UNITED STATES v. ARNETT (2026)

Oklahoma City Police Officers stopped a vehicle driven by Glen Delane Arnett, Jr. During the stop, they noticed Mr. Arnett was sitting on a handgun. He admitted he was a convicted felon. The officers arrested Mr. Arnett, searched the vehicle, and discovered several ounces of marijuana. The government indicted Mr. Arnett on three charges but the jury convicted him of just one: being a felon in possession of a firearm in violation of 18 U.S.C. § 922(g)(1). The district court imposed a 77-month prison sentence.

This appeal concerns Mr. Arnett's sentence. In calculating the advisory Guidelines range, the district court added four levels to Mr. Arnett's base offense under U.S.S.G. § 2K2.1(b)(6)(B), an enhancement that applies if the defendant “possessed any firearm ․ in connection with another felony offense[.]” The district court found “another felony offense” by treating Mr. Arnett's otherwise simple marijuana possession as a felony under 21 U.S.C. § 844 based on his prior Oklahoma drug convictions. That statute makes it a felony to possess a “controlled substance” after, as relevant here, “a prior conviction for any drug, narcotic, or chemical offense chargeable under the law of any State[.]” Id. § 844(a). Mr. Arnett now appeals, challenging the application of the § 2K2.1(b)(6)(B) enhancement. Mr. Arnett did not raise in the district court the arguments he makes on appeal, and he appropriately concedes our review is only for plain error. It is that standard of review that ultimately proves dispositive. Exercising jurisdiction under 18 U.S.C. § 3742(a) and 28 U.S.C. § 1291, we affirm.

I

A 1

One night in February 2023, Sergeant Cortland Glover and his partner were patrolling the parking lot of an “after-hours club.” RI.269. Police “monitor[ed] the area” because there had been “recent shootings there” and several “violent cases.” RI.269. Sergeant Glover grew suspicious after observing a black Hyundai Sonata back into a parking space. Nobody got out of the car. Sergeant Glover parked at a gas station across the street so he and his partner could “monitor” the large crowd outside the nightclub. RI.269, 285. The officers watched the Sonata for about five to ten minutes. During that time, they saw “[m]ultiple people come to and from [the] vehicle[.]” RI.269–70.

Sergeant Glover followed the Sonata when it left the nightclub parking lot. He initiated a traffic stop after the vehicle failed to signal a right turn. The Sonata “slow roll[ed]” to a stop, and Sergeant Glover observed “some odd movement in the vehicle” as the driver “reached back or something[.]” RI.270. He also noticed a “very strong[ ]” odor of marijuana coming from the Sonata. RI.274. Sergeant Glover approached the car and observed Mr. Arnett, the sole occupant, in the driver's seat. His behavior struck Sergeant Glover as “a little suspicious” because he was “staring at his phone” and would not “make eye contact.” RI.271.

Sergeant Glover asked Mr. Arnett for his driver's license. As Mr. Arnett “lean[ed] over to ․ his right side to retrieve his license from his pocket,” Sergeant Glover observed “a pistol tucked underneath his left thigh.” RI.271. The officers removed Mr. Arnett from the Sonata and placed him in handcuffs. The gun “was fully loaded” and had a “round in the chamber ready to fire.” RI.279. Sergeant Glover noticed the gun had marijuana “all over it.” RI.276. The chamber also was “matted with marijuana.” RI.276. Mr. Arnett told the officers he was a convicted felon, and they arrested him. RI.273.

After “securing” Mr. Arnett in the police cruiser, Sergeant Glover searched the Sonata. RI.274. He found a “brown paper sack ․ in the front passenger seat.” RI.275. There “were other bags of marijuana” inside the sack. RI.275. He also found a “Walmart-style sack in the front passenger side floorboard,” which contained “more bags of marijuana[.]” RI.275. In the backseat, Sergeant Glover discovered “a large digital scale” and “three or four cell phones.” RI.275. Officers seized 267 grams (or roughly eight ounces) of marijuana from the Sonata.

B

The government indicted Mr. Arnett on three counts: (1) being a felon in possession of a firearm, in violation of 18 U.S.C. § 922(g)(1); (2) possessing with intent to distribute “a quantity of marijuana,” in violation of 21 U.S.C. § 841(a)(1); and (3) possessing a firearm in furtherance of a drug-trafficking crime, in violation of 18 U.S.C. § 924(c)(1)(A). RI.18–20. Trial lasted three days. Mr. Arnett represented himself with the assistance of standby counsel.

Mr. Arnett put his girlfriend, Shawniece Latimer, on the stand. Ms. Latimer testified she owned the Sonata and that Mr. Arnett was driving it the morning of his arrest to “go see [his] brother in the hospital.” RI.434, 442. She also claimed to own the gun found in the car. Ms. Latimer told the jury she placed the gun in the Sonata “[p]robably ․ the day before” Mr. Arnett was arrested but did not tell him it was in the vehicle. RI.436–37. Ms. Latimer confirmed on cross examination that she “t[ook] care of” the gun and “k[ept] it clean.” RI.449. She explained there “shouldn't have been” any marijuana on the gun “unless it was in the bag with the marijuana.” RI.449. And she denied “leav[ing] the gun in the bag with the marijuana[.]” RI.449.

Ms. Latimer also testified the marijuana seized from the Sonata belonged to her. She purchased the marijuana “sometime around mid 2022” with a medical marijuana card but had not smoked it because it was “old” and she “got sick off of it.” RI.439, 444. Ms. Latimer confirmed she intended to “get rid of” the marijuana. RI.444–45. When pressed by the government about the packaging in which the marijuana was discovered, Ms. Latimer maintained she “put it all in one bag so [she] d[id not] have a bunch of little baggies just laying around.” RI.445–46. As with the handgun, Ms. Latimer did not inform Mr. Arnett there was marijuana in the vehicle.

After the jury retired for deliberations, Mr. Arnett moved for a judgment of acquittal under Federal Rule of Criminal Procedure 29. Mr. Arnett argued the evidence was insufficient to establish that he possessed the marijuana for purposes of the intent-to-distribute charge because “[t]here w[ere] no fingerprints” and “there was no analysis of” the suspected marijuana Sergeant Glover saw on the handgun. RI.518. The government insisted the evidence established beyond a reasonable doubt that Mr. Arnett knowingly possessed the marijuana. It reiterated Ms. Latimer's testimony “that she had not possessed th[e] gun” and did not leave “the marijuana on the firearm,” noted the marijuana's “close proximity” to Mr. Arnett, and emphasized Mr. Arnett was “the only person in the vehicle that night[.]” RI.516.

The district court denied the motion. The evidence of Mr. Arnett's possession of the marijuana was, in its view, “overwhelming.” RI.519. Mr. Arnett was “the only person in the car[,]” the district court reasoned. RI.519. “That's possession.” RI.519. The jury deliberated for two hours. It convicted Mr. Arnett of unlawful possession of a firearm but acquitted him of possession with intent to distribute and possession of a firearm in furtherance of a drug trafficking crime.

C

The case proceeded to sentencing.² In calculating Mr. Arnett's advisory Guidelines range, the PSR recommended a four-point enhancement to Mr. Arnett's base offense level under U.S.S.G. § 2K2.1(b)(6)(B) because he “possessed the [gun] in connection with the felony offense of possession of marijuana after a prior conviction for a drug offense.” RII.32. The PSR identified four prior drug-related state convictions: (1) a 2004 conviction for possession of a controlled dangerous substance (marijuana) with intent to distribute, in violation of Okla. Stat. Ann. tit. 63, § 2-401; (2) a 2004 conviction for possession of a controlled dangerous substance (cocaine) with intent to distribute, in violation of Okla. Stat. Ann. tit. 63, § 2-401; (3) a 2004 conviction for possession of a controlled dangerous substance (marijuana and cocaine) without a tax stamp, in violation of Okla. Stat. Ann. tit. 63, § 450.3; and (4) a 2008 conviction for possession of a controlled dangerous substance (marijuana), in violation of Okla. Stat. Ann. tit. 63, § 2-402.³

Mr. Arnett objected to the § 2K2.1(b)(6)(B) enhancement on three grounds. First, he claimed the handgun “neither facilitated” his possession of the marijuana “nor had the potential to facilitate this offense.” RII.53. Second, while Mr. Arnett “concede[d] that under federal law simple possession of drugs can be a felony (21 U.S.C. § 844),” he argued the Biden Administration “d[id] not appear” to be “enforcing this statutory provision with respect to marijuana offenses.” RII.54. And third, Mr. Arnett observed the jury acquitted him of possession with intent to distribute. In his view, this acquittal meant “the jury ostensibly accepted that the marijuana in the vehicle belonged to” Ms. Latimer, so a four-point enhancement “in this circumstance would have the effect of counting acquitted conduct as relevant conduct.” RII.55.

