United States v. Garfield Redd

Opinion

PUBLISHED

UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT

No. 20-6957

UNITED STATES OF AMERICA, Plaintiff - Appellee,

v. GARFIELD REDD, Defendant - Appellant.

Appeal from the United States District Court for the District of Maryland, at Baltimore. Richard D. Bennett, Senior District Judge. (1:07-cr-00470-RDB-1)

Argued: May 4, 2023 Decided: October 19, 2023

Before DIAZ, Chief Judge, WYNN, and QUATTLEBAUM, Circuit Judges.

Reversed and remanded by published opinion. Judge Wynn wrote the opinion, in which Chief Judge Diaz and Judge Quattlebaum joined. Judge Quattlebaum wrote a concurring opinion.

ARGUED: Paresh S. Patel, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Greenbelt, Maryland, for Appellant. Jason Daniel Medinger, OFFICE OF THE UNITED STATES ATTORNEY, Baltimore, Maryland, for Appellee. ON BRIEF: James Wyda, Federal Public Defender, OFFICE OF THE FEDERAL PUBLIC DEFENDER, Baltimore, Maryland, for Appellant. Erek L. Barron, United States Attorney, OFFICE OF THE UNITED STATES ATTORNEY, Baltimore, Maryland, for Appellee.

2

WYNN, Circuit Judge:

Garfield Redd was convicted of possession of a firearm by a felon in violation of 18 U.S.C. § 922(g)(1). The district court sentenced Redd to 240 months’ imprisonment, applying the sentencing enhancement provided by the Armed Career Criminal Act (“ACCA”), 18 U.S.C. § 924(e), after finding that Redd had three qualifying predicate convictions for a “serious drug offense” or “violent felony.”

But after the Supreme Court decided Johnson v. United States, striking down the “residual clause” of ACCA’s violent-felony definition as unconstitutional, 576 U.S. 591, 606 (2015), Redd filed a § 2255 motion to vacate his ACCA sentence. He argued that Maryland first-degree assault—which formed the basis for two of his ACCA predicate offenses—was not a “violent felony” under ACCA. The district court denied Redd’s motion, and he now appeals.

Upon review, we hold that Maryland first-degree assault is not a “violent felony”

under ACCA. Accordingly, we reverse the district court’s denial of Redd’s § 2255 motion, vacate Redd’s ACCA sentence, and remand for further proceedings.

I.

In 2008, a jury convicted Redd of possession of a firearm by a convicted felon, in violation of 18 U.S.C. § 922(g)(1). Normally, at the time, such a conviction carried a maximum ten-year sentence. 1 18 U.S.C. § 924(a)(2) (2008). But Redd’s presentence report

1

In June 2022, Congress increased the maximum penalty for a § 922(g) offense to fifteen years. Bipartisan Safer Communities Act, Pub. L. No. 117-159, § 12004(c), 136 Stat. 1313, 1329 (2022) (codified at 18 U.S.C. § 924(a)(8)).

3

recommended concluding that he was an armed career criminal under ACCA, which would raise his sentence to a minimum of fifteen years and a maximum of life. 18 U.S.C. § 924(e)(1); see United States v. Allred, 942 F.3d 641, 646 (4th Cir. 2019). Application of the ACCA enhancement would also raise the maximum authorized term of supervised release from three to five years. See 18 U.S.C § 3559(a)(1) (providing that an offense for which the maximum authorized term of imprisonment is life constitutes a Class A felony, while an offense with a maximum prison term of ten years is Class C); id. § 3583(b)(1), (2) (providing that maximum authorized term of supervised release for a Class A felony is five years, and for Class C is three years). At the sentencing hearing, the district court adopted the presentence report’s recommendation and applied the fifteen-year statutory minimum under ACCA.

In determining that Redd had reached armed-career-criminal status, the district court relied on four of Redd’s prior convictions: one for Maryland possession with intent to distribute cocaine in 1987, one for Maryland common-law assault in 1987, and two for Maryland first-degree assault, in 1998 and 2000. Redd argued that these convictions were not valid ACCA predicates, but the district court rejected his arguments and sentenced him to 240 months’ imprisonment, followed by five years of supervised release. 2

2

In 2021, the district court reduced Redd’s term of imprisonment to time served on a compassionate-release motion, but Redd is still serving a five-year term of supervised release. The fact that Redd is not presently imprisoned does not affect our ability to consider this appeal. United States v. Swaby, 855 F.3d 233, 238–39 (4th Cir. 2017) (explaining that the “custody” requirement of § 2255 is assessed at the time of filing and is in any event satisfied by serving a term of supervised release). And the controversy here is not mooted by Redd’s release from prison. United States v. Ketter, 908 F.3d 61, 66 (4th

4

On direct appeal, Redd challenged his designation as an armed career criminal under ACCA. This Court affirmed Redd’s sentence, concluding that three of his prior convictions—one for Maryland possession with intent to distribute cocaine and two for Maryland first-degree assault—“qualified as categorical predicate offenses” under ACCA. 3 United States v. Redd, 372 F. App’x 413, 415 (4th Cir. 2010) (per curiam).

In 2011, Redd filed his first 28 U.S.C. § 2255 motion, which the district court denied. Redd appealed the district court’s decision, but we declined to grant a certificate of appealability.

In 2015, the Supreme Court decided Johnson v. United States (“Johnson II”), holding that the residual clause of ACCA’s violent-felony definition was unconstitutional. 4 Johnson II, 576 U.S. at 606. Redd sought authorization to file a successive § 2255 motion in light of Johnson II, which we granted. Accordingly, in May 2016, Redd filed a second § 2255 motion to vacate his ACCA sentence. Redd argued that Maryland first-degree assault was an indivisible offense with alternative means and that, because one of the means could be committed in a way that would not constitute a categorically violent felony under the force clause of ACCA’s violent-felony definition, Maryland first-degree assault did not

Cir. 2018) (rejecting the argument that “an appeal by a defendant still serving his supervised release sentence is moot because he is no longer imprisoned”).

3

The Court did not address the fourth conviction—for Maryland common-law assault—because it found that Redd already had three valid predicates after considering the cocaine and first-degree-assault charges. Redd, 372 F. App’x at 416 n.*.

4

We will refer to this case as Johnson II because it followed an initial, 2010 decision that, while involving a different defendant with the same surname, also related to ACCA. See Johnson v. United States, 559 U.S. 133 (2010) (“Johnson I”).

5

qualify as a predicate violent felony. 5 Once again, the district court denied Redd’s § 2255 motion, concluding that this Court’s prior unpublished opinion in Redd’s direct appeal— which stated that Redd’s prior convictions qualified as predicate offenses under ACCA— foreclosed Redd’s arguments. Redd v. United States, No. CR RDB-07-0470, 2020 WL 2061509, at *3 (D. Md. Apr. 29, 2020) (citing Redd, 372 F. App’x at 415).

