938 F.3d 851

7th Cir.

2019

TODD R. CHAZEN, Petitioner-Appellee, v. MATTHEW MARSKE, Respondent-Appellant.

No. 18-3268

United States Court of Appeals For the Seventh Circuit

ARGUED FEBRUARY 8, 2019 — DECIDED SEPTEMBER 9, 2019

938 F.3d 851

Appeal from the United States District Court for the Western District of Wisconsin. No. 3:17-cv-447 — James D. Peterson, Chief Judge.

Before FLAUM, BARRETT, and SCUDDER, Circuit Judges.

SCUDDER, Circuit Judge. A federal jury in Minnesota convicted Todd Chazen of possessing a firearm following a prior felony conviction. The district court then sentenced Chazen pursuant to the Armed Career Criminal Act, which mandates a minimum 15-year sentence for a defendant who unlawfully possesses a firearm and has three prior convictions for a serious drug offense or violent felony. After an unsuccessful direct appeal and petition for post-conviction relief under 28 U.S.C. § 2255, Chazen turned to 28 U.S.C. § 2241 and sought a writ of habeas corpus, arguing that the application of a recent Supreme Court decision shows he no longer qualifies as an armed career criminal and is entitled to a lesser sentence. The district court agreed and granted habeas relief. We affirm.

I

A. Chazen‘s Sentencing and Direct Appeal

Following a 2011 trial in the District of Minnesota, a jury convicted Todd Chazen of being a felon in possession of a firearm, in violation of 18 U.S.C. § 922(g)(1). While this offense ordinarily carries a ten-year maximum sentence, 18 U.S.C. § 924(a)(2), the district court considered whether Chazen qualified for an enhanced sentence under the Armed Career Criminal Act.

The Act mandates a 15-year minimum sentence if a defendant has three or more prior convictions for a “violent felony” or “serious drug offense.” 18 U.S.C. § 924(e)(1). The Act defines violent felony as any federal or state felony that “has as an element the use, attempted use, or threatened use of physical force against the person of another” (the elements clause) or is “burglary, arson, or extortion” (the enumerated offenses clause). Id. § 924(e)(2)(B). At the time of Chazen‘s sentencing, the definition also included a residual clause, which encompassed any felony that “otherwise involves conduct that presents a serious potential risk of physical injury to another.” Id. In Johnson v. United States, the Supreme Court held the residual clause void for vagueness. 135 S. Ct. 2551, 2557 (2015).

Chazen has five felony convictions under Minnesota law: second-degree assault; second-degree manufacture of a controlled substance; escape from custody; and two convictions for second-degree burglary. At sentencing the government conceded that because Chazen‘s two burglary convictions occurred on the same day and involved the same course of conduct, they should not be counted as separate predicate convictions under the Armed Career Criminal Act. The government further conceded that Chazen‘s controlled substances conviction did not qualify as a serious drug offense within the meaning of the Act. But it nonetheless advocated for an enhanced sentence on the basis of Chazen‘s three remaining convictions—for assault, escape, and burglary. For his part, Chazen agreed that one of his burglary convictions and his assault conviction were qualifying felonies, but argued that he was not an armed career criminal because his escape conviction did not count as a violent felony predicate.

The district court in Minnesota sided with the government and sentenced Chazen to 21 years’ imprisonment. In doing so, the court did not specify which convictions qualified Chazen as an armed career criminal, noting only that “there [were] a number of possibilities” and finding (without further elaboration) that Chazen had “at least four” predicates.

On direct appeal in the Eighth Circuit, Chazen challenged his enhanced sentence, arguing that the district court erred by concluding that his Minnesota escape conviction was a qualifying felony under the Armed Career Criminal Act‘s residual clause. Relying on the government‘s concessions at sentencing, he maintained that without the improper consideration of his escape conviction, he no longer had three qualifying offenses. The government responded by withdrawing its concessions about Chazen‘s burglary convictions and his drug conviction, asserting that any error with his escape conviction was harmless because, even without it, at least three qualifying convictions remained. It also contended that Chazen‘s escape argument was directly foreclosed by Eighth Circuit precedent. The Eighth Circuit agreed with the government and affirmed Chazen‘s sentence, concluding that the escape conviction qualified as a predicate under the Act‘s residual clause. See United States v. Chazen, 469 F. App‘x 508, 509 (8th Cir. 2012).

B. Chazen‘s § 2255 Petitions

In 2013, Chazen invoked 28 U.S.C. § 2255 and petitioned for post-conviction relief in the District of Minnesota, again challenging his classification as an armed career criminal. Chazen persisted in his contention that the sentencing court erred in determining that his escape conviction qualified as a violent felony predicate. He also argued that the court was unable to rely on his drug conviction or one of his burglary convictions given the government‘s concessions at sentencing. The district court denied relief, concluding that the escape conviction qualified as a violent felony and finding Chazen‘s other arguments procedurally defaulted.

Two years after Chazen‘s unsuccessful § 2255 petition, the legal landscape shifted. First, the Supreme Court held that the Act‘s residual clause was unconstitutionally vague. See Johnson, 135 S. Ct. at 2557. Because Chazen‘s escape conviction was a residual-clause offense, he sought authorization from the Eighth Circuit to file a second § 2255 petition for post-conviction relief. He simultaneously proceeded to file a second § 2255 petition in the sentencing court (the District of Minnesota), arguing that because Johnson knocked out his escape conviction, he no longer qualified as an armed career criminal. The government agreed that Chazen‘s escape conviction no longer qualified as a predicate following Johnson. But the government maintained that Chazen still had the three strikes requisite to qualify as an armed career criminal owing to his Minnesota assault conviction and two burglary convictions. In May 2016, the Eighth Circuit summarily denied Chazen‘s application to file a second § 2255 petition. The Minnesota district court then concluded it lacked jurisdiction to consider his § 2255 petition.

