RECOMMENDED FOR PUBLICATION
                                 Pursuant to Sixth Circuit I.O.P. 32.1(b)
                                        File Name: 25a0079p.06

                     UNITED STATES COURT OF APPEALS
                                   FOR THE SIXTH CIRCUIT



 UNITED STATES OF AMERICA,                                    ┐
                                     Plaintiff-Appellee,      │
                                                              │
                                                               >        No. 23-3466
       v.                                                     │
                                                              │
                                                              │
 TYREN L. CERVENAK,                                           │
                                  Defendant-Appellant.        │
                                                              ┘

                                On Petition for Rehearing En Banc
            United States District Court for the Northern District of Ohio at Cleveland.
                      No. 1:22-cr-00204-1—John R. Adams, District Judge.

                              Argued En Banc: December 11, 2024

                                Decided and Filed: April 2, 2025

   Before: SUTTON, Chief Judge; MOORE, CLAY, GILMAN, McKEAGUE, GRIFFIN,
   KETHLEDGE, STRANCH, THAPAR, BUSH, LARSEN, NALBANDIAN, READLER,
        MURPHY, DAVIS, MATHIS, BLOOMEKATZ, and RITZ, Circuit Judges.
                                _________________

                                             COUNSEL

ARGUED EN BANC: Catherine Adinaro Shusky, OFFICE OF THE FEDERAL PUBLIC
DEFENDER, Cleveland, Ohio, for Appellant.       Jason Manion, UNITED STATES
ATTORNEY’S OFFICE, Cleveland, Ohio, for Appellee. ON SUPPLEMENTAL BRIEF:
Catherine Adinaro Shusky, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Cleveland,
Ohio, for Appellant. Jason Manion, James A. Ewing, UNITED STATES ATTORNEY’S
OFFICE, Cleveland, Ohio, for Appellee.

       MATHIS, J., delivered the opinion of the court, in which SUTTON, C.J., and MOORE,
CLAY, GILMAN, KETHLEDGE, STRANCH, LARSEN, DAVIS, and BLOOMEKATZ, JJ.,
joined in full, and BUSH, NALBANDIAN, MURPHY, and RITZ, JJ., joined in part.
NALBANDIAN, J. (pp. 23–28), delivered a separate opinion concurring in all but Part IV.A.2
and footnote 2, in which BUSH and MURPHY, JJ., joined. RITZ, J. (pp. 29–36), delivered a
separate opinion concurring in part and dissenting in part, in which GRIFFIN and READLER,
 No. 23-3466                       United States v. Cervenak                             Page 2


JJ., joined in Parts I–IV. GRIFFIN, J. (pp. 37–40), delivered a separate dissenting opinion in
which THAPAR, J., joined. THAPAR, J. (pp. 41–84), delivered a separate dissenting opinion in
which McKEAGUE, GRIFFIN, and READLER, JJ., joined.
                                     _________________

OPINION

                                     _________________

       MATHIS, Circuit Judge. Congress instructed the Sentencing Commission to enhance the
Sentencing Guidelines range for individuals convicted of a crime of violence or a controlled
substance offense after having been previously convicted of two or more such offenses.

(h). The Commission created the career-offender guideline in response to
Congress’s directive.

       We must decide if the district court properly increased Tyren Cervenak’s Guidelines
range under the career-offender guideline. Specifically, we consider whether Cervenak’s prior
convictions for robbery under Ohio law are “crimes of violence,” as the Guidelines use that term.
Because we hold that they are not, we vacate Cervenak’s sentence.

                                               I.

       In 2022, a grand jury indicted Cervenak on two counts of distributing controlled
substances, see 

(a)(1), (b)(1)(C), and on one count of being a felon in possession
of a firearm and ammunition, see 

(g)(1), 924(a)(2) (2021). Cervenak pleaded
guilty as charged.

       Cervenak had multiple prior felony convictions.         One of the convictions was for
trafficking heroin. Cervenak also had two convictions for robbery, in violation of Ohio Revised
Code § 2911.02(A)(2).

       The district court enhanced Cervenak’s sentence under the career-offender guideline.
The district court held that the guideline applied because Cervenak had been convicted of a
controlled substance offense (the two distribution-of-controlled-substances charges) after having
been convicted of a prior controlled substance offense (trafficking heroin) and two prior crimes
of violence (robbery). Application of the enhancement resulted in a Guidelines range of 151 to
 No. 23-3466                             United States v. Cervenak                                    Page 3


188 months’ imprisonment.              The district court sentenced Cervenak to 188 months’
imprisonment.

        Cervenak timely appealed, arguing that the career-offender guideline should not apply
because his robbery convictions are not crimes of violence, as defined by the Guidelines.
Relying on United States v. Carter, 

 (6th Cir. 2023), a panel of this court disagreed
with Cervenak. United States v. Cervenak, 

, 854–63 (6th Cir. 2024). Judge Gilman
dissented. 

 at 868–82. Around the same time, another panel of this court considered a similar
question involving a prior conviction under Ohio’s aggravated-robbery statute, concluding that
the defendant’s aggravated-robbery conviction was not a crime of violence under the Guidelines.
United States v. Ivy, 

 (6th Cir. 2024). We voted to rehear this case en banc to
resolve this tension in our caselaw. United States v. Cervenak, 

 (6th Cir. 2024)
(en banc) (mem).

                                                      II.

        The Guidelines create a sentence enhancement for career offenders. U.S.S.G. § 4B1.1
(2021).1 A defendant is a career offender if he was an adult when he committed, and was
convicted of, a federal felony offense “that is either a crime of violence or a controlled substance
offense,” and he “has at least two prior felony convictions of either a crime of violence or a
controlled substance offense.” Id. § 4B1.1(a). The parties dispute only whether Cervenak has
two prior qualifying convictions. More specifically, Cervenak contends that the career-offender
guideline does not apply because his state-law robbery convictions should not count as crimes of
violence, leaving him with only one qualifying conviction.

        A prior conviction does not constitute a “crime of violence” just because the crime is
generally considered violent in common parlance. The Guidelines define the term “crime of
violence,” and the Supreme Court has instructions on how to apply it that we must follow. A
violation of a federal or state criminal statute is a crime of violence if the offense is “punishable
by imprisonment for a term exceeding one year” and it:

        1
          We use the 2021 version of the Guidelines because that version was in effect at the time of Cervenak’s
sentencing. U.S.S.G. § 1B1.11(a).
 No. 23-3466                          United States v. Cervenak                           Page 4


       (1) has as an element the use, attempted use, or threatened use of physical force
       against the person of another; or

       (2) is murder, voluntary manslaughter, kidnapping, aggravated assault, a forcible
       sex offense, robbery, arson, extortion, or the use or unlawful possession of a
       firearm described in 

(a) or explosive material as defined in
       

(c).

 § 4B1.2(a). We will refer to subpart (1) as the elements clause and subpart (2) as the
enumerated-offenses clause.

       The dispute centers on whether Cervenak’s robbery convictions are crimes of violence
under the enumerated-offenses clause. In particular, the government argues that Cervenak’s
robbery convictions are a categorical match for Guidelines extortion and robbery. All agree that
the robbery convictions are not crimes of violence under the elements clause. See United States
v. Butts, 

 (6th Cir. 2022).

       We review de novo “a district court’s determination that a prior conviction is a crime of
violence under [§] 4B1.1.” United States v. Ruvalcaba, 

 (6th Cir. 2010)
(internal quotation marks omitted).

                                                III.

       We use the categorical approach to determine if Cervenak’s prior conviction matches
Guidelines extortion or robbery. United States v. Burris, 

, 399–400 (6th Cir.
2019) (en banc). Under the categorical approach, we “compare the elements of the statute
forming the basis of the defendant’s conviction with the elements of the ‘generic’ crime.”
Descamps v. United States, 

 (2013).              We look first to the Guidelines
themselves for the generic definition. If the Guidelines do not define the offense, then we must
discern the generic definition based on common understanding, such as from a survey of state
law. See Taylor v. United States, 

 (1990); United States v. Camp, 

, 600–01 (6th Cir. 2018). A defendant’s prior conviction is a crime of violence under
the Guidelines “only if the statute’s elements are the same as, or narrower than, those of the
generic offense.” Descamps, 

. “[A] state crime cannot qualify as” a crime of
violence “if its elements are broader than those of [the] generic offense.” Mathis v. United
 No. 23-3466                          United States v. Cervenak                                Page 5


States, 

 (2016). In making the comparison between a defendant’s prior
conviction and the generic offense, we do not consider how the defendant committed the crime
that led to the prior conviction. United States v. Taylor, 

 (2022).

        How does the categorical approach work in this case? We compare the elements of the
Ohio robbery statute with the elements of extortion and robbery under the Guidelines. If a
hypothetical defendant can be convicted of Ohio robbery without satisfying the elements of
Guidelines extortion or robbery, then the prior conviction is not a crime of violence. See Ivy, 

. We “must presume that the [prior] conviction rested upon nothing more than the
least of the acts criminalized under the” Ohio robbery statute. See Mellouli v. Lynch, 

 (2015) (internal quotation marks omitted).

        In applying the categorical approach, we must be careful to distinguish between the
elements of an offense and the means—“the underlying brute facts,” Mathis, 

(internal quotation marks omitted)—of the offense.           The elements of an offense “are the
constituent parts of a crime’s legal definition—the things the prosecution must prove to sustain a
conviction.” 

 (internal quotation marks omitted). At trial, a jury must find each
element of an offense beyond a reasonable doubt to convict a defendant. 

 “Means, on the
other hand, are mere real-world things—extraneous to the crime’s legal requirements.”
Richardson v. United States, 

 (6th Cir. 2018) (internal quotation marks
omitted). “They are circumstances or events having no legal effect or consequence.”                  

(quotation omitted). A jury does not have to find the means of an offense. 

        When applying the categorical approach to a state statute, we can consider several
sources to differentiate elements from means. We “look first to state law, including judicial
interpretations of the criminal statute by state courts.” United States v. Ritchey, 

 (6th Cir. 2016). This includes the text of the criminal statute because “the statute itself may
provide the answer.” 

 Where, for instance, the statute provides alternatives and assigns
different punishments to different alternatives, each alternative defines a separate crime. Mathis,

. If, by contrast, “a statutory list is drafted to offer ‘illustrative examples,’ then it
includes only a crime’s means of commission.” 

 (citation omitted). Finally, if state courts and
 No. 23-3466                           United States v. Cervenak                            Page 6


the state statute do not provide a clear answer, we can consider “the record of a prior conviction
itself.” 

                                                  IV.

        We start by examining the Ohio robbery statute and Cervenak’s robbery convictions. We
then compare Cervenak’s convictions with Guidelines extortion and Guidelines robbery. That
analysis reveals that Cervenak’s robbery convictions are not crimes of violence under the
Guidelines.

        Cervenak pleaded guilty to robbery under Ohio Revised Code § 2911.02. That statute
provides, in pertinent part:

        (A) No person, in attempting or committing a theft offense or in fleeing
        immediately after the attempt or offense, shall do any of the following:
        (1) Have a deadly weapon on or about the offender’s person or under the
        offender’s control;
        (2) Inflict, attempt to inflict, or threaten to inflict physical harm on another;
        (3) Use or threaten the immediate use of force against another.
        ...
        (C) As used in this section:
        ...
        (2) “Theft offense” has the same meaning as in section 2913.01 of the Revised
        Code.

. “A violation of division (A)(1) or (2)” is a second-degree
felony. 

 § 2911.02(B). “A violation of division (A)(3)” is a third-degree felony. Id.

        The robbery statute directs us to Ohio Revised Code § 2913.01 for the definition of “theft
offense.” Section 2913.01 in turn defines theft offense as a violation of any one of more than
thirty Ohio statutes. Id. § 2913.01(K)(1).

        One of the theft offenses identified in § 2913.01 is § 2913.02—Ohio’s theft statute. The
theft statute provides: “No person, with purpose to deprive the owner of property or services,
 No. 23-3466                        United States v. Cervenak                              Page 7


shall knowingly obtain or exert control over either the property or services” of another in any one
of five specified ways. Id. § 2913.02(A).

       But the theft-offense definition covers many other crimes too. Those crimes include
breaking and entering (§ 2911.13), trespassing in a habitation when a person is present
(§ 2911.12), forgery (§ 2913.31), tampering with records (§ 2913.42), and personating an officer
(§ 2913.44), among others. Id. § 2913.01(K)(1).

       We consider the Ohio robbery statute a divisible statute. A divisible “statute sets out one
or more elements of the offense in the alternative.” Descamps, 

. Such statutes
“define multiple crimes.” Mathis, 

. When dealing with divisible statutes, we
apply the modified categorical approach. 

 Under the modified categorical approach, courts
can look “to a limited class of documents (for example, the indictment, jury instructions, or plea
agreement and colloquy) to determine what crime, with what elements, a defendant was
convicted of.” 

 at 505–06; see also Shepard v. United States, 

 (2005). We
refer to these documents as the Shepard documents.

       As we see it, the Ohio robbery statute is divisible twice over. Butts, 40 F.4th at 770–71;
cf. Ivy, 

. It is divisible first by the type of robbery at issue—§ 2911.02(A)(1),
(A)(2), or (A)(3). It is then divisible by the type of theft offense the defendant committed, or
attempted to commit, to complete the robbery. All agree on the first step of divisibility. And the
parties agree that the statute is twice divisible. But Judge Thapar sees things differently. He
contends that the type of theft offense underlying an Ohio robbery conviction is only a means of
committing robbery, not an element of the robbery offense. Ohio law suggests otherwise.

       Start with Ohio caselaw. In State v. Tolliver, the Ohio Supreme Court opined that the
Ohio robbery statute “predicates every robbery on the elements of a completed or an attempted
‘theft offense.’” 

 (Ohio 2014) (emphasis added). Because the Ohio Supreme
Court says that a robbery conviction requires proving the elements of the underlying theft
offense, that should conclude the elements-or-means analysis. See State v. Pigg, No. 04CA2947,

, at *5 (Ohio Ct. App. Apr. 28, 2005) (holding that the trial court did not err in
 No. 23-3466                             United States v. Cervenak                                    Page 8


its jury instructions on robbery because the instructions required the jury to find that the
defendant committed the theft offense of § 2913.02 theft).

        Ohio’s pattern jury instructions confirm Ohio caselaw.                Those instructions state, in
pertinent part:

        The defendant is charged with robbery. Before you can find the defendant guilty,
        you must find beyond a reasonable doubt that . . . the defendant, in (committing)
        (attempting to commit) (fleeing immediately after [committing] [attempting to
        commit]) the offense of (insert name of applicable theft offense under R.C.
        2913.01[K]) . . . .

        UNDERLYING THEFT OFFENSE. Before you can find that the defendant was
        (committing) (attempting to commit) the offense of (insert name of applicable
        theft offense under R.C. 2913.01[K]), you must find beyond a reasonable doubt
        that the defendant (describe each element of applicable theft offense).

2 CR Ohio Jury Instructions 511.02 (emphasis omitted). Although the pattern instructions are
not “binding legal authority,” they “are helpful as an example of the generally accepted
interpretation of state law.” State v. Smith, 

 (Ohio Ct. App. 2023) (internal
quotation marks omitted).

        Based on Ohio law, the State “must prove—and a jury must find—the elements of a
particular predicate theft offense” to convict a defendant of robbery. Cf. United States v.
Wilson, 

 (6th Cir. 2020).2                 The theft offense underlying the robbery
convictions is, therefore, an element of the crime. Judge Thapar’s suggestion to the contrary—
that a jury could convict a person of robbery based on a statutorily defined theft offense without
having to find each element of that offense beyond a reasonable doubt—would run afoul of the
Due Process Clause. See In re Winship, 

 (1970). We thus apply the modified
categorical approach to determine: (1) which type of robbery encompassed Cervenak’s crimes of
conviction, and (2) which theft offense served as the predicate for the robbery convictions. Cf.
Wilson, 

.



        2
           Judge Thapar argues that we should certify to the Ohio Supreme Court the question of whether the
individual theft offenses are elements or means of robbery. Because Ohio law has already answered this question,
certification is unnecessary.
 No. 23-3466                             United States v. Cervenak                                      Page 9


        The Shepard documents give us a partial answer.                     The indictment that preceded
Cervenak’s robbery convictions shows Cervenak committed robbery under § 2911.02(A)(2).
But the indictment says nothing about which theft offense served as the predicate for the robbery
convictions. That silence means that we must look to “the least serious conduct” criminalized by
§ 2911.02(A)(2) with any of the various theft offenses as a predicate to determine if Cervenak’s
robbery conviction was a crime of violence under the Guidelines. See Borden v. United States,

 (2021) (plurality opinion).

                                                       A.

        Are Cervenak’s robbery convictions broader than Guidelines extortion? On this record,
the answer is yes.

        Ohio robbery, without a specified predicate theft offense, does not categorically match
Guidelines extortion. The elements of Guidelines extortion are “[1] obtaining something of
value from another [2] by the wrongful use of (A) force, (B) fear of physical injury, or (C) threat
of physical injury.” U.S.S.G. § 4B1.2 cmt. n.1. Based on the plain language of the robbery
statute, which incorporates the numerous Ohio theft offenses, a defendant can be convicted of
robbery without obtaining something of value.

        Several theft offenses do not require obtaining something of value from another person.
The theft offense of breaking and entering criminalizes the “trespass on the land or premises of
another, with purpose to commit a felony.” 

(B). The theft
offense of trespass in a habitation prohibits the “trespass in a permanent or temporary habitation
of any person when any person other than an accomplice of the offender is present or likely to be
present.” 

 § 2911.12(B). And a person is guilty of the theft offense of forgery if, “with
purpose to defraud,”3 he “[f]orge[s] any writing of another without the other person’s authority.”
Id. § 2913.31(A)(1). These theft offenses do not require the culprit to obtain anything. See Ivy,
93 F.4th at 946–47; State v. Stone, No. 2023-CA-23, 

, at *1 (Ohio Ct. App. Jan.
19, 2024) (trespass in a habitation); State v. Dunn, Nos. 112301 & 112916, 

, at

        3
          “‘Defraud’ means to knowingly obtain, by deception, some benefit for oneself or another, or to knowingly
cause, by deception, some detriment to another.” 

(B).
 No. 23-3466                             United States v. Cervenak                                     Page 10


*6 (Ohio Ct. App. Dec. 7, 2023) (breaking and entering). Nor do the additional elements of
§ 2911.02(A)(2) that are necessary to turn the theft offense into a robbery require obtaining
property from another person.           Stated another way, a defendant could commit robbery by
“[i]nflict[ing], attempt[ing] to inflict, or threaten[ing] to inflict physical harm on another,” 

(A)(2), by breaking and entering, trespassing in a habitation, or
committing forgery without “obtaining something of value from another,” U.S.S.G. § 4B1.2 cmt.
n.1.

        The relevant documents from Cervenak’s prior convictions do not specify which crime
served as the predicate theft offense. If, as caselaw requires, we presume the conviction rested
on the least serious conduct criminalized by the statute—in other words, conduct that does not
require obtaining something of value—Cervenak’s conviction is not a categorical match for
Guidelines extortion. See Moncrieffe v. Holder, 

 (2013). We may rely only on
“what the state conviction necessarily involved” when analyzing whether there is a categorical
match. 

        The government and Judge Thapar disagree with this analysis and make several
arguments in response. The crux of the government’s argument is that when a defendant is
convicted of robbery under § 2911.02(A)(2) and the defendant’s indictment fails to specify the
theft offense, Ohio courts assume that § 2913.02 theft was the predicate theft offense. And,
relying on United States v. Carter, the government further contends that if § 2913.02 theft is the
predicate theft offense for all such robbery convictions in Ohio, then an Ohio robbery conviction
would categorically match Guidelines extortion. See 69 F.4th at 363–64; cf. United States v.
Rice, No. 23-3771, 

, at *6 (6th Cir. Aug. 22, 2024).4 Judge Thapar parrots the
government’s arguments, but he couches them in constitutional terms. In his view, Fifth and
Sixth Amendment violations occur if § 2913.02 theft does not serve as the predicate offense for a
defendant’s robbery conviction when the indictment does not indicate the specific theft offense.
We reject the government’s and Judge Thapar’s arguments.



        4
          The government’s argument in this case followed Carter, as did the panel’s opinion. Carter’s holding and
analysis are incompatible with our decision today.
 No. 23-3466                         United States v. Cervenak                             Page 11


                                                 1.

       First, the government argues that, under Ohio caselaw, every § 2911.02(A)(2) robbery
involves § 2913.02 theft or attempted theft. Thus, the government concludes, we must assume
that the predicate offense for Cervenak’s robbery convictions was § 2913.02 theft or attempted
theft. The government claims that the Ohio Supreme Court established this principle in State v.
Smith, 

 (Ohio 2008). But the government’s argument is misguided.

       In Smith, a grand jury indicted the defendant for robbery under § 2911.02(A)(3). 884
N.E.2d at 596. The indictment indicated that the predicate theft offense was “theft of retail
merchandise from Macy’s.” D. 46-2 (caps and bold removed). After a bench trial, the trial-court
judge acquitted the defendant of robbery but convicted her of theft. Smith, 884 N.E.2d at 596–
97. At issue before the Ohio Supreme Court was whether theft is a lesser-included offense of
robbery. Id. at 596. Smith held that it is. Specifically, Smith said that “theft is a lesser included
offense of robbery as statutorily defined in the alternative of robbery by theft, because it would
be impossible to ever commit a robbery by theft without also committing a theft.” Id. at 599
(emphasis added).

       Smith does not help the government for two reasons.             One, the Smith indictment
identified theft as the predicate theft offense; Cervenak’s indictment does not. Two, the Smith
court concluded only that theft is a lesser-included offense of robbery. But that does not make
theft the lesser-included offense of all robbery convictions. See, e.g., State v. Castle, 

, 817 (Ohio Ct. App. 2017) (holding that the theft offense of receiving stolen property, 

(A), is a lesser-included offense of robbery).

       Indeed, this is the only reading that accords with the text of Ohio’s robbery statute.
Under Ohio law, an offense is a lesser-included offense only if “the greater offense as statutorily
defined cannot be committed without the lesser offense as statutorily defined also being
committed.” State v. Owens, 

 (Ohio 2020) (quotation omitted). But the
examples described above show instances where a robbery conviction does not require “theft”—
i.e., obtaining another’s property in violation of § 2913.02(A). See supra Part IV.A. And when
 No. 23-3466                        United States v. Cervenak                            Page 12


the predicate theft offense is not specified, Smith does not require Ohio courts to presume that
§ 2913.02 theft is the predicate.

