Harvard Law Review Harvard Law Review Harvard Law Review

Criminal Law

Categorical Mistakes: The Flawed Framework of the Armed Career Criminal Act and Mandatory Minimum Sentencing

The full text of this Comment may be found by clicking the PDF link to the left.

Congress fundamentally changed the punishment of federal crimes in the 1980s and almost entirely for the worse. The Comprehensive Crime Control Act of 19841×1. Pub. L. No. 98-473, tit. II, 98 Stat. 1976 (codified as amended in scattered sections of 18 and 28 U.S.C.). (CCCA) cabined the discretion of judges, eliminated parole, and gave greater power to prosecutors (through the use of mandatory minimum sentences, higher maximum sentences, and increased pretrial detention, all of which increased prosecutorial leverage to extract pleas).2×2. See id. The Sentencing Reform Act3×3. Pub. L. No. 98-473, tit. II, ch. II, 98 Stat. 1987 (1984) (codified as amended in scattered sections of 18 and 28 U.S.C.). was contained within the larger CCCA, and it created the United States Sentencing Commission, which was initially charged with creating mandatory sentencing guidelines for federal crimes based on its research and data.4×4. See id. § 217, 98 Stat. at 2017–26. But before the Commission even had a chance to get started, Congress made its own critical sentencing determinations for a range of crimes by passing sweeping mandatory minimum sentencing laws and harsh penalties for recidivists. This included passing the Armed Career Criminal Act5×5. Pub. L. No. 98-473, tit. II, ch. XVIII, 98 Stat. 2185 (1984) (codified as amended at 18 U.S.C. § 924(e) (2012)). (ACCA) as part of the 1984 package and later amending it in the Anti-Drug Abuse Act of 19866×6. Pub. L. No. 99-570, 100 Stat. 3207 (codified as amended primarily in scattered sections of 18 and 21 U.S.C.); see id. §§ 1401–1402, 100 Stat. at 3207-39 to -40. so that the statute imposed a fifteen-year mandatory minimum sentence on recidivist offenders convicted of a federal felon-in-possession offense who had three or more prior “serious drug” or “violent felony” convictions.7×7. 18 U.S.C. § 924(e)(1) (“In the case of a person who violates section 922(g) [criminalizing a felon’s possession of a firearm] and has three previous convictions by any court . . . for a violent felony or a serious drug offense, or both, such person shall be . . . imprisoned not less than fifteen years.”). The federal felon-in-possession provision, 18 U.S.C. § 922(g), makes it “unlawful for any person” who has been convicted of a felony to “ship or transport . . . or possess . . . any firearm or ammunition.”

The members of Congress who voted for these changes believed they would minimize unwarranted disparities in sentencing,8×8. 28 U.S.C. § 991(b)(1)(B) (2012) (listing “avoiding unwarranted sentencing disparities” as one of the purposes of creating the United States Sentencing Commission). make criminal sentences more transparent, and improve public safety.9×9. See Hillel R. Smith, Cong. Research Serv., R45220, The Federal “Crime of Violence” Definition: Overview and Judicial Developments (2018), https://fas.org/sgp/crs/misc/R45220.pdf [https://perma.cc/GE8S-G6L8] (noting that the CCCA aimed “to restore a proper balance between the forces of law and the forces of lawlessness”); see also Leslie Maitland Werner, Justice Department; Getting Out the Word on the New Crime Act, N.Y. Times, Nov. 16, 1984, at A24 (attributing that quote to Attorney General William French Smith). Unfortunately, Congress’s approach for achieving these goals was doomed to fail because of mistaken assumptions and premises. First, Congress incorrectly presumed that harsh penalties were merited for all repeat offenders, regardless of the underlying nature of their previous convictions or when they were committed. Yet by lumping together individuals with varying levels of culpability for the same mandatory punishments, Congress created disparities. Thousands of individuals received punishments disproportionate to their offenses because they were treated on par with the worst offenders Congress had in mind when passing its laws. Moreover, Congress ignored the central role played by prosecutors, who decide when and whether crimes with mandatory minimum sentences are charged. Prosecutors have not uniformly sought mandatory minimum sentences, which has led to greater disparities, particularly on the basis of race. These disparities result from a process even less transparent than the one it replaced, because all the action now takes place outside of courtrooms through plea negotiations in prosecutors’ offices. Second, Congress erroneously assumed that longer sentences and harsh collateral consequences would produce better safety outcomes, when in fact these policies often undermine public safety. Third, legislators failed to see how the new regime they created conflicted with key constitutional safeguards, paving the way for challenges in the courts that continue to this day.