C. Chazen‘s § 2241 Petition

The landscape shifted a second time in June 2016, when the Supreme Court decided Mathis v. United States, 136 S. Ct. 2243. Mathis was meaningful because it narrowed the range of state statutes that qualify as violent felony predicates under the Armed Career Criminal Act. Seeing this development, Chazen again pursued post-conviction relief, this time by filing a petition under 28 U.S.C. § 2241 in the Western District of Wisconsin, where he is serving his sentence.

In response to this new petition, the parties agreed that, due to intervening changes in the law, only three of Chazen‘s prior Minnesota convictions potentially qualified him as an armed career criminal: his second-degree assault conviction and his two second-degree burglary convictions. Relying on Mathis, Chazen argued that his burglary convictions no longer counted as violent felonies under the Act‘s enumerated offense clause. He also pointed to the Eighth Circuit‘s decision in United States v. McArthur, 850 F.3d 925, 940 (8th Cir. 2017), which held that the Minnesota crime of third-degree burglary no longer qualifies as a predicate under the Act after the Supreme Court‘s decision in Mathis.

Another legal development occurred in 2018. While Chazen‘s § 2241 petition was pending in the district court (in the Western District of Wisconsin, where he is confined), this court joined the Eighth Circuit in holding that because the Minnesota burglary statute covers more conduct than generic burglary, it does not qualify as a predicate violent felony under the Act. See Van Cannon v. United States, 890 F.3d 656, 665 (7th Cir. 2018). We reached this conclusion with respect to second-degree burglary, but the parties agree that for purposes of this appeal, there is no meaningful distinction between second- and third-degree burglary under Minnesota law.

Relying on both Mathis and Van Cannon, the district court granted Chazen‘s § 2241 petition, concluding that his burglary convictions no longer qualified as violent felony predicates. This left Chazen with only one qualifying predicate (his Minnesota assault conviction), which was not enough for an enhanced sentence under the Act. Accordingly, the district court ordered resentencing.

As part of affording Chazen habeas relief under § 2241, the district court took care to address a procedural point that defines much of the battleground in this appeal. The court determined that relief was available to Chazen under § 2241 because, at the time of his original § 2255 petition in 2013, Eighth Circuit precedent foreclosed any contention that his two prior Minnesota burglary convictions did not qualify as violent felony predicates under the Act. Being foreclosed like this meant that, following Mathis, Chazen was able through the so-called savings clause in § 2255(e) to pursue relief pursuant to § 2241.

In granting relief, the district court (in Wisconsin) transferred Chazen‘s case back to the District of Minnesota (the district of conviction) for resentencing. The government now appeals.

II

At this point, all agree that the validity of Chazen‘s sentence under the Armed Career Criminal Act depends on whether his two prior Minnesota burglary convictions qualify as generic burglary under the Act‘s enumerated offense clause. They take different positions on the merits of that question while also spilling substantial ink on the very knotty procedural issue of whether Chazen can pursue relief under § 2241. The complexity of the procedural question is staggering.

A. The Savings Clause and § 2241

As a general rule, a federal prisoner wishing to collaterally attack his conviction or sentence must do so under § 2255 in the district of conviction. See Light v. Caraway, 761 F.3d 809, 812 (7th Cir. 2014). If, however, § 2255 is “inadequate or ineffective to test the legality of [the prisoner‘s] detention,” relief may be granted under § 2241, the general federal habeas corpus statute, in the district of incarceration. 28 U.S.C. § 2255(e); see also Light, 761 F.3d at 812. In In re Davenport, we examined the scope of this provision, known as the “savings clause,” and concluded that whether § 2255 was “inadequate or ineffective” turns on whether a petitioner had “a reasonable opportunity [in a prior § 2255 proceeding] to obtain a reliable judicial determination of the fundamental legality of his conviction and sentence.” 147 F.3d 605, 609 (7th Cir. 1998); see also Webster v. Daniels, 784 F.3d 1123, 1136 (7th Cir. 2015) (en banc) (reiterating that Davenport is the law of this circuit). In considering Chazen‘s appeal, we are bound by Davenport.

In Davenport‘s wake, we have established a three-part test to determine whether a petitioner satisfies § 2255(e)‘s savings clause. See, e.g., Montana v. Cross, 829 F.3d 775, 784 (7th Cir. 2016). To pursue relief under § 2241, a petitioner must establish that “(1) the claim relies on a statutory interpretation case, not a constitutional case, and thus could not have been invoked by a successive § 2255 motion; (2) the petitioner could not have invoked the decision in his first § 2255 motion and the decision applies retroactively; and (3) the error is grave enough to be deemed a miscarriage of justice.” Beason v. Marske, 926 F.3d 932, 935 (7th Cir. 2019).

Everyone agrees that Chazen satisfies the first and third requirements. On the first prong, Chazen argues that the direction recently provided by the Supreme Court in Mathis—a decision interpreting the Armed Career Criminal Act—means that his Minnesota burglary convictions no longer qualify as violent felonies. And, as to the third prong for pursuing relief under § 2241, we have held that a defendant sentenced in error as an armed career criminal satisfies the “miscarriage of justice” requirement. See Light, 761 F.3d at 813.

