's] course of studies” and “present at any location” are vague. D. 67 (Pls.’ Corr. First Br. at 60–61). Yet “it is clear what the [statute] as a whole” requires—registrants must report every location that they travel to as students. Hill v. Colorado, 530 U.S. 703, 733 (2000) (quoting Grayned v. City of Rockford, 408 U.S. 104, 110 (1972)). Suffice it to say that the Michigan courts are competent to resolve some of the edge cases that the plaintiffs identify “if ․ ever confronted with [them] ․ on a complete factual picture.” Platt, 894 F.3d at 251.
Next, the plaintiffs challenge Michigan Compiled Laws §§ 28.725(1)(b), 28.727(1)(f). D. 67 (Pls.’ Corr. First Br. at 61–62). Section 28.725(1)(b) requires registrants to report when they “change[ ] [their] place of employment, or employment is discontinued.” And when registrants complete their periodic registration, they must report “[t]he name and address of each of [their] employers” pursuant to § 28.727(1)(f). “Employer” is then defined as “includ[ing] a contractor and any individual who has agreed to hire or contract with the [registrant] for his or her services.” Id. § 28.727(1)(f). The plaintiffs argue that the term “employer” is vague with respect to “de minimis paid labor,” like shoveling a neighbor's sidewalk in exchange for cash. D. 67 (Pls.’ Corr. First Br. at 61–62). This, they argue, then makes it unclear what constitutes a “place of employment”—if their neighbor no longer pays them to shovel snow, has their place of employment changed? Id. But again, the plaintiffs’ “clever hypotheticals” do not obscure what the section requires. Platt, 894 F.3d at 251. Registrants must report the names of those who procure their labor or services. And although the plaintiffs present an interesting edge case regarding “places of employment,” the phrase gives “people of ordinary intelligence a reasonable opportunity to understand what conduct [is] prohibit[ed]” and does not authorize arbitrary enforcement. Id. at 246 (quoting Hill, 530 U.S. at 732).
Finally, Michigan Compiled Laws § 28.727(j) requires registrants to report “[t]he license plate number and description of any vehicle” that they “owned or operated” when they complete their periodic registrations. The plaintiffs claim that the term “operated” is unclear because registrants do not know “whether they must report every car [they] drive[ ] even once.” D. 67 (Pls.’ Corr. First Br. at 62). As the district court found, however, “operated” “is a commonly used term with respect to motor vehicles.” Does IV, 773 F. Supp. 3d at 396. Indeed, it clearly requires registrants to report every car that they drive.
We thus affirm the district court's grant of summary judgment to the defendants on these aspects of Count VIII.
III. CONCLUSION
For the foregoing reasons, we AFFIRM IN PART, VACATE IN PART, and REMAND for further proceedings consistent with this opinion.
CONCURRENCE / DISSENT
I agree with the majority's analysis of the plaintiffs’ ex post facto, substantive-due process, equal-protection, and void-for-vagueness challenges, so I concur in those sections of the opinion. I would, however, send the procedural-due-process issue back to the district court for further proceedings, so I respectfully dissent from Part II.E of the majority opinion.
The plaintiffs assert that the Michigan State Police (MSP) violates the procedural-due-process rights of people convicted of non-Michigan offenses by requiring them to enroll in the 2021 version of Michigan's Sex Offenders Registration Act (SORA 2021) and deciding their tier without notice or an opportunity to be heard. Under the Fourteenth Amendment, states may not “deprive any person of life, liberty, or property, without due process of law.” U.S. Const. amend. XIV, § 1. We analyze procedural-due-process claims in two steps: First, we decide whether there is a constitutionally protected liberty or property interest at issue. Kaplan v. Univ. of Louisville, 10 F.4th 569, 577 (6th Cir. 2021). And second, we determine what procedures are needed to protect that interest. Id. To make the latter determination, we balance three factors:
“First, the private interest that will be affected by the official action; second, the risk of an erroneous deprivation of such interest through the procedures used, and the probable value, if any, of additional or substitute procedural safeguards; and finally, the Government's interest.” Gilbert v. Homar, 520 U.S. 924, 931–32 (1997) (quoting Mathews v. Eldridge, 424 U.S. 319, 335 (1976)).
Of note here, “[d]ue process is flexible and calls for such procedural protections as the particular situation demands.” Id. at 930. So the Supreme Court has “rejected the proposition that due process always requires the State to provide a hearing prior to the initial deprivation of property.” Kaplan, 10 F.4th at 578 (internal quotation marks omitted) (quoting Gilbert, 520 U.S. at 930). In cases “where a State must act quickly, or where it would be impractical to provide predeprivation process, postdeprivation process” will satisfy due process. Id.