At the sentencing hearing, Mr. Arnett focused on two issues: (1) whether he possessed the marijuana found in the car; and (2) whether his possession of the handgun was “connected to” his possession of the marijuana. Only the first issue is relevant to this appeal. On that front, the district court found the trial evidence “pretty clear” that Mr. Arnett was “in possession of the marijuana.” RIII.103. In fact, “more than a preponderance of the evidence” supported a finding of possession. RIII.104. The district court was persuaded by “the significant quantity of marijuana ․ in the car[,]” the marijuana being “pretty open and obvious in a bag,” Mr. Arnett “being the sole occupant of the car, the marijuana being in very close proximity to him and the firearm in the passenger seat, [and Mr. Arnett] telling the officer immediately after being pulled over that there was marijuana in the car[.]” RIII.104, 108, 111. These facts “indicate that Mr. Arnett was aware that there was marijuana in the car and [he] was in possession of the marijuana that was in the car.” RIII.111.

The district court addressed the implications of Mr. Arnett's acquittal for possession with intent to distribute. As the district court explained, “[t]he question is” whether Mr. Arnett was “in possession of the marijuana.” RIII.107. Possession with intent to distribute, however, is “a different crime.” RIII.103. The trial evidence was “pretty clear” Mr. Arnett “w[as] in possession,” the district court reasoned, “the jury just didn't believe that [he] w[as] selling or distributing.” RIII.107. The district court also rejected Mr. Arnett's ownership arguments, concluding ownership is “irrelevant to possession[.]” RIII.104. In the district court's view, “[e]ven if [it] believed” the marijuana in the Sonata belonged to Ms. Latimer, the dispositive inquiry is whether Mr. Arnett was “in possession of the marijuana as the sole occupant of the car ․ and whether there[ ] [was] a nexus between [him] and the marijuana.” RIII.108. It clarified Mr. Arnett “wouldn't have had to have touched [the marijuana] to possess it.” RIII.109.

Based on this reasoning, the district court overruled Mr. Arnett's objection to the § 2K2.1(b)(6)(B) four-point enhancement.⁴ Mr. Arnett had a total offense level of 24 and a criminal history category of IV. This resulted in an advisory Guidelines range of 77 to 96 months’ imprisonment.⁵ The district court ultimately sentenced Mr. Arnett to 77 months in prison followed by three years of supervised release. In imposing the sentence, the district court acknowledged “the plus-four enhancement [under § 2K2.1(b)(6)(B)] was a relatively close call,” but stated even if it had sustained Mr. Arnett's objection it “was going to end up at the exact same sentence[.]” RIII.127.

This timely appeal followed.

II

Mr. Arnett advances two challenges to the district court's application of the § 2K2.1(b)(6)(B) enhancement. He concedes that plain-error review applies to both arguments because he did not raise either in the district court. First, Mr. Arnett argues the district court plainly erred by treating his simple possession of marijuana as a felony under 21 U.S.C. § 844. Second, and alternatively, he contends the district court plainly erred by applying the wrong legal standard to conclude he possessed the marijuana found in the Sonata.

Where, as here, “an appellant raises a forfeited argument on appeal, we will reverse only if the appellant can satisfy our rigorous plain-error test.” United States v. Leffler, 942 F.3d 1192, 1196 (10th Cir. 2019). We will not reverse for plain error unless Mr. Arnett demonstrates “(1) error, (2) that is plain, which (3) affects substantial rights, and which (4) seriously affects the fairness, integrity, or public reputation of judicial proceedings.” United States v. Paycer, 154 F.4th 1261, 1271 (10th Cir. 2025). “Failure to establish any one of these elements precludes reversal.” United States v. Romero, 132 F.4th 1208, 1218 (10th Cir. 2025). Plain error is a “demanding standard,” United States v. McGehee, 672 F.3d 860, 876 (10th Cir. 2012), and “intentionally difficult to satisfy[,]” Paycer, 154 F.4th at 1271.

Applying this standard, we now consider Mr. Arnett's arguments.

III

Mr. Arnett argues the district court plainly erred by treating his simple possession of marijuana as a felony under 21 U.S.C. § 844. Mr. Arnett has demonstrated error, but we cannot agree the error is plain.

A

Under our plain error framework, Mr. Arnett must first show the district court “committed error.” United States v. Vazquez-Garcia, 130 F.4th 891, 897 (10th Cir. 2025) (quoting United States v. Magallanez, 408 F.3d 672, 678 (10th Cir. 2005)).

When, as here, a defendant is convicted of unlawfully possessing a firearm, § 2K2.1(b)(6)(B) adds four points to his base offense level if he “used or possessed any firearm ․ in connection with another felony offense[.]” U.S.S.G. § 2K2.1(b)(6)(B) (2023).⁶ The Guideline commentary defines “ ‘[a]nother felony offense’ for purposes of subsection (b)(6)(B)” as “any federal, state, or local offense ․ punishable by imprisonment for a term exceeding one year, regardless of whether a criminal charge was brought, or a conviction obtained.” Id. § 2K2.1, cmt. n.14(C).

In this case, Section 844 is the “other felony offense” for purposes of § 2K2.1(b)(6)(B).⁷ That statute says an individual may be imprisoned up to two years if he “knowingly or intentionally [ ] possess[es] a controlled substance ․ after a prior conviction under this subchapter ․ or a prior conviction for any drug, narcotic, or chemical offense chargeable under the law of any State[.]” 21 U.S.C. § 844(a). A “drug, narcotic, or chemical offense means any offense which proscribes the possession, distribution, manufacture, cultivation, sale, transfer, or the attempt or conspiracy to possess, distribute, manufacture, cultivate, sell or transfer any substance the possession of which is prohibited under this subchapter.” Id. § 844(c). Thus, when someone “knowingly or intentionally” possesses a “controlled substance” and already has (1) a conviction under § 844(a) or (2) a “conviction for any drug, narcotic, or chemical offense chargeable under the law of any State,” he can be imprisoned for over a year—the definition of “another felony offense” under § 2K2.1(b)(6)(B). See Carachuri-Rosendo v. Holder, 560 U.S. 563, 567–68, 130 S.Ct. 2577, 177 L.Ed.2d 68 (2010) (noting that “recidivist simple possession” “may be punishable as a felony” under the Controlled Substances Act).

Mr. Arnett's simple marijuana possession qualifies as a felony under § 844—and therefore can be considered “another felony offense” for purposes of § 2K2.1(b)(6)(B)—only if he has at least one prior “conviction for any drug, narcotic, or chemical offense chargeable under the law of any State[.]”⁸ 21 U.S.C. § 844(a). Mr. Arnett says he lacks a qualifying criminal history because none of his prior Oklahoma drug convictions is a “drug, narcotic, or chemical offense” as defined in § 844(c). Op. Br. at 13.