This appeal followed. We granted a certificate of appealability “on the question of whether Maryland first-degree assault is a violent felony under the force clause of the Armed Career Criminal Act.” J.A. 63. 6 II.

A.

“We review de novo a district court’s legal conclusions concerning a denial of § 2255 relief,” including “the question of whether a prior conviction qualifies as a ‘violent felony’ under [ACCA].” United States v. Proctor, 28 F.4th 538, 544 (4th Cir. 2022) (quoting United States v. Jones, 914 F.3d 893, 899 (4th Cir. 2019)).

5

Redd also argued that other prior convictions, including the Maryland common-

law-assault conviction, were not valid ACCA predicates. Of the four that the district court relied on in sentencing Redd as an armed career criminal, only the first-degree-assault convictions remain contested: intervening precedent has made clear that the common-law- assault conviction is not a valid ACCA predicate, United States v. Royal, 731 F.3d 333, 342 (4th Cir. 2013), and Redd does not argue in this appeal that the possession-with-intent- to-distribute-cocaine charge is not a valid ACCA predicate.

6

Citations to the “J.A.” refer to the Joint Appendix filed by the parties in this appeal.

6

B.

ACCA provides that anyone who violates 18 U.S.C. § 922(g) and has three previous convictions for “a violent felony or a serious drug offense, or both, committed on occasions different from one another” shall be fined and imprisoned for a minimum of fifteen years. 18 U.S.C. § 924(e)(1). Redd’s prior drug conviction and common-law-assault conviction are not at issue in this case; rather, the question is whether the two first-degree-assault convictions the district court relied on as ACCA predicates at sentencing are properly considered “violent felonies.”

The Act defines “violent felony” as “any crime punishable by imprisonment for a term exceeding one year” that (1) “has as an element the use, attempted use, or threatened use of physical force against the person of another” (the “force clause”); (2) “is burglary, arson, or extortion, [or] involves use of explosives” (the “enumerated-offenses clause”); or (3) “otherwise involves conduct that presents a serious potential risk of physical injury to another” (the “residual clause”). Id. § 924(e)(2)(B). But Johnson II declared the residual clause unconstitutionally vague. 576 U.S. at 606. And no one contends that Redd has a predicate enumerated offense. Thus, Redd’s predicate convictions for violent felonies may only remain valid predicates after Johnson II if they can satisfy the force clause.

In the context of the force clause, “the phrase ‘physical force’ means violent force—

that is, force capable of causing physical pain or injury to another person.” Johnson I, 559 U.S. at 140. And the force clause addresses only those offenses in which the perpetrator “direct[s] his action at, or target[s], another individual”—not those in which he merely acts recklessly. Borden v. United States, 141 S. Ct. 1817, 1825 (2021).

7

To determine whether a prior conviction is a violent felony for ACCA purposes, we apply the “now-familiar” categorical approach: we examine only the elements that must be proven for the offense in question, not the facts of how the defendant committed the crime, to determine whether the offense meets the force-clause definition. Borden, 141 S. Ct. at 1822; see Taylor v. United States, 495 U.S. 575, 600 (1990). “If any—even the least culpable—of the acts criminalized” does not require the kind of conduct that the force clause contemplates, “the statute of conviction does not categorically match the federal standard, and so cannot serve as an ACCA predicate.” Borden, 141 S. Ct. at 1822.

We employ a variant of the categorical approach—“labeled (not very inventively)

the ‘modified categorical approach’—when a prior conviction is for violating a so-called ‘divisible statute.’” Descamps v. United States, 570 U.S. 254, 257 (2013). A divisible statute sets out “one or more elements of the offense in the alternative,” effectively listing different crimes. Id. Where one alternative satisfies ACCA’s “violent felony” definition, but another does not, the modified categorical approach allows the court 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. The court then applies the garden-variety categorical approach to the specific alternative that served as the basis for the prior conviction. Id. If that alternative constitutes a categorically violent felony, then the conviction may serve as an ACCA predicate. If we conclude that the statute is divisible but are unable to discern which alternative served as the basis for the defendant’s prior conviction, we assume that the conviction was for the “least serious” conduct

8

encompassed by the statute, United States v. Vann, 660 F.3d 771, 775 (4th Cir. 2011) (en banc) (per curiam), and we apply the categorical approach to that alternative.

But if the statute is indivisible, the modified categorical approach has no role to play. Descamps, 570 U.S. at 258. When an indivisible statute criminalizes several different types of conduct, any one of which could constitute the offense, we simply ask whether an individual could be convicted for any of those alternative means based on conduct that is not categorically violent. Borden, 141 S. Ct. at 1822. If so, the crime may not serve as a predicate, regardless of the defendant’s own offense conduct.

Here, Redd argues that the Maryland first-degree-assault statute is an indivisible statute that merely sets out alternative means of committing a single offense. So, he contends, we must apply the categorical approach to the entire statute, asking whether any conduct criminalized by the statute can be committed in a way that is not categorically violent.

Accordingly, we must first determine whether Maryland first-degree assault is a divisible or an indivisible statute. If it is indivisible, we then ask whether any of the modalities by which it can be committed fails to satisfy ACCA’s force clause. 7 And,

7

The district court declined to address Redd’s divisibility argument, finding that this Court’s unpublished opinion in Redd’s direct appeal foreclosed the argument that Maryland first-degree assault is not a valid ACCA predicate. Redd, 2020 WL 2061509, at *3. But we addressed Redd’s direct appeal without the benefit of Borden, 141 S. Ct. 1817, which had not yet been decided and which clarified what convictions may serve as ACCA predicates. Thus, we are not constrained by our prior opinion on this issue.

9

ultimately, we conclude that the statute is indivisible and can be violated in a manner that does not satisfy the force clause. 8 C.

A divisible statute “list[s] elements in the alternative, and thereby define[s] multiple crimes.” Mathis v. United States, 579 U.S. 500, 505 (2016). Elements, of course, are the “‘constituent parts’ of a crime’s legal definition”—that is, “the things the ‘prosecution must prove to sustain a conviction.’” Id. at 504 (quoting Elements of Crime, Black’s Law Dictionary (10th ed. 2014)). “At a trial, they are what the jury must find beyond a reasonable doubt to convict the defendant.” Id.; see also Richardson v. United States, 526 U.S. 813, 818 (1999) (explaining that if a statute “makes each ‘violation’ a separate element, then the jury must agree unanimously about which . . . crimes the defendant committed”). And “at a plea hearing, [elements] are what the defendant necessarily admits when he pleads guilty.” Mathis, 579 U.S. at 504.