The disagreement comes on the middle prong of the savings clause test and whether Chazen was foreclosed in his prior § 2255 petition from asserting the claim he now raises. This question cannot be answered without evaluating the state of the law at that time and how it has changed. In this way, then, substance and process bleed together. Our analysis proceeds in two steps. First, we assess whether his second-degree burglary convictions under Minnesota law no longer qualify as violent felonies in light of Mathis. Second, we consider whether the law in place at the time of his prior § 2255 petition foreclosed the Mathis-based argument he now advances to challenge his enhanced sentence. Chazen is entitled to relief under § 2241 only if he can get to yes on both questions.

B. “Violent Felony” Under the Act

Recall that the Armed Career Criminal Act includes “burglary” in its list of enumerated violent felonies. See § 924(e)(2)(B)(ii). While the statute does not define the offense, the Supreme Court has held that Congress used “burglary” to refer to the generic version of the crime, meaning “an unlawful or unprivileged entry into, or remaining in, a building or other structure, with intent to commit a crime.” Taylor v. United States, 495 U.S. 575, 598 (1990).

To determine whether a prior conviction meets this definition, Taylor instructed courts to use what is known as the “categorical approach” by “look[ing] only to the statutory definitions of the prior offenses, and not to the particular facts underlying those convictions.” Id. at 600. If the statute under which a defendant was previously convicted “substantially corresponds” to or is narrower than Taylor‘s generic definition, the conviction qualifies as a violent felony under the Act. Id. at 602. Put another way, “[a] crime counts as ‘burglary’ under the Act if its elements are the same as, or narrower than, those of the generic offense.” Mathis, 136 S. Ct. at 2248. But if “the statute sweeps more broadly than the generic crime, a conviction under that law cannot count as an [Armed Career Criminal Act] predicate, even if the defendant actually committed the offense in its generic form.” Descamps v. United States, 570 U.S. 254, 261 (2013).

In so interpreting the Act, the Supreme Court has recognized a “narrow range of cases” in which sentencing courts are permitted to look beyond the statute of conviction to determine whether a prior conviction meets Taylor‘s generic definition of burglary. Taylor, 495 U.S. at 602. If a statute is divisible—meaning it “sets out one or more elements of the offense in the alternative“—courts may apply the so-called “modified categorical approach.” Descamps, 570 U.S. at 257. Under this approach, a sentencing court may “consult a limited class of documents, such as indictments or jury instructions, to determine which alternative formed the basis of the defendant‘s prior conviction,” id., “but only for the limited purpose of determining whether the elements of the crime of conviction match (or are narrower than) the elements of the generic offense.” Van Cannon, 890 F.3d at 663.

The Supreme Court has underscored an important qualifier: the modified categorical approach applies only to divisible statutes. The approach does not apply if a statute is indivisible—where the statute of conviction does not contain alternative elements, but rather “a single, ‘indivisible’ set of elements sweeping more broadly than the corresponding generic offense.” Descamps, 570 U.S. at 260. When faced with a statute that “defines a single offense, albeit one with multiple modes of conviction,” a sentencing court is confined to the statutory elements and therefore cannot rely on any documents from a defendant‘s criminal case or otherwise embark on fact finding to discern the defendant‘s actual offense conduct. Van Cannon, 890 F.3d at 663. If the analysis shows that “the alternative means listed in an indivisible statute cover a broader swath of conduct than the generic offense, then a conviction under the statute doesn‘t count as an [Armed Career Criminal Act] predicate.” Id.

The legal upshot, then, is that the question of whether a court may employ the modified categorical approach to an overbroad statute turns on the issue of divisibility. In the aftermath of Taylor and Descamps, lower courts grappled with how to determine whether a statute is divisible, a distinction that hinges on the often subtle and hairline difference between elements and means. See Najera-Rodriguez v. Barr, 926 F.3d 343, 348 (7th Cir. 2019) (“The difference between ‘elements’ and ‘means’ can seem slippery, sometimes almost metaphysical, but significant legal consequences flow from that difference.“).

Adding to the ambiguity, some circuits interpreted a footnote in Descamps to permit sentencing courts to continue to apply the modified categorical approach—by reviewing the record of conviction—whenever a statute lists alternative ways to commit a crime, even if the alternatives were means rather than elements. See Descamps, 570 U.S. at 264 n.2 (noting that “[w]hatever a statute lists (whether elements or means), the documents we approved in Taylor and Shepard—i.e., indictment, jury instructions, plea colloquy, and plea agreement—would reflect the crime‘s elements“). The Eighth Circuit, for example, read Descamps as “explicitly reject[ing]” any “means/elements distinction,” and cited footnote 2 for the proposition that courts “must apply the modified categorical approach” whether a statute listed alternative elements or alternative means. United States v. Mathis, 786 F.3d 1068, 1074, 1075 (8th Cir. 2015), rev‘d, 136 S. Ct. 2243 (2016). While the Sixth and Tenth Circuits adopted the same approach, see, e.g., United States v. Ozier, 796 F.3d 597, 600 (6th Cir. 2015); United States v. Trent, 767 F.3d 1046, 1057 (10th Cir. 2014), other circuits reached a contrary conclusion. See, e.g., Rendon v. Holder, 764 F.3d 1077, 1086 (9th Cir. 2014) (citing Descamps for the precept that “it is black-letter law that a statute is divisible only if it contains multiple alternative elements, as opposed to multiple alternative means“); Omargharib v. Holder, 775 F.3d 192, 198 (4th Cir. 2014) (explaining that “a crime is divisible under Descamps only if it is defined to include multiple alternative elements (thus creating multiple versions of a crime), as opposed to multiple alternative means (of committing the same crime)“).

