United States of America v. Danny Turner

In the
United States Court of Appeals
For the Seventh Circuit
____________________
No. 21‐2309
U NITED STATES OF A MERICA ,
Plaintiff‐Appellee,
v.
D ANNY T URNER ,
Defendant‐Appellant.
____________________
Appeal from the United States District Court for the
Western District of Wisconsin.
No. 3:20‐cr‐00038‐jdp‐1 — James D. Peterson, Chief Judge.
____________________
A RGUED A PRIL 8, 2022 — D ECIDED A UGUST 25, 2022
____________________
Before WOOD, H AMILTON, and J ACKSON‐A KIWUMI, Circuit
Judges.
HAMILTON , Circuit Judge. This appeal presents a new vari‐
ation on the familiar “categorical approach” to a defendant’s
prior criminal convictions under the Armed Career Criminal
Act (ACCA), 18 U.S.C. § 924(e). The question is whether a cat‐
egorical mismatch between a state criminal statute and the
federal recidivist statute can be based on physically impossi‐
ble conduct. Defendant Danny Turner was sentenced under

-- 1 of 32 --

2 No. 21‐2309
the ACCA based in part on two prior convictions under a Wis‐
consin drug trafficking statute. He contends that the Wiscon‐
sin statute sweeps more broadly than the definition of a “se‐
rious drug offense” under the ACCA because the state law
makes it a crime to deal in substances that the federal law does
not reach. The evidence before the district court shows, how‐
ever, that the supposed overbreadth concerns only substances
that, as a matter of chemistry, do not exist and cannot possibly
exist.
Like the Ninth Circuit in a similar case, “we opt for scien‐
tific reality over abstract legal doctrine.” United States v. Rodri‐
guez‐Gamboa, 972 F.3d 1148, 1150 (9th Cir. 2020). Under the
ACCA, a categorical mismatch cannot be based on truly im‐
possible conduct. Wisconsin’s drug statute does not expand
the scope of conduct actually treated as criminal beyond the
definition in the ACCA, despite superficial textual differ‐
ences. We therefore affirm Turner’s sentence as an armed ca‐
reer criminal.
I. Factual and Procedural Background
Defendant Turner was arrested after making four sales of
cocaine and one sale of heroin to an undercover police officer.
The arresting officers found a loaded handgun in Turner’s
waistband. A federal grand jury indicted Turner on eight
counts: six counts for distributing and possessing controlled
substances in violation of 21 U.S.C. § 841(a)(1); one for being
a felon in possession of a firearm in violation of 18 U.S.C.
§ 922(g)(1); and one for possessing a firearm in furtherance of
a drug trafficking crime in violation of 18 U.S.C. § 924(c). At
trial, Turner was convicted on all eight counts.

-- 2 of 32 --

No. 21‐2309 3
Turner’s conviction for being a felon in possession of a fire‐
arm would ordinarily carry a maximum sentence of ten years.
See § 924(a)(2). The district court found, however, that Turner
qualified as an armed career criminal under the ACCA, so he
faced a mandatory minimum sentence of fifteen years and a
maximum of life in prison. See § 924(e)(1). A defendant meets
that classification if his “prior criminal record includes at least
three convictions for ‘serious drug offense[s]’ or ‘violent
felon[ies].’” Shular v. United States, 140 S. Ct. 779, 783 (2020)
(alterations in original), quoting § 924(e)(1). The ACCA de‐
fines a “serious drug offense” to include an offense under
state law “involving manufacturing, distributing, or pos‐
sessing with intent to manufacture or distribute, a controlled
substance [as defined in 21 U.S.C. § 802], for which a maxi‐
mum term of imprisonment of ten years or more is prescribed
by law.” 18 U.S.C. § 924(e)(2)(A)(ii).
To determine whether a prior conviction qualifies as a se‐
rious drug offense or a violent felony under the ACCA, courts
apply what is known as the “categorical approach,” which
stems from Taylor v. United States, 495 U.S. 575 (1990). See John‐
son v. United States, 24 F.4th 1110, 1116 (7th Cir. 2022). The cat‐
egorical approach “focuses only on the elements of the crime
of conviction, not the actual facts of the defendant’s convic‐
tion.” United States v. Williams, 931 F.3d 570, 575 (7th Cir.
2019). If the elements of the crime of conviction sweep more
broadly than the definition of a “serious drug offense” in the
ACCA, “so that it is possible to violate the underlying statute
without committing a ‘serious drug offense’ within the mean‐
ing of the ACCA, then a conviction under the statute cannot
serve as a predicate ‘serious drug offense’ under the ACCA.”
Id. (emphasis added).

-- 3 of 32 --

4 No. 21‐2309
Before sentencing in this case, the presentence report rec‐
ommended that Turner be deemed an armed career criminal
under § 924(e) based on three prior drug convictions: two
state convictions related to trafficking cocaine and one federal
conviction for distributing crack cocaine. See Wis. Stat.
§§ 961.41(1)(cm)1 & (1m)(cm)1 (1997–98); 21 U.S.C.
§ 841(a)(1). Turner objected. He asserted that the state convic‐
tions could not qualify as predicate offenses because the Wis‐
consin drug statute is categorically broader than the ACCA’s
definition of a “serious drug offense.”
Turner identified two apparent mismatches between the
ACCA definition and the Wisconsin drug laws under which
he was convicted. The Wisconsin statute extends to dealing in
(1) narcotic analogs of cocaine and (2) esters and salts of esters
of cocaine. The government acknowledged the apparent tex‐
tual mismatches, but it argued the mismatches were “in name
only” because the allegedly overbroad conduct is factually
impossible. Relying on two declarations from expert chem‐
ists, the government explained that it is chemically impossible
for cocaine to have a “narcotic” effect and impossible to create
an ester or a salt of an ester of cocaine. Also, one of the gov‐
ernment’s experts testified in his declaration that a “narcotic
effect” is a type of “depressant effect.” If that is correct, then
even if cocaine could have a narcotic effect, federal law would
still apply to trafficking in it since federal law covers cocaine
analogs with depressant effects. See 21 U.S.C. § 802(32). The
government argued that the state statute’s nominal inclusion
of narcotic analogs and esters of cocaine thus did not actually

-- 4 of 32 --

No. 21‐2309 5
broaden the scope of prohibited conduct beyond the scope of
the ACCA definition of a serious drug offense.1
Our case law has expressly left the door open to just these
sorts of arguments of factual impossibility under the categor‐
ical approach. See United States v. Ruth, 966 F.3d 642, 648 (7th
Cir. 2020) (“There may be an occasion where a state statute
covers unquestionably nonexistent conduct, but we do not
need to predetermine how that analysis will look.”); see also
United States v. De La Torre, 940 F.3d 938, 952 n.5 (7th Cir. 2019)
(“Our opinion takes no position on the scientific merits, nor
should it be read as limiting the government’s ability to pre‐
sent [factual impossibility] argument[s] in future proceed‐
ings.”). The government relied on this point in Ruth in making
its arguments in the district court.
Turner responded by twice asking for time to consult an
expert in hopes of refuting the government’s experts in an ev‐
identiary hearing. The district court granted both of his exten‐
sion requests and scheduled a hearing. Two months after his
first request, though, Turner changed course. He told the dis‐
trict court that he did not plan to call witnesses or present ev‐
idence related to factual impossibility. The parties then filed a
joint motion to vacate the evidentiary hearing, which the
court granted.
In a written opinion prior to sentencing, the district court
overruled Turner’s objection to his classification as an armed
career criminal. Relying on the unrebutted testimony from the
1 In the district court, Turner also argued that the inclusion of analogs
of cocaine in the Wisconsin statute created a categorical mismatch, but the
court rejected that argument. On appeal, Turner has not challenged that
aspect of the court’s decision.

