United States v. Cantu

FILED United States Court of Appeals PUBLISH Tenth Circuit UNITED STATES COURT OF APPEALS July 6, 2020 Christopher M. Wolpert FOR THE TENTH CIRCUIT Clerk of Court _________________________________ UNITED STATES OF AMERICA, Plaintiff - Appellee, v. No. 19-6043 FRANCISCO CANTU, JR., Defendant - Appellant. _________________________________ Appeal from the United States District Court for the Western District of Oklahoma (D.C. No. 5:18-CR-00059-HE-1) _________________________________ Jacob Rasch-Chabot, Assistant Federal Public Defender, Denver, Colorado (Virginia L. Grady, Federal Public Defender, and Shira Kieval, Assistant Federal Public Defender, Denver, Colorado, on the briefs) for the Defendant-Appellant. Steven W. Creager, Assistant United States Attorney (Timothy J. Downing, United States Attorney, and Mark R. Stoneman, Assistant United States Attorney, with him on the brief), Oklahoma City, Oklahoma, for the Plaintiff-Appellee. _________________________________ Before HARTZ, MURPHY, and MATHESON, Circuit Judges. _________________________________ HARTZ, Circuit Judge. _________________________________ Defendant Francisco Cantu, Jr. appeals the enhancement of his sentence under the Armed Career Criminal Act (ACCA), 18 U.S.C. § 924(e)(1). Although he failed to preserve his challenge to the enhancement in district court, the enhancement was plainly contrary to the law of this circuit. Exercising jurisdiction under 18 U.S.C. § 3742(a) and 28 U.S.C. § 1291, we vacate his sentence and remand for resentencing. The ACCA enhancement rested in part on the characterization of Defendant’s two prior convictions for drug offenses under Okla. Stat. tit. 63, § 2–401(A)(1) as “serious drug offenses.” But there are multiple means by which the Oklahoma statute can be violated, and some of those means do not satisfy the ACCA definition of serious drug offense. Under the categorical/modified-categorical approach established by the United States Supreme Court for determining whether a state conviction can qualify as an ACCA predicate conviction, the two state convictions therefore cannot be predicate convictions supporting an ACCA enhancement. After describing the proceedings against Defendant, we summarize the relevant law under the ACCA, apply that law to Defendant’s prior state drug convictions, and then consider whether relief is proper even though Defendant did not preserve the issue in district court. I. BACKGROUND In 2019 Defendant pleaded guilty to being a felon in possession of a firearm. See 18 U.S.C. § 922(g). The presentence report (PSR) said that he was subject to an enhancement under the ACCA based on three prior Oklahoma convictions: (1) unlawful possession of a controlled drug with intent to distribute on August 21, 2008, (2) distribution of methamphetamine on December 13, 2010, and (3) distribution of methamphetamine on December 27, 2010. Defendant was convicted of the latter two state offenses on March 6, 2012, after pleas of nolo contendere. During sentencing for 2 his federal offense he did not object to his PSR, nor did he file a sentencing memorandum. The district court adopted the PSR and sentenced him to a prison term of 210 months, which was the bottom of the advisory sentencing guidelines range. Without the ACCA enhancement his maximum prison term would have been 120 months. See 18 U.S.C. § 924(a)(2). II. THE ACCA The ACCA increases the penalty for being a felon in possession of a firearm for any person who has “three previous convictions . . . for a violent felony or a serious drug offense.” 18 U.S.C. § 924(e)(1). Our concern in this appeal relates only to serious drug offenses. The statutory definition of serious drug offense includes “an offense under State law, involving manufacturing, distributing, or possessing with intent to manufacture or distribute, a controlled substance (as defined in section 102 of the Controlled Substances Act (21 U.S.C. § 802)), for which a maximum term of imprisonment of ten years or more is prescribed by law.” 18 U.S.C. § 924(e)(2)(A)(ii). The incorporated definition of controlled substance is “a drug or other substance, or immediate precursor, included in schedule I, II, III, IV, or V of part B of this subchapter [21 U.S.C §§ 811– 14].” 21 U.S.C. § 802(6). To determine whether a state conviction was for a serious drug offense, we generally begin with the categorical approach. See United States v. Smith, 652 F.3d 1244, 1246 (10th Cir. 2011) (categorical approach applies to both violent felonies and serious drug offenses). There are two key features of this approach. First, the court looks only to the elements of the state offense. The particular facts of the defendant’s prior 3 offense are irrelevant. All that counts is what the defendant had to do to be guilty of the offense. See Mathis v. United States, 136 S. Ct. 2243, 2248 (2016). Second, it is not enough that there is an overlap between the elements of the state offense and the definition of serious drug offense. It is necessary that essentially any conduct that satisfies the elements of the state offense also satisfy the definition of serious drug offense. If one can commit the state offense by conduct that is not a serious drug offense, then conviction of the state offense cannot be a predicate offense for the ACCA. See id. The Supreme Court recently illustrated these two points in Mellouli v. Lynch, 135 S. Ct. 1980, 1983–84 (2015), where it applied the categorical approach to a provision in the Immigration and Nationality Act authorizing the removal of an alien convicted of violating a state law “relating to a controlled substance (as defined in section 802 of Title 21),” 8 U.S.C § 1227(a)(2)(B)(i). Mr. Mellouli