United States Court of Appeals for the Fifth Circuit United States Court of Appeals Fifth Circuit
____________ FILED August 18, 2026
No. 25-30192 Lyle W. Cayce ____________ Clerk
United States of America,
Plaintiff—Appellee,
versus
Julius Augillard,
Defendant—Appellant.
______________________________
Appeal from the United States District Court for the Eastern District of Louisiana USDC No. 2:24-CR-90-1
______________________________
ON PETITION FOR REHEARING EN BANC
Before Clement, Douglas, and Ramirez, Circuit Judges. Per Curiam:
Treating the petition for rehearing en banc as a petition for panel rehearing (5th Cir. R.40 I.O.P.), the petition for panel rehearing is DENIED. The petition for rehearing en banc is DENIED because, at the request of one of its members, the court was polled, and a majority did not vote in favor of rehearing (Fed. R. App. P.40 and 5th Cir. R.40).
In the en banc poll, four judges voted in favor of rehearing: Chief Judge Elrod, and Judges Richman, Graves, and Higginson.
No. 25-30192
Thirteen voted against rehearing: Judges Jones, Smith, Stewart, Southwick, Haynes, Willett, Ho, Duncan, Engelhardt, Oldham, Wilson, Douglas, and Ramirez.
3
No. 25-30192
Andrew S. Oldham, Circuit Judge, concurring in the denial of rehearing en banc:
Thirty-six years ago, the Supreme Court announced the categorical approach. See Arthur Taylor v. United States, 495 U.S. 575 (1990). Ever since, “the categorical approach has developed a reputation for crushing common sense in any area of the law in which its tentacles find an inroad.” United States v. Escalante, 933 F.3d 395, 406 (5th Cir. 2019); accord post, at 11 (Elrod, C.J., dissenting from the denial of rehearing en banc). It often leads to “counterintuitive,” “bewildering,” and grotesque results. Alejos- Perez v. Garland, 93 F.4th 800, 811–12 (5th Cir. 2024) (Oldham, J., concurring). No federal judge has a nice thing to say about it. See, e.g., United States v. Cervenak, 135 F.4th 311, 342–44 (6th Cir. 2025) (en banc) (Griffin, J., dissenting) (collecting citations from dozens of federal judges). And yet, the inferior federal courts all too often encourage the doctrine’s metastasis. See Alejos-Perez, 93 F.4th at 811 (Oldham, J., concurring) (describing the expansion of the categorical approach across federal criminal law, immigration law, and sentencing).
But not today. In this case, our court voted overwhelmingly to retain a small but important limit on the categorical approach. That result accords with Supreme Court precedent. And it maintains a last bastion of reality in a doctrinal area that’s increasingly reminiscent of a Miró painting.
I
As everyone knows well by now, the categorical approach is counterintuitive at best. To determine whether Defendant A committed a crime of violence (for the purpose of certain sentencing enhancements), a court cannot ask whether Defendant A himself committed a crime of violence. Rather, the categorical approach requires courts to adjudicate
4
No. 25-30192
hypothetical and imaginary cases. 1 Why? Because the categorical approach commands us to ignore Defendant A’s actual violence and to focus instead on whether some imaginary non-defendant could hypothetically commit the same crime non-violently.
This leads to some absurd proceedings in federal court. To take just one example, we once debated far-fetched hypotheticals about poisoned orange juice and assisted suicide to determine whether a man committed a crime of violence by beating his brother-in-law to death with a baseball bat. United States v. Reyes-Contreras, 910 F.3d 169, 175, 185 (5th Cir. 2018) (en banc). And outrageously, our court once applied the categorical approach to hold that a “35-year-old who sexually abused a 14-year-old cannot be categorized as a tier II sex offender.” Escalante, 933 F.3d at 407.
The categorical approach is a windfall for violent criminals. It allows them to escape the reach of the law—even when their felonies are obviously violent. And the average law-abiding American would no doubt be shocked to learn that our courts deem convicted rapists, murders, carjackers, kidnappers, and domestic abusers “not violent.” See Alejos-Perez, 93 F.4th at 812 (Oldham, J., concurring) (collecting examples).
