In an unusual turn of events last summer, U.S. Fifth Circuit Court of Appeals Judge James E. Graves, Jr. wrote the majority opinion in Alexis v. Barr, then separately wrote a concurrence criticizing his own opinion. Writing for the majority, Judge Graves denied in part and dismissed in part petitioner Alexis’s request for review of the Board of Immigration Appeals (BIA) order affirming an immigration judge’s removal order. In his concurrence, Judge Graves explained that he felt bound by Fifth Circuit precedent in writing the opinion; however, he found the result to be both “illogical and unfair.” At issue in the case was whether Alexis’s conviction under Texas law for possession of cocaine qualified as a controlled substance offense under federal law, making him, a noncitizen, removable from the United States. To answer that question, courts apply a “categorical approach” in which they compare a state criminal law statute to the federally recognized, generic definition of the crime, to determine whether the state law is a categorical match. Where a match is found, certain federal law consequences are triggered, including immigration consequences. In this case, the immigration judge, the BIA, and the Fifth Circuit all agreed that the definition of cocaine is broader under Texas law than federal law. In most circuits, the inquiry would have ended there: the court would declare the Texas law overbroad, and refrain from applying federal immigration consequences. Instead, Judge Graves went on to apply unique Fifth Circuit precedent which dictates that, in order to establish overbroadness, a petitioner must also show a “realistic probability” that Texas will prosecute conduct that falls outside the generic, federal definition of the crime, and ultimately found that Alexis failed to meet that test.
In his concurrence, Judge Graves noted that the Fifth Circuit’s application of the realistic probability test is inconsistent with recent U.S. Supreme Court and circuit court opinions. At least one immigration law expert agrees with that assessment, and adds that the Fifth Circuit’s application of the realistic probability test is also applied inconsistently within the Fifth Circuit. Judge Graves suggested that the Supreme Court should resolve the circuit split, or the Fifth Circuit should revisit the matter. Alexis petitioned the Supreme Court for certiorari, but the Court denied his request on November 16, 2020. In light of the Court’s denial, and the overwhelming evidence that the Fifth Circuit has misapplied the realistic probability test, this piece argues that the matter is ripe for reconsideration by the Fifth Circuit.
I. Background: Applying State Criminal Law to Federal Immigration Law
An individual’s conviction for certain crimes under state law can trigger federal law consequences, including additional incarceration, required registration as a sex offender, and mandatory removal from the country under immigration law, as was the case in Alexis v. United States. But state criminal statutes vary, and an individual could be convicted of a crime in one state for the same conduct that would not be illegal in another state. In an attempt to remedy these discrepancies and promote some uniformity in the application of federal law consequences to state criminal convictions, federal courts employ the “categorical approach” to determine if a state criminal statute fits within the federal, generic definition of the crime. The approach is not without its critics—some scholars argue that it creates more problems than it solves. Still, courts have widely employed the categorical approach since the Supreme Court first articulated it in 1990, and the Court has reaffirmed its use in several opinions since then.
A.The Categorical Approach Explained
In determining whether federal immigration consequences flow from a conviction under state law, courts will “compare the elements of the statute forming the basis of the defendant’s conviction with the elements of the ‘generic’ crime—i.e. the offense as commonly understood.” Courts find the elements of the federal, generic crime in a federal statute, common law, or other sources such as jury instructions. In assessing whether the state statute is a categorical match with the generic crime, “[s]entencing courts may ‘look only to the statutory definitions’—i.e. the elements—of a defendant’s prior offenses, and not ‘to the particular facts underlying those convictions.’” The Supreme Court in Descamps v. United States applied the categorical approach to a California burglary statute. The Court found the following basic elements of generic burglary: (1) unlawful or unprivileged entry into, or remaining in; (2) a building or structure; (3) with intent to commit a crime. In contrast, California Penal Code § 459 provided that a “person who enters” certain locations “with intent to commit grand or petit larceny or any felony is guilty of burglary.” Missing from the California statute was the requirement that the entry be unlawful—it was broad enough to include a shoplifter who lawfully entered a store. Therefore, the California burglary statute was overbroad, was not a categorical match within the generic definition of burglary, and no federal law consequences flowed from conviction under it.