had been convicted in Kansas of violating a state law prohibiting possession of drug paraphernalia to store or conceal a substance that was a controlled substance as defined by state law. See Mellouli, 135 S. Ct. at 1983. The controlled substance involved in the offense was Adderall, which is also a controlled substance under federal law. See id. at 1985. What mattered for purposes of the categorical approach, however, was not the specific substance involved in Mellouli’s offense but (1) the fact that the Kansas statute could be violated with any controlled substance under Kansas law and (2) the Kansas statute of conviction “was not confined to federally controlled substances,” because “Kansas’ schedules of controlled substances included at least nine substances—e.g., salvia and jimson weed—not defined in § 802.” Id. at 1988. The Court rejected the government’s argument that state drug offenses can 4 qualify as serious drug offenses if the state schedules have a “substantial overlap” with the federal schedules. Id. at 1990 (internal quotation marks omitted). “In short, the state law under which [Mr. Mellouli] was charged categorically related to a controlled substance, but was not limited to substances defined in § 802.” Id. at 1988 (original brackets and internal quotation marks omitted). Thus, “[u]nder the categorical approach . . . [Mr.] Mellouli’s drug-paraphernalia conviction d[id] not render him deportable.” Id. Because the government did not argue otherwise, see id. at 1986 n.4, the Supreme Court in Mellouli treated the Kansas statute as stating a single criminal offense, which could be violated in a number of alternative ways, depending on what controlled substance was involved. But sometimes the alternatives (say, each controlled substance) in a statute define distinct criminal offenses. We then say that the statute is divisible and apply the modified-categorical approach, in which the categorical approach is applied separately to the relevant sub-crime within the statute. See Mathis, 136 S. Ct. at 2249. A court can ascertain which sub-crime the defendant was convicted of by examining a limited category of court records, such as the charging document, jury instructions, a plea agreement, or a plea colloquy. See id. To determine whether a statute is divisible, it is essential to distinguish between elements and means. “Elements are the constituent parts of a crime’s legal definition— the things the prosecution must prove to sustain a conviction.” Id. at 2248 (internal quotation marks omitted). They are what the jury must unanimously find beyond a reasonable doubt to render a guilty verdict, and what a defendant must admit when pleading guilty. See id. Means, in contrast, “spell[] out various factual ways of 5 committing some component of the offense—a jury need not find (or a defendant admit) any particular item.” Id. at 2249. For example, a statute may have as an element the use of a “deadly weapon” and “further provide[] that the use of a ‘knife, gun, bat, or similar weapon’ would all qualify [as means].” Id. at 2249 (further internal quotation marks omitted). A jury could convict even if half believed the defendant used a knife and the other half thought he used a gun, so long as there was unanimity on the relevant element—namely, that he used a “deadly weapon.” See id.; United States v. Degeare, 884 F.3d 1241, 1251–52 (10th Cir. 2018) (noting that “Mathis makes jury unanimity the touchstone of the means-or-elements inquiry” and that “we have likewise adopted a unanimity-focused approach to the means-or-elements question”). With this background we now examine Defendant’s state drug convictions to see whether they qualify as serious drug offenses. III. APPLICATION TO DEFENDANT’S CONVICTIONS A. Divisibility of § 2–401(A)(1) Defendant’s 2012 convictions were for violations of an Oklahoma statute that made it unlawful “[t]o distribute, dispense, transport with intent to distribute or dispense, possess with intent to manufacture, distribute, or dispense, a controlled dangerous substance or to solicit the use of or use the services of a person less than eighteen (18) years of age to cultivate, distribute or dispense a controlled dangerous substance.” Okla. Stat. Ann. tit. 63, § 2–401(A)(1) (West 2011). Controlled dangerous substance is defined by Oklahoma as “a drug, substance or immediate precursor in Schedules I through V of the [Oklahoma] Uniform Controlled Dangerous Substances Act.” Id. § 2– 6 101(8). It is undisputed that in December 2010, when Defendant committed the two state offenses at issue here, at least three substances (e.g., salvinorin A) that satisfied this definition were not controlled substances under federal law. Compare Okla. Stat. Ann. tit. 63, § 2–204(C) (West Supp. 2010), with 21 C.F.R. §§ 1308.11–15 (2019). Therefore, under the categorical approach Defendant’s convictions under § 2–401(A)(1) were not convictions for serious drug offenses under the ACCA. See Melllouli, 135 S. Ct. at 1988. The government argues, however, that the Oklahoma statute is divisible based on each individual drug listed in Oklahoma’s drug schedules and that Defendant was convicted of offenses involving methamphetamine. Since methamphetamine is a federal controlled substance, see 21 C.F.R. § 1308.12 (2019), Defendant’s state convictions would be for serious drug offenses. The principal issue on appeal thus is whether the state statute is divisible, with a violation for each controlled substance being a distinct offense. To help us answer the question, the Supreme Court has set forth a framework for analysis. First, we should begin by examining “authoritative sources of state law,” including the statute on its face and state-court