By contrast, an indivisible statute simply lists different ways a single crime may be committed, Mathis, 579 U.S. at 504–05; it does not “contain[] alternative elements” that must be proven by the prosecution and agreed upon unanimously by a jury, but rather

8

We note that, in another case that was recently pending before this Court, the Government conceded that Borden—in which the Supreme Court held that reckless conduct does not satisfy the ACCA violent-felony definition—“impact[ed]” the question of “whether first-degree assault with a firearm under subsection (a)(2) of [the Maryland first-degree-assault statute]” was a “crime of violence” under 18 U.S.C. § 16(a) and therefore asked that the Court remand the case to the Board of Immigration Appeals to ultimately be terminated. Respondent’s Unopposed Motion to Remand and Stay Briefing Schedule at 2, Walcott v. Garland, No. 19-2469 (4th Cir. July 23, 2021), ECF No. 45.

10

simply lists alternative means by which a defendant might commit the crime, Descamps, 570 U.S. at 258. For example, “suppose a statute requires use of a ‘deadly weapon’ as an element of a crime and further provides that the use of a ‘knife, gun, bat, or similar weapon’ would all qualify.” Mathis, 579 U.S. at 506. Because that list of weapons “merely specifies diverse means of satisfying a single element of a single crime—or otherwise said, spells out various factual ways of committing some component of the offense—a jury need not find (or a defendant admit) any particular item.” Id. Indeed, the jury could convict even if some jurors found that the defendant committed the crime with a knife and others found that he used a gun, “so long as all agreed that the defendant used a ‘deadly weapon.’” Id. (citing Descamps, 570 U.S. at 270). That statute, then, is indivisible.

Thus, “[a] prosecutor charging a violation of a divisible statute must generally select the relevant element from its list of alternatives,” and “the jury, as instructions in the case will make clear, must then find that element, unanimously and beyond a reasonable doubt.” Descamps, 570 U.S. at 272. An indivisible statute, by contrast, does not require a jury to unanimously agree as to the means by which the defendant committed the crime. Mathis, 579 U.S. at 506. In sum, “the focal point of the analysis is what the jury must find (or a defendant must admit) to convict.” United States v. Al-Muwwakkil, 983 F.3d 748, 755 (4th Cir. 2020), as amended (Dec. 28, 2020); see also United States v. Fuertes, 805 F.3d 485, 498 (4th Cir. 2015) (“Any statutory phrase that . . . refers to multiple, alternative means of commission must still be regarded as indivisible if the jurors need not agree on which method of committing the offense the defendant used.” (quoting Rendon v. Holder, 764 F.3d 1077, 1085 (9th Cir. 2014))).

11

To discern whether a statute is divisible or indivisible, courts may consult several sources for guidance. Mathis, 579 U.S. at 518–19. For one, the statute itself, “on its face[,] may resolve the issue.” Id. at 518. For example, where a statute assigns “different punishments” to its alternatives, those alternatives are elements, effectively setting out different crimes. Id. Additionally, where a statute identifies components that must be included in the charging documents, those components are elements. Id.

Although the language and structure of a statute may suggest divisibility—for instance, by employing disjunctive language or setting out multiple modalities of the offense in different subsections—statutory language alone “may not invariably answer the divisibility question.” Allred, 942 F.3d at 649; see also Omargharib v. Holder, 775 F.3d 192, 194 (4th Cir. 2014) (explaining that the “mere use of the disjunctive ‘or’ in the definition of a crime does not automatically render it divisible”). Still, where one statutory alternative “differs . . . significantly” from the other, the court “must treat the two as different crimes.” United States v. Roof, 10 F.4th 314, 403 (4th Cir. 2021) (quoting Allred, 942 F.3d at 650).

In addition to the statutory text, we may consider the practices of the state itself for guidance. Indeed, a decision of a state court may “definitively answer[] the question” of whether the statutory alternatives are treated by state courts as elements or means. Mathis, 579 U.S. at 517. Because state courts are the authoritative interpreters of state-law issues, if a state court has definitively answered the elements-or-means question, “we simply are not at liberty to ignore that determination.” Schad v. Arizona, 501 U.S. 624, 636 (1991). A state’s charging practices may also prove instructive, Al-Muwwakkil, 983 F.3d at 756, as

12

might documents used in prosecuting the offense, like pattern jury instructions, see Omargharib, 775 F.3d at 199.

If, after consulting the statutory text and other state-law sources, it remains unclear whether the alternatives are means or elements, we may “peek” at the record of the prior conviction—for instance, the charging documents and jury instructions used in the defendant’s case—to determine whether the statute is divisible. Mathis, 579 U.S. at 518– 19; see also Shepard v. United States, 544 U.S. 13, 20–21 (2005). We consult these sources only “if state law fails to provide clear answers” and “for the sole and limited purpose” of determining whether the statutory alternatives are elements or means, not to examine how the defendant’s particular crime was committed. Mathis, 579 U.S. at 518 (internal quotation marks omitted).

If at the end of the day, after employing the tools set out in Mathis, a court is still unclear on the divisibility of the statute, the statute must be treated as indivisible. That is so because, to find that a state statute is divisible, we must discern “clear signals” indicating as much—“signals that convince us to a certainty that the elements are correct and support divisibility.” United States v. Hope, 28 F.4th 487, 503–04 (4th Cir. 2022) (emphasis added) (quoting Najera-Rodriguez v. Barr, 926 F.3d 343, 356 (7th Cir. 2019)). We demand certainty because, if we interpret state law incorrectly in a single case by “finding that state laws include essential elements that state courts have not treated as such,” we run the risk of “mistakenly cast[ing] doubt on the much higher volume of state criminal prosecutions under those same state statutes.” Najera-Rodriguez, 926 F.3d at 356.

13

1.

As stated previously, to determine if the Maryland first-degree-assault statute, Md.

Code, Art. 27 § 12A-1 (1996), is divisible, Mathis instructs us to first look to state law. At the time of Redd’s prior convictions, the statute provided, in relevant part 9:

(a) Serious physical injury; use of a firearm – (1) A person may not intentionally cause or attempt to cause serious physical injury to another.

(2) A person may not commit an assault with a firearm, including [any of the enumerated kinds of firearms].

(b) Penalty. – A person who violates this section is guilty of the felony of assault in the first degree and on conviction is subject to imprisonment for not more than 25 years.

Md. Code, Art. 27 § 12A-1 (1996).

Here, the statutory text alone does not make clear whether the statute is divisible or indivisible. True, the statutory alternatives are distinct, set out as separate sentences in different subsections. But while statutory language can suggest divisibility, language alone will not “invariably answer the divisibility question.” Allred, 942 F.3d at 649. And the statute lacks further indicia of divisibility—for instance, it assigns only one punishment for committing first-degree assault, rather than assigning separate punishments for each subsection. See Mathis, 579 U.S. at 518.