The crux of Mr. Arnett's argument is that the district court should have applied “the categorical approach” to § 844(c) to determine whether any of his prior Oklahoma drug convictions qualify as “drug, narcotic, or chemical offense[s].” Op. Br. at 14–15. The categorical approach “focus[es] solely on whether the elements of the crime of conviction sufficiently match the elements of [the federal statute], while ignoring the particular facts of the case.” Mathis v. United States, 579 U.S. 500, 504, 136 S.Ct. 2243, 195 L.Ed.2d 604 (2016). And under the categorical approach, Mr. Arnett claims all four of his prior Oklahoma convictions are “overbroad” because the underlying statutes of conviction proscribed possessing or trafficking substances that were not “controlled substances” under federal law. Op. Br. at 18–22; see Johnson v. Barr, 967 F.3d 1103, 1106 (10th Cir. 2020) (explaining that under the categorical approach, when the state statute of conviction “criminalizes more substances” than those proscribed by the federal statute, “then the [state] statute is ‘overbroad,’ and no ‘categorical match’ exists between the [state] statute and its federal analogue” (quoting Descamps v. United States, 570 U.S. 254, 276–77, 133 S.Ct. 2276, 186 L.Ed.2d 438 (2013))).

The government suggests the factual approach—not the categorical approach—applies to § 844(c). “[T]he factual approach ‘look[s] not only at the terms of the statute of conviction, but also at the underlying facts.’ ”⁹ United States v. Abeyta, 877 F.3d 935, 940 (10th Cir. 2017) (alteration in original) (quoting United States v. Martinez-Hernandez, 422 F.3d 1084, 1087 (10th Cir. 2005)). Using this approach, the government argues Mr. Arnett's convictions for possession with intent to distribute qualify as “drug, narcotic, or chemical offense[s]” because “a federally banned substance was actually involved[.]” Ans. Br. at 16 (emphasis added). For those offenses, Mr. Arnett “was convicted of possession with intent to distribute ․ marijuana and cocaine”—both of which “are controlled substances under federal law.” Ans. Br. at 19.

According to Mr. Arnett, “the government ․ tacitly concedes that, if the categorical approach applies, Mr. Arnett lacks any qualifying predicate.” Reply Br. at 1–2. He suggests “the issue” on appeal therefore “reduces to whether § 844(c) requires a categorical or factual approach.” Reply Br. at 2. We agree with Mr. Arnett's framing of the dispositive issue.¹⁰ And for the reasons that follow, we conclude Mr. Arnett has established error.

Whether the categorical approach or the factual approach applies “depend[s] on the language” of the statute at issue. Martinez-Hernandez, 422 F.3d at 1087. The “determination of the applicable approach,” we have said, “is a question of statutory interpretation.” Abeyta, 877 F.3d at 940 (citing Taylor v. United States, 495 U.S. 575, 600, 110 S.Ct. 2143, 109 L.Ed.2d 607 (1990)). “We employ the categorical approach ‘when the language of the [federal statute] confines the court's inquiry to the terms of the [state] statute of conviction.’ ” Id. (quoting Martinez-Hernandez, 422 F.3d at 1086). But “we employ the factual approach ‘when the language of the [federal statute] requires courts to look at the specific facts underlying the prior [state] offense.’ ” Id. (quoting Martinez-Hernandez, 422 F.3d at 1087).

Mr. Arnett makes a convincing case for why the categorical approach applies to § 844(c). He argues “the language” of § 844(c) “unambiguously directs the court to look to the statute of conviction, not to the facts.” Op. Br. at 16. Recall, the statute defines “drug, narcotic, or chemical offense” as “any offense which proscribes the possession, distribution, manufacture, cultivation, sale, [or] transfer ․ [of] any substance the possession of which is prohibited under” federal law. 21 U.S.C. § 844(c) (emphasis added). Mr. Arnett claims § 844(c)’s “directive to focus on what the offense ‘proscribes’ settles” what approach applies in this case. Op. Br. at 16. “Facts,” he says, “do not ‘proscribe.’ Rather, the statute of conviction ‘proscribes.’ Thus, § 844(c) tells the court to base its analysis on the statute of conviction—i.e., to apply the categorical approach.” Op. Br. at 16.

In support of his textual argument, Mr. Arnett points to United States v. Herrera-Roldan, 414 F.3d 1238 (10th Cir. 2005) and United States v. Martinez-Hernandez, 422 F.3d 1084 (10th Cir. 2005), two cases applying the categorical approach in the context of the Armed Career Criminal Act (ACCA) to enhancements with text similar to § 844(c)’s. In Herrera-Roldan, we applied the categorical approach to an enhancement defining “ ‘drug trafficking offense’ as ‘an offense under federal, state, or local law that prohibits the ․ possession of a controlled substance[.]’ ” 414 F.3d at 1240 (emphasis added) (quoting U.S.S.G. § 2L1.2(b)(1), cmt. n.1(B)(iv) (2003)). The “focus” of the “drug trafficking offense” definition, we said, was “not on the defendant's conduct, but on what the state law prohibits,” meaning “our inquiry” was “confine[d] ․ to the terms of the statute of conviction[.]” Id. at 1241. The enhancement at issue in Martinez-Hernandez had identical “prohibits” language. See 422 F.3d at 1087. There too we applied the categorical approach because the enhancement defined “firearms offense” as “[a]n offense under Federal, state, or local law that prohibits the possession of” certain kinds of firearms. See id. (emphasis added) (quoting U.S.S.G. § 2L1.2(b)(1), cmt. n.1(B)(v)(II) (2003)). We once again concluded the phrase “that prohibits” “focus[es] on the terms of the statute of conviction” by “ask[ing] us to look at what the statute prohibits.” Id.

Mr. Arnett contends the word “proscribes is just a synonym for ‘prohibits’ ” and “confines the inquiry to the terms of the statute of conviction in just the same way[.]” Op. Br. at 17. We agree.

Like Mr. Arnett, we see little—if any—difference between § 844(c)’s use of “proscribes” and the “that prohibits” language at issue in Herrera-Roldan and Martinez-Hernandez. The term “proscribes” is more or less “synonymous with ‘prohibits[.]’ ” Op. Br. at 30; see Proscribe, Black’s Law Dictionary (12th ed. 2024) (defining “proscribe” as “[t]o outlaw or prohibit; to forbid officially” (emphasis added)). Thus, just as the enhancement language in Herrera-Roldan and Martinez-Hernandez focused our inquiry on what the statute of conviction prohibited, here too § 844(c) confines us to the terms of the Oklahoma statutes underlying Mr. Arnett's prior drug convictions.¹¹ All that is to say, the categorical approach applies to § 844(c).

The government advances three counterarguments, but none is persuasive.

First, in support of the factual approach, the government likens § 844(c) to the statute at issue in Nijhawan v. Holder, 557 U.S. 29, 129 S.Ct. 2294, 174 L.Ed.2d 22 (2009). There the Supreme Court examined whether the categorical approach or the “circumstance-specific” approach ¹² applies to the “fraud and deceit” subprovision of 8 U.S.C. § 1101(a)(43). See Nijhawan, 557 U.S. at 34–37, 129 S.Ct. 2294. That subprovision defines “aggravated felony” to encompass—as relevant here—“an offense that ․ involves fraud or deceit in which the loss to the victim or victims exceeds $10,000[.]” 8 U.S.C. § 1101(a)(43)(M)(i). The Court concluded this language “is consistent with a circumstance-specific approach.” Nijhawan, 557 U.S. at 38, 129 S.Ct. 2294. “The words ‘in which’ (which modify ‘offense’),” the Court reasoned, “can refer to the conduct involved ‘in’ the commission of the offense of conviction, rather than to the elements of the offense.” Id. at 39, 129 S.Ct. 2294. Thus, § 1101(a)(43)(M)(i)’s “monetary threshold applies to the specific circumstances surrounding an offender's commission of a fraud and deceit crime on a specific occasion.” Id. at 40, 129 S.Ct. 2294.

