United States v. Carr

Full title: UNITED STATES OF AMERICA, Plaintiff-Appellee, v. ANTHONY CARR…

Court: United States Court of Appeals, Seventh Circuit

Date published: Jul 8, 2024

No. 22-1245 (7th Cir. Jul. 8, 2024)

Opinion

22-1245

07-08-2024

UNITED STATES OF AMERICA, Plaintiff-Appellee, v. ANTHONY CARR, Defendant-Appellant.

ROVNER, CIRCUIT JUDGE

Argued April 20, 2023

Appeal from the United States District Court for the Northern District of Illinois, Eastern Division. No. 1:20-cr-00559-1 - Charles R. Norgle, Sr., Judge.

Before Easterbrook, Rovner, and St. Eve, Circuit Judges.

ROVNER, CIRCUIT JUDGE

Anthony Carr's offense level and sentencing range for being a felon in possession of a firearm were increased substantially in view of his three prior convictions in Illinois for armed robbery, which the district court treated as a crime of violence. See U.S.S.G. §§ 2K2.1(a)(1), 4B1.2(a)(2). Carr contends that this was error. The Sentencing Guidelines identify robbery as a crime of violence, § 4B1.2(a)(2), and Carr does not dispute that the elements of Illinois robbery, including armed robbery, line up with the elements of generic robbery, such that a conviction for Illinois robbery qualifies as a crime of violence for sentencing purposes under the categorical framework established by Taylor v. United States, 495 U.S. 575 (1990). Thus, assuming that Carr was convicted of robbery as a principal, there would be no doubt that his conviction was one for a crime of violence.

Carr's contention that he was not convicted of a crime of violence hinges on the theoretical possibility that he could have been convicted as an aider and abettor to robbery rather than a principal. Although Carr was not charged as an accomplice in any of the three Illinois robberies, accomplice liability is implicit in every Illinois criminal charge and a defendant can thus be convicted as an aider and abettor even if he is not expressly charged as such.

Illinois has a form of aiding and abetting liability known as common design, which holds an accomplice liable for any criminal act committed in furtherance of the target offense the accomplice intended to aid and abet, even if that secondary criminal act was one that the accomplice did not foresee or agree to. See Monroe v. Davis, 712 F.3d 1106, 1120 (7th Cir. 2013) (describing the Illinois rule and collecting cases). So, for example, if the accused aids and abets the burglary of a home while the owners are out, and during the burglary the owners arrive home unexpectedly, at which point one of the accused's cohorts points a gun at the owners and demands that they hand over their money, smartphones, and jewelry, the accused will be held liable for armed robbery (in addition to burglary) even if he did not anticipate or intend to aid a robbery and did not know that his cohort was armed.

Invoking Gonzales v. Duenas-Alvarez, 549 U.S. 183 (2007), Carr contends that Illinois' common-design rule of accomplice liability is out of step with comparable accomplice-liability rules in other jurisdictions, in that the rule as articulated by Illinois courts does not require that a secondary criminal act be foreseeable to the aider and abettor; other states, by contrast, expressly limit an aider and abettor's liability to secondary criminal acts that are reasonably foreseeable to him. Given the breadth of the Illinois rule, and that accomplice liability is implicit in every Illinois charged offense, including robbery, Carr contends that Illinois robbery must be deemed a categorical mismatch for generic robbery, such that federal courts cannot treat Illinois robbery as a crime of violence for sentencing purposes.

We reject the argument. Each of Carr's Illinois convictions is for armed robbery, and the pertinent elements of that offense are a match for generic robbery. The abstract possibility that Carr could have been convicted as an aider and abettor rather than a principal is neither here nor there: Illinois, like every other state, does not recognize a distinction between principals and accomplices. Although it is true that Illinois articulates its "common-design" iteration of accomplice liability in broad terms, we are not convinced that Illinois is an outlier in the way that it applies common-design liability. Duenas-Alvarez requires Carr to establish a "realistic probability" that Illinois would hold a defendant liable for a crime as an accomplice when most other states would not, 549 U.S. at 193, and in this respect, what Illinois courts do is more important than what they say. As we shall see, the results of Illinois common-design cases are consistent with those in jurisdictions that nominally apply narrower accomplice liability principles. Further, however broad Illinois common-design liability may be, liability as an accomplice is not an element of the offense of robbery, and as such it is not something that factors into the Taylor categorical analysis as the Supreme Court has framed and applied that analysis to date. Duenas-Alvarez hints that a state's accomplice liability might be so "special" as to require different treatment of that state's convictions in federal court; but it has yet to give lower courts guidance as to precisely how we should factor a state's "special" accomplice liability into Taylor's categorical framework. Carr's position is that a state's unusually broad formulation of accomplice liability will necessarily disqualify any robbery conviction (and quite probably, any other conviction) from that state as a basis for an enhanced federal penalty-be it an increase in the Guidelines sentencing range, an increased statutory minimum or maximum prison term, or in the case of an immigrant, removal from the United States. That aberrant outcome cannot be the result that Congress, the Sentencing Commission, or the Supreme Court intended.

Carr makes two other challenges to his sentence. One of them requires a remand to the district court; the other fails on plain-error review.

I.

On July 8, 2020, Carr was at a block party in Chicago when police officers arrived and began patting down guests for contraband. Carr attempted to flee from the officers but was unsuccessful: he was discovered to be in possession of a loaded semi-automatic pistol with an extended magazine with the capacity to hold 34 rounds of ammunition. Carr had previously been convicted of a felony, which made it unlawful for him to possess a firearm as a matter of both federal and state law. See 18 U.S.C. § 922(g)(1); 720 ILCS 5/24-1.1(a) and 5/24-1.7(a). Carr was initially held in the Cook County Correctional Center on state charges. He was on parole from his prior Illinois convictions for armed robbery at the time of his arrest, and because his gun possession was in violation of the terms of his parole, his parole was revoked and he was transferred to the custody of the Illinois Department of Corrections on September 1, 2020, for the remainder of his parole term.

On September 3, 2020, a federal grand jury returned an indictment charging Carr with the unlawful possession of a firearm by a felon, in violation of section 922(g)(1). Local authorities promptly dismissed the pending state charges against Carr, and pursuant to a writ of habeas corpus ad prosequendum, Carr was taken into federal custody on September 28, 2020, for purposes of prosecution on the felon-in-possession charge. However, while Carr was in federal custody, an Illinois Department of Corrections detainer remained in place until Carr completed his (revoked) term of parole on the armed robbery convictions on October 19, 2021.

Carr pleaded guilty to the felon-in-possession offense in June 2021. The record before the court at sentencing, which includes both the federal pre-sentence investigation report prepared by the probation officer and the plea colloquy underlying the three armed robbery convictions in state court, details the facts underlying the armed robbery convictions.

On May 20, 2007, Carr and two of his friends, with the aim of obtaining gas money, perpetrated an armed robbery on two individuals sitting in a pickup truck: the victims were ordered out of the truck at gunpoint and relieved of their cash, a cell phone, and other personal items. Carr played a central role in the crime: he and one of his accomplices had obtained the guns used in the attempted robbery from a friend, Carr made the initial approach to the victims, he drove the getaway car, and he stored the guns at his house after the incident. R. 61 at 9-10 ¶ 32; R. 58-4 at 7-10.

On May 6, 2008, Carr committed two armed robberies without accomplices. In the first, he robbed a victim at gunpoint of $4 and a cellphone. In the second, he robbed a different victim, again at gunpoint, of a coin purse. R. 61 at 10-11 ¶¶ 33-34.

In April 2010, Carr was sentenced in state court for all three robberies, to which he had pleaded guilty. He was ordered to serve a 12-year prison term for the 2007 robbery and two 12-year terms for the 2008 robberies, with the latter terms running concurrently with one another but consecutively to the term imposed for the 2007 robbery, for an aggregate prison term of 24 years. Carr was out on parole from that prison term in 2020 when he was arrested with the loaded semi-automatic pistol underlying the federal charge in this case.

For federal sentencing purposes, the pre-sentence report treated each of these three Illinois convictions for armed robbery as a "crime of violence" pursuant to sections 2K2.1(a)(2) and 4B1.2(a) of the Sentencing Guidelines. R. 61 at 7-8 ¶ 15. That characterization had the effect of increasing Carr's base offense level by two levels, from 24 to 26. Following a two-level enhancement for reckless endangerment based on Carr's effort to flee the police, see § 3C1.2,¹ and a three-level reduction for acceptance of responsibility, see § 3E1.1, Carr's total, adjusted offense level was 25. Coupled with a category IV criminal history, the advisory sentencing range was 84 to 105 months in prison. Had the base offense level not been increased based on the robbery convictions, Carr's total offense level would have been 23, and the resulting advisory sentencing range would have been 70 to 87 months.

1 Carr challenged that enhancement below but does not raise the issue here.

Carr objected to the treatment of his prior armed robbery convictions, arguing that they were not "crimes of violence." R. 45, 56. Carr's argument, as we discussed at the outset of this opinion, was premised on the breadth of accomplice liability in Illinois. Although each of Carr's Illinois convictions was for armed robbery, and there was no indication that Carr had been charged or convicted as anything but a principal for any of the robberies, Carr postulated that one charged with armed robbery in Illinois could be convicted of that offense as an accomplice not because he agreed to commit robbery-or the use of a firearm or other dangerous weapon to commit the robbery-but rather because the armed robbery was committed in furtherance of a separate crime to which he had agreed (in our prior example, residential burglary), however unexpected the robbery may have been to him. The breadth of Illinois accomplice liability, in Carr's view, places it outside of the mainstream of secondary criminal liability in the United States, such that a conviction for an Illinois offense-in this case, armed robbery-cannot be treated as a categorical match for its generic counterpart. See Duenas-Alvarez, 549 U.S. 183. So, although the Sentencing Guidelines treat robbery as a crime of violence, Carr argued that his Illinois convictions for (armed) robbery are a categorical mismatch for generic robbery and thus could not be treated as crimes of violence for purposes of enhancing his offense level.

The district court summarily rejected Carr's argument. The court briefly remarked:

The Seventh Circuit has dealt directly with the issue. And to the extent that there is any undecided aspect of the challenge, this Court overrules the defendant's objections and finds that he has been convicted of constitutionally supportable crimes of violence.

R. 70 at 11. The court adopted the PSR as relevant here and imposed a prison term at the bottom of the advisory sentencing range-84 months-to be followed by a three-year term of supervised release.

Carr also asked the court to reduce the sentence imposed by 15 months to reflect the time he had spent in pretrial and pre-sentencing detention from the date of his arrest by local authorities on July 8, 2020, until the expiration of his state parole term on October 19, 2021. Although Carr had nominally been in federal custody for most of that time period, he noted that the Bureau of Prisons would not credit that time against his sentence because, as a matter of law, he was still in the primary custody of Illinois as he served out the remainder of his revoked state parole term. Carr nonetheless proposed that a reduction in his sentence was warranted to reflect the fact that his initial arrest and detention by local authorities, and the revocation of his state parole, were based on the same conduct (his unlawful possession of a firearm) as his federal felon-in-possession conviction. The district court would not entertain that argument, emphasizing that the question of sentence credits was one exclusively for the Bureau of Prisons and that Carr would be credited for any time spent in federal custody. R. 70 at 23, 28. The court apparently did not appreciate Carr's argument that a substantial portion of his pre-sentencing detention would be characterized as state rather than federal custody, and that what Carr was asking for was a reduction in his sentence rather than a credit.

II.

The Sentencing Guideline applicable to the unlawful possession of a firearm specifies a base offense level of 26 if "the defendant committed any part of the instant offense subsequent to sustaining at least two felony convictions of ... a crime of violence" as defined in section 4B1.2 of the Guidelines. U.S.S.G. § 2K2.1(a)(1)(B) &comment. (n.1) (Nov. 2018). Section 4B1.2(a) in turn identifies two ways in which a prior offense can qualify as a "crime of violence": (1) the "elements clause" deems an offense to be a violent crime if it has as an element the use, attempted use, or threatened use of physical force against the person of another, whereas (2) the "enumerated offense clause" lists certain offenses which per se qualify as violent crimes, and those enumerated offenses include robbery.