-- 5 of 32 --

6 No. 21‐2309
government’s experts, the court found that esters of cocaine
and narcotic analogs of cocaine simply do not exist. The court
also found that even if a narcotic analog of cocaine could exist,
the ACCA definition of a serious drug offense would reach it
under the Controlled Substances Act, 21 U.S.C. § 802, because
a narcotic effect is a type of depressant effect. Because the two
apparent mismatches did not actually expand the scope of
criminalized conduct under state law beyond the ACCA def‐
inition, Turner’s state convictions qualified as serious drug of‐
fenses under the ACCA. The district court then imposed a to‐
tal sentence of 20 years for Turner’s convictions. Turner has
appealed. 2
II. Analysis
We review de novo the district court’s legal determination
that Turner had three prior serious drug offenses and there‐
fore qualified as an armed career criminal. United States v.
Lockett, 782 F.3d 349, 352 (7th Cir. 2015). The district court’s
factual findings related to esters and narcotic analogs of co‐
caine are subject to review for clear error, id., though the evi‐
dence before the court was not disputed.
Turner’s two prior state drug convictions were under Wis‐
consin Statutes §§ 961.41(1)(cm)1 & (1m)(cm)1.3 As explained,
2 The ACCA’s mandatory minimum made up the first fifteen years,
and Turner’s § 924(c) conviction required a five‐year consecutive manda‐
tory minimum. The court imposed a 97‐month sentence on each of the six
drug counts to run concurrently with each other and with the other two
sentences.
3 Section 961.41(1)(cm)1 bars possession with the intent to traffic co‐
caine or a controlled substance analog of cocaine. Section 961.41(1m)(cm)1
bars trafficking cocaine or a controlled substance analog of cocaine. For
our purposes, the relevant parts of the Wisconsin statute are the

-- 6 of 32 --

No. 21‐2309 7
we apply the categorical approach here, focusing only on the
elements of the Wisconsin drug law—not the actual facts of
Turner’s prior convictions—to determine if those elements
sweep more broadly than the definition of a “serious drug of‐
fense” in the ACCA. Williams, 931 F.3d at 575. Turner identi‐
fies two apparent textual mismatches between that statute
and the ACCA definition: (1) narcotic analogs of cocaine and
(2) esters and salts of esters of cocaine. Factual impossibility
is dispositive for only one of those apparent mismatches.4
definitions of cocaine and its analogs, see §§ 961.16(2)(b) & 961.01(4m), not
the differences between possessing with the intent to distribute and actu‐
ally distributing. If there is an apparent mismatch between the definition
of cocaine or controlled substance analog in the state statute as compared
to the ACCA definition, that mismatch would apply equally to each of
Turner’s prior state convictions. As a result, we adopt the parties’ ap‐
proaches and focus on the provision that bars trafficking cocaine rather
than the provision that prohibits possession with intent to distribute.
4 In his opening brief on appeal, Turner also raised an apparent mis‐
match based on the Wisconsin statute’s inclusion of “any salt, compound,
derivative or preparation of coca leaves.” Wis. Stat. § 961.16(2)(b). But fed‐
eral law covers those substances. Compare id. with 21 U.S.C. §§ 812(c) n.1
& 811 (authorizing attorney general to promulgate regulations identifying
controlled substances), and 21 C.F.R. § 1308.12(b)(4) (identifying “any salt,
compound, derivative or preparation of coca leaves” as a Schedule II con‐
trolled substance). Turner conceded this point in his reply brief.
For the first time on appeal, Turner offers an additional apparent mis‐
match related to narcotic analogs of cocaine. Wisconsin law defines ana‐
logs to include those in which the defendant “represents or intends” the
substance as having a narcotic effect. Wis. Stat. § 961.01(4m)(a)2. The
ACCA definition does not. Turner did not raise this argument in the dis‐
trict court. Even assuming for the sake of our analysis that this argument
was not waived and that Turner can invoke plain‐error review, he cannot
meet that demanding standard. See United States v. Jones, 22 F.4th 667, 675
(7th Cir. 2022) (“To prevail on plain error review, [a defendant] must show

-- 7 of 32 --

8 No. 21‐2309
We proceed as follows. First, we analyze the apparent tex‐
tual mismatches between the Wisconsin drug statute and the
ACCA’s definition of a serious drug offense. We conclude that
the district court’s factual findings related to narcotic analogs
and esters of cocaine were not clearly erroneous. Then, we
turn to factual impossibility under the categorical approach.
We conclude that categorical mismatches cannot be based on
impossible conduct. Finally, we deny Turner’s alternative re‐
quest that we order a limited remand for an evidentiary hear‐
ing in which he could challenge the government’s evidence of
factual impossibility. The district court was prepared to hold
such a hearing, but Turner deliberately chose not to pursue it.
A. Cocaine Analog with a “Narcotic” Effect
The first apparent mismatch applies to a cocaine analog
with a supposedly “narcotic” effect. The Wisconsin drug law
criminalizes dealing in an analog of cocaine, defined as a
chemically similar substance that has a “stimulant, depres‐
sant, narcotic or hallucinogenic effect.” Wis. Stat.
§ 961.01(4m)(a)1. Under the federal definition used in the
ACCA, by comparison, a “controlled substance analogue” in‐
cludes a chemically similar substance with a “stimulant, de‐
pressant, or hallucinogenic effect.” 21 U.S.C. § 802(32). Turner
(1) an error, (2) that was plain, (3) that affected his substantial rights, and
(4) that seriously affected the fairness, integrity, or public reputation of the
proceedings.”). First, it is not clear that the district court even committed
an error. A defendant could represent an analog of cocaine as having a
narcotic effect in a street sale—even if that is factually impossible—and
still be convicted under federal or Wisconsin law since, as we explain be‐
low, a narcotic effect is a type of depressant effect. Second, even if there
was an error, it could not qualify as “plain” because of the debate about
whether a narcotic effect should be deemed a subset of depressant effects.