decisions. Mathis, 136 S. Ct. at 2256. If, for example, “a state court decision definitively answers the question, . . . a sentencing judge need only follow what it says.” Id. Next, “if state law fails to provide clear answers, federal judges have another place to look: the record of a prior conviction itself.” Id. For example, “an indictment and jury instructions could indicate, by referencing one alternative term to the exclusion of all others, that the statute contains a list of elements, each one of which goes toward a separate crime.” Id. at 2257 (“Of course, such record materials will not in every 7 case speak plainly. . . .”). If, however, we cannot ultimately say with certainty that the statute is divisible, we will not apply the modified-categorical approach. See Degeare, 884 F.3d at 1248 (“[U]nless we are certain that a statute’s alternatives are elements rather than means, the statute isn’t divisible and we must eschew the modified categorical approach.”). Federal courts have interpreted statutes similar to Oklahoma’s on a number of occasions and have relied on state-court authority to determine divisibility, although the authorities in different states point in different directions. See Cucalon v. Barr, 958 F.3d 245, 253 (4th Cir. 2020) (collecting cases). Here, too, a state-court decision provides the answer to divisibility. Before discussing the decision, however, we should say a little more about the Oklahoma statutory scheme. Section 2–401(A)(1) did not distinguish among the various substances defined as controlled dangerous substances by Oklahoma law. But the penalties were not the same for every drug offense. Section 2–401(B) introduced the penalty provisions with: “Any person who violates the provisions of this section with respect to” and then set different penalties for three different categories of controlled dangerous substances. See Okla. Stat. Ann. tit. 63, § 2–401(B)(1) (West 2011) (punishing offenses involving Schedule I or II narcotic drugs, LSD, or one of five other substances related to gamma hydroxybutyrate (GHB) with a “term of imprisonment for not less than five (5) years nor more than life”); 1 id. § 2–401(B)(2) (punishing offenses 1 Paragraph (B)(1) stated: 8 involving any other Schedule I, II, III, or IV drug with a “term of imprisonment for not less than two (2) years nor more than life”); 2 id. § 2–401(B)(3) (punishing offenses involving Schedule V drugs with “a term of imprisonment for not more than five (5) years”). 3 [Any person who violates the provisions of § 2–401(A) with respect to] [a] substance classified in Schedule I or II which is a narcotic drug, lysergic acid diethylamide (LSD), gamma butyrolactone, gamma hydroxyvalerate, gamma valerolactone, 1,4 butanediol, or gamma-hydroxybutyric acid as defined in Sections 2–204 and 2–208 of this title, upon conviction, shall be guilty of a felony and shall be sentenced to a term of imprisonment for not less than five (5) years nor more than life and a fine of not more than One Hundred Thousand Dollars ($100,000.00), which shall be in addition to other punishment provided by law and shall not be imposed in lieu of other punishment. Any sentence to the custody of the Department of Corrections shall not be subject to statutory provisions for suspended sentences, deferred sentences, or probation except when the conviction is for a first offense[.] Okla. Stat. Ann. tit. 63, § 2–401(B)(1) (West 2011). 2 Paragraph (B)(2) stated: [Any person who violates the provisions of § 2–401(A) with respect to] [a]ny other controlled dangerous substance classified in Schedule I, II, III, or IV, upon conviction, shall be guilty of a felony and shall be sentenced to a term of imprisonment for not less than two (2) years nor more than life and a fine of not more than Twenty Thousand Dollars ($20,000.00), which shall be in addition to other punishment provided by law and shall not be imposed in lieu of other punishment. Any sentence to the custody of the Department of Corrections shall not be subject to statutory provisions for suspended sentences, deferred sentences, or probation except when the conviction is for a first offense[.] Okla. Stat. Ann. tit. 63, § 2–401(B)(2) (West 2011). 3 Paragraph (B)(3) stated: 9 In our view, Oklahoma case law makes it impossible to say with certainty that the Oklahoma statute is divisible by individual drug. The leading opinion on the point is Watkins v. State, 855 P.2d 141 (Okla. Crim. App. 1992). The defendant had been convicted on one count of possession with intent to distribute cocaine and one count of intent to distribute phencyclidine (PCP) based on his shipping both substances in a single package. See id. at 141. The Oklahoma Court of Criminal Appeals (OCCA) determined that the defendant had committed only one violation of § 2–401 and could not be sentenced separately for two offenses. See id. at 142. It rejected the argument that there were “two separate offenses with different elements requiring different ranges of punishment”; “Possession with Intent to Distribute is a single offense under Section 2- 401. Part A of Section 2-401 sets forth the substantive offense while Part B sets forth the parameters of punishment based on the type of drug involved.” Id. The court acknowledged that “different punishment levels are provided for different drugs,” but said that the State’s “argument does not take into consideration the statutory language of Section 2–401(A)(1) which is the substantive criminal prohibition, i.e., it is unlawful for any person to possess with the intent to distribute ‘a controlled dangerous substance.’” Id. In short, “[t]he statutory prohibition does not distinguish between types or [Any person who violates the provisions of § 2–401(A) with respect to] [a] substance classified in Schedule V, upon conviction, shall be guilty of a felony and shall be sentenced to a term of imprisonment for not more than five (5) years and a fine of not more than One Thousand Dollars ($1,000.00), which shall be in addition to other punishment provided by law and shall not be imposed in lieu of other punishment[.] Okla. Stat. Ann. tit. 63, § 2–401(B)(3) (West 2011). 