         The government then pivots to argue that Ohio cases have relied on Smith, explicitly and
implicitly, to establish a default that § 2913.02 theft is the predicate theft offense when a
defendant’s indictment is silent as to the predicate theft offense. But Smith “creates no default.”
See Moncrieffe, 

. Nor do the other cases the government cites establish such a
default. Although in those cases the indictment failed to specify which predicate theft offense
the defendant was charged with, the Ohio courts analyzed the trial testimony for sufficiency,
weight, or jury-instruction purposes under the assumption that the predicate offense was
§ 2913.02 theft. See e.g., State v. Gunn, No. L-20-1034, 

, at *6 (Ohio Ct.
App. June 30, 2021) (sufficiency); State v. Tenbrook, No. CA2020-01-005, 

,
at *2 (Ohio Ct. App. Nov. 9, 2020) (weight); State v. Villani, No. CA2018-04-080, 

, at *5 (Ohio Ct. App. May 13, 2019) (availability of jury instructions). The fact that
evidence sufficient to prove § 2913.02 theft satisfies the theft-offense element of a robbery
conviction does not mean that proving § 2913.02 theft is the only way to prove a theft offense.
We must conduct a fundamentally different analysis than the Ohio courts conducted in those
cases.    Unlike the Ohio courts, we cannot review the entire record to determine whether
§ 2913.02 could be an appropriate predicate based on the facts in the state-court record. Instead,
we must look to a much narrower class of documents and determine whether § 2913.02 theft
must be the predicate. See Moncrieffe, 569 U.S. at 190–91.

         None of the government’s cases limits the statutory definition of theft offense to just
§ 2913.02 theft. And, critically, many of the cases cited by the government acknowledge that
§ 2913.02 theft is just one type of theft offense. See, e.g., State v. Singleton, No. 28457, 

, at *2 (Ohio Ct. App. May 13, 2020) (“A ‘theft offense’ includes any of the
multiple enumerated offenses listed in R.C. 2913.01(K).”); State v. McCloud, No. 27825, 

, at *3 (Ohio Ct. App. Aug. 17, 2018) (“The applicable meaning of the term ‘theft
offense’ is provided by R.C. 2913.01(K)(1) and includes violations of R.C. 2913.02(A).”); State
v. Pullen, No. 25829, 

, at *4 (Ohio Ct. App. Feb. 13, 2015) (describing the
 No. 23-3466                        United States v. Cervenak                             Page 13


definition of “‘theft offense’ under the robbery statute” as “applying the definition in R.C.
2913.01(K)(1), which includes a violation of R.C. 2913.02”).

       Second, the government argues that Cervenak failed to show, through Ohio caselaw, that
a “realistic probability” exists that Ohio courts would apply the robbery statute to conduct
broader than Guidelines extortion. See Gonzales v. Duenas-Alvarez, 

 (2007).
But a defendant need not provide caselaw when they can point to the text of the state statute.
“We cannot ignore the statutory text.” Camp, 

 (quotation omitted). Thus, where,
as here, a statute’s meaning “is plain, the defendant need not provide a case to demonstrate a
realistic probability that the statute is broader than the generic offense.” 

 (internal quotation
marks omitted). The plain language of a criminal statute can demonstrate, like in this case, that a
state would apply it to a broader swath of conduct than the generic definition of a Guidelines
offense. On this point, we are in good company. See United States v. Proctor, 

,
552–53 (4th Cir. 2022); Aguirre-Zuniga v. Garland, 

, 450–51 (7th Cir. 2022);
Aspilaire v. U.S. Att’y Gen., 

 (11th Cir. 2021); Gonzalez v. Wilkinson, 

 (8th Cir. 2021); Lopez-Aguilar v. Barr, 

 (9th Cir. 2020); Zhi
Fei Liao v. Att’y Gen. U.S., 

 (3d Cir. 2018); Hylton v. Sessions, 

 (2d Cir. 2018); United States v. Titties, 

, 1274–75 (10th Cir. 2017); Swaby v.
Yates, 

 (1st Cir. 2017). But see Alexis v. Barr, 

 (5th Cir. 2020).

       Here, “the least serious conduct” criminalized by Ohio’s robbery statute, see Borden, 593
U.S. at 441—such as when a robbery conviction is predicated on the theft offenses of breaking
and entering, trespass in a habitation, or forgery—falls outside the Guidelines definition of
extortion. Subject to a contrary interpretation by Ohio courts, see Taylor, 

, the
plain text of the robbery statute thus obviates the need for caselaw to show a realistic probability
that Ohio criminalizes robbery beyond the Guidelines extortion definition.

       To hold otherwise would place defendants in the “odd[]” situation of having to procure
and “present empirical evidence about the [State’s] own prosecutorial habits.” 

. That
creates “practical challenges . . . in a world where most cases end in plea agreements.” 

 So,
“in applying the categorical approach, state law crimes should . . . be given their plain meaning.”
Swaby, 

.
 No. 23-3466                                United States v. Cervenak                                         Page 14


         The government’s reading of Duenas-Alvarez is unavailing. In United States v. Taylor,
the Supreme Court explained how the Duenas-Alvarez realistic-probability inquiry operates. 

. The Court noted that, “[i]n Duenas-Alvarez, there was no doubt that the relevant
state offense overlapped significantly with the” definition of the federal generic offense. 

 But
the defendant nonetheless argued that state courts interpreted the state offense to reach conduct
beyond the federal offense, in a manner not clearly contemplated by the statute’s text. 

Therefore, the defendant had to point to state caselaw to show a “realistic probability” that the
State would apply its statute as the defendant suggested. Id.; Duenas-Alvarez, 

.
The Court further observed, “the elements of the relevant state and federal offenses clearly
overlapped and the only question the Court faced was whether state courts also appl[ied] the
statute in [a] special (nongeneric) manner.” Taylor, 

 (alterations in original)
(internal quotation marks omitted). But when the text of the statute itself clearly reaches a
broader swath of conduct than the federal offense, “Congress tasked the courts with a much more
straightforward job: Look at the elements of the underlying crime and ask” whether they match.5

.

         Put differently, in Duenas-Alvarez, the Court required the defendant to point to state
caselaw because the dispute was about how far the elements of the state offense reached. The
Court did not look to caselaw to determine the elements themselves, as the government urges us
to do here. Although caselaw might provide clarity on the breadth of conduct that the statutory
elements proscribe, the statute itself defines the elements. See Descamps, 

. We
have followed these precedents, looking to the statutory text to determine the statutory elements,


         5
           Mellouli v. Lynch casts further doubt on the notion that Duenas-Alvarez requires a defendant to present
caselaw when the statutory text itself creates a realistic probability that a state would apply its statute to conduct not
covered by a Guidelines offense. In Mellouli, the Court considered whether a state drug-possession conviction made
a lawful permanent resident eligible for removal. The criminal charge and plea agreement did not identify which
controlled substance Mellouli pleaded guilty to possessing under Kansas law. 

. And Kansas’s drug
“schedules included at least nine substances not included in the federal lists.” 

. Applying the categorical
approach, the Court concluded that Mellouli’s conviction was not a categorical match because Kansas’s drug
schedules included drugs not listed in the federal counterpart. 

. The Court reached this conclusion without
citing Duenas-Alvarez, and the Court never considered whether Mellouli had presented caselaw showing that
Kansas had prosecuted defendants for possession of substances not listed under the federal drug schedule. But that
is exactly what you would expect the Court to have done, were Duenas-Alvarez to require a defendant to present a
case even when the statutory text clearly renders a state statute broader than its federal comparator. See Gonzalez,

 (making this point); Alexis, 

 (Graves, J., concurring) (same).
 No. 23-3466                       United States v. Cervenak                             Page 15


and then looking to caselaw to assess “the full range of conduct encompassed by those statutory
elements.” See Cradler v. United States, 

, 667–70 (6th Cir. 2018).

       As explained above, there are several instances in which the elements of Ohio’s robbery
statute plainly cover conduct that Guidelines extortion does not. Take it from Judge Thapar, who
tells us that Ohio’s robbery statute, on its “face[],” “criminalize[s] conduct that doesn’t
necessarily entail ordinary theft (taking something of value from another).” Thapar Dissent at
59. Cervenak thus does not ask us to find that Ohio applies its robbery statute in a “special
(nongeneric) manner.” Taylor, 

 (quotation omitted). He asks us to simply read
the statute’s plain text. He does not have to point to caselaw for that. And unless Ohio courts
tell us otherwise, we will give the text its plain meaning. Doing so does not require any
“application of legal imagination to [the] state statute’s language.” Duenas-Alvarez, 

.

       But even applying the government’s reading of Duenas-Alvarez, the result does not
change. In State v. Dixon, the Ohio Court of Appeals upheld a robbery conviction based on
forgery as the predicate theft offense when the indictment did not specify the predicate. No. 89-
CA-22, 

, at *2–3 (Ohio Ct. App. June 20, 1990). And in State v. Swarthout, the
same court upheld a robbery conviction predicated on attempted unauthorized use of property, a
theft offense under Ohio Revised Code §§ 2913.01(K)(1); 2913.04(A), where the indictment did
not name a predicate theft offense. No. 10CA0107-M, 

, at *3–4 (Ohio Ct.
App. Sept. 30, 2011). The elements for robbery convictions predicated on forgery or attempted
unauthorized use do not require a defendant to obtain something of value from another. This
further shows that Cervenak’s robbery conviction is broader than Guidelines extortion.

       Third, the government argues that the Ohio robbery statute substantially corresponds to
the Guidelines extortion definition even if the least criminalized acts under the statute do not
require a defendant to obtain something of value from another. But Guidelines extortion cannot
categorically match a state statute that does not require that a defendant obtain something of
value from another person, so this argument fails. Cf. Sanchez-Perez v. Garland, 

 (6th Cir. 2024).
 No. 23-3466                        United States v. Cervenak                             Page 16


                                                 2.

       Judge Thapar argues that a defendant’s Fifth and Sixth Amendment rights to receive fair
notice of the charges against him would be violated if § 2913.02 theft did not serve as the
predicate offense for a defendant’s robbery conviction when the indictment does not indicate the
specific theft offense. But time and again the Ohio Supreme Court has explained that “an
indictment is not defective as long as it tracks the language of the criminal statute describing the
offense, because that suffices to provide the defendant with adequate notice of the charges
against him.” State v. Whitaker, 

 (Ohio 2022) (alterations and internal
quotation marks omitted); see also State v. Horner, 

 (Ohio 2010). And if the
indictment tracks the statute, the defendant can remedy “the omission of the underlying [offense]
in the indictment” through a “bill of particulars.” State v. Skatzes, 

, 233 (Ohio
2004); cf. State v. Vaughn, No. 10CA009820, 

, at *2–3 (Ohio Ct. App. Jan. 31,
2011) (concluding that a charge of violating Ohio’s theft-in-office statute which, like robbery,
requires an underlying theft offense, did not require the indictment to name a particular theft
offense); State v. Scott, No. 22745, 

, at *8 (Ohio. Ct. App. Apr. 30, 2010)
(“The fact that the indictment did not name the [predicate] theft offense [to a charge of
aggravated robbery] did not render the indictment fatally defective.” (citation omitted)). Ohio
prosecutors must provide “the defendant with a bill of particulars setting up specifically the
nature of the offense charge and of the conduct of the defendant alleged to constitute the offense”
if the defendant so requests. Ohio R. Crim. P. 7(E).

       Thus, an indictment that tracks Ohio’s robbery statute provides a defendant with
constitutionally adequate notice, even if it does not name the predicate theft offense.         See
Williams v. Haviland, 

 (6th Cir. 2006) (“[I]n a state prosecution, due process
mandates only that the indictment provide the defendant with fair notice of the charges against
him.” (internal quotation marks omitted)). That is because a bill of particulars cures any possible
confusion as to the predicate theft offense, thus alleviating any constitutional concerns.
See Warner v. Zent, 

 (6th Cir. 1993) (holding that an “indictment, as fleshed
out . . . by a bill of particulars,” informed an Ohio defendant “of the ‘nature and cause’ of the
accusations made against him”).
 No. 23-3466                        United States v. Cervenak                             Page 17


       Ultimately, there is a critical difference between a due-process inquiry and a categorical-
approach inquiry. In evaluating a due-process challenge, a court must determine whether the
defendant had adequate notice of the charges against him. As described above, an indictment
that does not specify the predicate theft offense can still provide such adequate notice. By
contrast, in applying the modified categorical approach, the court must determine whether “the
jury necessarily found” the defendant guilty of a particular offense. See Descamps, 

. Presuming that a defendant’s Ohio robbery conviction was always predicated on § 2913.02
theft not only fails to meet this test, but would also “raise serious Sixth Amendment concerns”
by going beyond “what the jury in a prior trial must have accepted as the theory of the crime” or
“what the defendant and state judge must have understood as the factual basis of the prior plea.”
Id. at 269 (quotation omitted). And presuming that a jury convicted a defendant of a predicate
offense that was not specified to the jury at the time of conviction would raise its own due-
process and fair-notice concerns.

       Judge Thapar also argues that if we do not agree with his conclusion that Ohio law
presumes § 2913.02 theft is always the predicate theft offense when the indictment charging a
defendant with robbery does not specify the theft offense, then we should ask the Ohio Supreme
Court to clarify this point. We can certify questions to state high courts when a question of state
law is “novel[]” or “unsettle[d].” See Lehman Bros. v. Schein, 

 (1974).
However, we “will not trouble our sister state courts every time an arguably unsettled question of
state law comes across our desks.” Pennington v. State Farm Mut. Auto. Ins. Co., 

 (6th Cir. 2009) (quoting Pino v. United States, 

 (10th Cir. 2007)
(Gorsuch, J.)). Instead, “[w]hen we see a reasonably clear and principled course” to resolving
the question, we will answer it without certification. 

 (quoting Pino, 

).

       We do not need to certify a question to the Ohio Supreme Court in this case. Instead, the
current statutory text and relevant caselaw provide “sufficient guidance.” See 

 The plain text
of the statute is clear that Ohio robbery can be committed without obtaining something of value
from another person. We have a “reasonably clear and principled course” to resolving this case
without certification, see 

 and we can and should resolve it on the merits. And we do not
countenance an approach that would necessitate searching for purported trends in state cases and,
 No. 23-3466                          United States v. Cervenak                                 Page 18


in every case where a handful of decisions might suggest that the statute is not applied as broadly
as it plainly reaches, certifying the question to the state supreme court. The categorical approach
does not invite such a freewheeling inquiry that requires the analysis of state prosecutorial habits
that certification here would condone. See Taylor, 

.

                                                     B.

        Our final inquiry is whether Cervenak’s robbery convictions are broader than Guidelines
robbery. Again, the answer is yes.

        Ohio robbery does not categorically match Guidelines robbery just because those
offenses share the same name. Taylor, 495 U.S. at 588–89. We look only at the elements of the
offenses. 

 We discussed above the elements of Ohio robbery when a defendant’s charging
document fails to specify a particular theft offense. For Guidelines robbery, we use the generic
meaning of robbery because the 2021 version of the Guidelines does not define robbery. Camp,

. We have held previously that generic robbery involves: (1) the
“misappropriation of property under circumstances involving immediate danger to the person,”
United States v. Yates, 

 (6th Cir. 2017) (quotation omitted), (2) by a defendant
acting at least recklessly, Ivy, 

.6

        Ohio    robbery    does    not    always     require   the   misappropriation     of    property.
“‘[M]isappropriation’ is ordinarily understood as the unlawful taking or use of another person’s
property.” Sport Supply Grp., Inc. v. Columbia Cas. Co., 

, 461–62 (5th Cir. 2003);
see also Misappropriation, Black’s Law Dictionary (12th ed. 2024) (“The application of
another’s property or money dishonestly to one’s own use.”). Just as Ohio robbery without a
specified predicate offense does not require a defendant to obtain something of value from
another, it also does not require a defendant to unlawfully take or use the property of another.

        A few examples prove the point.               As discussed above, a defendant commits
§ 2911.02(A)(2) robbery by inflicting, attempting to inflict, or threatening to inflict physical
harm on another person while committing, or attempting to commit, a theft offense. And recall

        6
          Because we hold that Ohio robbery does not require the misappropriation of property, we need not
consider whether Ohio robbery requires a defendant to have a mental state of at least recklessness.
 No. 23-3466                        United States v. Cervenak                            Page 19


that we look to the least serious conduct. Suppose an offender’s robbery conviction is predicated
on the theft offense of trespassing in a habitation under Ohio Revised Code § 2911.12(B), which
provides: “No person, by force, stealth, or deception, shall trespass in a permanent or temporary
habitation of any person when any person . . . is present or likely to be present.” The offender
commits robbery the moment he inflicts, attempts to inflict, or threatens to inflict physical harm
on another person while in that person’s home. He does not have to take or use that person’s
property to satisfy the elements of Ohio robbery.

       Or suppose a robbery conviction is predicated on the theft offense of cheating. The
cheating statute provides, in pertinent part: “No person shall knowingly . . . [e]ngage in conduct
designed to corrupt the outcome of an athletic or sporting event.” Id. § 2915.05(B)(2). So if a
person threatens to inflict harm on someone else to “corrupt” a football game, he has committed
robbery. But he has not necessarily taken or used another person’s property.

       One last example—suppose a robbery conviction is predicated on the theft offense of
personating an officer. The personating-an-officer statute provides: “No person, with purpose to
defraud or knowing that he is facilitating a fraud, or with purpose to induce another to purchase
property or services, shall personate a law enforcement officer, or an inspector, investigator, or
agent of any governmental agency.” Id. § 2913.44(A). Again, even if an offender inflicts,
attempts to inflict, or threatens to inflict physical harm while personating a government official,
he does not necessarily unlawfully take or use the property of another. For instance, in State v.
Powell, the Ohio Court of Appeals upheld the conviction of a chiropractor for personating a
police officer. 

, 1171–72 (Ohio Ct. App. 1999). There, the chiropractor made
calls to “accident victims in order to solicit potential patients” by representing that he worked
with the police department. 

 at 1170–71. The court did not mention a requirement that the
defendant take or use another person’s property when sustaining the conviction. So even if the
defendant threatened physical harm while making the solicitations (which would turn the crime
into robbery), there would still be no misappropriation.

       Ohio robbery thus captures more conduct than generic robbery, meaning there is no
categorical match. Judge Ritz and Judge Thapar push back on this analysis.
 No. 23-3466                        United States v. Cervenak                           Page 20


                                                 1.

       Judge Ritz, although he does not concede that generic robbery involves the
misappropriation of property, argues that his definition of misappropriation would sweep in all
the theft offenses. This would make Ohio robbery a categorical match for Guidelines robbery.
Rather than use the ordinary meaning of misappropriation, Judge Ritz relies on the following
definition of “misappropriate”—“to wrongly make use of without authority or right.” Ritz
Dissent at 35.

       But even using Judge Ritz’s definition, not all the Ohio theft offenses involve
misappropriation. For example, cheating, see 

(B), does not
mandate the wrongful use of property. The same rationale applies to the personating-an-officer
theft offense, which also does not require an offender to wrongly use someone else’s property.
In State v. Powell, the defendant violated the personating-an-officer statute by “making
solicitation calls to accident victims” while “stat[ing] that he was a police officer.” 

.

       Judge Ritz also argues that Cervenak must identify cases where defendants were
convicted of robbery “for conduct extending beyond ‘misappropriation.’” Ritz Dissent at 35.
But as we explained in Part IV.A.1 above, because the plain language of the Ohio robbery statute
shows the statute applies to conduct broader than generic robbery, Cervenak did not need to
explore, or provide evidence of, the State’s prosecutorial habits.

                                                 2.

       Judge Thapar bases his argument that Ohio robbery is narrower than generic robbery on
two contentions: (1) the notion that the Ohio robbery statute is not twice divisible, and (2) his
own broad definition of generic robbery.
 No. 23-3466                               United States v. Cervenak                     Page 21


       First, as discussed above, Judge Thapar believes that the theft offenses are means, not
elements, of robbery. Therefore, he argues, the robbery statute has only one level of divisibility.
The plain text of the statute,7 Ohio caselaw,8 Ohio’s pattern jury instructions,9 our precedent,10
the government, Cervenak, and Judge Ritz all rebuff Judge Thapar’s argument. And because the
Ohio robbery statute is doubly divisible, Judge Thapar’s Ohio-robbery-is-Guidelines-robbery
argument falls apart.

       Second, Judge Thapar argues that many states have definitions that rely on “theft
offenses” that are distinct from ordinary theft. Thus, he asserts that this court’s conception of
generic robbery should encompass robbery predicated on a wide range of theft-adjacent offenses.
But neither this court nor any of our sister circuits have adopted his approach. Some courts
define generic robbery like we do, as “the misappropriation of property under circumstances
involving immediate danger to the person.” United States v. Fluker, 

 (4th Cir.
2018) (quotation omitted); United States v. Prigan, 

, 1120 (9th Cir. 2021) (same);
United States v. Montiel-Cortes, 

 (5th Cir. 2017) (per curiam) (same); United
States v. House, 

 (8th Cir. 2016) (same). Others define it as “the taking of
property from another person or from the immediate presence of another by force or by
intimidation.” United States v. Scott, 

 (3d Cir. 2021) (quotation omitted);
United States v. Eason, 

 (11th Cir. 2020) (same); United States v. Moore,

 (2d Cir. 2019) (same); United States v. Ball, 

 (1st Cir. 2017)
(same). The Seventh Circuit defines generic robbery as “the taking of property from a person by
force or the threat of force.” United States v. Carr, 

, 646 (7th Cir. 2024) (quotation
omitted). But none defines it like Judge Thapar.

       The overwhelming weight of circuit caselaw to the contrary notwithstanding, Judge
Thapar (and Judge Ritz) points to Ohio robbery’s similarity to other states’ robbery statutes,
insisting that we cannot define generic robbery in a way that excludes how many states define

       7
           

, 2913.01(K).
       8
           Tolliver, 

.
       9
           2 CR Ohio Jury Instructions 511.02.
       10
            Butts, 40 F.4th at 770–71; cf. Wilson, 978 F.3d at 997–98.
 No. 23-3466                         United States v. Cervenak                           Page 22


the offense. Although his premise is true, our decision here—like the decisions of several other
circuits that define generic robbery similarly—does not restrict the definition of generic robbery.
Rather, in the cases where the predicate offense is not specified, we presume the least culpable
conduct criminalized by the statute, which does not necessarily involve misappropriating
property. Where the Shepard documents specify one of the many predicate theft offenses that
involve the misappropriation of property, that offense will fit within the generic definition. Our
approach ensures that, before substantially enhancing a defendant’s sentence, the predicate
conviction is one for which the factfinder “necessarily found” the defendant guilty of a theft
offense captured by the generic definition of robbery. Descamps, 

. That is what
the categorical approach requires.

          Without law on his side, Judge Thapar resorts to scare tactics—he claims Americans will
suffer if we do not see things his way. In so doing, he “strike[s] a tone of chilling doom that is
wholly disproportionate to what” we do today. See Trump v. United States, 

(2024).