Additionally, the statutory alternatives here do not “differ[] so significantly” from each other that we “must treat the two as different crimes.” Roof, 10 F.4th at 403 (quoting

9

The statute has since been recodified and amended, including by adding a third mechanism by which first-degree assault may be committed, that is, “intentional[] strang[ulation].” Md. Code, Crim. Law § 3-202(b)(3) (2020). Nevertheless, as to the substance of the first two modalities and penalty, the statute remains identical to the version that was in place when Redd was convicted. See id. § 3-202(b)(1)–(2), (c).

14

Allred, 942 F.3d at 650). The two means of accomplishing Maryland first-degree assault overlap such that either might apply to a single criminal act: for instance, if a person strikes another with a firearm, that could constitute “an assault with a firearm,” or it could constitute “intentionally caus[ing] . . . serious physical injury to another.” Md. Code, Art. 27 § 12A-1.

Another Maryland statute, Md. Code, Crim. Law § 3-206(a), also sheds light on our interpretation of the first-degree-assault statute by “identify[ing] which things must be charged (and so are elements) and which need not be (and so are means).” Mathis, 579 U.S. at 518. That statute indicates that an indictment for first-degree assault is “sufficient” if it states that the defendant “assaulted (name of victim) in the [first] degree . . . in violation of (section violated) against the peace, government, and dignity of the State.” Md. Code, Crim. Law § 3-206(a). 10 So the State need not specify which modality of the statute it contends the defendant violated. A Maryland statute explicitly approving of an indictment for first-degree assault that does not indicate under which modality the defendant is charged strongly suggests that the statute is indivisible. See Mathis, 579 U.S. at 518.

We also look to the decisions of Maryland courts for additional insight. Because the divisibility question is a component of the application of ACCA, a federal criminal statute, the Maryland courts have not had occasion to address whether the first-degree-assault

10

A similar version of this statute has existed since at least 1996. See Md. Code, Art. 27 § 12A-4 (1996).

15

statute is divisible. Thus, the Government urges us to look to the reasoning of those courts in analogous contexts for guidance.

In the primary case relied upon by the Government—Dixon v. State—the Court of Appeals 11 analyzed the first-degree-assault statute in the context of merger for double- jeopardy purposes, asking whether a defendant’s conviction for first-degree assault merged into attempted voluntary manslaughter. 772 A.2d 283, 297 (Md. 2001). The court concluded that the first modality of first-degree assault—intentionally causing serious physical injury—did merge with attempted voluntary manslaughter, but that the second modality—assault with a firearm—did not. Id. at 301–02. The court then attempted to determine under which modality the defendant had been convicted. Id. at 303–04.

The court concluded that the record was “not clear” on this point. Id. at 303. The trial judge had instructed the jury that it must decide whether “the defendant used a firearm to commit the assault or the defendant intended to cause serious physical injury in the commission of the assault.” Id. at 305. But, “[u]pon returning its [guilty] verdict, the jury did not identify verbally under which modality it had convicted [the defendant],” nor did the verdict sheet indicate which modality the conviction was based on. Id. at 305–06. Thus, the court “[could] not determine under which modality the jury convicted [the defendant]

11

We note that, effective December 14, 2022, the Court of Appeals of Maryland has been renamed the Supreme Court of Maryland, and the Court of Special Appeals of Maryland has been renamed the Appellate Court of Maryland. Press Release, Md. Courts, Voter-Approved Constitutional Change Renames High Courts to Supreme and Appellate Court of Maryland (Dec. 14, 2022), https://www.courts.state.md.us/media/news/2022/pr20221214 [https://perma.cc/MU4W- CAQ2].

16

of first degree assault.” Id. at 306. The court “recommended” that, “to avoid this type of ambiguity” in the future, trial courts use special verdict forms. Id. at 303 n.33.

The Government reads the Court of Appeals’ suggestion in Dixon—that trial courts use special verdict forms to avoid difficult merger questions—as evincing the statute’s divisibility. But we think the court’s recommendation—rather than requirement—that trial courts use special verdict forms for first-degree-assault charges suggests that the statute is in fact indivisible. Indeed, if the jury were required to unanimously agree as to the modality of first-degree assault that the defendant was guilty of—as would be true if the statute were divisible—then special verdicts would always be appropriate in cases where the jury was instructed as to both modalities of first-degree assault.

The Maryland Criminal Pattern Jury Instructions for first-degree assault indicate that juries are typically not instructed that they must unanimously agree as to the modality of which the defendant is guilty. The current instructions provide, in relevant part 12:

In order to convict the defendant of first degree assault, the State must prove all of the elements of second degree assault and also must prove that:

(1) the defendant used a firearm to commit assault; or (2) the defendant intended to cause serious physical injury in the commission of the assault . . . .

Md. Crim. Pattern Jury Instructions 4:01.1A. The Court of Special Appeals has indicated that an instruction to this effect—which did not instruct the jury that it had to unanimously

12

As noted, the statute now includes a third modality—intentional strangulation—

which was not part of the statute at the time of Redd’s prior convictions. Thus, we have omitted that portion of the jury instructions here.

17

agree as to the modality—was “literally a correct statement of the law.” Dickerson v. State, 40 A.3d 1122, 1127 (Md. Ct. Spec. App. 2012).

The Notes on Use accompanying the jury instructions are particularly helpful because they suggest courts use special verdict forms in some, but not all, first-degree- assault cases. The Notes instruct that the court should require a special verdict on first- degree assault in the specific scenario where a defendant is charged with both first-degree assault and the use of a handgun in the commission of a felony or crime of violence. Md. Crim. Pattern Jury Instructions 4:01.1A, Notes on Use. Again, the direction that courts employ special verdicts in a narrow set of enumerated circumstances suggests that courts generally need not require special verdicts, and therefore that jurors are generally not required to unanimously find a defendant guilty under any particular modality.

Further examination of Maryland case law makes clear that declining to instruct the jury that it must agree unanimously on a modality of first-degree assault is not an anomaly. Indeed, despite the Dixon court’s suggestion, Maryland courts still decline to impress upon juries the distinctions between the first-degree-assault modalities, let alone instruct that the juries must unanimously agree as to one of the modalities. See, e.g., Wright v. State, 282 A.3d 284, 291 (Md. Ct. Spec. App. 2022) (noting that jury was instructed as to both modalities of first-degree assault and found the defendant guilty but that “the record [did] not reveal which modality of first-degree assault the jury accepted or whether the jury unanimously agreed upon a modality at all”); Dickerson, 40 A.3d at 1127 (finding that the trial court did not abuse its discretion in instructing the jury as to both modalities of first- degree assault and declining to answer a question from the jury in a manner that would

18

have effectively taken one modality out of consideration); Snyder v. State, 63 A.3d 128, 138 (Md. Ct. Spec. App. 2013) (concluding that “there was sufficient evidence presented to the jury at trial to convict [the] appellant of first-degree assault” without stating under which alternative).