10 classifications of drugs.” Id. It explained, “While we recognize the Oklahoma Legislature has the power to create separate penal provisions prohibiting different acts which may be committed at the same time, it was not exercised in the passage of the provisions of Section 2–401(A)(1) as this statute applies to a ‘controlled dangerous substance.’” Id. Because the “elements of each count . . . were the same,” double- jeopardy doctrine prohibited the defendant’s multiple punishments for the same offense. Id. at 142. The court’s language at the very least is potent support for the proposition that the alternative ways in which the statutory violation can be committed (by distributing any one of a number of controlled substances) are alternative means, rather than alternative elements. This and other courts have held that decisions like Watkins resolve the divisibility question. See United States v. McKibbon, 878 F.3d 967, 975 (10th Cir. 2017) (Colorado Supreme Court double-jeopardy decision that defendant could not be convicted of both possession and distribution “addresses exactly th[e] question” of whether statute was divisible); United States v. Garcia, 948 F.3d 789, 794 (7th Cir. 2020) (considering an Indiana intermediate-court case that held “possessing marijuana and hashish is only one violation” to be “the authoritative resolution” of the divisibility issue); Martinez v. Sessions, 893 F.3d 1067, 1071 (8th Cir. 2018) (concluding that Missouri double-jeopardy court decisions show that each controlled substance is an element); see also Najera- Rodriguez v. Barr, 926 F.3d 343, 352–353 (7th Cir. 2019) (interpreting Illinois case describing element as a “controlled substance” to “signal[]” that “identity of the controlled substance is not an element”). But see United States v. Burris, 912 F.3d 386, 11 403–04 (6th Cir. 2019) (en banc), cert. denied, 140 S. Ct. 90 (2019) (interpreting Ohio law as treating felonious assault and aggravated assault as distinct crimes even though a separate state statute prohibits punishment under both statutes for a “single act undertaken with a single animus” as a matter of legislative intent because they are “allied offenses of similar import” (internal quotation marks omitted)). The government points out, however, that Watkins recognized that punishment for violations of § 2–401(A)(1) is “based on the type of drug involved,” Watkins, 855 P.2d at 142, and that in light of United States Supreme Court authority postdating Watkins, the statute must therefore be divisible. There is much truth in what the government says, but it does not affect our conclusion. First, the government is correct that the Oklahoma punishment for offenses involving controlled dangerous substances depends on the substance. We have already noted that § 2–401(B) divides substances into three categories with different maximum sentences. Second, it is true that after Watkins was decided the Supreme Court held that “[o]ther than the fact of a prior conviction, any fact that increases the penalty for a crime beyond the prescribed statutory maximum must be submitted to a jury, and proved beyond a reasonable doubt.” Apprendi v. New Jersey, 530 U.S. 466, 490 (2000). In other words, a fact that must be proved to increase the penalty is an element of the offense. See Mathis, 136 S. Ct. at 2256 (“If statutory alternatives carry different punishments, then under Apprendi they must be elements.”). Thus, the category to which an Oklahoma controlled dangerous substance belongs is an element of a violation of § 2–401(A)(1). And third, because different offenses under § 2– 401(A)(1) have different elements, the statute must be divisible. See United States v. 12 Maldonado-Palma, 839 F.3d 1244, 1247 (10th Cir. 2016) (“New Mexico’s aggravated assault statute is a divisible statute because it sets out alternative elements for aggravated assault in three subsections.”). That syllogism, however, does not take the government far enough to prevail in this case. It is not as if the penalty is different for each controlled dangerous substance. Those substances are divided into only three categories for purposes of punishment. Therefore, § 2–401(A)(1) is divisible based on those three categories. But any further division into individual substances appears inconsistent with Watkins. It effectively held that the jury did not need to agree on which controlled dangerous substance was involved in the defendant’s crime. From Apprendi we now know that the jury must agree on which category the substance belongs to. But Apprendi still leaves to Oklahoma to determine whether juries must agree on which substance within a category was involved. Although the precise holding in Watkins is incorrect because cocaine and PCP were in different categories for sentencing purposes, the language in Watkins appears inconsistent with the notion that now, in light of Apprendi, the specific controlled dangerous substance is an element of § 2–401(A)(1). Indeed, just last year the OCCA, agreeing with a concession by the State, held that a defendant could not be convicted of both unlawful possession with intent to distribute cocaine and unlawful possession with intent to distribute methamphetamine (which belong to the same category for sentencing) when