II
The actual-case rule is a glimmer of sunshine in the otherwise-bleak landscape of the categorical approach.
_____________________
1 It is unclear how this is consistent with Article III. “[A]djudicating hypothetical cases can be tough to reconcile with both the prohibition on hypothetical jurisdiction, see Steel Co. v. Citizens for a Better Env’t, 523 U.S. 83, 101 (1998), and also the prohibition on advisory opinions, see Correspondence of the Justices, in William Baude, et al., Hart and Wechsler’s Federal Courts and the Federal System 67–69 (8th ed. 2025).” Nathan v. Alamo Heights Indep. Sch. Dist., 173 F.4th 576, 614 (5th Cir. 2026) (Oldham, J., concurring).
5
No. 25-30192
Here is how it works. Federal law imposes sentencing enhancements for certain predicate felonies. For example, the Arthur Taylor Court considered the federal Armed Career Criminal Act (“ACCA”), which imposed a sentencing enhancement for the predicate felony of “burglary.” 495 U.S. at 589, 598. What is “burglary” in the federal sense? Arthur Taylor held it’s burglary in “the generic sense in which the term is now used in the criminal codes of most States.” Id. at 598. So if a defendant is convicted of burglary in State A, that conviction counts as a burglary conviction under ACCA—so long as the elements of burglary in State A match the elements of “generic” burglary in the States more generally. And in performing that element-matching analysis, courts must apply the categorical approach: The relevant inquiry is whether the elements categorically match—not necessarily whether the defendant committed an act that everyone on the planet earth would recognize as burglary.
But what if the relevant State’s law is broader than the generic offense?
The categorical approach suggests a windfall for the defendant. If State B criminalizes, say, theft in a way that’s broader than the generic elements for theft in the States more generally, then the defendant could argue that his offense is not categorically considered theft. And under the categorical approach, it would not matter that the defendant obviously stole something. All that would matter, our defendant would say, is that State B has a broader definition of theft than the generic definition in the States more generally, so theft in State B does not categorically constitute theft.
The Supreme Court placed important limits on such hypothetical adjudication in Gonzales v. Duenas-Alvarez, 549 U.S. 183 (2007). In Duenas- Alvarez, the Court made clear that if a defendant argues that State B’s definition of a crime is broader than the generic definition, he must show a “realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside the generic definition of a crime.” Id.
6
No. 25-30192
at 193. To do so, a defendant “must at least point to his own case or other cases in which the state courts in fact did apply the statute in the special (nongeneric) manner for which he argues.” Ibid.
The actual-case rule imposes a reality check on the categorical approach. As the Court recognized in Duenas-Alvarez, the actual-case requirement ensures that the existence of a categorical match does not merely depend on “the application of legal imagination to a state statute’s language.” Ibid. In this way, the Duenas-Alvarez rule helps ameliorate the “byzantine-like” process that often forces judges to engage in “rigorous abstract reasoning” and reach results that are “almost certainly contrary to any plain reading of the statute.” Escalante, 933 F.3d at 406–07.
Our circuit faithfully applied this rule in United States v. Castillo-
Rivera, 853 F.3d 218 (5th Cir. 2017) (en banc). Citing Duenas-Alvarez, we rejected the defendant’s position that he did not have to point to an “actual case” in which the State has applied its criminal statute in a way that was broader than the generic offense. Id. at 223. The defendant’s position, we emphasized, “does not comply with the Supreme Court’s directive in Duenas-Alvarez.” Ibid. Castillo-Rivera has provided the governing rule in our circuit for the last nine years.
The Supreme Court has repeatedly demonstrated that the Castillo-
Rivera approach is correct. Time and again, the Court has required defendants to show that States actually apply their laws in ways that are broader than the relevant federal statute. See James v. United States, 550 U.S. 192, 202 (2007), overruled on other grounds by Samuel Johnson v. United States, 576 U.S. 591 (2015) (recognizing that “while [Florida’s] statutory language is broad, the Florida Supreme Court has considerably narrowed its application in the context of attempted burglary”); Curtis Johnson v. United States, 559 U.S. 133, 138 (2010) (“We are . . . bound by the Florida Supreme
7
No. 25-30192
Court’s interpretation of state law, including its determination of the elements” of the statute in question); Moncrieffe v. Holder, 569 U.S. 184, 206 (2013) (“To defeat the categorical comparison in this manner, a noncitizen would have to demonstrate that the State actually prosecutes the relevant offense in cases involving antique firearms.”).