Alexis v. United States entailed a straightforward application of the categorical approach. The petitioner, Richard Lawrence Alexis, was a native of Trinidad and Tobago who had been living in the United States as a lawful permanent resident for nearly 30 years. Many of his immediate family members were U.S. citizens, including his mother, step-father, three siblings, and young daughter. In November 2016, Alexis pled guilty and was sentenced under Texas law to one year in prison for the possession of a controlled substance. On January 23, 2018, the Department of Homeland Security initiated removal proceedings against Alexis, under 8 U.S.C. § 1227(a)(2)(B)(i), which makes noncitizens removable from the country for a conviction relating to a controlled substance. Alexis contested the removal proceeding on the grounds that his state conviction did not qualify as a federal controlled substance offense because the Texas definition of cocaine was broader than the federal definition. The immigration judge (IJ) applied a categorical approach and determined that the Texas statute was indeed overbroad—it included position isomers of cocaine, which the federal statute did not cover. However, the IJ then applied the Fifth Circuit’s interpretation of the realistic probability test and found that Alexis failed to establish overbroadness under that test.
B.The Realistic Probability Test
The categorical approach is noteworthy for its complexity, and the Supreme Court has repeatedly considered questions about how it should be applied since it was first established in 1990. In 2007, the Supreme Court in Gonzales v. Duenas-Alvarez established the realistic probability test to clear up confusion over how courts should apply the categorical approach when the state statute is a categorical match on its face, but may not be a match in its application. In that case, respondent Luis Duenas-Alvarez was a lawful, permanent resident convicted of vehicular theft under California law. The Ninth Circuit Court of Appeals found that no immigration consequences flowed from Duenas-Alvarez’s conviction because the California law, which recognizes liability for a party acting as an “accessory or accomplice” to the crime, is broader than federal law. On review, the Supreme Court disagreed, finding that the generic definition of theft at common law has long included aiders and abettors.
Interestingly, Duenas-Alvarez chose not to defend the Ninth Circuit’s position, but rather posed a novel argument that focused on the element of intent. Specifically, he argued that while the plain text of the California statute was not overbroad on its face, it was overbroad in practice because California credits commission of a crime that “naturally and probably” results from intended conduct as intent. To analyze that question, the Duenas-Alvarez Court articulated a new “realistic probability” test for courts to apply when considering whether a state’s application of its criminal statute makes it overbroad even where the plain text of the state statute does not. The majority wrote:
[T]o find that a state statute creates a crime outside the generic definition of a listed crime in a federal statute requires more than the application of legal imagination to a state statute’s language. It requires a realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside the generic definition of a crime. To show realistic probability, an offender, of course, may show that the statute was so applied in his own case. But he must at least point to his own case or other cases in which the state courts in fact did apply the statute in the special (nongeneric) manner for which he argues.
The Court went on to find that California’s application of the “natural and probable consequences” doctrine failed to meet that test: it found no evidence that California applied its statute overbroadly, and the California statute was therefore a categorical match both on its face and in practice.
II. The Fifth Circuit’s Application of Realistic Probability
The Fifth Circuit established new precedent when it applied the realistic probability test in the 2017 case of United States v. Castillo-Rivera. The appellant, Juan Castillo-Rivera, was born in Mexico and came to the United States shortly after his birth. In 2013, he pled guilty to a third felony under Texas law and was deported under federal immigration law. He then reentered the United States unlawfully and was arrested again in Texas in 2014. This last arrest led to an illegal reentry charge and enhanced sentence. Castillo-Rivera argued that the enhanced sentence did not apply to him because the Texas definition of “aggravated felony” was overbroad. A Fifth Circuit panel disagreed, and Castillo-Rivera petitioned for rehearing en banc.