both drugs were found in the same receptacle. See Alexander v. State, 449 P.3d 860, 868 (Okla. Crim. App. 2019) (“Based on Watkins, we must vacate Count 2 . . . .”); id. at 871 (Lumpkin, J., concurring in part and dissenting in part) (continuing to apply Watkins 13 since “the Oklahoma Legislature failed to set out that possession of each separate drug constitutes a separate offense and is punishable individually” and the “Legislature has had a substantial amount of time to correct the drafting and by its silence has elected not to do so”). We need add only that methamphetamine, the drug involved in Defendant’s state prosecutions, is in the same category of drugs as three Oklahoma controlled dangerous drugs that have not been controlled substances under federal law. (We express no view on whether a defendant’s conviction would qualify as a conviction for a serious drug offense under the ACCA if the sentence had been imposed under § 2–401(B)(1) or (B)(3).) Although we consider Watkins dispositive, we also address three arguments made by the government based on other state-law sources. None is persuasive. First, the government relies on the Oklahoma Uniform Jury Instruction for Defendant’s offenses, 4 which at the time of his conviction provided: No person may be convicted of distributing a controlled dangerous substance unless the State has proved beyond a reasonable doubt each element of the crime. These elements are: First, knowingly/intentionally, Second, distributing/(transporting with the intent to distribute)/[soliciting the use]/[using the services] of a person less 4 Mathis speaks of examining the jury instructions as part of the record of conviction that may be turned to when state-law sources do not provide clear answers on the issue of divisibility. See 136 S. Ct. at 2256–57. This court has gone beyond that limited use and has also analyzed a state’s uniform jury instructions as a potential source of state law even when a defendant pleads guilty. See Titties, 852 F.3d at 1270 n.15, 1271 (“[W]here the defendant pled guilty and no jury instructions were given, the state’s uniform jury instructions can provide insight into the means/elements question.”). 14 than 18 year of age to cultivate/distribute/ manufacture/(attempt to manufacture); Third, the controlled substance of [Name of Substance]. Okla. Unif. Jury Instr. CR (2d) 6–2 (bold lettering omitted). The government contends that because the third element is singular the instruction clearly indicates that only one substance can be inserted, which in turn shows that the individual drug is an element. If the jury instruction or a usage note with the instruction expressly required that only one substance be stated in the instruction on the third element, the government would have a better argument, even though Oklahoma jury instructions are not treated as definitive, see Mitchell v. State, 387 P.3d 934, 943 (Okla. Crim. App. 2016) (“Trial courts should use the uniform jury instructions if they state the applicable law.” (emphasis added)); Okla. Stat. Ann. tit. 12, § 577.2 (“Whenever Oklahoma Uniform Jury Instructions (OUJI) contains an instruction applicable in a civil case or a criminal case, giving due consideration to the facts and the prevailing law, and the court determines that the jury should be instructed on the subject, the OUJI instructions shall be used unless the court determines that it does not accurately state the law.”). But as the uniform instruction is written, we think it does not confront the particular question of whether multiple substances could be treated as alternative means in the third element. We note that circuit courts have generally refused to treat such “singular” language in itself as establishing that every alternative that could be placed in the template is an element of the offense. See, e.g., Jimenez v. Sessions, 893 F.3d 704, 714 n.4 (10th Cir. 2018) (“[W]e are not persuaded that Colorado’s pattern jury instructions shed much light on the 15 question before us. The first-degree-trespass instruction includes a space for courts to insert the name of the ulterior offense . . . . But the pattern instruction does not tell us whether a jury would have to reach a unanimous determination if multiple ulterior offenses are at issue.”); Martinez, 893 F.3d at 1072 (“Whether means or element, the identity of a controlled substance will be inserted to complete the [Missouri approved jury] instructions when the charge involves a substance other than marijuana. But we must look elsewhere to understand whether the name of the controlled substance describes an element of the offense.”); Harbin v. Sessions, 860 F.3d 58, 68 (2d Cir. 2017) (“Although the [New York pattern jury instructions] include a blank with the word ‘specify’ in it, allowing a judge to name the substances at issue in the case, the instructions do not say it is impermissible to identify more than one substance. No rule of law or language requires that the term ‘specify’ cover only one object.”). But cf. Guillen v. U.S. Att’y Gen., 910 F.3d 1174, 1184 (11th Cir. 2018) (divisibility indicated by jury instructions requiring entry of “(specific substance)” and state intermediate appellate court’s citing instructions as setting out elements of the drug offense). Similarly, even if we were to turn to the records of Defendant’s prior convictions, we would attach little weight to the fact that the charging documents for Defendant’s two methamphetamine convictions explicitly mention methamphetamine rather than just charging him with distribution of a substance described in § 2–401(B)(2). The prosecution had no occasion to consider how to charge the joint distribution of two substances. We address