                                          *      *      *

          In sum, Cervenak’s robbery convictions are not “crimes of violence” under the
Guidelines.     The district court therefore erred in applying the career-offender guideline to
enhance Cervenak’s Guidelines range.

                                                V.

          For these reasons, we VACATE Cervenak’s sentence and REMAND for resentencing
consistent with this opinion. We also GRANT Cervenak’s and the government’s motions to
take judicial notice.
 No. 23-3466                           United States v. Cervenak                          Page 23


                               _____________________________

                                 CONCURRENCE / DISSENT
                               _____________________________

       NALBANDIAN, Circuit Judge, concurring in part and dissenting in part. I write briefly
to explain my position in this difficult case. For reasons that I’ll discuss below and as discussed
by Judge Thapar in dissent, I believe that we should have certified the question of whether an
Ohio robbery indictment that does not list a specific “theft offense” should be regarded as
charging theft under Ohio Revised Code § 2913.02(A).               But a majority of my colleagues
disagree. So I join all but Part IV(A)(2) and footnote two of the majority opinion because I agree
with its reading of Ohio law—such that it has the better view on applying the categorical
approach.

       We should have certified for two reasons.          First, although I don’t agree with the
government’s reading of State v. Smith, 

 (Ohio 2008), I believe Ohio law is
uncertain enough on this point to warrant certification. And second, like Judge Thapar in
dissent, I believe that there may be possible constitutional problems with a state-court indictment
that fails to specify a theft offense. Certification would allow the Ohio Supreme Court the
opportunity to address these issues.

       Start with Smith. We must figure out what state law criminalizes to apply the categorical
approach. As far as I’m concerned, we start with state criminal statutes, given the modern
uneasiness with common-law crimes. See United States v. Lanier, 

 (6th Cir.
1996). And here, I agree with the majority’s explanation and interpretation of Ohio Revised
Code § 2911.02. This statutory scheme might seem a little odd to us, but it’s Ohio’s prerogative
as a separate sovereign to define crimes as it wishes. See Commonwealth of Puerto Rico v.
Sanchez Valle, 

, 68–69 (2016).

       But we don’t necessarily stop there. As with the Rules of Decision Act and the Erie
doctrine, federal courts are often asked to answer state-law questions. See Smith v. Gen. Motors
LLC, 

, 878–79 (6th Cir. 2021). And state-court decisions play a pivotal role in this
process. See In re Dow Corning Corp., 

 (6th Cir. 2005). Because “state courts
 No. 23-3466                          United States v. Cervenak                         Page 24


are the ultimate expositors” of that state’s law, their interpretations can narrow or broaden the
meaning of state statutes. Mullaney v. Wilbur, 

 (1975). These interpretations
can be the difference between a categorical match or not. United States v. Taylor, 

,
858–59 (2022).

       In this context, the Supreme Court has warned against artificially broad interpretations of
state law that take a state statute from categorical match to mismatch. Moncrieffe v. Holder, 

 (2013). In those theoretical situations, the defendant must cite “at least . . . his
own case or other cases” that support the broad interpretation. Gonzales v. Duenas-Alvarez, 

 (2007). In other words, the broad interpretation must be “a realistic probability,
not a theoretical possibility.” 

       But I don’t think this is one of those cases.         We aren’t imagining a law-school
hypothetical that might fit within the broad language of the statute. Instead, Ohio has chosen to
include predicate “theft offenses” within its robbery statute that, naturally interpreted, do not
involve the taking of property from another. Ohio Rev. Code. § 2913.01(K). This comes from
the text of the statutes. For example, Ohio’s aggravated burglary and burglary statutes have
nothing to do with taking someone’s property. Id. §§ 2911.11, 2911.12. Nor would it seem,
does unlawfully using a telecommunications device or cheating at sports.          Id. §§ 2913.06,
2915.05. But the plain language of the robbery statute says you commit robbery if you do any of
these theft offenses while harming someone or threatening to harm someone. Id. § 2911.02.
And that’s apparent from the language of the statutes—not some unnatural interpretation of the
words or doctrinal convolution.

       So when an Ohio indictment charges someone with a robbery without specifying an
underlying theft offense, it could refer to a crime that did not involve taking anyone’s property.
And that’s the way the categorical approach works. No matter what we know about what the
defendant actually did—we compare the Guidelines offense with the “least culpable conduct”
criminalized under the state statute. United States v. Havis, 

 (6th Cir. 2019)
(en banc) (per curiam).
 No. 23-3466                               United States v. Cervenak                                       Page 25


         The government claims that the Ohio Supreme Court’s Smith decision takes care of this
problem by providing the needed narrowing construction of Ohio robbery in cases when the
exact theft offense is not specified in the indictment. It says that the Ohio Supreme Court has
held that an indictment with an unspecified theft offense should be read as charging
§ 2913.02(A) as the predicate. But I disagree.

         That wasn’t the issue in Smith, which addressed whether an individual may be convicted
of a lesser-included offense not specified in the indictment. 884 N.E.2d at 599. In doing so, the
Ohio Supreme Court needed to answer whether “the greater offense [could], as statutorily
defined, ever be committed without the lesser offense, as statutorily defined, also being
committed.” Id. (quoting State v. Deem, 

 (Ohio 1988)). But applying this
test to the robbery statute posed a problem because the theft offense element could “be proved
alternatively—as theft or attempted theft.” Id. at 598. The Ohio Supreme Court determined that
“when a statute sets forth mutually exclusive ways of committing the greater offense, a court is
required” to determine whether the greater offense requires committing the lesser offense for
“each alternative method of committing the greater offense.” Id. at 599.

         So the Ohio Supreme Court held that theft is “a lesser included offense of robbery.” Id.
But it did not hold that theft is the only lesser-included offense of robbery. See State v. Castle,

, 816–17 (Ohio Ct. App. 2017) (applying Smith to hold that receiving stolen
property under Ohio Revised Code § 2913.51(A) is a lesser included offense of robbery).
Indeed, the government’s theory that Smith held that robberies categorically involve completed
or attempted theft under § 2913.02(A) asks us to “read judicial opinions in a vacuum.” 1 Faber
v. Ciox Health, LLC, 

 (6th Cir. 2019). While Smith at times uses sweeping
language, “we don’t read precedents like statutes” and instead must evaluate the text of the
opinion based on the reasoning and legal question at issue. Herr v. U.S. Forest Serv., 

           Part of the government’s confusion comes from how the “lesser included offense” test applies to “a statute
written in the alternative, such as robbery.” Castle, 86 N.E.3d at 816. A statute written in the alternative “creates a
separate offense for each alternative.” Smith, 884 N.E.2d at 599 (quoting Pandelli v. United States, 

 (6th Cir. 1980)). In Smith, the Ohio Supreme Court did not hold that theft—as defined in § 2913.02(A)—will
always be the lesser included offense of any robbery under § 2911.02. Instead, it held that “it would be impossible
to ever commit a robbery by theft without also committing a theft.” Id. (emphasis added). Just like it would be
impossible to ever commit a robbery by receiving stolen property without also committing the crime of receiving
stolen property. Castle, 86 N.E.3d at 817. So statutory theft isn’t always the lesser included offense of robbery.
 No. 23-3466                          United States v. Cervenak                             Page 26


809, 819 (6th Cir. 2015); see also Loper Bright Enters. v. Raimondo, 

, 426 (2024)
(Gorsuch, J., concurring). Extending statements from the Ohio Supreme Court far beyond their
original scope does little to safeguard the interests of federalism. And when deciding matters of
state law, we should tread lightly.

       So why certify?       Even though the best reading of the statute does not track the
government’s view, I believe that there’s enough uncertainty (both because of Smith and other
cases) to warrant certification. It appears to me that Ohio robbery prosecutions without a
specified theft offense might always involve the taking of property—even though no Ohio court
has held that the robbery statute always requires such a taking. The government points to several
Ohio cases that seem to presume that theft, as defined in § 2913.02(A), is the typical predicate
“theft offense.” See, e.g., State v. Gunn, No. L-20-1034, 

, at *6 (Ohio Ct.
App. June 30, 2021). But those same courts have noted that § 2913.01(K)’s definition of theft
offense “includes any” of the “enumerated offenses,” even those without a taking. State v.
Singleton, No. 28457, 

, at *2 (Ohio Ct. App. May 13, 2020). When the state
court precedents are unclear, “discretion, to say nothing of respect for a co-equal sovereign, is
the better part of valor.” See Carey v. Wolnitzek, 

 (6th Cir. 2010). Given the
unsettled nature of Ohio caselaw on this issue, I’d certify.

       Additionally, I think that not having a specified, predicate theft offense in a criminal
indictment—without more—might create constitutional concerns. See Russell v. United States,

 (1962). After all, where is the fair notice to the defendant of what he is
accused of doing? Where is the “statement of the facts and circumstances” that “inform the
accused of the specific offense, coming under the general description, with which he is
charged”? 

 (quoting United States v. Hess, 

 (1888)). True, in Smith, the to-
wit clause of the indictment told the defendant that she was likely accused of theft, as defined by
§ 2913.02(A). But Cervenak’s indictment isn’t as clear. This may or may not be a realistic
concern, but letting the Ohio Supreme Court weigh in could remove any lingering constitutional
doubt. So also for that reason, I would support certification.

Finally, as an aside, I will make the perhaps obvious point that district judges retain the ability to
impose an appropriate sentence. In other words, whether a past conviction constitutes a “crime
 No. 23-3466                        United States v. Cervenak                           Page 27


of violence” only impacts a defendant’s career-offender status for calculating the Guidelines
range. U.S.S.G. § 4B1.1(a). But the range is purely “advisory.” United States v. Hensley, 

 (6th Cir. 2024) (quoting United States v. Booker, 

 (2005)).
While the sentencing court “must include the Guidelines range in the array of factors warranting
consideration,” its recommendation is not binding. United States v. Thomas-Mathews, 

, 541–42 (6th Cir. 2023) (quoting Kimbrough v. United States, 

 (2007)). And
the sentencing court has both the prerogative and responsibility to impose an appropriate
sentence that is “sufficient, but not greater than necessary” to account for the factors Congress
set forth in 

(a). The advisory-Guidelines range does nothing to limit that
discretion. For this reason, a court “may vary from the Guidelines [range] for policy reasons
alone.” United States v. Allen, 

 (6th Cir. 2024).

       Going forward, today’s holding instructs sentencing courts in this circuit that Ohio’s
robbery statute is not a categorical match for Guidelines extortion or generic robbery—as
relevant to the version of the Guidelines in effect at the time of Cervenak’s federal conviction.
So sentencing courts must follow this decision in determining whether a defendant is a career
offender under the Guidelines. But today’s holding only addresses the judge’s calculation of the
advisory-Guidelines range. Once that has been properly determined, the district judge still has
discretion to impose a sentence outside that range based on everything considered during
sentencing.

       In Cervenak’s case, the district court’s discretion was only limited by the language of the
statutes. Given the quantity of drugs in his possession, the maximum term of incarceration for
each of Cervenak’s two federal drug-trafficking convictions was twenty years.          

(a)(1), (b)(1)(C). And the maximum for his federal firearm conviction was ten years. 

(g)(1), 924(a)(2). It is these statutes, and not the Guidelines range, that cabin the
possible term of imprisonment.

       At resentencing, the district court must apply our holding and so find that Cervenak is not
a career offender under § 4B1.1(a) of the Guidelines. This conclusion will impact both the
offense level and the final Guidelines range. But it is here that guidance from our holding runs
out. Once the advisory range has been recalculated, the district judge must still conduct an
 No. 23-3466                       United States v. Cervenak                            Page 28


“individualized assessment based on the facts presented” and considering the relevant § 3553(a)
factors. United States v. Bolds, 

 (6th Cir. 2007) (quoting United States v. Gall,

 (2007)). Of which the Guidelines range is one among many (albeit an important
one). 18 U.S.C § 3553(a)(4)(A).

       So although we hold that Cervenak’s Ohio robbery conviction does not qualify him for
the career-offender enhancement under the Guidelines, that does not mean that his many prior
convictions will have no relevance for the other § 3553(a) factors. Indeed, if the sentencing
court believes that Cervenak’s unknown underlying theft offense was a regular theft, the court
could reject the categorical approach and apply something akin to the career-offender
enhancement as a matter of its residual discretion.      The growing list of judges who have
criticized the categorical approach—as highlighted by Judge Griffin in his dissent—shows that
such a policy-based variance would be eminently reasonable.
 No. 23-3466                        United States v. Cervenak                            Page 29


                               _____________________________

                                 CONCURRENCE / DISSENT
                               _____________________________

       RITZ, Circuit Judge, concurring in part and dissenting in part. I agree with the majority
that Ohio’s robbery statute is twice divisible—first by the three subsections of Ohio Rev. Code
§ 2911.02, and then by the underlying theft offenses cross-referenced in Ohio Rev. Code
§ 2913.01(K). See United States v. Butts, 

 (6th Cir. 2022). I also agree that
in cases such as Cervenak’s, where the state-court documents do not specify an underlying theft
offense, we must assume that any of the underlying theft offenses could have been the predicate
crime. See Moncreiffe v. Holder, 

 (2013); United States v. Burris, 

 (6th Cir. 2019) (en banc). I further agree with the majority that not all of Ohio’s
predicate theft offenses require a defendant to “obtain[] something of value from another,” so
Cervenak’s robbery conviction cannot qualify as extortion under the guidelines. See U.S.S.G. §
4B1.2. And finally, I agree that there is no need to certify any question to the Ohio Supreme
Court. See Pennington v. State Farm Mut. Auto. Ins. Co., 

 (6th Cir. 2009)
(citation omitted).

       But for me, there is a straightforward way to resolve this case. We should hold, as the
district court found, that Cervenak’s robbery conviction qualifies as robbery under the guidelines
enumerated-offenses clause.

I.     Defining generic robbery

       To determine whether a prior conviction qualifies as a crime of violence under the
enumerated-offenses clause, we “compare the elements of the offense in question with the
elements of the ‘generic’ enumerated offense to determine whether the two are a categorical
match.” United States v. Camp, 

 (6th Cir. 2018) (citing Descamps v. United
States, 

 (2013)). If the conduct criminalized by the offense in question is
broader than the conduct covered by the generic definition, the offense is not a crime of violence.
United States v. Yates, 

 (6th Cir. 2017).
 No. 23-3466                             United States v. Cervenak                                    Page 30


        A key step, then, is to define the generic enumerated offense. At the time of Cervenak’s
sentencing, the guidelines listed “robbery” as an enumerated offense but did not define it. See
U.S.S.G. § 4B1.2(a)(2) (2022).1 Relying on Supreme Court decisions addressing the Armed
Career Criminal Act (ACCA), 

(e), we have held that to determine the generic
definition of an enumerated offense not otherwise defined in the guidelines, we “survey[] how
the crime is described across jurisdictions, as well as consult[] sources such as the Model Penal
Code.” United States v. Rede-Mendez, 

 (6th Cir. 2012) (citing Taylor v.
United States, 

, 598 & n.8 (1990)); cf. Camp, 

 (citing Descamps,

).

        Applying this approach, we have defined generic robbery as the “misappropriation of
property under circumstances involving immediate danger to the person.” Yates, 

. That definition, which marries a force element to a theft element, derives from a 2006 Fifth
Circuit opinion, which itself relied on a criminal-law treatise and a survey of state robbery
statutes. See United States v. Santiesteban-Hernandez, 

 (5th Cir. 2006) (citing
Wayne R. LaFave, Substantive Criminal Law § 20.3 intro., (d)(2) (2d ed. 2003)).

        As we and the Fifth Circuit have observed, states have implemented the force element of
the generic robbery definition in two main ways.                 Yates, 

; Santiesteban-
Hernandez, 

. The majority of states require property to be taken “by means of
force or putting in fear,” while a minority of states and the Model Penal Code define the
immediate-danger requirement “in terms of ‘bodily injury’ or ‘committing violence’ or ‘physical
harm.’” Yates, 

 (quoting Santiesteban-Hernandez, 

; United States
v. Lockley, 

 (11th Cir. 2011)). The latter is a “narrower” definition, in that
the minority “jurisdictions require something more than mere intimidation or undifferentiated
fear; the defendant must directly threaten the victim’s bodily integrity.” Lockley, 

. In any event, “[t]he difference in the majority and minority definitions is slight,” and
we have declined to adopt one reading over the other. Yates, 

 (quoting Lockley,

        1
          A definition of robbery was added to the career-offender guideline in November 2023. See U.S.S.G.
§ 4B1.2(e)(3) (2024). The Sentencing Commission made this change to clarify that federal Hobbs Act robbery
convictions are crimes of violence, and in response to court of appeals decisions holding otherwise. See U.S.S.G.
app. C, amt. 822; see also 

(b)(1).
 No. 23-3466                           United States v. Cervenak                                 Page 31

); see Camp, 

; see also Lockley, 

 (explaining
that “the fear or intimidation contemplated by the majority definition in nearly all instances is the
fear of bodily harm”).

        Many circuits have adopted the same definition of generic robbery we adopted from the
Fifth Circuit. See, e.g., United States v. Gattis, 

 (4th Cir. 2017); United
States v. House, 

 (8th Cir. 2016); United States v. Becerril-Lopez, 

 (9th Cir. 2008). At least two circuits have adopted the “majority approach” and define
generic robbery as “the taking of property from another person or from the immediate presence
of another person by force or intimidation.” Lockley, 

; United States v. Walker,

 (2d Cir. 2010).

        The bottom line is that courts, including this one, have defined generic robbery to include
two primary components: a theft element and a force element.

II.     Mens rea

        Recently, and for the first time, we added a mens rea element to the definition of generic
robbery. United States v. Ivy, 

 (6th Cir. 2024). In Ivy, we stated that the force
or fear element used by the majority of states “suggests a mens rea of purposefulness.” 

(quoting United States v. Beard, No. 22-3398, 

, at *3 (6th Cir. May 22,
2023)). We also stated that the bodily-injury element used by a minority of states and the Model
Penal Code “suggest[s] a mens rea of recklessness.” 

 (quoting Beard, 

, at
*3). Thus, we concluded, “generic robbery has a mens rea of at least recklessness as to its force
requirement.” 

.

        I am not convinced that generic robbery requires a mens rea of recklessness as to the
force element. First, many states’ robbery statutes do not specify a mens rea for this element.2
See, e.g., 

; Miss. Stat. § 97-3-73; Wash. Stat. § 9A.56.190; see also, e.g.,

; Colo. Stat. § 18-4-301; Ill Stat. 5/18-1(a); 

(providing a mens rea requirement for one element of robbery, but not the force or physical harm

        2
         The new guidelines robbery definition also does not identify a mens rea requirement. See U.S.S.G. §
4B1.2(e)(3) (2024).
 No. 23-3466                         United States v. Cervenak                             Page 32


element). Some of these states require a recklessness mens rea via operation of statute. See, e.g.,

(b); 720 Ill. Comp. Stat. Ann. 5/4-3. Crucially, however, many do not.
See 

 (defining robbery as a common law offense); State v.
Wilkerson, 

 (W. Va. 2013); People v. Anderson, 

 (Cal. 2011);
State v. Day, 

 (R.I. 2007); State v. Wright, 

, 796 (Mo. Ct.
App. 2020); State v. Moulin, 

 (Kan. Ct. App. 2019) (table); State v. Lewis, 

 (Wash. App. 2010).

       The lack of a mens rea requirement in these states does not mean that defendants can
commit robbery accidentally.      They must still commit the underlying theft offense with a
culpable state of mind. See, e.g., Anderson, 

; People v. Miller, 

 (N.Y. 1995); see also State v. Tolliver, 

 (Ohio 2014) (noting
that Ohio robbery “always includes the mens rea of the predicate theft offense”). This approach
of requiring a felonious intent for the underlying theft offense but not for the force element traces
back to the common law. See 2 William Oldnall Russell, A Treatise on Crimes and Indictable
Misdemeanors, 64 (2d ed. 1828); Wm. L. Clark & Wm. L. Marshall, A Treatise on the Law of
Crimes 559 (Herschel Bouton Lazell ed., 2d ed. 1905); 4 William Blackstone, Commentaries on
the Laws of England 239, 242 (1769).

       It would be odd to require recklessness for the force element of generic robbery when so
many states decline to do so. In an analogous context, when assessing whether particular state
robbery convictions qualified as “violent felonies” under the ACCA, the Supreme Court
cautioned against construing the federal sentencing enhancement “in a way that would render it
inapplicable in many States.” Stokeling v. United States, 

 (2019). The same
principle applies here.

       Second, and relatedly, our pre-Ivy precedent defining generic robbery did not mention a
recklessness requirement. See Johnson v. United States, 

, 722 (6th Cir. 2023);
Camp, 

; Yates, 

. Ivy discerned such a requirement in Yates’s
summary of states’ definitional approaches to robbery. But, as noted above, just because the
majority of states define the force requirement as using “means of force or putting in fear” does
not mean that those states require that conduct to occur intentionally or purposefully. And just
 No. 23-3466                        United States v. Cervenak                            Page 33


because a minority of states and the Model Penal Code define the force requirement “in terms of
‘bodily injury’ or ‘committing violence’ or ‘physical harm’” does not mean that those states
require a mens rea of recklessness as to that element.

       Third, Ivy’s holding as to recklessness is inconsistent with the Fifth Circuit decision that
we used to define generic robbery in the first place. The Fifth Circuit has confirmed that its
precedent does not require a particular mens rea as to the force element of generic robbery.
United States v. Adair, 

 (5th Cir. 2021) (citing Santiesteban-Hernandez, 

). The Ninth Circuit has likewise concluded there is no mens rea requirement for
the force element. United States v. Flores-Mejia, 

 (9th Cir. 2012). And
no other court of appeals has imposed a mens rea requirement like ours.

       For these reasons, I would not add a recklessness requirement to the force element of
generic robbery. I would instead adhere to our pre-Ivy case law on that point.

III.   Comparing Ohio robbery to generic robbery

       The next question is whether an Ohio robbery conviction like Cervenak’s qualifies as
generic robbery. For all the debate in this case over how to apply the categorical approach,
everyone agrees that at the first level of divisibility, Cervenak was convicted under subsection
(A)(2) of the Ohio robbery statute. That provision makes it a crime to “[i]nflict, attempt to
inflict, or threaten to inflict physical harm on another” while “attempting or committing a theft
offense or in fleeing immediately after the attempt or offense.”                 Ohio Rev. Code
§ 2911.02(A)(2). At the second level of divisibility, as the majority opinion explains, we assume
that any of the underlying theft offenses listed in Ohio Rev. Code § 2911.02(K) could have
served as the predicate for Cervenak’s conviction.