The district court, having already decided that SORA 2021 constitutes punishment, concluded that the current framework for registering individuals with non-Michigan convictions is insufficient under the Due Process Clause. Does v. Whitmer (“Does III”), 751 F. Supp. 3d 761, 787, 819 (E.D. Mich. 2024). “Under the MSP's current procedure,” the court explained, non-Michigan offenders “have no notice or opportunity to be heard before the MSP decides whether an individual must register.” Id. But the court missed an important step in its analysis: It failed to identify what liberty interest, precisely, Michigan is depriving out-of-state offenders of—or when. Instead, presumably assuming some interest in not being classified within what it had just deemed a punitive registration system, the court skipped directly to an examination of MSP's existing procedures, and where they may be lacking. Id. at 818–20. But this puts the cart before the horse: A court determines whether the deprivation of an interest has violated procedural due process “[o]nly after identifying” that interest as “a protected liberty or property right under the Fourteenth Amendment.” Midkiff v. Adams County Regional Water District, 409 F.3d 758, 762–63 (6th Cir. 2005) (emphasis added). Yet here, the court identified no such interest at all.
The plaintiffs contend that those convicted of non-Michigan offenses have a liberty interest in “correct classification.” D. 74 (Pls.’ Corr. Third Br. at 47). Relying on Hart v. Hillsdale County, the majority accepts this contention, agreeing that “misclassification implicates liberty interests under the stigma-plus test.” Maj. Op. at 22. True, in Hart, we recognized that the listing of an individual as a sex offender on the public registry, even though he was no longer a covered “offender” under the 2011 amendments, passed the “stigma plus” test. 973 F.3d 627, 644 (6th Cir. 2020). But of course, there we confronted a SORA significantly different from its 2021 counterpart—one that included the student-safety zones that registrants found so debilitating in seeking housing and employment. See Does #1-5 v. Snyder (“Does I”), 834 F.3d 696, 701–02 (6th Cir. 2016). And it is on this aspect of the then-existing scheme that we focused when describing the “collateral consequences” of registration—the “plus” of “stigma plus.” Hart, 973 F.3d 627, 644.
But, as the majority recognizes elsewhere in its opinion, SORA is not what it used to be. See Maj. Op. at 7. Not only did SORA 2021 eliminate the student-safety zones of its earlier iterations, Mich. Comp. Laws §§ 28.733–28.735 (2006) (repealed 2021); it also removed registrants’ tier designations from public view in the online registry, id. § 28.728(3)(e), and allowed the MSP to remove in-person reporting requirements for registrants updating their names, addresses, places of work, schools, vehicles, emails, internet identifiers, or temporary residences, id. § 28.725(1)–(2). Yet despite these changes to the statutory scheme, the majority would have us accept, without further examination, that the stigma-plus test applies here precisely the same way it did in Hart.
Even assuming that SORA 2021 would still implicate stigma-plus for individuals misclassified as publicly registered sex offenders, additional questions arise when we consider other members of the non-Michigan offense subclass: Most immediately, does placement in the non-public registry still deprive non-Michigan offenders of a liberty interest? The plaintiff in Hart found himself placed incorrectly in SORA 2011's public database, and our analysis of the reputational harm he suffered focused on this “representation to the public.” 973 F.3d at 644–45. But this court has said nothing about the interest implicated by misclassification as a non-public sex offender. And it seems clear that individuals placed in tier I and excluded from SORA 2021's public registry would, at the very least, suffer reputational harm at a far lower level than those in tiers II or III—if they suffer any at all.
What about those individuals misclassified as tier II, when they should be tier III, or vice versa? Do they still have as strong an interest in correct classification, even though they would be on the public registry either way? There may be an argument that misclassification in a higher tier than one should be does implicate a cognizable interest when “erroneous labeling” is more “stigmatizing to one's reputation” than the correct label would be. See Schepers v. Commissioner, 691 F.3d 909, 914 (7th Cir. 2012). But again—does this argument really hold up under SORA 2021's scheme, which no longer publicizes a registrant's specific tier?
The majority acknowledges the first issue, at least. It concedes that “[i]t is not obvious ․ whether tier I registrants excluded from the public registry suffer stigmatization.” Maj. Op. at 23. But no matter, it says, because these registrants still must comply with in-person reporting requirements—a liberty interest—and pay registration fees—a “quintessential” property interest. Id. To be sure, each one of SORA 2021's tiers carries some not-insignificant obligations with it. But the plaintiffs’ procedural-due-process claim relies exclusively on the type of reputational interest described in Hart. D. 74 (Pls.’ Corr. Third Br. at 48–49). And besides, whatever interest individuals misplaced in a non-public tier—or placed in one public tier versus another—may have, it would seem to be quantifiably different than the interest of individuals incorrectly placed in a public-registry tier.