summarily the government’s remaining two arguments, largely because their logic escapes us. One argument is that the penalty sections of § 2–401 16 indicate divisibility because they refer not only to drug schedules but to specific drugs. Recall that § 2–401(B)(1) sets the penalty for any “substance classified in Schedule I or II which is a narcotic drug, lysergic acid diethylamide (LSD),” or which is one of five other substances related to GHB. Okla. Stat. Ann. tit. 63, § 2–401(B)(1) (West 2011). But we fail to see how it is of any moment how Oklahoma decided to define any specific group of drugs for purposes of punishment—whether in lockstep with the statutory schedules, or by picking and choosing among the schedules. The important thing is that the separate paragraphs of § 2–401(B) make § 2–401(A)(1) divisible into groups based on the penalty imposed because of the Supreme Court’s decision in Apprendi, but any further divisibility would be inconsistent with the OCCA’s decision in Watkins. The government’s other argument is based on the Oklahoma Methamphetamine Registry Act, which establishes a registry for methamphetamine offenders and makes it unlawful for them to have Schedule V compounds or pseudoephedrine. See Okla. Stat. Ann. tit. 63, § 2–701 (West 2011). We acknowledge that if there were a unique punishment under § 2–701 for methamphetamine offenders, then methamphetamine offenses would be divisible from the other offenses encompassed by the statute. But, as the government admits, the registry is not part of the formal punishment for the underlying offense. See Bivens v. State, 431 P.3d 985, 995 (Okla. Crim. App. 2018) (“The requirements of the Oklahoma Methamphetamine Registry Act were not part of the range of punishment for Appellant’s offense nor did any statutory provision permit a judge or a jury to impose, delay, alter, or suspend registration and no provision within the Oklahoma Methamphetamine Registry Act authorizes a sentencing judge or jury to 17 require or preclude compliance with the Act. . . . [T]he Oklahoma Methamphetamine Registry Act is a regulatory scheme that is entirely separate and distinct from the applicable punishment range.”). Accordingly, the registry act is irrelevant to our divisibility analysis. B. Realistic-Probability Test Next, the government argues that even if § 2–401(A)(1) is not divisible by individual drug, the statute nevertheless satisfies the categorical approach because we can ignore the fact that Oklahoma characterizes three substances as controlled dangerous substances which are not federal controlled substances. It urges us to apply the realistic- probability test to reach that conclusion. The realistic-probability test has been applied by the Supreme Court to preclude defendants from arguing that a state statute does not satisfy the categorical approach because under certain imagined circumstances it would be possible to violate the state statute and not satisfy requirements under the ACCA for a violent felony or a serious drug offense. The test requires the defendant to show that there is “a realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside the [definition in the federal statute].” Gonzales v. Duenas-Alvarez, 549 U.S. 183, 193 (2007). In Gonzales the Supreme Court rejected the defendant’s argument that his California vehicle-theft offense did not satisfy the federal statutory definition of theft because California courts could apply state accomplice-liability doctrine to non-theft conduct. See id. at 190. The Court held that his “application of legal imagination to a state statute’s language” was insufficient. Id. at 193. Instead, he was required to point to 18 cases in which the statute had been applied to such conduct. See id. The government argues here that Defendant would fail the realistic-probability test because the parties have not identified any case prosecuting a defendant under § 2–401(A)(1) for conduct involving the three non-federally controlled substances. In this circuit, however, we have held that a defendant need not come forward with instances of actual prosecution when the “plain language” of the statute proscribes the conduct at issue. Titties, 852 F.3d at 1274. In Titties the government argued that the defendant was required to come forward with a “case in which Oklahoma has prosecuted someone under [the purported ACCA predicate statute] for pointing a firearm in obvious jest.” Id. We said that it was enough that the statute explicitly “reaches conduct undertaken for purposes of ‘whimsy, humor or prank.’” Id. So too here. The plain language of § 2–401(A)(1) expressly criminalized drugs that were not federally controlled and thus falls “outside the ACCA’s ambit.” Id.; see United States v. Abeyta, 877 F.3d 935, 944 (10th Cir. 2017) (applying this aspect of Titties in the context of a sentencing-guidelines enhancement). It requires no “legal imagination,” Gonzales, 549 U.S. at 193, to see that Oklahoma law forbade distribution of three substances that are not federal controlled substances. In sum, we conclude that § 2–401(A)(1) is not divisible by individual drug. Because that statute prohibits the distribution of three drugs that were not federally controlled, Defendant’s conviction under that statute was not a “serious drug offense” 19 under the ACCA. The enhancement of Defendant’s sentence under the ACCA was therefore error. 