None of this is surprising. The law in this area is complex in the extreme; unexpected difficulties and nuances surface in new cases; and it often takes the Supreme Court deciding additional cases for clarity and stability to begin to emerge. Law develops iteratively like this in many areas, which is exactly what Chazen saw happen here.

To that end, in Mathis v. United States, the Supreme Court granted review to clarify when a statute qualifies as “divisible” and to resolve the circuit split that had developed over the “elements” versus “means” distinction in the context of predicate offenses under the Armed Career Criminal Act. See 136 S. Ct. at 2251.

At issue in Mathis was whether Iowa‘s burglary statute qualified as a violent felony predicate. See id. at 2250. The parties agreed that the statute extended to conduct beyond generic burglary because it criminalized unlawful entry into a broader range of places, including “any building, structure, [or] land, water, or air vehicle.” Id. (quoting Iowa Code § 702.12 (2013)). The debate centered on whether the sentencing court erred in consulting Mathis‘s conviction records to determine which type of burglary he committed and, relatedly, whether the different locations in the Iowa statute amounted to alternative means or alternative elements. See id.

On this latter question, the Eighth Circuit had concluded this was a distinction without a difference, explaining that “[w]hether [the different locations] amount[ed] to alternative elements or merely alternative means to fulfilling an element, the statute is divisible, and [the court] must apply the modified categorical approach.” Mathis, 786 F.3d at 1075. In so holding, the Eighth Circuit added to the circuit split rooted in confusion about what makes a statute divisible for purposes of applying the modified categorical approach.

The Supreme Court resolved the split by rejecting the Eighth Circuit‘s approach—which other courts had also adopted—and held that the modified categorical approach is unavailable when, as with the Iowa law, a statute lists multiple “means” to satisfy just one element. Mathis, 136 S. Ct. at 2253. In reaching this conclusion, the Court stressed that “[d]istinguishing between elements and facts is ... central to [the Act‘s] operation,” id. at 2248, because if a statute lists alternative means (rather than elements) it is not divisible, and, as a consequence, a court may not rely on the modified categorical approach to determine which of the alternatives formed the basis of a defendant‘s conviction. See id. at 2257. Instead, the Court explained, the sentencing court is confined to the text of the statute and if, as with burglary under Iowa law, any of the alternative means of satisfying a statutory element is broader than generic burglary, the conviction cannot qualify as a predicate under the Armed Career Criminal Act. See id. at 2248.

The Court also provided guidance for distinguishing between elements and means, explaining that “elements” are a crime‘s “constituent parts,” which a jury must find beyond a reasonable doubt or a defendant must admit if he pleads guilty. Id. at 2248. “Means,” on the other hand, are merely “various factual ways of committing some component of the offense—a jury need not find (or a defendant admit)” them. Id. at 2249. A prior conviction counts as generic burglary under the Act only if its elements—not the means of satisfying the elements—are the same as or narrower than the generic definition of burglary. See id. at 2256.

C. Chazen‘s Minnesota Burglary Convictions

Returning to the facts here, Chazen argues that his Minnesota second-degree burglary convictions no longer qualify as “violent felonies” in light of Mathis. We recently considered the Minnesota burglary statute in Van Cannon v. United States and reached this exact conclusion. 890 F.3d 656, 665 (7th Cir. 2018).

Our analysis proceeded within the framework the Supreme Court supplied in Mathis. We first considered whether the statute contained alternative elements—in which case the statute is divisible, and the modified categorical approach is available—or alternative means of committing a single element—which ends the inquiry. The Minnesota legislature defined second-degree burglary this way: “Whoever enters a building without consent and with intent to commit a crime, or enters a building without consent and commits a crime while in the building, either directly or as an accomplice, commits burglary in the second degree” if certain aggravating factors are present. Minn. Stat. § 609.582(2)(a) (emphasis added). We concluded that “[t]he disjunctive phrasing describes two different factual ways“—or means—“of committing a single crime,” meaning the statute is indivisible. Van Cannon, 890 F.3d at 664. “Because the statute is indivisible,” we explained, “a conviction under it does not count as an [Armed Career Criminal Act] predicate if one of the listed alternatives is broader than generic burglary.” Id.

We concluded that the Minnesota statute contained such overbreadth. Specifically, we determined that while the first alternative qualifies as generic burglary—a person commits second-degree burglary if he “enters a building without consent and with intent to commit a crime,” Minn. Stat. § 609.582(2)(a)—the second alternative criminalizes conduct that is broader than Taylor‘s definition of generic burglary because it ”doesn‘t require proof of intent to commit a crime at all—not at any point during the offense conduct.” Id. (emphasis in the original).

After Van Cannon, then, our law is clear that a conviction for Minnesota second-degree burglary does not qualify as a predicate violent felony under the Armed Career Criminal Act. Nothing about this appeal requires us to revisit that holding.

But we do need to take a breath and address a procedural point. Remember that Chazen committed his felon-in-possession offense and was convicted and sentenced in the District of Minnesota—in other words, within the Eighth Circuit. But he is currently incarcerated in the Western District of Wisconsin—in our circuit. These locations matter, for it is well-established that if relief under § 2241 is available at all, it must be sought in the district of incarceration. See Light, 761 F.3d at 812. Here, then, that means Chazen had to file his § 2241 petition in a district court in this circuit, which he did by going to the Western District of Wisconsin. What is less clear, however, is whether, in evaluating the merits of Chazen‘s petition, we should apply our own precedent or the precedent of the circuit of conviction. See Shepherd v. Krueger, 911 F.3d 861, 863 (7th Cir. 2018) (explaining that the parties debate “whether § 2241 should be available to [petitioner] at all, and if so whether this court should apply our own precedent or Sixth Circuit precedent,” but “elect[ing] to bypass these procedural hurdles for relief because this case can be resolved most simply on the merits“).

