576 F.3d 400
7th Cir.
2009

Lead Opinion

WOOD, Circuit Judge.

Vernon Woods was convicted of two counts of distributing ecstasy, in violation of 21 U.S.C. § 841(a)(1), and one count of possession of a weapon by a felon, in violation of 18 U.S.C. § 922(g). The district court found that Woods was a career offender and thus was subject to an enhanced sentence under § 4B1.1 of the United States Sentencing Guidelines (“U.S.S.G.”). The court imposed a sentence of 192 months, well above the 84-month sentence Woods might have received without the career offender enhancement. Woods now appeals his sentence, challenging whether, following the Supreme Court’s decision in Begay v. United States, - U.S. -, 128 S.Ct. 1581, 170 L.Ed.2d 490 (2008), and this court’s decision in United States v. Smith, 544 F.3d 781 (7th Cir.2008), his prior conviction for involuntary manslaughter— which required only a finding of recklessness — qualifies as a prior violent felony conviction for the purpose of the Guidelines. We conclude that Begay and Smith resolve this question in Woods’s favor, and we therefore vacate the judgment of the district court and remand for further proceedings.

I

After being caught in October and November 2006 distributing methylenedioxymethamphetamine (commonly known as ecstasy) to an undercover police officer, Woods pleaded guilty both to that offense and the offense of being a felon in possession of a firearm and ammunition that had previously been transported in interstate commerce. In the presentence investigation report (“PSR”), the Probation Service concluded that Woods was a career offender as defined by U.S.S.G. § 4B1.1. In so doing, the Probation Service relied on two prior convictions in Woods’s record: (1) a 1993 Illinois conviction for possession of cocaine with intent to deliver; and (2) a 2001 Illinois conviction for involuntary manslaughter. It is the second conviction that concerns us here.

The facts underlying Woods’s earlier conviction for involuntary manslaughter were contested at crucial points. Woods had been babysitting his infant son for several days. At a change of plea hearing (held after a jury had failed to convict him *402upon a first trial), Woods admitted that the child was five weeks old, and that he called 911 on the afternoon of March 18, 1999, when the child became unresponsive. The emergency authorities responded and took the baby to the hospital; six months later, he died. The state was prepared to call the deputy medical examiner, who would have testified that the child died of water on the brain (hydrocephalus due to subdural hematoma) as a result of blunt head trauma. She also would have testified that there were other signs of “wanton cruelty,” including a clinical history of cerebral palsy and a clinical history of severe mental retardation (although there is no explanation of how she came to the latter conclusion with respect to a five-week-old child). She described the manner of death as “homicide” — but as Woods notes, involuntary manslaughter is classified under Illinois law as a homicide offense. See 720 ILCS, Act 5, Part B, Article 9 (Homicide); 720 ILCS 5/9-3 (involuntary manslaughter).

One possible explanation of those- facts is that Woods took violent action against the child, shaking him and causing injury that resulted in his death six months later. But Woods, in his response to the PSR, gave an alternative explanation. According to Woods, he had dropped the baby and never intended to hurt him. When the baby lost consciousness, he shook the baby in an effort to revive him, and then he called 911 and requested an ambulance.1 Nothing in the plea colloquy before the state court resolved which version was true, nor were there any facts that might have shown whether the blunt head trauma could have resulted from being dropped as opposed to being shaken.

At the sentencing hearing, Woods objected to the Government’s characterization of his involuntary manslaughter conviction as a crime of violence under the Guidelines. (He conceded that the first conviction fell within the definition of § 4B1.1 because it was a controlled substance offense.) Woods argued that his involuntary manslaughter offense was not a crime of violence for two principal reasons: first, because his actions did not create a “serious potential risk of physical injury to another”; and second, because the mens rea for involuntary manslaughter in Illinois requires only criminal recklessness, and recklessness was insufficient to trigger the enhanced sentencing range recommended by the Guidelines. Further, Woods argued that even if the court were to look beyond the statute of conviction, the transcript of the plea hearing did not demonstrate that he acted in a way that presented a serious potential risk of physical injury to another.

The district court rejected all of these arguments, holding that the involuntary manslaughter statute described behavior presenting a risk analogous to the Illinois offense of reckless discharge of a firearm. This court held that the latter offense fell within the scope of § 4B1.1 of the Guidelines in United States v. Newbem, 479 F.3d 506, 508 (7th Cir.2007). The district court also held that although Newbem did not require it to go any further, that it would if necessary find the underlying facts of Woods’s conviction sufficient to support a finding that his conviction for involuntary manslaughter was a crime of violence as the Guidelines define that term. On November 16, 2007, the district court sentenced Woods to 192 months’ imprison*403ment, a sentence in the middle of the career offender range of 188 to 235 months. After Woods brought his appeal, the Supreme Court decided Begay, which cast new light on the Court’s interpretation of career offender enhancements like the one found in § 4B1.1. Whether Woods is entitled to succeed or fail in this appeal turns on the proper understanding of the Supreme Court’s decisions in Begay and the cases that have followed it.

II

The Sentencing Guidelines designate any defendant convicted of a “crime of violence or a controlled substance offense” who also has at least two prior felony convictions of either a crime of violence or a controlled substance offense as a “career offender.” § 4B1.1. Career offenders are subject to an enhanced base offense level and are automatically assigned to Criminal History Category VI. A great deal therefore hangs on the proper characterization of a defendant’s past encounters with the law. For Woods, it meant the difference between an advisory Guidelines range of 84 to 105 months (without career criminal status) and a range of 188 to 235 months (with career criminal status).2

In Woods’s case, as in many, we are concerned with the question whether the defendant’s prior offenses are properly characterized as crimes of violence. The Guidelines define a crime of violence as “any offense under federal or state law” that

(1) has as an element the use, attempted use, or threatened use of physical force against the person of another, or
(2) is burglary of a dwelling, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.

U.S.S.G. § 4B1.2(a). In deciding whether a crime fits that definition, the Supreme Court has instructed lower courts to use a categorical approach. In James v. United States, 550 U.S. 192, 127 S.Ct. 1586, 167 L.Ed.2d 532 (2007), a case dealing with the closely analogous Armed Career Criminal Act (“ACCA”), 18 U.S.C. § 924(e), the Court explained what it meant by a “categorical approach”:

Under this approach, we “ ‘look only to the fact of conviction and the statutory definition of the prior offense,’ ” and do not generally consider the “particular facts disclosed by the record of conviction.” Shepard v. United States, 544 U.S. 13, 17, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005) (quoting Taylor [v. United States], 495 U.S. [575, 602, 110 S.Ct. 2143, 109 L.Ed.2d 607 (1990)]). That is, we consider whether the elements of the offense are of the type that would justify its inclusion within the residual provision, without inquiring into the specific conduct of this particular offender.