And thankfully so. The actual-case rule that comes from Duenas-
Alvarez and Castillo-Rivera means that in at least one corner of the categorical approach, a defendant can win relief only by pointing to actual cases. Imaginary hypotheticals, here at least, are insufficient. Our court wisely decided not to jettison the actual-case rule in favor of still more fanciful extensions of the categorical approach.
III
Finally, a few words about today’s dissenting opinion. I address (A) the Supreme Court’s Justin Taylor case and (B) the existence of the circuit split.
A
The dissenting opinion posits that the Supreme Court sub silentio overruled Duenas-Alvarez in Justin Taylor. Post, at 15–17 (Elrod, C.J., dissenting from the denial of rehearing en banc) (citing United States v. Justin Taylor, 596 U.S. 845 (2022)). I respectfully disagree.
The question in Justin Taylor was whether Hobbs Act robbery—a federal offense, not a state one—was a crime of violence under 18 U.S.C. § 924(c)(3)(A). 596 U.S. at 148. The Court rejected the Government’s argument, premised on Duenas-Alvarez, that federal prosecutors had never applied Hobbs Act robbery in an overbroad fashion. Id. at 858–59. The Court found Duenas-Alvarez inapplicable. Because a federal court was construing a
8
No. 25-30192
federal statute, the Court obviously had no reason to consider the elements of generic robbery in the States. Ibid.
But when federal judges interpret state law, the Court made clear, federalism requires us to defer to state courts’ actual applications of state law. Far from overruling Duenas-Alvarez, Justin Taylor reaffirmed it: “Appreciating the respect due state courts as the final arbiters of state law in our federal system, [Duenas-Alvarez] reasoned that it made sense to consult how a state court would interpret its own State’s laws.” Id. at 859 (emphasis added). So, the fact that the Justin Taylor Court did not look past the text of the federal Hobbs Act says nothing about what a federal judge must do when interpreting a state statute. If anything, the Justin Taylor decision reaffirmed the rule announced in Duenas-Alvarez, and repeated in James, Curtis Johnson, and Moncrieffe.
It’s no answer to say that the actual-case rule is limited to cases of a “facial mismatch” between the state statute and the relevant federal counterpart. See post, at 16 (Elrod, J., dissenting from denial of rehearing en banc). The Supreme Court in James rejected this attempted distinction. In that case, the question was whether Florida’s attempt statute swept beyond ACCA’s then-operative residual clause. 550 U.S. at 201–02. While Florida’s law appeared broader “[o]n its face,” the Court’s inquiry did not end with statutory text. Id. at 202. As the Court recognized, state courts are free to “considerably narrow[]” a statute’s application, so the Court followed the “consistent[] appli[cation]” of Florida law in Florida courts. Id. at 202–03; see also Cervenak, 135 F.4th at 365 (Thapar, J., dissenting) (observing that the actual-case rule “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’” (quoting James, 550 U.S. at 202)).
9
No. 25-30192
In this way, the Supreme Court has instructed that the actual-case rule does not disappear in the face of a facial mismatch. 2
B
Next, the dissent points to the supposed existence of a circuit split.
Post, at 17–18. Even assuming the dissenting opinion is correct about the nature and division of that split, I commend the en banc court for not trying to resolve it here. Consider three points.
First, the categorical approach has been with us for almost two generations. In that time, it has bedeviled the inferior federal courts. Predicting how it should apply is as hard as knowing the position and momentum of an electron. So if further innovation is warranted in this area, we should await direction from the Supreme Court.
Second, if and when the Supreme Court reconsiders the actual-case rule, it’s not obvious that it’ll overrule Duenas-Alvarez and Castillo-Rivera. True, the story of the categorical approach is generally one of ever-wider expansion. See Mathis v. United States, 579 U.S. 500, 536–38 (2016) (Alito, J., dissenting). But not always. Duenas-Alvarez, James, Curtis Johnson, and Moncrieffe all illustrate important limits on the doctrine.