A.     Force element

       Remember, though, that whatever the underlying theft offense in Cervenak’s case was,
Cervenak necessarily “inflicted, attempted to inflict, or threatened to inflict physical harm on
another” person while committing or attempting the offense. Ohio defines “physical harm to
persons” as “any injury, illness, or other physiological impairment, regardless of its gravity or
 No. 23-3466                         United States v. Cervenak                             Page 34


duration.” Id. § 2901.01(A)(3). This physical-harm requirement is enough to satisfy the force
element of generic robbery.      See, e.g., State v. Brown, 

, 167 (Ohio 2024)
(affirming that holding a third party at gunpoint threatens physical harm for the purpose of
(A)(2)). In fact, Ohio requires even more culpability as to the force element than the majority of
states do. See Lockley, 

. And since we should not require any particular mens
rea for this element, it does not matter whether Ohio requires recklessness. Compare Tolliver, 

 (holding that the force element of Ohio robbery does not require a culpable
state of mind); with State v. Cunningham, No. 21AP-470, 

, at *13-15 (Ohio
Ct. App. May 28, 2024) (holding that recklessness is required, due to a post-Tolliver change in
Ohio statutory law).

B.     Theft element

       That leaves the crux of the issue, as I see it: whether any of the potential underlying theft
offenses in the Ohio robbery statute extend further than the theft component of generic robbery.
Again, we have defined the theft component as “misappropriation of property”—a definition we
borrowed from another circuit. It is not clear to me that this is the right or only way to define the
theft element, especially considering (as noted below) those are not the words states tend to use
when defining robbery.      But even if we use “misappropriation” as the relevant definition,
Cervenak’s conviction qualifies as a crime of violence.

       The Ohio statute defines “theft offense” to encompass dozens of crimes. Yates, 

; Ohio Rev. Code § 2913.01(K)(1). Some of these predicate offenses fit squarely in the
category of “misappropriation.” See, e.g., Ohio Rev. Code § 2913.02 (theft); Ohio Rev. Code §
2913.03 (unauthorized use of a vehicle). Others, though, are further afield. For example, the
statute lists trespass, passing bad checks, and altering liquor control labels as potential predicate
theft offenses. See, e.g., Ohio Rev. Code § 2911.13 (breaking and entering); Ohio Rev. Code §
2913.11 (passing bad checks); Ohio Rev. Code § 2913.31 (forgery); Ohio Rev. Code § 2913.32
(criminal simulation, including fraudulently altering liquor control labels and deceptively
retouching audiovisual material). Whether these qualify as “misappropriation” is certainly a
closer call than whether “theft” qualifies. Of course, one interpretive challenge we face is that
we are not applying a textual definition of “misappropriation.” Rather, we are interpreting and
 No. 23-3466                         United States v. Cervenak                              Page 35


applying a term set out in the case law, and the case law provides little guidance on what the
term means.

       But it is not hard to conceive of a definition of misappropriation that sweeps in all of
Ohio’s predicate theft offenses. See Misappropriation, Black’s Law Dictionary (12th ed. 2024)
(“The application of another’s property or money dishonestly to one’s own use”); see also
Merriam-Webster’s Collegiate Dictionary (11th ed. 2019) (defining “misappropriate” as “to
appropriate wrongly,” and defining “appropriate” as “to take or make use of without authority or
right”). If misappropriation is “to wrongly make use of without authority or right,” then offenses
like altering liquor control labels or deceptively retouching audiovisual equipment would meet
the definition, as both involve using property beyond one’s scope of authority. The same is true
for forgery, which depends on making a false claim of authority over a writing, or breaking and
entering, which involves making wrongful use of another’s dwelling. See Ohio Rev. Code §
2911.13; Ohio Rev. Code § 2913.31.

       Of course, we could try to hypothesize scenarios in which a defendant could commit an
(A)(2) Ohio robbery in a way that extends further than generic robbery. Could he, for instance,
tamper with records, or pass a bad check, or deceptively retouch audiovisual material in a way
that would not constitute “misappropriation”—while inflicting, attempting, or threatening
physical harm at the same time? It seems unlikely. But regardless, the Supreme Court has
steered us away from using “legal imagination” in this fashion. Gonzales v. Duenas-Alvarez,

 (2007). To show overbreadth a defendant must show “a realistic probability,
not a theoretical possibility, that a State would apply its statute to conduct that falls outside” the
relevant federal definition. E.g., Moncrieffe, 

. And to do so, he “must at least
point to his own case or other cases in which the state courts in fact did apply” the statute to
conduct that is outside the federal definition. Duenas-Alvarez, 

; see also Cradler
v. United States, 

 (6th Cir. 2018) (holding that courts must look to state case
law to assess the full range of conduct encompassed by criminal statutory elements). Here,
neither Cervenak nor the majority has identified any examples of convictions under subsection
(A)(2) for conduct extending beyond “misappropriation.” Cf. State v. Dixon, No. 89-CA-22,

, at *2-3 (Ohio Ct. App. June 20, 1990) (upholding a robbery conviction under
 No. 23-3466                         United States v. Cervenak                             Page 36


subsection (A)(3) of the Ohio robbery statute where forgery was the predicate theft offense).
That is not just significant but, in my view, determinative.

        As a final observation about the scope of Ohio robbery’s theft element, it bears noting
that many states, like Ohio, have robbery statutes that define the theft element to encompass a
wide range of conduct. See, e.g., Tex. Penal Code § 29.02 (defining a predicate theft offense as
“Chapter 31 Theft,” which lists twenty related theft offenses, including advertising devices to
unscramble encrypted signals, id. § 31.03, and communicating a trade secret, id. § 31.05(b)(3));
Mich. Penal Code § 750.530 (defining robbery as “in the course of committing a larceny,” which
includes offenses like Mich. Penal Code § 750.530, “injuring a fruit tree”). Whether and how
those statutes are divisible, and whether the various permutations of those statutes qualify as
generic robbery under the majority’s analysis, are issues beyond the scope of this case.

        But the fact that Ohio’s robbery statute is not an outlier should give us pause in holding
that Ohio physical-harm robbery is not robbery under the guidelines. After all, the definition of a
guidelines enumerated offense must “be capacious enough to cover the states’ varied
formulations of those elements.” United States v. Brown, 

 (6th Cir. 2020)
(citing Quarles v. United States, 

 (2019)); see also 

 (generic
definition must give states “flexibility” and “‘breathing space’” in defining their state offenses);
Stokeling, 

; United States v. Castleman, 

 (2014). Many of
our sister circuits have held that robbery statutes that define the theft element in ways similar to
Ohio qualify as generic robbery. United States v. Harrison, 

 (11th Cir. 2023)
(Georgia); Adair, 16 F.4th at 470–71 (Texas); United States v. McCants, 

 (3d
Cir. 2020) (New Jersey); United States v. Stovall, 

 (8th Cir. 2019) (Arkansas).
We should do the same.

IV.     Conclusion

        For these reasons, I would hold that Ohio physical-harm robbery is “robbery” and thus a
crime of violence under the guidelines. As the majority sees it otherwise, I respectfully dissent
on this point.
 No. 23-3466                        United States v. Cervenak                            Page 37


                                      _________________

DISSENT

                                      _________________

       GRIFFIN, Circuit Judge, dissenting.

       I join Judge Thapar’s dissent and Sections I through IV of Judge Ritz’s dissent. I write to
add my voice to the overwhelming list of jurists urging the Supreme Court to discard the absurd,
convoluted, and nonsensical categorical approach.

       When defendant Tyren Cervenak pulled a gun from his waistband and ordered two
people to “give them everything they had,” he violently committed two counts of felonious
armed robbery in violation of Ohio law. Yet the majority’s application of the categorical
approach to the Guidelines’ career-offender provision renders yet another result divorced from
common sense—today we hold that Cervenak’s pointing of a gun at his victims does not qualify
as a “crime of violence” under U.S.S.G. § 4B1.2(a). Although a higher sentencing guideline is
mandated for defendant Cervenak because of his prior crimes of violence, today’s decision tells
his victims (and the public at large) that Cervenak did not commit crimes of violence because a
hypothetical criminal could commit a hypothetical crime that does not fit a generic definition of
robbery or extortion. In Judge Thapar’s words in his separate writing for the panel, this result is
“a testament to the categorical approach’s well-documented absurdity.”           United States v.
Cervenak, 

 (6th Cir. 2024) (Thapar, J., concurring); see also Bains v. Garland,

, at *5 (6th Cir. Aug. 27, 2024) (Thapar, J., concurring) (collecting cases).

       Judge Thapar and I are in good company in decrying the categorical approach. See, e.g.,
United States v. Taylor, 

, 860–73 (2022) (Thomas, J., dissenting); Mathis v. United
States, 

, 520–21 (2016) (Kennedy, J., concurring); 

 at 536–44 (Alito, J.,
dissenting). In this regard, I concur in this excellent citation from Judge Michael Park of the
United States Court of Appeals for the Second Circuit:

       As a growing number of judges across the country have explained, the categorical
       approach perverts the will of Congress, leads to inconsistent results, wastes
       judicial resources, and undermines confidence in the administration of justice.
       See, e.g., Mathis v. United States, — U.S. —, 

, 195 L.Ed.2d
No. 23-3466                       United States v. Cervenak                             Page 38


     604 (2016) (Kennedy, J., concurring) (“[T]oday’s decision is a stark illustration of
     the arbitrary and inequitable results produced by applying an elements based
     approach to this sentencing scheme.”); 

 at 2268–69 (Alito, J., dissenting) (“The
     Court’s approach calls for sentencing judges to delve into pointless abstract
     questions. . . . A real-world approach would avoid the mess that today’s decision
     will produce.”); Lopez-Aguilar v. Barr, 

 (9th Cir. 2020)
     (Graber, J., concurring) (“I write separately to add my voice to the substantial
     chorus of federal judges pleading for the Supreme Court or Congress to rescue us
     from the morass of the categorical approach. The categorical approach requires
     us to perform absurd legal gymnastics, and it produces absurd results.” (citations
     omitted)); United States v. Battle, 

, 163 n.2 (4th Cir. 2019)
     (Quattlebaum, J.) (“Through the Alice in Wonderland path known as the
     ‘categorical approach,’ we must consider whether Battle’s assault of a person with
     the intent to murder is a crime of violence. While the answer to that question
     might seem to be obviously yes, it is not so simple after almost 30 years of
     jurisprudence beginning with Taylor.”); United States v. Escalante, 

,
     406–07 (5th Cir. 2019) (Elrod, J.) (“In the nearly three decades since its inception,
     the categorical approach has developed a reputation for crushing common sense
     in any area of the law in which its tentacles find an inroad. . . . Perhaps one day
     the Supreme Court will consider revisiting the categorical approach and setting
     the federal judiciary down a doctrinal path that is easier to navigate and more
     likely to arrive at the jurisprudential destinations that a plain reading of our
     criminal statutes would suggest.” (footnotes omitted)); United States v. Burris,
     

 (6th Cir. 2019) (en banc) (Thapar, J., concurring) (“A casual
     reader of today’s decision might struggle to understand why we are even debating
     if ramming a vehicle into a police officer is a crime of violence. The reader’s
     struggle would be understandable. The time has come to dispose of the long-
     baffling categorical approach.”); United States v. Douglas, 

 (1st
     Cir. 2018) (Lynch, J.) (“On the whole, it is at least as practical to allow a jury to
     parse carefully between crimes based on specific real-world conduct rather than,
     under a categorical approach, to force judges to be willfully blind to particular
     facts and thus to go down the rabbit hole to a realm where we must close our eyes
     as judges to what we know as men and women.” (cleaned up)); United States v.
     Williams, 

 (3d Cir. 2018) (Roth, J., concurring) (“I write
     separately because of my concern that the categorical approach, along with its
     offspring, the modified categorical approach, is pushing us into a catechism of
     inquiry that renders these approaches ludicrous.”); Cradler v. United States, 

 (6th Cir. 2018) (Kethledge, J., concurring) (“Whatever the merits of
     [the categorical] approach, accuracy and judicial efficiency are not among them
     . . . .”); United States v. Brown, 

 (9th Cir. 2018) (Owens, J.,
     concurring) (“All good things must come to an end. But apparently bad legal
     doctrine can last forever, despite countless judges and justices urging an end to
     the so-called Taylor categorical approach.”); Ovalles v. United States, 

 (11th Cir. 2018) (en banc) (Pryor, J., concurring) (“How did we ever
     reach the point where this Court, sitting en banc, must debate whether a
 No. 23-3466                        United States v. Cervenak                             Page 39


       carjacking in which an assailant struck a 13-year-old girl in the mouth with a
       baseball bat and a cohort fired an AK-47 at her family is a crime of violence? It’s
       nuts. And Congress needs to act to end this ongoing judicial charade.”); United
       States v. Chapman, 

 (3d Cir. 2017) (Jordan, J., concurring)
       (“Forcing judges to close their eyes to what is obvious promotes inefficiency and
       guarantees difficult-to-explain sentences.”); United States v. Valdivia-Flores, 

 (9th Cir. 2017) (O’Scannlain, J., specially concurring) (“I write
       separately to highlight how [this case] illustrates the bizarre and arbitrary effects
       of the ever-spreading categorical approach for comparing state law offenses to
       federal criminal definitions.”); United States v. Faulls, 

 (4th Cir.
       2016) (Shedd, J., concurring) (“[T]he categorical approach is the antithesis of
       individualized sentencing; we do not consider what the individual to be sentenced
       has actually done, but the most lenient conduct punished by his statute of
       conviction.”); United States v. Doctor, 

 (4th Cir. 2016)
       (Wilkinson, J., concurring) (“[T]he categorical approach can serve as a protracted
       ruse for paradoxically finding even the worst and most violent offenses not to
       constitute crimes of violence.”); United States v. Aguila-Montes de Oca, 

 (9th Cir. 2011) (en banc) (Bybee, J.) (“In the twenty years since Taylor,
       we have struggled to understand the contours of the Supreme Court’s [categorical
       approach] framework. Indeed, over the past decade, perhaps no other area of the
       law has demanded more of our resources.”).

United States v. Scott, 

, 126–27 (2d Cir. 2021) (en banc) (Park, J., concurring).

       Further, since Judge Park’s compilation, many more distinguished jurists have added
their voices to our chorus. See United States v. DeFrance, 

, 824–26 (9th Cir. 2024)
(Rakoff, J., concurring); Alejos-Perez v. Garland, 

, 809–12 (5th Cir. 2024) (Oldham,
J., concurring in the judgment); United States v. Harris, 

, 459–80 (3d Cir. 2023)
(Jordan, J., concurring in the denial of rehearing en banc) (order); United States v. Lopez, 

 (11th Cir. 2023) (Grant, J., concurring); Alfred v. Garland, 

,
1066–67 (9th Cir. 2023) (en banc) (McKeown, J., dissenting); 

 at 1072 n.1 (VanDyke, J.,
dissenting); United States v. Jenkins, 

 (3d Cir. 2023); United States v. Morris,

, 317–18 n.9 (2d Cir. 2023); United States v. Turner, 

, 518–19 (7th Cir.
2022); United States v. Rice, 

, 579, 587 (4th Cir. 2022); United States v. Castillo, 

 (2d Cir. 2022); Valdez v. Garland, 

, 85–86 (9th Cir. 2022) (Graber, J.,
concurring in part and dissenting in part); United States v. Scott, 

, 200–02 (3d Cir.
2021) (Phipps, J., dissenting); Alfred v. Garland, 

, 987–89 (9th Cir. 2021) (England,
J., concurring).
 No. 23-3466                       United States v. Cervenak                           Page 40


       Although some common sense may be on the horizon, see Sentencing Guidelines for
United States Courts, 

-01, 128 (proposed Jan. 2, 2025) (proposing the
elimination of “the use of the categorical and modified categorical approaches by providing a
definition for ‘crime of violence’ that is based on a defendant’s conduct and a definition of
‘controlled substance offense’ that lists specific federal drug statutes”), any action by the
Sentencing Commission will not stop the madness in applying the categorical approach to
criminal statutes, see, e.g., Lowe v. United States, 

, 416–20 (6th Cir. 2019) (using
the categorical approach to hold rape under Tennessee law is not a violent felony for purposes of
the Armed Career Criminal Act’s violent-felony enhancement).

       Until the Supreme Court reverses its precedent or Congress intervenes, we will forever be
uncovering more “strange things,” Taylor, 

 (Thomas, J., dissenting) (quoting L.
Carroll, Alice in Wonderland and Through the Looking Glass 227 (J. Messner ed. 1982)), and
denying crime victims the justice they deserve.

       I respectfully dissent.
 No. 23-3466                             United States v. Cervenak                                   Page 41


                                           _________________

DISSENT

                                           _________________

        THAPAR, Circuit Judge, dissenting. The majority decides that when Tyren Cervenak
robbed two people at gunpoint, he did not commit a violent crime. Outside the walls of the
Potter Stewart courthouse, robbery is always violent. Consider a few recent Ohio robbery cases:
robbery at gunpoint in a park; running over someone with a car while stealing a diamond ring;
and stealing car keys and a wallet after a stabbing.1 That’s why the majority’s decision is not
only a blow to common sense, but also to public safety.

        The American people—Ohioans in particular—will suffer the consequences. Violent
criminals recidivate more quickly and often than their non-violent counterparts.2 Cervenak is a
good example: he committed robbery in 2016, trafficked heroin in 2020, and distributed crack
cocaine in 2022. As a result of today’s decision, recidivating violent felons like Cervenak will
be back on Ohio’s streets sooner rather than later. The consequences are as predictable as they
are tragic.

        The majority claims that the categorical approach is responsible for today’s outcome. In
fairness to the majority, the categorical approach often forces us to reach conclusions that defy
common sense. But not here. Responsibility lies at the feet of this court.

        It didn’t have to be this way. The majority passed over three defensible answers before
reaching an indefensible one. First, the majority could have held that Cervenak’s robbery
conviction was a categorical match for generic robbery under the Sentencing Guidelines.
Second, the majority could have recognized that Cervenak’s robbery conviction was a
categorical match for extortion as defined in the Guidelines. And third, at the very least, the
majority could have followed the Supreme Court’s instructions and other circuits’ examples by

        1
         See State v. Thomas, No. 113641, 

, at *1 (Ohio Ct. App. Nov. 21, 2024); State v.
Powell, No. 2024-CA-8, 

, at *1 (Ohio Ct. App. Oct. 25, 2024); State v. Flitcraft, No. 2023-L-113,

, at *5 (Ohio Ct. App. Aug. 19, 2024).
        2
           See U.S. Sent’g Comm’n, Recidivism Among Federal Violent Offenders 3 (Jan. 2019), available at
https://perma.cc/SF87-CYB9.
 No. 23-3466                         United States v. Cervenak                           Page 42


seeking clarity from the Ohio Supreme Court on Ohio law before reaching a result that opens
Ohio robbery indictments to serious constitutional concerns.        It chose none of the above.
I respectfully dissent.

                                   I. Background & Summary

          How did we get here? Start with the facts. Cervenak pled guilty to dealing drugs and
possessing a firearm.      At sentencing, the district court noted that Cervenak had previous
convictions for trafficking heroin and for robbing two people at gunpoint.           Those prior
convictions are crucial: the federal Sentencing Guidelines recommend a longer sentence when
“the defendant has at least two prior felony convictions of either a crime of violence or a
controlled substance offense.” See U.S.S.G. §§ 4B1.1, 4B1.2. Nobody disputes that Cervenak’s
heroin trafficking could serve as a predicate offense for an enhancement. The district court
concluded that Cervenak’s robbery also qualified because it was a “crime of violence.”
Accordingly, it gave him a sentencing enhancement.

          Cervenak argues that his previous Ohio robbery conviction isn’t a “crime of violence.”
Thus, he says, he shouldn’t have received the enhancement.

          To assess whether Ohio robbery is a crime of violence, courts employ the “categorical
approach.” In doing so, courts compare the elements of Ohio robbery with the elements of the
crimes that the Guidelines specify as violent. If the elements of Ohio robbery are “the same as,
or narrower than” the elements of one of the Guidelines crimes, then Ohio robbery is a “crime of
violence” under the Sentencing Guidelines. See Descamps v. United States, 

(2013).

          To run that comparison, the court first uses Shepard documents to determine the elements
of the version of Ohio robbery that Cervenak committed. Second, it pinpoints the relevant
“crime[s] of violence” listed in the Guidelines: robbery and extortion. Finally, it compares Ohio
robbery with both Guidelines robbery and extortion.

          Based on his Shepard documents—specifically, his indictment—Cervenak was convicted
of Ohio robbery under § 2911.02(A)(2). That provision prohibits “[i]nflict[ing], attempt[ing] to
 No. 23-3466                         United States v. Cervenak                               Page 43


inflict, or threaten[ing] to inflict physical harm on another” while “attempting or committing a
theft offense or in fleeing immediately after the attempt or offense.” 

(A)(2). Ohio defines a “theft offense,” in turn, to be any of thirty-one offenses listed in
§ 2913.01(K)(1). But Cervenak’s indictment didn’t specify which “theft offense” he committed.

        So, before comparing Cervenak’s robbery offense to crimes of violence in the Guidelines,
there’s a critical threshold question: Is each of those thirty-one offenses a distinct “element” of
the crime of Ohio robbery, or is each merely a different “means” of committing the single theft
offense element of Ohio robbery?

        The best reading of Ohio law—and the reading that insulates Ohio’s robbery indictment
practices from serious constitutional issues—is that the different offenses are means, not
elements. If so, then Ohio robbery is a categorical match for generic robbery because Ohio
robbery substantially corresponds to both the common law and a mass of other states’ laws. That
closes this case.

        But adopting the majority’s assumption that each particular theft offense is itself an
element—such that there are at least thirty-one different criminal offenses that all go by the name
of “robbery” in Ohio—then there’s a constitutional conundrum. Constitutional grand jury and
fair notice requirements mandate that indictments list all the elements of the crime. But Ohio
routinely doesn’t specify the individual theft offense on which the robbery charge is predicated.
If an individual theft offense is an element, the majority’s approach would mean that Ohio is
violating the state and federal constitutions when it fails to specify the theft offense in its robbery
indictments. To avoid that constitutional concern, precedent leaves two reasonable choices.

        Under option one, turning to state courts’ interpretations of state law and practice
demonstrates that Ohio courts presume that an unspecified “theft offense” is ordinary theft. This
makes Ohio robbery a clean match with Guidelines extortion. And it avoids any constitutional
concerns.

        As for option two, before figuring that Ohio’s criminal procedure is on a collision course
with the Ohio and federal constitutions, precedent commands that we ask the Ohio Supreme
Court for clarification on Ohio law.       That is, the court should have certified whether the
 No. 23-3466                         United States v. Cervenak                         Page 44


individual theft offenses are means or elements.        The court also should have certified the
question of whether a robbery indictment resting on an unspecified theft offense means ordinary
theft (and thus gives the defendant the constitutional notice that he’s due).