Courts apply the categorical approach first delineated by the Supreme Court in Taylor, 495 U.S. at 602, to ascertain whether a given offense qualifies as a crime of violence for purposes of the Guidelines. E.g., United States v. Dixon, 27 F.4th 568, 570 (7th Cir 2022). When applying the enumerated offense clause of section 4B1.2(a)(2), this means that we compare the elements of the state statute forming the basis of the defendant's conviction with the elements of the generic offense (i.e., the offense as it is commonly understood). Descamps v. United States, 570 U.S. 254, 257 (2013); Taylor, 495 U.S. at 598, 602. Under this approach, the defendant's prior conviction under the state statute will qualify as a crime of violence only if the statute's elements are the same as, or narrower than, those of the generic offense. Descamps, 570 U.S. at 257; Taylor, 495 U.S. at 599 . We do not consider the facts underlying the defendant's conviction in applying the categorical approach. Id. at 600-02.

Carr was convicted of armed robbery in violation of section 18-2 of the Illinois criminal code. 720 ILCS 5/18-2(a). Section 18-1 outlines the offense of simple robbery: "A person commits robbery when he or she knowingly takes property .. from the person or presence of another by the use of force or by threatening the imminent use of force." 720 ILCS 5/18-1(a). Section 18-2 in turn specifies that robbery becomes armed robbery when, as relevant here, the perpetrator carries about his person, or is otherwise armed with, a firearm, 720 ILCS 5/18-2(a)(2), or a dangerous weapon other than a firearm, 720 ILCS 5/18-2(a)(1). Armed robbery thus incorporates the offense of simple robbery and adds an element: carrying or otherwise arming oneself with a firearm or other dangerous weapon.²

2 Although the record indicates that Carr used a firearm in each of the three armed robberies, as part of his plea agreement, the charges were amended to replace each reference to a firearm with the term "bludgeon." R. 58-4 at 2-3. Carr was thus convicted pursuant to section 18-2(a)(1) (armed robbery committed with a dangerous weapon other than a firearm), and the judgments of conviction each indicate that the offense of conviction was "ARMED ROBBERY/NO FIREARM." R. 58-1 at 12; R. 58-2 at 5; R. 58-3 at 5.

There is no dispute that Illinois robbery on its face is a categorical match for generic robbery, and thus constitutes a "crime of violence" under the enumerated offense clause of section 4B1.2(a)(2). "Robbery as ordinarily understood is the taking of property from a person by force or the threat of force ... ." United States v. Bedell, 981 F.2d 915, 915 (7th Cir. 1992). The elements of simple robbery in Illinois conform to the elements of generic robbery; the offense of armed robbery simply adds an element (carrying or otherwise possessing a firearm or other dangerous weapon during the robbery) which further narrows the offense. This satisfies the categorical inquiry, which has led us to conclude previously that Illinois armed robbery is a crime of violence under the enumerated offense clause of the guideline, and, for that matter, the elements clause as well. See United States v. Chagoya-Morales, 859 F.3d 411, 422 (7th Cir. 2017); Bedell, 981 F.2d at 915-16; see also Klikno v. United States, 928 F.3d 539, 544-47 (7th Cir. 2019) (Illinois robbery necessarily involves the use of physical force); United States v. Brown, 916 F.3d 706, 707-08 (8th Cir. 2019) (per curiam) (Illinois robbery) (citing Dembry v. United States, 914 F.3d 1185, 1187-88 (8th Cir. 2019)). To simplify matters going forward, we shall refer to "Illinois robbery" as we address Carr's contentions that his convictions for Illinois armed robbery are not a match for generic robbery.

A. Mental State of Recklessness

Carr suggests first that we revisit the foregoing holdings deeming Illinois robbery to be a crime of violence in light of Borden v. United States, 593 U.S. 420 (2021). Borden holds that a criminal offense with a mens rea of recklessness does not qualify as a "violent felony" under the elements clause of the Armed Career Criminal Act, 18 U.S.C. § 924(e). That act specifies that an offense qualifies as a violent felony if it "has as an element the use, attempted use, or threatened use of physical force against the person of another[.]" § 924(e)(2)(B)(i). The Court reasoned that the use of force "against another" "demands that the perpetrator direct his action at .. another individual," 593 U.S. at 429 (emphasis supplied), which bespeaks purposeful or knowing action and is therefore inconsistent with a mens rea of recklessness, id. at 429-34.

Prior to 2013, the Illinois robbery statute did not specify a mens rea necessary to commit the offense,³ and in the absence of a specified mental state the default rule in Illinois was that such offenses could be completed with either purpose, knowledge, or recklessness. See generally 720 ILCS 5/4-3 through 5/4-6; see also People v. Jones, 595 N.E.2d 1071, 1075 (Ill. 1992) ("[E]ither intent, knowledge or recklessness is an element of robbery even though the statutory definition of robbery does not expressly set forth a mental state."). This presents the possibility that in 2007 and 2008, when Carr engaged in multiple armed robberies, it might have been sufficient if his use of force to accomplish the robberies was merely reckless as opposed to knowing or intentional. See United States v. Love, 2023 WL 2546507, at *3 (7th Cir. Mar. 17, 2023) (non-precedential decision).

3 The 2007 version of section 18-1(a) provided: "A person commits robbery when he or she takes property . from the person or presence of another by the use of force or by threatening the use of force." The term "knowingly" was added to the statute in 2013. See P.A. 97-1108, § 10-5, effective Jan. 1, 2013.

However, Carr did not make this particular argument below, notwithstanding the fact that Borden was decided in June 2021, nearly eight months prior to his February 2022 sentencing. Carr therefore forfeited the argument, limiting our review to plain error. E.g., United States v. Acox, 595 F.3d 729, 730 (7th Cir. 2010). Among other elements, a showing of plain error requires that the error in question be, in retrospect, obvious. Henderson v. United States, 568 U.S. 266 (2013).

As we concluded in Love, it is not obvious that the crime of Illinois armed robbery prior to 2013 encompassed the reckless use of force. Although it is possible that the default minimum mental state of recklessness applied to both the taking element of robbery as well as the force element of the offense, it is also possible that the default minimum mental state of recklessness applied only to the former and not to the force element. 2023 WL 2546507, at *3-*4. The Illinois pattern instruction on robbery at that time was consistent with this possibility. Id. at *3 (citing ILLINOIS PATTERN JURY INSTRUCTIONS-CRIMINAL § 14.02). See also United States v. Brown, 74 F.4th 527, 532-33 (7th Cir. 2023) (finding that similar language in Illinois carjacking statute suggests that use or threat of force must be knowing or intentional), cert. denied, 144 s. Ct. 1019 (2024).

Given the possibility that Illinois required the use of force to be knowing or intentional at the time Carr was convicted, he cannot show that the district court plainly erred in treating his armed robbery convictions as crimes of violence in establishing his offense level for sentencing purposes. Id. at *4.

B. Illinois common-design accomplice liability

1. Duenas-Alvarez and conviction as an accomplice rather than a principal

The final judgments for each of Carr's three armed robbery convictions cite armed robbery in violation of 720 ILCs 5/18-2(a)(1) as the offense of conviction; no other statute is cited. see R. 58-1 at 12; R. 58-2 at 5; R. 58-3 at 5. We have said thus far that the elements of Illinois robbery are a categorical match for generic robbery, thus making the offense a crime of violence for purposes of Guidelines sections 4B1.2(a)(2) and 2K2.1(a)(1). Arguably, our analysis could end there: Carr was convicted on two or more occasions of robbery in Illinois, the Guidelines identify "robbery" as a crime of violence, and the elements of Illinois armed robbery line up with the elements of generic robbery. See United States v. Gamez, 77 F.4th 594, 599 (7th Cir. 2023) ("Having concluded that Indiana arson is a crime of violence under [the Armed Career Criminal Act], we could end our analysis there."); see also Alfred v. Garland, 64 F.4th 1025, 1051 (9th Cir. 2023) (en banc) (Collins, J., concurring in judgment in part and dissenting in part); United States v. Valdivia-Flores, 876 F.3d 1201, 1212 (9th Cir. 2017) (Rawlinson, J. dissenting), overruled by Alfred, 64 F.4th at 104748.

However, based on Duenas-Alvarez, Carr wants us to consider the possibility that he might have been convicted of robbery on an aiding and abetting theory. As discussed, his argument is based on the breadth of Illinois' common-design rule of accomplice liability as it is described by Illinois courts, which he believes places Illinois law outside of the norm for accomplice liability vis-a-vis the federal government and other states. So, to the extent that Carr theoretically might have been convicted as an aider and abettor rather than as a principal, he believes his convictions for robbery are not a match for generic robbery. Before we reach the merits of that argument, we must first review what the Court in Duenas-Alvarez did and did not hold and how we have applied Duenas-Alvarez to cases in which accomplice liability was at issue.

An individual's prior conviction for certain types of felony offenses, including crimes of violence, can subject him to various criminal and civil penalties, including increased statutory maximum and minimum terms of imprisonment, an increased sentencing level, or removal from the United States. See, e.g., 18 U.S.C. § 924(e) (Armed Career Criminal Act); 21 U.S.C. §§ 841(b) and 843(d) (penalties for controlled substances offenses); Guidelines § 2K2.1(a) (robbery guideline); 8 U.S.C. § 1227(a)(2)(A)(iii) (Immigration and Nationality Act) (subjecting immigrant to removal if he has committed an aggravated felony).

The principal question presented in Duenas-Alvarez was whether it makes a difference that the individual may have been convicted as an aider and abettor of the felony offense at issue-and thus may not have personally committed each element of the offense-rather than as a principal. The government was seeking to remove Duenas-Alvarez from the United States on the ground that he had committed a "theft offense," which is among the list of aggravated felonies in the INA which subject an immigrant to removal. See 8 U.S.C. § 1101(a)(43)(G). Duenas-Alvarez had been convicted of automobile theft in California pursuant to a statute that expressly imposed criminal liability on accessories and accomplices to such a theft as well as principals. Relying entirely on its prior decision in Penuliar v. Ashcroft, 435 F.3d 961 (9th Cir. 2006), j. vacated & remanded, 549 U.S. 1178 (2007), the Ninth Circuit answered this question in the negative. Duenas-Alvarez v. Gonzales, 176 Fed.Appx. 820 (2006). Penuliar, applying the categorical inquiry set forth in Taylor, examined the elements of California vehicle theft to see whether they were a match for generic theft, such that the conviction under the state statute would qualify as a "theft offense" for purposes of the INA. It concluded that they were not. Generic theft requires, among other things, that a defendant have taken or controlled another's property; and although the California vehicle theft statute included that element, by reaching aiders and abettors, it imposed liability on an individual who might have helped another person take or control someone else's car but did not do so himself. Without having personally taken or controlled another's property, the Ninth Circuit reasoned, a defendant has not committed generic theft, and therefore the California statute was a categorical mismatch for the type of "theft offense" referenced by the INA. 435 F.3d at 969-70. The Supreme Court disagreed. Although historically, the common law had drawn distinctions among the four categories of participants in a felony offense-first-degree principals, second-degree principals, accessories before the fact, and accessories after the fact-the Court noted that over time, American jurisdictions have eliminated any distinctions between the first three of these categories. "Indeed, every jurisdiction-all states and the Federal Government-has 'expressly abrogated the distinction' among principals and aiders and abettors who fall into the second and third categories." 549 U.S. at 189-90 (quoting 2 W. LaFave, SUBSTANTIVE CRIMINAL LAW § 13.1(e), at 333 (2d ed. 2003)). Consequently, it was appropriate to treat aiders and abettors of a theft as having committed theft for purposes of the federal statute. Id. Duenas-Alvarez's conviction for California theft thus rendered him removable regardless of whether he was convicted as a principal or as an aider and abettor.