The ACCA illustrates each of these flaws. Starting from the premise that a small group of repeat offenders were committing a disproportionate share of violent crimes, Congress set out to stop these so-called “career criminals” with a fifteen-year mandatory minimum sentence and a maximum sentence of life imprisonment.10×10. See 18 U.S.C. § 924(e). But instead of creating a precision regime that pinpointed and targeted the small number of people who repeatedly exhibited a propensity for violence, Congress enacted a sweeping law that ended up including individuals without any violence in their past and lumping them together with individuals who had committed numerous previous acts of violence. All of them were treated as the most dangerous type of repeat offender meriting the harsh minimum sentence. Further, as with other laws imposing harsh mandatory punishments, the ACCA has been erratically and discriminatorily applied.

In its haste to create a far-reaching new punishment regime for repeat offenders, Congress also set up a host of vexing constitutional and statutory interpretation questions for the courts. Because Congress sought to turn the prosecution of individuals with previous convictions for violent crime into a federal problem, it had to create a regime that accounted for the variety in state laws. Moreover, because Congress wanted to include a range of prior offenses as eligible for triggering the ACCA’s mandatory minimum, it used sweeping and imprecise language.

The result has been chaos in the federal courts. The Supreme Court ultimately struck down as “unconstitutionally vague” the catchall residual clause in the ACCA that had included as a “violent felony” any offense that “otherwise involves conduct that presents a serious potential risk of physical injury to another.”11×11. Johnson v. United States, 135 S. Ct. 2551, 2557 (2015); see id. at 2555–56 (quoting 18 U.S.C. § 924(e)(2)(B)(ii)). The Court last Term similarly struck down the residual clause in 18 U.S.C. § 924(c)(3)(B), see United States v. Davis, 139 S. Ct. 2319, 2336 (2019), and did the same to the residual clause in 18 U.S.C. § 16(b) during the 2017 Term in Sessions v. Dimaya, 138 S. Ct. 1204, 1216 (2018). But even with the residual clause out of commission, federal courts continue to struggle with questions about whether past state convictions meet the ACCA’s other definitions for “violent felony.” Under one prong of the definition, often referred to as the “elements clause,”12×12. See Stokeling v. United States, 139 S. Ct. 544, 549 (2019). a violent felony qualifies under the Act if it “has as an element the use, attempted use, or threatened use of physical force against the person of another.”13×13. 18 U.S.C. § 924(e)(2)(B)(i). The elements clause thus requires judges to determine what counts as “physical force.”14×14. Id. Under another prong of the definition, known as the “enumerated-offenses clause,”15×15. See Quarles v. United States, 139 S. Ct. 1872, 1880 (2019) (Thomas, J., concurring). Congress lists specific crimes as meeting the definition, and includes among the items on the list “burglary, arson, or extortion.”16×16. 18 U.S.C. § 924(e)(2)(B)(ii). This language requires courts to determine if a given state law defining those crimes matches up with the ACCA’s understanding of those crimes. Thus, some state burglary laws will qualify and some will not, depending on what the ACCA means when it uses the term “burglary.”