-- 8 of 32 --

No. 21‐2309 9
argues that the state law’s application to a cocaine analog with
a “narcotic” effect creates a textual mismatch because a de‐
fendant could be convicted for distributing such drugs under
Wisconsin state law, but not under federal law.
The district court rejected that argument, and we do too.
The court relied on the government’s expert to reach its con‐
clusion. As a preliminary matter, there is a pharmacological
question of whether cocaine or a cocaine analog can produce
a “narcotic” effect at all. According to the government’s ex‐
pert testimony, “cocaine is not a narcotic.” Narcotics are de‐
fined to have depressant effects on the central nervous sys‐
tem. Cocaine, by contrast, is classified as a stimulant and pro‐
duces a “stimulatory” effect on the central nervous system. It
is therefore chemically impossible for cocaine to have a truly
narcotic effect.
We need not decide that issue on that basis, however. As
the district court explained, “there is a more fundamental
problem with Turner’s argument.” According to the undis‐
puted expert testimony, a narcotic effect is a type of depres‐
sant effect. That means that even if cocaine could have a “nar‐
cotic effect,” it would be a subset of “depressant effects.” In
that sense, a defendant could be convicted in federal court of
distributing an analog of cocaine that had a supposedly nar‐
cotic effect because that conduct would fall under the classifi‐
cation of distributing a cocaine analog with a depressant ef‐
fect, which is treated as criminal under federal law.
Turner argues that we should not deem a narcotic effect a
subset of depressant effects. He relies on the text of the Wis‐
consin statute, arguing that since the state legislature chose to
list “depressant” and “narcotic” separately, the two terms
must have different meanings and capture distinct types of

-- 9 of 32 --

10 No. 21‐2309
conduct. This textual argument has its roots in the anti‐sur‐
plusage canon, under which courts read statutory text with
an eye toward avoiding rendering any language superfluous.
See, e.g., In re Southwest Airlines Voucher Litig., 799 F.3d 701,
710 (7th Cir. 2015).
There are two problems with this argument. First, Turner
has not articulated an interpretation of narcotic effect that
would not also be considered a depressant effect from a phar‐
macological standpoint. In that sense, he has failed to supply
a plausible interpretation of “narcotic” and “depressant” that
would separate the two terms and avoid redundancy. Second,
the anti‐surplusage canon “is not an absolute rule.” Marx v.
General Revenue Corp., 568 U.S. 371, 385 (2013). As we have ex‐
plained, “drafters of legislation often ‘intentionally err on the
side of redundancy’ as a precautionary measure and as a re‐
sponse to political demands.” Schutte v. Ciox Health, LLC, 28
F.4th 850, 862 (7th Cir. 2022), quoting Abbe R. Gluck & Lisa
Schultz Bressman, Statutory Interpretation from the Inside—An
Empirical Study of Congressional Drafting, Delegation, and the
Canons: Part I, 65 Stan. L. Rev. 901, 934 (2013); see also Schutte,
28 F.4th at 862–63 (collecting sources discussing why strict ap‐
plication of the anti‐surplusage canon is often misguided and
unwarranted). The most natural reading of the Wisconsin
drug law fits that description. The district court’s finding that
a narcotic effect is a type of depressant effect was not clearly
erroneous. There is no overbreadth due to this apparent tex‐
tual mismatch.
B. Salts and Esters of Cocaine—Factual Impossibility
The second apparent textual mismatch squarely presents
the problem of factual impossibility. Wisconsin’s definition of
cocaine includes “any of [its] salts, esters, isomers and salts of

-- 10 of 32 --

No. 21‐2309 11
esters and isomers that are theoretically possible within the
specific chemical designation.” Wis. Stat. § 961.16(2)(b). The
parties agree that the ACCA definition of a serious drug of‐
fense does not reach dealing in esters and salts of esters of co‐
caine.5
As one of the government experts explained, esters and
salts of esters of cocaine simply do not exist as a matter of
chemistry. Cocaine itself is an ester of a compound called ec‐
gonine, and a cocaine molecule contains two functional
groups that are classified as esters. Because of this chemical
structure, there “are no functional groups … that can form
any additional esters,” which means “there are no esters of
cocaine.” And since it is not chemically possible to have an
ester of cocaine, it also is not possible to have a salt of an ester
of cocaine.
The district court relied on this expert’s declaration to con‐
clude that esters and salts of esters of cocaine are factually im‐
possible. Turner did not offer any evidence to dispute that
finding, and it was not clearly erroneous. The district court
5 According to one of the government experts, “Esters are a chemical
functional group where an acyl group (carbon double bonded to an oxy‐
gen, C=O) joins an alkyl or aryl group (R‐) to an alkoxy or aryloxy group
(‐OR’), often written as RCOOR’ or RCO 2 R’.” See also Esters, IUPAC Com‐
pendium of Chemical Terminology Online, https://goldbook.iu‐
pac.org/terms/view/E02219 (last visited August 24, 2022) (defining esters
as “Compounds formally derived from an oxoacid RkE(=O)l(OH)m, (l ≠ 0)
and an alcohol, phenol, heteroarenol, or enol by linking with formal loss
of water from an acidic hydroxy group of the former and a hydroxy group
of the latter”). Salt, in this context, is a “chemical compound consisting of
an assembly of cations and anions.” Salt, IUPAC Compendium of Chemi‐
cal Terminology Online, https://goldbook.iupac.org/terms/view/S05447
(last visited August 24, 2022).

-- 11 of 32 --

12 No. 21‐2309
then concluded that a categorical mismatch cannot be based
on impossible conduct. Unlike the court’s factual findings,
that legal determination does not receive deference. Lockett,
782 F.3d at 352. We turn to that legal issue now.
The central legal question in this appeal is whether a de‐
fendant can avoid being deemed an armed career criminal by
showing that he could have been convicted under a state stat‐
ute for conduct that is factually impossible because the state
statute treats as criminal impossible conduct that the federal
definition does not reach. That question may strike at least a
casual reader as strange. It may not seem so surprising to
those familiar with the categorical approach and its some‐
times‐maddening abstract exercises. See, e.g., Rachel E. Bar‐
kow, Categorical Mistakes: The Flawed Framework of the Armed
Career Criminal Act and Mandatory Minimum Sentencing, 133
Harv. L. Rev. 200, 202–08, 206 n.66, 207 n.69 (2019) (describing
how the ACCA and various categorical approaches have cre‐
ated “chaos” in federal courts, leading judges and commenta‐
tors to decry outcomes as “unsatisfying and counterintuitive”
and “not based in reality” (citations omitted)); Alfred v. Gar‐
land, 13 F.4th 980, 987 n.1, 988 & n.3 (9th Cir. 2021) (England,
J., specially concurring) (collecting cases and explaining that
“as we have seen countless times, the categorical approach is
untethered from common sense,” and “[a]bsurd results are
far from an anomaly”), rehearing en banc granted, 35 F.4th
1218 (9th Cir. 2022); see also United States v. Taylor, 142 S. Ct.
2015, 2026 (2022) (Thomas, J., dissenting) (“This holding ex‐
emplifies just how this Court’s ‘categorical approach’ has led
the Federal Judiciary on a ‘journey Through the Looking

