Three decades ago, Congress set the courts upon a new course for the sentencing of federal defendants, moving away from a long-in-place system that gave wide discretion to federal judges to impose sentences from nigh no prison time to effective life sentences.
But this discretion was not so wide in practice as in appearance-the judge's sentence gave way when the prisoner left the court for prison. The total time served by the prisoner was on his arrival determined in the main by a parole commission. The commission determined release dates, and in a rough and crude way-relative to the work of the Sentencing Commission-anticipated the system now in place by using a scoring system that looked in part to a defendant's criminal history. In response to charges from the Left of disparate and from the Right of anemic sentencing, and thus with the support of both ends of the political spectrum, Congress shifted the focus to a defendant's individual circumstances on the one hand and mandatory minimum sentences tailored to particular crimes on the other. With much work from its newly erected Sentencing Commission, nourished by reflection, essential empirical study, and judges tasked with applying its regulations, this reform effort appears to now be understood by bench and bar, enjoying a measure of well-earned credibility. Yet its relatively calibrated system *520of adjustments struggles with rifle-shot statutory efforts deploying an indeterminate calculus for identification of repetitive, sentence-enhancing conduct that add on to the sentence produced by the guidelines, such as the Armed Career Criminal Act. In setting a federal criminal sentence the district judge looks, in part, to both the number and type of a defendant's prior convictions, a task complicated by the statute's effort to draw on criminal conduct bearing differing labels and boundaries set by the various states. Today, we continue to refine our efforts.
In this case, we consider questions posed by the use of Texas's burglary statute, Texas Penal Code § 30.02, to enhance a federal sentence. First, we confront whether two provisions of the statute, Texas Penal Code §§ 30.02(a)(1) and (3), are indivisible for the purposes of categorical analysis. Second, we consider whether either of these two provisions is broader than the federal generic definition of burglary encoded in the Armed Career Criminal Act, 18 U.S.C. § 924(e)(1). We answer each of these questions in the affirmative, and VACATE the appellant's sentence and REMAND for resentencing consistent with this decision.
On November 5, 2012, Dallas police officers stopped Michael Herrold for failing to signal a right turn. An officer approaching his car saw a handgun on the floor and arrested him. Herrold pled guilty to possession of a firearm by a former felon.1 This latest conviction came on top of a series of past felonies, including three convictions for Texas offenses that his PSR listed as making him eligible for the sentence enhancement imposed by the Armed Career Criminal Act ("ACCA")2 : (1) unlawful possession of LSD with intent to distribute; (2) burglary of a building; and (3) burglary of a habitation. Herrold argued that none of these offenses qualified as ACCA-predicate offenses, such that a sentence enhancement was therefore improper. The trial judge disagreed; he adopted the recommendation of the PSR and sentenced Herrold to 211 months in prison, including the ACCA enhancement. The judge observed, however, that Herrold had made "forceful arguments" that the enhancement should not apply, and he requested guidance from our court on the question. Without the enhancement, Herrold faces a statutory maximum of ten years3 -the enhancement added at least 91 months to his sentence and subjected him to a statutory minimum of fifteen years.4
We considered Herrold's arguments on direct appeal and affirmed his sentence on the basis of circuit precedent.5 The Supreme Court vacated our judgment and remanded for renewed consideration in light of Mathis v. United States , --- U.S. ----, 136 S.Ct. 2243, 195 L.Ed.2d 604 (2016).6 On remand, Herrold argued that Mathis forecloses the possibility that his two Texas burglary convictions can serve as ACCA predicates.7 We affirmed his sentence once again, this time on the basis of an earlier post- Mathis decision, *521United States v. Uribe , 838 F.3d 667 (5th Cir. 2016).8 We now reconsider this argument en banc and, in doing so, revisit Uribe and its progeny as well.
I.
The ACCA enhances the sentences of defendants with at least three previous convictions for certain crimes. Not all convictions trigger the enhancement-the ACCA specifies that a previous conviction must be for a "violent felony" or a "serious drug offense" for it to count as an ACCA predicate.9 "Violent felony," the sole category under which Herrold's burglary convictions could plausibly fall, is defined in part by reference to other crimes, and the ACCA tells us that "burglary, arson, [and] extortion" fit the bill.10
That said, "burglary" is confined to a federal definition of "generic burglary" unbound by a state's decision to label criminal conduct by that term.11 The fact that two of Herrold's convictions arose under a provision of Texas's burglary statute, Texas Penal Code § 30.02(a)(1), is therefore not dispositive. Labels aside, we must determine whether Texas's burglary statute sweeps more broadly in its application than the generic form of burglary encoded in the ACCA. Only then may we decide whether Herrold's convictions qualify as "violent felonies" that trigger an accompanying federal sentence enhancement.
II.