Last Term, the Supreme Court provided a window into the issues the ACCA poses for federal courts. The Court had three ACCA cases that demonstrate how, even decades after its passage, the ACCA continues to clog the courts with questions about which state felonies qualify as ACCA predicates for an increased sentence. In Stokeling v. United States,17×17. 139 S. Ct. 544. the Court had to determine if Denard Stokeling’s prior state conviction for robbery satisfied the ACCA’s elements clause.18×18. Id. at 549. Stokeling pleaded guilty to being a felon in possession of a firearm when the police found a gun and ammunition in his backpack while investigating him as a suspect in a burglary of his place of employment.19×19. Id. Stokeling would have been subject to the sentencing enhancements in the ACCA only if he had three qualifying predicate felonies.20×20. See id. at 556 (Sotomayor, J., dissenting). He did not challenge the statute’s applicability to his prior convictions for kidnapping or home invasion, but he argued that the third offense necessary for him to qualify, a 1997 robbery conviction under Florida law, failed to meet the ACCA’s terms.21×21. Id. at 549 (majority opinion). The Florida robbery statute, as interpreted by the Florida Supreme Court, requires “force sufficient to overcome a victim’s resistance,” which can be satisfied by minimal force if that is all that is needed to overcome the victim’s resistance.22×22. Id. at 558 (Sotomayor, J., dissenting) (quoting Robinson v. State, 692 So. 2d 883, 887 (Fla. 1997); then citing McCloud v. State, 335 So. 2d 257, 258 (Fla. 1976)). The district court concluded that Stokeling did not merit the ACCA sentencing enhancement.23×23. Id. at 549 (majority opinion). The Eleventh Circuit reversed and rejected Stokeling’s argument that Florida’s robbery law did not require sufficient force to be deemed a “violent felony” under ACCA.24×24. Id. at 549–50.

The Supreme Court affirmed Stokeling’s ACCA conviction in a decision authored by Justice Thomas and joined by Justices Breyer, Alito, Gorsuch, and Kavanaugh.25×25. Id. at 550. Justice Thomas’s opinion concluded that robbery convictions count as ACCA predicates as long as the relevant law defining robbery requires the defendant to “overcome the victim’s resistance,”26×26. Id. even if the force necessary to do so is minimal.27×27. See id. at 553. The Court’s opinion relied heavily on the common law definition of robbery as well as an earlier version of the ACCA that expressly listed robbery as a qualifying predicate.28×28. See id. at 550–52. Even though robbery was removed in a subsequent amendment to the ACCA, the majority did not interpret that change as an attempt to remove robbery as a qualifying offense, but instead as a way to expand the law’s reach.29×29. See id. at 551.

Justice Sotomayor’s dissent, which was joined by Chief Justice Roberts, Justice Ginsburg, and Justice Kagan, relied heavily on the Court’s 2010 decision in Johnson v. United States,30×30. 559 U.S. 133 (2010). which concluded that a Florida battery law failed to qualify as an ACCA predicate because it could be satisfied by nominal contact.31×31. Stokeling, 139 S. Ct. at 557 (Sotomayor, J., dissenting) (citing Johnson, 559 U.S. at 138–43). Johnson emphasized that the force required for a felony to count as an ACCA “violent felony” had to be “violent,” “substantial,” and “strong.”32×32. Id. (quoting Johnson, 559 U.S. at 140). Florida’s robbery statute, however, could be satisfied with minimal force, thereby including “glorified pickpockets, shoplifters, and purse snatchers.”33×33. Id. at 559. The dissent thus concluded that the Florida robbery law fell far short of the violent and substantial force the ACCA required for a predicate offense.34×34. Id.

In two other ACCA cases, the Court dealt with the meaning of “burglary” in the ACCA’s enumerated-offense clause. In Quarles v. United States,35×35. 139 S. Ct. 1872 (2019). Jamar Quarles pleaded guilty to being a felon in possession of a firearm after his girlfriend called 911 and said he had threatened her at gunpoint.36×36. Id. at 1875–76. Quarles had three prior felony convictions,37×37. All three previous offenses involved altercations with ex-girlfriends. Id. at 1876. One conviction was for a home invasion where Quarles attempted to chase down an ex-girlfriend who was seeking refuge in an apartment, and the other two convictions were for assault with a dangerous weapon. Id. but he argued that his conviction for third-degree home invasion failed to qualify as an ACCA predicate because it did not meet the ACCA’s definition of burglary.38×38. Id. Specifically, Quarles claimed that the Michigan law did not satisfy the ACCA because it covered situations where a defendant forms the intent to commit a crime at any time while unlawfully remaining in a dwelling, instead of covering only those instances where the intent is formed right at the moment when the defendant first becomes unlawfully present in a dwelling.39×39. Id. The district court disagreed and sentenced him to seventeen years.40×40. Id. The Sixth Circuit affirmed.41×41. Id. The Supreme Court took the case to resolve a circuit split.42×42. Id.