_____________________
2 The dissent claims that my citation to Moncrieffe is “particularly puzzling”
because “[t]he Court did not look to see if there were cases in Georgia that had actually” been prosecuted in a way that was broader than the federal law. Post, at 18 n.3 (Elrod, C.J., dissenting from the denial of rehearing en banc). That itself is a puzzling assertion. In determining that the Georgia statute was overbroad, the Court declared, “we know that Georgia prosecutes this offense” in an overbroad way—and then cited two state court cases that proved the Georgia statute’s actual overbreadth. Moncrieffe, 569 U.S. at 194 (citing Taylor v. State, 581 S.E.2d 386, 388 (Ga. App. 2003), and Hadden v. State, 353 S.E.2d 532, 533–534 (Ga. App. 1987)); see also Brief for the Petitioner at 19, Moncrieffe v. Holder, 569 U.S. 184 (No. 11-702) (citing Georgia cases).
10
No. 25-30192
Third, it’s obviously the Supreme Court’s prerogative to decide whether to take a case and whether to extend a doctrine. Just as it’s the Supreme Court’s prerogative alone to overrule Duenas-Alvarez. See Rodriguez de Quijas v. Shearson/Am. Express, Inc., 490 U.S. 477, 484 (1989). But from the perspective of an inferior court, it’s noteworthy that the Supreme Court has considered and refused at least 23 cert petitions challenging the actual-case rule from Duenas-Alvarez. See Brief for the Respondent in Opposition at 14 n.3, Zuniga-Ayala v. Garland, 145 S. Ct. 768 (No. 24-103) (collecting 22); Zuniga-Ayala v. Garland, 145 S. Ct. 768 (2024) (mem.) (the 23rd). So it’s not immediately obvious why the 24th would be different.
*
I therefore concur in the en banc court’s lopsided vote to deny rehearing.
11
No. 25-30192
Jennifer Walker Elrod, Chief Judge, joined by Higginson, Circuit Judge, dissenting from the denial of rehearing en banc:
“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.” United States v. Escalante, 933 F.3d 395, 406 (5th Cir. 2019) (Elrod, J.). It often leads to “counterintuitive,” “absurd” conclusions and “reality-defying distinctions.” Id. at 407 (citation omitted). But we are bound to follow the decisions of the Supreme Court. In this area, our precedent does not.
In an attempt to tame the categorical approach’s tentacles, this court in United States v. Castillo-Rivera imposed an “actual case” requirement: A defendant must identify an actual case in which a state prosecuted conduct falling outside the federal statute even where the state statute is, on its face, broader than its federal counterpart. See 853 F.3d 218 (5th Cir. 2017) (en banc). I was in the majority in that case. At the time, I believed the “actual case” requirement to be a faithful application of the law. However, the Supreme Court in United States v. Taylor made clear that the “actual case” requirement is not the law. 596 U.S. 845 (2022). All ten of our regional sister circuits agree. We do not need to wait for the Supreme Court to correct us when it provides us with clear guidance. Accordingly, I respectfully dissent from the denial of rehearing en banc.
I
Defendant Julius Augillard pleaded guilty to two counts of unlawful possession of a firearm, in violation of 18 U.S.C. § 922(g)(1). Because one count involved a semiautomatic firearm capable of accepting a large-capacity magazine and Augillard had a prior felony for a “controlled substance offense”—a 2011 Louisiana conviction for possession with intent to
12
No. 25-30192
distribute cocaine—the district court set his base offense level at 22. See U.S.S.G. § 2K2.1(a)(3). It sentenced him to 70 months’ imprisonment.
Section 2K2.1 borrows the definition of “controlled substance offense” from Section 4B1.2(b): an offense “under federal or state law” that prohibits “possession of a controlled substance” with intent to distribute. U.S.S.G. § 4B1.2(b). The Guidelines do not further define “controlled substance.” We have held that “controlled substance” in the Guidelines refers only to substances covered by the federal Controlled Substances Act. United States v. Gomez-Alvarez, 781 F.3d 787, 794 (5th Cir. 2015).