We are not persuaded. There are meaningful textual differences between §§ 844(c) and 1101(a)(43)(M)(i) which render the government's comparison unconvincing. Most notably, § 1101(a)(43)(M)(i) does not “prohibit” or “proscribe” anything. It also lacks any comparable language confining a court's review to the terms of the statute of conviction. See Herrera-Roldan, 414 F.3d at 1241 (“confin[ing] our inquiry to the terms of the statute of conviction”); Martinez-Hernandez, 422 F.3d at 1087 (“focus[ing] on the terms of the statute of conviction”). And noticeably absent from § 844(c) is anything like § 1101(a)(43)(M)(i)’s “monetary threshold” language that might require us to examine “the specific circumstances surrounding an offender's commission of” a drug offense. See Nijhawan, 557 U.S. at 40, 129 S.Ct. 2294. The government does not point to any persuasive similarities between the statutes either. Instead, it recites Nijhawan’s facts and the text of § 1101(a)(43)(M)(i) before asserting, without more, that we “should conclude [§ 844(c)]—read in context—requires a court to determine what actually happened and whether a federally banned substance was actually involved[.]” Ans. Br. at 15–16. We cannot do so.

Second, the government insists Mr. Arnett “omit[s] one of the strongest textual clues against his position: the word ‘any.’ ” Ans. Br. at 16. “Congress used the word ‘any’ throughout § 844 when addressing the enhancement for recidivist drug offenders[,]” the government notes, and this “use of the word ‘any’ distinguishes § 844 from the other laws that were interpreted by the cases on which Mr. Arnett relies.” Ans. Br. at 16–17. The government emphasizes the “expansive meaning” of the word “any.” Ans. Br. at 17 (internal quotation marks omitted). It claims Congress “had ready language available in the ACCA” if it “wished to limit who qualified for a felony sentence under § 844(a) to only those with state convictions [that] were a categorical match or less restrictive than federal law[.]” Ans. Br. at 17.

We disagree. The word “any” does not inform our central inquiry—whether § 844(c) is referring to the statute underlying the prior conviction (the categorical approach) or to the defendant's particular conduct (the factual approach). As Mr. Arnett puts it, “the expansiveness of the word ‘any’ might suggest a liberal approach to deciding whether the thing we have to analyze—be it the statute or the facts—ultimately satisfies the terms of the enhancement.” Reply Br. at 5. It does not, however, “tell us which thing to analyze in the first instance.” Reply Br. at 5.

Third, the government offers policy considerations. It dismisses the categorical approach as “a maligned interpretative approach” and cautions us against “expanding the categorical approach further than it has already gone.” Ans. Br. at 17, 19; see also id. at 18 (“[S]cores of jurists have decr[ied] the categorical approach.” (second alteration in original) (internal quotation marks omitted)). We have recognized “the categorical approach can sometimes generate counterintuitive results,” “occasional anomalies,” and “odd outcomes.”¹³ United States v. Adams, 40 F.4th 1162, 1170 & n.7 (10th Cir. 2022). But even a righteous critique does not allow us to abandon the dictates of applicable law. Without more, we cannot ignore what the law requires.

Accordingly, we conclude Mr. Arnett has carried his burden on the first prong of the plain error standard. But that does not end our inquiry. For Mr. Arnett to prevail, he also must show the district court plainly erred by not applying the categorical approach to § 844(c). We turn to that question now.

B

Generally, “[a]n error is plain if it is clear or obvious under current, well-settled law.” United States v. Warrington, 78 F.4th 1158, 1167 (10th Cir. 2023) (quoting United States v. Wolfname, 835 F.3d 1214, 1221 (10th Cir. 2016)). And the “law is well-settled ․ if there is precedent directly on point from the Supreme Court or the Tenth Circuit, or if there is a consensus in the other circuits.” United States v. Egli, 13 F.4th 1139, 1146 (10th Cir. 2021). When the defendant does not produce “binding caselaw—either from the Supreme Court or from our court—that plainly answers the question,” United States v. Johnson, 157 F.4th 1309, 1323 (10th Cir. 2025), that failure usually “sounds the death knell ․ on plain-error review,” id. (alteration in original) (quoting United States v. Garcia, 946 F.3d 1191, 1210 (10th Cir. 2020)).

Still, “even without controlling caselaw, an error can be plain when the plain language of a rule or statute clearly settles the question.” United States v. Jones, 74 F.4th 1065, 1070 (10th Cir. 2023); see also Johnson, 157 F.4th at 1321 (recognizing “another legal source can render an error plain, too: the text of a statute or rule”); United States v. Story, 635 F.3d 1241, 1248 (10th Cir. 2011) (“[E]ven where there is no such precedent, we may find plain error where the district court has engaged in a ‘clearly erroneous’ application of statutory law.” (quoting United States v. Poe, 556 F.3d 1113, 1129 (10th Cir. 2009))). “[N]o matter the source relied on, the error must be ‘so clear or obvious that it could not be subject to any reasonable dispute.’ ” Johnson, 157 F.4th at 1321 (emphasis omitted) (quoting United States v. Berryhill, 140 F.4th 1287, 1302 n.3 (10th Cir. 2025)). We assess plainness based on the law “at the time of appeal.” United States v. Cordery, 656 F.3d 1103, 1107 (10th Cir. 2011).

1

Mr. Arnett contends “it is clear or obvious that a court must use the categorical approach to determine whether a defendant has a prior conviction that elevates his simple possession of drugs from a misdemeanor to a felony.” Op. Br. at 29. He acknowledges no case in the Tenth Circuit or Supreme Court has so held, so he relies directly on the statutory text. He points to the word “proscribes” in § 844(c), which he says “clearly settles the question” of whether the categorical approach applies. Op. Br. at 29. Mr. Arnett again invokes Herrera-Roldan and Martinez-Hernandez in support of his position. He claims those cases “are clearly generalizable to § 844(c)” and compel a finding of plainness here. Op. Br. at 29.

For its part, the government contends “Mr. Arnett has not cited to any Supreme Court or Tenth Circuit case where either court has found that the categorical approach applies to § 844.” Ans. Br. at 20. It acknowledges Mr. Arnett's reliance on “several cases involving the application of the categorical approach in other contexts”—i.e., Herrera-Roldan and Martinez-Hernandez—but emphasizes “there is no direct precedent on point.” Ans. Br. at 20. Rather, “there is something approaching the opposite,” the government explains, pointing to United States v. Shaw, No. 22-3251, 2023 WL 3335596, at *4 (10th Cir. May 10, 2023) (unpublished), which held the categorical approach does not apply to § 844. Ans. Br. at 20. The government recognizes Shaw is unpublished and not precedential. But when it comes to assessing the plainness prong, the government insists Shaw “is still evidence that the law was not clearly established in favor of applying the categorical approach to § 844.” Ans. Br. at 20. And the government reminds us “even an unpublished opinion can show that the law is not clearly to the contrary.” Ans. Br. at 20 (quoting Romero, 132 F.4th at 1218). After this appeal was submitted for decision, we issued another relevant unpublished decision—United States v. Simmons, No. 24-6077, 2025 WL 2374700, at *4 (10th Cir. Aug. 15, 2025) (unpublished)—also holding the categorical approach does not apply to § 844.¹⁴

Given the legal landscape, we cannot say Mr. Arnett has carried his burden to show the error is plain. To be sure, Herrera-Roldan and Martinez-Hernandez are “generalizable” to § 844(c) even though those cases involved enhancements in the ACCA context. See United States v. Venjohn, 104 F.4th 179, 187 (10th Cir. 2024) (noting that “to succeed on plainness,” a defendant need not “point to Supreme Court or circuit cases that precisely deal with the subject statute at issue” because “a case that sets forth a principle clearly generalizable to the subject statute” will suffice (quoting United States v. Faulkner, 950 F.3d 670, 680 (10th Cir. 2019))). The problem for Mr. Arnett is we have two unpublished decisions holding the categorical approach does not apply to § 844: Shaw and Simmons. We discuss them in turn.