The Government also argues that language used by Maryland courts to describe the first-degree-assault statute indicates divisibility. Maryland courts have, at times, referred to the different forms of first-degree assault as “elements.” See Wright, 282 A.3d at 290 (“First-degree assault with the use of a firearm includes the element of possessing a firearm.”); Dickerson, 40 A.3d at 1125 (“In the second of the aggravating modalities, the handgun, or other firearm, is a legal element of the crime itself.”); Dixon, 772 A.2d at 303 (explaining that “multi-purpose criminal statute[s],” like the first-degree-assault statute, contain “alternative elements”). Of course, if the different modalities were, in fact, “elements,” then the statute would be divisible. See Mathis, 579 U.S. at 505.

But the mere fact that Maryland courts have used the word “element” is not dispositive—indeed, Maryland courts have used a number of different words in referring to the statute. See, e.g., Middleton v. State, 192 A.3d 777, 781 n.7 (Md. Ct. Spec. App. 2018) (referring to the statutory alternatives of first-degree assault as “variet[ies]”); Wright, 282 A.3d at 287 (referring to the statutory “modalit[ies]”). Ultimately, the State’s actions in charging first-degree assault, as well as the trial courts’ actions in instructing juries, matter far more than the mere words the appellate courts use in describing the offense. The “focal point of the [divisibility] analysis” is “what the jury must find (or a defendant must admit) to convict.” Muwwakkil, 983 F.3d at 755 (emphasis added). And Maryland juries

19

are plainly not required to agree unanimously that the defendant violated the statute pursuant to a particular modality of the statute.

For the same reason, we reject the Government’s argument that Dixon’s characterization of the first-degree-assault statute as a “multi-purpose criminal statute” is dispositive of the divisibility question. Dixon, 772 A.2d at 303. True, the court indicated that the first-degree-assault statute is a “multi-purpose” statute for purposes of a merger analysis, which requires courts to treat the alternatives of the offense “as it would treat separate statutes.” Id. (quoting Nightingale v. State, 542 A.2d 373, 376 (Md. 1988)). But the prosecutor in Dixon “did not . . . mention either modality of the assault statute,” id. at 306, the jury was instructed that it “could find petitioner guilty of first degree assault under either modality” without being instructed that it needed to unanimously agree, id. at 305, and the jury “returned a general verdict,” id. at 306. And the Dixon court did not find any of those things, in and of themselves, to render the conviction defective.

Thus, the statutory language, the decisions of Maryland courts, and the State’s charging practices strongly indicate that the Maryland first-degree-assault statute is indivisible.

2.

To the extent doubt remains regarding whether a statute is divisible or indivisible after looking at the statutory text and the decisions of the Maryland courts, Mathis indicates that we may “peek” at the record to determine whether the alternative modalities of the statute “[are] element[s] of the offense.” Mathis, 579 U.S. at 518–19 (internal quotation marks omitted). But we need not do so here, because the statutory text and the decisions of

20

the Maryland courts make clear that the Maryland first-degree-assault statute is indivisible. 13 Thus, having found no “clear signals” that the statute is divisible, Hope, 28 F.4th at 503–04 (quoting Najera-Rodriguez, 926 F.3d at 356), and multiple signals that the statute is indivisible, we conclude that Maryland first-degree assault is an indivisible offense.

D.

Because the Maryland first-degree-assault statute is indivisible, we apply the categorical approach to the entire statute, asking whether any of its alternative modalities can be committed in a way that does not constitute a violent offense. Redd argues that Maryland first-degree assault is not a categorically violent felony because the second of its modalities—“assault with a firearm”—can be committed with a mens rea of recklessness. 14 And a criminal offense that can be committed with a mens rea of recklessness—“a less culpable mental state than purpose or knowledge”—cannot qualify as a “violent felony” under ACCA. Borden, 141 S. Ct. at 1821–22.

Our analysis proceeds in three steps. First, “we look for the least culpable conduct”

encompassed by the “assault” element of the Maryland first-degree-assault statute.

13

The Government moved to supplement the Joint Appendix to provide what it characterizes as the charging documents from Redd’s 1998 and 2000 first-degree-assault convictions. Redd objects, arguing that the documents are incomplete and that some of the documents are not proper Shepard materials. We deny the Government’s motion and remind the Government that the proper setting for developing the record is the district court, not this Court in the first instance. E.g., Vann, 660 F.3d at 775 n.6.

14

Redd also contends that the Maryland first-degree-assault statute does not meet ACCA’s violent-felony definition because the crime can be committed with merely de minimis force. We need not reach this argument because we agree with Redd that the crime in question can be committed with a mens rea of recklessness.

21

Proctor, 28 F.4th at 545–46 (quoting United States v. Middleton, 883 F.3d 485, 489 (4th Cir. 2018)). Second, we ask whether an “added mens rea element” of the first-degree- assault statute itself “circumscribes the reach of the statute” such that it only punishes conduct conforming to ACCA’s definition of a violent felony. Id. at 546. Finally, we assess “whether there is a ‘realistic probability’ that Maryland would have punished the minimally culpable conduct that can satisfy both the assault and intent elements.” Id.

1.

To start, we must identify the “least culpable conduct” encompassed by the “assault”

element of the Maryland statute. Id. at 545–46. For an individual to be convicted of Maryland first-degree assault, “the State must prove all the elements of assault in the second-degree, and, to elevate the offense to first-degree, at least one of the statutory aggravating factors.” Snyder, 63 A.3d at 134. Thus, to secure a conviction under the firearm modality of the statute, the State must prove all the elements of second-degree assault, and also must prove that the assault was committed with a firearm. Id.

Maryland second-degree assault, in turn, “encompasses three types of common law assault and battery: (1) the ‘intent to frighten’ assault, (2) attempted battery and (3) battery.” Id. Assault and battery were common-law crimes in Maryland until they were codified in 1996. However, even after they were replaced with statutory first- and second- degree assault, the crimes “retain[ed] their judicially determined meanings.” Proctor, 28 F.4th at 547 n.8 (quoting Snyder, 63 A.3d at 135); see also Md. Code, Crim. Law § 3- 201(b). Those meanings had been “extensively developed at common law.” Watts v. State, 179 A.3d 929, 936 (Md. 2018).

22

At common law, battery—the third type of Maryland second-degree assault—was defined as the “unlawful application of force to the person of another,” which could be “intentional or unintentional.” Elias v. State, 661 A.2d 702, 709 (Md. 1995) (quoting Epps v. State, 634 A.2d 20, 23 (Md. 1993)). Indeed, an unintentional battery could “arise from contact that [was] the result of a person’s criminal negligence that legally cause[d] injury to another.” Id. (emphasis added). And although unintentional battery at common law required “more . . . than ordinary negligence sufficient to support a civil action,” a defendant could be culpable for unintentional battery if his conduct was “so reckless as to constitute a gross departure from the standard of conduct that a law-abiding person would observe.” Id. (emphases added) (first quoting Duckworth v. State, 594 A.2d 109, 113 (Md. 1991); and then quoting State v. Albrecht, 649 A.2d 336, 348 (Md. 1994)). Today’s second- degree-assault statute encompasses this common-law understanding of unintentional battery. See Md. Code, Crim. Law § 3-201(b).