Here is the wrinkle: The current federal CSA expressly excludes Ioflupane (a radioactive cocaine derivative used in medical imaging), while Louisiana’s 2011 cocaine definition did not. See United States v. Minor, 121 F.4th 1085, 1089 (5th Cir. 2024) (holding that we compare the predicate conviction with the federal CSA schedules as they exist “at the time of the current sentencing”). 1 Augillard contends that this facial mismatch disqualifies his prior conviction under the categorical approach. Faithfully applying our en banc precedent, the district court and the panel held that the facial mismatch did not disqualify the prior conviction because Augillard could not point to a Louisiana prosecution for possession of Ioflupane. See Castillo-Rivera, 853 F.3d at 223 (holding that there is “no exception to the actual case requirement” even “where a court concludes a state statute is broader on its face”).
_____________________
1 The federal CSA did not exclude Ioflupane until 2015, so if we were to compare the 2011 Louisiana Controlled Substances Act with the 2011 federal Controlled Substances Act, there would be no mismatch.
13
No. 25-30192
II
Our “actual case” requirement not only places us on the short side of a 10-1 circuit split, but, more importantly, puts our circuit’s precedent in tension with the Supreme Court’s. First, I provide some background on the categorical approach and the origins of our “actual case” requirement. I then discuss why we should have revisited the issue en banc.
A
The “categorical approach” is an analytical tool for classifying offenses in specific immigration and criminal law contexts where prior offenses matter. Under the categorical approach, a court must compare “the statutory definitions of the prior offenses” to the relevant federal definition. Shular v. United States, 589 U.S. 154, 157 (2020). “A prior state conviction . . . does not qualify” as a predicate if the definition of the crime in the relevant state statute is “broader than” that of the relevant federal category. United States v. Stitt, 586 U.S. 27, 32 (2017) (citing Mathis v. United States, 579 U.S. 500, 504 (2016)).
Applying the categorical approach, courts must identify “‘the least of th[e] acts’ criminalized by the state statute” and “determine whether that conduct would fall within the federal definition of the crime.” Esquivel-Quintana v. Sessions, 581 U.S. 385, 389 (2017) (alteration in original) (quoting Johnson v. United States, 559 U.S. 133, 137 (2010)). “If any—even the least culpable—of the acts criminalized” by the state statute falls outside the federal definition, “the statute of conviction does not categorically match the federal standard, and so cannot serve as a[] [qualifying] predicate.” Borden v. United States, 593 U.S. 420, 424 (2021).
In United States v. Castillo-Rivera, we held en banc that a defendant cannot “simply point to certain crimes that may be included in [the state offense] but not the [federal offense].” 853 F.3d at 222. Rather, we held that
14
No. 25-30192
a defendant must show that the state courts “have actually applied” the state statute in the way that falls outside the federal statute. Id.
A word on where this rule comes from: In Gonzalez v. Duenas-Alvarez, 549 U.S. 183 (2007), an immigration case, Duenas-Alvarez faced removal for an “aggravated felony,” which the INA defined to include “a theft offense.” 8 U.S.C. § 1101(a)(43)(G). His predicate conviction was for vehicle theft under California law, which expressly swept in “a party or an accessory to or an accomplice in the driving or unauthorized taking.” Cal. Vehicle Code § 10851(a). The Ninth Circuit held that § 10851(a) was categorically broader than generic theft because it reached aiders and abettors, not just principal takers. See Penuliar v. Gonzalez, 435 F.3d 961 (9th Cir. 2006).
The Supreme Court reversed. It held that generic theft includes accomplice liability. 549 U.S. at 189–90. Then after concluding that the text of the California statute was a match, the Court turned to Duenas-Alvarez’s argument that California applies its accomplice-liability doctrine so expansively that § 10851(a) reaches conduct beyond generic theft. Id. at 190–93. The Court concluded that a state crime outside the generic definition “requires more than the application of legal imagination to a state statute’s language”; it requires “a realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside the generic definition of a crime.” Id. at 193. The offender makes that showing by “point[ing] to his own case or other cases in which the state courts in fact did apply the statute in the special (nongeneric) manner for which he argues.” Id. Our court looked to this language in Duenas-Alvarez when devising our actual case requirement. Castillo-Rivera, 853 F.3d at 223.