        Neither of these options involves declaring that robbing two people at gunpoint isn’t a
violent crime. Yet that’s what the majority has done.

                                           II. Robbery

        When Cervenak was sentenced for his federal offense, the Sentencing Guidelines did not
define robbery. See U.S.S.G. § 4B1.2(a) (2021). Thus, a court must “compare the elements” of
Ohio robbery “with the elements of the ‘generic’ crime—[that is,] the offense as commonly
understood.” Descamps, 

. Here, that would involve comparing Ohio robbery
under § 2911.02(A) with generic robbery.

        But running that comparison first requires knowing the elements of Ohio robbery. And
that, in turn, requires knowing whether the individual theft offenses are themselves elements or
are mere means of committing the single “theft offense” element. They are means.

        A. Ohio Robbery’s Different Means of Committing the “Theft Offense” Element

        Elements “are the constituent parts of a crime’s legal definition—the things the
prosecution must prove to sustain a conviction.” Mathis v. United States, 

(2016) (citations omitted). Means are “various factual ways” of committing a single element.

.

        To be sound under the federal and Ohio constitutions, the indictment must list all the
elements of the crime. Russell v. United States, 

 (1962); State v. Troisi, 

, 703–04 (Ohio 2022). But it doesn’t have to list the means. And if the defendant
goes to trial, the jury must unanimously agree that each element of the alleged crime has been
satisfied. Mathis, 

; Ohio Crim. R. 31(A). But the jurors don’t have to all agree
on the means a defendant used.

        Consider an example. A statute prohibits the use of a “deadly weapon,” and then lists
various weapons that are “deadly”—such as a “knife,” a “gun,” and so on. See Mathis, 579 U.S.
 No. 23-3466                        United States v. Cervenak                            Page 45


at 504. The use of a deadly weapon is an element; the specific weapons are different means of
satisfying that element. 

 The indictment would need to state that the defendant used a deadly
weapon. But it would not need to specify whether he used a knife, gun, or the like. To convict
the defendant, the jury would have to find he used a deadly weapon. But the jurors would not
need to agree on which deadly weapon he used.

       Here, the text of Ohio’s robbery statute; state court caselaw; constitutional avoidance;
and our categorical approach precedent show that the individual theft offenses are means of
committing Ohio robbery’s theft offense element.

       Start with the plain text. The statute criminalizes “attempting or committing a theft
offense.” 

(A). It never specifies which “theft offense” the
defendant must commit. Instead, the statute points to a different provision, which in turn lists
several qualifying theft offenses—just like the knife, gun, and other deadly weapons listed in the
hypothetical statute above. See 

 § 2911.02(C)(2) (defining “theft offense” with reference to
§ 2913.01). Thus, a defendant could be convicted of robbery by means of any theft offense listed
in § 2913.01(K)(1). All that matters for a conviction is that the defendant “[i]nflict, attempt to
inflict, or threaten to inflict physical harm on another” while committing any “theft offense.” Id.
§ 2911.02(A)(2).

       Ohio courts confirm what the text reveals. Ohio courts routinely refer to “the ‘theft
offense’ element of the robbery statute,” rather than referring to the multiple predicate theft
offenses. See State v. Tolliver, 

 (Ohio 2014) (emphasis added). And they
have done the same for aggravated robbery, which mirrors Ohio robbery’s reliance on a
predicate “theft offense.” See State v. Horner, 

 (Ohio 2010) (noting that
aggravated robbery “includes as an element an underlying theft offense” (emphasis added)).
Thus, these cases suggest that the underlying theft offense isn’t a set of alternative elements;
rather, the theft offense is a single element of Ohio robbery. The individual theft offenses are
alternative means of satisfying that single element.

       What’s more, when an Ohio crime rests on a predicate offense, the Ohio Supreme Court
has held that “an indictment’s reference to the statute number of a predicate offense provides
 No. 23-3466                        United States v. Cervenak                           Page 46


sufficient notice to a defendant.” Troisi, 

 (citing State v. Buehner, 

, 1164 (Ohio 2006)). Thus, the indictment “need not also include each element of the
predicate offense.” Buehner, 853 N.E.2d at 1164. In Buehner, the defendant was charged with
the offense of “ethnic intimidation,” which requires the commission of one of various predicate
offenses.   Id.   The defendant’s indictment “specifically identified” one of those predicate
offenses by its “statute number.” Id. Thus, the Ohio Supreme Court found that the indictment
was adequate. Id.

       Buehner teaches that if the § 2913.01(K)(1) theft offenses were “elements” of robbery,
the robbery indictment would need to list the “statute number” of the specific predicate theft
offense that the defendant is accused of committing. Here, for example, that would include
listing “§ 2913.02” for ordinary theft. But the robbery counts in Cervenak’s indictment—like
many such Ohio robbery indictments—don’t mention any predicate by name or by statute
number. Rather, Cervenak’s indictment just accuses him of “attempting or committing a theft
offense.” R. 32-1, Pg. ID 141–42. The fact that the indictment doesn’t specify the theft offense
by name or by number indicates that the specific theft offense isn’t an element. So, to convict a
defendant of Ohio robbery, the jury must find that he committed a theft offense—it can be any of
the thirty-one listed in § 2913.01(K)(1).

       Constitutional avoidance also supports reading the individual theft offenses as alternative
means of committing the single “theft offense” element. The federal and Ohio constitutions
mandate that indictments give fair notice by listing the elements of the crime. Russell, 

 (relying on due process and the Sixth Amendment’s notice clause, among other
provisions); Troisi, 

 (interpreting the Ohio constitution as requiring indictment
to “contain[] the elements of the offense charged” (citation omitted)).

       But Ohio robbery indictments often don’t charge the predicate theft offense. Instead,
they state that the defendant was “attempting or committing a theft offense”—just as Cervenak’s
indictment did. R. 32-1, Pg. ID 141–42. That’s a constitutional problem unless the individual
theft offenses are mere means of committing a single theft offense element.
 No. 23-3466                               United States v. Cervenak                                      Page 47


         So, the “means” reading squares with our duty to avoid concluding that states are
violating the Constitution—as well as their own constitutions. As Justice Story wrote when
confronted with a state constitutional challenge to a state law while riding circuit: If “there is a
construction [of the state law], which although not favored by the exact letter . . . give[s] it a
constitutional character,” then “this construction ought to be adopted.”                           Soc’y for the
Propagation of the Gospel v. Wheeler, 

 (Story, Circuit Justice,
C.C.D.N.H. 1814) (No. 13,156).

         Justice Story’s admonition flowed from earlier Supreme Court opinions. In Cooper v.
Telfair, 

 (1800), a diversity suit raised the question whether a Georgia statute
violated the Georgia Constitution. See David P. Currie, The Constitution in the Supreme Court:
The First Hundred Years, 1789–1888, at 51 (1992). The Court found no state constitutional
violation. Justice Washington explained that the state law’s unconstitutionality would need to be
“clearly demonstrated” before he would conclude that “the law [was] void.” Cooper, 

 (Washington, J.).          Similarly, Justice Paterson wrote that there must be a “clear and
unequivocal breach of the [Georgia] constitution” to warrant pronouncing the Georgia “law
void.” 

 (Paterson, J.). If founding-era Supreme Court Justices hesitated before finding a
conflict between a state’s law and its constitution, we should not manufacture such a conflict
here.

         A final argument for concluding that the individual theft offenses are means of
committing the theft offense element sounds in uncertainty: If there’s doubt about the elements-
versus-means question, then “the statute is indivisible.” United States v. Stitt, 

(6th Cir. 2017) (en banc), rev’d on other grounds, 

 (2018). In other words, the theft
offenses are various means of committing the same “indivisible” crime, rather than separate
“divisible” elements.3


         3
           The majority, for its part, points to State v. Tolliver in support of the proposition that Ohio robbery
requires proof of “the elements” of the underlying theft offense. Maj. Op. at 7 (quoting Tolliver, 

).
If the majority’s reading of Tolliver is correct, that places Tolliver in conflict with the Ohio Supreme Court’s
reasoning in State v. Horner. See 

 (stating that “an underlying theft offense” is “an element” of
Ohio robbery (emphases added)). It also places Tolliver in tension with itself: Tolliver elsewhere refers to “the
‘theft offense’ element of the robbery statute.” 

 (emphasis added). So, if the majority is right about
Tolliver, then the Ohio caselaw is at least muddled on this question—which is yet another reason to either certify the
 No. 23-3466                               United States v. Cervenak                                      Page 48


                          B. Comparing Ohio Robbery & Generic Robbery

         The individual theft offenses are not themselves elements of Ohio robbery. Thus, the
next step is to compare the elements of Ohio robbery with the elements of generic robbery. See
Descamps, 

.

         Start with the elements of Ohio robbery. Recall that Cervenak was convicted under
§ 2911.02(A)(2), which prohibits “[i]nflict[ing], attempt[ing] to inflict, or threaten[ing] to inflict
physical harm on another” while “attempting or committing a theft offense or in fleeing
immediately after the attempt or offense.” 

(A)(2). So, Ohio
robbery consists of two sets of elements. First, there’s a force element. That’s the inflicting,
attempting to inflict, or threatening to inflict physical harm. And second, there’s the theft
element. That’s the § 2913.01(K)(1) “theft offense.”

         Generic robbery has those same two elements: force and theft. To determine generic
robbery’s elements, courts must first look to the common law and then at other states’
contemporary practice. See Stokeling v. United States, 

 (2019); Taylor v. United
States, 

, 598–99 (1990). Both the common law and other states’ contemporary
practice indicate that Ohio robbery is a match for generic robbery.

                                                     1. Force

         Ohio robbery “substantially corresponds” to generic robbery as defined by the common
law and other states’ laws. Taylor, 

. First, both Ohio robbery and generic
robbery require an actual or threatened infliction of injury. Second, both Ohio robbery and
generic robbery also entail some sort of theft or theft-related offense.

                                      i. Ohio Robbery’s Force Element

         Ohio robbery requires “[i]nflict[ing], attempt[ing] to inflict, or threaten[ing] to inflict
physical harm on another.” 

(A)(2). These requirements match



elements-versus-means question, see infra, Part IV, or conclude that the distinct theft offenses are means, see Stitt,

.
 No. 23-3466                        United States v. Cervenak                             Page 49


the requirement of common law robbery that property be taken through either “force or putting
[the victim] in fear.” 3 Wayne R. LaFave, Substantive Criminal Law § 20.3 (3d ed. 2017).

       And Ohio robbery’s mens rea requirement isn’t to the contrary. Despite what prior
panels have said, see, e.g., United States v. Butts, 

, 770–71 (6th Cir. 2022), Ohio
robbery requires that its force element be committed at least recklessly. That’s been true since
March 2015, the effective date of a bill that abrogated a 2014 Ohio Supreme Court ruling to the
contrary. Unfortunately, our court has failed to recognize this critical change in Ohio law.

       In the 2014 case of State v. Tolliver, the Ohio Supreme Court held that “the state need not
prove a culpable mental state with respect to the force element” of Ohio robbery. 

. And Tolliver made clear that its holding applied to § 2911.02(A)(2), the type of robbery for
which Cervenak was convicted.        See Butts, 

 (“Although Tolliver analyzed
convictions under § (A)(3) rather than § (A)(2) of § 2911.02, it made clear that its reasoning
applies equally to all of § 2911.02’s subdivisions.” (citing Tolliver, 19 N.E.3d at 874–75)).

       Our court has relied on Tolliver to conclude that a defendant can commit robbery’s force
element with a mental state less culpable than recklessness. See, e.g., id. at 770–71.

       But the Ohio legislature abrogated Tolliver just months after the state high court handed it
down. As a leading treatise on Ohio criminal law explains, the new law “put to rest” Ohio
courts’ “struggles,” like in Tolliver, to determine an element’s mens rea when the statute didn’t
specify one. 2 Lewis R. Katz et al., Baldwin’s Ohio Practice, Criminal Law § 85:10, at 187 (3d
ed. Supp. 2024). Therefore, State v. Tolliver “should be considered abrogated by the statutory
amendment.” Id.

       How did the legislature abrogate Tolliver? Tolliver had reasoned that under then-existing
Ohio law, if a section of the Revised Code (like robbery) “already require[d] proof of a culpable
mental state” for one element of the offense, then “the state need prove” culpability only for that
element. 

. So, because the robbery section of the Code already required proof
of a culpable mental state for the predicate theft offense, the prosecution only had to prove a
culpable mental state as to that theft offense element. 

 Therefore, the prosecution didn’t have
to prove a culpable mental state as to robbery’s force element. 

 No. 23-3466                                United States v. Cervenak                                       Page 50


         In response to Tolliver, the Ohio legislature amended state law as follows: “When
language defining an element of an offense that is related to knowledge or intent or to which
mens rea could fairly be applied neither specifies culpability nor plainly indicates a purpose to
impose strict liability, the element of the offense is established only if a person acts recklessly.”

(C)(1). “The amendment made clear that mens rea is to be
applied on an element-by-element basis.” 2 Katz et al., supra, § 85:10, at 188. In other words,
the legislature abrogated Tolliver’s focus on the section: the revised law “applies to the elements
of an offense, rather than the entire section defining an offense.” State v. Cunningham, No.
21AP-470, 

, at *15 (Ohio Ct. App. May 28, 2024) (emphasis added); see also
State v. Ewing, No. 29685, 

, at *7 (Ohio Ct. App. June 30, 2021) (observing
that § 2901.21(C)(1) “now omits any reference to the ‘section’ that defines an offense”). After
that amendment, the fact that the theft offense element requires a culpable mental state doesn’t
mean that the force element lacks such a requirement. Rather, at the very least, the force element
requires a “reckless” mental state.

         That’s a critical change in Ohio law. The force element of Ohio robbery is an element
“to which mens rea could fairly be applied,” doesn’t “specif[y] culpability,” and doesn’t “plainly
indicate[] a purpose to impose strict liability.” Cunningham, 

, at *15 (quoting
§ 2901.21(C)(1)). Therefore, the force element “is established only if a person acts recklessly.”

(C)(1). It no longer matters that the Ohio robbery statute
specifies a mental state for the theft offense element but not the use-of-force element. Section
2901.21(C)(1) fills the gap.

         Ohio courts of appeals have recognized this crucial statutory change. See Cunningham,

, at *15. Therefore, in Cunningham, the jury instructions correctly specified
that the defendant “had to commit the physical harm element of robbery with the mental state of
recklessness.” Id.; see also State v. Morris, No. C–150421, 

, at *3 (Ohio Ct.
App. Aug. 24, 2016) (recognizing the same effect of § 2901.21(C)(1) on aggravated robbery).4



         4
           To be sure, other Ohio courts of appeals have failed to notice the Ohio legislature’s abrogation of Tolliver.
See, e.g., State v. Smith, No. 27560, 

, at *2 (Ohio Ct. App. July 13, 2018). But we should follow
the lead of the courts of appeals like Cunningham that noticed the critical change in Ohio law. And to the extent
 No. 23-3466                              United States v. Cervenak                                     Page 51


                                   ii. Generic Robbery’s Force Element

          Ohio robbery requires a mens rea of at least recklessness for the force element, which
makes it an easy match with generic robbery. Cf. United States v. Adair, 

 (5th
Cir. 2021); United States v. Flores-Mejia, 

 (9th Cir. 2012); United States v.
Martin, 

, 883–85 (8th Cir. 2021). How so?

          The Supreme Court has repeatedly told courts to look at two sources to define a generic
offense. First, the common law. Stokeling, 

; cf. Neder v. United States, 

, 21–22 (1999). Second, other states’ definitions of the crime. Why look there? Because by
omitting a definition of robbery, the Sentencing Commission “did not wish to specify an exact
formulation that an offense must meet in order to count” for an enhancement. Taylor, 

.     Rather, the Commission designated certain predicate offenses by using “uniform,
categorical definitions to capture all offenses of a certain level of seriousness.” 

. So,
when applying the categorical approach, the Court has “repeatedly declined to construe” a
federal criminal statute “in a way that would render it inapplicable in many States.” Stokeling,

; see also United States v. Castleman, 

 (2014) (reading “physical
force” to include common-law force, in part because a different reading would render federal law
“ineffectual in at least 10 States”). The same holds true for the Sentencing Guidelines.

          Thus, the common law and other states’ robbery statutes provide a definition of generic
robbery. See United States v. Brown, 

 (6th Cir. 2020) (doing the same for
generic burglary). The common law didn’t require a mens rea with respect to robbery’s force
element. And, like Ohio, many states require only a mens rea of recklessness for the force
element.

                                               a. Common Law

          A defendant committed common law robbery by force when he applied enough force to
overcome the victim’s resistance. His intent didn’t matter.

there is any doubt, we should certify to the Ohio Supreme Court the question whether the force element of Ohio
robbery must be committed with a mens rea of at least recklessness, consistent with the 2015 statutory change. See,
e.g., Somers v. United States, 

, 1054–56 (11th Cir. 2021) (certifying state law questions to state
supreme court in light of a “split” in state lower courts), certifying questions to 

 (Fla. 2022).
 No. 23-3466                         United States v. Cervenak                             Page 52


       At common law, “[t]he intent of the perpetrator of violence was not a factor in
determining whether the offense of robbery ha[d] been committed.” United States v. Morris, No.
CR 20-438, 

, at *6 (E.D. Pa. Sept. 21, 2022), aff’d, No. 22-2781, 

 (3d Cir. Oct. 27, 2023). For example, Blackstone observed that “open and violent
assault” distinguished robbery from simple theft. 4 William Blackstone, Commentaries on the
Laws of England 241 (1769). But the amount of force needed for this violence was not required
to be of “any great degree.” 

. All the law required was that the force “oblige[d] a man
to part with his property.” 

 It didn’t require a mens rea.

       A later treatise describes the force element similarly: “if any injury be done to the
person . . . there will be a sufficient actual ‘violence’” to establish robbery. 1 William Oldnall
Russell & Charles Sprengel Greaves, Crimes and Misdemeanors 875 (8th ed. 1857). This
treatise lists some contemporary examples, none of which required intentional use of force.
They include snatching an earring from a lady’s ear such that “blood [was] drawn”; twisting a
diamond pin from a lady’s hair such that “part of the hair [was] torn away”; or snatching a watch
from a steel chain and “overcom[ing] the resistance made by the steel chain.” 

 at 875–76.
One could do so negligently.

       Other treatises shared this conception of the force element. See, e.g., Wm. L. Clark &
Wm. L. Marshall, 2 Law of Crimes 861–62 (1900) (“If there is any injury to the person of the
owner, or if he resists the attempt to rob him, and his resistance is overcome, there is sufficient
violence to make the taking robbery, however slight the resistance.”). Force at common law only
required overcoming the resistance of a victim. It didn’t require any intent or purpose.

       The Supreme Court shares this reading of common law.              In Stokeling, the Court
explained that “[i]f an act physically overcame a victim’s resistance, however slight that
resistance might be, it necessarily constituted violence.” 586 U.S. at 78 (quotation omitted).

       Many state supreme courts have read the common law in the same way. Consider West
Virginia. There, robbery “derive[s] from the common law.” State v. Wilkerson, 

 (W. Va. 2013) (citation omitted). Accordingly, the state supreme court has explained: “The
intent to injure . . . is not an element of robbery.” 

. Similarly, the New Jersey Supreme
 No. 23-3466                         United States v. Cervenak                              Page 53


Court has observed that common law robbery cases “uniformly required the presence of violence
or threat but did not address the mental state that must accompany that conduct. Thus, the
common law did not place separate or discrete significance on the culpability associated with the
assaultive aspect of robbery.” State v. Sewell, 

 (N.J. 1992). All told, common
law robbery didn’t require intent for the force element.

       The common law is consistent with Ohio practice, both under the old Tolliver regime and
after § 2901.21(C)(1) supplied a minimum mens rea of recklessness. Remember the common
law examples of snatched earrings, torn pins, and stolen watches? These examples look a lot like
Ohio robbery committed with reckless force.

       Take State v. Pellegrini, No. 1-12-30, 

 (Ohio Ct. App. Jan. 22, 2013).
There, the defendant grabbed a moneybag from the victim’s hand. 

. In doing so, the
defendant caused the moneybag to catch on a ring on the victim’s finger, which in turn caused
her finger to swell. 

 The Ohio Court of Appeals upheld the conviction, finding that the
defendant “recklessly caused the victim physical harm.” 

. This incident mirrors those
described in the common law treatises: The defendant may not have intended to use force, but
he recklessly (or without any culpable mens rea with respect to his use of force, for that matter)
caused an injury when trying to take the property away from the victim. Just like how grabbing
the pin (which in turn caught the victim’s hair) constituted incidental force sufficient for robbery,
so too does grabbing the moneybag (which in turn caught the ring) constitute incidental force
sufficient for robbery. The upshot, then, is that Ohio robbery, with its minimum mens rea of
recklessness, clears the minimum culpability required at common law.

                                          b. Other States

       Ohio’s use of recklessness as a baseline mens rea also matches up with at least twenty-
one other states’ robbery statutes. That’s critical because the Supreme Court has declined to
define a generic crime in a way that would exclude statutes “in at least 10 States.” Castleman,

. So, if our court were to conclude that Ohio robbery isn’t a match with generic
robbery because it can be committed recklessly, we’d be concluding that the robbery statutes of
at least twenty-two states fall outside the generic definition. This court can’t do that. See 

 No. 23-3466                               United States v. Cervenak                                      Page 54


         To start, three states explicitly allow robbery to be committed by reckless use of force.5

         Six more states, including Ohio, don’t specify a mens rea for the force element of their
robbery statute. But elsewhere in their criminal codes, they provide that if an offense fails to
specify a mens rea, then the requisite mens rea is at least recklessness (or, in one state,
negligence).6 And some of these states’ courts have recognized that robbery’s force element thus
takes on a minimum mens rea of recklessness. See, e.g., McElyea v. State, 

(Ark. Ct. App. 2004) (holding that “[f]or purposes of the robbery statute, it is immaterial whether
appellant ever intended to use physical force”), aff’d, 

 (Ark. 2005); People v.
Jones, 

 (Ill. 1992) (holding that intent, knowledge, or recklessness can
serve as the mental state for robbery). So too with their jury instructions. See, e.g., Ill. Pattern
Jury Instrs.-Criminal 14.01 (providing that a person can “recklessly” commit the offense of
robbery). And in at least one of these states, Pennsylvania, a state court has suggested that strict
liability suffices. See Commonwealth v. Payne, 

 (Pa. Super. 2005) (holding
that robbery, unlike aggravated assault, “simply does not require [a specific] mental state when
the ‘serious bodily injury’ is inflicted”).              Meanwhile, federal courts have reasoned that
recklessness would suffice in Pennsylvania. See, e.g., United States v. Parrott, 

, 667 (E.D. Pa. 2022); United States v. Giles, No. CR 94-20-01, 

, at *4
(E.D. Pa. Nov. 19, 2021). So far, that’s nine states, including Ohio, that define robbery to
include a reckless use of force.