5 C. Plain Error Because Defendant raised no objection to the ACCA enhancement during sentencing, the last obstacle he faces is plain-error review. “Under Federal Rule of Criminal Procedure 52(b), a plain error that affects substantial rights may be considered even though it was not brought to the court’s attention.” United States v. Faulkner, 950 F.3d 670, 672 (10th Cir. 2019) (brackets and internal quotation marks omitted). But to prevail, Defendant must show that “(1) an error occurred; (2) the error was plain; (3) the error affected his substantial rights; and (4) the error seriously affected the fairness, integrity, or public reputation of a judicial proceeding.” Id. (brackets and internal quotation marks omitted). Our discussion above establishes error in Defendant’s sentencing. We further conclude that the error was plain. “An error is plain if it is clear or obvious under current, well-settled law” of this court or the Supreme Court. Id. at 678 (internal quotation marks omitted). That standard has been satisfied. Watkins is definitive and, in every material respect, is still good law in Oklahoma. See McKibbon, 878 F.3d at 976 (resolution of divisibility issue was plain, based in part on the “clear holding of the Colorado Supreme Court”). We reject the government’s contention that the absence of a previous case from this court specifically addressing § 2–401(A)(1) made the issue debatable. See Titties, 5 Having found error on this basis, we do not address Defendant’s alternative argument for reversal. 20 852 F.3d at 1272 n.19 (rejecting similar argument); Faulkner, 950 F.3d at 680 (“[T]here need not be an in-circuit case dealing with the precise state statute at issue if there is a case that sets forth a principle clearly generalizable to the subject statute.”). It is worth recalling that it falls on the government to establish divisibility with certainty, and we have suggested that this standard may benefit a defendant on plain-error review. See Titties, 852 F.3d at 1272 n.19. But see United States v. Bain, 874 F.3d 1, 31 (1st Cir. 2017) (holding, based on the interplay between plain error and the divisibility question’s certainty requirement, that a defendant “can only succeed on the plain error standard if the . . . statute was clearly indivisible at the time of his conviction”). But Defendant needs no such benefit in this case, because Watkins clearly answers the divisibility question. Nor is the error in this case less plain because of prior precedential opinions of this court holding that a conviction under § 2–401 was a serious drug offense under the ACCA. See United States v. Villanueva, 821 F.3d 1226, 1239–40 (10th Cir. 2016); United States v. McMahon, 91 F.3d 1394, 1397–98 (10th Cir. 1996). In neither case, nor in any other published or unpublished decision of this court, did we address a claim that § 2–401 could not be a serious drug offense because some Oklahoma controlled dangerous substances are not controlled substances under federal law. Those opinions are therefore not in point. See Lowe v. Raemisch, 864 F.3d 1205, 1209 (10th Cir. 2017) (“If an issue is not argued, . . . the decision does not constitute a precedent to be followed.” (internal quotation marks omitted)). 21 The government also attacks plainness on a ground other than divisibility, arguing that § 2–401(A)(1) did not clearly fail under the categorical approach because the Fifth Circuit has held that the realistic-probability test would apply to that very statute. See Aplee. Br. at 32 (“Similarly, Mr. Cantu cannot establish plain error as to the issue of the application of the realistic probability test in determining whether a statute is categorically a serious drug offense. Neither this Court nor the Supreme Court has addressed this issue, and, as the Fifth Circuit noted in Vazquez [v. Sessions, 885 F.3d 862, 872–74 (5th Cir. 2018),] the law on this issue is not well-settled.”). But whatever another circuit has to say about the realistic-probability test does not make the law in this circuit any less clear. As discussed above, this court does not apply the realistic-probability test when the statute on its face clearly proscribes the relevant conduct. Thus, the error was plain. Because Defendant’s sentence exceeded the statutory maximum, he has presumptively satisfied the remaining elements of plain error. See Rosales-Mireles v. United States, 138 S. Ct. 1897, 1907–09 (2018) (sentencing- guidelines error that satisfies first three prongs of plain-error review will ordinarily satisfy fourth prong); Faulkner, 950 F.3d at 673 n.3 (noting “the now well-established principle that a plain error leading to the adoption of an incorrect, higher Guidelines range will ordinarily satisfy plain error review’s third and fourth prongs”). And the government has not suggested any reason why this case should not be encompassed by the general rule. 22 IV. CONCLUSION We VACATE the sentence imposed by the district court and REMAND for resentencing. 23 19-6043, United States v. Cantu HARTZ, J., Circuit Judge, concurring I write separately to note two issues that this court may need to resolve in future cases that concern whether a prior conviction is a conviction for a violent felony or a serious drug offense under the ACCA. First, this court has been inconsistent about whether we apply the categorical approach based on the version of the state statutory offense in effect at the time of commission of the offense or at the time of conviction. Compare United States v. Titties, 852 F.3d 1257, 1262 n.2 (10th Cir. 2017) (“[W]e focus on the [state] law as it applied to Mr. Tittle when he committed the offense.”), with United States v. Degeare, 884 F.3d 1241, 1255 n.7 (10th Cir. 2018) (recognizing the relevant state law as that existing at the time of the defendant’s state conviction). The timing issue can be subtle. As I understand the ACCA and Supreme Court authority, the comparison that must be made is between what the defendant could have been convicted of at the time of the commission of the