Duenas-Alvarez was not about a situation where the texts of the state and federal statutes conflict. The text of the state statute in that case was consistent with the federal statute. The claimed overbreadth rested entirely
15
No. 25-30192
on how California courts might atypically construe its aider-and-abettor doctrine. The “actual case” requirement was the Court’s solution to speculative arguments about state-court interpretation. The Court was not confronting a statute that was overbroad on its face. In other words, the Court in Duenas-Alvarez left open the question of whether courts should apply the “actual case” requirement when the text of the state statute demonstrates that the state statute reaches conduct that the federal statute does not.
Our en banc court in Castillo-Rivera later grappled with that open question. Our court held in an 8-7 decision that there “is no exception to the actual case requirement articulated in Duenas-Alvarez where a court concludes a state statute is broader on its face.” 853 F.3d at 223. “[I]nterpreting a state statute’s text alone is simply not enough to establish the necessary ‘realistic probability,’” even when, as in Castillo-Rivera, the text of the state statute “is plainly broader than its federal counterpart.” Id.
B
We should have revisited our “actual case” holding in Castillo-Rivera en banc for three reasons.
First, our holding in Castillo-Rivera is inconsistent with intervening Supreme Court precedent: United States v. Taylor, 596 U.S. 845 (2022).
In Taylor, the Supreme Court addressed whether attempted Hobbs Act robbery qualifies as a “crime of violence” under 18 U.S.C. § 924(c)(3)(A)’s elements clause, which requires that the offense have “as an element the use, attempted use, or threatened use of physical force.” Applying the categorical approach, the Court said no. Taylor, 596 U.S. at 851. Attempted Hobbs Act robbery requires only: (1) an intent to take property by force or threat of force, and (2) a substantial step toward that end. Id. (citing United States v. Resendiz-Ponce, 549 U.S. 102, 107 (2007)).
16
No. 25-30192
The Court posited a hypothetical in which a defendant plans to rob a store, buys supplies, plots an escape route, recruits a getaway driver, and writes a threatening note. Id. at 851–52. When the defendant crosses into the store, the police immediately arrest him. Id. at 852. There is no question that the defendant committed the crime of an attempted Hobbs Act robbery. Id. And there is equally no question that the defendant did not use, attempt to use, or threaten physical force. Id.
The government invoked Duenas-Alvarez, faulting the defendant “for failing to identify a single case in which it has prosecuted someone for attempted Hobbs Act robbery without proving a communicated threat.” Id. at 857. The Court rejected that argument, stressing the “oddity of placing a burden on the defendant to present empirical evidence about the government’s own prosecutorial habits,” particularly in “a world where most cases end in plea agreements, and not all of those cases make their way into easily accessible commercial databases.” Id.
The Court distinguished Duenas-Alvarez on two bases. First, the issue in Duenas-Alvarez would have required a federal court to make a judgment about the meaning of a state statute, which implicates federalism concerns. Id. at 858–59. Second, “in Duenas-Alvarez 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.’” Id. at 859 (quoting Duenas-Alvarez, 549 U.S. at 193).