Consistent with Duenas-Alvarez, we concluded in United States v. Worthen, 60 F.4th 1066 (7th Cir.), cert. denied, 144 S.Ct. 91 (2023), that an individual convicted of aiding and abetting a Hobbs Act robbery is guilty of committing a crime of violence, even if he did not personally use force or threaten to use force against the person or property of another-the element of Hobbs Act robbery that makes it a crime of violence. See 18 U.S.C. § 1951(b)(1). We noted that "[i]t is 'hornbook law' that convicting an aider and abettor first requires showing that the underlying crime (here, Hobbs Act robbery) 'was actually committed.'" 60 F.4th at 1069 (quoting United States v. Motley, 940 F.2d 1079, 1081) (7th Cir. 1991)). Moreover, the government must further show that the aider and abettor took some affirmative act in furtherance of the offense with the intent of facilitating the commission of that offense. Id. Once these showings are made, the aider and abettor is culpable for the offense on the same terms as the principal. Aiding and abetting an offense, we emphasized, is not a separate crime from the underlying offense itself:

[It] is instead an alternative theory of liability for the commission of the principal offense. Put more directly, "an aider and abettor of a Hobbs Act robbery necessarily commits all the elements of a principal Hobbs Act robbery," In re Colon, 826 F.3d 1301, 1305 (11th Cir. 2016). And because the principal offense of Hobbs Act robbery satisfies the force clause of § 924(c), aiding and abetting a Hobbs Act robbery qualifies as a crime of violence too. See id.

60 F.4th at 1069-70; see also Mwendapeke v. Garland, 87 F.4th 860, 866 (7th Cir. 2023) (complicity to Kentucky first-degree robbery is an aggravated felony crime of violence subjecting immigrant to removal); Gamez, 77 F.4th at 599 (aiding and abetting Indiana arson is a crime of violence for purposes of the Armed Career Criminal Act); United States v. Groce, 999 F.2d 1189, 1191-92 (7th Cir. 1993) (Wisconsin conviction for burglary as a party to a crime rather than as a principal qualifies as conviction for a violent felony for purposes of Armed Career Criminal Act).

What Carr argues here, however-invoking a second aspect of Duenas-Alvarez-is that Illinois has embraced a form of aiding and abetting liability that is well outside of the mainstream of accomplice liability. At this point, we should pause to say a few words about the different ways in which federal and state jurisdictions formulate aiding and abetting liability.

2. Accomplice liability in the United States: from narrow to broad

In the classic articulation of federal accessory liability under 18 U.S.C. § 2,⁴ the Second Circuit in United States v. Peoni, 100 F.2d 401, 402 (1938) (L. Hand, J.), said that an individual must "in some sort associate himself with the [criminal] venture, ... participate in it as in something that he wishes to bring about, [and] . seek[ ] by his action to make it succeed." This formulation is understood to require a shared intent between the principal and the accomplice as to the offense, in the sense that the accomplice is aware of the principal's criminal purpose and takes action to help the principal, with the desire that the principal will succeed. See Rosemond v. United States, 572 U.S. 65, 76-77 (2014); United States v. Wu, 668 F.3d 882, 885 (7th Cir. 2011); United States v. Zafiro, 945 F.2d 881, 887-88 (7th Cir. 1991), j. aff'd on other grounds, 506 U.S. 534 (1993); United States v. Beck, 615 F.2d 441, 448-49 (7th Cir. 1980). This is a relatively narrow formulation of accomplice liability, in that it holds the accomplice liable only for the intended or target crime, and not for a secondary offense that the principal might commit in pursuit of the target crime-to take our prior example, committing armed robbery in the course of an agreed-upon residential burglary-without the accomplice's knowledge and approval. See Commonwealth v. Hanright, 994 N.E.2d 363, 369-70 (Mass. 2013), overruled in part on other grounds by Commonwealth v. Brown, 81 N.E.3d 1173 (Mass. 2017); Wilson-Bey v. United States, 903 A.2d 818, 830-39 (D.C. 2006); State v. Carrasco, 946 P.2d 1075, 1079-80 (N.M. 1997).

4 Subsection (a) of the statute provides that "[w]hoever commits an offense against the United States or aids, abets, counsels, commands, induces or procures its commission, is punishable as a principal."

A much broader formulation of accomplice liability is found in the natural and probable consequences doctrine, which has been adopted in a minority of states. Under this formulation, if a principal commits a secondary crime in the course of carrying out the target crime, an aider and abettor is liable for the secondary crime so long as it is a natural and probable consequence of the target crime, even if the aider and abettor did not anticipate that the secondary crime would occur and had no intent that such a crime occur. John F. Decker, The Mental State Requirement for Accomplice Liability in American Criminal Law, 60 S. C. L. REV. 237, 242 (2008). One way to describe "a natural and probable consequence" is a result that, in light of ordinary experience, is one to be expected, rather than an extraordinary or surprising result. See WISCONSIN JURY INSTRUCTIONS - CRIMINAL, No. 411 (2005). On this understanding, a natural and probable consequence is one that is reasonably foreseeable to the defendant at the time he aids and abets the principal in committing the target crime. See id. &comment. (n.11) (citing William L. Prosser, THE LAW OF TORTS, at 252 (West 4th ed. 1971)). Consistent with that understanding, many of the jurisdictions that either expressly embrace the natural and probable consequences doctrine or apply some version of that doctrine without invoking it by name require that a secondary offense committed by the principal have been reasonably foreseeable to his accomplice before the accomplice will be held liable for that offense. See, e.g., Beck, 615 F.2d at 453 ("aiding and abetting [the filing of a false customs export declaration in violation of 18 U.S.C. § 1001] is proven if a defendant voluntarily gives false information or participates in a plan such that it is foreseeable that false information will be used in statements made to a government agency in order to further the plan");⁵ Ala. Code § 13A-2-23, cmt. ("Regarding the extent of liability as a consequence of the principal's actions, Alabama case law has held that as a general rule each person entering upon an unlawful purpose is responsible for everything which may consequently and proximately flow from the unlawful purpose (whether committed by defendant or not, and whether specifically intended or not[.]")) (collecting cases) (emphasis added), and D.L. v. State, 625 So.2d 1201, 1204 (Ala.Crim.App.1993) (juvenile who participated in burglary of residence and theft was not also liable for arson committed by cohort in anger, where "it [did] not appear from the testimony presented that arson was a reasonably foreseeable circumstance of the burglary and theft" and "there [was] no evidence that the arson was the proximate, natural, and logical result of the criminal adventure of burglary and theft upon which the appellant, [cohort] and the other juveniles were engaged"); People v. Favor, 279 P.3d 1131, 1134 (Cal. 2012) ("A person who knowingly aids and abets criminal conduct is guilty of not only the intended crime [target offense] but also of any other crime the perpetrator actually commits [nontarget offense] that is a natural and probable consequence of the intended crime. The latter question is not whether the aider and abettor actually foresaw the additional crime, but whether, judged objectively, it was reasonably foreseeable. Liability under the natural and probable consequences doctrine is measured by whether a reasonable person in the defendant's position would have or should have known that the charged offense was a reasonably foreseeable consequence of the act aided and abetted.") (cleaned up) (emphasis and bracketed language in original); Rodriguez v. State, 147 So.3d 1066, 1068 (Fla. Dist. Ct. App. 2014) ("The independent act doctrine arises when one cofelon, who previously participated in a common plan, does not participate in acts committed by his cofelon, which fall outside of, and are foreign to, the common design of the original collaboration. An independent act instruction is appropriate only when the actions of the cofelon who allegedly acted outside the scope of the original plan were not foreseeable based on the actions a defendant set in motion.") (cleaned up); State v. McFarland, 960 So.2d 1142, 1148 (La. Ct. App. 2007) ("The defendant also contends the evidence was insufficient to convict him [of aggravated burglary] because he did not know Luther was armed. However, under general principles of accessorial liability, all parties to a crime are guilty for the deviations from the common plan, which are the foreseeable consequences of carrying out the plan."); Tex. Penal Code Ann. § 7.02(b) ("If, in the attempt to carry out a conspiracy to commit one felony, another felony is committed by one of the conspirators, all conspirators are guilty of the felony actually committed, though having no intent to commit it, if the offense was committed in furtherance of the unlawful purpose and was one that should have been anticipated as a result of the carrying out of the conspiracy. In this subsection, 'conspiracy' means an agreement between two or more persons to commit a felony."); Wis.Stat. § 939.05 ("(1) Whoever is concerned in the commission of a crime may be charged with and convicted of the commission of the crime although the person did not directly commit it ... . (2) A person is concerned in the commission of the crime if the person: . (c) Is a party to a conspiracy with another to commit it or advises, hires, counsels or otherwise procures another to commit it. Such a party is also concerned in the commission of any other crime which is committed in pursuance of the intended crime and which under the circumstances is a natural and probable consequence of the intended crime ... ."), and State v. Ivy, 350 N.W.2d 622, 627-28 (Wis. 1984) (because robbery is a violent crime involving the taking of a person's property against his or her will by the use or threat of force, there are many situations in which it would be reasonable to find that an armed robbery was the natural and probable consequence of robbery; in such situations, even if defendant did not actually know that principal was armed, the defendant "would at least be on notice of the likelihood that the person who directly committed the robbery would be armed with a dangerous weapon and might use the weapon").

5 Rosemond acknowledges but takes no position as to the propriety of holding a defendant liable for the natural and probable consequences of the crime that he intended to aid and abet. 572 U.S. at 76 n.7.

3. Duenas-Alvarez and the prospect of "special," overbroad accomplice liability

With that bit of background, we can return to Duenas-Alvarez and Carr's argument as to the breadth of Illinois' accomplice liability. Recall that the primary question presented in Duenas-Alvarez-the one on which the Supreme Court had granted certiorari-was whether the Immigration and Nationality Act's reference to a "theft offense" includes not only those convicted as principals to theft but also aiders and abettors.⁶ By the time the case was briefed on the merits and argued before the Supreme Court, Duenas-Alvarez conceded that the answer to this question was "yes." 549 U.S. at 190. He argued nonetheless that California stood apart from its sister states in embracing the natural and probable consequences doctrine. Thus, although an aider and abettor in California might have intended and acted to support only one particular target offense, he could nonetheless be held liable for an entirely separate crime that his cohort committed, so long as that additional secondary crime was the natural and probable result of the crime he agreed to aid and abet. E.g., People v. Prettyman, 926 P.2d 1013, 1020 (Cal. 1996).⁷ At the time of the Duenas-Alvarez decision, California was among a minority of States to have expressly adopted the doctrine (10 by the count of the LaFave treatise, which the Court cited in Appendix C to its decision). 549 U.S. at 196.

6 See Gonzales v. Duenas-Alvarez, 2006 WL 1723979, at *1 (June 22, 2006) (government's petition for writ of certiorari); Gonzales v. Duenas-Alvarez, 548 U.S. 942 (U.S. Sep. 26, 2006) (granting certiorari).

7 California has since abandoned the natural and probable consequences doctrine in murder cases. See Cal. Penal Code § 188(a)(3) (as amended effective Jan. 1, 2019) ("[I]n order to be convicted of murder, a principal in a crime shall act with malice aforethought. Malice shall not be imputed to a person based solely on his or her participation in a crime."); People v. Lopez, 305 Cal.Rptr.3d 93, 99 (Cal.App. 2023) (noting that amendment to § 188 "wholly eliminated the natural and probable consequences doctrine for first and second degree murder").

But the Supreme Court was not convinced that California did in fact define aiding and abetting liability so broadly as to place it outside the national mainstream of accomplice liability. In order to make such a showing, the Court explained, an offender must be able to identify "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. ... [H]e 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." Id. at 193. And Duenas-Alvarez had not made that showing. Although California was among a minority of jurisdictions to have affirmatively embraced the natural and probable consequences doctrine, only another 10, by Duenas-Alvarez's own count, had gone so far as to expressly reject the doctrine. Id. at 190-91. Most states and the federal government, the Court explained, apply some version of that doctrine or, in the alternative, permit a jury to infer a defendant's shared intent to commit a second, non-target offense in circumstances similar to those in which California has applied the natural and probable consequences doctrine. Id. at 191. Having conducted its own review of the California cases cited by Duenas-Alvarez, the Court was satisfied that there was nothing "somehow special" about the way in which California courts articulated or applied accomplice liability. Id.

The Court's analysis ended there; it had no occasion to consider how a finding that a state's aiding and abetting liability is "somehow special" enough to place it outside the mainstream would affect the categorical analysis of a state offense under Taylor. Duenas-Alvarez does make clear that it is a defendant's burden to show that a given state's accomplice liability rules are so broad as to render them "special": the Court speaks solely in terms of what the defendant, not the government, must show. And the decision helpfully adds that the defendant must demonstrate "a realistic probability, not a theoretical possibility," that the state would find an aider and abettor liable for an offense under circumstances where most other jurisdictions would not. But the decision is opaque as to what a court should do if and when a defendant satisfies that burden.