Without a proper assessment of these various interests—the first step of our procedural-due-process analysis—I do not see how we can move to step two. But that is what the district court did, concluding that “a judicial determination” should be afforded to every person with a non-Michigan offense before registration under SORA is required. Does III, 751 F. Supp. 3d at 819–820. 1 And it is what the majority seems content to do in its opinion, too.
But a court cannot properly apply the Mathews balancing test without a clear view of what it is that it's balancing. Put differently, it can't decide how to weigh the likely value of more procedural safeguards or the government's interest against the plaintiffs’ private interests until it knows what those interests are. Mathews, 424 U.S. at 335. To give just one example of why this matters: Michigan's public-safety interest in prompt registration of out-of-state offenders would surely hold more weight against an individual misplaced in tier II when he should be in tier III than a person who shouldn't be on the public registry at all. At the very least, it seems premature to conclude that every individual included in the non-Michigan offense subclass needs a full judicial determination before being required to register under SORA.
In short, “[w]ithout the benefit of” a more complete analysis by the district court on these issues, I feel “we are unable to decide in the first instance whether [the non-Michigan offenders] received the process [they were] due.” Hart, 973 F.3d at 645. And since “we are a court of review, not first view,” I would vacate and leave this issue to the district court and parties to consider on remand. Id. For that reason, I respectfully dissent from Part II.E of the majority opinion.
FOOTNOTES
1. Alexander William Furtaw, Note, Sex Offender Legislation Ex Post Facto: The History and Constitutionality of Michigan's Sex Offenders Registration Act, 48 J. Legis. 301, 303 (2022).
2. SORNA set an initial compliance deadline in 2009, see 34 U.S.C. § 20926(a)(1), but the Attorney General granted Michigan two extensions, pushing its deadline to 2011, see id. § 20926(b); U.S. Dep’t of Just., SORNA Extensions Granted (2010).
3. “The requirement to report any change in electronic mail addresses and internet identifiers applies only to an individual required to be registered under [SORA 2021] after July 1, 2011.” Mich. Comp. Laws § 28.725(2)(a).
4. Tier I offenses, among others, include: possession of child sexually abusive material, Mich. Comp. Laws § 750.145c(4); engaging the services of a person under eighteen for prostitution, id. § 750.449a(2); surveilling or photographing nude children, id. § 750.539j; and sexual contact with a corpse, id. § 750.160d(1). See id. § 28.722(r).Tier II offenses, among others, include: accosting, enticing or soliciting a child under sixteen for immoral purposes, id. § 750.145a; producing or financing child-sex abuse activity or material, id. § 750.145c(2); sodomy of a minor, id. § 750.158; and procuring or inducing a person to engage in prostitution, id. § 750.455. See id. § 28.722(t). Tier I offenders who go on to commit a subsequent Tier I offense are designated Tier II offenders. Id. § 28.722(s)(i).Tier III offenses, among others, include: first-degree criminal sexual conduct, id. § 750.520b; second-degree criminal sexual conduct against a minor under thirteen, id. § 750.520c; third-degree criminal sexual conduct, id. § 750.520d; and sexual penetration of a corpse, id. § 750.160d(2). See id. § 28.722(v). Tier II offenders who then commit a subsequent Tier I or II offense are designated Tier III offenders. Id. § 28.722(u)(i).
5. Another plaintiff named in the operative complaint, John Doe F, is not involved in this appeal. John Doe F had his underlying conviction expunged and was removed from the registry after this suit was filed. R. 156 (Expungement Notice at 1) (Page ID #8657). The district court, accordingly, dismissed him from the case in the final judgment, and he did not file a notice of appeal. See R. 177 (Am. Judgment at 14) (Page ID #9390); R. 181 (Pls.’ Notice of Appeal at 1) (Page ID #9405).Two named plaintiffs, moreover, have been—or appear to have been—removed from the registry. John Doe H was removed from the registry while this appeal was pending and is thus no longer subject to SORA 2021. See D. 84 (Pls.’ Notice Regarding John Doe H's Removal from the Mich. Registry at 1). He is one of several representatives of the primary class, as well as the sole representative of a subclass of individuals convicted of registerable offenses after 2011. R. 35 (5/18/2022 Order Grant'g Class Certification ¶¶ 2, 8) (Page ID #1116–18). The defendants, in addition, claimed below that John Doe A was removed from the registry in light of the Michigan Supreme Court's decision in People v. Lymon, 29 N.W.3d 58 (Mich. 2024). R. 129-2 (Defs.’ SOMF Resp. ¶¶ 3, 6, 8) (Page ID #7234–37). John Doe A is also a representative of the primary class and the sole representative of a subclass of registrants who have never committed an offense with “a sexual component.” R. 35 (5/18/2022 Order Grant'g Class Certification ¶¶ 2, 6) (Page ID #1116–18).The claims involving John Doe H's and A's subclasses are not at issue on appeal. And although they have (or may have) been removed from the registry, they may continue to serve as representatives of the primary class. The primary class was certified well before their claims were (or appear to have been) mooted. See id. ¶ 6 (Page ID #1116–17). “Once a class is certified, the mooting of the named plaintiff's claim does not moot the action, the court continues to have jurisdiction to hear the merits of the action if a controversy between any class member and the defendant exists.” Wilson v. Gordon, 822 F.3d 934, 942 (6th Cir. 2016) (citation modified). The controversy between the primary class and the defendants persists, so John Does H and A remain viable class representatives. See, e.g., id.; Binta B. ex rel. S.A. v. Gordon, 710 F.3d 608, 619 (6th Cir. 2013).