Taylor thus instructs that the realistic-probability test is inapplicable when either: (1) the text of the predicate statute is a mismatch with the federal definition, or (2) where the predicate offense is a federal crime. Neither circumstance implicates the federalism concern that animated Duenas- Alvarez. A facial mismatch requires no prediction because there is “no
17
No. 25-30192
overlap to begin with.” Taylor, 596 U.S. at 859. The court need not “make a judgment about the meaning of a state statute” at all—the court needs only to read it. Id. The second situation does not implicate federalism concerns because even if there is a facial match between the predicate offense and the federal statute, if the predicate offense is a federal law, federalism is not implicated because the court is making a judgment about the meaning of a federal statute, not a state one. 2 Second, we are alone in a 10-1 circuit split. Every other regional circuit has rejected an “actual case” requirement when a state statute is on its face broader than its federal counterpart. See Swaby v. Yates, 847 F.3d 62, 66 (1st Cir. 2017) (holding that there is no need to show “a realistic probability that the state actually will prosecute offenses involving [a] particular drug” when “drug schedules make clear that the Rhode Island offense covers at least one drug not on the federal schedules”); Hylton v. Sessions, 897 F.3d 57, 63 (2d Cir. 2018) (“The realistic probability test is obviated by the wording of the state statute, which on its face extends to conduct beyond the definition of the corresponding federal offense.”); Salmoran v. Att’y Gen., 909 F.3d 73, 81–82 (3d Cir. 2018) (holding that there is no need to identify “actual prosecution where the statute expressly authorizes the state government to enforce broader conduct”); Gordon v. Barr, 965 F.3d 252, 260 (4th Cir. 2020) (holding that “when the state, through plain statutory language, has defined the reach of a state statute to include conduct that the federal offense does not, the categorical analysis is complete”); United States v. Cervenak, 135 F.4th 311, 326 (6th Cir. 2025) (en banc) (holding that the “defendant need
_____________________
2 The concurring opinion states that I “posit[] that the Supreme Court sub silentio overruled Duenas-Alvarez” in Taylor. Ante, at 7. That is incorrect. Taylor calls out Duenas- Alvarez by name, distinguishes it, and clarifies the conditions that must be met for it to apply. 596 U.S. at 858–59.
18
No. 25-30192
not provide caselaw when they can point to the text of the state statute”); Aguirre-Zuniga v. Garland, 37 F.4th 446, 450 (7th Cir. 2022) (“If the statute is overbroad on its face under the categorical approach, the inquiry ends.”); Gonzalez v. Wilkinson, 990 F.3d 654, 660 (8th Cir. 2021) (similar); Lopez- Aguilar v. Barr, 948 F.3d 1143, 1147 (9th Cir. 2020) (holding the realistic- probability test met if “state statute expressly defines a crime more broadly than the generic offense”); United States v. Titties, 852 F.3d 1257, 1274–75 (10th Cir. 2017) (holding that there is no need to point to an actual case where the statute is facially overbroad); Aspilaire v. Att’y Gen., 992 F.3d 1248, 1255 (11th Cir. 2021) (holding that the defendant may demonstrate that statutory language itself, “rather than the application of legal imagination to that language,” satisfies realistic-probability test). 3
_____________________
3 Each of these decisions post-dates the three cases that the concurring opinion cites for the proposition that the Supreme Court “has required defendants to show that States actually apply their laws in ways that are broader than the relevant federal statute.” Ante, at 6–7. And those three cases do not prove the concurring opinion’s point. The first two are consistent with my proposed approach—we look to the statutory text to determine overbreadth unless either party shows that the state reads the text idiosyncratically. See James v. United States, 550 U.S. 192, 202 (2007) (noting that the Florida Supreme Court had narrowed the reach of certain statutory text); Johnson v. United States, 559 U.S. 133, 138 (2010) (noting that court is bound by the Florida Supreme Court’s construction of state statute).
The third case that the concurring opinion cites, Moncrieffe v. Holder, 569 U.S. 184 (2013), is particularly puzzling. The holding of Moncrieffe was that the Georgia Controlled Substances Act was not a categorical match to the federal Controlled Substances Act because the Georgia statute did not have an exception for social sharing of marijuana without remuneration, as the federal CSA did. Id. at 188–89. The Court did not look to see if there were cases in Georgia that had actually prosecuted social sharing a small amount of marijuana—it relied on the plain language of the statute. The language from Moncrieffe that the concurring opinion quotes was responding to a hypothetical situation without statutory text to guide it. Here, the plain text reveals the overbreadth of the state statute.