         But that’s not all. Two states impose a strict liability mens rea on robbery’s force
element. California, for one, has no mens rea when it comes to use of force against the victim.

         5
          See 

(1)(c) (robbery includes “recklessly inflict[ing] serious bodily injury
upon another”); Me. Rev. Stat. Ann. tit. 17-A, § 651(1)(A) (robbery includes “recklessly inflict[ing] bodily injury on
another”); 

(a)(1) (robbery includes “intentionally, knowingly, or recklessly caus[ing]
bodily injury to another”).
         6
            See 

 (defining robbery); 

 § 5-2-203(b) (applying at least recklessness absent
a prescribed mens rea); 720 Ill. Comp. Stat. Ann. 5/18-1 (defining robbery); id. 5/4-3(b) (requiring that a defendant
act with “any mental state defined in Sections 4-4, 4-5 or 4-6,” the least culpable of which is recklessness); 

(1) (defining robbery); 

 § 12.1-02-02(1)(e), (2) (applying at least recklessness
absent a prescribed mens rea); 

 (defining robbery); 

 § 302(c) (applying at
least recklessness absent a prescribed mens rea); 

 (defining robbery); 

 cmt. (noting
that Montana robbery “parallels that of” the Model Penal Code, which can rest on a mens rea of recklessness, see
infra); 

 § 45-2-103(1) (applying at least negligence absent a prescribed mens rea).
 No. 23-3466                              United States v. Cervenak                                       Page 55


People v. Anderson, 

 (Cal. 2011) (upholding a robbery conviction even though
the defendant “did not intend to strike” the victim “but did so accidentally”); see also United
States v. Martinez-Hernandez, 

 (9th Cir. 2019) (noting that California
robbery may be committed with an accidental use of force). Kansas uses this scheme, too. See
State v. Moulin, 

 (Kan. Ct. App. 2019) (table) (explaining that robbery’s “force
element does not require a culpable mental state”). That’s eleven states.

         Further, at least three other states define the force required for robbery by the injury
caused, not the perpetrator’s mens rea.7 Thus, these robbery statutes “include[] no element of
intent to inflict bodily injury.” State v. Lewis, 

 (Wash. Ct. App. 2010), as
amended (June 2, 2010) (emphasis in original). Some courts in these states accordingly suggest
that a person could commit robbery “by forcibly stealing property and injuring a person in the
course thereof recklessly or with criminal negligence.” State v. Wright, 

, 796
(Mo. Ct. App. 2020). This makes sense: When one part of the statute requires a specific intent
to take the property of another but is silent on the force element, “[t]he statute thus does not
explicitly impose a requirement that the force or violence have been intentionally deployed.”
Gray v. United States, 

 (D.C. 2017) (McLeese, J., concurring in the judgment
in part and dissenting in part). The tally is up to fourteen states, then, where robbery can be
committed by reckless force.

         And, as discussed earlier, West Virginia robbery is a common law crime. Wilkerson, 

. So, that state’s highest court has made clear that “[t]he intent to injure . . . is not an
element of robbery.” 

. That’s fifteen states.

         West Virginia is not alone. Robbery in North Carolina, South Carolina, Rhode Island,
and Virginia also flows from the common law. See State v. Herring, 

 (N.C.
1988); 

; State v. Day, 

, 977–78 (R.I. 2007);
Commonwealth v. Jones, 

 (Va. 2004).8 As explained above, the common law


         7
         See, e.g., 

(1) (the defendant must “cause[] physical injury”); Wash. Rev. Code
Ann. § 9A.56.200(1)(a)(iii) (the defendant must “[i]nflict[] bodily injury”); 

(a)(1) (same).
         8
           To be sure, certain federal courts have reasoned that some of these common law robbery statutes don’t
entail a realistic probability of conviction for a reckless use of force. Why? Because there’s no evidence of state
 No. 23-3466                               United States v. Cervenak                                       Page 56


didn’t impose a culpable mens rea requirement on robbery’s force element. So, absent any
indication from these states’ supreme courts, there’s no evidence that they have departed from
the common law. Indeed, the evidence shows that they have stayed true to the common law
requirement that only a minimal quantum of force be used when committing robbery to satisfy
the force element. See, e.g., State v. Sawyer, 

 (N.C. 1944) (“Although actual
force implies personal violence, the degree of force used is immaterial, so long as it is sufficient
to compel the victim to part with his property. . . .”); see also United States v. Gardner, 

, 803–04 (4th Cir. 2016) (collecting more recent examples of North Carolina caselaw),
abrogated on other grounds as recognized in United States v. White, 

 (4th Cir.
2021).

         That adds up to nineteen states.

         Finally, there are three more. Maryland robbery can be committed with reckless force.
That state’s caselaw characterizes robbery as “larceny from the person accomplished by either an
assault (putting in fear) or a battery (violence).” Snowden v. State, 

 (Md.
1991). And battery in Maryland can be committed recklessly or with criminal negligence. See
Duckworth v. State, 

, 112–13 (Md. 1991). It need not be intentional. Elias v. State,

 (Md. 1995). Thus, Maryland robbery can be committed through a reckless,
unintentional battery.        The same goes for Delaware.                The Delaware Supreme Court has
explained: “Although physical injury may be an element of first degree robbery in some cases, it
need not be caused intentionally or recklessly. . . .” Hackett v. State, 

 (Del.
1990); accord Fedorkowicz v. State, 

 (Del. 2010) (table). And based on the text of
its robbery statute, Florida is also worthy of inclusion. The statute specifies a mens rea for the
theft element, but not the force element. See 

(1) (requiring only that


court convictions to that effect. See, e.g., United States v. Doctor, 

 (4th Cir. 2016). But this case
doesn’t require determining whether robbery in these states is a categorical match with a federal statute or
Guideline. Instead, it requires looking at these laws to get a sense of the universe of state robbery laws to reach a
generic definition of robbery—much like the Supreme Court in Castleman with respect to assault or battery. In
Castleman, the Court relied on the Government’s listing of the relevant state laws, the majority of which included
only citations to state statutes and didn’t mention state caselaw. Castleman, 572 U.S. at 167–68 (citing App. B to
Brief for United States 10a–16a). Similarly, in his concurrence, Justice Scalia cited the state statutes alone, without
any reference to state court caselaw. 

 at 178 n.5 (Scalia, J., concurring in part and concurring in the judgment).
After all, the goal here is to get a general idea of what “robbery” is so as not to rule out a large batch of states.
 No. 23-3466                               United States v. Cervenak                                       Page 57


“when in the course of the taking there is the use of force, violence, assault, or putting in fear”).
That’s twenty-two total states.

         In addition, the Model Penal Code ties robbery’s force element to a reckless mens rea.
Cf. Taylor, 

 & n.8 (noting how the MPC squares with other state statutes in
defining a generic crime). The MPC doesn’t specify the mens rea for the use of force during a
robbery. See Model Penal Code § 222.1. So, knowledge or recklessness suffices. See id.
§ 2.02(3).

         All told, at least twenty-two states, plus the MPC, say a defendant can commit robbery by
a reckless use of force. Again, that’s critical because the Supreme Court has declined to define
generic crime in a way that would exclude the relevant statutes “in at least 10 States.”
Castleman, 

; see also Stokeling, 

 (declining to construe a federal
criminal statute “in a way that would render it inapplicable in many States”). Following the
Supreme Court, generic robbery includes a reckless use of force.9

                                                 *        *        *

         In sum, Ohio’s robbery statute is by no means “idiosyncratic” in defining force with a
mens rea of at least recklessness. Brown, 

. For one, the common law didn’t
require intent for robbery’s force element. And Ohio robbery is in the good company of at least
twenty-one other states.10


         9
           Ohio robbery under § 2911.02(A)(2) can also involve “threaten[ing] to inflict physical harm on
another.” This version of Ohio robbery is consistent with the common law and other states’ practices. Blackstone
explained that the necessary violence for robbery required “so much force, or threatening by word or gesture, be
used, as might create an apprehension of danger.” 4 Blackstone, supra, at 242. Just like robbery by force, robbery
by “threatening” did not come with an attached mens rea requirement. See Sewell, 

 (explaining that
common law robbery “cases uniformly required the presence of violence or threat but did not address the mental
state that must accompany that conduct. Thus, the common law did not place separate or discrete significance on
the culpability associated with the assaultive aspect of robbery.” (emphasis added)). And the robbery codes across
states today include threats to inflict bodily injury without specifying a mens rea. See, e.g., 720 Ill. Comp. Stat.
Ann. 5/18-1(a) (robbery includes “threatening”); 

 (robbery can be accomplished by “force or
fear”); see also Model Penal Code § 222.1(b) (robbery includes “threaten[ing] another with . . . bodily injury”).
         10
           Of course, this does not mean that a state robbery statute that lacks a mens rea for the force element can’t
be a categorical match for generic robbery. After all, common law robbery had no mens rea requirement for the
force element. And robbery is the “quintessential” crime of violence. Stokeling, 

. But because the
minimum mens rea for Ohio robbery’s force element is recklessness, that’s not at issue here.
 No. 23-3466                                  United States v. Cervenak                           Page 58


                                                   2. Theft Element

          Ohio robbery is also a categorical match for generic robbery with respect to the theft
offense element. Ohio robbery’s theft offense element is not an outlier compared to that of other
states.

          Here, recall that the underlying “theft offenses” represent different means of committing
Ohio robbery’s “theft offense” element. Given the categorical approach’s “indifference to how a
defendant actually committed a prior offense,” a court “may ask only whether the elements of the
state crime and generic offense” provide a match.                     Mathis, 

 (emphasis in
original). So, a court must compare the “theft offense” element of Ohio’s robbery statute to the
theft element of generic robbery to see if they align.

          Compared to other states, Ohio’s “theft offense” need only have the same “basic”
substance as its state counterparts to fit within the generic definition, “regardless of its exact
definition or label.” Taylor, 

. In order words, Ohio’s conception of a “theft
offense” need not be a dead ringer for other states’ laws; it need only “substantially
correspond[]” to them. Quarles v. United States, 

 (2019) (quoting Taylor, 

).

          It does. Ohio is not alone in defining the theft offense element of robbery to include a
broad range of means. Just like the text of the Ohio statute that lists the thirty-one “theft
offenses,” many other states—seventeen at least—define “theft” more broadly by way of an
umbrella term.11 And just like Ohio, many of these states seem to understand their umbrella
theft offense to involve various means of committing a theft. They don’t treat the facts that give
rise to the theft offenses as elements that must be proven to the jury. This is because their
robbery indictments often don’t specify an underlying “theft offense.” They instead refer to that
state’s umbrella “theft offense” term.12 So, the umbrella theft offense element in these states


          11
               Such terms include “theft” and “larceny.” See Table infra.
          12
           See, e.g., Indictment, State v. Clark, No. 9014039, 

 (Travis Cnty. Apr. 17, 2001)
(Texas); Indictment, State v. Howard, No. CR20032139, 

 (Pulaski Cnty. June 19, 2003)
(Arkansas); Indictment, State v. Tyner, No. 11SC104368, 

 (Fulton Cnty. Sept. 28, 2011)
(Georgia); Indictment, State v. Mau, No. 06-1-1393, 

 (Honolulu Cnty. July 12, 2006) (Hawaii);
 No. 23-3466                            United States v. Cervenak                                 Page 59


serves as a comparator with Ohio robbery’s umbrella theft offense element. See Mathis, 

. And such a comparison demonstrates that at least on their faces, the robbery statutes
of several other states criminalize conduct that doesn’t necessarily entail ordinary theft (taking
something of value from another)—just like Ohio. The chart below offers a sampling of such
state robbery statutes and the breadth of the theft offenses on which they’re predicated.


          State             Umbrella Theft Offense                    Examples of the Breadth
                                for Robbery                            of That Theft Offense

 Arkansas                “theft.” 

- receiving stolen property. Ark. Code
                         12-102(a).                   Ann. § 5-36-106(a).

 Connecticut             “larceny.” Conn. Gen. Stat. “failing to pay the prevailing rate of
                         Ann. § 53a-133.             wages,” Conn. Gen. Stat. Ann. § 53a-
                                                     119(14), receiving stolen property, id.
                                                     § 53a-119(8),       and      installing
                                                     “counterfeit” or “nonfunctional” air
                                                     bags, id. § 53a-119(16).

 Delaware                “theft.” Del. Code Ann. tit. paying employees below the minimum
                         11, § 831(a).                wage and misclassifying an employee
                                                      as an independent contractor with the
                                                      intent to evade taxes. See Del. Code
                                                      Ann. tit. 11, § 841D(a); id. tit. 19,
                                                      § 1102A(a)(4), (a)(5).

 Georgia                 “theft.” 

- “abandon[ing] a shopping cart upon
                         8-40(a).                     any public street,” 

-
                                                      8-21(c), receiving stolen property, 

                                                      § 16-8-7, and “bringing stolen property
                                                      into” Georgia, id. § 16-8-9.

 Hawaii                  “theft.” Haw. Rev. Stat. Ann. receiving stolen property. Haw. Rev.
                         § 708-840(1).                 Stat. Ann. § 708-830(7).


Indictment, Commonwealth v. Derkson, No. 05-CR-220-002, 

 (Kenton Cnty. Apr. 15, 2005)
(Kentucky); Indictment, State v. Harris, No. BANSCCR200601108, 

 (Penobscot Cnty. Jan. 2,
2007) (Maine); Indictment, State v. Perry, No. 16-02-0138I, 

 (Mercer Cnty. Mar. 24, 2021) (New
Jersey); Indictment, State v. Toland, No. 09C48275, 

 (Marion Cnty. Sept. 17, 2009) (Oregon);
Indictment, Commonwealth v. Myers, No. 2015CR0003265, 

 (Chester Cnty. Oct. 8, 2015)
(Pennsylvania).
No. 23-3466                 United States v. Cervenak                         Page 60


Iowa           “theft.”    Iowa Code Ann. “[a]ny act that is declared to be theft by
               § 711.1(1).                any provision of the Code,” 

(10), exercising control
                                          over property that one knows to have
                                          been stolen, 

 § 714.1(4), and
                                          destroying, concealing, or disposing of
                                          property in which somebody else has a
                                          security interest, id. § 714.1(5).

Kentucky       “theft.” Ky. Rev. Stat. Ann. receiving stolen property, Ky. Rev.
               § 515.020(1).                Stat. Ann. § 514.110, and possessing or
                                            receiving stolen letters or postcards, id.
                                            § 514.150.

Maine          “theft.” Me. Rev. Stat. Ann. receiving stolen property.      Me. Rev.
               tit. 17-A, § 651(1).         Stat. Ann. tit. 17-A, § 359.

Michigan       “larceny.” Mich. Comp. Laws “injuring” somebody else’s “fruit tree”
               Ann. § 750.530(1).          and “injuring” another’s “ornamental
                                           shrub.”    

.

Montana        “theft.”  Mont. Code Ann. pawn shop owner allowing stolen
               § 45-5-401(1).            property to be bartered there after a
                                         peace officer has called for the property
                                         to be held for thirty days. 

.

New Jersey     “theft.”    N.J.   Stat.   Ann. receiving stolen property or bringing
               § 2C:15-1(a).                   stolen property into New Jersey. N.J.
                                               Stat. Ann. § 2C:20-7(a).

New York       “larceny.”   N.Y. Penal Law “acquiring lost property,” 

.                   Law § 155.05(2)(b), and paying
                                           employees below the minimum wage,
                                           id. § 155.05(2)(f).

North Dakota   “theft.” N.D. Cent. Code Ann. receiving stolen property. N.D. Cent.
               § 12.1-22-01(1).              Code Ann. § 12.1-23-02(3).

Oregon         “theft.” Or. Rev. Stat. Ann. receiving stolen property.      Or. Rev.
               § 164.395(1).                Stat. Ann. § 164.095(1).

Pennsylvania   “theft.” 18 Pa. Stat. and Cons. receiving stolen property. 18 Pa. Stat.
               Stat. Ann. § 3701(a)(1).        and Cons. Stat. Ann. § 3925(a).
 No. 23-3466                              United States v. Cervenak                                       Page 61


 Texas                     “theft as defined in Chapter communicating trade secrets, Tex.
                           31.” 

(b)(3), selling
                           § 29.02.                     property with an obliterated serial
                                                        number, id. § 31.11(a)(2), and
                                                        manufacturing devices that are
                                                        designed to access encrypted signals,
                                                        id. § 31.13.

 Wyoming                   “theft.” 

- receiving property of another.                      Wyo.
                           2-401 (citing Wyo. Stat. Ann. Stat. Ann. § 6-3-402(a).
                           § 6-3-402).


         In sum, at least seventeen other states’ robbery statutes involve an umbrella theft offense
broader than ordinary theft because that element doesn’t necessarily entail taking something of
value from another. And note how similar many of these theft offenses are to Ohio’s broad array
of theft offenses. For example, many of the above states’ theft statutes encompass receiving
stolen property. So does Ohio’s. See 

(K)(1), 2913.51.13

         By predicating its robbery statute on a broad definition of a “theft offense,” Ohio hasn’t
“completely redefine[d]” the crime of robbery. Brown, 

. Were Ohio robbery’s
theft element not a categorical match with generic robbery’s theft element, that would mean that
reams of state robbery statutes don’t match up with generic robbery. Supreme Court precedent
forecloses such a conclusion. See Castleman, 

. After all, the Commission didn’t
promulgate such a “self-defeating” Guideline. Quarles, 

.

         Defining generic robbery to account for the broad statutes of many states also respects the
state legislatures that draft criminal codes. States are “independent sovereigns.” Medtronic, Inc.
v. Lohr, 

 (1996). They enact criminal laws to advance public safety and
safeguard individual liberty within their borders, not to help the federal sentencing regime run
smoothly. In other words, states don’t fashion their criminal codes with the categorical approach
in mind. Case in point: Ohio’s current robbery law dates to 1973, almost two decades before the

         13
           What’s more, our sister circuits have found the robbery statutes of many of these states to be categorical
matches for generic robbery. See, e.g., United States v. Stovall, 

, 759–61 (8th Cir. 2019) (Arkansas);
United States v. Harrison, 

, 1331–36 (11th Cir. 2023) (Georgia); United States v. McCants, 

, 424–29 (3d Cir. 2020) (New Jersey); United States v. Carey, No. 20-2723, 

, at *2 (3d Cir.
July 13, 2021) (Pennsylvania); Adair, 16 F.4th at 470–71 (Texas).
 No. 23-3466                              United States v. Cervenak                                      Page 62


Supreme Court in Taylor told courts to use the categorical approach in federal sentencing. So,
when the Commission provides a generic, undefined term like robbery, courts must incorporate
into that definition the many states that define robbery as a combination of a force element and
an umbrella “theft offense.” After all, those are the “basic elements” of a robbery, and courts
cannot get hung up on “technical definitions and labels” when determining its generic definition.
Taylor, 

; see Quarles, 

.14

                                                *        *        *

        Thus, because the individual theft offenses are mere means of committing a singular
“theft offense” element, Ohio robbery is a categorical match for generic robbery.

                III. Individual Theft Offenses as Elements: Guidelines Extortion

        The majority assumes that the individual theft offenses are themselves elements, not
means. Even if that were correct, it doesn’t lead to the majority’s conclusion that robbing two
people at gunpoint is not a violent crime.

        If the individual theft offenses are themselves elements, then the Ohio robbery statute
criminalizes thirty-one distinct robbery offenses.15 Because Ohio robbery indictments often
don’t charge the underlying theft offense element, the majority’s approach suddenly sets off
constitutional alarm bells: The Ohio and United States constitutions require an indictment to
charge each element of a crime.

        Consider the relevant constitutional provisions, both state and federal. Article I, Section
10 of the Ohio Constitution provides that “no person shall be held to answer for a capital, or

        14
           The majority’s definition of generic robbery flouts this guidance. Drawing on panel precedent, the
majority defines generic robbery to involve “misappropriation of property.” Maj. Op. at 18. This requires the
defendant to “unlawfully take or use the property of another.” 

 But the majority’s reading of that standard would
eliminate the robbery statutes of at least seventeen states from the generic definition, so it’s too narrow. See
Castleman, 

; Stokeling, 

.
            The majority also criticizes me for defining generic robbery based on the definitions of “many states.”
Maj. Op. at 21. But as the majority itself recognizes, the generic definition of robbery derives from sources “such
as . . . a survey of state law.” Id. at 4.
        15
           It’s likely twenty-nine. Two of the “theft offense” predicates of Ohio robbery are robbery and
aggravated robbery themselves. See 

(K)(1). It’s hard to imagine how a robbery
indictment could be predicated on a robbery or an aggravated robbery as the theft offense.
 No. 23-3466                         United States v. Cervenak                               Page 63


otherwise infamous, crime, unless on presentment or indictment of a grand jury.” Ohio Const.
art. I, § 10. In light of that provision, the Ohio Supreme Court has explained that an indictment
that doesn’t contain the elements of the crime charged “is defective and cannot be cured.” State
v. Headley, 

 (Ohio 1983). Otherwise, the defendant could be convicted “on
a charge essentially different from that found by the grand jury.” 

       On the federal side, the Sixth Amendment provides in part that “[i]n all criminal
prosecutions, the accused shall enjoy the right to . . . be informed of the nature and cause of the
accusation.” U.S. Const. amend. VI. This protection has been incorporated against the states by
the Fourteenth Amendment’s Due Process Clause. Gannett Co. v. DePasquale, 

 (1979). Similarly relevant is the Due Process Clause itself, which prevents states from
depriving people of their liberty “without due process of law.” U.S. Const. amend. XIV, § 1.

       Given these provisions, the Supreme Court has explained that an indictment is
constitutionally defective if it fails to “apprise the defendant with reasonable certainty, of the
nature of the accusation against him”—even if the indictment recites the applicable statutory
language. Russell, 

 (citation omitted). “In an indictment upon a statute, it is not
sufficient to set forth the offence in the words of the statute, unless those words of themselves
fully, directly, and expressly, without any uncertainty or ambiguity, set forth all the elements
necessary to constitute the offence intended to be punished.” 

 (citation omitted).