In its analysis, the Fifth Circuit relied on Duenas-Alvarez to assert that Castillo-Rivera’s argument that the Texas statute was not a categorical match with the federal statute was insufficient to relieve him of the enhanced sentencing imposed by federal law. The court wrote that “Castillo-Rivera cannot simply point to certain crimes that may be included in one [statute] but not the other.” In reality, however, that is exactly how the categorical approach has been applied since the Supreme Court first articulated it in 1990. The Castillo-Rivera court went on to hold that the petitioner “must also show that Texas courts have actually applied [Texas law] in this way.” With this statement, the Fifth Circuit appears to have interpreted Duenas-Alvarez to impose a new requirement that Castillo-Rivera satisfy the realistic probability test even though he already established that the state statute was not a categorical match with federal, generic law. That is how Castillo-Rivera has been applied in subsequent Fifth Circuit cases, including Alexis. However, that interpretation is inconsistent with the actual articulation of the test in Duenas-Alvarez.
Judge Graves in his Alexis concurrence noted that the Fifth Circuit diverges from at least seven other circuit courts in its application of the realistic probability test. To illustrate this point, Judge Graves listed cases collected by the Third Circuit which demonstrate that the Fifth Circuit is alone in its approach. The Second Circuit clearly described the way that the test is applied in other circuits when it held: “The ‘realistic probability’ test articulated in Duenas-Alvarez has no role to play in the categorical analysis, however, when the state statute of conviction on its face reaches beyond the generic federal definition.” In fact, prior to the Castillo-Rivera decision, even the Fifth Circuit applied the test this way.
Judge Graves also noted that, in addition to being inconsistent with other circuit courts, the current Fifth Circuit approach is inconsistent with the Supreme Court’s application of the realistic probability test. He cited to at least two Supreme Court cases decided after Duenas-Alvarez that did not apply the realistic probability test when considering whether state convictions trigger federal consequences. Given this precedent, it seems appropriate for the Fifth Circuit to revisit its approach.
III. Conclusion: The Fifth Circuit Should Reconsider Realistic Probability
The Supreme Court established the realistic probability test to clarify whether unique application of a state’s criminal law statute could make it broader than the federal, generic law. That a facially broader state law is overbroad was never in question. As Judge Graves noted in his Alexis concurrence, the Fifth Circuit’s application of Duenas-Alvarez is inconsistent with subsequent U.S. Supreme Court cases and cases in other circuits. As such, the Fifth Circuit should revisit its decision in Castillo-Rivera, and adjust its application of Duenas-Alvarez to be more consistent with the actual meaning of the case.
 Alexis v. Barr, 960 F.3d 722 (5th Cir. 2020).
 Id. at 731 (Graves, J., concurring).
 Under the Immigration and Nationality Act, a noncitizen who has been convicted of a violation of law relating to a controlled substance is removable from the United States. 8 U.S.C. § 1227(a)(2)(B)(i). Alexis also sought review of his applications for asylum, withholding of removal, and relief under the Convention Against Torture; however, those issues are beyond the scope of this piece.
 See Taylor v. United States, 110 S. Ct. 2143 (1990).
 Texas law includes “position isomers” in its definition of “cocaine,” but federal law does not: “The parties do not dispute that Texas’s definition of ‘cocaine’ is facially broader than the federal definition of ‘cocaine.’” Alexis, 960 F.3d at 726.
 Id. (citing Vetcher v. Barr, 953 F.3d 361, 367 (5th Cir. 2020)).
 Id. at 732–33 (Graves, J., concurring).
 Immigration Review, Precedential Decisions from 6/8/20 – 6/14/20 (controlled substance, realistic probability test, CAT, changed country condition motions to reopen, unaccompanied alien minor) (June 14, 2020), https://open.spotify.com/episode/0YZkNGzMdP9AIQPA9fzX5U?si=Vv4blBpxTmmtkji8WSN1ZQ [https://perma.cc/PSE8-7JH7].
 Alexis, 960 F.3d at 734 (Graves, J., concurring).
 Denial of Petition for Writ of Certiorari, No. 20–11, 2020 WL 6701082 (S. Ct. Nov. 16, 2020).