Tying it all together: Maryland first-degree assault is simply second-degree assault plus a statutorily defined aggravating factor, such as the use of a firearm. And Maryland second-degree assault encompasses unintentional battery, which can be committed with a mens rea of recklessness. So, if an individual commits a reckless battery with a firearm, he has committed first-degree assault. 15

15

This Court has previously found that several other states’ assault-with-a-deadlyweapon statutes can be violated by reckless conduct. See United States v. Simmons, 917 F.3d 312, 321 (4th Cir. 2019) (North Carolina statute), as amended (Mar. 6, 2019); United States v. Barcenas-Yanez, 826 F.3d 752, 756–57 (4th Cir. 2016) (Texas statute); Garcia v. Gonzales, 455 F.3d 465, 468 (4th Cir. 2006) (New York statute). Other circuits have

23

2.

Having established that Maryland battery includes reckless conduct, we must next ask whether any “added mens rea element” of the first-degree-assault statute “circumscribes [its] reach” such that it only punishes conduct conforming to ACCA’s definition of a violent felony. Proctor, 28 F.4th at 546.

The firearm modality of the statute provides that “[a] person may not commit an assault with a firearm.” Md. Code, Art. 27 § 12A-1(a)(2). On its face, nothing in this text circumscribes the reach of the statute to exclude reckless conduct. The Government argues, however, that the statute’s reference to a firearm means we can infer an intent-to-cause- harm element. We disagree.

The Government cites two Maryland appellate-court cases for the proposition that the reference to a firearm in the statute allows us to infer intent to cause physical injury: Jenkins v. State, 806 A.2d 682 (Md. Ct. Spec. App. 2002), and Jones v. State, 114 A.3d 256 (Md. Ct. Spec. App. 2015). But both opinions have since been vacated by the Supreme Court of Maryland. See Jenkins v. State, 825 A.2d 1008 (Md. 2003); Jones v. State, 155 A.3d 492 (Md. 2017). Consequently, we give them no weight.

Nor are we persuaded by the Government’s citation to cases indicating that “specific intent to harm may be inferred from the pointing of a firearm at the body of a victim.”

reached similar conclusions about similar statutes. See United States v. Garner, 28 F.4th 678, 683–84 (5th Cir. 2022) (Louisiana statute); United States v. Rose, 896 F.3d 104, 114 (1st Cir. 2018) (Rhode Island statute); United States v. Kennedy, 881 F.3d 14, 19 (1st Cir. 2018) (Massachusetts statute).

24

Government’s Br. at 24; e.g., State v. Raines, 606 A.2d 265, 269 (Md. 1992) (instructing that “the trier of fact may infer the intent to kill from the surrounding circumstances,” considering “the accused’s acts, conduct and words,” such as “the use of a deadly weapon directed at a vital part of the human body”). This kind of fact-based assessment of how the accused used a firearm in committing the offense is at odds with our task under the categorical approach, which does not permit us to look at the “facts underlying the prior convictions.” Taylor, 495 U.S. at 600. Further, the cited cases did not instruct that courts may infer intent to kill from a mere assault with a firearm; they addressed “the use of a deadly weapon directed at a vital part of the human body,” Raines, 606 A.2d at 269, “the pointing of a gun toward another human and discharging it in random fashion,” Hoes v. State, 368 A.2d 1080, 1088 (Md. Ct. Spec. App. 1977), and “the use of deadly force,” Evans v. State, 349 A.2d 300, 340 (Md. Ct. Spec. App. 1975). None of these components are necessary elements of first-degree assault with a firearm. 16

16

The Government also argues that Maryland courts have held that first-degree assault is a specific-intent crime, so it may not be committed recklessly. But a closer look at the cases cited by the Government makes clear that the courts were addressing the first modality of first-degree assault—“intentionally caus[ing] or attempt[ing] to cause serious physical injury to another,” Md. Code, Art. 27 § 12A-1(a)(1) (emphasis added)—which plainly requires specific intent. Take Chilcoat v. State. There, the Maryland Court of Special Appeals stated that “the State must prove that an individual had a specific intent to cause a serious physical injury.” Chilcoat v. State, 843 A.2d 240, 246 (Md. Ct. Spec. App. 2004). But the surrounding context makes clear that the court was discussing only the first modality of the first-degree-assault statute. Id. (“[T]he statute prohibits not only causing, but attempting to cause, a serious physical injury to another. Although the State must prove that an individual had a specific intent to cause a serious physical injury, a jury may infer the necessary intent from an individual’s conduct and the surrounding circumstances . . . .” (citations omitted)). To avoid “mistakenly cast[ing] doubt on . . . state criminal prosecutions” by imputing additional elements not recognized by the state courts, we will

25

The Maryland Criminal Pattern Jury Instructions bolster this conclusion. When a defendant is charged with only first-degree assault, and when the “theory of assault is a battery” that “involves a firearm,” Md. Crim. Pattern Jury Instructions 4:01.1B, Notes on Use, the jury should be instructed that the State must prove that the defendant caused offensive physical contact to the victim that was “the result of an intentional or reckless act,” Md. Crim. Pattern Jury Instructions 4:01.1B(C) (alteration adopted) (emphasis added). The comment to this jury instruction makes this even clearer: “[R]eckless physical contact can only constitute a first degree assault if it involves a firearm. Serious-physical- injury and strangulation type first degree assaults require specific intent.” 17 Id., Comment. And we find the pattern jury instructions especially persuasive because Maryland courts place great weight on these instructions. Johnson v. State, 115 A.3d 668, 682 (Md. Ct. Spec. App. 2015) (explaining that Maryland appellate courts strongly favor the trial court’s use of pattern jury instructions).

not rely on isolated statements from the Maryland courts devoid of their context. Najera- Rodriguez, 926 F.3d at 356.

17

The fact that some—but not all—modalities of the first-degree-assault statute require intent might be read to suggest that the statute is divisible; indeed, it suggests that that the elements the State must prove vary based on which modality the defendant is accused of violating. But any suggestion of divisibility that we might find here cannot overcome the dispositive fact that juries are not required to unanimously agree as to the modality. See Descamps, 570 U.S. at 272 (explaining that, when a statute is divisible, the prosecutor “must generally select the relevant element from its list of alternatives,” and “the jury . . . must then find that element unanimously and beyond a reasonable doubt”); Mathis, 579 U.S. at 506 (explaining that a jury need not unanimously agree as to the means by which defendant committed an indivisible offense).