-- 12 of 32 --

No. 21‐2309 13
Glass,’ during which we have found many ‘strange things.’”
(citation omitted)).6
The Wisconsin statute’s inclusion of esters and salts of es‐
ters of cocaine creates an apparent textual mismatch with the
ACCA definition here. In the government’s view, when the
conduct that creates the mismatch is factually impossible, the
mismatch is “in name only” and there is no overbreadth.
Turner argues that this inquiry into factual impossibility is not
permitted under the conduct‐based categorical approach, as‐
serting that this court has already rejected such “‘theoretical
challenges’ to state‐statute overbreadth.” We have done no
such thing. As explained next, we have expressly left open the
question whether a categorical mismatch can be based on
“unquestionably nonexistent conduct.” United States v. Ruth,
966 F.3d 642, 648 (7th Cir. 2020).
1. Seventh Circuit Case Law
In the past few years, we have been asked twice to address
factual impossibility under the categorical approach. Each
time we declined the invitation. First, in United States v. De La
6 The categorical method is firmly in place under the ACCA as a mat‐
ter of statute and Supreme Court precedent. The advisory Sentencing
Guidelines now in effect also use the categorical method, particularly in
the career‐offender guidelines. In 2018 the Sentencing Commission pro‐
posed amendments to the Guidelines that would free courts from rigid
adherence to the categorical method and allow them to consider reliable
information about the defendant’s actual conduct leading to an earlier
conviction. Sentencing Guidelines for United States Courts, 83 Fed. Reg.
65400, 65407–11 (Dec. 20, 2018). While action was pending on those pro‐
posals, however, the Sentencing Commission lost its quorum. Earlier this
month, the United States Senate confirmed nominations needed to restore
a quorum to the Commission.

-- 13 of 32 --

14 No. 21‐2309
Torre, the state statute in question was categorically broader
than the applicable federal definition because the state law
made it a crime to deal in geometric isomers of methamphet‐
amine while federal law did not. 940 F.3d 938, 951 (7th Cir.
2019). The government argued that the two “statutes actually
mirror[ed] each other” because geometric isomers of metham‐
phetamine did not exist. Id. The issue had not been raised in
the district court, however, so the government had (under‐
standably) not presented evidence in the district court as to
the factual impossibility of geometric isomers of metham‐
phetamine. In applying plain‐error review, we refused to re‐
solve the factual issue by taking judicial notice on appeal of
two expert declarations from other cases.
In De La Torre we said that the government’s argument re‐
garding impossibility was not “pertinent … when the plain
language chosen by the [state] legislature dictates that the
[state] statute is categorically broader than the federal defini‐
tion of felony drug offense.” 940 F.3d at 952. Relying on this
language, Turner argues that De La Torre expressly rejected
factual impossibility arguments under the categorical ap‐
proach.
That sentence cannot carry the weight Turner gives it.
First, in rejecting the government’s attempt to rely on evi‐
dence presented in other cases because the issue had not been
raised before the appeal, we explained: “Our opinion takes no
position on the scientific merits, nor should it be read as lim‐
iting the government’s ability to present [factual impossibil‐
ity] argument[s] in future proceedings.” 940 F.3d at 952 n.5.
Second, De La Torre itself acknowledged that the issue of im‐
possibility did not make a difference in the outcome since the
decision was not “solely dependent on the definition of

-- 14 of 32 --

No. 21‐2309 15
methamphetamine and which of its isomers do or do not ex‐
ist.” Id. at 952. The state statute was overbroad in other re‐
spects. Third, in Ruth—decided less than a year later—the
government again argued that an apparent mismatch be‐
tween a state and federal statute was based on factually im‐
possible conduct (dealing in positional isomers of cocaine).
Ruth, 966 F.3d at 648. In Ruth we recognized that De La Torre
had “left the door ajar for future science based arguments.”
Id.
In Ruth, the government’s factual impossibility argument
ultimately fell short because the evidence was deficient. As in
De La Torre, the government failed to supply convincing evi‐
dence that positional isomers of cocaine did not exist. The ex‐
pert declaration in Ruth asserted that the expert had never de‐
tected positional isomers in analyzing over 50,000 samples of
cocaine. That is not quite the same as declaring that certain
compounds cannot exist. And contrary to Turner’s assertions,
Ruth did not shut the door to factual impossibility arguments
in rejecting this evidence: “There may be an occasion where a
state statute covers unquestionably nonexistent conduct, but
we do not need to predetermine how that analysis will look.”
966 F.3d at 648.7
7 Turner quotes one sentence in Ruth as foreclosing theoretical chal‐
lenges to apparent textual mismatches:
It is enough for us to say that where, as here, the state
statute of conviction is plain and intentional, our job is
straightforward: we compare the state statute to the fed‐
eral recidivism statute at issue and ask only if the state
law is the same as or narrower than federal law.
966 F.3d at 648. If this sentence were read in isolation, it would tend to
support his argument. But read with the rest of Ruth in mind, this

-- 15 of 32 --

16 No. 21‐2309
Nor did Aguirre‐Zuniga v. Garland, 37 F.4th 446 (7th Cir.
2022), a case decided after oral argument in this appeal, close
the door that Ruth and De La Torre left open. The state statute
in Aguirre‐Zuniga criminalized dealing in optical and posi‐
tional isomers of methamphetamine. Federal law, by contrast,
criminalized only optical isomers of methamphetamine. We
observed that positional isomers exist as a matter of chemis‐
try: “Methamphetamine has optical and positional isomers,
and methamphetamine itself exists in two isomeric forms, l‐
methamphetamine and d‐methamphetamine, which them‐
selves can be combined into a potential third iteration known
as a ‘racemic mixture,’ dl‐methamphetamine.” Id. at 452 (em‐
phases added and citation omitted). Since the statute crimi‐
nalized this real conduct, the government could not defeat the
mismatch by arguing that there was no realistic probability
that someone would be prosecuted for such conduct. Like
Ruth and De La Torre, however, Aguirre‐Zuniga did not deal
with truly “unquestionably nonexistent conduct.” Ruth, 966
F.3d at 648 (declining to “predetermine” the analysis for state
statute overbreadth from “unquestionably nonexistent con‐
duct”). Accordingly, factual impossibility under the categori‐
cal approach has remained an open question in our circuit.
This case requires us to answer it.
2. Impossible Conduct
Unlike in Ruth, De La Torre, and Aguirre‐Zuniga, the district
court here made an express factual finding that the conduct—
dealing in esters and salts of esters of cocaine—that creates
argument fails to persuade. The sentence did not trump everything else in
the opinion, which expressly left open the question of genuine factual im‐
possibility.