predicate state offense and what constitutes a federal drug offense at the time of the federal offense. Two examples may clarify the matter. Say, at the time of the state conviction substance A was a state controlled substance but has never been a federal controlled substance. A quick look might lead to the conclusion that under the categorical approach the state conviction was not for a serious drug offense under the ACCA. But what if substance A had been added as a state controlled substance only a short time before the conviction and was not a state controlled substance when the defendant committed the state offense? In that circumstance, under ex post facto doctrine the defendant could not have been convicted of an offense involving substance A, so it would seem that the offense of which the defendant was convicted would have to have been a serious drug offense. Or, say that at the time of the state offense, substance A was a state controlled substance but not a federal controlled substance, yet substance A was a federal controlled substance by the time of the defendant’s federal offense. Then the defendant was on notice when he committed his federal crime that he had a serious drug offense on his record, and I would think that the state offense would be a proper predicate serious drug offense under the ACCA. The panel opinion does not address these possibilities because there is no dispute that at all relevant times Oklahoma law included three substances as controlled dangerous substances which were not controlled substances under federal law. The second issue is whether plain-error review is fully applicable when the alleged error would make the sentence illegal. This court has applied traditional plain-error analysis in that context. See United States v. Gonzales, 558 F.3d 1193, 1198–1200 (10th Cir. 2009) (error was not clear), abrogation on other grounds recognized by United States v. Snyder, 871 F.3d 1122, 1130 n.4 (10th Cir. 2017); United States v. Cernobyl, 255 F.3d 1215, 1220 (10th Cir. 2001) (Apprendi error); United States v. Keeling, 235 F.3d 533, 539 (10th Cir. 2000) (Apprendi error but sentence affirmed because of failure to establish fourth prong). But some opinions by this court have suggested that we should always reverse when there has been an unpreserved error resulting in an illegal sentence, without the necessity of considering the second, third, or fourth prongs of plain-error review. See 2 Titties, 852 F.3d at 1275, following United States v. Gonzalez-Huerta, 403 F.3d 727, 739 n.10 (10th Cir. 2005) (en banc), following United States v. Smith, 156 F.3d 1046, 1057 (10th Cir. 1998), following United States v. Wainwright, 938 F.2d 1096, 1098 (10th Cir. 1991), following United States v. Vance, 868 F.2d 1167, 1169 (10th Cir. 1989) (stating that “the imposition of an illegal sentence would constitute plain error”). (It should be pointed out, however, that in several opinions the court has conducted traditional plain- error analysis even while also invoking cases suggesting that we always reverse when there has been an illegal sentence. See, e.g., United States v. Mendenhall, 945 F.3d 1264, 1267–70 (10th Cir. 2019); Titties, 852 F.3d at 1272 n.19; United States v. Moyer, 282 F.3d 1311, 1317–20 (10th Cir. 2002).) This circuit’s special rule originated in our opinion in Vance, which predated the Supreme Court’s decision in United States v. Olano, 507 U.S. 725, 733–37 (1993), where the Supreme Court first set forth the four elements of plain-error review. Since Olano the Supreme Court has repeatedly insisted that unpreserved error can be grounds for reversal only if the four-prong test is satisfied, 1 unless the error was structural or 1 See United States v. Marcus, 560 U.S. 258, 262–67 (2010) (failure to instruct jury that it could convict defendant only if he had engaged in prohibited conduct after effective date of statute); Puckett v. United States, 556 U.S. 129, 134–43 (2009) (government’s failure to meet its obligations under plea agreement); United States v. Dominguez Benitez, 542 U.S. 74, 80–85 (2004) (failure by district court to give a warning required by Federal Rule of Criminal Procedure 11); United States v. Cotton, 535 U.S. 625, 629–31, 633 (2002) (failure of indictment to allege element of the offense); United States v. Vonn, 535 U.S. 55, 60–74 (2002) (trial judge’s failure to advise represented defendant that he had right to counsel if he went to trial); Jones v. United States, 527 U.S. 373, 388–89 (1999) (rejecting argument that Federal Death Penalty Act created exception to requirement that defendant show that failure to give instruction was plain error); Johnson v. United States, 520 U.S. 461, 465–67 (1997) (failure to instruct jury on element of offense). 3 jurisdictional. 2 Perhaps all our decisions applying the circuit’s special rule would have reached the same result under traditional plain-error analysis. But maintaining a special rule for sentencing error may no longer be tenable. See United States v. Zangari, 677 F.3d 86, 95 (2d Cir. 2012) (criticizing this court’s failure to conduct four-prong plain- error analysis of sentencing error). If a case arises in which our special rule would make a difference, this court may wish to consider the matter en banc. 2 See Marcus, 560 U. S. at 263 (noting possibility that structural errors may automatically satisfy third prong of plain-error review); Nguyen v. United States, 539 U.S. 69 (2003) (invalidating judgment of court-of-appeals panel that included non-Article III judge, without assessing plain error); Cotton, 535 U.S. at 630 (“[D]efects in subject- matter jurisdiction require correction regardless of whether the error was raised in district court.”). 4