550 U.S. at 202, 127 S.Ct. 1586. See also Begay, 128 S.Ct. at 1584. In United States v. Templeton, 543 F.3d 378 (7th Cir.2008), taking note of the identity of language between the ACCA and §§ 4B1.1 *404and 4B1.2, we held that the James analysis also applies to the Guidelines’s career offender provisions. In the discussion that follows, we therefore refer to the ACCA and the career offender provisions of the Guidelines interchangeably.

In applying the categorical approach, James recognized that the specific facts underlying certain offenses might reflect either a greater or a lesser degree of violence. The Court rejected the idea that a crime can never be one of violence, using the categorical approach, unless “every conceivable factual offense covered by [the] statute must necessarily present a serious potential risk of injury....” 550 U.S. at 208, 127 S.Ct. 1586. Rather, it wrote, “the proper inquiry is whether the conduct encompassed by the elements of the offense, in the ordinary case, presents a serious potential risk of injury to another.” Id. As we understand it, this means that a crime must be categorized as one of violence even if, through some freak chance, the conduct did not turn out to be violent in an unusual case. Importantly, the focus remains on the elements of the offense, not the particular facts surrounding each defendant’s conduct.

Although the categorical approach, as it has developed, suffices to answer most questions about the proper characterization of a prior offense, it is not enough by itself in one class of cases: when a statute covers more than one offense. In such cases, the Court has permitted courts to consult “the terms of the charging document, the terms of a plea agreement or transcript of a colloquy between judge and defendant in which the factual basis for the plea was confirmed by the defendant, or to some comparable judicial record of this information,” in order to determine what the defendant’s prior conviction was for (i.e., generic burglary or some lesser offense). Shepard, 544 U.S. at 26, 125 S.Ct. 1254. What the sentencing court cannot do is to look at the particular facts underlying the defendant’s conviction. Taylor, 495 U.S. at 600, 110 S.Ct. 2143.

Problems often arise when statutes describe more than one offense, but only some parts of the statute would qualify as a crime of violence. In Smith, we explained how that issue must be approached, in light of the governing Supreme Court cases:

Under the categorical approach, we consider the offense generieally; we may not inquire into the specific conduct of a particular offender. Begay, 128 S.Ct. at 1584; James, 127 S.Ct. at 1594. When a statute encompasses multiple categories of offense conduct — some of which would constitute a violent felony and some of which would not — we may expand our inquiry into a limited range of additional material [as set forth in Shepard, supra] in order to determine whether the jury actually convicted the defendant of (or, in the case of a guilty plea, the defendant expressly admitted to) violating a portion of the statute that constitutes a violent felony.... Such an examination, however, is “only to determine which part of the statute the defendant violated.” United States v. Howell, 531 F.3d 621, 623 (8th Cir.2008); see also [United States v.] Mathews, 453 F.3d [830, 834 (7th Cir.2006)]. This rule is not meant to circumvent the categorical approach by allowing courts to determine whether the actual conduct of the individual defendant constituted a purposeful, violent and aggressive act. See Shepard, 544 U.S. at 25, 125 S.Ct. 1254 (discussing the problems inherent in judicial fact-finding, particularly after Apprendi v. New Jersey, 530 U.S. 466, 120 S.Ct. 2348, 147 L.Ed.2d 435 (2000), and noting that looking further into the facts surrounding a prior conviction likely would violate the standard set forth in *405Apprendi); Mathews, 453 F.3d at 834 n. 8 (discussing Shepard, and noting that “this limitation preserves the categorical approach of Taylor and ensures that a defendant was ‘necessarily’ convicted of a generic burglary”).

544 F.3d at 786-87 (some citations omitted) (final emphasis added). In short, the additional materials permitted by Shepard may be used only to determine which crime within a statute the defendant committed, not how he committed that crime.

We emphasize the latter point because some confusion has arisen in our recent cases about the proper way to apply the modified categorical approach. In particular, our decision in Templeton, 543 F.3d at 383-84, can be read to rely on the actual conduct of the defendant, rather than which part of a divisible statute the defendant violated. In Templeton, the defendant was charged under Wisconsin’s escape offense, which states that

[a] person in custody who intentionally escapes from custody under any of the following circumstances is guilty of a Class H felony....

Wis. Stat. § 946.42(3). We stated — accurately enough, as far as it goes — that “it is possible to violate Wis. Stat. § 946.42 in a manner that constitutes a crime of violence under § 4B1.1, and possible to do so in a way that does not.” Templeton, 543 F.3d at 383-84. The problem arises in the next part of that passage, where we wrote, “Taylor holds that when a state statute can be violated in a way that is, or is not, the basis of federal recidivist treatment, a court may look at the indictment or other charging papers to determine in what way the defendant committed the offense.” Id. at 384.

If the words “in a way” in that sentence mean under one distinct portion of the statute or another, then Templeton is consistent with the line of Supreme Court decisions discussed above. If, however, the words “in a way” refer to the facts of the individual defendant’s case, then it is inconsistent with that line of cases and with our own Smith decision. In Temple-ton itself, the Wisconsin offense of escape covers a wide variety of conduct, some of which may pose a risk of violence and some of which may not, but the statute is not divisible in the sense called for by the modified categorical approach. Rather than specifying various subcategories of conduct, it simply states that “escape” is an offense and defines “escape” broadly to mean “leave [custody] in any manner without lawful permission or authority.” Wis. Stat. § 946.42(l)(b). We recognize that this definition can, as a factual matter, include conduct that would constitute a crime of violence as well as conduct that would not. Some may think that this is enough to justify a finding that the violent conduct is covered, if the charging papers or other permissible sources show that the particular offense was violent. One could argue that it is artificial to draw a line between, on the one hand, general statutes that prohibit both violent and nonviolent conduct, and, on the other, statutes that differentiate between violent and nonviolent offenses.