Justice Kavanaugh wrote the opinion for a unanimous Court.43×43. Id. The Court had already concluded in Taylor v. United States44×44. 495 U.S. 575 (1990). that the meaning of “burglary” in the ACCA should be based on “the generic sense in which the term is now used in the criminal codes of most States.”45×45. Id. at 598. The Court in Taylor further elaborated that this meant “unlawful or unprivileged entry into, or remaining in, a building or structure, with intent to commit a crime.”46×46. Id. at 599. Thus, the only question for the Court in Quarles was the timing of the intent requirement, and it agreed with the government that such intent could be formed at any time while a defendant is unlawfully present in a building or structure.47×47. Quarles, 139 S. Ct. at 1877. The Court reached this conclusion based on the ordinary meaning of “remaining in,” which refers to a “continuous activity,”48×48. Id. as well as the fact that every state appellate court facing this issue when the ACCA was passed concluded as much.49×49. Id. at 1878.

Justice Thomas wrote separately to call into question the Court’s approach to the ACCA.50×50. Id. at 1880 (Thomas, J., concurring). The Supreme Court has instructed courts to use what is called a “categorical approach” in deciding whether a predicate offense is a violent felony under either the elements clause or the enumerated-offenses clause.51×51. See Taylor, 495 U.S. at 600–02. If the case involves the elements clause, courts determine if a given conviction qualifies as a predicate felony by looking to the statutory elements of the offense of which the defendant was convicted, as opposed to looking at whether the defendant’s underlying conduct when he or she committed the crime actually involved violence.52×52. Mathis v. United States, 136 S. Ct. 2243, 2251 (2016); Taylor, 495 U.S. at 600; Office of Gen. Counsel, U.S. Sentencing Comm’n, Primer: Categorical Approach 16 (2017). Similarly, if a case involves one of the enumerated offenses, such as burglary, courts must compare the language in the statute of conviction with the generic definition of burglary that the Supreme Court concluded applies to the ACCA.53×53. Quarles, 139 S. Ct. at 1877; Taylor, 495 U.S. at 599. Justice Thomas called the categorical approach “difficult to apply” and prone to “yield[ing] dramatically different sentences depending on where a burglary occurred.”54×54. Quarles, 139 S. Ct. at 1881 (Thomas, J., concurring). He urged the Court to reconsider “whether its approach is actually required in the first place for ACCA’s enumerated-offenses clause.”55×55. Id.

A third ACCA case of the last Term, United States v. Stitt,56×56. 139 S. Ct. 399 (2018). also raised the question of the meaning of burglary in the enumerated-offenses clause. Victor Stitt and Jason Daniel Sims had consolidated cases before the Court.57×57. Id. at 404. Both of them had been convicted of being felons in unlawful possession of a firearm, and both raised questions about whether their prior burglary convictions qualified as ACCA predicates.58×58. Id. Stitt challenged the ACCA’s applicability to a prior conviction for aggravated burglary under Tennessee law because the state statute, which required burglary of a habitation, defined “habitation” to include a structure or vehicle that had been “designed or adapted for the overnight accommodation of persons.”59×59. Id. (quoting Tenn. Code Ann. § 39-14-401(1)(A) (1997)). Sims similarly argued that his prior conviction for burglary of a residential occupiable structure under Arkansas law failed to qualify because it included in its definition of “residential occupiable structure” a vehicle, building, or other structure “which is customarily used for overnight accommodation of persons whether or not a person is actually present.”60×60. Id. (quoting Ark. Code Ann. § 5-39-101(1) (1997)). In both cases, the issue was “[w]hether burglary of a nonpermanent or mobile structure that is adapted or used for overnight accommodation” qualifies as an ACCA burglary.61×61. Id. (alteration in original) (citation omitted). Both defendants lost in the district court and won on appeal (Stitt in the Sixth Circuit and Sims in the Eighth).62×62. Id. The Court granted certiorari because of disagreement in the circuits about the scope of the term “burglary.”63×63. Id.