26

We similarly reject the Government’s argument that the canon of noscitur a sociis 18 suggests we should read an intent-to-harm requirement into the firearm modality. Contrary to the Government’s assertion that it “would make no sense” to include a crime that can be committed with a mens rea of recklessness in a statute that otherwise covers intentional crimes, Government’s Br. at 26; see Md. Code, Crim. Law § 3-202(b)(1) (“A person may not intentionally cause or attempt to cause serious physical injury to another.”), (b)(3) (“A person may not commit an assault by intentionally strangling another.”), we conclude that it would make little sense to specify that the first and third modalities—but not the second—require intentional conduct, if the legislature had in fact intended for all three modalities to require intent. 19

18

Under the canon of noscitur a sociis, statutory terms are “known by the company they keep.” Lagos v. United States, 138 S. Ct. 1684, 1688–89 (2018).

19

The Board of Immigration Appeals recently reached the same conclusion—that the firearm modality of Maryland first-degree assault can be committed with a mens rea of recklessness—in an unpublished decision based on substantially similar reasoning as that which we set out above, and therefore concluded that the firearm modality is not categorically a “crime of violence” under 18 U.S.C. § 16. M.T.B., AXXX-XXX-564 (B.I.A. Mar. 8, 2023) (saved as ECF opinion attachment); see Johnson I, 559 U.S. at 140 (citing a case concerning the “crime of violence” definition in 18 U.S.C. § 16 in interpreting the “violent felony” definition in ACCA). And although two of our sister circuits have reached the opposite conclusion regarding the Maryland first-degree-assault statute—one in an unpublished opinion—both of those cases were decided prior to Borden, which held that a crime did not satisfy the violent-felony definition if it could be committed with a mens rea of recklessness, and therefore are of limited persuasive value to our decision today. See United States v. Haight, 892 F.3d 1271, 1281–82 (D.C. Cir. 2018) (holding that “the use of violent force includes the reckless use of such force” under ACCA and concluding that Maryland first-degree assault is categorically a violent felony); United States v. Zavala Romero, 742 F. App’x 21, 22 (5th Cir. 2018) (per curiam) (finding no plain error in the conclusion that Maryland first-degree assault is a “crime of violence” under the Sentencing Guidelines).

27

3.

Having concluded that the Maryland first-degree-assault statute, as interpreted by Maryland courts, extends to reckless conduct for the firearm modality, we assess “whether there is a ‘realistic probability’ that Maryland would have punished [that] minimally culpable conduct.” Proctor, 28 F.4th at 546.

Finding a state statute to fall outside the ACCA violent-felony definition requires “more than the application of legal imagination to a state statute’s language.” Gonzales v. Duenas-Alvarez, 549 U.S. 183, 193 (2007). Instead, it requires “a realistic probability, not a theoretical possibility, that the State would apply its statute to conduct that falls outside” the federal definition. Id. “To show that realistic probability,” the defendant must point to a case “in which the state courts in fact did apply the statute” in the manner the defendant contends they might. Id.; see also United States v. Battle, 927 F.3d 160, 164 (4th Cir. 2019) (applying the realistic-probability test).

But as we have held—and as the Supreme Court clarified last year—a defendant’s failure to identify a case in which the state actually prosecuted the conduct that the defendant alleges they could have “is by no means dispositive” in determining whether a state statute meets the ACCA violent-felony definition. Proctor, 28 F.4th at 552; accord United States v. Taylor, 142 S. Ct. 2015, 2024 (2022). Rather, where “a state statute plainly covers” the conduct that the defendant alleges it does, we will not fault the defendant for failing to identify a case in which the state successfully prosecuted the conduct. Proctor, 28 F.4th at 552; cf. Taylor, 142 S. Ct. at 2024–25 (looking to whether the “elements of the relevant state and federal offenses . . . overlapped”). Nor will we hold against the defendant

28

a failure to identify a successful prosecution where “state courts have clearly greenlit” the application of the statute proposed by the defendant. Proctor, 28 F.4th at 553; accord United States v. Aparicio-Soria, 740 F.3d 152, 158 (4th Cir. 2014) (en banc) (“We do not need to hypothesize about whether there is a ‘realistic probability’ that Maryland prosecutors will charge [the conduct in question]; we know that they can because the state’s highest court has said so.”); cf. Taylor, 142 S. Ct. at 2025 (noting that in Gonzales, “[a]ppreciating the respect due [to] state courts as the final arbiters of state law in our federal system, th[e Supreme] Court reasoned that it made sense to consult how a state court would interpret its own State’s laws”).

Although Redd has not identified a specific successful prosecution for Maryland first-degree assault with a firearm based on reckless conduct, we do not have to squint to discern how the statute might cover such conduct; indeed, it is quite plain from Maryland’s statutory scheme, and from the case law interpreting it, that reckless conduct is included. Therefore, Redd’s inability to point to a specific case is not dispositive. See Proctor, 28 F.4th at 552. Rather, the terms of the statute and the decisions of Maryland courts show that a Maryland prosecutor could bring charges for first-degree assault against a defendant for recklessly committing an assault with a firearm.

III.

We conclude that the Maryland first-degree-assault statute sets out an indivisible offense and that one of the modalities of that offense—assault with a firearm—can be committed with a mens rea of recklessness. Because reckless conduct does not satisfy

29

ACCA’s definition of “violent felony,” Maryland first-degree assault cannot serve as an ACCA predicate crime. 20 Based on the foregoing, we reverse the district court’s denial of Redd’s § 2255 motion, vacate Redd’s ACCA sentence, and remand to the district court for further proceedings. See Proctor, 28 F.4th at 553 (where district court relied on an invalid ACCA predicate, reversing denial of § 2255 motion and remanding with instructions that the district court grant the motion, vacate and set aside the judgment, and correct the sentence).

REVERSED AND REMANDED

20

Our conclusion accords with that of at least one sister circuit that has applied Borden to find that a state statute criminalizing assault “committed with a firearm” did not satisfy the Sentencing Guidelines’ definition of a “crime of violence” because it could be committed with a mens rea of recklessness. Garner, 28 F.4th at 683–84 (5th Circuit, analyzing Louisiana statute); see id. at 682 (explaining that the relevant definition of “crime of violence” contained “identical operative language” to ACCA’s force clause).

30

QUATTLEBAUM, Circuit Judge, concurring:

I concur in the majority’s conclusion that we must reverse the district court’s denial of Redd’s § 2255 motion, vacate Redd’s ACCA sentence and remand for further proceedings. But I get there following a slightly different path. And I write separately to explain that alternative route.

I.