Whether this viewpoint would have merit on its own is, however, no longer open to us. The Supreme Court has spoken to the issue in a line of cases including Shepard, James, Begay, Chambers v. United States, - U.S. -, 129 S.Ct. 687, 172 L.Ed.2d 484 (2009), and, most recently, Nijhawan v. Holder, - U.S. -, 129 S.Ct. 2294, 174 L.Ed.2d 22 (2009). In all of them, it has opted for a rule that precludes deciding on a case-by-case basis whether a particular prior violation of a general statute posed the kind of risk of violence that would justify the recidivism enhancements provided by the ACCA or the career of*406fender Guidelines. We see no other way to read the operative language of James. As we pointed out in Smith, “This rule [permitting the expanded inquiry] is not meant to circumvent the categorical approach by allowing courts to determine whether the actual conduct of the individual defendant constituted a purposeful, violent and aggressive act.” 544 F.3d at 786. The Supreme Court used similar language in Chambers, where it emphasized the need to interpret the statutes underlying prior convictions in light of the normal way in which a crime is committed and commented, “by so construing the statute, one avoids the practical difficulty of trying to ascertain at sentencing, perhaps from a paper record mentioning only a guilty plea, whether the present defendant’s prior crime, as committed on a prior occasion, did or did not involve violent behavior.” 129 S.Ct. at 690.

Chambers made a point of noting that the failure-to-report offense at issue there was identified in a separate part of the statute. Thus, in Nijhawan, in the course of distinguishing between a statute like the ACCA, which uses a categorical approach, and a statute like the provision of the Immigration and Nationality Act directly at issue in Nijhawan’s case (8 U.S.C. § 1101(a)(43)(M)(i)), which uses a circumstance-specific approach, the Court discussed the categorical approach at length:

[T]he categorical method is not always easy to apply. That is because sometimes a separately numbered subsection of a criminal statute will refer to several different crimes, each described separately. And it can happen that some of these crimes involve violence while others do not. A single Massachusetts statute section entitled “Breaking and Entering at Night,” for example, criminalizes breaking into a “building, ship, vessel or vehicle.” Mass. Gen. Laws, ch. 266, § 16 (West 2006). In such an instance, we have said, a court must determine whether an offender’s prior conviction was for the violent, rather than the nonviolent, break-ins that this single five-word phrase describes (e.g., breaking into a building rather than into a vessel), by examining “the indictment or information and jury instructions,” Taylor, [495 U.S.] at 602 [110 S.Ct. 2143], or, if a guilty plea is at issue, by examining the plea agreement, plea colloquy or “some comparable judicial record” of the factual basis for the plea. Shepard v. United States, 544 U.S. 13, 26, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005).

129 S.Ct. at 2299. Later in the Nijhawan opinion, the Court (speaking of the ACCA) wrote, “Taylor; James, and Shepard, the cases that developed the evidentiary list to which petitioner points, developed that list for a very different purpose, namely that of determining which statutory phrase (<contained within a statutory provision that covers several different generic crimes) covered a prior conviction.” Id. at 2303 (emphasis added). (Nijhawan thus also demonstrates that it does not matter whether the earlier statute placed the statutory phrase in its own subsection, or if it merely made it part of a list. The point is that the statute itself is “divisible” — that is, it expressly identifies several ways in which a violation may occur.)

Nijhawan supports our understanding that the permissible additional materials may be consulted only for the purpose of determining under which part of a divisible statute the defendant was charged. In the Massachusetts example given by the Court, that material could be used to determine whether the crime fit under the “building” or “vessel” part of the statute, but it could not be used to see whether a particular act of breaking into a vessel gave rise to a substantial risk of injury to a person. To the extent that *407Templeton may be read as permitting reference to those materials to ascertain whether the particular crime was a violent one, we reject its analysis as inconsistent with the Supreme Court’s decisions. Because this opinion has the effect of changing the approach this court has taken to the application of the ACCA and U.S.S.G. § 4B1.1, it was circulated to the full court pursuant to Circuit Rule 40(e). Judges Flaum, Kanne, Rovner, Evans, Williams, and Sykes voted to approve this opinion and its understanding of Begay and the cases that have followed it. Chief Judge Easterbrook, Judge Posner, and Judge Tinder voted to hear the case en banc.

The dissent argues that Taylor cannot be reconciled with this approach because it deals with a non-divisible statute (one defining burglary as entry into a building with intent to commit a felony), yet it permits a sentencing judge to consider the charging papers or guilty-plea colloquy. We do not see Taylor this way. As the dissent concedes, post at 414-15, the statute before the Court in Taylor was a divisible one, as we are using that term. See 495 U.S. at 578 n. 1, 110 S.Ct. 2143. As the Court noted, “Missouri had seven different statutes under which one could be charged with second-degree burglary. All seven offenses required entry into a structure, but they varied as to the type of structure and the means of entry involved.” Id. One of those statutes, as described by the Court, even included a list. Id. Missouri adopted a more generic statute only after the defendant’s second-degree burglary conviction. Id. At the end of its opinion, the Taylor Court observed that in Taylor’s case “most but not all the former Missouri statutes defining second-degree burglary include all the elements of generic burglary.... Despite the Government’s argument to the contrary, it is not apparent to us from the sparse record before us which of those statutes were the bases for Taylor’s prior convictions.” 495 U.S. at 602, 110 S.Ct. 2143. Second, the dissent suggests that the Court was indifferent to the change in Missouri’s laws. Dissent, post, at 415. But this sheds little light on the problem before us, for the simple reason that Missouri’s amended statute mirrors the generic definition of burglary that the Court endorsed in Taylor: “an unlawful or unprivileged entry into, or remaining in, a building or other structure, with intent to commit a crime.” 495 U.S. at 598, 110 S.Ct. 2143.

Once the prior crime has properly been identified, the court must ascertain whether it is expressly identified by the ACCA or Guidelines, or if it is covered (if at all) only by the residual clause, describing an offense that “otherwise involves conduct that presents a serious potential risk of physical injury to another.” We can put to one side for purposes of this opinion those offenses that are more specifically identified, either because they have as an element the use (or attempted or threatened use) of physical force against the person of another, or because they are included in the statutory list (burglary, arson, extortion, or something involving the use of explosives). It is the residual clause that has posed most of the problems, and it is the residual clause that we are concerned with in this case. In Begay, the Supreme Court held that the residual clause was limited to offenses that were similar to the listed crimes, both in kind as well as in degree of risk posed. 128 S.Ct. at 1585. Only offenses that reflect the same “purposeful, violent, and aggressive manner” as the listed offenses satisfy the definition. Id. at 1586.