We need not decide this issue here because the government took the position in the district court that the law of this circuit governs the merits of Chazen‘s claim. See, e.g., Prevatte v. Merlak, 865 F.3d 894, 898, 901 (7th Cir. 2017) (accepting the government‘s concession on an element of the savings clause test and explaining that § 2241 and § 2255 address remedies, not jurisdiction). Given this concession, we see no reason why our holding in Van Cannon does not resolve the merits of Chazen‘s claim.

Our course of analysis is unaffected by the Supreme Court‘s most recent decision in Quarles v. United States, 139 S. Ct. 1872 (2019). In Quarles, the Court considered the “exceedingly narrow question” whether Taylor‘s definition of generic burglary requires proof that a defendant intended to commit a crime at the time he unlawfully entered, or instead whether it is sufficient that he formed the intent to commit a crime at some point while unlawfully “remaining in” the building or structure. Id. at 1875. The Court held that “generic remaining-in burglary occurs under § 924(e) when the defendant forms the intent to commit a crime at any time while unlawfully remaining in a building or structure.” Id. at 1880. As the government has pointed out in its supplemental briefing, this holding may well abrogate the Eighth Circuit‘s decision in McArthur, which rested on the “contemporaneous intent” reasoning rejected in Quarles. See McArthur, 850 F.3d at 939. But the Eighth Circuit has not resolved this question. See Raymond v. United States, No. 18-2349, 2019 WL 3783103, at *3 (8th Cir. Aug. 13, 2019) (observing without deciding that the Supreme Court‘s decision in Quarles may have abrogated McArthur).

What we can say with confidence is that Quarles did not abrogate Van Cannon‘s conclusion that Minnesota burglary is broader than generic burglary because the state statute does not require proof of any intent at any point. Indeed, the Court expressly declined to address this issue in Quarles. See 139 S. Ct. at 1880 n.2.

What all of this means, then, is that Chazen‘s Minnesota burglary convictions no longer qualify as violent felony predicates under the reasoning of Van Cannon. And with there being no contrary law in the Eighth Circuit, we believe Chazen has done enough to show that he no longer qualifies as an armed career criminal.

D. Chazen‘s Eligibility for § 2241 Relief

Reaching this conclusion on the merits of Chazen‘s claim, however, is not the end of the inquiry. The reason is because Chazen is pursuing relief under § 2241 and his ability to do so depends on establishing that he was unable in his prior § 2255 proceedings to advance the arguments he now raises to challenge his sentence. See Beason, 926 F.3d at 935 (explaining the requirements for pursuing relief under § 2241 via the savings clause in § 2255(e)). On this point, the analysis encounters yet additional complexity.

The government insists that Chazen cannot rely on the savings clause to pursue relief under § 2241 because his claim does not rely on a “new, retroactive rule.” Gov‘t Br. at 16. In its view, because Mathis did not announce a substantive change in the law, but rather clarified the circumstances under the Armed Career Criminal Act in which a sentencing court may apply the modified categorical approach, Chazen could have raised his arguments about Minnesota‘s burglary statute in his initial § 2255 petition by relying on pre-Mathis Supreme Court decisions like Taylor and Descamps. As the government sees it, then, because Chazen was not foreclosed from raising these arguments in his initial collateral attack, he cannot now bring his claims under § 2241.

As an initial matter, the government conceded in the district court that Mathis is retroactive. We have likewise suggested (without deciding) that Mathis is retroactive. See Holt v. United States, 843 F.3d 720, 722 (7th Cir. 2016) (observing that “substantive decisions such as Mathis presumptively apply retroactively on collateral review“). But assessing the remainder of the government‘s position—that Chazen may not rely on Mathis because it did not announce a “new rule“—is complicated by the fact that we have not been consistent in our articulation of the second requirement of the savings clause test. While we have repeatedly stressed that a petitioner seeking to invoke the savings clause must establish that he was unable to raise his statutory claim at the time of his original § 2255 petition—thus showing that § 2255 was “inadequate or ineffective to test the legality of [the prisoner‘s] detention“—we have “employed various formulations of this inquiry.” Montana, 829 F.3d at 784. In some instances, we have said that satisfying this condition requires a petitioner to show that he is relying on a “new rule” that applies “retroactively to cases on collateral review and could not have been invoked in his earlier proceeding.” Camacho v. English, 872 F.3d 811, 813 (7th Cir. 2017); see also Davis v. Cross, 863 F.3d 962, 964 (7th Cir. 2017) (same); Prevatte, 865 F.3d at 897 (same).

Yet we have never explained what “new” means or requires in this context.

In other instances, we have articulated the requirement for satisfying the savings clause, not in terms of a “new rule,” but rather in terms of whether the petitioner “relies on a retroactive decision that he could not have invoked in his first § 2255 motion.” Brown v. Caraway, 719 F.3d 583, 586 (7th Cir. 2013); see also Beason, 926 F.3d at 935 (same); Poe v. LaRiva, 834 F.3d 770, 772–73 (7th Cir. 2016) (same); Light, 761 F.3d at 812–13; Brown v. Rios, 696 F.3d 638, 640 (7th Cir. 2012). More to it, we have held that a petitioner seeking relief under § 2241 need only show that the case on which he relies had not yet been decided at the time of his § 2255 petition. See, e.g., Rios, 696 F.3d at 640. Under this formulation, Chazen would meet the second criteria because the Supreme Court decided Mathis two years after his initial § 2255 petition.