6. The district court below granted summary judgment for the plaintiffs on Counts I and II and required the defendants to “provide written notice to all members of the pre-2011 ex post facto subclass.” R. 177 (Am. Judgment at 3) (Page ID #9379).
7. The plaintiffs quibble with the notion that tier II and III offenders commit more serious offenses than tier I offenders. See, e.g., D. 67 (Pls.’ Corr. First Br. at 55–56). But taxonomizing offenses and deciding that some are more serious than others “involves ․ substantive penological judgment[s] that, as a general matter, [are] ‘properly within the province of legislatures, not courts.’ ” Harmelin v. Michigan, 501 U.S. 957, 998 (1991) (Kennedy, J., concurring in part and concurring in the judgment) (quoting Rummel v. Estelle, 445 U.S. 263, 275–76 (1980)).
8. Count XI also includes an equal-protection component. The district court found that the MSP's registration procedures for individuals with non-Michigan convictions violates the Equal Protection Clause because it “treats people with non-Michigan convictions more harshly than people with Michigan convictions.” Does III, 751 F. Supp. 3d at 820. Michigan does not challenge this aspect of the district court's ruling, so we do not address it.
9. When an out-of-state offense is not listed on this spreadsheet, the MSP staff tasked with making these determinations are instructed to consult an “analyst.” R. 129-2 (Defs.’ SOMF Resp. ¶ 548) (Page ID #7539). These analysts are not lawyers. Id. ¶ 549 (Page ID #7539). It appears that analysts used to “make tier determinations for ‘very simple’ offenses.” Id. Under current policy, however, the defendants assert that when MSP staff encounter a non-Michigan offense that has not yet been assigned a tier designation, they must consult with the MSP's legal department, which then assigns the non-Michigan offense to a tier. Id.
10. The defendants further contend that the Full Faith and Credit Clause requires Michigan to honor out-of-state convictions. D. 70 (Defs.’ Second Br. at 49–50). The district court, however, found that the defendants forfeited this argument by failing to brief it adequately. Does III, 751 F. Supp. 3d at 820 n.60. The defendants do not contest this finding on appeal. See D. 70 (Defs.’ Second Br. at 49–50). Thus, the defendants have abandoned their ability to challenge it. See Berry v. Experian Info. Sols., Inc., 115 F.4th 528, 540 (6th Cir. 2024).
1. To be fair, the district court's decision came after it had decided that the challenged provisions were akin to criminal penalties in its ex post facto analysis. If you take that as a given, then its procedural-due-process conclusion is less of a leap. But the court still failed to consider the distinctions between registrants and their respective interests that I raise here. In any case, we are vacating and remanding the plaintiffs’ Ex Post Facto Clause claims for potential mootness, which makes a remand for further consideration on the procedural-due-process issue all the more appropriate.
KAREN NELSON MOORE, Circuit Judge.
Thank you for your feedback!
As the largest network of trusted legal brands, we help firms build authority across the platforms consumers and AI systems rely on most. Our network helps attorneys strengthen visibility, credibility, and preference where legal decisions begin.
Docket No: Nos. 25-1413 /1414
Decided: November 13, 2026
Court: United States Court of Appeals, Sixth Circuit.
Search our directory by legal issue
Enter information in one or both fields (Required)
Harness the power of our directory with your own profile. Select the button below to sign up.
Learn more about FindLaw’s newsletters, including our terms of use and privacy policy.
Make It a Preferred Google Search Source
Add to GoogleGet help with your legal needs
FindLaw’s Learn About the Law features thousands of informational articles to help you understand your options. And if you’re ready to hire an attorney, find one in your area who can help.
Search our directory by legal issue
Enter information in one or both fields (Required)