The Supreme Court recently addressed the issue of violence, for these purposes. As it had already noted in James, the offense must in the ordinary run of cases describe behavior that poses a sufficiently *408great risk of physical injury to another before it will satisfy the ACCA or § 4B1.1. Chambers v. United States, - U.S. -, 129 S.Ct. 687, 691-93, 172 L.Ed.2d 484 (2009). Relying in part on empirical data collected by the U.S. Sentencing Commission, the Court concluded that the crime of failure to report for penal confinement did not pose the required degree of risk.

The aspect of Begay that has come to the fore in Woods’s appeal is the requirement that the crime involve “purposeful” conduct. In Smith, we held that “those crimes with a mens rea of negligence or recklessness do not trigger the enhanced penalties mandated by the ACCA [or § 4B1.1].” 544 F.3d at 786. We therefore must decide whether the crime of involuntary manslaughter, as defined by Illinois law, is a crime of violence. This is a question of law that this court reviews de novo. United States v. Franco-Fernandez, 511 F.3d 768, 769 (7th Cir.2008).

Before turning to the specifics of Woods’s case, it is helpful to review the general law of mens rea. The first point is one of the most important: the state of mind in question must exist, as the Model Penal Code (“MPC”) puts it, “with respect to each material element of the offense.” MPC § 2.02(1). It is possible, however, that the mental state required might differ with regard to each element of the crime. See generally 1 Wayne R. LaFave, Substantive Criminal Law § 5.1(d), at 338 (2d ed.2003). The important point is to match the mental state in question to the conduct that is being made criminal by the statute, rather than to incidental steps along the way. As another treatise puts it, “[C]on-duct is a neutral or indifferent term in the sense that it may or may not constitute a crime. It constitutes a crime only if the ‘act or omission’ is voluntary and penally prohibited, and only if the ‘accompanying mental state’ is a recognized culpable mental state.” 1 Wharton’s Criminal Law § 25, at 146 (Charles E. Torcia ed., 15th ed.1993).

A number of “recognized culpable mental states” exist. The MPC refers to these as the “kinds of culpability.” Section 2.02 identifies four levels or categories: purposely, knowingly, recklessly, and negligently. (Exceptions to this rule are set forth in MPC § 2.05, but they are not pertinent here.) Once again, the state of mind (or kind of culpability) must be linked to each material element of the crime. So, for example, here is the MPC language describing a reckless state of mind:

A person acts recklessly with respect to a material element of an offense when he consciously disregards a substantial and unjustifiable risk that the material element exists or will result from his conduct. The risk must be of such a nature and degree that, considering the nature and purpose of the actor’s conduct and the circumstances known to him, its disregard involves a gross deviation from the standard of conduct that a law-abiding person would observe in the actor’s situation.

MPC § 2.02(2)(c). It is noteworthy that the person must consciously disregard the risk in question. It is also important to bear in mind that he must be disregarding the risk that the material element exists or will result from his conduct.

The Supreme Court’s decision in Begay shows how these distinctions operate in practice. The petitioner, Larry Begay, pleaded guilty to being a felon in possession of a firearm, in violation of 18 U.S.C. § 922(g)(1). Begay, 128 S.Ct. at 1583-84. He had a dozen prior convictions for Driving Under the Influence (“DUI”). The sentencing judge concluded that these were crimes that fell within the residual clause of the ACCA, because they involved *409conduct that presented a serious potential risk of physical injury to another. Based on that finding, the judge sentenced Begay to the mandatory minimum of 15 years in prison. The Court did not question the fact that DUI involves conduct with the necessary serious risk of physical injury to another. Importantly, however, it placed no weight on the fact that the driver consciously intended to drink, id. at 1587, or that he intended to get behind the wheel of a car. What did matter was that the state law at issue did not criminalize any intentional behavior. Instead, it “criminalizfed] conduct in respect to which the offender need not have had any criminal intent at all.” Id. at 1586-87. Underscoring the point, the Court noted that “the conduct for which the drunk driver is convicted (driving under the influence) need not be purposeful or deliberate.” Id. at 1587. This is so even though there is ample evidence that drunk drivers often inflict physical injuries — even death — on others in them pathway. In Begay, therefore, the Court rejected a reading of the ACCA that would have allowed the drunk driver’s intentional acts of drinking and driving, followed by recklessness with regard to the behavior that the statute made criminal (behavior that represented the consequences of the intentional act of drinking), to satisfy the statute.

Nothing in Begay, and nothing we say here, is meant to suggest that the presence of any recklessness component in a crime means that the conviction cannot be one of violence. An example that roughly reverses the facts of Begay illustrates the point. In this example, the drinking is reckless, and the dangerous driving is intentional; in Begay the opposite was true. Suppose that Jane goes to a party at which there are two large bowls of punch: one is nonalcoholic, and one is spiked with a clear, odorless, tasteless alcoholic drink. Knowing that one has a high alcoholic content, Jane nevertheless recklessly drinks from both bowls, paying no attention to which one she is using. At the party, she spots her ex-husband Tom leaving; she decides to follow him in her car. She does so, intending to bash her car into Tom’s so that he will have an expensive repair bill; unfortunately, however, Jane causes Tom’s car to swerve off the road and he dies in the resulting crash. The fact that Jane recklessly drank the alcohol would in no way insulate her from prosecution for the intentional assault on Tom in which she used her car as a weapon. That is true even though there is a reckless component to these events, since Jane might have had the sense not to assault Tom had she been sober.

What does remain important is the precise crime for which the defendant was convicted in the earlier case. It often will happen that a course of conduct could be charged under either a greater crime, such as murder or voluntary manslaughter, or a lesser crime, such as involuntary manslaughter. We are well aware that prosecutors sometimes begin with the greater charge and settle for the lesser charge after plea bargaining. That said, the only thing that counts for purposes of the ACCA or the career offender Guidelines is the prior crime for which the defendant was actually convicted. There is nothing that this court either could or should do about the prosecutorial discretion that is exercised at the charging stage. See In re United States, 503 F.3d 638, 642 (7th Cir. 2007). Thus, a prosecutor might charge one person who shoots a gun into a crowd with reckless endangerment, but then charge another person who does the same thing under a statute that prohibits aggravated assault with intent to cause bodily injury to another. If the defendant is convicted and then later is foolish enough to commit a crime that brings him before a *410federal court, the former crime would not count for purposes of the ACCA or the career offender guidelines, but the latter one would.