       An Ohio robbery indictment resting on an unspecified theft offense thus seems to violate
both the state and federal constitutions if the specific theft offense is itself an element. As for the
Ohio Constitution, the indictment doesn’t “contain[] the elements of the offense charged.”
Troisi, 206 N.E.3d at 703–04 (citation omitted). As for the United States Constitution, the
indictment doesn’t “fully, directly, and expressly, without any uncertainty or ambiguity, set forth
all the elements necessary to constitute the offence intended to be punished.” Russell, 

 (citation omitted). Maybe the defendant is accused of selling a device that’s manufactured
to gain access to cable TV. See 

(B). Or maybe the defendant is
accused of possessing a counterfeit telecommunications device with the purpose of using it
criminally. See 

 § 2913.06(A). Or maybe he’s accused of having issued a bad check. See id.
§ 2913.11(B). The list goes on.
 No. 23-3466                         United States v. Cervenak                            Page 64


         There were two ways to silence the constitutional alarm bells created by the majority’s
interpretation of Ohio’s robbery statute.

         First, the majority could have looked to state court interpretations and practice to see if
they resolve the constitutional problem.       They do:     Ohio courts presume that a robbery
indictment’s unspecified theft offense is ordinary theft. This presumption renders Ohio robbery
a categorical match with Guidelines extortion. It also avoids any grand jury and fair notice
concerns created by the majority’s interpretation of Ohio’s robbery law.

         Second, the majority could have certified the Ohio state law questions that are giving rise
to the constitutional conundrum: (1) whether the individual theft offenses are means or elements
and (2) whether an unspecified theft offense in an Ohio robbery indictments refers to ordinary
theft.

                     A. Ohio State Court Practice: The Theft Presumption

         Ohio courts presume that an unspecified theft offense in a robbery indictment refers to
run-of-the-mill theft.

                                    1. The Ohio Supreme Court

         The Ohio Supreme Court has equated “theft offense” with ordinary theft. In State v.
Smith, a defendant was found guilty of “theft” under 

, even
though she was indicted only for robbery. 

, 596–97 (Ohio 2008). The robbery
indictment alleged that the defendant had committed a predicate “theft offense”—but didn’t
specify which one. The Ohio Supreme Court held that the defendant’s conviction for ordinary
“theft” was proper because, in Ohio, theft is a lesser included offense of robbery. 

. The
court reasoned that, whatever the facts may be, “it would be impossible” to commit robbery
under § 2911.02 without also committing “theft” under § 2913.02. Id. (emphasis added).

         The majority looks to facts in the Smith indictment to claim that the defendant was
indicted for theft. But when the defendant moved for reconsideration, the Ohio Supreme Court
doubled down, explaining that the defendant was “not indicted for theft. Rather, the indictment
charged her with robbery, and the trial court convicted her of the lesser included offense of
 No. 23-3466                       United States v. Cervenak                           Page 65


theft.” State v. Smith, 

 (Ohio 2009). We must accept what the Ohio
Supreme Court says. See Mullaney v. Wilbur, 

 (1975). This reading also
makes sense. “Theft of retail merchandise from Macy’s” is a fact supporting the charge. It’s not
a crime in the Ohio Revised Code. All told, the Ohio Supreme Court said that the defendant
could be convicted of theft because her indictment for robbery meant she was also charged with
ordinary theft. See Smith, 905 N.E.2d at 151–52.

       State v. Smith controls Cervenak’s case. Cervenak, like the defendant in Smith, was
charged with robbery by “theft offense” under § 2911.02.        The Ohio Supreme Court has
explained that it’s “impossible” to commit robbery by theft offense under § 2911.02 without also
committing regular “theft” under § 2913.02. Smith, 884 N.E.2d at 599 (emphasis added). And
as all agree, “theft” under § 2913.02 necessarily requires “obtaining something of value from
another.” U.S.S.G. § 4B1.2 cmt. n.1 (2021). Thus, under the rule of State v. Smith, Cervenak
committed Guidelines extortion, a crime of violence.

       Smith isn’t a one-off. In State v. Horner, the defendant challenged his indictment’s
failure to specify what mental state was required for aggravated robbery. The Ohio Supreme
Court said that aggravated robbery “includes the element of attempting or committing a theft
offense, which incorporates all the elements of theft.” 

 (emphasis added).
Likewise, in State v. Thomas, the court again explained that a defendant commits robbery when
he “flee[s] immediately after a theft or an attempted theft.” 

, 1191 (Ohio 2005).

                                     2. Ohio Lower Courts

       What the Ohio Supreme Court has suggested, the Ohio Courts of Appeals have explicitly
confirmed: “theft” is an element of Ohio robbery when the indictment doesn’t specify the
predicate theft offense.   For example, in State v. Tellis, the court acknowledged that the
defendant’s robbery indictment “did not specify which theft offense the state accused [the
defendant] of committing.” 

, 842 n.3 (Ohio Ct. App. 2020). The court proceeded
 No. 23-3466                               United States v. Cervenak                                      Page 66


to “presume that the theft offense was theft.” 

 Cervenak’s situation is no different from the
situation confronted by the Ohio Court of Appeals in Tellis.16

         That similarity is critical: As the Supreme Court has made clear, “the Ohio courts have
the final authority to interpret . . . that State’s” law, and “the Ohio Court of Appeals” may
“authoritatively” construe state law.           Brown v. Ohio, 

 (1977) (quotation
omitted). So, in the absence of a state supreme court holding, we must respect the appellate
courts’ interpretation of state law as authoritative unless there’s an indication the state supreme
court would rule differently. Hisrich v. Volvo Cars of N. Am., Inc., 

, 449 n.3 (6th
Cir. 2000). And there’s no indication that the state supreme court would part ways with the
state’s lower courts: On the contrary, the Ohio Supreme Court itself has equated the “theft
offense” underlying robbery with regular “theft.” See Smith, 884 N.E.2d at 599. Accordingly,
our court must follow that appellate court’s lead—even if we believe its decision was “unsound.”
Ziebart Int’l Corp. v. CNA Ins. Cos., 

, 251 (6th Cir. 1996).

         And two types of lower-court practices—how trial courts instruct juries on robbery and
how appeals courts analyze sufficiency-of-the-evidence challenges to robbery convictions—also
reveal that Ohio courts equate unspecified “theft offenses” with “theft.”

         When Ohio’s trial courts hear cases in which the defendant is charged with a crime
involving a “theft offense,” but the state does not specify the theft offense, the courts charge
juries using the definition for generic theft. For example, an Ohio grand jury indicted the
defendant in State v. Tillison for robbery premised on an unspecified theft offense.                             See
Indictment at 2, State v. Tillison, No. 2016 CRC-I 000371 (Wayne Cnty. Ct. C.P. Nov. 21,




         16
            And in State v. Martin, the court of appeals recited the elements of Ohio aggravated robbery, including
its “theft offense” clause that mirrors that of Ohio robbery. No. 13CA010356, 

, at *2 (Ohio Ct.
App. Mar. 10, 2014). Having recited that definition, it then cited the definition of run-of-the-mill Ohio “theft.” 

So, the court jumped straight from “theft offense” to ordinary “theft” to explain what’s necessary to commit robbery.
Similarly, in State v. Frazier, in assessing a sufficiency-of-the-evidence challenge to a defendant’s robbery
conviction, the court equated a “theft offense” with “the actual taking of . . . property.” 

, 640 (Ohio
Ct. App. 2016).
 No. 23-3466                               United States v. Cervenak                                  Page 67


2016). The jury instructions defined the underlying “theft offense” as § 2913.02(A) theft. Jury
Instrs. at 4–5, 7 (Feb. 14, 2017). Tillison is no outlier in this respect.17

        The Ohio Supreme Court has implicitly endorsed this practice.                        In Tolliver, the
defendant’s robbery indictment didn’t specify his underlying theft offense. 

.
But the trial court instructed the jury on run-of-the-mill theft—including its “knowing[]” mental
state for obtaining property.        

    The Ohio Supreme Court accepted that the defendant’s
predicate theft offense was theft as the jury instructions specified. 

; see also 

(“R.C. 2911.02(A) predicates every robbery on the elements of a completed or an attempted
‘theft offense,’ which, for [the defendant], included the mental states of ‘purpose’ and
‘knowingly.’ R.C. 2913.02(A).”). And the court did so over a dissent critiquing its assumption
that ordinary theft necessarily applied, given that the “theft offense” statute has “a variety of
meanings” beyond theft alone. See 

 (Lanzinger, J., dissenting). What’s more, at least
one Ohio court of appeals has understood Tolliver to stand for the proposition that “every
robbery is predicated on a theft.” State v. Brock, 

, 1245 (Ohio Ct. App. 2019).

        Likewise, when Ohio appeals courts resolve defendants’ sufficiency-of-the-evidence
challenges to their robbery convictions, the courts ask whether the state offered sufficient
evidence of ordinary theft. See, e.g., United States v. Cervenak, 

, 860–61 (6th Cir.
2024) (collecting examples); State v. Holloway, No. 112129, 

, at *7 (Ohio Ct.
App. July 13, 2023) (finding sufficient evidence of the items that the defendant “took from [the
victim’s] possession”); Indictment at 3–4, State v. Holloway, No. CR-22-668155-A (Cuyahoga
Cnty. Ct. C.P. Mar. 4, 2022).18


        17
            See State v. Pigg, No. 04CA2947, 

, at *3–5 (Ohio Ct. App. Apr. 28, 2005) (upholding
the trial court’s jury instruction on § 2913.02(A) theft where the elements of the underlying theft offense were
otherwise unspecified); Jury Instr. at 14, State v. Stovall, No. CR17-3132 (Lucas Cnty. C.P. Ct. Mar. 7, 2018)
(instructing the jury that § 2913.02(A) “‘[t]heft’ is an essential element of robbery” even in a case where the
defendant was also charged with a different theft predicate—there, aggravated burglary); Indictment at 2–3, id.
(Dec. 11, 2017).
        18
           The majority doesn’t cite a single case to the contrary. It cites two cases where a qualifying “theft
offense” didn’t involve taking something of value. See Maj. Op. at 9–10 (citing State v. Stone, No. 2023-CA-23,

, at *1 (Ohio Ct. App. Jan. 19, 2024); State v. Dunn, Nos. 112301 & 112916, 

,
at *6 (Ohio Ct. App. Dec. 7, 2023)). Robbery wasn’t charged in either case. So, they say nothing whatsoever about
whether Ohio courts presume that an Ohio robbery based on an unspecified theft offense rests on theft.
 No. 23-3466                          United States v. Cervenak                          Page 68


                            B. The Impact of the Theft Presumption

        Ohio courts’ presumption that an unspecified theft offense in a robbery indictment refers
to ordinary theft (1) establishes that Ohio robbery is a categorical match for Guidelines extortion
and (2) avoids any constitutional concerns.

                               1. Matching with Guidelines Extortion

        The presumption that “theft offense” means ordinary theft confirms that Ohio robbery is
a categorical match with Guidelines extortion.

                                       i. Guidelines Extortion

        A crime punishable by more than one year of prison time is a “crime of violence” if its
elements are “the same as, or narrower than” the elements of Guidelines “extortion.” See
Descamps, 

; U.S.S.G. § 4B1.2(a)(2). The Guidelines define “extortion” as
“obtaining something of value from another by the wrongful use of (A) force, (B) fear of
physical injury, or (C) threat of physical injury.” U.S.S.G. § 4B1.2 cmt. n.1 (2021).

                                          ii. Ohio Robbery

        To assess the elements of Ohio robbery, our court is “bound by” the state supreme court’s
interpretation of the relevant state statute, “including its determination of the elements of” the
offense. Johnson v. United States, 

 (2010). In doing so, we keep in mind “that
state courts are the ultimate expositors of state law.” Mullaney, 

.

        So, to win, Cervenak must show that the Ohio courts have interpreted their robbery
statute as broader than Guidelines extortion (such that Ohio robbery isn’t a categorical match).
See Gonzales v. Duenas-Alvarez, 

 (2007). As the Supreme Court explained in
Duenas-Alvarez, Cervenak must provide evidence of this broader interpretation in “at least
. . . his own case or other cases.” 

 Thus, even if our court could “legal[ly] imagin[e]” conduct
under Ohio’s robbery statute that didn’t constitute Guidelines extortion, that isn’t enough. Id.;
see also Cradler v. United States, 

 (6th Cir. 2018) (Keith, J.) (explaining that
federal courts “cannot . . . look[] only to the words of the violated state statute”).
 No. 23-3466                              United States v. Cervenak                                      Page 69


        This requirement from Duenas-Alvarez is rooted in the categorical approach itself. For
our court to find a categorical match, it’s not necessary “that every conceivable factual offense
covered by” Ohio’s robbery statute aligns with Guidelines extortion. See James v. United States,

 (2007) (citing Duenas-Alvarez, 

), overruled on other grounds
by Johnson v. United States, 

 (2015). The question is whether, when an Ohio
robbery indictment doesn’t specify the underlying theft offense, “the ordinary case” of Ohio
robbery matches up with Guidelines extortion. 

 So, Cervenak must show that Ohio, on
average, “actually prosecutes” robbery with an unspecified theft offense in a way that’s broader
than Guidelines extortion. Moncrieffe v. Holder, 

 (2013).

        Requiring Cervenak to point to state court cases in support of his argument also accords
state courts the respect they deserve as “the final arbiters of state law in our federal system.”
United States v. Taylor, 

 (2022). And it recognizes that no matter how clear
the state statute’s text may appear “[o]n its face,” the state courts are free to “considerably
narrow[] its application.” James, 

.19

        The majority, for its part, claims that the defendant can meet the realistic probability
requirement by just pointing to the statutory text. But that’s not what the Supreme Court did in
Duenas-Alvarez. Indeed, the majority’s text-only approach flies in the face of Duenas-Alvarez,
James, Johnson, and Moncrieffe: when a federal court interprets a state law while applying the
categorical approach, it is “bound by” the state supreme court’s interpretation.                         Johnson,

. While it’s true that “the statute itself defines the elements,” Maj. Op. at 14, the
state courts tell us what the statute means—and thus what its elements are. We ought to listen to
them: states’ “authoritative tribunal[s]” often find “in a statute less than meets the outsider’s eye.
Informed local courts may find meaning not discernible to the outsider.” La. Power & Light Co.
v. City of Thibodaux, 

 (1959).




        19
           The deference to state courts’ narrowing constructions mandated by Duenas-Alvarez and James mirrors
the deference to state courts’ narrowing constructions in the First Amendment overbreadth context. See, e.g., R.A.V.
v. City of St. Paul, 

 (1992). Both forms of deference flow from the fact that state courts’
constructions of state laws are “binding” on federal courts. New York v. Ferber, 

, 769 n.24 (1982).
 No. 23-3466                              United States v. Cervenak                                      Page 70


        The majority tries to escape this line of precedent by citing Taylor. 596 U.S. at 858–59.
But Taylor considered whether Hobbs Act robbery—a federal statute—was a crime of violence.

. So, the Court itself was the final arbiter of the law’s meaning. Not so when we
interpret state law. As Taylor expressly recognized, we’re not the “final arbiter” of state law; the
state courts are. 

. Nothing in Taylor, then, supports the majority’s conclusion that our
inquiry can end at the plain text.20

        The majority’s invocation of Taylor is part of its broader text-only approach to Ohio’s
robbery statute. When interpreting federal laws, far be it from me to adopt anything other than a
textualist approach.        But when interpreting state laws, far be it from us to adopt any
interpretation that doesn’t square with that of the state courts.

        Thus, Cervenak must show that Ohio regularly prosecutes robbery in a way that’s broader
than Guidelines extortion. He can’t make that showing. Given that his robbery indictment
didn’t specify his theft offense, the question is whether Ohio robbery with an unspecified “theft
offense” can be committed without “obtaining something of value from another.” Because of
Ohio courts’ presumption that “theft offense” means ordinary theft, it can’t. Ordinary theft
involves taking something of value from another. 

(A). Thus,
Ohio robbery matches up with Guidelines extortion.

                                          iii. The Majority’s Cases

        The majority is unable to rebut this reading of Ohio’s caselaw. For starters, it can’t
escape Smith’s grip by pointing to State v. Castle. In that case, an Ohio lower court claimed:
“As receiving stolen property is a theft offense and it is impossible for an offender to commit a
robbery without committing a theft offense, it follows that a robbery cannot be committed
without also committing the offense of receiving stolen property.” State v. Castle, 

, 817 (Ohio Ct. App. 2017). Castle is correct that an offender can’t commit robbery without

        20
           The majority also cites to Mellouli v. Lynch, but that case too is unavailing. See Maj. Op. at 14 n.5
(citing 

 (2015)). The Court in Mellouli never passed judgment on the applicability of Duenas-Alvarez.
It thus didn’t “consider[] the issue and consciously reach[] a conclusion about” Duenas-Alvarez. See Wright v.
Spaulding, 

 (6th Cir. 2019). That silence is of scant persuasive value compared to the Court’s
repeated invocation of the need to look to state court interpretations of state law in Duenas-Alvarez itself, James,
Johnson, and Moncrieffe.
 No. 23-3466                               United States v. Cervenak                                      Page 71


also committing a theft offense, but that doesn’t mean that the offender necessarily committed
the specific theft offense of receiving stolen property. He could have committed a different theft
offense, such as ordinary theft. So, Castle can’t be right when it reasons that an offender who
commits robbery necessarily commits every single one of the thirty-one theft offenses
(including, among others, receiving stolen property). Aggravated robbery, for example, is a theft
offense. Under Castle’s logic, a defendant who commits robbery necessarily committed
aggravated robbery. Obviously, that logic doesn’t add up.

         Further, Smith, Horner, and Thomas provide “persuasive data that the [Ohio] Supreme
Court would decide otherwise.” Kingsley Assocs., Inc. v. Moll PlastiCrafters, Inc., 

 (6th Cir. 1995).21 Thus, Castle doesn’t move the needle.

         In rare instances—the majority cites two—Ohio courts don’t treat an unspecified “theft
offense” as ordinary theft. But even then, Ohio courts still require the government to show that
the robbery defendant took something of value, thereby satisfying the definition of Guidelines
extortion.

         In one of the majority’s cases, the court presumed that “theft offense” entails the
unauthorized use of another’s property. See State v. Swarthout, No. 10CA0107-M, 

, at *3 (Ohio Ct. App. Sept. 30, 2011). Unauthorized use involves using or operating
another’s property without consent. 

(A). That may not entail a
permanent deprivation of property, but a temporary one.                      For example, in Swarthout, the
defendant seized the victim’s phone and flipped through it for some time before leaving it at the
crime scene with the victim. See 

, at *1, *3.

         Such a temporary taking still entails “obtaining something of value from another.”
U.S.S.G. § 4B1.2 cmt. n.1 (2021). To “obtain” something is to “bring [it] into one’s own
possession,” even if not permanently. Obtain, Black’s Law Dictionary (10th ed. 2014). And
other courts have interpreted Guidelines extortion not to require a deprivation of property to be

         21
            Another wrinkle associated with the majority’s citation to Castle is that the state amended the charges to
include receiving stolen property under § 2913.51(A). See Charge, 2014 CR 00550, State v. Castle,
https://ecourts.mahoningcountyoh.gov/eservices/searchresults.page?x=OSxTm6C1*rNNdibNTFvmcKoe1sKhls8tk
Mi8l3NPynT7HjMaGcc8h76HhobykYc3tfeNGj7nviMNvPZAZ63gfQ.
 No. 23-3466                               United States v. Cervenak                                       Page 72


permanent. See United States v. Velasquez-Bosque, 

 (9th Cir. 2010) (holding
that extortion does not “contain any requirement that property be taken permanently”).
Therefore, even in the one instance when an Ohio court presumed a robbery’s unspecified theft
offense was unauthorized use (that is, a temporary taking), the robbery was still a categorical
match for Guidelines extortion.

         The lone other case the majority cites for the proposition that Ohio cases exist where
robbery is committed without taking something of value is State v. Dixon. No. 89-CA-22, 

, at *2–3 (Ohio Ct. App. June 20, 1990). In that case, the underlying theft offense was
forgery: the defendant tried to cash a forged check at a convenience store. 

. This
involved attempting to take something of value from another—that is, he tried to take cash from
the store.     Dixon does nothing to take Ohio robbery outside the purview of Guidelines
extortion.22

         In sum, among the parties’ sentencing memoranda before the district court, the parties’
briefing before the original three-judge panel, the parties’ en banc petition filings, the parties’ en
banc briefing, the three separate panel opinions, and the four opinions here, no one can identify a
single case of Ohio robbery that didn’t entail taking something of value from another.

         That failure should close this case. As a defendant arguing that Ohio’s law is broader
than Guidelines extortion, Cervenak bears the burden of showing a “realistic probability” that
Ohio robbery covers conduct that doesn’t fall within the federal crime to which it’s being
compared. Duenas-Alvarez, 

. Demonstrating “a realistic probability is not an
exercise in educated guessing.” United States v. Castillo-Rivera, 

 (5th Cir.
2017) (en banc). Cervenak had to come armed with evidence that “the reality of prosecutions”
for Ohio robbery covers more conduct than Guidelines extortion. Butts, 

 (citing

         22
            To be sure, the majority’s two cases that rely on unauthorized use and forgery instead of theft still raise
constitutional questions. Although the Ohio lower courts aren’t clear, unauthorized use could be a lesser included
offense of theft. Compare State v. Vrazalica, No. 84412, 

, at *4 (Ohio Ct. App. Mar. 17, 2005)
with State v. Mumford, No. 17-81-28, 

, at *2 (Ohio Ct. App. Feb. 4, 1983). If so, that would avoid
any constitutional defect and is just one more question the Ohio Supreme Court could answer if we had certified. As
to forgery, the single case the majority cites, State v. Dixon, was decided thirty-five years ago—well before Smith.
To the extent one decades-old case could have constitutional defects, that wreaks far less havoc on Ohio than the
majority’s solution. Of course, if we were willing to certify the question, the Ohio Supreme Court (the ultimate
arbiter of Ohio law) could resolve this issue.
 No. 23-3466                         United States v. Cervenak                              Page 73


Moncrieffe, 

). But Cervenak has provided no evidence that Ohio robbery
premised on an unspecified theft offense encompasses any conduct beyond that covered by
Guidelines extortion. So, under binding Supreme Court precedent, he loses.

                                2. Avoiding Constitutional Concerns

       The majority, for its part, refuses to recognize Ohio’s theft presumption. Instead, it
assumes that any one of thirty-one separate theft offenses can serve as a predicate for a robbery
indictment that doesn’t specify the theft offense.       In so doing, the majority manufactures
numerous constitutional problems. For one, if an indictment doesn’t specify a theft offense, and
there’s no presumption of theft, that indictment violates Ohio’s Grand Jury Clause by failing to
list the elements of the crime charged. See Ohio Const. art. I, § 10.

       Further, without a theft presumption, the defendant cannot have fair notice of the charges
being brought against him, thereby allowing him to prepare his defense. See U.S. Const. amend.
VI; id. amend. XIV, § 1. His theft offense could be ordinary theft, but it could also be one of the
more esoteric theft offenses listed in § 2913.01(K)(1). He might prepare a defense geared
towards one of the thirty-one theft offenses, only to find out at trial that the prosecutor is proving
up a different theft offense.