In Shaw, the government petitioned the district court for revocation of Mr. Shaw's supervised release based on positive drug tests and “alleged these tests establish[ed] Grade B violations for possessing a controlled substance[.]” 2023 WL 3335596, at *1. Grade B violations encompass “conduct constituting any [ ] federal, state, or local offense punishable by a term of imprisonment exceeding one year.” Id. (alteration in original) (quoting U.S.S.G. § 7B1.1(a)(2)). The revocation petition “explained that [Mr. Shaw's] state drug conviction under Missouri law from 2002 made his simple drug possession during supervised release eligible for a prison term exceeding a year under federal law.” Id. The district court agreed. See id. at *2. Mr. Shaw then moved to correct his sentence, arguing “the district court should have applied the ‘categorical approach’ to determine whether his Missouri drug offense was a ‘prior [drug] conviction’ under 21 U.S.C. § 844(a).” Id. (alteration in original). The district court denied the motion. Id.

On appeal, Mr. Shaw, proceeding pro se, argued the district court erred in determining his supervised release violation was Grade B because his 2002 Missouri conviction was “ ‘not a controlled substance offense under the Federal Controlled Substance Act’ and thus c[ould not] serve as a predicate offense for calculating his criminal history category.” Id. A panel of this court rejected Mr. Shaw's argument. See id. at *3. The panel held “§ 844(a) does not require the prior offense to be a violation of” the Federal Controlled Substance Act because “[i]t simply refers to ‘any drug, narcotic, or chemical offense chargeable under the law of any State.’ ” Id. at *4. Notably, the panel was unpersuaded by Mr. Shaw's argument “that the elements of” the Missouri statute underlying his 2002 conviction “do not match the elements of § 844(a).” Id. “Although the categorical approach applies to serious drug offenses under the ACCA,” the panel explained, “it does not pertain to § 844(a).” Id. Mr. Shaw's argument for applying the categorical approach thus “ha[d] no relevance to whether his Missouri conviction constitute[d] ‘a prior conviction for any drug, narcotic, or chemical offense chargeable under the law of any State’ under § 844(a).” Id.

In Simmons, Mr. Simmons pleaded guilty to unlawful possession of a firearm after a search of his vehicle yielded “three firearms, ammunition, and small quantities of various drugs.” 2025 WL 2374700, at *1. His PSR detailed his “long rap sheet” of “numerous state drug-related offenses” and recommended a four-point enhancement under § 2K2.1(b)(6)(B) because he “possessed a firearm in connection with the felony offense of possession of [a] controlled dangerous substance[.]” Id. at *1, 3 (emphasis omitted). Mr. Simmons did not object to the four-point enhancement. Id. at *1.

On appeal, he argued the district court plainly erred in applying § 2K2.1(b)(6)(B)’s enhancement. See id. at *2. He insisted “his drug possession offense could not have been elevated from a misdemeanor to a felony under 21 U.S.C. § 844 because the categorical approach plainly applies to the statute.” Id. Like Mr. Arnett does here, Mr. Simmons “lean[ed] on cases in which the categorical approach was applied to allegedly analogous statutes.” Id. at *4 & n.5 (citing United States v. Cantu, 964 F.3d 924, 926 (10th Cir. 2020) and Herrera-Roldan, 414 F.3d at 1241).

The Simmons panel disagreed “that the categorical approach plainly applies to” § 844. Id. at *2. “[E]ven if there was an error,” the panel explained, “it does not surpass the ‘plain’ prong because it is not clearly established that the categorical approach applies to § 844.” Id. at *4. The panel reasoned Mr. Simmons’ analogous cases “did not address the federal statute” at issue and pointed to Shaw’s holding. Id. The panel then went a step further and reiterated Shaw’s holding:

We clarified in Shaw that “§ 844(a) does not require the prior offense to be a violation of [the Federal Controlled Substance] Act. It simply refers to ‘any drug, narcotic, or chemical offense chargeable under the law of any State.’ ” And tellingly, we found that “[a]lthough the categorical approach applies to serious drug offenses under the ACCA, it does not pertain to § 844(a).”

Id. (alterations in original) (internal citation omitted). Under the circumstances, the panel determined Mr. Simmons’ arguments “analogizing § 844 to seemingly parallel statutes” was “unconvincing,” and “even if the district court erred, it was not plain error given a persuasive circuit case on the matter.” Id.

Mr. Arnett tries to sidestep Shaw and Simmons. As to Shaw, he maintains that decision is “wrong” and says the panel's analysis was “conclusory” because it “did not discuss—or even cite to—the definition of ‘drug, narcotic, or chemical offense’ contained in § 844(c).” Op. Br. at 30. Mr. Arnett emphasizes that his plain error argument “hinges on the definition of ‘drug, narcotic, or chemical offense’ ․ set forth in § 844(c)”—an issue the Shaw panel “did not consider[.]”¹⁵ Reply Br. at 10. Because “Shaw does not mention § 844(c) or deal with any of the arguments now before the Court[,]” Mr. Arnett says that case “cannot undermine [his] showing that the error is plain.” Reply Br. at 10. He seeks distance from Simmons on similar grounds. In Mr. Arnett's view, “Simmons does not mention, much less articulate a basis for rejecting, [his] primary argument for the conclusion that 21 U.S.C. § 844 plainly requires the categorical approach.” Dkt. No. 68 at 1. He observes “Simmons does not cite § 844(c)” or discuss the significance of the statute's use of the word “proscribes.” Dkt. No. 68 at 1–2.

We are ultimately unpersuaded we can reverse on plain error review. Though Shaw and Simmons are unpublished and therefore not binding circuit precedent, that does not mean we can disregard them—especially when evaluating plainness. “[E]ven an unpublished opinion can show that the law is not clearly to the contrary.” Romero, 132 F.4th at 1218; see also, e.g., Story, 635 F.3d at 1249 (concluding the district court's alleged sentencing error was not plain in part because “two (albeit non-precedential unpublished) cases to directly address the issue in our circuit interpreted the statutes to allow the sentence imposed”). Notwithstanding the text of § 844(c), two unpublished decisions saying the categorical approach does not apply to that statute at least renders the law on this issue “unsettled.” United States v. Whitney, 229 F.3d 1296, 1310 (10th Cir. 2000) (concluding defendant could not “establish plain error” because the “law surrounding” the issue was “unsettled”); see also, e.g., United States v. Randall, 661 F.3d 1291, 1299 (10th Cir. 2011) (holding instructional error could not be plain “[i]n light of the unsettled state of the law on th[e] issue in our circuit”).

We do not see how—given both Shaw and Simmons—the categorical approach's applicability to § 844(c) can be beyond “any reasonable dispute.” Berryhill, 140 F.4th at 1302 n.3 (emphasis omitted) (internal quotation marks omitted). True enough, Shaw did not grapple with the nuances of the categorical approach. See 2023 WL 3335596, at *4. Nor did the panel directly reject Mr. Arnett's textual argument as to § 844(c). See id. But Simmons did. The panel found “unconvincing” Mr. Simmons’ reliance on cases—including Cantu and Herrera-Roldan—involving “seemingly parallel statutes” because they “did not address” § 844. Simmons, 2025 WL 2374700, at *4. He relies on the same set of cases to analogize the “that prohibits” enhancement language to § 844(c)’s “proscribes” language. At base, he makes the same core arguments as Mr. Simmons, except now there are two persuasive circuit cases on point instead of just one. Because we cannot say the categorical approach plainly applies to § 844(c), Mr. Arnett has not carried his burden on plain error review.¹⁶

Accordingly, the district court did not plainly err in treating Mr. Arnett's simple possession of marijuana as a felony for purposes of § 2K2.1(b)(6)(B)’s four-point enhancement.

IV

We next consider Mr. Arnett's alternative challenge to the § 2K2.1(b)(6)(B) enhancement—that the district court “applied the wrong legal standard in ruling that [he] possessed his girlfriend's marijuana.” Op. Br. at 37 (heading format omitted). According to Mr. Arnett, the district court “held that [he] had constructive possession of his girlfriend's marijuana, and therefore committed ‘another felony,’ simply because he had knowledge that the drugs were next to him in the car.”¹⁷ Op. Br. at 38. He contends this was plain error because “constructive possession is not established by knowledge and access alone but additionally requires intent to actually exercise control—something the [district] court never found.” Op. Br. at 38. We are not persuaded. Even assuming the district court erred, and plainly so, Mr. Arnett has not carried his burden on the third prong of the plain error standard.