Ill

We turn now to Woods and his prior conviction for involuntary manslaughter. Illinois defines involuntary manslaughter as follows:

A person who unintentionally kills an individual without lawful justification commits involuntary manslaughter if his acts lohether lawful or unlaioful which cause the death are such as are likely to cause death or great bodily harm to some individual, and he performs them recklessly.

720 ILCS 5/9-3(a) (emphasis added). The Illinois Criminal Code defines the term “recklessness” more precisely:

A person is reckless or acts recklessly, when he consciously disregards a substantial and unjustifiable risk that circumstances exist or that a result will follow, described by the statute defining the offense; and such disregard constitutes a gross deviation from the standard of care which a reasonable person would exercise in the situation.

720 ILCS 5/4-6 (emphasis added). No one argues that Illinois involuntary manslaughter falls within either U.S.S.G. § 4B1.2(a)(l), which requires that the offense have as an element the use, attempted use, or threatened use of physical force against the person of another, or the first part of U.S.S.G. § 4B1.2(a)(2), which covers burglary of a dwelling, arson, extortion, and the use of explosives. The question is whether Illinois’s involuntary manslaughter offense should be categorized as a crime of violence under the residual clause in U.S.S.G. § 4B1.2(a)(2) as one that otherwise involves conduct that presents a serious potential risk of physical injury to another.

Applying the categorical approach required by the Supreme Court, we held in Smith that crimes with the mens rea of negligence or recklessness do not trigger the enhanced penalties mandated by the ACCA. Woods argues that this holding disposes of his case as well. He relies on the fact that the statute under which he was convicted covers unintentional killings, resulting from either lawful or unlawful acts that are performed recklessly.

Recognizing that Smith poses a problem for its position, the Government suggests that Smith categorically excludes only some crimes of recklessness from the ambit of the residual clause. Specifically, the Government claims that the Illinois involuntary manslaughter statute is excluded from Smith’s scope because, under the statute, a defendant must consciously disregard a substantial and unjustifiable risk, and this conscious disregard is itself the kind of voluntary and purposeful act that Begay had in mind. That is, the Government claims that if a defendant, such as Woods, intends the act but was reckless as to the consequences of that act, then the crime is not excluded from the scope of the residual clause under Begay.

In our view, this is precisely the distinction that the Begay Court rejected. In Begay itself, the defendant intended both the act of drinking alcoholic beverages and the act of driving his car; he was reckless only with respect to the consequences of those acts. As we have explained at more length above, this position was entirely consistent with the classic line that has been drawn between the actus reus and the mens rea of a criminal offense. The Government’s argument not only blurs that line; it obliterates it. The proposed ground on which the Government attempts to distinguish Smith would require this *411court to find that as long as a defendant’s act is volitional, he or she has acted purposefully under Begay’s interpretation of the career offender guidelines, even if the mens rea for the offense is recklessness. Unlike the dissent, post at 416-17, we understand this to cover criminal recklessness, which is the type of recklessness involved in most crimes. Every crime of recklessness necessarily requires a purposeful, volitional act that sets in motion the later outcome. Indeed, when pressed at oral argument to provide an example of a situation where a defendant would be reckless as to the outcome and not begin with an intentional act, the Government could not provide one.

It is worth underscoring, as we did in Smith, that the enhanced sentencing range under the ACCA or the career offender guidelines is imposed in addition to any punishment that already has been imposed on a defendant. (Nor, in this post-Booker world, does our interpretation prevent a sentencing judge from taking the circumstances of the prior crime into account in the process of selecting a reasonable sentence for the current crime.) Within very broad constitutional bounds, the legislature is entitled to establish a penalty as harsh as it believes is warranted for the prior crime. In separating out purposeful, violent, and aggressive crimes as the bases for enhancement of a later, unrelated criminal sentence, Congress was attempting to focus on “those offenders whose criminal history evidenced a high risk for recidivism and future violence ... [who] exhibited a special need for an increased sentence in order to deter future violent crimes.” Id. at 785. The overbreadth of the Government’s proposed principle can be seen in a simple example. Suppose a physician prescribes penicillin to a patient but consciously disregards the risk that the patient had an allergy to penicillin. Suppose then that the patient does have an allergy and dies as a result of the medication, and the physician is convicted of involuntary manslaughter under the Illinois statute in question here (because, under the Illinois statute, even lawful acts, such as a physician’s prescribing medication to a patient, can be the foundation for an involuntary manslaughter conviction). It seems clear that this physician is not the type of violent and aggressive criminal that the sentencing enhancements are intended to encompass, yet, under the principle espoused by the Government, this conviction would be the basis for a sentencing enhancement.

The Government also urges us to apply the “modified categorical approach,” but we do not agree with it that the Illinois involuntary manslaughter statute is one to which the modified categorical approach applies. As we explained earlier, James, Taylor, and Shepard permit a court to go beyond the statutory definition of the crime to consult judicial records (charging documents, plea colloquy, etc.) only where the statute defining the crime is divisible, which is to say where the statute creates several crimes or a single crime with several modes of commission. By “modes of commission” we mean modes of conduct identified somehow in the statute. The Illinois involuntary manslaughter statute is not divisible in this way, and we have no occasion to consult the record further in order to resolve Woods’s appeal.

The approach we take today invites comparison with the one adopted by the en banc court in United States v. Shannon, 110 F.3d 382 (7th Cir.1997). In that case, we had to decide whether the defendant, Shannon, had a prior felony conviction of a crime of violence for purposes of the offense of being a felon in possession of a firearm, 18 U.S.C. § 922(g)(1). Whether Shannon had such a conviction or not depended on the same Guidelines provision we are considering here, U.S.S.G. *412§ 4B1.2(a), and in particular the meaning of a “serious potential risk of physical injury to another.” Shannon’s prior conviction was for second-degree sexual assault of a child, which is committed by anyone who “has sexual contact or sexual intercourse with a person who has not attained the age of 16.” Wis. Stat. § 948.02(2). We acknowledged that we could not “peek behind the information” and look at the particulars of Shannon’s conduct. Shannon, 110 F.3d at 384. We did, however, consult the information, which indicated that the child in question was 13 years old — a fact that Shannon had admitted. That fact made it unnecessary to decide whether any felonious sexual act with a child was, by definition, a crime of violence, though at the time we expressed doubt about the wisdom of such a conclusion. Since Shannon was decided, the Supreme Court has handed down Shepard (2005), James (2007), Begay (2008), Chambers (2009), and Nijhawan (2009). Statutes addressing sexual contact with children vary widely. We acknowledge that we have already found that a violation of Wis. Stat. § 948.02(2) is not necessarily a crime of violence. See Xiong v. I.N.S., 173 F.3d 601, 607 (7th Cir.1999) (finding that consensual sex between an 18-year-old and his 15-year-old girlfriend was not a crime of violence for purposes of 18 U.S.C. § 16(b)). Because we do not have a concrete statute of that type before us in this case, we leave for a more appropriate occasion any further consideration of the effect of today’s decision on the general approach taken in Shannon and the question whether Xiong is consistent with Shannon.