First, I address whether Maryland first-degree assault is a divisible or an indivisible statute. The majority makes several reasonable arguments in concluding it is indivisible. But a number of Maryland cases suggest the statute is divisible. In Marlin v. State, 993 A.2d 1141 (Md. Ct. Spec. App. 2010), the court commented that the two modalities of first- degree assault are “regarded as distinct statutes.” Id. at 1155. And in Dickerson v. State, 40 A.3d 1122 (Md. Ct. Spec. App. 2012), the court explained,

In the second of the aggravating modalities, the handgun, or other firearm, is a legal element of the crime itself. In this case, the acquittal’s negating of the element of using a handgun would be legally inconsistent with the necessary presence of such an element for a first-degree conviction based upon the use of such a weapon. If the first-degree assault conviction were necessarily based on this aggravating factor, it would, indeed, be fatally legally inconsistent according to Price.

For the other aggravating modality of first-degree assault, however, there is no required element involving a handgun, or any firearm. The aggravation is all in the mens rea of the assault. In an assault based on this theory of aggravation, there may well have been factual or logical inconsistency but no legal inconsistency. There was no necessary finding of a legal element which the acquittal had removed from the case.

31

Id. at 1125 (emphasis added). And as the majority notes, Dixon v. State, 772 A.2d 283 (Md. 2001), refers to § 12A-1 as a “multi-purpose criminal statute” with “separate offense[s]” and urges trial judges to use special interrogatories so the jury can clarify which offense it decides. Id. at 303 n.33 (stating “[w]e have recommended, and continue to recommend . . . a special verdict” so that the record is clear on which modality the jury found the defendant guilty).

Also, parts of the Maryland Criminal Pattern Jury Instructions point towards divisibility. True, the instructions indicate that for first-degree assault, juries are not always instructed that they must unanimously agree as to the modality of which the defendant is guilty. Maj. Op. 17. But the commentary to those instructions provides that the trial court should require a special verdict form and that the jury should provide a finding on whether the defendant is guilty of (a)(1), (a)(2) or both when a defendant is charged with both first- degree assault and the use of a firearm in the commission of a felony or crime of violence. 1 Md. Crim. Pattern Jury Instructions 4:01.1A, Notes on Use. Further, the commentary directs the court to charge the jury on only the modality “for which a prima facie case has been established.” Id.

So Maryland law is not all at clear on the question of whether Md. Code, Art. 27 § 12A-1 (1996) points more towards divisibility than to indivisibility.

1

Both first and second-degree assault are included in Maryland’s definition of crime of violence. Md. Code Ann., Public Safety § 5-101(c)(3).

32

Normally, in the case of a close call, we treat the statute as indivisible. United States v. Hope, 28 F.4th 487, 503–04 (4th Cir. 2022) (explaining that to find that a state statute is divisible, we must discern “clear signals” indicating as much— “signals that convince us to a certainty that the elements are correct and support divisibility”). In other words, a tie goes to indivisibility. But here, I question the wisdom of making this decision in this case. Even if we find the statute divisible, the Shepard 2 documents are not part of the record on appeal. And without any such documents to provide clarity as to whether Redd was charged under a particular modality of first-degree assault, we end up applying the categorical approach anyway. Said differently, applying the tie-goes-to-indivisibility rule does not change the outcome of this case. But it would establish that the statute is indivisible going forward in the face of what we all seem to agree is ambiguous Maryland law. I would hold off making that decision until we have to—in other words, when it makes a difference. By that time, Maryland may have brought clarity to this murky issue.

II.

As to whether Maryland first-degree assault can be committed recklessly, I join the majority’s opinion, which ultimately concludes that it can. As the majority points out, to convict a defendant of Maryland first-degree assault, “the State must prove all the elements

2

In Shepard v. United States, 544 U.S. 13, 16 (2005), the Supreme Court defined such documents to include the “charging document, written plea agreement, transcript of plea colloquy, and any explicit factual finding by the trial judge to which the defendant assented.”

33

of assault in the second-degree, and, to elevate the offense to first-degree, at least one of the statutory aggravating factors.” Snyder v. State, 63 A.3d 128, 134 (Md. Ct. Spec. App. 2013). As to second-degree assault, the state must only prove that the defendant acted with “criminal negligence,” which Maryland courts have explained includes reckless conduct. Elias v. State, 661 A.2d 702, 709 (Md. 1995) (quoting State v. Albrecht, 649 A.2d 336, 348 (Md. 1994)). At the time Redd pled guilty, there were only two statutory aggravating factors—intent to cause or attempt to cause serious physical injury or assault committed with a firearm. As noted above, Dickerson tells us that the former requires an elevated mens rea—something more than “criminal negligence”—while the latter does not. 40 A.3d at 1125. The majority combines these principles of Maryland law to suggest Maryland first- degree assault by using a firearm can be committed recklessly.

To me, this analysis seems a bit mechanical. After all, it seems at least plausible that the added requirement of committing an assault with a firearm might imply intentionality. So, were this all there was to go on, I might resist this reasoning. But there is more. Under the Maryland Criminal Pattern Jury Instructions, when a defendant is charged with first- degree assault under subsection (a)(2), the state must prove physical contact with the victim that was the “the result of an intentional or reckless act.” Md. Crim. Pattern Jury Instructions 4:01.1B(C). And the comments to those instructions are consistent. See Md. Crim. Pattern Jury Instructions 4:01.1B, Comment (“One potentially confusing consequence of this scheme is that it necessarily must include the possibility of a ‘reckless act’ causing offensive physical contact or harm under the element (2) of instruction ‘C’ (Battery). Importantly, reckless physical contact can only constitute a first degree assault

34

if it involves a firearm. Serious-physical-injury and strangulation type first degree assaults require specific intent.”). What’s more, the pattern jury instructions are strongly favored by Maryland courts. Johnson v. State, 115 A.3d 668, 682 (Md. Ct. Spec. App. 2015). Given this additional evidence from Maryland law that first-degree assault with a firearm can be committed recklessly, federalism compels that I respect what Maryland has to say on this issue. As a result, I join in the majority’s conclusion that Maryland first-degree assault with a firearm can be committed recklessly and thus does not qualify under the force clause of the ACCA.

But only reluctantly. Frankly, I cannot come up with a scenario where one commits an assault with a firearm that is not the result of an intentional act. And Redd has not identified an actual specific prosecution of first-degree assault with a firearm based on reckless conduct. 3 In fact, Redd’s hypothetical example of a gun discharging and injuring a bystander while being mishandled would seem to be accidental, which the Maryland Criminal Pattern Jury Instructions say is not sufficient for first-degree assault. Md. Crim. Pattern Jury Instructions 4:01.1B(C) (“In order to prove that the defendant committed [first-degree assault], the State must prove . . . that the contact was the result of an intentional [or reckless] act of the defendant and was not accidental[.]”) (third alteration in

3

I recognize this failure is not dispositive. United States v. Proctor, 28 F.4th 538, 544, 552 (4th Cir. 2022).

35

original). Even so, on “the Alice in Wonderland path known as the ‘categorical approach,’” 4 that is not enough to affirm the district court.

4

United States v. Battle, 927 F.3d 160, 163 n.2 (4th Cir. 2019).