In still other cases, we have employed a “slightly higher standard,” requiring a petitioner to show not only that he relies on a newly decided case of statutory interpretation, but also that, at the time of his initial § 2255 petition, his claim was “foreclosed by binding precedent” in the circuit of his conviction. Brown, 719 F.3d at 595 (quoting Hill v. Werlinger, 695 F.3d 644, 648 (7th Cir. 2012)). And most recently we have explained that “if it ‘would have been futile’ for a petitioner to raise these arguments in his § 2255 motion because the ‘law was squarely against him,’ then the savings clause applies and [a petitioner] may proceed and pursue resentencing under § 2241.” Beason, 926 F.3d at 936 (quoting Webster, 784 F.3d at 1136).

Stepping back, we can plant our feet in this appeal on easier ground. There is no doubt that at the time of Chazen‘s § 2255 motion in 2013, his claim for relief was foreclosed by Eighth Circuit precedent concluding that Minnesota burglary qualified as a violent felony for federal sentencing purposes. See United States v. LeGrand, 468 F.3d 1077 (8th Cir. 2006). In LeGrand the Eighth Circuit had held that the Minnesota burglary statute “includes the elements of a generic burglary as outlined in Taylor.” Id. at 1081–82. LeGrand also held that the statute fit within the Act‘s residual clause because “burglary always has a ‘serious potential risk of physical injury to another.‘” Id. at 1082 (quoting United States v. Mohr, 407 F.3d 898, 901 (8th Cir. 2005)). While LeGrand assessed the statute in the context of the Sentencing Guidelines definition of a “crime of violence,” the Eighth Circuit treated cases on one as binding on the other. See United States v. Sonczalla, 561 F.3d 842, 846 (8th Cir. 2009) (applying LeGrand‘s analysis of the Sentencing Guidelines to the Armed Career Criminal Act‘s language). And in 2012, the Eighth Circuit reiterated its conclusion that Minnesota burglary qualifies as a “violent felony” for purposes of the Act. See United States v. Constantine, 674 F.3d 985, 990 (8th Cir. 2012). In short, in 2013—at the time Chazen first moved for post-conviction relief under § 2255—“the law was squarely against” him in that it foreclosed the position he currently advances—that Minnesota burglary is not a violent felony under the Act.

We also conclude that Mathis can provide the basis for Chazen‘s § 2241 petition. While our caselaw is not a model of clarity with respect to whether a petitioner must rely on a “new rule,” the thrust of our precedent has focused on whether an intervening case of statutory interpretation opens the door to a previously foreclosed claim. Mathis fits the bill. Mathis injected much-needed clarity and direction into the law under the Armed Career Criminal Act: before Mathis, the Eighth Circuit had held that Minnesota burglary was a violent felony and it applied the modified categorical approach regardless of whether a statute listed distinct elements or alternative means. See, e.g., Mathis, 786 F.3d at 1075. It is only after Mathis—a case decided after Chazen‘s § 2255 petition that the government concedes is retroactive—that courts, including our court and the Eighth Circuit, have concluded that Minnesota burglary is indivisible because it lists alternative means of committing a single crime. See Van Cannon, 890 F.3d at 664; McArthur, 850 F.3d at 938.

In this way, Mathis is “new” as a functional and practical matter for federal inmates seeking relief from a mandatory minimum sentence under the Act. This is particularly true for those petitioners, like Chazen, who were convicted in the Eighth Circuit, whose interpretation of “divisibility” was overruled by Mathis. See, e.g., Brown v. United States, 929 F.3d 554, 559 (8th Cir. 2019) (explaining that pre-Mathis cases holding that a conviction under the Armed Career Criminal Act qualifies as a violent felony are no longer controlling because “[c]ontrary to the approach [the Eighth Circuit] took in those decisions, Mathis clarified that if any of the means by which a defendant can commit a crime fall outside the definition of the generic offense, then it is not a violent felony, even if the defendant actually committed the generic crime“).

Before Mathis it would have been futile for Chazen to argue that his Minnesota burglary convictions did not qualify as violent felonies. It was only after the Supreme Court decided Mathis that Chazen had any daylight to seek relief by raising the claim embodied in his § 2241 petition. In these circumstances, where the government has conceded that Mathis is retroactive and Chazen was so clearly foreclosed by the law of his circuit of conviction at the time of his original § 2255 petition, we conclude that Chazen has done enough to satisfy the savings clause requirements. This conclusion finds support in our prior observation that “[a]n independent claim based on Mathis must be brought, if at all, in a petition under 28 U.S.C. § 2241.” Dawkins v. United States, 829 F.3d 549, 551 (7th Cir. 2016). And in light of our prior decision in Van Cannon, it is clear that Chazen‘s two Minnesota burglary convictions no longer qualify as predicates under § 924(e).

III

For these reasons, we AFFIRM the district court‘s judgment awarding Chazen habeas relief under 28 U.S.C. § 2241.

BARRETT, Circuit Judge, concurring. I join the panel‘s opinion because it has support in our precedent. I write separately, though, to express concern about the state of our precedent. As the opinion observes, the complexity of our cases in this area is “staggering.” We have stated the “saving clause” test in so many different ways that it is hard to identify exactly what it requires. And the resulting confusion has caused our law to drift beyond the course we set in In re Davenport. See 147 F.3d 605, 611 (7th Cir. 1998). At this point, our definition of “inadequacy” and “ineffectiveness” under § 2255(e) undermines the limits that § 2255(h) imposes on second or successive motions. Our patch for statutory cases has grown larger than the hole we identified in the statute.