4. Common design accomplice liability in Illinois

Carr contends that Illinois law is "special" in the way that the Court in Duenas-Alvarez envisioned. Illinois, like California, has adopted the natural and probable consequences doctrine. 2 Wayne R. LaFave, SUBSTANTIVE CRIMINAL LAW § 13.3(b) n.29 (3d ed. 2017 & supp. Oct. 2023); Decker, 60 S. C. L.Rev. at 334-36; Brennan v. People, 15 Ill. 511, 516 (1854); People v. Morgan, 364 N.E.2d 56, 59-60 (Ill. 1977); People v. Green, 535 N.E.2d 413, 421-22 (Ill.App.Ct. 1988). But what sets Illinois apart from other such jurisdictions, Carr argues, is that Illinois does not require that a secondary crime committed by a principal in connection with the target crime have been reasonably foreseeable to his accomplice.

Carr's argument begins with Illinois' accountability statute and its provision for common design culpability. The statute opens with the proposition that "[a] person is responsible for conduct which is an element of an offense if the conduct is either that of the person himself, or that of another and he is legally accountable for such conduct as provided in Section 5-2, or both." 720 ILCS 5/5-1. Section 5-2 goes on to provide:

A person is legally accountable for the conduct of another when _ either before or during the commission of an offense, and with the intent to promote or facilitate that commission, he or she solicits, aids, abets, agrees, or attempts to aid

that other person in the planning or commission of the offense.

When 2 or more persons engage in a common design or agreement, any acts in furtherance of that common design committed by one party are considered to be the acts of all parties to the common design or agreement and all are equally responsible for the consequence of those further acts. ..

720 ILCS 5/5-2(c) (emphasis ours). Note that the first of the two paragraphs of section 5-2 reflects a standard shared-intent approach to accomplice liability.⁸ It is the second paragraph, setting forth the common-design rule, that reflects Illinois' take on the natural and probable consequences doctrine.⁹ Although the common-design provision was not added to the accountability statute until 2009, the addition simply codified a doctrine that was already well-established in Illinois common law. See Gov. James R. Thompson, et al., The Illinois Criminal Code of 2009: Providing Clarity in the Law, 41 J. MARSHALL L. REV. 815, 823-24 (2008).

8 The Fifth Circuit in United States v. Sandoval-Ruiz, 543 F.3d 733 (5th Cir. 2008), addressed an argument that what is now the first paragraph of section 5/5-2(c) sets Illinois apart from federal accomplice liability because it renders a defendant liable for soliciting the commission of an offense in addition to aiding and abetting the offense. The court rejected that argument:

We agree with the district court judge that there is no realistic probability that Illinois would apply the statute to conduct outside the scope of the federal statute. The face of the statute itself supports this point. As in the federal statute, the Illinois statute requires the commission of the substantive offense, the intent to promote the offense, and some form of participation in the offense. Unlike the federal statute, participation may be in the form of solicitation. Solicitation is further defined as "to command, authorize, urge, incite, request, or advise another to commit an offense." The practical meaning of this list of verbs does not differ from the practical meaning of the federal list in such a way as to make the Illinois statute more broad. For instance, a defendant under federal law found to have factually "requested" or "advised" a crime could reasonably be found guilty of having "induced" or "counseled" the crime.

Id. at 736-37. The court also rejected the defendant's contention that one can be held liable as an accomplice in Illinois courts based on nothing more than his approving presence at the scene of the crime. Id. at 737. The Fifth Circuit did not address the argument that Carr makes here, however, that common-design liability places Illinois out of the mainstream with accomplice liability in other jurisdictions.

9 The two are distinct theories of accomplice liability. See People v. Carr-McKnight, 166 N.E.3d 866, 887 (Ill.App.Ct. 2020).

The doctrine dates at least as far back as the Illinois Supreme Court's 1854 decision in Brennan v. People. Albert Story, a contractor for the Illinois Central Railroad, had been murdered in LaSalle, Illinois in December 1853. When laborers angered by a reduction in their daily wage confronted Story at his office, Story fatally wounded one of them with a gun. An enraged mob then chased Story out of his office and tracked him down at a stable near his home, where he was beaten and killed. See Dreadful Riot and Loss of Life on the Illinois Central Railroad, N.Y. DAILY TIMES, Dec. 19, 1853, at 1; Riot at LaSalle- Full Particulars, ALTON TELEGRAPH, Dec. 21, 1853 (transcribed by Nancy Piper), available at http://genealogytrails.com/ill/lasalle/history/1853Riot.html. Multiple men were tried and convicted for Story's murder. In the course of affirming their convictions, the Illinois Supreme Court noted an error in certain jury instructions that were given at the defendants' request:

These instructions required the jury to acquit the prisoners, unless they actually participated in the killing of Story, or unless the killing happened in pursuance of a common design on the part of prisoners and those doing the act to take his life. Such is not the law. The prisoners may be guilty of murder, although they neither took part in the killing, nor assented to any arrangement having for its object the death of Story. It is sufficient that they combined with those committing the deed to do an unlawful act, such as to beat or rob Story; and that he was killed in the attempt to execute the common purpose. If several persons conspire to do an unlawful act, and death happens in the prosecution of the common object, all alike are guilty of the homicide. The act of one of them done in furtherance of the original design is, in consideration of law, the act of all. And he who advises or encourages another to do an illegal act is responsible for all the natural and probable consequences that may arise from its perpetration.

15 Ill. at 516 (citations omitted).

People v. Kessler, 315 N.E.2d 29 (Ill. 1974), decided 120 years after Brennan, is considered a "textbook application of the common-design rule" in Illinois. See People v. Fernandez, 6 N.E.3d 145, 150 (Ill. 2014). Kessler was convicted as aider and abettor on two counts of attempted murder committed in the course of a burglary. Kessler was out drinking with two other men, one of whom had mentioned earlier in the day that he needed $1,800; Kessler had told his companions that he previously worked at a particular tavern and had seen the cash receipts, although they were not as large as $1,800. From Kessler's remark, a plan developed among the three men to commit an after-hours burglary of that tavern. Kessler remained in a car outside of the tavern while his two companions, whom we shall refer to as the principals, went inside to commit the burglary. In the course of the burglary, the two principals were surprised by one of the tavern's owners, who had stopped by to investigate after noticing an unfamiliar car in the parking lot. One of the principals shot the owner in the neck with a pistol they had discovered on the premises and intended to take with them. The principals then left the tavern in the car with Kessler. Eventually, state police forced the car off the road into a ditch. The principals fled the car on foot, leaving Kessler behind. In their attempt to evade capture, one of the principals fired a shot at a state trooper who was chasing them. The Illinois Supreme Court held that Kessler was liable for the two attempted murders (of the tavern owner and police officer) on a straightforward reading of Illinois' accountability statute, even if he did not participate in or intend the shootings:

[T]he burglary was the offense which [Kessler and the principals] had jointly planned and were jointly committing, and each was legally accountable for the conduct of the other in connection therewith. The result was the offense of attempted murder of ... the tap owner, and of

[the] State Trooper ... who answered a report of the incident and who tried to apprehend the fleeing parties.

315 N.E.2d at 33.

As in Kessler, the court in Fernandez imposed liability on an individual who aided and abetted a burglary for a shooting committed by the principal in furtherance of the burglary. Fernandez agreed to drive his friend Gonzales to burgle parked cars near Chicago's Maxwell Street market. Finding no such cars, as the market was closed, they drove to a church parking lot, where Gonzalez decided to steal a car radio from a car parked in the lot. Unbeknownst to Fernandez, Gonzalez had brought a firearm with him. An off-duty police officer planning to attend services at the church was parking his car in the lot when he heard the sound of breaking glass as Gonzalez was smashing the window of the car he intended to burgle. When the police officer walked up to Gonzalez, announced himself, and showed his badge, Gonzalez attempted to flee the scene with Fernandez and fired at the officer. Fernandez was found guilty by accountability of both the burglary of the car and two counts of aggravated discharge of a firearm in the direction of a police officer. Fernandez argued that he could not be held liable for the firearm discharge when he did not know that his friend was armed, let alone that Gonzales would fire the gun at the police officer, and thus could not have intended that the shots be fired. The Illinois Supreme Court disagreed. The court noted that the statute imposes liability for any criminal act done in furtherance of the planned and intended act:

In its brief before this court, the State begins by arguing that, "by conceding his guilt for the

burglary, under the facts of this case, defendant has effectively conceded his guilt for aggravated discharge of a firearm." This is exactly right. As Kessler clearly establishes, section 5-2(c) means that where one aids another in the planning or commission of an offense, that person is legally accountable for the conduct of the person he aids; and that the word "conduct" encompasses any criminal act done in furtherance of the planned and intended act. Here, defendant concedes that he aided Gonzalez in the planning and commission of the burglary. That being the case, defendant is legally accountable for any criminal act that Gonzalez committed in furtherance of the burglary, which in this case was the aggravated discharge of a firearm in the direction of a peace officer. Under well-settled accountability principles, the evidence in this case more than supports defendant's conviction.

6 N.E.3d at 151 (emphasis in original).

These cases by their terms certainly make clear that a person who aids and abets another in committing one offense need not know about other acts committed by the principal in connection with that offense, let alone share the principal's intent as to those acts; once he has agreed to help commit the target offense, he becomes liable for any and all secondary criminal acts committed by the principal in furtherance of that target offense, even if the additional acts were unexpected and even if the defendant would not have agreed to participate in those acts. See also Morgan, 364 N.E.2d at 60 (where plan formed among multiple individuals to rob victim, who was behind on his rent, defendant was present while victim was beaten and killed, and it could be inferred from defendant's presence at beating and his receipt of share of robbery proceeds that he was a participant in robbery, he could also be held liable for murder of victim on accountability theory, notwithstanding defendant's purported remark to cohorts that he would accompany them to robbery but that "I don't want nothing to do with it"); Green, 535 N.E.2d at 422-23 (where defendant was party to plan to commit home invasion of victim's apartment in order to force victim to pay money he owed to a co-defendant, and it was defendant who convinced victim to open burglary gates at entry to apartment on pretext that defendant wanted to purchase cocaine from victim, at which point co-defendants rushed into apartment, demanded money from victim, and ultimately killed him and three young girls sleeping in apartment, defendant was liable for murders on accountability theory: "Defendant's role in the home invasion was integral to its success and clearly he intended to promote or facilitate the commission of that offense by pretending to buy cocaine from Rule [the victim] so that House and Bobo [the co-defendants] could gain entry into the apartment they could not otherwise lawfully enter. _ In the absence of any evidence that defendant detached himself from the criminal enterprise, defendant was accountable for the conduct of House and Bobo after his initial action in getting Rule to open the burglary gates.") (citation omitted); People v. Rogers, 461 N.E.2d 511, 516 (Ill.App.Ct. 1984) (defendant held liable for robbery and murder of victim notwithstanding his contention that he only agreed to participate in burglary of victim's house and that he fled premises before his companions entered home and killed victim: "Whatever is done in furtherance of the design is the act of all and each is guilty of the crimes that are committed. .. Even if defendant was not present at the time victim was threatened and murdered, his knowledge and planning of the criminal scheme and subsequent participation and execution are sufficient to hold him accountable.").

More than that, the broad, unqualified language of the Illinois cases signals that an accomplice might be liable for additional criminal acts committed by the principal in furtherance of a common criminal design even when such acts were not reasonably foreseeable to him. Indeed, Illinois courts have gone so far as to say that "the concept of reasonable foreseeability (i.e. proximate cause) . has no place in accountability analysis." People v. Cooper, 743 N.E.2d 32, 38 (Ill. 2000) (quoting the appellate court's unpublished order in the case); see also People v. Watson, 201 N.E.3d 63, 74 (Ill.App.Ct. 2021) ("Foreseeability is not an element of the common-design theory of accountability."); People v. Ivy, 37 N.E.3d 945, 953 (Ill.App.Ct. 2015); United States ex rel. Lovett v. Schomig, 2001 WL 558052, at *2-*3 (N.D. Ill. May 21, 2001). Reasonable foreseeability does come into play with felony murder-a theory of liability that exposes a defendant to liability for deaths caused not only by his accomplices but by other actors, including criminal rivals, victims, and police officers-but not with accountability, see Cooper, 743 N.E.2d at 38; the two doctrines have different theoretical underpinnings, as the Illinois Supreme Court explained in People v. Dennis, 692 N.E.2d 325 (Ill. 1998):

Felony murder seeks to deter persons from committing forcible felonies by holding them responsible for murder if a death results. Because of the extremely violent nature of felony

murder, we seek the broadest possible bounds for the attachment of criminal liability. For that reason, in felony murder, a defendant's liability is not limited to his culpability for commission of the underlying felony. A defendant may be found guilty of felony murder regardless of a lack either of intent to commit murder or even connivance with a codefendant. Our continued adherence to a proximate cause approach is further exemplary of how broadly we seek to extend the reaches of criminal liability in the case of felony murder.