-- 16 of 32 --

No. 21‐2309 17
the apparent mismatch cannot occur. As noted, that finding
was not clearly erroneous; it was based on unchallenged ex‐
pert testimony. Thus, whether Turner qualifies as an armed
career criminal under the ACCA turns on whether state stat‐
ute overbreadth can stem from conduct that is factually im‐
possible. It cannot.
Every Supreme Court decision applying the categorical
approach has contemplated real rather than imaginary dis‐
crepancies between the elements of state and federal statutes.
Starting with Taylor v. United States, the Court instructed
lower courts to evaluate prior state court convictions for “bur‐
glary” to see whether the “statutory definition substantially
corresponds to ‘generic’ burglary,” or if “the charging paper
and jury instructions actually required the jury to find all the
elements of generic burglary in order to convict the defend‐
ant.” 495 U.S. 575, 602 (1990) (emphases added). That analysis
is “abstract” in the sense that it directs courts to evaluate the
elements of a prior offense, not the actual underlying facts of
the defendant’s prior conviction. But Taylor cautioned against
invoking “technical definitions and labels under state law” to
create faux mismatches. Id. at 590. Asking whether the ele‐
ments “substantially correspond” or if a jury was “actually
required” to find certain elements was designed to keep the
objective inquiry at least tethered to reality.
As noted above, since Taylor the categorical approach has
been criticized as not sufficiently tethered to reality. See also,
e.g., United States v. Doctor, 842 F.3d 306, 312–13 (4th Cir. 2016)
(Wilkinson, J., concurring) (explaining that the categorical ap‐
proach has moved “beyond what the Supreme Court origi‐
nally anticipated” and “has pushed criminal sentencing to the
very last place that sentencing ought to be, that is at an

-- 17 of 32 --

18 No. 21‐2309
untenable remove from facts on the ground”). But at least one
reality check remains. In Gonzales v. Duenas‐Alvarez, the Su‐
preme Court offered a significant clarification:
[T]o find that a state statute creates a crime out‐
side the generic definition of a listed crime in a
federal statute requires more than the applica‐
tion of legal imagination to a state statute’s lan‐
guage. It requires a realistic probability, not a the‐
oretical possibility, that the State would apply its
statute to conduct that falls outside the generic
definition of a crime.
549 U.S. 183, 193 (2007) (emphases added); see also Moncrieffe
v. Holder, 569 U.S. 184, 191 (2013) (“[O]ur focus on the mini‐
mum conduct criminalized by the state statute is not an invi‐
tation to apply ‘legal imagination’ to the state offense….” (ci‐
tation omitted)). The Duenas‐Alvarez reality check focuses the
inquiry on real—not imaginary—state statute overbreadth. It
also supports the rejection of any approach that would re‐
quire courts to wade into such imaginative waters. Turner’s
approach, however, would have us dive in.
We decline Turner’s invitation to ignore this general lesson
from Duenas‐Alvarez. We also reject his argument that Duenas‐
Alvarez does not apply here because it concerned only the ge‐
neric‐offense branch of the categorical approach, as distinct
from a conduct‐based test.
To explain this difference, the Supreme Court recently
clarified that the categorical approach differs slightly based
on the federal statutory language at issue. Shular, 140 S. Ct. at
783. Where the federal recidivist statute lists covered offenses,
courts must identify “generic” versions of the crimes with the

-- 18 of 32 --

No. 21‐2309 19
elements of “the offense as commonly understood,” and then
compare those generic elements to the elements of the crime
of the prior conviction. Id. We have described that approach
as the “generic‐offense method.” Ruth, 966 F.3d at 646. The
second approach, and the one Shular directed courts to use in
cases like Turner’s, is the “conduct‐based method.” Id. As ex‐
plained, that requires us to ask “only whether [Turner’s] prior
conviction’s elements necessarily entail the conduct identi‐
fied” in the ACCA’s definition of a serious drug offense. Id. at
647; see 18 U.S.C. § 924(e)(2)(A)(ii).
In Ruth we said that rules derived from cases applying the
generic‐offense method did “not apply equally under the con‐
duct‐based method at play” in that case. 966 F.3d at 648. And
as mentioned, Duenas‐Alvarez applied the offense‐based
method. This language in Ruth, Turner says, prevents any
consideration of Duenas‐Alvarez or cases like it because his
case also concerns only the conduct‐based method. We disa‐
gree. Ruth explained that offense‐based method precedents
do “not apply equally” to conduct‐based method cases, not
that they do not apply at all. Id. (emphasis added). The warn‐
ing from Duenas‐Alvarez to avoid imaginative inquiries
reaches beyond the case’s categorization. Accord, United
States v. Rodriguez‐Gamboa, 972 F.3d 1148, 1153–54 (9th Cir.
2020) (“teaching of Duenas‐Alvarez” is not limited to cases in‐
volving generic offenses). We do not read Shular or Ruth as
barring courts from distinguishing between possible and im‐
possible conduct when applying either method.
To be clear, asking about genuine factual impossibility is
not the “realistic probability” test stated in different terms.
Factual impossibility is distinct and comes one step before
such an inquiry into realistic probabilities, if any. See Aguirre‐

-- 19 of 32 --

20 No. 21‐2309
Zuniga, 37 F.4th at 450 (explaining that courts apply the real‐
istic probability test only when the statutory language is am‐
biguous or has an indeterminate reach).
After identifying an apparent mismatch, the next logical
question—though one that rarely needs to be asked out
loud—is whether that supposedly overbroad conduct can ac‐
tually occur. If it cannot, then there is plainly no overbreadth.
The reason is simple. Where a state criminal statute seems to
apply to conduct that would be factually impossible while the
federal statute does not, the state statute does not actually
make criminal any conduct that its federal counterpart does
not. The conduct that the federal statute would not reach
simply cannot happen. Factual impossibility prevents courts
from venturing into the land of the imaginary and reaching
absurd outcomes.
The point that overbreadth must concern conduct that can
actually occur is so basic that decisions applying the categor‐
ical approach have arguably already incorporated it, at least
implicitly. Consider United States v. Taylor, 142 S. Ct. 2015
(2022), the Supreme Court’s most recent case concerning the
categorical approach. The question before the Court was
whether attempted Hobbs Act robbery qualified as a “crime
of violence” under the ACCA. The Act defines “crime of vio‐
lence” to include an offense that is a felony and that “has as
an element the use, attempted use, or threatened use of phys‐
ical force against the person or property of another.” 18 U.S.C.
§ 924(c)(3)(A).
“Without question,” the Court said, “many who commit
the crime of attempted Hobbs Act robbery do use, attempt to
use, or threaten to use force.” Taylor, 142 S. Ct. at 2022. But
“some cases are not all cases.” Id. As the court explained,