A

We first ask whether the district court “committed error.” Vazquez-Garcia, 130 F.4th at 897. To justify application of § 2K2.1(b)(6)(B)’s four-point enhancement, the government was required to prove by a preponderance of the evidence that Mr. Arnett possessed the marijuana. See United States v. Alqahtani, 73 F.4th 835, 850 (10th Cir. 2023) (holding “the other [felony] offense need only be proven by a preponderance of the evidence” to apply the four-point enhancement under § 2K2.1(b)(6)(B)).

“Possession may be actual or constructive.” United States v. Stepp, 89 F.4th 826, 832 (10th Cir. 2023). “Actual possession exists when a person has direct physical control over a thing. Constructive possession is established when a person, though lacking such physical custody, still has the power and intent to exercise control over the object.” Henderson v. United States, 575 U.S. 622, 626, 135 S.Ct. 1780, 191 L.Ed.2d 874 (2015) (internal citation omitted). Thus, to prove constructive possession, “the government must demonstrate the defendant knew of, had access to, and intended to exercise dominion or control over the contraband.” United States v. Johnson, 46 F.4th 1183, 1187 (10th Cir. 2022).

When, as here, contraband “can be attributed to more than one person, there must be a nexus between the defendant and the [contraband] supported by ‘reasonable inferences from direct or circumstantial evidence[.]’ ” United States v. Brown, 85 F.4th 1291, 1296 (10th Cir. 2023) (second alteration in original) (quoting United States v. Reece, 86 F.3d 994, 996 (10th Cir. 1996)); see also United States v. Samora, 954 F.3d 1286, 1290 (10th Cir. 2020) (applying the nexus requirement when the defendant “borrowed the car from his ex-girlfriend” even though he was “the sole occupant of the vehicle”). A defendant's “[p]roximity to the [contraband] alone is not enough” to establish the knowledge of and intent required for constructive possession. Brown, 85 F.4th at 1296.

Mr. Arnett argues “a driver's knowledge that contraband is in the car with him is not enough to deem him in constructive possession” because he “must also have an intent to actually exercise control over the contraband.” Op. Br. at 39. He claims the district court “did not comply with this rule” and instead “erroneously deemed him in constructive possession of his girlfriend's marijuana simply because he had access to it and knew it was there, without requiring proof that [he] had an intent to actually exercise control over the drugs.” Op. Br. at 39.

In Mr. Arnett's view, there are “six different ways” in which “the record shows that the district court applied the wrong standard.”¹⁸ Op. Br. at 41. He argues (1) the district court “repeatedly” said it was “deeming Mr. Arnett in possession of [the marijuana] based on nothing more than his proximity to it and knowledge of its presence”; (2) the government's sentencing memorandum “urg[ed] the district court to apply the defunct knowledge-and-access standard” by “cit[ing] and quot[ing] the ․ standard from United States v. Lopez, 372 F.3d 1207, 1211 (10th Cir. 2004),” a pre-Henderson case; (3) the district court never discussed intent when addressing the possession issue; (4) the district court found “it irrelevant that Mr. Arnett did not own the marijuana”; (5) “the district court deemed Mr. Arnett in possession of the marijuana by process of elimination”; and (6) the district court “treated a finding that Mr. Arnett possessed the marijuana as the only outcome that was even plausible[.]” Op. Br. at 41–46 (emphasis omitted) (citations omitted).

The government disagrees. In its view, “[t]he [trial] record is equally clear that the district court knew” the correct constructive possession standard “because it instructed the jury to apply that definition of constructive possession.” Ans. Br. at 33. “Mr. Arnett must demonstrate based on the record that the district court at sentencing plainly rejected” that definition, says the government. Ans. Br. at 33. It also characterizes the district court's “language” as “consistent with what this [c]ourt has said in addressing challenges to the sufficiency of the evidence of constructive possession.” Ans. Br. at 34–35. The government urges us to disregard its “regrettable” citation to Lopez in its sentencing memorandum; claims “the district court had no need to address” intent because Mr. Arnett did not; reiterates “[o]wnership is not necessary for possession”; denies the district court found possession by process of elimination; and accuses Mr. Arnett of “mischaracteriz[ing] the district court's view of the evidence.” See Ans. Br. at 36–42 (header format omitted).

The district court correctly framed the dispositive issue as “whether there[ ] [was] a nexus between [Mr. Arnett] and the marijuana.” RIII.108; see Samora, 954 F.3d at 1290. Its analysis of the evidence, however, appears improperly focused on Mr. Arnett's proximity to and knowledge or “awareness” of the marijuana. See RIII.104, 108, 111. The district court never mentioned intent. See United States v. Little, 829 F.3d 1177, 1182 (10th Cir. 2016) (“[C]onstructive possession requires intent to exercise control[.]”). We cannot tell from the record whether its possession finding encompassed the requisite antecedent determination that Mr. Arnett intended to exercise dominion and control over the marijuana. Accordingly, we assume without deciding that Mr. Arnett has established the error prong. That is, we assume the district court applied the wrong constructive possession standard by failing to consider whether Mr. Arnett intended to exercise control over the marijuana.

B

We must next determine whether the error was plain. Mr. Arnett contends “the legal principle on which [he] relies—that constructive possession requires an intent to actually exercise control over the item—is now settled by binding Supreme Court and Tenth Circuit precedent.” Op. Br. at 47. The government responds that “[t]he question of whether any error is plain is moot.” Ans. Br. at 42 (heading format omitted). “In light of the proper definition of constructive possession in the jury instructions and the fact that the district court never rejected that definition at sentencing,” says the government, “Mr. Arnett has failed to show any clear or obvious error.”¹⁹ Ans. Br. at 43.

We readily conclude any error was plain. Constructive possession requires “the power and intent to exercise control over the object.” Henderson, 575 U.S. at 626, 135 S.Ct. 1780; accord Little, 829 F.3d at 1182. If the district court did not find Mr. Arnett intended to exercise control over the marijuana, the failure to do so was contrary to well-settled law. See Whitney, 229 F.3d at 1309 (“An error is clear and obvious when it is contrary to well-settled law.”).

C

We now proceed to the third prong of plain error review. This is where Mr. Arnett encounters difficulties. “To satisfy the third prong of plain-error review, a defendant generally must demonstrate that an error was ‘prejudicial, meaning that there is a reasonable probability that, but for the error claimed, the result of the proceeding would have been different.’ ” United States v. Bustamante-Conchas, 850 F.3d 1130, 1138 (10th Cir. 2017) (en banc) (quoting United States v. Algarate-Valencia, 550 F.3d 1238, 1242 (10th Cir. 2008)). The reasonable probability standard “is a lesser standard than proof by a preponderance of the evidence, and it is satisfied by ‘a probability sufficient to undermine confidence in the outcome.’ ” United States v. Perez-Perez, 992 F.3d 970, 974 (10th Cir. 2021) (quoting Bustamante-Conchas, 850 F.3d at 1138); see also United States v. Dominguez Benitez, 542 U.S. 74, 83 n.9, 124 S.Ct. 2333, 159 L.Ed.2d 157 (2004) (“The reasonable-probability standard is not the same as, and should not be confused with, a requirement that a defendant prove by a preponderance of the evidence that but for error things would have been different.”). Here, Mr. Arnett must show a reasonable probability that “he would have received a lesser sentence” but for the district court's alleged error. United States v. Trujillo-Terrazas, 405 F.3d 814, 819 (10th Cir. 2005).