Id. at 335 (citations omitted). By contrast, "accountability focuses on the degree of culpability of the offender and seeks to deter persons from intentionally aiding or encouraging the commission of offenses." Id. Illinois accountability doctrine is broad to the extent it deems the acts of one party to a common criminal design to be the acts of all, but it also has limits that felony murder does not: the secondary criminal acts must have been committed by one's accomplice, and the acts must have been committed in furtherance of the common design. Compare Cooper, 743 N.E.2d at 38 (defendants who engaged in shoot-out with opposing gang members could not be held liable under accountability theory for death resulting from return fire by opposing gang members, however foreseeable such a death might have been to defendants), with People v. Lowery, 687 N.E.2d 973 (Ill. 1997) (defendant properly held liable for felony murder where, in course of attempting to rob victim at gunpoint, defendant dropped the gun in struggle with victim, victim seized the gun, and when defendant fled the scene, victim shot at defendant but struck and killed bystander).¹⁰

10 See also Butler v. People, 18 N.E. 338, 339 (Ill. 1888) (where marshal attempted to arrest one of two defendants who were breaching peace at horse show and fight ensued among the defendants, marshal, and others who attempted to break up fight, and during fight marshal discharged his revolver, striking and killing bystander, defendants could not be held liable for murder of bystander on accountability theory: "There was no common design or purpose existing between the two defendants and [the marshal]. They had not assembled or come together for the commission of any unlawful act. They were enemies, belonging to opposite factions .. . [The defendants] would be responsible for what they did themselves, and such consequences as might naturally flow from their acts and conduct; but they never advised, encouraged, or assented to the acts of [the marshal], nor did they combine with him to do any unlawful act, nor did they in any manner assent to anything he did, and hence they could not be responsible for his conduct toward the deceased.").

Carr thus makes a decent case for the notion that the breadth of the common-design rule as articulated in Illinois places it out of the mainstream of other natural and probable acts jurisdictions, which typically require that the secondary criminal acts for which an accomplice is being held to account be reasonably foreseeable. Time and again, Illinois courts have emphasized that once an individual has embraced and acted to support the target offense as something he wishes to succeed, he is criminally liable for any secondary offense that his cohort may commit in furtherance of the target offense. Fernandez, 6 N.E.3d at 151. Indeed, Illinois' common-design rule is sometimes referred to colloquially as a "misdemeanor murder rule," in that a defendant may even be held to account for a murder committed in furtherance of a misdemeanor target offense. See People v. Terry, 460 N.E.2d 746, 749 (Ill. 1984) ("We agree that the rule does impose liability for murder even though a misdemeanor was originally intended.").

Defendants in Illinois frequently argue that they agreed only to the target offense, that they neither knew of nor intended for any secondary crimes to take place, that they did not know their cohorts were armed, and that they participated in the target offense on the assurance that no dangerous weapon would be used. These types of arguments consistently fail. See, e.g., People v. Nelson, 89 N.E.3d 725, 735 (Ill. 2017) (where defendant and three other women had engaged in common design to batter male victim, Wilson, and during the beating one of the women, Hall, fatally stabbed victim, defendant was liable for murder notwithstanding her statement to police that she told Hall that they should only beat the victim up, that she did not know that Hall intended to kill Wilson, that during the beating Hall "suddenly and without explanation" stabbed Wilson, and that she tried to stop Hall once the stabbing began (although defendant did not remove herself from the attack): "If the four codefendants' common design was to commit a criminal assault on Wilson, then it would not matter under Illinois law whether Hall 'suddenly' stabbed Wilson during the attack or whether, at some point, defendant told Hall to stop and attempted to grab the knife."); People v. Tarver, 45 N.E.2d 630, 632 (Ill. 1942) (where "Tarver group" of young people, including defendant Mack, confronted "Walker group" of youths in order to avenge a beating suffered by a member of Tarver group, and during confrontation, a Tarver group member took gun from Mack and fatally shot a member of the Walker group, defendant Mack was accountable notwithstanding fact that Mack had agreed to join confrontation on condition there would be no shooting: "A shot fired by one of the defendants, under the circumstances shown, was a shot fired by all and all of them must answer for the result."); People v. Phillips, 14 N.E.3d 1, 56 (Ill.App.Ct. 2014), cert. denied, 574 U.S. 1173 (2015) (where defendant and his three cohorts, including Grimes, went to confront and administer a black eye to woman who had struck defendant's girlfriend in face with a brick, but discovered that there was a crowd near woman's house that was too large to deal with, and they were in the process of leaving when Grimes said "wait for a minute," walked out of defendant's view and then fired a single shot into crowd which struck and killed a man, defendant was guilty of murder on accountability theory: "[D]efendant cannot escape liability merely because his criminal intentions did not rise to the level of murder. By attaching himself to a group bent on illegal acts, defendant became accountable for all the crimes of his companions, including the shooting of [the victim]."); People v. Duncan, 698 N.E.2d 1078, 1083 (Ill.App.Ct. 1998) (where sidewalk scuffle between current and former girlfriends of Orlando Potts escalated into plan to "kick [Potts'] ass," and defendant was part of group that walked to Potts' apartment with that aim in mind, and on arrival at apartment defendant's cohort Lemont took out gun that defendant had given him and shot up apartment, killing 14 year-old victim Gilmore inside apartment rather than Potts, defendant was liable for murder of Gilmore: "The fact that Lemont shot Gilmore during the execution of the group's common plan to hurt Orlando Potts is sufficient to hold defendant accountable for the shooting.").

Even so, although Illinois courts may articulate the Illinois common-design rule in a way that appears uniquely broad, we are not convinced that application of the common-design rule in Illinois is truly out of step with the application of the natural and probable consequence doctrine in jurisdictions where foreseeability is required. Foreseeability, of course, entails an objective rather than a subjective assessment; a court asks not what the individual defendant anticipated, but rather what a reasonable person in his position reasonably could have foreseen. See United States v. Yates, 98 F.4th 826, 839 (7th Cir. 2024); Stanphill v. Ortberg, 129 N.E.3d 1167, 1176-77 (Ill. 2018). With that point in mind, it is worth noting that courts deem all sorts of grave consequences to be foreseeable once a defendant has agreed to engage in a forcible target offense like robbery.

5. Comparing the application and results of Illinois common-design cases with cases from other jurisdictions that require foreseeability

The Illinois Supreme Court's own decision in Hami/ton v. Peop/e, 113 Ill. 34 (1885), which involved an attempted murder committed in the course of a botched watermelon heist, nicely illustrates why the results of Illinois common-design cases are consistent with the holdings of other jurisdictions that require secondary crimes to be reasonably foreseeable to an accomplice. Hamilton was among three men who climbed over a fence into a farmer's watermelon patch one night, intending to steal some of the melons. One of the men had a pistol with him. As it happened, the farmer and his son (armed with a shotgun) were keeping watch over the patch, having noticed thefts over the course of the prior two evenings. A verbal confrontation between the intruders and the farmer escalated into a physical scuffle. The farmer's wife, armed with a hoe, joined the fray. One of the men seized the hoe from her and struck the farmer in the head with it, knocking him out. The court sustained the conviction of Hamilton as well as the principal who struck the farmer with the hoe for assault with intent to commit murder:

The fact is undisputed that the three defendants, one of whom was armed with a pistol, invaded the premises of the prosecuting witness [the farmer] with a criminal purpose. The business upon which the parties had deliberately entered was a hazardous one. They had a right to expect that in the event they were detected in stealing the melons, it would result in violence endangering life or limb[ ]-as it actually turned out afterwards. That they were all coconspirators in a dangerous criminal enterprise, is an undisputed fact. Such being the case, whatever was done by one, in contemplation of law was done by all, and all are therefore equally responsible.

Id. at 37-38. Note that the court appears to have actually considered whether attempted murder was a foreseeable consequence of watermelon theft, and answered that question in the affirmative, emphasizing that having deliberately embarked on a hazardous criminal scheme, the defendants "had a right to expect" that violence might erupt in the event they were caught stealing the melons. Id. (emphasis added); see also Morgan, 364 N.E.2d at 60 ("Where one attaches himself to a group bent on illegal acts which are dangerous or homicidal in character, or which will probably or necessarily require the use of force and violence that could result in the taking of life unlawfully, he becomes accountable for any wrongdoings committed by other members of the group in furtherance of the common purpose, or as a natural or probable consequence thereof even though he did not actively participate in the overt act itself." (quoting the lower court's decision, 350 N.E.2d 27, 34 (Ill.App.Ct. 1976)).

Cases from other natural and probable consequence jurisdictions have reached similar conclusions. To take a prominent example, the California Appellate Court's decision in People v. Nguyen, 26 Cal.Rptr.2d 323 (1993), was one of the cases that the U.S. Supreme Court considered in Duenas-Alvarez in assessing (and rejecting) the contention that California's natural-and-probable-consequence jurisprudence was an outlier. In that case, the defendants, who had aided and abetted robberies of tanning and relaxation spas-which evidently were really operating as illicit massage parlors-were held liable for a sexual assault that the principal committed on the owner of the tanning spa as a way of frightening the owner into cooperation. The California court rejected the defendants' argument that they had not intended for the assault to occur and that a sexual assault was not a foreseeable consequence of robbery:

Robbery is a crime that can be committed in widely varying circumstances. It can be committed in a public place, such as on a street or in a market, or it can be committed in a place of isolation, such as in the victim's home. It can be committed in an instant, such as in a forcible purse snatching, or it can be committed over a prolonged period of time in which the victim is held hostage. During hostage-type robberies in isolated locations, sexual abuse of victims is all too common. As Presiding Justice Gardner observed (with respect to residential robbery) in his concurring opinion in

People v. Lopez (1981) 116 Cal.App.3d 882, at page 891, 172 Cal.Rptr. 374: "When robbers enter the home, the scene is all too often set for other and more dreadful crimes such as that committed on Mrs. H. in this case. In the home, the victims are particularly weak and vulnerable and the robber is correspondingly secure. The result is all too often the infliction of other crimes on the helpless victim. Rapes consummated during the robbery of a bank or supermarket appear to be a rarity, but rapes in the course of a residential robbery occur with depressing frequency." With respect to residential robbery and other isolated-victim and hostage-type robberies, we agree with Presiding Justice Gardner's observation, both as a reflection of the reported decisional authorities, and as a synopsis of our experience with criminal cases which this court is regularly required to consider. Robbery victims are sexually assaulted far too often for this court to conclude, as a matter of law, that sexual offenses cannot be a reasonably foreseeable consequence of a robbery.

In turning to the facts of this case we find ample evidence to support a theory that the sexual offenses were a reasonably foreseeable result of the defendants' participation in the group criminal endeavor. The defendants and their cohorts chose to commit robberies in businesses with a sexual aura, both from the types of services they held themselves out as providing and from the strong suspicion, repeatedly expressed by the

participants at the trial, that they were actually engaged in prostitution. The businesses were arranged much like a residence, with separate rooms furnished as bedrooms might be. The businesses operated behind locked doors, which both added to their sexual aura and gave the robbers security against intrusion or discovery by outsiders. The robbers went to the businesses in sufficient numbers to easily overcome any potential resistance and to maintain control over the victims for as long as they desired.

Id. at 332-33 (additional citations omitted).¹¹ Cf. People v. James, 93 N.E.3d 626, 636-38 (Ill.App.Ct. 2017) (concluding that sexual assault was not in furtherance of original common design to invade a residence and rob the occupants, but going on to find evidence was sufficient to permit jury to find that that defendant became part of a secondary design with co-defendant to commit sexual assault on victim).