Davenport aims to fix a “glitch” in § 2255(h)(2). See Brown v. Caraway, 719 F.3d 583, 587 n.1 (7th Cir. 2013) (quoting Unthank v. Jett, 549 F.3d 534, 536 (7th Cir. 2008)). That provision permits a prisoner to bring a second or successive challenge to his detention if the motion is based on “a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable.” This permission does not apply, however, to a second or successive motion based on a new statutory interpretation—perhaps because of a congressional oversight. As one noted treatise explains, Congress appears to have modeled § 2255(h)(2) on § 2244(b), which governs second and successive collateral attacks by state prisoners. HART & WECHSLER‘S THE FEDERAL COURTS AND THE FEDERAL SYSTEM 1362 (Richard H. Fallon Jr. et al. eds., 7th ed. 2015). But “Congress seems to have lost sight of the fact that federal convicts more often can raise federal statutory claims in their collateral attacks—notably in cases in which the federal criminal statute under which a prisoner was convicted has since been authoritatively interpreted more narrowly.” Id. Davenport holds that § 2255‘s failure to provide a federal prisoner “any opportunity for judicial rectification of so fundamental a defect in his conviction as having been imprisoned for a nonexistent offense” renders § 2255 “inadequate” or “ineffective” for purposes of § 2255(e)‘s saving clause. 147 F.3d at 611. Thus, a prisoner in that situation can seek a writ of habeas corpus under § 2241. Id.¹

Davenport‘s test for “inadequacy” and “ineffectiveness” largely tracks § 2255(h)(2). That makes sense. The “structural problem” in § 2255 is that § 2255(h)(2) doesn‘t authorize second or successive motions based on statutory claims. See Poe v. LaRiva, 834 F.3d 770, 773 (7th Cir. 2016) (“Where Davenport recognized a structural problem in § 2255(h) is in the fact that it did not permit a successive petition for new rules of statutory law made retroactive by the Supreme Court.“). Davenport fixes that problem by effectively giving such prisoners the relief that they would have had if § 2255(h)(2) had included them.

Our cases, however, have phrased Davenport‘s test inconsistently, and little by little, they have stopped tracking § 2255(h)(2). Most significantly, we have moved away from Davenport‘s requirement that the change in law come from the Supreme Court. 147 F.3d at 611. Sometimes we have included that requirement in our restatement of the test. See, e.g., Light v. Caraway, 761 F.3d 809, 812 (7th Cir. 2014) (allowing for “a § 2241 challenge based on a new statutory interpretation by the Supreme Court” (emphasis added)). Sometimes we have not. See, e.g., Montana v. Cross, 829 F.3d 775, 783 (7th Cir. 2016). To be sure, almost all of our post-Davenport cases have dealt with petitions based on Supreme Court decisions, so in the mine-run of cases, nothing has turned on the source of the rule.² Still, we stopped explicitly saying that the rule must come from the Supreme Court, and recently, we allowed a petitioner to obtain relief based on a circuit-level decision. Beason v. Marske, 926 F.3d 932, 935 (7th Cir. 2019). Under our circuit‘s law, therefore, a prisoner with a second or successive statutory claim can secure relief based on a court of appeals case, while a prisoner with a second or successive constitutional claim can secure relief only when the Supreme Court acts. That is an odd state of affairs.

It also raises the stakes for the choice-of-law question that we do not resolve today: which circuit‘s law applies to a Davenport petition. Chazen‘s case illustrates the point. Chazen was convicted in the Eighth Circuit, so its law governed his trial, sentencing, direct appeal, and first § 2255 motion. Even now, there is no controlling law in the Eighth Circuit on which Chazen could base a claim for relief, so a collateral attack filed there would fail.³ But a § 2241 petition is filed in the circuit of confinement, and seeking the benefit of Van Cannon v. United States, 890 F.3d 656 (7th Cir. 2018), Chazen argues that we should apply Seventh Circuit law to resolve his petition.

We should be skeptical of this argument. Applying the law of the circuit of confinement risks recreating some of the problems that § 2255 was designed to fix. See United States v. Hayman, 342 U.S. 205, 210–19 (1952) (explaining that § 2255 directs post-conviction litigation to the district of conviction to avoid the practical problems that existed before § 2255, when all litigation was brought by a writ of habeas corpus in the district of confinement); see also HART & WECHSLER at 1356 (cataloguing the issues courts faced before § 2255‘s enactment). We have never decided the issue, but district courts confronting it have concluded that the law of the circuit of conviction applies. As one court explained:

The rule ensures that the law that prevails in the judicial circuit of any federal prisoner‘s conviction, or a substantially similar law, is the law that will be applied to the prisoner‘s § 2241 petition seeking vacation of a conviction. Application of the law of the place of conviction is a consistent, reasonable rule, as is evidenced by the requirement that § 2255 motions be filed in the district of conviction. The rule [petitioner] believes appropriate—applying the substantive law of the place of confinement—is actually far more arbitrary. Such a rule would base the choice of law decision on the fortuitous placement of a prisoner by the Bureau of Prisons, not the more rational factor of the place of conviction.

Hernandez v. Gilkey, 242 F. Supp. 2d 549, 554 (S.D. Ill. 2001); see also Cano v. Warden USP–Terre Haute, No. 2:17-cv-441, 2018 WL 3389746 (S.D. Ind. July 12, 2018); Roberts v. Watson, No. 16-cv-541, 2017 WL 6375812 (W.D. Wis. Dec. 12, 2017); accord Burgess v. Williams, No. 4:18-cv-2643, 2019 WL 2641902 (N.D. Ohio June 27, 2019); Eames v. Jones, 793 F. Supp. 2d 747, 749 (E.D.N.C. 2011). This position has force.