Writing for a unanimous Court, Justice Breyer concluded that the burglary of a structure or vehicle that has been adapted or is customarily used for overnight accommodations counts as a qualifying ACCA burglary.64×64. Id. at 407. The Court again relied on Taylor and its admonition to look at the definition of burglary in the criminal codes of the states when the ACCA was passed, and it found that a majority of state burglary statutes included vehicles that had been adapted or customarily used for lodging.65×65. Id. at 405–06.

These three ACCA cases from last Term are emblematic of how these state statutory questions end up clogging the federal court dockets, as judges struggle to determine whether various statutes from the fifty states meet the ACCA’s definition of “violent felony.” The issue comes up frequently and the use of the categorical approach often results in similar crimes being treated differently because of slight differences in state statutory language.66×66. See, e.g., United States v. Mayo, 901 F.3d 218, 230 (3d Cir. 2018) (explaining that the categorical approach often produces “unsatisfying and counterintuitive” outcomes because it is “concerned only with the elements of the statute of conviction, not the specific offense conduct of an offender” (quoting United States v. Ramos, 892 F.3d 599, 606 (3d Cir. 2018))). Justice Thomas is hardly alone in his criticism of the categorical approach. Other jurists and commentators have criticized the Supreme Court’s approach for both being “extremely complicated” and producing inconsistent results that vary based on the state statute at issue.67×67. Rebecca Sharpless, Finally, a True Elements Test: Mathis v. United States and the Categorical Approach, 82 Brook. L. Rev. 1275, 1277 (2017) (citation omitted); see also United States v. Aguila-Montes de Oca, 655 F.3d 915, 917 (9th Cir. 2011) (“[W]e have struggled to understand the contours of the Supreme Court’s framework. Indeed, over the past decade, perhaps no other area of the law has demanded more of our resources.”). Justices Thomas and Alito have been particularly vocal critics on the current Court. See Quarles, 139 S. Ct. at 1881 (Thomas, J., concurring) (“[T]he categorical approach employed today is difficult to apply and can yield dramatically different sentences depending on where a burglary occurred . . . .”); Mathis v. United States, 136 S. Ct. 2243, 2269–70 (2016) (Alito, J., dissenting) (“A real-world approach would avoid the mess that today’s decision will produce.” Id. at 2269.). Use of the categorical approach also leads in some cases to someone who has what appears to be violent behavior in his or her past nevertheless not qualifying under the elements test.68×68. See, e.g., United States v. Davis, 875 F.3d 592, 595 (11th Cir. 2017). Judges bristle at having to “go down the rabbit hole . . . to a realm where we must close our eyes as judges to what we know as men and women.”69×69. Id.; see also Sheldon A. Evans, Punishing Criminals for Their Conduct: A Return to Reason for the Armed Career Criminal Act, 70 Okla. L. Rev. 623, 643–45 (2018) (describing complaints of courts and agreeing with them by calling the categorical approach “contrived” and “not based in reality,” id. at 645).

These criticisms have merit, but the target is misplaced. The fault does not lie with the Supreme Court.70×70. To be sure, the Supreme Court at times exacerbates the problems by stretching the language of the ACCA and failing to apply the rule of lenity in favor of defendants. See, e.g., infra pp. 232–35 (explaining the flaws with the majority’s approach in Stokeling and the expansive reading it gives the ACCA). The blame for this regime falls squarely on Congress and the statutory framework it elected to adopt. The ACCA and the categorical rule are pieces of a much broader, irrational federal framework put in place in the 1980s that persists to this day. The categorical approach and mandatory minimum punishments both fail to recognize important individual differences in cases. In neither setting do judges have the flexibility they need to match sentences with relevant facts to create proportionate outcomes. The problem in both contexts stems from Congress’s desire to strip judges of discretion, to take on a greater institutional role for itself in dictating sentences in individual cases without evaluating data and evidence, and to make the prosecution of violent crime a federal issue instead of leaving it to the states. While Congress instituted these reforms in the name of public safety, its actual policies have ended up making recidivism more likely, while creating glaring disparities and disproportionate sentences.