11 The Nguyen court went on to add that the defendants continued to participate in the robbery after the sexual assault occurred, a circumstance the court viewed as confirming their acceptance of and shared responsibility for the assault. Id. at 333. It was this sort of observation that the Supreme Court had in mind in Duenas-Alvarez when it noted that California's application of the natural and probable consequences doctrine in Nguyen and like cases was consistent with the precedents of jurisdictions that premise accomplice liability on facts supporting an inference that the accomplice shared the principal's intent to commit a secondary crime. 549 U.S. at 191.

Cases from still other jurisdictions apply the natural and probable consequences doctrine in a comparable way, deeming secondary offenses committed by a principal in furtherance of the target crime to have been reasonably foreseeable to the aider and abettor even when those secondary offenses represented a dramatic and unexpected escalation of force by the principal, and even when the defendant did not realize the principal was armed. See Howell v. State, 339 So.2d 138, 140 (Ala.Crim.App.1976) (defendant liable for assault with intent to murder after he and two others conspired to rob gas station in early morning hours, and when police officer arrived to investigate, principal shot him: "[a]ssault with intent to murder would be a foreseeable consequence of the joint enterprise in which [the defendant and the two others] were engaged"); Johnson v. State, 687 N.E.2d 345, 349-50 (Ind. 1997) (notwithstanding defendant's contention that murder was beyond the scope of a plan with principal to rob defendant's father-in-law, defendant was liable as accomplice for felony murder where during robbery at father-in-law's home, principal killed both father-in-law and his wife: "The criminal liability of an accomplice is not negated by the principal's commission of an offense greater in severity than the offense originally planned, if the resulting offense is a probable and natural consequence of the planned offense. Johnson admitted that he was the instigator of the planned robbery and that he left the door to the Smiths' house unlocked with the intent that [the principal] would enter and rob the Smiths. _ Planning a robbery of a person in that person's own home is bound to create a risk that violence may ensue when the homeowner predictably attempts to protect himself and his family.") (citations omitted); State v. Edwards, 498 P.2d 48, 52 (Kan. 1972) (where defendant and three companions had agreed to rob victim, and defendant upon entry to victim's house looked at furniture and remarked, "we could make a killing here" and personally stole a radio, and all four left house in defendant's car after principal stabbed victim, defendant was liable for aggravated battery of victim notwithstanding his acquittal on burglary charge and his contention that it was unforeseeable principal would stab victim with paring knife: "Robbery is a crime of violence committed by threat or force. There was evidence that defendant participated in the aggravated robbery by taking a radio from the premises. From the facts it may be readily inferred that violence, if necessary, was contemplated when the four entered the house. .... Under these circumstances, the defendant can hardly be considered an innocent bystander in the whole affair."); State v. Smith, 748 So.2d 1139, 1143 (La. 1999) (where co-defendants set out with principal to burglarize temporarily unoccupied home of their occasional employer, victim and wife arrived and interrupted burglary suddenly and unexpectedly, and co-defendants fled house while principal stayed behind and fatally shot victim, co-defendants were liable for murder notwithstanding their contention (bolstered by principal's testimony) that they were unaware principal was armed: "[U]nder general principles of accessorial liability, all parties to a crime are guilty for deviations from the common plan which are the foreseeable consequence of carrying out the plan. The risk that an unauthorized entry of an inhabited dwelling may escalate into violence and death is a foreseeable consequence of burglary which every party to the offense must accept no matter what he or she actually intended.") (cleaned up); People v. Robinson, 715 N.W.2d 44, 5051 (Mich. 2006) (where defendant and principal went to victim's house to "f*** the victim up," both defendant and principal struck victim until defendant told principal "that's enough" and walked back to car, whereupon principal shot and killed victim, defendant was liable for murder: "The victim's death is clearly within the common enterprise the defendant aided because a homicide might be expected to happen if the occasion should arise within the common enterprise of an aggravated assault. _ A natural and probable consequence of leaving the enraged [principal] alone with the victim is that [he] would ultimately murder the victim.") (cleaned up); State v. Filippi, 335 N.W.2d 739, 742 (Minn. 1983) (defendant held liable for assault with a dangerous weapon after he and principal attempted to rob drugstore and principal shot at police officers who responded to robbery, notwithstanding principal's testimony that defendant had never before seen principal with gun and principal did not tell him that he was carrying the gun with him during the burglary: court reasoned in dicta that it is "reasonably foreseeable" that police might arrive in course of burglary, and that because "a burglary of this sort carries with it the possibility of violence, it seems reasonable to infer from defendant's mere intentional participation in the burglary that defendant knew or could foresee that the burglary might result in violence"); State v. Jackson, 601 N.W.2d 741, 751 (Neb. 1999) (where defendant and principal, while awaiting a taxi in front of mini mart, engaged in a hostile verbal exchange with two other men who had walked out of store, principal followed men across street and began fighting with one of the men-the victim-defendant crossed street ostensibly to break up fight but fought back when second man accosted him, and first man later died as a result of blows inflicted by principal, defendant could be held liable for manslaughter: "the jury could have determined from the evidence that [defendant's] participation in the assault on [the second man] could also be characterized as assistance or encouragement to [the principal's] assault on [the victim]. The jury could also have reasonably determined that [the victim's] death was a natural and expected consequence of the beating inflicted by [the principal].").

6. Illinois: not so "special" after all

The application of the natural and probable consequences doctrine in these cases strikes us as entirely consistent with the application of Illinois' common-design rule. In other words, although Illinois courts may frame the common-design rule of accomplice liability in very broad terms, the results of the Illinois cases are wholly consistent with the cases from other jurisdictions which require that a secondary offense have been foreseeable to the accomplice. In practice, then, Illinois courts are not "apply[ing] its [accomplice liability] statute to conduct that falls outside the generic definition of a crime." Duenas-Alvarez, 549 U.S. at 193.

For example, returning to Kessler-the "textbook" Illinois common-design case-where the accomplice was convicted of attempted murder based on the actions of his cohorts in the course of an after-hours tavern burglary, it is not difficult to construct an argument that it would have been reasonably foreseeable to Kessler, having agreed to a plan in which the two principals would break into a tavern and raid the till, that someone, be it the tavern owner or a police officer, might catch them in the act and attempt to forcibly stop the burglary, that such an individual might be armed, and that there might be a gun kept on the premises. Even if Kessler's accomplices had not found a gun in the tavern, it is not at all hard to imagine them picking up some other dangerous implement-a baseball bat or a knife, for example-or snatching a gun away from an officer or other intervenor-and using that weapon to complete the burglary and make a getaway.

Likewise with Fernandez: Fernandez's principal smashed the window of a car in a church parking lot in broad daylight, while worship services were commencing. One could readily foresee that a churchgoer or a police officer might happen by, and then what? It is anything but improbable that violence might occur. And, of course, in both instances, Kessler and Fernandez remained with their accomplices once gunfire erupted and to that extent could be read as endorsing the use of violence. See Duenas-Alvarez, 549 U.S. at 191 (noting that many states permit jury inferences of intent in circumstances similar to those in which California has applied the natural and probable consequences doctrine).

In short, although Carr has arguably established that Illinois' formulation of common-design doctrine is unusually broad to the extent it does not require that a secondary criminal act committed in furtherance of the agreed-upon target offense have been reasonably foreseeable to the defendant, he has not shown that Illinois courts have applied the doctrine in a way that is at odds with the natural and probable consequences jurisprudence in other states; in practice, Illinois courts do not hold accomplices liable for secondary offenses that were not reasonably foreseeable to them. There is thus nothing "special" about the results of the common-design cases in Illinois-as we have seen, the same results could just as easily have been reached in other natural and probable consequences jurisdictions. To be sure, one can find cases in other jurisdictions where a defendant was held not to be culpable for a secondary criminal act committed by the principal in connection with the target offense on the ground that the secondary crime was not reasonably foreseeable to the defendant. See, e.g., State v. Davis, 604 P.2d 68, 72-73 (Kan.Ct.App. 1979) (defendant who stood as lookout while principal attempted to pry open vending machine in unoccupied apartment-building laundry room could not be held liable for aggravated battery after security guard interrupted the attempted theft, ordered both men up against wall, and then principal ran back into laundry room, retrieved gun, fired at guard, and struck him: "[T]he intended crime, misdemeanor theft, was to take place in the early morning hours in a laundry room that the defendant knew to be unoccupied when the theft occurred. ... The record on appeal does not even hint that the defendant knew or had any reason to suspect [the principal] had a weapon. . In our opinion, it is mere speculation to say that a person who with another is planning to commit a misdemeanor theft in an unoccupied room can reasonably foresee in the absence of other facts that the coconspirator will shoot someone in an effort to avoid apprehension."). But Carr has pointed us to no Illinois case which has applied the common-design rule to hold a defendant liable for secondary crimes that would not have been (at least arguably) reasonably foreseeable to him.¹²

12 Carr points us to the Illinois Court of Appeals' decision in People v. Stinde, 2011 WL 10457946 (Mar. 23, 2011), for the proposition that Illinois courts will apply the common-design doctrine so broadly as to impose liability on an accomplice for a secondary, violent crime even when the target offense of the common design was a non-violent offense. The Stinde decision is both non-precedential and not citable even for its persuasive value. See Ill. Sup. Ct. Rule 23(e)(1). In any case, the facts at issue in Stinde reveal that violence was an apparent possibility from the start of the criminal plan. Stinde had agreed to help his sister's boyfriend buy some marijuana. But in fact, Stinde and his two cohorts intended to rip off the boyfriend. While Stinde and the boyfriend were sitting in a parked car, ostensibly waiting for the "weed man" to arrive, one of Stinde's cohorts walked up to the car and shot and killed the boyfriend. In contesting his conviction for murder on an accountability theory, Stinde would later contend that at most, the plan was to commit theft by fraud by selling the boyfriend substandard marijuana, and that violence was wholly outside the scope of that common design. But the Court of Appeals was plainly unconvinced by the premise of that argument. The court pointed out that the principal who shot the victim was desperate for money, that the principal's modus operandi was to commit robbery when he needed cash, and that when Stinde's sister had contacted Stinde about her boyfriend's wish to buy marijuana, Stinde had remarked to her, "I hope you don't like him." Stinde is yet another Illinois case, then, in which it could be said that the prospect of violence was reasonably foreseeable to the accomplice.

Given Duenas-Alvarez's insistence that there be a realistic probability that a state would apply its accountability statute to conduct falling outside the generic definition of a crime, 549 U.S. at 193; Moncrieffe v. Holder, 569 U.S. 184, 191, 206 (2013), we must look past what the courts of that state say to what they actually do. In the abstract, given the broad language that Illinois courts use to describe the common-design rule, one can of course hypothesize scenarios in which an aider and abettor theoretically might be held liable to account for a completely unforeseeable secondary offense that his principal commits in furtherance of the agreed-upon target offense: strangling a store clerk who refuses to accept a counterfeit bill or a forged check that the accomplice prepared and the principal attempted to pass, for example. But Illinois courts have decided hundreds of common-design cases since the Illinois Supreme Court decided the Brennan case 170 years ago. Among them we have been unable to find one in which a court has held an aider and abettor liable for a secondary offense that one could not say was reasonably foreseeable to him.

Perhaps another way of putting this is that although other states nominally narrow the scope of the natural and probable consequences doctrine by insisting that the secondary offense be reasonably foreseeable to the accomplice, that additional requirement does not grant the accomplice as much protection as Carr might like to think. As we have seen, once an individual has agreed to aid and abet a forcible offense-any-thing from residential burglary to bank robbery-courts are likely to deem almost any violent act committed in furtherance of that offense to be reasonably foreseeable, however sudden and unexpected it may have been to the aider and abettor.

As Duenas-Alvarez makes clear, it is Carr's burden to show that accomplice liability in Illinois is truly special. He has not made that showing. He has identified a different, perhaps unique, formulation of aiding and abetting liability, but he has not shown that the application of the common-design rule in Illinois has actually resulted in convictions that would not be sustained in other jurisdictions. We therefore reject Carr's contention that Illinois robbery is a categorical mismatch for generic robbery.

7. Assuming that Illinois accomplice liability is "special"-what is next?

Suppose, however, that we are wrong in this conclusion, and that Illinois law is properly treated as the sort of outlier vis-a-vis aiding and abetting liability that the Supreme Court had in mind in Duenas-Alvarez when it referred to state accountability doctrine that is "somehow special." Where does this lead us?