We note in this connection that the Court has just granted certiorari in another case in this line, United States v. Johnson, 528 F.3d 1318 (11th Cir.2008), cert. granted, - U.S. -, 129 S.Ct. 1315, 173 L.Ed.2d 583 (2009). In Johnson, the Court accepted two questions for review: first, whether a holding from the highest court of a state that a given offense does not have as an element the use or threatened use of physical force is binding on federal courts applying the ACCA; and second, whether a prior state conviction for simple battery is in all cases a “violent felony” no matter how little force is used. Johnson may throw further light on the Court’s application of its modified categorical approach. We note, however, that it is not scheduled for argument until October 6, 2009. If the Court holds in Johnson that simple battery is in all cases a violent felony, then our en banc judgment in Shannon will be ratified, but on a broader ground, since one cannot commit the offense of sexual assault of a child without physical sexual contact or intercourse. If the Court adopts a different approach in Johnson, then it will be our duty to follow it. We have considered the possibility of holding this case and the many others presently before this court involving similar issues, but we have concluded that the costs of doing so outweigh the benefits. It may be of some assistance to the Supreme Court to know how we have interpreted the decisions it has issued thus far, and both the parties and the district courts deserve a disposition from us sooner rather than later.

That observation takes us back to where we began. Our best effort to read the applicable Supreme Court decisions leads us to the conclusion that the Court has rejected the technique of categorizing prior crimes based on the particular way in which they were committed. That observation guides our categorization of the conduct involved in a prior offense as something fitting the residual clause of the ACCA or the career offender guidelines, or not. As for the mental state requirement, we adhere to our holding in Smith that the residual clause encompasses only *413purposeful crimes; crimes with the mens rea of recklessness do not fall within its scope. In Woods’s case, these conclusions mean that the district court should not have included his conviction for involuntary manslaughter when it applied the career offender guidelines. Woods thus does not have the requisite number of predicate convictions to authorize sentencing him as a career offender.

The judgment of the district court is Vacated and the case is Remanded for further proceedings consistent with this opinion.

Notes

. The dissent assumes that Woods admitted that this act of shaking was 'purposeful, violent, and aggressive.” Dissent, post, at 416. But the record is not clear on that critical point. Whether the shaking was gentle or violent is a question of fact; in order to resolve it, we would have to conduct an independent investigation of the event.

. It should go without saying that nothing in this opinion means that a sentence above the advisory Guidelines range cannot be imposed. See, e.g., Kimbrough v. United States, 552 U.S. 85, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007). On that point, we agree with the dissent that the sentencing judge is entitled to impose a higher sentence based on "the history and characteristics of the defendant.” 18 U.S.C. § 3553(a)(1). This case therefore is only about the advisory guideline range with which the judge begins; it is not about the judge’s ability to impose a reasonable sentence, subject only to statutory minima and maxima. The sentencing judge's ability to exercise this discretion allows for whatever fine-tuning is necessary after the categorical approach has been applied.



Dissenting Opinion

EASTERBROOK, Chief Judge,

with whom

POSNER and TINDER, Circuit Judges, join, dissenting.

Begay v. United States, - U.S. -, 128 S.Ct. 1581, 170 L.Ed.2d 490 (2008), called into question many of this court’s decisions interpreting U.S.S.G. § 4B1.2(a)(2) and similar recidivism provisions, such as 18 U.S.C. § 16(b) and § 924(e)(2)(B)(ii). Last January the court set for argument before two panels several appeals that presented issues affected by Begay. As it happens, the six judges on those panels do not agree on how Begay applies, so proposed opinions in two cases were circulated to the full court under Circuit Rule 40(e). We decided to resolve the disputes through this circulation, without argument en banc. The approach proposed by the panel in Woods has the support of a majority and becomes the law of the circuit. I disagree with some aspects of the panel’s analysis and would proceed differently.

The career-offender portion of the Sentencing Guidelines, like 18 U.S.C. § 16 and § 924(e)(2)(B), counts toward the total of the defendant’s “crimes of violence” or “violent felonies” any conviction of an offense that has as an element the use or attempted use of force against the person of another. These provisions also include a residual category. The Guidelines define as a “crime of violence” any offense that:

is burglary of a dwelling, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.

U.S.S.G. § 4B1.2(a)(2). Begay understands the “otherwise involves ...” language of § 924(e)(2)(B)(ii) to cover only crimes “similar” to burglary, arson, extortion, and explosives offenses in the sense that they involve “purposeful, violent, and aggressive conduct”. 128 S.Ct. at 1586. This led the Court to hold in Begay that drunk driving does not qualify, because the driver does not set out to harm anyone, and in Chambers v. United States, - U.S. -, 129 S.Ct. 687, 172 L.Ed.2d 484 (2009), that failure to report to prison and walkaway escapes are not violent felonies. See also United States v. Templeton, 543 F.3d 378 (7th Cir.2008) (anticipating the holding of Chambers).

Begay creates problems of classification. It may be easy to tell when a person’s conduct was violent and aggressive, but whether a crime of conviction entails such conduct can be tricky, because it is necessary to think through the many varieties of behavior within a law’s domain. States did not write their statutes with Begay in mind. Many laws penalize categories of activity, some violent and some not. Or they may penalize reckless conduct. Criminal recklessness is a form of intent, see Farmer v. Brennan, 511 U.S. 825, 114 S.Ct. 1970, 128 L.Ed.2d 811 (1994), but particular laws may employ that mental state in idiosyncratic ways. Still other laws may apply a negligence standard to one element of an offense, recklessness to a second, knowledge to a third, and purpose to a fourth. There are many thousands of state and federal criminal stat*414utes. The judiciary needs a list or a set of categories rather than an open-ended standard, but for now we must make do.