-- 20 of 32 --

No. 21‐2309 21
individuals arrested “before they can threaten anyone may be
convicted too.” Id. Since people could commit attempted
Hobbs Act robbery without actually threatening force, the
Court held that the offense swept more broadly than the
ACCA definition and could not qualify as a predicate. The
key point for purposes of this case is that it was actually pos‐
sible to commit the overbroad conduct. Id. at 2035 (Alito, J.,
dissenting) (“[T]he Court infers that attempted Hobbs Act
robbery is not a ‘crime of violence’ under § 924(c)(3)(A) be‐
cause it is possible to commit that offense without attempting
to use force.”); cf. Lopez‐Aguilar v. Barr, 948 F.3d 1143, 1148
(9th Cir. 2020) (“Because it is possible to commit theft by de‐
ception with the consent of the owner, Oregon’s theft statute
expressly includes conduct outside of the generic definition.”
(emphasis added)).
In Turner’s case, by contrast, it simply is not possible to deal
in esters or salts of esters of cocaine—since esters and salts of
esters cannot exist chemically—so Wisconsin’s drug law does
not sweep in more conduct than the ACCA’s definition of a
“serious drug offense.” 8
8 We have not forgotten that the ACCA imposes heavy penalties on
defendants who qualify as armed career criminals. The categorical ap‐
proach requires an objective analysis that may well be underinclusive in
terms of the Act’s purposes, but which conforms to the Act’s text, which
focuses on prior convictions rather than prior conduct. The categorical ap‐
proach bars judges from potentially imposing a mandatory minimum sen‐
tence based on conduct that could not have landed a defendant in federal
court. Accord, Rodriguez‐Gamboa, 972 F.3d at 1153 (“The purpose of the
categorical approach is to ascertain whether the defendant was necessarily
convicted in state court of conduct that would also violate the relevant
federal law.”). But when the overbroad conduct cannot possibly occur,

-- 21 of 32 --

22 No. 21‐2309
The Ninth Circuit has agreed. In Rodriguez‐Gamboa, the
court opted for “scientific reality over abstract legal doctrine”
in considering a factual impossibility rebuttal to an over‐
breadth challenge. 972 F.3d at 1150. The state statute at issue
there criminalized dealing in geometric isomers of metham‐
phetamine while the federal counterpart did not, creating a
similar apparent mismatch. As the district court in that case
found, however, geometric isomers of methamphetamine do
not exist as a matter of chemistry. In a persuasive opinion, the
Ninth Circuit held that a categorical mismatch cannot be
based on nonexistent conduct. Id. at 1152–54. As the court put
it:
Because we know as a scientific fact that drag‐
ons have never existed, we would not find over‐
broad a state statute criminalizing the posses‐
sion of dangerous animals, defined to include
dragons, if the relevant federal comparator out‐
lawed possession of the same animals but did
not include dragons. We see no reason to reach
a different result here.
Id. at 1155. We agree.
If we had doubts on this score, they would be assuaged
because the Wisconsin statute here acknowledges the ques‐
tion of factual impossibility. The statute defines cocaine as
“any of [its] salts, esters, isomers and salts of esters and iso‐
mers that are theoretically possible within the specific chemical
designation.” Wis. Stat. § 961.16(2)(b) (emphasis added). This
language invites consideration of factual impossibility. It
there is no risk that a defendant convicted under the state statute commit‐
ted conduct that federal law does not also criminalize.

-- 22 of 32 --

No. 21‐2309 23
extends the definition of cocaine to include variants that are
“theoretically possible” within cocaine’s “specific chemical
designation.” That language does not reach variants that are
not theoretically possible. By including esters and salts of es‐
ters in its definition of cocaine, the state legislature was
plainly using a catch‐all approach rather than vouching (in‐
correctly) for a scientific fact.
To avoid this result, Turner would have us end the analy‐
sis after identifying one type of conduct that is not in the fed‐
eral counterpart, even if that conduct could not possibly occur
and the plain language of the state statute indicated that the
state legislature did not intend to convey that it thought such
conduct occurs. It is easy to think of some resulting absurdi‐
ties from this approach. The Ninth Circuit, for its part, hy‐
pothesized a state statute outlawing possession of dangerous
animals defined to include dragons. Rodriguez‐Gamboa, 972
F.3d at 1155. We hypothesized at argument here a statute
making it a crime to assault human beings, leprechauns, and
zombies. The mismatch based on a theoretically possible as‐
sault on a leprechaun or zombie would not reach beyond the
actual scope of a violent felony under the ACCA. The categor‐
ical approach does not require courts to ignore whether mis‐
matched “conduct” is actually impossible as a matter of sci‐
entific fact.
3. Turner’s Additional Arguments
Turner offers two additional arguments. First, he asserts
that a factual impossibility inquiry violates the Sixth Amend‐
ment. It does not. The Sixth Amendment prohibits judges
from making factual findings at sentencing that increase a de‐
fendant’s minimum or maximum penalty other than a prior
conviction. See Apprendi v. New Jersey, 530 U.S. 466, 490 (2000);

-- 23 of 32 --

24 No. 21‐2309
Alleyne v. United States, 570 U.S. 99, 115–16 (2013). By finding
that Turner could not have been convicted under the Wiscon‐
sin statute for possessing cocaine esters, Turner argues, the
district court explored the facts of his particular prior offense,
which Apprendi and Alleyne should bar. We disagree.
Apprendi and Alleyne bar courts from going “beyond iden‐
tifying the crime of conviction to explore the manner in which
the defendant committed that offense.” Mathis v. United States,
579 U.S. 500, 511 (2016). A judge cannot “mak[e] a disputed
determination about ‘what the defendant and state judge
must have understood as the factual basis of the prior plea’ or
‘what the jury in a prior trial must have accepted as the theory
of the crime.’” Id., first quoting Shepard v. United States, 544
U.S. 13, 25 (2005) (plurality opinion), and then quoting
Descamps v. United States, 570 U.S. 254, 269 (2013). Such find‐
ings address adjudicative facts related to the specific manner
in which a defendant committed his prior offense or how a
jury and/or judge might have viewed his conduct. Adjudica‐
tive facts are typically left to a jury to find. The Supreme Court
has long understood the Sixth Amendment to bar adjudica‐
tive fact‐finding under the categorical approach. See Mathis,
579 U.S. at 510‐11.
The type of fact‐finding required to understand whether
conduct included in statutory language can possibly occur is
different. The district court here was not making findings
about Turner’s individual conduct. Instead, it was finding
what amounts to a legislative fact: that cocaine esters do not
exist and cannot exist, so neither Turner nor anyone else could
be convicted for trafficking in them. This is a type of legisla‐
tive fact‐finding that judges conduct regularly in challenges
to legislation. “The concept of a legislative fact comes into its