To Carr, the answer is clear. An Illinois defendant can be convicted on an accountability theory even if he is charged as a principal; as in other jurisdictions, Illinois treats accountability not as a separate offense, but rather as an alternative way of proving the defendant guilty of the substantive offense. People v. Ceja, 789 N.E.2d 1228, 1247 (Ill. 2003). Consequently, conviction as an accomplice pursuant to the common-design rule is a theoretical possibility in every Illinois case, as Carr argues.¹³ On the assumption that he theoretically might have been convicted of Illinois armed robbery on a common-design theory of aiding and abetting rather than as principal, Carr would have us graft the common-design rule onto the elements of Illinois robbery, plug these grafted elements into the Taylor categorical analysis and-if we accept the premise that Illinois common design is outside the main-stream-deem Illinois robbery to be a categorical mismatch for generic robbery in view of the "special" nature of Illinois accountability theory.

13 Looking only at the face of Carr's prior robbery convictions, we would not be able to tell whether he was convicted as aider and abettor, as we would have been able to do had he been charged as such-and, on occasion, people are charged as aiders and abettors in Illinois. E.g., People v. Dennis, supra, 692 N.E.2d at 332; People v. Kijowski, 377 N.E.2d 1324, 1327 (Ill.App.Ct. 1978). If Carr had been charged as an accomplice, then we could apply the Taylor categorical approach in the usual way, focusing on the elements of the charged offense. Carr's argument, however, is premised on the notion that he might have been convicted as an aider and abettor without having been charged as such.

This certainly was the premise of the petitioner's argument in Duenas-Alvarez, and arguably it was the premise on which the Supreme Court's analysis proceeded. But because the Court found that California's accomplice liability was not outside of the mainstream, it did not need to consider next steps. If, as Carr argues, Illinois accomplice-liability really is unique, we do not have that luxury.

The first problem we have with Carr's invocation of Taylor is that accountability is simply a theory of liability. But Taylor's categorical analysis is reserved for the elements of the offense, as Carr himself agrees. See United States v. (Justin) Taylor, 596 U.S. 845, 850 (2022); Descamps, 570 U.S. at 261. Accountability is not an element of robbery or any other substantive offense: not only is accountability a liability theory that need not be charged, but as Carr points out, the jury need not even agree unanimously whether a defendant is guilty as an accomplice or as a principal. People v. Dunbar, 127 N.E.3d 604, 614-15 (Ill.App.Ct. 2018) ("We agree that it is difficult to ascertain whether a particular juror signed the guilty verdict after finding defendant acted as the principal or acted as an accomplice ... [but] unanimity is not required concerning alternate ways in which the crime can be committed."); People v. Jackson, 874 N.E.2d 123, 129 (Ill.App.Ct. 2007) ("Defendant was not entitled to a unanimous verdict on whether he fired the weapon or whether Bascomb fired the weapon. The jury need only be unanimous with respect to the ultimate question of defendant's guilt or innocence of the crime charged, and unanimity is not required concerning alternate ways in which the crime can be committed."); People v. Travis, 525 N.E.2d 1137, 1147-48 (Ill.App.Ct. 1988). So, by insisting that we import Illinois accountability principles into the elements of Illinois robbery, Carr would have us input something into the elements-only categorical analysis for which it was not designed.

Second, treating Illinois robbery as a mismatch for generic robbery based on the breadth of Illinois accomplice liability would have the effect of disqualifying all Illinois robbery convictions as predicates for either an increase in the minimum and maximum statutory penalties applicable to a defendant, see, e.g., 18 U.S.C. § 924(e) (Armed Career Criminal Act; 21 U.S.C. §§ 841(b) and 843(d) (penalties for controlled substances offenses), or an increase in the Sentencing Guidelines offense level (and corresponding sentencing range), see, e.g., Guidelines § 2K2.1(a) (robbery guideline), regardless of whether or not those convictions in fact rested on accountability theory. This would, in turn, reduce the sentencing exposure for defendants with Illinois robbery convictions in their criminal histories, against the obvious intent of Congress and the Sentencing Commission, which both recognized that defendants with prior violent felonies in their past are, as a group, more dangerous to society than those without such convictions in their histories. Sentencing judges presumably could compensate for the distortion in the Guidelines offense level and advisory sentencing range by taking into account prior Illinois robbery convictions as an aspect of "the history and characteristics of the defendant" under 18 U.S.C. § 3553(a)(1) and as a way of avoiding unwarranted sentencing disparities among defendants with similar backgrounds under section 3553(a)(6); but they could do nothing about the effects on the statutory minimum and maximum penalties to which the defendant is subject. Similarly, in immigration cases, there would be no way for a judge to compensate for the effect of disqualifying a predicate conviction as the basis for removal. See Valivia-Flores, 876 F.3d at 1211 (O'Scanlain, J., concurring specially).

It also bears mentioning that there is nothing about the logic of Carr's argument that is limited to robbery; because liability as an accomplice is inherent in every Illinois offense, regardless of whether it is charged, each and every Illinois offense could be labeled as a mismatch for its generic counterpart based on the putatively "special" nature of the common-design rule in Illinois. Under Carr's approach, then, no Illinois offense-even first-degree murder-could qualify as a crime of violence. The Supreme Court has consistently cautioned against a reading of federal recidivism and immigrant-removal statutes in a manner that would defeat the purpose of those statutes. See Pugin v. Garland, 599 U.S. 600, 607 (2023); Quarles v. United States, 139 S.Ct. 1872, 1879 (2019); Stokeling v. United States, 586 U.S. 73, 81-82 (2019); Taylor, 495 U.S. at 593-94. And as we observed in Worthen, "[t]here is no indication that Taylor intended the categorical approach to apply to aiding and abetting liability in th[e] way" that Carr proposes. 60 F.4th at 1071 (citing United States v. Cammorto, 859 F.3d 311, 316 (4th Cir. 2017) (rejecting a similar challenge as "untenable" because it would preclude any categorical match)); see also Gamez, 77 F.4th at 600; Valdivia-Flores, 976 F.3d at 1210-11 (O'Scanlain, J., concurring specially).

As the government suggests, one way to mitigate this problem, while accounting for the concerns that Carr has raised about the breadth of Illinois accomplice liability, would be to employ a modified categorical approach in order to determine whether it is possible that the defendant was actually convicted as an accomplice rather than as a principal. But whatever merit there might be in this notion, the Supreme Court's decision in in Mathis v. United States, 579 U.S. 500 (2016), dooms it from the start.

Courts traditionally use the modified categorical approach when a defendant has been convicted of violating a divisible statute, that is, a statute which lists one or more offense elements in the alternative, and one of those elements matches the corresponding element of the generic offense and one does not. See Descamps, 570 U.S. at 257.

[T]he 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. The court can then do what the categorical approach demands: compare the elements of the crime of conviction (including the alternative element used in the case) with the elements of the generic crime.

Id.; see Shepard v. United States, 544 U.S. 13, 26 (2005) (describing the limited class of documents a court may consult when applying this approach); Najera-Rodriguez v. Barr, 926 F.3d 343, 348 (7th Cir. 2019); United States v. Edwards, 836 F.3d 831, 835 (7th Cir. 2016); Abdelqadar v. Gonzales, 413 F.3d 668, 672 (7th Cir. 2005). If an Illinois prosecutor argued to a jury that the defendant was guilty of robbery as an accomplice, then the jury would be instructed on Illinois accountability principles. See ILL. PATTERN JURY INSTRUCTIONS - CRIMINAL, No. 5.03. And if a defendant pleaded guilty to Illinois robbery as an accomplice, then the plea colloquy would likely reflect as much. in such cases, where the Shepard documents confirm that the defendant might have been convicted of Illinois robbery as an accomplice, then it arguably would be appropriate to disqualify the Illinois conviction as one for a crime of violence, given the purported mismatch between Illinois common-design liability and generic accomplice liability. On the other hand, where a review of the Shepard documents rules out the possibility that a defendant might have been convicted of Illinois robbery as an accomplice, then there would be no need to deem the conviction a categorical mismatch for generic robbery, for the common-design rule will not have come into play at all. The robbery conviction could then be relied upon as a valid predicate for sentence enhancements. See Moncrieffe, 569 U.S. at 219-20 (Alito, J., dissenting) (advocating for a more flexible, common-sense approach to assessment of state convictions so as to avoid results that Congress did not intend).

Following that approach here would of course confirm that Carr was not convicted as an aider and abettor, let alone as a party to a common design to commit some lesser crime that evolved into armed robbery. The plea colloquy leading to his convictions on the three Illinois robberies reveal that he was a principal who played a key role in the 2007 armed robbery and was the sole actor in the two 2008 robberies.

But here we run into the third problem with Carr's approach: As we discussed, above, Taylor's categorical inquiry is reserved for offense elements, and accountability is not an element of robbery, but rather an alternative theory of liability, which suggests that we cannot employ the modified categorical approach here. Indeed, the Supreme Court's decision in Mathis resolves any doubt on this score. At issue in Mathis was Iowa burglary, and whether it qualified as a predicate "violent felony" for purposes of triggering enhanced statutory penalties under the Armed Career Criminal Act, 18 U.S.C. § 924(e). The Iowa burglary statute had a broader reach at that time than generic burglary: whereas generic burglary requires unlawful entry into a building or other structure, the Iowa statute also encompassed entries into land, water, and air vehicles (i.e., cars, boats, and planes). Those various places were listed in the state statute, and although they were properly understood as alternative means of satisfying the locational element of the offense, 579 U.S. at 506,¹⁴ the question was whether they could be treated as the equivalent of alternative elements for purposes of the Taylor inquiry, in which case a court could employ the modified categorical approach and consult the Shepard documents to confirm that a defendant was convicted of burglarizing a building or other structure, such that the conviction would count as burglary for purposes of a sentence enhancement. The Court answered this question firmly in the negative, and reversed the lower courts for having erroneously applied the modified categorical approach to conclude that Mathis had, in fact, burglarized a structure. "How a given defendant actually perpetrated the crime-what we have referred to as the underlying brute facts or means of commission-makes no difference [to the categorical inquiry]; even if his conduct fits within the generic offense, the mismatch of elements saves the defendant from an ACCA sentence." Id. at 509 (cleaned up). The modified categorical approach, the Court admonished, "is not to be repurposed as a technique for discovery whether a defendant's prior conviction, even though for a too-broad crime, rested on facts (or otherwise said, involved means) that also could have satisfied the elements of a generic offense. Id. at 513-14. Here, as in Mathis, we are dealing with something that is not an element of the offense-whether the defendant committed the crime as an aider and abettor rather than as a principal-and Mathis precludes us from "repurpos[ing]" the modified categorical approach to determine whether a defendant might have been convicted on a particular theory of liability. See Alfred, 64 F.4th at 1051-55 (concurrence in part, dissent in part); Bourtzakis v. United States, 940 F.3d 616, 627 (11th Cir. 2019) (Robreno, D.J., concurring in part and concurring in judgment).

14 With respect to the distinction between the elements of an offense and the means of committing an offense, we observed in Najera-Rodriguez:

The difference between "elements" and "means" can seem slippery, sometimes almost metaphysical, but significant legal consequences flow from that difference. "'Elements' are the 'constituent parts' of a crime's legal definition-the things the 'prosecution must prove to sustain a conviction.'" Mathis, 136 S.Ct. at 2248, quoting Black's Law Dictionary 634 (10th ed. 2014). And just as a prosecutor must prove beyond a reasonable doubt every element of a crime to a jury, a defendant pleading guilty necessarily admits every element of the crime. Id. By way of contrast, the facts of the offense "are mere real-world things-extraneous to the crime's legal requirements," and "need neither be found by a jury nor admitted by a defendant." Id. In addition, the jury cannot convict without agreeing unanimously on each element of the crime, while jurors need not reach any agreement on subsidiary facts or "means" of committing the crime. United States v. Edwards, 836 F.3d 831, 836 (7th Cir. 2016). This distinction is critical under this body of law for collateral consequences, as well as for other purposes. For example, a crime's elements affect multiplicity challenges because the "Double Jeopardy Clause permits successive punishment or prosecution of multiple offenses arising out of the same conduct only if each offense contains a unique element." Id.

926 F.3d at 348-49. Mathis gives guidance as to how a court can distinguish between means and elements, and our opinions in Edwards, 836 F.3d at 836-37, and Najera-Rodriguez, 926 F.3d at 349-50, address that guidance. Of course, in this case, it is undisputed that the theory of liability on which a state seeks conviction of the defendant-including an aiding and abetting theory of accountability-is not an element of the offense.