       By contrast, the presumption of regular theft gives the defendant fair notice of the
charges and enables him to prepare a defense. When confronted with an indictment’s allegation
that he had used or threatened force while committing a “theft offense,” the defendant knows that
he can retort that he had not in fact been stealing or trying to steal the victim’s property. He
won’t later be blindsided by a prosecutor proving he in fact had committed any one of §
2913.01(K)(1)’s theft offenses, because theft offense presumptively means ordinary theft.

       Thus, by not recognizing Ohio’s theft presumption, the majority runs roughshod over the
Ohio and United States constitutions.

       In response, the majority (1) misreads Ohio caselaw, (2) disregards Ohio statutes and
rules of criminal procedure, and (3) misunderstands constitutional avoidance.
 No. 23-3466                        United States v. Cervenak                            Page 74


                          i. The Majority’s Misreading of Ohio Caselaw

       The majority’s response proceeds in two steps. First, the majority notes that the Ohio
Supreme Court has said that indictments that track the relevant statutory language provide a
defendant with adequate notice of the crime charged. Second, the majority states that if the
indictment tracks the statute but omits an element—such as the specific theft offense—a bill of
particulars can cure that defect.

       At step one, the Ohio Supreme Court has never said that a valid indictment can omit an
element of the crime, even if it tracks the statutory language. Rather, the Ohio Supreme Court
has made clear that an indictment must “contain[] the elements of the offense charged.” Troisi,
206 N.E.3d at 703–04 (citation omitted). This rule is grounded in Article I, Section 10 of the
Ohio Constitution (which contains both a Grand Jury Clause and a Notice Clause) and the Notice
Clause of the Sixth Amendment. Id. at 703.

       An indictment can satisfy this requirement in different ways. For one, it can track the
language of the statute—that is, it can track the statutory language that lists the elements. State
v. Whitaker, 

 (Ohio 2022). Thus, in the context of robbery (and assuming,
as we do here, that the specific type of theft offense is itself an element), the indictment would
have to track the language of the specific theft offense. Alternatively, if the charged offense
relies on a predicate offense as an element, the indictment can refer to the statute number of the
predicate offense. Troisi, 

 (citing Buehner, 853 N.E.2d at 1164). In the
robbery context, the indictment would refer to the theft offense by statute number, such as
specifying § 2913.02 to indicate regular theft.

       Therefore, the majority is wrong to imply that an indictment suffices when it tracks the
statutory language, even if it omits a statutory element. If a predicate offense is an element, the
indictment must either recite the statutory language with all the elements or cite the predicate
offense by statute number.

       At step two, the majority is wrong to conclude that a bill of particulars can save an
indictment that doesn’t contain each element of the crime charged. A bill of particulars can put
meat on the bones of an indictment that fails to give the defendant “a fair and reasonable
 No. 23-3466                        United States v. Cervenak                             Page 75


opportunity to prepare his defense.” Id. (citation omitted). But it can’t save an indictment that is
not “legally sufficient” in the first place: the indictment itself must “describ[e] the elements of
the charged offense.” Id. (citation omitted).

       Given the Ohio Constitution’s Grand Jury Clause, it’s unsurprising that the Ohio
Supreme Court has never said that an indictment can fail to list the elements of a crime or that a
bill of particulars can cure such a failure.     Recall that Article I, Section 10 of the Ohio
Constitution guarantees that “no person shall be held to answer for a capital, or otherwise
infamous, crime, unless on presentment or indictment of a grand jury.” Ohio Const. art. I, § 10.
So, a grand jury indictment that doesn’t include the elements of the crime charged “is defective
and cannot be cured.” Headley, 

. As an Ohio appellate court put it: “It is
elementary that averments in a bill of particulars may not be used to cure fundamental defects in
an indictment.” State v. Gingell, 

 (Ohio Ct. App. 1982); accord State v.
Nelson, 

 (Ohio Ct. App. 2023); cf. Buehner, 853 N.E.2d at 1165 (Moyer,
C.J., dissenting) (noting that Article I, Section 10 of the Ohio Constitution “requires a grand jury
to consider every element of a charged offense before issuing an indictment”).

       Ohio’s Grand Jury Clause and caselaw thus make clear that a bill of particulars can detail
an indictment’s factual basis but can’t fix an indictment’s failure to contain the elements of the
crime charged.    To better grasp this point, consider a hypothetical.       A man commits two
robberies on January 1—robbery A and robbery B. The indictment states that the defendant
committed a robbery on January 1. The indictment contains the elements of robbery, but it
doesn’t provide supporting facts. Thus, the defendant does not know whether he’s being indicted
for robbery A or for robbery B. A bill of particulars can provide that clarity by laying out the
underlying facts in more detail. Then, the defendant can make an informed choice whether to
plead guilty or not. By contrast, if the indictment fails to “describ[e] the elements of the charged
offense” of robbery, it’s legally insufficient from the start; the grand jury did not indict the
defendant with the charged crime. Troisi, 

 (citation omitted). No bill of
particulars can save that robbery indictment. 

 If it could, then the prosecutor, who provides
the bill of particulars, would be indicting the defendant for the crime charged.             That’s
nonsensical under Ohio’s guarantee of an indictment by grand jury.
 No. 23-3466                              United States v. Cervenak                                      Page 76


        While the majority relies on the Ohio Supreme Court’s decision in State v. Skatzes to
support its novel reading of Ohio’s caselaw and Grand Jury Clause, that decision is not to the
contrary.     In Skatzes, the defendant was charged with kidnapping, among other charges.

, 232 (Ohio 2004).                 Ohio’s kidnapping statute criminalizes removing or
restraining someone for several listed purposes. 

(A)(1)–
(6). Critically, the enumerated purposes are means of committing kidnapping; they are not
elements. How do we know? The Ohio Supreme Court endorsed the lower court’s treatment of
the purposes “as alternative means to satisfy the mental element” of kidnapping. Skatzes, 819
N.E.2d at 237. Moreover, Ohio jury verdicts must be “unanimous.” Ohio Crim. R. 31(A). And
Skatzes held that the jury was “not required to unanimously agree upon any one purpose for [the]
kidnapping.” 819 N.E.2d at 238. Thus, the specific purpose underlying a kidnapping is not an
element.

        Now, consider the facts of Skatzes.                The indictment charged the defendant with
kidnapping in service of four distinct purposes—including to “facilitate the commission of any
felony.” 

(A)(2). The defendant took issue with the indictment
for not specifying the felony that his alleged kidnapping sought to facilitate. Skatzes, 819 N.E.2d
at 233. The Ohio Supreme Court rejected this challenge because the bill of particulars could
remedy the omission of that felony. Id. Of course it could—as one of the listed purposes,
facilitating the commission of a felony is a means of committing kidnapping. So, the indictment
didn’t have to specify the underlying felony. As the Ohio Supreme Court put it, the underlying
felony was a “basis for the grand jury’s findings,” not an element. Id. In sum, Skatzes does not
stand for the proposition that an indictment can omit an underlying offense when that offense is
an element of the crime charged. Rather, it stands for the uncontroversial point that an
indictment need not include every possible means of committing an offense.23


        23
           The majority also cites two Ohio lower court cases that upheld indictments that did not specify a
predicate offense. Of course, neither of these unreported lower court cases can upend the Ohio Supreme Court’s
consistent mandate that Ohio indictments “contain[] the elements of the offense charged.” Troisi, 206 N.E.3d at
703–04 (citation omitted).
         Moreover, both cases can be squared with Ohio Supreme Court precedent. In Scott, the defendant stabbed
the victim to death and stole his wallet. State v. Scott, No. 22745, 

, at *1 (Ohio. Ct. App. Apr. 30,
2010). The defendant challenged his aggravated robbery indictment on the grounds that it didn’t specify the
 No. 23-3466                                United States v. Cervenak                                        Page 77


         All in all, Ohio caselaw does not support the majority’s claim that an Ohio indictment can
be valid even if it doesn’t name all the charged crime’s elements. Rather, it supports the
opposite conclusion.

           ii. The Majority’s Disregard of Ohio Statutes and Rules of Criminal Procedure

         The majority’s approach is also at odds with Ohio’s statutes and rules of criminal
procedure. The Ohio statute providing for bills of particulars doesn’t allow a bill of particulars
to cure an indictment that fails to list all the elements of a crime. Under Ohio law, a bill of
particulars may detail “the nature of the offense charged and the conduct of the defendant which
is alleged to constitute the offense.” 

. Ohio Rule of Criminal
Procedure 7(E) says the same. Ohio Crim. R. 7(E). So, although a bill of particulars can provide
details regarding the nature of the offense charged, the indictment must charge the offense in the
first place. And to do that, the indictment must contain all the elements. Troisi, 206 N.E.3d at
704.

                                          iii. Constitutional Avoidance

         At bottom, a bill of particulars cannot cure an indictment that fails to comply with Ohio’s
Grand Jury Clause. See Ohio Const. art. I, § 10. That should be enough to give the majority
pause before it insists that a bill of particulars can cure an invalid indictment. After all, the Ohio
Supreme Court, not the Sixth Circuit, is the ultimate expositor of Ohio’s constitution.




underlying theft offense. Id. at *8. The court rejected his challenge. It reasoned that the indictment adequately
tracked the words of the statute and that a bill of particulars could supply additional details. Id. Scott can be read to
rightly (though implicitly) conclude that the specific theft offense is a mere means of committing the theft offense
element. See supra Part II.A. After all, Scott relied on State v. Landrum, 

, 724–25 (Ohio 1990),
which in turn cited an Ohio Supreme Court case holding that an indictment is “defective and cannot be cured” if it
fails to contain the elements of the charged crime. Headley, 453 N.E.2d at 720. Alternatively, the theft presumption
could’ve been at work in Scott. Cf. Tellis, 165 N.E.3d at 842 n.3.
        The majority also cites State v. Vaughn. There, the court found that an indictment for “theft in office,” see

(A)(2), that didn’t specify the underlying theft offense wasn’t defective. See No.
10CA009820, 

, at *3 (Ohio Ct. App. Jan. 31, 2011). If the specific theft offense is a means, then
Vaughn poses no problem. If it’s an element, then Vaughn was incorrect: as the dissenting judge put it, “the State
should have identified the predicate theft offense.” Id. at *6 (Belfance, P.J., dissenting). Still, because the Ohio
Supreme Court is the ultimate expositor of Ohio law, this unreported lower court case cannot overcome the Ohio
Supreme Court precedents discussed above.
 No. 23-3466                         United States v. Cervenak                            Page 78


       As for the due process and Sixth Amendment concerns raised by an indictment that
doesn’t specify a predicate theft offense, I agree with the majority that a bill of particulars can
cure those defects. But the existence of a potential cure does not change the majority’s disregard
for constitutional avoidance.     This principle of interpretation—dating back to founding-era
Supreme Court precedent, see Mossman v. Higginson, 

 (1800)—is about not
creating a constitutional issue in the first place. As Chief Justice Marshall put it: “The question,
whether a law be void for its repugnancy to the constitution, is, at all times, a question of much
delicacy, which ought seldom, if ever, to be decided in the affirmative, in a doubtful case.”
Fletcher v. Peck, 

 (1810). The fact that Ohio might have procedures
to cure some of the constitutional issues that our interpretation creates is of no moment; that the
state may be able to clean up our mess says nothing about our duty to avoid creating the mess to
begin with.     Thus, the only interpretation of Ohio’s robbery statute that accords with
constitutional avoidance is to presume an unspecified theft offense refers to theft (assuming, of
course, that each theft offense is an alternative element).

                                          *       *       *

       In the end, the majority is determined to read Ohio’s robbery indictments in a way that
leaves them defective as a matter of both state and federal constitutional law. First, the majority
reasons that the specific theft offense underlying a robbery indictment is an element as opposed
to a means of committing robbery. If true, that’s a big problem, since an indictment must contain
the elements of the charged crime. Luckily, Ohio caselaw offers a solution to the majority’s self-
inflicted constitutional conundrum: the theft presumption. Yet the majority doubles down,
papering over reams of Ohio cases that apply the theft presumption and allay constitutional
concerns.

       The best explanation for the majority’s persistence is that it doesn’t believe it’s creating
any constitutional issues. The most unfortunate consequence of that mistaken belief is that it
leads the majority to forgo the last defensible way to resolve this case and the one chosen by our
sister circuits when facing similar uncertainty: certification.
 No. 23-3466                        United States v. Cervenak                            Page 79


                                          IV. Certification

       Even if the majority refuses to follow Ohio courts’ interpretation of state law, we could
have at least asked the Ohio Supreme Court for clarification. United States Supreme Court
precedent instructs us to certify state law questions to the state supreme court to avoid
constitutional concerns—like the fair notice concerns that spring from the majority’s
understanding of Ohio law. Other circuits follow the Supreme Court’s directions and have
regularly certified questions in analogous situations. Our court should have followed their lead.

       First, we could have certified whether the individual theft offenses that underlie Ohio
robbery convictions are different means of committing a singular theft offense or are themselves
alternative elements of robbery. If the individual theft offenses are means, that would cure any
constitutional concerns with Ohio robbery indictments. Second, if each theft offense is an
alternative element of robbery, we could have asked the Ohio Supreme Court whether it meant
what it said in Smith: that robbery entails theft. That too would resolve any constitutional
issues, as explained above.

       The majority’s unwillingness to certify is particularly troubling in this context.
Normally, we have no choice but to definitively rule on thorny legal issues. But today, we didn’t
have to do that. This case turns on multiple intricacies of Ohio law. Certification allows us to
turn to the state supreme court to resolve those nuances. Accordingly, the majority should have
recognized its own limitations and asked the Ohio Supreme Court for help—especially since its
ruling has constitutional implications.

                                  A. Supreme Court Precedent

       The Supreme Court has directed us to certify state-law questions to state courts before
resolving those questions in a way that raises constitutional concerns.

       Take Mckesson v. Doe, 

 (2020) (per curiam). There, the dispositive question
of state law was whether Louisiana tort law permitted a court to hold a protest organizer liable
for injuries that protest attendees caused. 

 at 3–4. If the answer was “yes,” there might have
 No. 23-3466                         United States v. Cervenak                             Page 80


been First Amendment issues. Without consulting Louisiana, the Fifth Circuit initially answered
“yes.” 

.

       The Supreme Court vacated the Fifth Circuit’s judgment. In doing so, the Supreme Court
admonished that the “Fifth Circuit should not have ventured into so uncertain an area of tort
law—one laden with value judgments and fraught with implications for First Amendment
rights—without first seeking guidance on potentially controlling Louisiana law from the
Louisiana Supreme Court.” 

. So too here. The Sixth Circuit should not venture into an
area of Ohio criminal law—especially one fraught with state constitutional, due process, and
Sixth Amendment implications—without first consulting the Ohio Supreme Court.

       To be sure, the source of the constitutional concern here isn’t a state statute, but a state’s
indictment practice. But that poses no bar to applying the well-trod principle of constitutional
avoidance. It didn’t stop the Supreme Court from doing so in Mckesson, where the source of the
concern wasn’t a statute, but the common law.

       The Supreme Court was similarly insistent about the need to certify before creating
constitutional pitfalls in Arizonans for Official English v. Arizona, 

 (1997). There,
Arizona passed a constitutional amendment making English the official state language, including
“the language of . . . all government functions and actions.” 

 (citation omitted). A state
employee sued.     

.    Throughout the litigation, Arizona’s Attorney General sought
certification to the Arizona Supreme Court to obtain an authoritative construction of the
amendment that could save it from constitutional concerns. 

. The lower courts refused,
instead finding the relevant state amendment to be “clear.” See Yniquez v. Arizonans for Off.
Eng., 

, 929 (9th Cir. 1995), vacated sub nom. Arizonans for Off. Eng. v. Arizona, 

 (1997). But the Supreme Court made clear that certification was the proper approach
when a saving construction was possible. Arizonans for Off. Eng., 

. So, the Court
vacated the lower court’s judgment and allowed the Arizona Supreme Court to issue a definitive
ruling on the construction of the state law. 

.

       This procedure makes sense. After all, a court’s “task is to seek harmony, not to
manufacture conflict,” between laws and the Constitution. United States v. Rahimi, 602 U.S.
 No. 23-3466                              United States v. Cervenak                                      Page 81


680, 701 (2024) (citation omitted). And that’s no less true when the law at issue is state law.
See Soc’y for the Propagation of the Gospel, 22 F. Cas. at 766 (Story, J.) (stating that the federal
courts “cannot but entertain the most entire respect and confidence” in state courts); Defiance
Water Co. v. City of Defiance, 

 (1903) (noting “the presumption” that “in all
cases . . . the state courts will do what the Constitution and laws of the United States require”);
Fox v. Washington, 

 (1915) (Holmes, J.) (“So far as statutes fairly may be
construed in such a way as to avoid doubtful constitutional questions they should be so
construed; and it is to be presumed that state laws will be construed in that way by the state
courts.” (citation omitted)); DeVillier v. Texas, 

 (2024) (“We should not
‘assume the States will refuse to honor the Constitution,’ . . . because ‘States and their officers
are [also] bound by obligations imposed by the Constitution.’” (alteration in original) (quoting
Alden v. Maine, 

 (1999))).

        And this analysis says nothing of the potential state constitutional concerns raised by the
majority’s approach. Our court must avoid creating any such problems. See Soc’y for the
Propagation of the Gospel, 22 F. Cas. at 769 (Story, J.). So, the state constitutional angle is yet
another reason to certify.

        Instead, the majority takes it upon itself to incorrectly answer questions of Ohio law. The
majority decides that Ohio didn’t intend robbery to be a crime of violence. The majority decides
to flout the Ohio Supreme Court’s decision in State v. Smith and hold that Ohio robbery doesn’t
always require taking something of value from another. The majority decides that an unspecified
“theft offense” could refer to any of the thirty-one enumerated offenses in § 2913.01(K)(1), when
Ohio’s caselaw shows otherwise. And the majority implicitly concludes that reams of Ohio
robbery indictments are constitutionally insufficient.24

                                B. Other Circuits’ Certification Practice

        Worse still, in defying Supreme Court precedent and misinterpreting state law, our court
is in a league of its own: our sister circuits have often certified questions of state law when

        24
           If they’re not constitutionally deficient, that means the individual theft offenses are themselves means—
which in turn means that Ohio robbery is a categorical match with generic robbery. See supra Part II.
 No. 23-3466                         United States v. Cervenak                            Page 82


wading through the categorical approach’s thicket. See United States v. Franklin, 

 (7th Cir. 2018) (per curiam) (certifying a means-versus-elements question of state law in
light of significant “practical consequences” for federal sentencing and state criminal procedure),
certifying question to 

 (Wis. 2019); United States v. Conage, 

 (11th Cir. 2020) (seeking clarification on how Florida defined “purchase” in a drug
trafficking statute), certifying questions to 

 (Fla. 2022); United States v. Glispie,

 (7th Cir. 2019) (asking whether the limited-authority doctrine applies to
Illinois residential burglary), certifying question to 

 (Ill. 2020); United States v.
Lawrence, 

 (9th Cir.), certifying questions to 

 (Or. 2018), on
reconsideration, 

 (Or. 2019) (en banc) (accepting, and then declining, questions of
whether Oregon robbery was divisible).

       And they have done so even when “[t]he most natural reading” of the relevant state law
cuts in one direction. Somers v. United States, 

 (11th Cir. 2021), certifying
questions to 

 (Fla. 2022). Because the definitive interpretation of state law is up
to state courts—not us, no matter how clear we may think the text of the state law is—other
circuits have “respectfully [sought] the guidance” of the relevant state supreme court. Id. at
1056. That deference squares with “the fundamental principle that we are not free to substitute
our own interpretations of state statutes for those of a State’s courts.” Schad v. Arizona, 

 (1991) (plurality opinion). Even if we couldn’t bring ourselves to reach a defensible
answer, we didn’t have to lurch for the indefensible.

       In the end, state legislatures don’t write their criminal laws with the categorical approach
in mind. Certain drafting practices can seem efficient and useful to state legislators. But they
can create issues when we cram them into federal sentencing; it’s like trying to fit a square peg in
a round hole. So, if the plain text and state caselaw don’t point in the same direction, and our
decision has significant practical consequences and raises serious constitutional concerns, we
should ask the state for a definitive answer.
 No. 23-3466                        United States v. Cervenak                           Page 83


                                        *       *      *

       Not all explanations are excuses. But for lower court judges, unsavory precedents can
sometimes offer both. At times, “precedent made me do it” can explain and excuse holdings that
otherwise seem unjust or just plain wrong. See, e.g., Nat’l Republican Senatorial Comm. v.
FEC, 

, 395 (6th Cir. 2024) (en banc).

       But precedent didn’t make the majority conclude that robbing two people at gunpoint
isn’t a crime of violence. In fact, precedent counseled against it. Precedent makes clear that,
whether individual theft offenses are “means” of committing robbery’s “theft offense” element
or are themselves “elements,” Tyren Cervenak committed a crime of violence when he held two
people up at gunpoint. If they’re means, then Ohio robbery is a categorical match for generic
robbery. If they’re elements, then Ohio robbery is a categorical match for Guidelines extortion.

       The majority’s choice is not without real-world consequences.         By concluding that
robbery indictments resting on an unspecified theft offense may rest on any of thirty-one distinct
offenses, the majority risks opening up countless Ohio robbery indictments (and in turn,
convictions) to constitutional challenges.     If those challenges succeed, countless violent
criminals in Ohio may serve less time in prison and spend more time on the streets. The majority
accuses me of resorting to “scare tactics” by worrying about these consequences. Maj. Op. at 22.
If following Supreme Court and Ohio precedent—and then pointing out the practical effects of
ignoring that precedent—amount to “scare tactics,” then the majority is right. At the end of the
day, we should take “every reasonable precaution” to know we’re right on the law before letting
violent criminals out of jail sooner. Brown v. Plata, 

 (2011) (Alito, J.,
dissenting).

       Faced with these potentially disastrous real-world consequences, the majority should
have at least certified the question to the Ohio Supreme Court rather than rip the issue out of
Ohio’s hands. “It is hard to fathom . . . why [our court] would countenance such an outcome so
divorced from reality”—especially when we had the opportunity to follow our sister circuits and
ask Ohio for answers. Taylor, 596 U.S. at 863 (Thomas, J., dissenting).
 No. 23-3466                       United States v. Cervenak                          Page 84


       The categorical approach may compel us to shield our eyes from reality, but it didn’t
compel us to blind ourselves to controlling Supreme Court caselaw. The categorical approach
may force us to cover our ears when common sense tells us that violent crimes are crimes of
violence, but it didn’t force us to ignore both how a mass of states define robbery and how the
Ohio Supreme Court interprets Ohio law. I respectfully dissent.