-- 24 of 32 --

No. 21‐2309 25
own when there is no reason to believe that certain facts per‐
tinent to a case vary from locality to locality, or from person
to person….” Frank v. Walker, 773 F.3d 783, 795 (7th Cir. 2014)
(Posner, J., dissenting from denial of rehearing en banc). Leg‐
islative facts “are not unique to a particular case and provide
therefore an appropriate basis for legislation of general appli‐
cation.” Id. Asbestosis, for example, either is a progressive dis‐
ease or it is not. Nothing supports the idea that it is progres‐
sive in some cases but not in others. Similarly here, we see no
risk that cocaine esters could exist in Illinois but not Wiscon‐
sin. The question of factual impossibility in this case presents
a legislative fact that applies broadly to the state of the world;
it does not require making a finding specific to Turner’s prior
convictions. Apprendi and Alleyne concern the latter.
In addition, courts regularly engage in analogous fact‐
finding under the categorical approach. Consider the “realis‐
tic probability” test. See Duenas‐Alvarez, 549 U.S. at 193. That
test asks whether there is “a realistic probability … that the
State would apply its statute to conduct that falls outside the
generic definition of a crime.” Id.; see also Portee v. United
States, 941 F.3d 263, 271–73 (7th Cir. 2019) (applying the real‐
istic probability test to Indiana’s “felony intimidation” stat‐
ute). While ultimately a legal determination, this analysis re‐
quires courts to determine a fact about the world: whether
“State A” realistically prosecutes “Conduct B.” It could not be
the case that, for one defendant, State A would realistically
apply its statute to Conduct B, but for another defendant, it
would not. Apprendi and Alleyne do not bar judges from mak‐
ing such a “factual” determination related to state prosecuto‐
rial practices. The Supreme Court has expressly endorsed
such an inquiry. Following that guidance, determining

-- 25 of 32 --

26 No. 21‐2309
whether overbroad conduct actually exists must be permissi‐
ble fact‐finding under the Sixth Amendment.
Finally, Turner argues that a factual impossibility ap‐
proach will lead to inconsistent outcomes, which the categor‐
ical approach was designed in part to prevent. We are not con‐
vinced. If district courts split over whether it is factually pos‐
sible to create esters of cocaine, we will resolve the issues on
appeal. That possible future challenge does not convince us
that we should make a decision here by assuming that a sci‐
entific impossibility is real, just as we would not base a deci‐
sion on the assumption that leprechauns or zombies are real.
To sum up, a categorical mismatch cannot be based on fac‐
tually impossible conduct. Since esters of cocaine cannot exist
chemically, there is no genuine mismatch between the Wis‐
consin statute under which Turner was convicted and the
ACCA’s definition of a “serious drug offense.” Turner was
properly classified as an armed career criminal based on his
three predicate convictions for serious drug offenses.
C. Evidentiary Hearing?
There is one issue left. In his appellate briefs, Turner as‐
serts that if we reject his arguments against considering fac‐
tual impossibility, we should order a limited remand for an
evidentiary hearing on the chemistry question of factual im‐
possibility. We deny this request. After the government pre‐
sented its evidence of factual impossibility in the district
court, Turner twice asked for more time to consult an expert.
He also suggested that an evidentiary hearing might be nec‐
essary to address the government’s evidence. The district
court granted his requests and scheduled a hearing. Two
months after his first request for more time, Turner changed

-- 26 of 32 --

No. 21‐2309 27
his mind and together with the government asked the court
to vacate the hearing since he did not plan on presenting evi‐
dence to rebut the government’s evidence of factual impossi‐
bility. That’s textbook waiver. United States v. Jaimes‐Jaimes,
406 F.3d 845, 848 (7th Cir. 2005) (“The touchstone of waiver is
a knowing and intentional decision.”).
As we have explained: “To waive an argument on appeal,
a defendant must have had some strategic reason for waiving
the argument in the trial court.” United States v. Hammond, 996
F.3d 374, 399 (7th Cir. 2021). Ruth and De La Torre put Turner
on notice that the government could argue factual impossibil‐
ity. After having been allowed ample time to track down con‐
trary evidence, Turner chose not to dispute the factual impos‐
sibility of esters of cocaine in the district court (or on appeal)
and chose to rely instead on the legal arguments we have re‐
jected above. Turner cannot now assert that we should give
him the evidentiary hearing that he forwent because he “had
no reason to rebut the government’s evidence.” Turner was
not entitled first to try out his legal theory to block considera‐
tion of facts and then, upon the failure of that strategy on ap‐
peal, start again in the district court with a different strategy.
Forgoing the promised evidentiary hearing was a strategic
choice, not an inadvertent one.
AFFIRMED.

-- 27 of 32 --

28 No. 21‐2309
J ACKSON‐A KIWUMI, Circuit Judge, concurring. I agree with
the majority opinion regarding Turner’s argument about the
“narcotic effect” of cocaine: The plain meaning of the Wiscon‐
sin statute, as confirmed by the record before us, shows that
narcotic effects are a subset of depressant effects. So, the Wis‐
consin statute is not categorically broader than its federal
counterpart in this way.
As for Turner’s esters and salts of esters of cocaine argu‐
ment, I agree with the majority opinion that the Wisconsin
statute explicitly contemplates the question of “theoretical
possibility.” As I see it, that statutory language permits us to
decide this case using the traditional categorical approach.
After doing so, I, like my colleagues, conclude that the Wis‐
consin statute does not sweep more broadly than its federal
counterpart on this axis either.
I concur only in the judgment because I worry that the ma‐
jority opinion’s approach of adding a new inquiry to the anal‐
ysis—determining whether there is a factual impossibility be‐
fore reaching the realistic probability test—risks plunging
trial courts into fact‐intensive inquiries and risks appellate
courts creating binding precedent on rapidly changing scien‐
tific developments.
Before I explain these two concerns, I briefly summarize
how, in my view, this case can be resolved using the tradi‐
tional categorical approach: When comparing a state statute
to its federal counterpart under the categorical approach, first
we determine whether there is any facial mismatch. Aguirre‐
Zuniga v. Garland, 37 F.4th 446, 450 (7th Cir. 2022). Only if we
find that our comparison is hindered because the language of
the state statute is ambiguous or has indeterminate reach, do
we move to the second step to determine whether there is a