So now we are back to Carr's view, which is that if Illinois accomplice liability is overbroad, then all Illinois robberies are necessarily disqualified as predicates for federal statutory and Guidelines sentence enhancements. His is an all-or-nothing proposition which forecloses any look behind the face of a defendant's conviction for Illinois robbery even for the limited purpose of determining whether it is possible the defendant was or was not convicted as an accomplice under a common-design theory. (This is not surprising: as we have said, his challenge would fail if we applied anything resembling a modified categorical analysis of his robbery convictions.) The results of Carr's approach are unacceptable, as we have previously signaled. Gamez, 77 F.4th at 600; Worthen, 60 F.4th at 1070-71. And Duenas-Alvarez does not, in our view, compel that result.

Duenas-Alvarez, as we have discussed, certainly envisions the possibility that a state might define its aiding and abetting in such an expansive way as to render the substantive offense that a defendant aided and abetted a mismatch for the generic offense. The Court rejected Duenas-Alvarez's contention that California was such a state, but it left the door to the possibility that another defendant might be able to establish "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. 549 U.S. at 193. But the Court did not answer the questions we have raised here: How does the Taylor categorical inquiry address an accomplice theory of liability that has not been charged and indeed need not be charged? To the extent that liability as an aider and abettor is inherent in every substantive offense, if a state's accomplice liability is so broad as to place it outside of the mainstream, does that disqualify all of the state's convictions as drug or violent-crime predicates for federal sentencing purposes? To avoid that result, can a court apply some analogue of the modified categorical approach in order to assess whether accomplice liability was even at issue in the case resulting in the defendant's conviction? See Alfred, 64 F.4th at 1057 (concurrence in part and dissent in part) ("because the Court [in Duenas-Alvarez] rejected th[e] premise [that California's aiding and abetting doctrine was an outlier], the Court never addressed what would have followed if Duenas-Alvarez had been correct in his analysis of California law") (emphasis in original).

Until such questions are answered, we do not believe it is appropriate to discard Carr's prior convictions for robbery as sentencing predicates based solely on the theoretical possibility that Carr might have been convicted as an accomplice. Carr points us to no court that has taken such a step.¹⁵ Carr's prior convictions were for Illinois armed robbery, period. Taylor asks, are the elements of Illinois robbery a match for the elements of generic robbery? The answer to that question is yes. Until we are told otherwise, that is the end of our inquiry.

15 The Ninth Circuit briefly took this course in two cases, Valdivia-Flores, supra, 876 F.3d 1201 (deeming Washington conviction for possessing a controlled substance with intent to deliver to be a categorical mismatch for federal counterpart based on Washington's aiding and abetting law which, at time of petitioner's conviction, did not require defendant's intent to aid in commission of offense; consequently, conviction did not qualify as "aggravated felony" for purposes of petitioner's removability under Immigration and Nationality Act), and United States v. Franklin, 904 F.3d 793 (9th Cir. 2018) (likewise deeming Washington convictions for delivery of controlled substance to be categorical mismatch for their generic counterpart based on Washington's aiding and abetting law; consequently, convictions did not qualify as predicate "serious drug offenses" for purposes of Armed Career Criminal Act), abrogated in part by Shular v. United States, 589 U.S. 154 (2020). More recently, however, the en banc Ninth Circuit overruled those cases in Alfred, 64 F.4th 1025. We acknowledge that there are other circuits which have applied the Taylor categorical approach to a state's accomplice liability rules even when the defendant was not charged as an aider and abettor. See United States v. Coats, 8 F.4th 1228, 1247-61 (11th Cir. 2021); United States v. Capelton, 966 F.3d 1, 6-10 (1st Cir. 2020); Bourtzakis, 940 F.3d at 621-25; United States v. Boleyn, 929 F.3d 932, 936-40 (8th Cir. 2019). In those cases, however, the courts were able to conclude, without too much difficulty, that the accomplice liability of the state in question was not outside of the mainstream, just as the Supreme Court did in Duenas-Alvarez. By contrast, the Ninth Circuit's fractured opinion in Alfred, 64 F.4th 1025, shows how challenging application of the Taylor framework can be where there is a plausible case to be made that a state's accomplice liability is somehow broader or different from that typically applied in other jurisdictions. See id. at 1031-48 (lead opinion) (reasoning that Washington's accomplice liability must be factored into the Taylor categorical analysis of a Washington robbery conviction under Duenas-Alvarez when, as a matter of state law, liability as an accomplice is implicit in every criminal charge, and holding that Washington's accomplice liability was not out of step with generic accomplice liability, including federal aiding and abetting liability under 18 U.S.C. § 2); id. at 1050-66 (Collins, J., concurring in part and dissenting in part) (asserting that a state's accomplice liability should not be factored into categorical analysis of robbery conviction; rather, court should confine its analysis to the elements of Washington robbery and whether those elements line up with those of generic robbery, which they do; also rejecting the court majority's articulation of federal aiding and abetting liability under § 2); id. at 1067-72 (McKeown, J., dissenting) (agreeing with lead opinion that state's accomplice liability must be factored into categorical analysis of whether Washington robbery constitutes a theft offense for federal purposes, but concluding that Washington accomplice liability is not a categorical match with generic accomplice liability, and so Washington robbery is not a categorical match with generic robbery); id. at 1071-74 (Van Dyke, J., dissenting) (agreeing that Washington robbery is a categorical mismatch with generic robbery once accomplice liability is taken into consideration; although finding "a lot to commend" in concurrence/dissent's exclusive focus on elements of Washington robbery alone, nonetheless concluding that Duenas-Alvarez forecloses that approach).

C. Sentence reduction for time in federal custody serving state sentence

In his written sentencing memorandum, Carr asked the district judge to reduce his federal sentence so as to effectively give him credit for the time for the roughly 15-month period commencing with his July 8, 2020 arrest by local authorities on state charges, continuing through his September 1, 2020 transfer from the Cook County Correctional Center to the Illinois Department of Corrections on the state parole violation, and through his transfer from state to federal detention on September 28, 2020 pursuant to the writ of habeas corpus ad prosequendum, and concluding on October 19, 2021 with the completion of his state parole term. Prior to his transfer to federal detention on the federal felon-in-possession charge, of course, Carr was exclusively in local and state detention, and that time would not be credited against his federal sentence; but even after his transfer to federal detention pursuant to the writ of habeas corpus ad prosequendum, although Carr was nominally in federal custody at that point, as a matter of law, he remained in the primary custody of state officials pending the completion of his (revoked) state parole, and consequently that period too would not be credited against his federal sentence. "A writ of habeas corpus ad prosequendum enables [the receiving jurisdiction] to take temporary custody of a prisoner confined within another jurisdiction, and indict, prosecute and sentence such [a] prisoner," Flick v. Blevins, 887 F.2d 778, 781 (7th Cir. 1989) (per curiam), while "the sending state retains full jurisdiction over the prisoner," id. "[F]ederal custody commences only when the state authorities relinquish the prisoner on satisfaction of the state obligation." United States v. Evans, 159 F.3d 908, 912 (4th Cir. 1998); see also Sinito v. Kindt, 954 F.2d 467, 469 (7th Cir. 1992) (per curiam) ("The issuance of the writ of habeas corpus ad prosequendum did not alter [the prisoner's] custody status. It merely changed the location of his custody for the sentence he was already serving."). Carr's position was that it was appropriate to in some way credit this time against his federal sentence because local authorities had arrested him and state authorities had revoked his parole based on the same offense conduct (his unlawful possession of a gun) underlying the federal felon-in-possession charge. Indeed, Carr argued, had federal authorities issued the writ before he entered into the custody of the Illinois Department of Corrections on the parole violation, all of the time he had spent in federal pretrial confinement would automatically be counted toward his federal sentence. But because Carr instead entered state custody before the federal writ was issued, the federal Bureau of Prisons (which is charged with the responsibility to calculate a defendant's sentence, including any credit for pretrial detention) would view this 15-month period as the completion of a distinct, state-imposed obligation, and it would only credit Carr for the time spent in federal custody after October 19, 2021. See 18 U.S.C. § 3585(b) ("[a] defendant shall be given credit toward the service of a term of imprisonment for any time he has spent in official detention prior to the date the sentence commences ... that has not been credited against another sentence") (emphasis added); Manuel v. Terris, 803 F.3d 826, 828 (7th Cir. 2015); Flick, 887 F.2d at 781. Carr therefore asked the judge to reduce whatever sentence he decided to impose by 15 months as a way to effectively credit Carr for this period. See Manuel, 803 F.3d at 828-29 (noting that "only the Bureau [of Prisons] is authorized to give credit against a prison sentence for prior time spent in custody") (citing United States v. Wilson, 503 U.S. 329, 333-36 (1992)); United States v. Whitlow, 740 F.3d 433, 440 (7th Cir. 2014) (district court has discretion to reduce defendant's sentence to account for time spent in custody prior to commencement of sentence that Bureau of Prisons will not otherwise credit) (collecting cases).

Carr's counsel advocated for the same relief at the sentencing hearing. But the district judge declined to entertain this request. The judge remarked that only the Bureau of Prisons had the authority to credit Carr for his pretrial detention, see id., and that as a matter of course, the Bureau would give Carr credit for the time he had spent in federal custody on the felon-in-possession charge. It appears that the judge did not appreciate that Carr was seeking a form of recognition for a segment of his federal detention that the Bureau would count as time spent in state rather than federal custody. Nor did the judge appear to understand that Carr was asking the court to reduce his sentence as an alternative means of granting him credit for a time period that the Bureau of Prisons itself would not grant him.¹⁶

16 We view this case as being distinct from United States v. Kimble, 2023 WL 2259153 (7th Cir. Feb. 28, 2023) (non-precedential decision), which the government has cited to us pursuant to Federal Rule of Appellate Procedure 28(j). The issue in Kimble was whether the district court had committed procedural error by failing to address one of the defendant's principal arguments in mitigation when it passed over in silence the defendant's request for a sentence reduction to account for the time he had spent in custody before beginning his federal sentence. We held that the district court had committed no such error, because although that request had been presented in the defendant's sentencing memorandum, defense counsel had only mentioned it in passing in one sentence of her remarks at sentencing. Id. at *3. We added that counsel's use of the word "credit" to describe the request was "somewhat misleading," because only the Bureau of Prisons could credit the defendant for time spent in pre-sentencing custody, and what the defendant was asking for was a reduction in his sentence. In this case, by contrast, there is no question that Carr presented his request for a sentence reduction as a major argument in mitigation in both his written sentencing memorandum and in his counsel's oral remarks at sentencing. (And the district court did, in fact, address the request.) It is true that Carr's counsel used the term "credit" in describing the relief Carr was asking for, but we are satisfied that his memorandum and his counsel's remarks made sufficiently clear that he was actually asking for a reduction in his sentence. The error here was that the district court nonetheless misapprehended the nature of Carr's request.

Although the government rightly argues that the district court was powerless to grant Carr formal sentencing credit for the full length of his pre-sentence detention (a credit is something that only the Bureau of Prisons may grant), it does not dispute that the court had the authority and discretion to recognize the time Carr spent serving out the remainder of his revoked parole for the gun possession as a mitigating factor that might warrant a reduction in whatever prison term the court ordered Carr to serve. See Gov't Br. 55; Whitlow, 740 F.3d at 440.

Because the district court evidently did not appreciate the true nature of Carr's request, we will vacate his sentence and remand to the district court for the limited purpose of addressing Carr's request for a sentence reduction. We express no opinion as to whether Carr is entitled to such a reduction. That request is addressed to the district court's discretion in the first instance.

III.

For the foregoing reasons, we reject Carr's contentions that his prior convictions for Illinois armed robbery were a categorical mismatch for generic robbery and could not be considered as predicates for an enhanced offense level pursuant to section 2K2.1(a)(1) of the Sentencing Guidelines; we affirm the district court's sentencing decision in that respect. However, we vacate and remand the sentence for the limited purposes of reconsidering Carr's request for a sentencing reduction to account for the time he initially spent in Cook County detention on state charges and thereafter while serving out the remainder of his revoked parole term on the prior state convictions for armed robbery.

AFFIRMED IN PART, VACATED IN PART, and REMANDED.