Begay requires us to ask whether a crime that poses a “serious potential risk of physical injury” to another person is also sufficiently intentional, violent, and aggressive that it is similar to burglary and arson. Woods concludes that homicide does not meet this definition.

How can homicide not be an intentional, violent, and aggressive act? How can it be that burglary is a crime of violence, even though people rarely are injured in burglaries, and homicide is not, even though a person’s death is an element of the offense? The panel’s answer is that involuntary manslaughter, though treated in Illinois as a form of homicide (effectively third-degree murder), see 720 ILCS 5/9-3, has a definition broad enough to include some killings in which the mental element is recklessness rather than knowledge or purpose. Illinois calls the offense “involuntary” manslaughter when the defendant, though intending to perform the acts that end in death, does not want the victim to die, but is recklessly indifferent to the risk of death. This causes a problem for classification because federal recidivism statutes (and similar parts of the Guidelines) ask what crime the defendant has been convicted of, not what he did in fact. That categorical approach sends us to the state statute’s text rather than the facts of the defendant’s conduct.

The panel in Woods understands the categorical approach to ask whether a crime is “divisible”: unless all (or almost all) varieties of conduct within a law’s domain meet the Begay standard, then any conviction under the statute must be deemed one for a nonviolent offense. As it is possible to commit involuntary manslaughter in Illinois without purposeful, violent, and aggressive conduct, the panel concludes that no conviction for involuntary manslaughter may be the basis of a federal recidivism enhancement. We know that Woods was violent toward the victim. He concedes dropping and then shaking the baby, who died as a result. But because the drop may have been thoughtless, and conviction did not require proof that Woods intended the baby’s death, the panel holds that his federal sentence is too high.

I think that the sentencing judge should be allowed to look at the charging papers and plea colloquy in the criminal prosecution whether or not the statute is “divisible” in the panel’s sense. To see why, it is essential to start with Taylor v. United States, 495 U.S. 575, 110 S.Ct. 2143, 109 L.Ed.2d 607 (1990), which established the categorical approach to recidivist enhancements.

Taylor holds, and Shepard v. United States, 544 U.S. 13, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005), reiterates, that the question to ask is: of what crime does the defendant stand convicted? Taylor holds that federal recidivist statutes use a charge-offense rather than a real-offense approach. The Justices wrote: “Congress intended the sentencing court to look only to the fact that the defendant had been convicted of crimes falling within certain categories, and not to the facts underlying the prior convictions.” 495 U.S. at 600, 110 S.Ct. 2143. But how can one tell what the conviction represents, when a single state crime covers acts both within and without the federal domain?

Taylor concluded that the federal statute covered what it called “generic burglary”: only entering a residence with the intent to commit a felony is the crime of “burglary” for a federal recidivist enhancement. Some states have a statute with these elements. Other states use lists, as in “any person who enters a tent, railroad *415car, chicken coop, or dwelling with intent to commit a felony within” commits burglary. The panel treats statutes with lists as divisible. Still a third kind of statute provides that “any person who enters a building with intent to commit a felony therein” commits burglary. There’s nothing “divisible” about that law: the word “building” covers barns, ships, and dwellings. Yet Taylor says that here, too, the sentencing judge may look at the charging papers or guilty-plea colloquy to see whether the person was convicted of entering a house rather than a barn. A “divisibility” principle that excludes this aspect of Taylor is incompatible with the Supreme Court’s understanding.

Here is how the Justices summed up them conclusion:

We think the only plausible interpretation of § 924(e) (2) (B) (ii) is that, like the rest of the enhancement statute, it generally requires the trial court to look only to the fact of conviction and the statutory definition of the prior offense. This categorical approach, however, may permit the sentencing court to go beyond the mere fact of conviction in a narrow range of cases where a jury was actually required to find all the elements of generic burglary. For example, in a State whose burglary statutes include entry of an automobile as well as a building, if the indictment or information and jury instructions show that the defendant was charged only with a burglary of a building, and that the jury necessarily had to find an entry of a building to convict, then the Government should be allowed to use the conviction for enhancement.

495 U.S. at 602, 110 S.Ct. 2143 (footnote omitted). The Justices observed that Missouri had seven different burglary statutes, some with subdivisions or lists and some without. 495 U.S. at 578-79 n. 1, 110 S.Ct. 2143. They did not suggest that the difference mattered; instead the Justices adopted the approach quoted above as the approach to all seven. So instead of asking whether a state law is “divisible,” we should ask whether the jury (or judge) necessarily found all the elements required to classify the crime as “violent” for federal purposes.

What Taylor excludes is calling something “burglary” because that is what the defendant did, even if he was convicted of something else (such as unlawful entry of a residence, after a plea bargain that excluded the “with intent to commit a felony therein” element). And Shepard blocks using anything other than the charging papers and plea colloquy to establish what the defendant was convicted of. Neither opinion makes “divisibility” indispensable to classification. And, as far as I can see, no other circuit treats “divisibility” as a sine qua non. In this circuit the word first appeared in United States v. Mathews, 453 F.3d 830, 833 (7th Cir.2006), which used it as shorthand for Taylor's categorical approach rather than as a stand-alone requirement. Two other circuits have used the word, though neither has treated divisibility as a legal requirement. See Dulal-Whiteway v. Department of Homeland Security, 501 F.3d 116 (2d Cir.2007); Huerta-Guevara v. Ashcroft, 321 F.3d 883 (9th Cir.2003).

Woods suggests that decisions after Taylor create a divisibility requirement, even though the Justices themselves have not used the word or its functional equivalent. Yet Nijhawan v. Holder, - U.S. -, 129 S.Ct. 2294, 174 L.Ed.2d 22 (2009), shows that Taylor’s original analysis remains the Court’s approach. Nijhawan posed the question whether immigration officials may consider the size of a fraud when implementing 8 U.S.C. § 1227(a)(2)(A)(iii), which allows removal *416based on certain felony convictions, when the dollar value of the fraud is not an element of the offense. The court of appeals gave an affirmative answer; a dissenting opinion invoked a “divisibility” requirement in support of an argument that value must be ignored. Nijhawan v. Attorney General, 523 F.3d 387, 402-05 (3d Cir.2008) (Stapleton, J., dissenting). The Supreme Court affirmed in Nijhawan without mentioning “divisibility.” The Justices explained the categorical approach this way:

We also noted [in James v. United States, 550 U.S. 192, 127 S.Ct. 1586, 167 L.Ed.2d 532 (2007)] that the categorical method is not always easy to apply. That is because sometimes a separately numbered subsection of a criminal statute will refer to several different crimes, each described separately. And it can happen that some of these crimes involve violence while others do not. A single Massachusetts statute section entitled “Breaking and Entering at Night,” for example, criminalizes breaking into a “building, ship, vessel or vehicle.” Mass. Gen. Laws, ch. 266, § 16 (West 2006). In such an instance, we have said, a court must determine whether an offender’s prior conviction was for the violent, rather than the nonviolent, break-ins that this single five-word phrase describes (e.g., breaking into a building rather than into a vessel), by examining “the indictment or information and jury instructions,” Taylor, supra, at 602, 110 S.Ct. 2143, or, if a guilty plea is at issue, by examining the plea agreement, plea colloquy or “some comparable judicial record” of the factual basis for the plea. Shepard v. United States, 544 U.S. 13, 26, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005).