Mr. Arnett argues “there is at least a reasonable probability that, had the district court applied the correct standard, it would not have found that [he] constructively possessed the marijuana—and, thus, would not have increased his Guidelines range under § 2K2.1(b)(6).” Op. Br. at 48. He insists “a factfinder could easily conclude that [he] did not intend to control the medical marijuana that his girlfriend had purchased and left in her car.” Op. Br. at 48. Mr. Arnett says this “reasonable probability that the district court would have ruled differently under the correct standard ․ triggers the presumption that a Guidelines error affects a defendant's substantial rights.” Op. Br. at 48 (citation omitted); see United States v. Sjodin, 139 F.4th 1188, 1205 (10th Cir. 2025) (“[W]hen a district court sentences a defendant under an incorrect Guidelines range, we presume prejudice.”). And, he points out, “[t]he district court's conclusory assertion that the enhancement in question didn't matter does not rebut th[at] presumption[.]” Op. Br. at 49.

The government counters that Mr. Arnett cannot show “his sentencing would have turned out differently if the correct standard was applied.” Ans. Br. at 43. It insists “there was plenty of evidence” for the district court to “arrive at this conclusion.” Ans. Br. at 43. We interpret the government's argument as suggesting that, even under the correct standard, the district court would have found Mr. Arnett intended to exercise dominion and control over the marijuana given the evidence presented at trial. We agree.

“[P]roof of knowledge and intent must be based on drawing inferences from the defendant's actions, the testimony and actions of others, and documentary evidence.” United States v. Rufai, 732 F.3d 1175, 1191–92 (10th Cir. 2013). And we are not precluded from reviewing the evidence to determine whether the district court's application of the correct constructive possession standard to the evidence might have yielded a different result. See, e.g., United States v. Caraway, 534 F.3d 1290, 1299 (10th Cir. 2008) (reviewing trial testimony to determine whether there was a reasonable probability the defendant would not have been found guilty had the erroneously admitted statement not been presented to the jury; concluding that, given the “strong case” against the defendant, it was not “reasonably probable” the admitted statement “was a critical factor for the jury”).

Recall, the district court identified five pieces of evidence in support of its conclusion that Mr. Arnett possessed the marijuana: (1) Mr. Arnett was the “sole occupant” of the Sonata; (2) the “significant quantity of marijuana”; (3) the marijuana was “pretty open and obvious in a bag”; (4) the marijuana, which was in the front passenger seat, was “in very close proximity to [Mr. Arnett] and the firearm”; and (5) Mr. Arnett told Officer Glover “immediately after being pulled over that there was marijuana in the car[.]” RIII.104, 108, 111. Given this record, Mr. Arnett's intent to exercise dominion and control over the marijuana is supported by at least a preponderance of the evidence, so we are confident the district court still would have found constructive possession under the correct legal standard. See, e.g., Brown, 85 F.4th at 1296 (holding “the district court could certainly find from the totality of the evidence that [the defendant] knew of and intended to possess” the handgun because he was the driver of the car, the handgun was nearby on the floor of the passenger's seat, the handgun was in plain view, a loaded magazine was in the driver's side compartment, and the ammunition in the defendant's pocket was compatible with the handgun); see also United States v. Judkins, No. 24-1088, 2025 WL 1693686, at *3 (10th Cir. June 17, 2025) (unpublished) (concluding defendant's physical possession of the car and power to exercise “control over the items in it,” coupled with the accessibility of the handgun on the car's floor, provided sufficient evidence for the jury to conclude the defendant constructively possessed the firearm).²⁰

We do not mean to suggest the third prong of plain error review is simply a sufficiency test. But the inquiry that resolves the third prong will often depend on the particular nature of the error alleged. Here, Mr. Arnett bears the burden of establishing a reasonable probability of a lesser sentence absent the district court's purported error: application of an incorrect legal standard. To establish a reasonable probability of a lesser sentence, then, Mr. Arnett must show a reasonable probability that the district court would have found intent lacking—and thus would not have increased his Guidelines range under § 2K2.1(b)(6)(B)—had it applied the correct legal standard. We must examine the evidence to determine whether such a probability exists. On the record before us, we cannot say Mr. Arnett has made this showing.²¹

In sum, even if the district court applied the wrong standard to determine Mr. Arnett constructively possessed the marijuana, it did not plainly err in imposing § 2K2.1(b)(6)(B)’s four-point enhancement.

V

We AFFIRM Mr. Arnett's 77-month sentence for unlawful possession of a firearm.²²

FOOTNOTES

2.   The district court initially appointed counsel for purposes of sentencing at Mr. Arnett's request. Later, Mr. Arnett again decided to represent himself with the assistance of standby counsel. Mr. Arnett requested and received full assistance of counsel midway through sentencing proceedings.

3.   “Whenever a prior conviction is relevant to sentencing, the government must establish the fact of that conviction by a preponderance of the evidence.” United States v. Cooper, 375 F.3d 1041, 1052 (10th Cir. 2004). Mr. Arnett does not dispute the existence or validity of his state drug convictions identified in the PSR. We therefore treat those predicate facts as established. See United States v. Hooks, 551 F.3d 1205, 1217 (10th Cir. 2009) (“If a defendant fails to specifically object to a fact in the PSR, the fact is deemed admitted by the defendant and the government need not produce additional evidence in support of the admitted fact.”).

4.   The district court also resolved other PSR objections not relevant to this appeal.

5.   Had the district court sustained Mr. Arnett's objection to § 2K2.1(b)(6)(B)’s four-point enhancement, his Guidelines range would have been 51 to 63 months.

6.   Effective November 1, 2025, the United States Sentencing Commission recodified § 2K2.1(b)(6)(B) as § 2K2.1(b)(7)(B). See U.S. Sent'g Guidelines Manual § 2K1.1(b)(7)(B) (U.S. Sent'g Comm'n 2025). It also renumbered the attendant application notes. The November 2025 amendments are not substantive with respect to the provisions relevant here. We nonetheless cite and refer to the 2023 Guidelines because they were in effect at the time of Mr. Arnett's sentencing in October 2024 and were used by the United States Probation Officer to calculate Mr. Arnett's offense level. See United States v. Sullivan, 255 F.3d 1256, 1259 (10th Cir. 2001) (noting “a sentencing court must apply the guidelines in effect at the time of sentencing ․ unless such application would violate the ex post facto clause”); U.S.S.G. § 1B1.11(a) (“The court shall use the Guidelines Manual in effect on the date that the defendant is sentenced.”).

7.   The district court never mentioned § 844. The PSR does not reference it either. Only the PSR addendum cites that statute, but even then, it merely acknowledges Mr. Arnett's “conc[ession] that under federal law simple possession of drugs can be a felony (21 U.S.C. § 844)[.]” RII.47. Mr. Arnett does not take issue with this omission on appeal. The parties proceed under the mutual assumption that the district court must have relied on § 844 to elevate Mr. Arnett's simple possession of marijuana to a felony. We take the same approach.

8.   Because “Mr. Arnett has no prior federal drug conviction[,]” he reasons that “the district court's application of § 2K2.1(b)(6) embodies an implicit conclusion that at least one of [his] several Oklahoma convictions for possessing a ‘controlled dangerous substance’ qualifie[s] as a conviction for a ‘drug, narcotic, or chemical offense.’ ” Op. Br. at 13. We agree. The government never argued Mr. Arnett has a prior § 844 conviction. And the PSR does not identify one, much less specify which of Mr. Arnett's four Oklahoma drug convictions qualify as “drug, narcotic, or chemical offense[s]” under § 844. It is therefore unclear whether the district court believed one, some, or all four of those convictions satisfy § 844’s definition of a “drug, narcotic, or chemical offense.” Mr. Arnett does not focus on this in his briefing. On appeal, and as we will discuss, the government argues only that Mr. Arnett's convictions for possession with intent to distribute under Okla. Stat. Ann. tit. 63, § 2-401 qualify as “drug, narcotic, or chemical offense[s].”

9.   The government never uses the term “factual approach” in its brief. But its arguments necessarily implicate the factual approach. For example, the government claims § 844(c) requires us to “determine what actually happened and whether a federally banned substance was actually involved in the underlying offense.” Ans. Br. at 16. This is the hallmark of the factual approach.

10.   The government essentially conceded as much at oral argument. When asked whether it agreed Mr. Arnett lacks a qualifying predicate conviction if the categorical approach applies to § 844(c), the government indicated it “did not challenge that” point. See Oral Arg. at 32:57–3:20.