Texas's burglary statute, Texas Penal Code § 30.02(a), reads:
A person commits an offense if, without the effective consent of the owner, the person:
(1) enters a habitation, or a building (or any portion of a building) not then open to the public, with intent to commit a felony, theft, or an assault; or
(2) remains concealed, with intent to commit a felony, theft, or an assault, in a building or habitation; or
(3) enters a building or habitation and commits or attempts to commit a felony, theft, or an assault.12
As is evident, Texas's burglary statute is alternatively phrased, comprised of a list of several disjunctive subsections. Statutes taking this form pose a preliminary question-and its answer switches us to the appropriate analytical track. We must determine whether the statute sets forth alternative means of committing a single substantive crime, or separate elements, effectively defining distinct offenses.13 We refer to the former sort of statutes as "indivisible," and we call the latter "divisible."14 If a statute describes alternative means of committing one offense (i.e., if a statute is indivisible), we compare the whole thing to its federal generic counterpart and determine whether any part falls outside the federal template. In other words, we perform *522the classic categorical approach.15 If the alternative terms of a statute outline elements of distinct offenses (i.e., if a statute is divisible), we isolate the alternative under which the defendant was convicted and apply the federal template to only that alternative. This second analytical track has come to be known as the modified categorical approach.16
After the first time we upheld Herrold's sentence, Mathis v. United States provided a more fine-grained trace between statutory means and elements.17 In doing so, it also offered a typology of the authorities that federal courts may look to in determining whether a statute is divisible or indivisible.
Our first task is to determine whether state law sources resolve the question.18 If state court decisions dictate that a jury need not unanimously agree on the applicable alternative of the statute, the statute is indivisible and its alternative terms specify different means of committing a single offense.19 And if state courts have decided a jury must unanimously agree on the alternative, the alternatives describe separate offenses comprised of distinct elements.20 We may also look to the text of the statute. If the statute lists different punishments for each of its alternatives, they must be elements of distinct offenses.21 And the statute may also simply tell us "which things must be charged (and so are elements) and which need not be (and so are means)."22
If one of these authorities resolves the question, our inquiry ends. If state law fails to answer the question, we may look at the record of the defendant's prior convictions "for the sole and limited purpose of determining whether [the listed items are] element[s] of the offense."23 The record is relevant because if all statutory alternatives are charged in a single count of an indictment or lumped together in a jury instruction, this is evidence "that each alternative is only a possible means of commission, not an element that the prosecutor must prove to a jury beyond a reasonable doubt."24 And if an indictment or jury instruction contains only one of the statute's alternatives, this is evidence that the statute lists elements and is therefore divisible.25
Should our dual forays into state law and the record leave the question of divisibility inconclusive, the tie goes to the defendant-because the ACCA demands certainty that a defendant indeed committed a generic offense,26 any indeterminacy on the question means the statute is indivisible.27
A.
Conducting this inquiry leads us to the conclusion that *523Texas Penal Code §§ 30.02(a)(1) and (a)(3) are indivisible. While the Texas burglary statute itself lacks any trait that the Supreme Court deemed relevant to the divisibility inquiry,28 Texas case law settles the question. Indeed, Texas courts have repeatedly held that a jury need not unanimously agree on whether Texas Penal Code § 30.02(a)(1) or (a)(3) applies in order to sustain a conviction for burglary.29
In Martinez v. State ,30 the Texas Court of Appeals squarely faced the question of whether jury instructions charging Texas Penal Code §§ 30.02(a)(1) and (a)(3) in the alternative foul Texas's constitutional requirement for jury unanimity. And the Texas Court of Appeals rejected the application of that requirement in crystalline terms: "We must decide whether the legislature intended, through this single substantive distinction between burglary as defined under subsections (a)(1) versus (a)(3), to create two distinct criminal offenses. Guided by the court of criminal appeals' prior analysis of section 30.02, we conclude it did not."31 Accordingly, said the Martinez court, jurors are free to choose between subsections 30.02(a)(1) and (a)(3) without imperiling a conviction.32 This decision is no outlier-it was neither the first nor last Texas state court decision to come to the clear conclusion that jury unanimity between subsections (a)(1) and (a)(3) of Texas's burglary statute is not needed.33 Under Mathis , when state law does not require jury unanimity between statutory alternatives, the alternatives cannot be divisible.