129 S.Ct. at 2299. If a judge may look at the charging papers and plea colloquy to ascertain whether the defendant burgled a house or a vessel, even though a single subsection covers both, what role can “divisibility” play? Surely not that there must be a list (as there was in the Massachusetts law quoted in Nijhawan); recall that Taylor dealt with statutes that did not have lists and made it a crime to break into any building with intent to commit a felony inside. Nijhawan shows that Taylor has not been modified by later decisions.

Woods dropped a five-week-old baby on his head, then shook the comatose child to death. That is purposeful, violent, and aggressive conduct. The possibility that Woods did not intend to drop the child need not detain us; the state statute requires some knowing conduct, a standard satisfied by the shaking if not the dropping. (The state judge did not pin this down, because it was not relevant as a matter of state law.) The Woods panel concludes that recklessness does not meet Begay’s requirement of intentional conduct, but Farmer holds that criminal recklessness — the kind involved here — is a form of intent, and I think it likely that the Justices will deem it sufficient for recidivism enhancements too.

Recklessness in criminal law means creating a risk of serious harm, usually by knowingly doing dangerous things with eyes closed to consequences. See generally Model Penal Code § 220.2(2). That mental state has been equated with intent not only in eighth-amendment cases but also in securities law, where proof of fraud depends on showing intent to deceive. Ernst & Ernst v. Hochfelder, 425 U.S. 185, 96 S.Ct. 1375, 47 L.Ed.2d 668 (1976). We held in Sundstrand Corp. v. Sun Chemical Corp., 553 F.2d 1033, 1044-45 (7th Cir.1977), that recklessness equates to intent when danger is so obvious that a reasonable person must be aware of it. Every other court of appeals has concluded that *417recklessness (appropriately defined) is a form of intent, and, though the question remains open in the Supreme Court, the Justices have not suggested restiveness. See Tellabs, Inc. v. Makor Issues & Rights, Ltd., 551 U.S. 308, 319 n. 3, 127 S.Ct. 2499, 168 L.Ed.2d 179 (2007). Why, then, should we declare that for recidivist enhancements a reckless indifference to the danger caused by one’s deliberate acts cannot satisfy the Begay standard?

The Woods panel writes: “nothing we say here is meant to suggest that the presence of any recklessness component in a crime means that the conviction cannot be one of violence.” Op. 409 (emphasis in original). The opinion then gives an elaborate example involving spiked punch and a purpose to commit mayhem. “Purpose” is the most exacting standard of intent under the Model Penal Code’s typology, it suffices by any standard. I grant that recklessness is not universally equivalent to intent; statutory context matters. But in the main a violent or aggressive crime that produces injury or death should meet the Begay standard, even if the actor recklessly ignored the risks to others.

Take a person who draws a gun and fires six shots into a crowded night club, not caring whether anyone is injured or killed. The intentional discharge of a gun is a violent and aggressive act; that the shooter is indifferent to the consequences shows his danger and is a good reason for a recidivist enhancement following his next conviction; it is not a reason to ignore the conduct. Likewise a person who drops a baby on his head, and intentionally shakes the inert body violently, has committed an aggressive and dangerous act; the person’s indifference to consequences should not prevent counting the conviction. I disagree with the approach of Woods to the extent that it commits this circuit to a contrary course.

One final observation. Taylor, Shepard, James, Chambers, and Nijhawan all involve the interpretation of statutes. This appeal involves the interpretation and application of the Sentencing Guidelines. We have held that the career-offender enhancement must be treated as a statute to the extent it implements 28 U.S.C. § 994(h), which requires the sentences for recidivists who commit specified crimes to be “at or near the maximum term authorized” for the new federal crimes. And this means, as we have also held, that the language in § 4B1.2 should be understood the same way as the language in § 16(b) and § 924(e), to the extent that the Guideline and the statutes use the same words. But Guideline 4B1.2 goes beyond § 994(h) by including federal offenses for which Congress has not specified a sentence “at or near the maximum” for recidivists. For these other offenders, the Guideline is merely advisory, and district judges may disagree. See United States v. Knox, 573 F.3d 441, 447-50 (7th Cir. July 20, 2009).

Moreover, a conclusion that a particular prior conviction is not one for a “crime of violence” does not limit the judge’s discretion to give a higher sentence based on the defendant’s actual criminal history. See Spears v. United States, - U.S. -, 129 S.Ct. 840, 172 L.Ed.2d 596 (2009); Kimbrough v. United States, 552 U.S. 85, 128 S.Ct. 558, 169 L.Ed.2d 481 (2007). The career-offender guideline does not affect the maximum allowable sentence. Since “[n]o limitation shall be placed on the information concerning the background, character, and conduct of a person convicted of an offense which a court of the United States may receive and consider for the purpose of imposing an appropriate sentence” (18 U.S.C. § 3661), a district judge is entitled to ask what Woods actually did.

*418Neither the categorical approach of Taylor nor the divisibility approach of Woods prevents the judge from giving Woods the same sentence that would be appropriate if involuntary manslaughter in Illinois were a “crime of violence” under § 4B1.2(a)(2). Woods just requires a more roundabout approach at sentencing. Is that sensible? When a recidivism enhancement raises the statutory floor under a sentence, or the maximum allowable sentence, a court should be punctilious about ensuring that the enhancement applies. But when the prior conviction just affects an exercise of discretion, the approach should be more flexible: when selection of the sentence is not governed by rule, why employ elaborate rules about “divisibility” and “recklessness” that the district judge may elect to bypass in the end?