11.   Although § 844(c)’s text “unavoidably demands the categorical approach,” Mr. Arnett says it is “also worth noting that interpreting § 844 to require a factual approach would make it either unconstitutional or unworkable.” Op. Br. at 17. This is because the “statutory maximum” under § 844 “turns on whether the defendant has a prior conviction for a ‘drug, narcotic, or chemical offense,’ ” and “only a jury, and not a judge, may find facts that increase a maximum penalty, except for the simple fact of a prior conviction.” Op. Br. at 17 (emphasis omitted) (quoting Mathis, 579 U.S. at 511, 136 S.Ct. 2243). But at the same time, Mr. Arnett suggests, “interpreting § 844 to require a jury to determine the facts underlying the defendant's prior conviction(s) would entail mini trials on events that ․ may be decades old.” Op. Br. at 18. Given our conclusion that the categorical approach applies to § 844(c), we need not address this argument.

12.   The “circumstance-specific” approach is another way to describe the factual approach—i.e., they are one and the same. See, e.g., United States v. Cooper, 159 F.4th 768, 775 (10th Cir. 2025) (“[I]f the court applies a circumstance-specific approach, then it is free to look at the specific facts and conduct that led to the conviction in order to determine whether the conviction qualifies.”); United States v. Gieswein, 887 F.3d 1054, 1059 (10th Cir. 2018) (“When a statute or Guideline ‘was intended to refer to the specific acts in which a defendant has engaged on a prior occasion, we use a circumstance-specific approach’ under which we ‘may look beyond the elements of the prior offense and consider the facts and circumstances underlying an offender's conviction.’ ” (quoting United States v. White, 782 F.3d 1118, 1131 (10th Cir. 2015))).

13.   In fact, “a growing number of judges across the country” have criticized the categorical approach as, among other things, a “waste[ ] of judicial resources” that “undermines confidence in the administration of justice.” United States v. Scott, 990 F.3d 94, 126 (2d Cir. 2021) (Park, J., concurring) (collecting cases).

14.   Simmons was decided before oral argument. The parties addressed that case in letters of supplemental authority under Federal Rule of Appellate Procedure 28(j).

15.   Mr. Arnett suggests the panel in Shaw may not have had fully developed briefing on the issue because Mr. Shaw appeared pro se.

16.   Mr. Arnett makes two other plainness arguments, but we need not address them. First, he claims “it is clear or obvious that each of [his] prior [Oklahoma] convictions is overbroad.” Op. Br. at 31. He argues the Oklahoma statutes of conviction “make it undeniable that Oklahoma prohibited substances that were not on the federal schedules of controlled substances at the time of [his] prior offenses.” Op. Br. at 31. This argument speaks to our analysis under the categorical approach. For the reasons we have explained, however, it is not “clear or obvious” the categorical approach applies to § 844(c) in the first place, so we do not consider whether Mr. Arnett's prior Oklahoma drug convictions “sweep[ ] in conduct” that § 844(c) does not. See United States v. Griffin, 164 F.4th 782, 785 (10th Cir. 2026) (holding that under the categorical approach, “if the state statute sweeps in conduct that the ․ federal offense does not, the state conviction does not qualify” under the federal statute). Second, Mr. Arnett maintains “it is clear or obvious that none of the [Oklahoma] statutes underlying [his] convictions is divisible down to the individual substance involved.” Op. Br. at 31. This argument likewise sprints past our threshold inquiry: whether the categorical approach applies to § 844(c). Only if the categorical approach plainly applied to § 844(c) would we then examine whether Mr. Arnett's Oklahoma statutes of conviction are divisible by the individual drug.

17.   The district court never specified whether it found actual or constructive possession, but the parties proceed on the assumption that it found constructive possession. We agree that, based on its discussion of proximity and knowledge, the district court determined Mr. Arnett constructively possessed the marijuana. This conclusion is bolstered by the district court's statement to Mr. Arnett that he “wouldn't have had to ․ touch[ ] [the marijuana] to possess it.” RIII.109.

18.   Mr. Arnett refers to the “wrong” standard purportedly applied by the district court here as “the pre-Henderson standard.” Op. Br. at 41. As he explains, “[f]or generations, the law in [our] circuit was that mere knowledge of a nearby item's presence was a sufficient mens rea to put the defendant in constructive possession of it.” Op. Br. at 39; see, e.g., United States v. Ledford, 443 F.3d 702, 714 (10th Cir. 2005) (“[T]he government need only prove knowledge and access to the weapon, not the intention to exercise control[.]”), abrogated by Henderson, 575 U.S. at 626, 135 S.Ct. 1780. Mr. Arnett correctly observes this “changed with Henderson,” when the Supreme Court “held that constructive possession requires a person to ‘ha[ve] the power and intent to exercise control over the object.’ ” Op. Br. at 40 (alteration in original) (quoting Henderson, 575 U.S. at 626, 135 S.Ct. 1780). We subsequently recognized “that Henderson change[d] the law of constructive possession in our circuit[.]” United States v. Little, 829 F.3d 1177, 1182 (10th Cir. 2016).

19.   The government's argument is difficult to follow. It does not explain how mootness is implicated, and the government's reassertion of its jury instruction argument appears to conflate the error and plainness prongs.

20.   We can rely on unpublished decisions for their persuasive value. United States v. Engles, 779 F.3d 1161, 1162 n.1 (10th Cir. 2015) (citing 10th Cir. R. 32.1).

21.   According to the government, affirmance is required because the district court would have imposed the same sentence even if it had not applied the four-point enhancement. The district court, says the government, “evaluated Mr. Arnett's entire record ․ and concluded that the guideline range represented the appropriate range of punishment[.]” Ans. Br. at 23, 43–44. The government further relies on the district court's statement that it would have imposed a 77-month sentence even if the Guidelines range was 51 to 63 months—i.e., the Guidelines range the district court would have calculated absent § 2K2.1(b)(6)(B)’s four-point enhancement.This is not the appropriate way to analyze the third prong of the plain-error standard. A district court's Guidelines calculation forms the “framework for sentencing” and “anchor[s] ․ the district court's discretion[.]” Peugh v. United States, 569 U.S. 530, 542, 549, 133 S.Ct. 2072, 186 L.Ed.2d 84 (2013). Thus, “[w]hen a defendant is sentenced under an incorrect Guidelines range—whether or not the defendant's ultimate sentence falls within the correct range—the error itself can, and most often will, be sufficient to show a reasonable probability of a different outcome absent the error.” Molina-Martinez v. United States, 578 U.S. 189, 198, 136 S.Ct. 1338, 194 L.Ed.2d 444 (2016); see also United States v. Sabillon-Umana, 772 F.3d 1328, 1333 (10th Cir. 2014) (“When the court's starting point is skewed a ‘reasonable probability’ exists that its final sentence is skewed too.”). Nor are we persuaded by the district court's statement that it would have imposed a 77-month sentence even if Mr. Arnett's Guidelines range was 51 to 63 months. As shown by the many cases Mr. Arnett cites in his opening brief, we have repeatedly “rejected the notion that district courts can insulate sentencing decisions from review by making such statements.” United States v. Burris, 29 F.4th 1232, 1238 (10th Cir. 2022) (internal quotation marks omitted).

22.   We deny Mr. Arnett's Opposed Motion for Leave to File Pro Se Supplemental Brief and Supplement to Opposed Motion for Leave to File Pro Se Supplemental Brief because he is represented by counsel. See United States v. Pearl, 324 F.3d 1210, 1216 (10th Cir. 2003) (“As [the appellant] is represented by counsel, we deny his motion to file an additional pro se supplemental brief[.]”); United States v. Coleman, 9 F.3d 1480, 1487 (10th Cir. 1993) (declining to address defendant's arguments raised in his pro se supplemental brief because “[i]t is our policy on direct appeals ․ only to address issues raised by counsel”).

ROSSMAN, Circuit Judge.