Today‘s opinion avoids resolving the choice-of-law problem because the government conceded in the district court that Seventh Circuit law applies.⁴ I have reservations about whether we should accept this concession. While we have the discretion to accept a concession on a point of law, we are not obligated to do so. Krieger v. United States, 842 F.3d 490, 499 (7th Cir. 2016) (“Of course we are not bound to accept the government‘s concession when the point at issue is a question of law.“); see also Costello v. BeavEx, Inc., 810 F.3d 1045, 1061 n.4 (7th Cir. 2016) (rejecting a party‘s concession on a question of law). Here, I worry that accepting the concession risks giving the impression that we settled the issue. Lest there be any confusion, we have not. In a later case, this is an issue that deserves our careful consideration; it is not one that we should resolve for all cases based on the government‘s litigating position in this one.

Chazen‘s case implicates other issues that the full court should clarify at some point. For example, as today‘s opinion explains, we have articulated at least three different standards for determining whether a recently decided case qualifies as one that the petitioner could not have raised in his initial § 2255 motion. Maj. Op. at 18–19. We need to pick one. And when we do, we should focus on the fact that some of our formulations specify that the recently decided case must announce a “new rule” without explaining what this means. See, e.g., Light, 761 F.3d at 812 (asserting that a § 2241 petition must be “based on a new statutory interpretation by the Supreme Court” (emphasis added)). In particular, we have not discussed whether “newness” in the Davenport context bears any relationship to the “newness” requirements in § 2255(h)(2) and § 2255(f)(3). A “new” rule for purposes of these subsections is one that “breaks new ground” rather than clarifying established law. Headbird v. United States, 813 F.3d 1092, 1095 (8th Cir. 2016) (quoting Teague v. Lane, 489 U.S. 288, 301 (1989)) (explaining that “newness” has the same meaning in both § 2255(h)(2) and § 2255(f)(3)). Other circuits have held that Mathis, which the Supreme Court characterized as dictated by Taylor, does not satisfy this test. See, e.g., Washington v. United States, 868 F.3d 64, 65–66 (2d Cir. 2017); Arazola-Gates v. United States, 876 F.3d 1257, 1259 (9th Cir. 2017); In re Conzelman, 872 F.3d 375, 376–77 (6th Cir. 2017); United States v. Taylor, 672 F. App‘x 860, 863 (10th Cir. 2016). Thus, if the Davenport test uses “new” in the standard habeas sense that it has in § 2255(h)(2) and § 2255(f)(3), Chazen‘s claim presumably cannot succeed. See Guenther v. Williams, No. 17-cv-231, 2017 WL 5054731 (W.D. Wis. Nov. 2, 2017) (holding that a Mathis claim is not cognizable under § 2241 because Mathis applied established law rather than announcing a “new” rule).

But as Chazen points out, we have not expressly embraced this definition of “new” for Davenport petitions—and we have sometimes implied that a “new” case is nothing more than one that opens a previously foreclosed position. If that‘s right, cases like Mathis, which correct misapplications of “old” rules, qualify. Maybe that is the position that we want to take, but if so, we should expressly adopt and defend it. Doing so would require us to think through the implications of forgoing a “newness” requirement in this saving clause context. For example, that position might mean that § 2255(f)‘s statute of limitations renders § 2255 “inadequate and ineffective” when it blocks even a first motion relying on Mathis. But see Boatwright v. Warden Fairton FCI, 742 F. App‘x 701, 704 (3d Cir. 2018) (rejecting this position).

In sum, this body of law is plagued by numerous complex issues. This might not be the case in which to untangle them, but at some point, we need to give litigants and district courts better guidance.

Notes

There is a circuit split on whether the lack of relief for statutory claims is a feature or a bug of § 2255. Compare (en banc) (holding that a change in caselaw does not make a § 2255 motion to vacate a prisoner‘s sentence “inadequate or ineffective” under the saving clause), and (rejecting a statutory interpretation challenge under § 2241 as barred by § 2255), with (authorizing saving-clause relief when an intervening statutory interpretation decision rendered the defendant‘s conduct non-criminal), (same), (same), (same), and (same).

The requirement that the new case be made retroactive by the Supreme Court—as opposed to the requirement that the case have been decided by the Supreme Court in the first place—is not a problem in Davenport cases. Statutory interpretations that narrow the range of conduct made criminal are always substantive and therefore retroactive. Because of that rule, a Supreme Court decision narrowing a criminal statute applies retroactively even if the Supreme Court doesn‘t expressly say so. ; .

To be clear, while Chazen‘s petition invokes as well as , Mathis is not the case that justifies granting him relief. See and . Van Cannon is the case that renders his burglary conviction insufficient to support the enhancement he received under the Armed Career Criminal Act. He invokes Mathis to justify his failure to raise this issue in his first § 2255 motion. Mathis itself does not offer him relief, because he does not complain that the sentencing court‘s decision to count his burglary conviction was the result of the error Mathis identifies—looking at the facts underlying his crime to conclude that he committed it by a violent means. Chazen needs both Mathis and Van Cannon to succeed.

At the time the government took this position, Seventh and Eighth Circuit law were identical on the relevant point. Compare , with . The ground shifted after the Supreme Court decided . At that point, McArthur was abrogated, the choice-of-law question became outcome-determinative, and the government backtracked.