 See generally Amit Jain & Philip Dane Warren, An Ode to the Categorical Approach, 67 UCLA L. Rev. Discourse 132 (2019).
 See generally Taylor v. United States, 110 S. Ct. 2143 (1990).
 See States’ Commandeered Convictions: Why States Should Get a Veto over Crime-Based Deportation, 132 Harv. L. Rev. 2322 (2019) (arguing that tying federal immigration consequences to state criminal law is an unconstitutional commandeering of state police power; that it is unfair to the extent that immigration consequences are triggered by state adjudication of crime and not the actual criminal act; and that the process lacks political accountability); see also Sheldon A. Evans, Categorical Nonuniformity, 120 Colum. L. Rev. 1771 (2020) (noting that the categorical approach actually results in nonuniformity because the same conduct that triggers federal immigration consequences in one state would not trigger immigration consequences in a neighboring state whose statute is broader than the federal, generic definition of the crime). But see Jain & Warren, supra note 13 (arguing that the categorical approach is good federalist policy because it incentivizes states to ensure that the most serious conduct triggers the most serious consequences).
 The Supreme Court has decided over 20 cases on the categorical approach in the past 13 years, with two more pending in the 2020–2021 term. Evans, supra note 15, at 1777.
 Descamps v. United States, 133 S. Ct. 2276, 2281 (2018).
 See generally Taylor, 110 S. Ct. 2143.
 Descamps, 133 S. Ct. at 2283 (citing Taylor, 110 S. Ct. at 2159). An exception to this rule is the “modified categorical approach,” which is applied to a “divisible” statute, which the Court describes as a statute that:
sets out one or more elements of the offense in the alternative—for example, stating that burglary involves entry into a building or an automobile. If one alternative (say, a building) matches an element in the generic offense, but the other (say, an automobile) does not, the modified categorical approach permits sentencing courts to consult a limited class of documents, such as indictments and jury instructions, to determine which alternative formed the basis of the defendant’s prior conviction.
Id. at 2281.
 Id. at 2283.
 Id. at 2282.
 Id. at 2283.
 Alexis v. Barr, 960 F.3d 722, 724 (5th Cir. 2020).
 Id. at 725.
 See generally Evans, supra note 15.
 Gonzales v. Duenas-Alvarez, 127 S. Ct. 815 (2007).
 Id. at 819.
 Id. at 820.
 Id. at 820–21.
 Id. at 821.
 Id. at 822. Notably, Justice Stevens concurred with the majority in part but dissented specifically to Part III-A, which articulates the realistic probability test. He wrote that the Court should limit its decisions to the question it granted certiorari to answer and withhold comment on issues of California law until after the court of appeals has addressed them. Id. at 825 (Stevens, J., concurring in part and dissenting in part).
 United States v. Castillo-Rivera, 853 F.3d 218 (5th Cir. 2017).
 Id. at 220.
 Id. at 221.
 Id. at 222.
 See generally Taylor v. United States, 110 S. Ct. 2143 (1990).
 Castillo-Rivera, 853 F.3d at 222.
 See generally Alexis v. Barr, 960 F.3d 722 (5th Cir. 2020).
 Id. at 733 (Graves, J., concurring).
 Id. at 733 n.1 (Graves, J., concurring) (citing Zhi Fei Liao v. Att’y Gen. U.S., 910 F.3d 714, 723 n.11 (3d Cir. 2018)).
 Williams v. Barr, 960 F.3d 68, 78 (2d Cir. 2019).
 See Mercado v. Lynch, 823 F.3d 276 (5th Cir. 2016) (rejecting the government’s argument that the petitioner had to meet the realistic probability test where the statute on its face was overbroad).
 Alexis, 960 F.3d at 732 (Graves, J., concurring).
 Id. at 732 (Graves, J., concurring) (citing Mellouli v. Lynch, 135 S. Ct. 1980 (2015); Mathis v. United States, 136 S. Ct. 2243 (2016)).
 Id. at 732–33 (Graves, J., concurring).