The ACCA cases last Term show how this regime puts the federal courts in general and the Supreme Court in particular in the almost impossible position of trying to make the ACCA a coherent punishment regime, given the irrational and poorly researched foundation on which it rests. Consider the Court’s choices in Stokeling in determining whether a robbery statute requiring minimal force should be included as an ACCA predicate. On the one hand, there is the fact that Congress originally listed robbery as one of only two predicate offenses that trigger the ACCA,71×71. See Stokeling, 139 S. Ct. at 551 (noting the “two enumerated crimes of ‘robbery or burglary’” in the original statute). and the statute in Stokeling mirrored the traditional common law definition of robbery that Congress likely had in mind.72×72. See id. at 550–52. Thus, Congress may well have intended to include an offense as minor as the one in Stokeling because it did not take the time to consider the variety in state robbery statutes. On the other hand, as the Court stated in Johnson and the dissent reiterated in Stokeling, it makes little sense to give a fifteen-year mandatory minimum sentence based on a prior record that includes slightly aggravated pickpocketing and purse snatching.73×73. Id. at 559 (Sotomayor, J., dissenting); see Johnson v. United States, 559 U.S. 133, 140–41 (2010). The Court faced the dilemma of which of these arguments should prevail because Congress simultaneously wanted to target the most serious repeat offenders but did not bother to research how best to do that. It ended up using slapdash language that fails to recognize the complexity in the laws of the fifty states and forced the federal courts and ultimately the Supreme Court to work out the details.

While the Court had an easier time reaching unanimous decisions in the two cases that required it to define burglary,74×74. Quarles, 139 S. Ct. at 1875; Stitt, 139 S. Ct. at 403–04. that was only because the Court previously settled on the idea that a generic definition of burglary governed and should be based on the dominant approach in the states when the ACCA was passed.75×75. See Taylor v. United States, 485 U.S. 575, 598 (1990). Thus, in the burglary cases, applying the ACCA was only a matter of deciding the prevailing approach to burglary and seeing if the state laws at issue in Stitt and Quarles matched up. But the Justices’ unanimity masks broader disagreement in the federal courts about whether judges should take a narrower approach to some of the ACCA’s enumerated offenses, precisely because the harsh fifteen-year sentence can be too easily applied to individuals with criminal histories nothing like those of the repeat offenders Congress discussed when it initially passed the ACCA. For example, while Congress may have wanted a broad definition of burglary because it assumed burglaries are “inherently dangerous,”76×76. Stitt, 139 S. Ct. at 406. the reality is that more than 97% of burglaries involve no physical harm to anyone.77×77. See infra p. 231. Congress passed a law focused on a crime bearing very little relationship to the violence it sought to prevent, leaving courts to figure out whether they should interpret the scope of that offense narrowly to limit imposition of the mandatory minimum, or whether they should take a more expansive approach based on Congress’s flawed premises.

Solving this dilemma is about much more than keeping or jettisoning the categorical rule. The complexity of the ACCA cases does not stem from the Supreme Court’s categorical rule but from Congress’s failure to wrestle with any of the tough questions that go along with effectively deciding to turn state crimes into federal ones and to impose harsh consequences as a blanket matter. Instead of seeking to discard or limit the categorical rule, as Justice Thomas and others advocate, those who see the problems with the categorical approach should recognize that a true fix to the mess created by the ACCA requires abandoning a legislative framework of punishment that over-federalizes crime, relies on mandatory minimum sentences, and makes assumptions not grounded in fact and research.

Congress should have allowed the Sentencing Commission to use data and evidence to guide sentencing policy and to identify how best to address previous state convictions in the contexts of sentencing people for violating federal crimes, instead of trying to take that task on for itself. A guideline model also has the virtue of giving judges more leeway to make punishments fit the facts before them. Unfortunately, Congress chose a different path and created a regime that is fundamentally flawed because of its own mistaken assumptions. The categorical rule is but one example of its unsound approach.



* Vice Dean and Segal Family Professor of Regulatory Law and Policy and Faculty Director, Center on the Administration of Criminal Law, NYU School of Law. I owe a huge thanks to Leah Litman for her constructive comments. I am grateful to Albert Huber and Jonathan Spratley for excellent research assistance and to the editors of the Harvard Law Review for their outstanding comments and suggestions.