The concurring opinion is incorrect that Moncrieffe “declared, ‘we know that Georgia prosecutes this offense’ in an overbroad way.” Ante, at 9 n.2 (quoting Moncrieffe, 569 U.S. at 194). The full quote from the opinion is: “we know that Georgia prosecutes
19
No. 25-30192
On facts materially identical to ours—a state cocaine statute that swept in Ioflupane—the Eleventh Circuit held that the prior conviction did not qualify, despite the total absence of any Ioflupane prosecution. United States v. Miller, 157 F.4th 1365, 1378 (11th Cir. 2025). Three other circuits have agreed. United States v. Wilkes, 133 F.4th 600, 603 n.3 (6th Cir. 2025) (noting that “it does not matter” that the defendant’s conviction did not involve Ioflupane and rejecting argument that Duenas-Alvarez required defendant to show case in which Ioflupane possession was prosecuted); United States v. Perez, 46 F.4th 691, 701 (8th Cir. 2022), abrogation on other grounds recognized by, United States v. Gordon, 111 F.4th 899, 901 n.4 (8th Cir. 2024) (“Because Perez’s state statute of conviction included Ioflupane whereas the CSA specifically excludes Ioflupane as a controlled substance, the state statute is overbroad on its face, and our categorical analysis ends with the text of the statute.”); United States v. Holliday, 853 F. App’x 53, 55 (9th Cir. 2021).
Our across-the-board “actual case” requirement has created a geographic disparity in immigration and criminal law. En banc review was warranted. See Fed. R. App. P. 40(b)(2)(C).
Third, this issue is a “question[] of exceptional importance.” Fed. R.
App. P. 40(b)(2)(D). The categorical approach applies to a variety of different situations. See, e.g., Descamps v. United States, 570 U.S. 254 (Armed
_____________________
this offense when a defendant possesses only a small amount of marijuana, see, e.g., Taylor v. State, 260 Ga. App. 890, 581 S.E.2d 386, 388 (2003) (6.6 grams), and that ‘distribution’ does not require remuneration, see, e.g., Hadden v. State, 181 Ga. App. 628, 628–629 (1987).” This not the showing that Castillo-Rivera requires. The Court cited Taylor to establish that Georgia prosecutes distribution offenses involving small amounts of marijuana and Hadden to establish that the Georgia law does not require remuneration. But it did not identify an actual Georgia prosecution involving the relevant combination—distribution of a small amount of marijuana for no remuneration (i.e., social sharing).
20
No. 25-30192
Career Criminal Act); Moncrieffe v. Holder, 569 U.S. 184 (2013) (INA); United States v. Taylor, 596 U.S. 845 (2022) (statutory sentencing enhancement for use of firearm in crime of violence); United States v. Sereal, 153 F.4th 493, 496–99 (5th Cir. 2025) (Sentencing Guidelines).
And as the panel opinion notes, we have, post-Taylor, employed the “actual case” requirement to the detriment of immigration petitioners and criminal defendants in cases regarding Guidelines enhancements, United States v. Russell, 136 F.4th 606, 612 (5th Cir. 2025), removability under the INA, Ponce v. Garland, 70 F.4th 296 (5th Cir. 2023), and the Armed Career Criminal Act, United States v. Kerstetter, 2025 WL 1079071 (5th Cir. Apr. 10, 2025). The district courts within our circuit (and the Board of Immigration Appeals) confront the categorical approach even more often than we do.
III
The categorical approach crushes common sense. The legally correct result here can fairly be described as counterintuitive, to say the least: A defendant’s cocaine-trafficking conviction loses its force as a predicate offense because of a radioactive diagnostic compound that, so far as anyone can tell, no person has ever abused. If the question before us was whether I would design such a system, I would vote no. But the question is not what we would prefer the law to be. Instead, it is about what the law requires.
If the categorical approach’s consequences are intolerable, the remedy lies with Congress and the Sentencing Commission, which remain free to write enhancements and removability requirements in terms of conduct rather than convictions. Until they do, the law requires what it requires—and it requires the same thing in this circuit as it does everywhere else.
21
No. 25-30192
***
I joined the majority opinion in Castillo-Rivera because I believed at the time that adherence to Supreme Court precedent required it. My commitment to following Supreme Court precedent has not changed, but the Court’s guidance on the categorical approach has. Accordingly, I respectfully dissent from denial of rehearing en banc.