-- 28 of 32 --

No. 21‐2309 29
“realistic probability” of prosecution. 1 Id. Here, Wisconsin de‐
fines cocaine to include “salts, esters, isomers, and salts of es‐
ters and isomers that are theoretically possible within the specific
chemical designation.” Wis. Stat. § 961.16(2)(b) (emphasis
added). The phrase “theoretically possible” circumscribes all
the different iterations of cocaine the statute lists. Therefore,
the statute is ambiguous because it is not clear whether the
1 The reasoning in the Supreme Court’s recent decision in United States
v. Taylor, 142 S. Ct. 2015 (2022), could change the contours of the realistic
probability test. The Court began its discussion by noting “the oddity of
placing a burden on the defendant to present empirical evidence about the
government’s own prosecutorial habits” and “the practical challenges [the
realistic probability test] present[s] in a world where most cases end in
plea agreements, and not all of those cases make their way into easily ac‐
cessible commercial databases.” Id. at 2024 (citations omitted). The Court
then rejected the government’s argument to use the realistic probability
test in Taylor for two core reasons: (1) Duenas‐Alvarez involved a state stat‐
ute, while the statute at issue in Taylor was federal, so there was no feder‐
alism rationale “to consult how a state court would interpret its own
State’s laws”; and (2) “the elements of the relevant state and federal of‐
fenses [in Duenas‐Alvarez] clearly overlapped,’” while the statutes in Tay‐
lor had “no overlap to begin with.” Id. at 2025 (citations omitted).
Taylor suggests that the realistic probability test applies when a state
statute “clearly overlap[s]” with its federal counterpart and does not ap‐
ply when two federal statutes have “no overlap to begin with.” But it is
unclear how Taylor’s logic affects the realistic probability test in situations
where two federal statutes “clearly overlap” or where a state statute and
its federal counterpart have “no overlap to begin with.” We need not parse
these nuanced situations today. Neither Turner nor the government be‐
lieve Taylor is of any consequence—neither side offered any explanation
of the case’s relevance in a letter pursuant to Federal Rule of Appellate
Procedure 28(j) or other filing. Instead, the parties chose to stake their
claims solely on the issue of factual impossibility, a question Taylor does
not touch. Whether Taylor has changed the landscape of the categorical
approach in other ways, therefore, is left for another day.

-- 29 of 32 --

30 No. 21‐2309
Wisconsin statute is facially broader than federal law—it
could be broader, or not. Without clarity in the statute (or in
other statutes that inform the statute of conviction) on how to
determine theoretical possibility under the statute, I turn to
the realistic probability test. Turner does not point to any
cases where someone was prosecuted for esters or salts of es‐
ters of cocaine (because, as best as we know right now, esters
of cocaine do not exist). Thus, there is no realistic probability
that the state would prosecute someone for esters or salts of
esters of cocaine. Consequently, the Wisconsin statute does
not criminalize more conduct than its federal counterpart and
Turner’s prior convictions under the Wisconsin statute qual‐
ify as ACCA predicates. See Gonzales v. Duenas‐Alvarez, 549
U.S. 183, 193 (2007).
The majority opinion does not travel this road. Instead, the
majority develops an intermediary step where “[a]fter identi‐
fying an apparent mismatch,” a court must determine
“whether that supposedly overbroad conduct can actually oc‐
cur.” This rule invites two practical problems that courts may
now face: (1) categorical approach cases can turn into full‐
blown battle‐of‐the‐expert mini‐trials over chemistry con‐
cepts at the district court level; and (2) this case and future
appellate decisions will act as precedent on scientific facts, de‐
spite the fact that chemistry is an ever‐evolving field of study.
As for the first potential problem, one of the core reasons
the Supreme Court adopted the categorical approach was to
avoid “the practical difficulties and potential unfairness of a
factual approach.” Taylor v. United States, 495 U.S. 575, 601
(1990). The Court was concerned that trial courts would have
to sift through the charging papers, review trial evidence, and
even present additional witnesses—all to show that an

-- 30 of 32 --

No. 21‐2309 31
underlying conviction qualified, in the case of Taylor, as ge‐
neric burglary. Id. This concern, among others, is why the Su‐
preme Court declined to apply the modified categorical ap‐
proach to indivisible statutes. See Descamps v. United States,
570 U.S. 254, 270–71 (2013). Although today’s decision does
not deal with the papers and evidence underlying a prior state
conviction, it nonetheless welcomes a factual inquiry into cat‐
egorical approach cases. This seems as if it will result in par‐
ties presenting competing expert testimony as to whether an
ester, isomer, or salt of cocaine, methamphetamine, fentanyl,
or another illegal drug exists. In turn, the parties will file
Daubert motions or otherwise attempt to undermine the other
side’s view. All in all, district courts at sentencing will now
have to engage in lengthy, drawn‐out factual discussions of
the chemistry of these substances. This did not happen in
Turner’s case, but it easily could in another case. Such a result
undermines the spirit of the categorical approach. 2
As for the second potential problem, the majority opinion
states that “[t]he question of factual impossibility in this case
presents a legislative fact that applies broadly to the state of
the world,” and that “[i]f district courts split over whether it
is factually possible to create esters of cocaine, we will resolve
the issues on appeal.” This pledge turns issues of evolving sci‐
ence into precedent. If we as an appellate court determine
that, as a matter of law, there are no esters of cocaine or
2 The majority opinion emphasizes that it is not applying the realistic
probability test in this new step. And while the analytical process the ma‐
jority proposes differs from Duenas‐Alvarez, it is still a rule that under‐
mines the categorical approach’s focus on the statutory language and in‐
vites extraneous facts, ultimately turning a pure legal question into a fight
over who has the better expert.

-- 31 of 32 --

32 No. 21‐2309
whatever the substance at issue is, we fail to account for the
constant efforts of drug manufacturers to iterate and develop
their products. Take, for example, the history of fentanyl ana‐
logues. Just between 2012 and 2016, “seventeen fentanyl ana‐
logues were reported to the UNODC … [but] only one of
them, acetylfentanyl, has been placed under international
control.” United Nations Office on Drugs & Crime, Global
SMART Update: Fentanyl and its Analogues – 50 Years On, at 4
(Mar. 2017), https://tinyurl.com/8efp6wck; see also Harold E.
Schueler, Emerging Synthetic Fentanyl Analogs, Academic Fo‐
rensic Pathology 36, 38 (2017) (listing six fentanyl analogues
developed over two years). In UNODC’s view, “[t]he count‐
less possibilities to create new compounds by small changes
in chemical structures pose a growing challenge to interna‐
tional control of the opioid trade.” United Nations Office on
Drugs & Crime, Global SMART Update: Fentanyl and its Ana‐
logues – 50 Years On, at 4. The rapid development of fentanyl
analogues highlights how the scientific methods involved in
these illicit substances are ever‐changing. Therefore, I worry
that our caselaw will ossify these “legislative facts” of chem‐
istry in a way that may not comport with the near‐future’s re‐
ality.
Because I would resolve this case using the classic appli‐
cation of the categorical approach, and the majority’s new
rule potentially presents two applicability problems, I concur
only in the judgment.

-- 32 of 32 --