The Uribe court relied on different Texas state court decisions to reach the contrary conclusion, believing that Day v. State34 and Devaughn v. State35 compelled *524its finding that Texas Penal Code § 30.02(a) is divisible.36 With respect, and aware that their language can mislead, we must disagree. These cases, as we read them, are not "ruling[s] of th[e] kind" deemed relevant by the Mathis Court, and they cannot resolve the divisibility question.37
In Day , the Texas Court of Criminal Appeals described "the elements of the three types of burglary" outlined by Texas Penal Code § 30.02(a) in comparing them to the offense of criminal trespass.38 However, its choice of the word "elements" is not imbued with any apparent legal significance-its division of Texas Penal Code § 30.02(a) into different "elements" was in service of determining whether criminal trespass is properly considered a lesser included offense of burglary. The Day court's analysis thus simply speaks to the different kinds of facts necessary to prove each individual burglary variant. In fact, the Day court also used language that could be read to suggest that the burglary statute is indivisible.39
Similarly, in Devaughn , the Court of Criminal Appeals occasionally used the word "element" in describing the provisions of Texas Penal Code § 30.02(a). Under Texas Penal Code § 30.02(a)(1), it explained that "[p]roof of the intent to commit either theft or a felony ... is[ ] a necessary element in the State's case."40 And it noted that "intent to commit a felony or theft is not an element of the offense proscribed by § 30.02(a)(3)."41 As in Day , however, the court's choice to use the word "element" in this context is of uncertain legal significance; Devaughn ultimately concerns the right of criminal defendants to notice of charges guaranteed under the Texas constitution. The analysis of that right does not turn on a distinction between elements and means.42 Once more-and likely for this very reason-the Devaughn court also chose to use language describing the different provisions of Texas Penal Code § 30.02(a) as alternative means of committing a single offense.43
*525Of course it is true that Day and Devaughn reflect decisions from Texas's highest criminal court while Martinez and the others come from intermediate courts. But this fact is of no real consequence- Day and Devaughn are simply concerned with questions that are different in nature from the ones that Mathis tells us are relevant. What's more-and driving this point home-it is not as if the Martinez court and the other Texas courts addressing jury unanimity ignored the existence of Day and Devaughn . Quite the contrary. The jury unanimity decisions explicitly and repeatedly invoke those two cases.44 We are not confronted with a situation, then, in which we must manage conflicting state decisions or decide how to deal with a rogue lower court's holding. Instead, we face the utterly workaday situation in which a state's highest court has articulated some principles about the nature of a statute to answer one question, and a series of state lower court decisions has drawn on those principles to answer a different question. Put another way, the lower courts have fleshed out Day and Devaughn and told us what they mean in this precise context: jury unanimity, the issue that Mathis deems dispositive, is not required between Texas Penal Code §§ 30.02(a)(1) and (a)(3).
Besides Day and Devaughn , the jury unanimity cases draw on the reasoning of another kindred case: the Supreme Court's opinion in Schad v. Arizona .45 Schad recognized and upheld the Arizona Supreme Court's treatment of premeditated murder and felony murder as different means of committing a single offense, such that jury unanimity between those alternatives is not required.46 And the Mathis Court cited Schad as an appropriate example of a federal court looking to state law on jury unanimity for answers on the question of divisibility.47 That the Texas courts also cite Schad indicates that they saw themselves performing the same role as the Arizona Supreme Court and makes their relevance to our inquiry all the more unmistakable. Under Mathis , they must pass muster.
The government argues that the Texas jury unanimity cases are nevertheless wrongly decided, and that we should disregard them. Small wonder-the government conceded at oral argument that if Martinez and its ilk accurately describe Texas burglary law, then its position would be "dead in the water." But Mathis does not contemplate federal substantive review of state decisions on jury unanimity for correctness on the merits; it directly informs us that where there is controlling case law, our inquiry is at an end.48 Layering *526an additional level of substantive review on the tasks Mathis assigns to sentencing courts would only deepen their descent into what some have described as a "time-consuming legal tangle."49
These cases all present something of a cautionary tale. Courts may speak of "elements" and "means" in myriad ways; to take just the first word, the cases cited to us contain references to the "element[s] in the State's case,"50 the "main element[s] of burglary,"51 and the " 'same elements' test" of Blockburger v. United States ,52 among other variations on that theme. No doubt recognizing these words' context-shifting nature,53 the Mathis Court did not send us on a search for state cases that describe a disjunctively phrased statute using either the word "elements" or "means."54 It demanded certainty. It demanded that we find "ruling[s] of th[e] kind" it relied on-rulings that may "definitively answer[ ] the question" of divisibility.55 Those, it held, are decisions considering whether jury unanimity is required between statutory alternatives. There is Texas case law concerning the need for jury unanimity between Texas Penal Code §§ 30.02(a)(1) and (a)(3), and it points in just one direction-that Texas Penal Code §§ 30.02(a)(1) and (a)(3) are indivisible.
B.
State case law on jury unanimity notwithstanding, the government brings other arguments that the two statutory alternatives should be treated as divisible. These arguments are foreclosed by Mathis .
First, the government makes several statutory claims about the nature and structure of Texas Penal Code § 30.02(a). It asserts that indivisible statutes should generally be limited to ones that consist of illustrative examples of conduct satisfying a listed offense. For example, a hypothetical indivisible "deadly weapon" offense might proscribe the use of a "knife, gun, bat, or similar weapon " to commit a crime.56 This assertion reflects misplaced emphasis on a statement in Mathis . As we have explained, Mathis does suggest that several features of a statute might resolve the question of its divisibility-of relevance here, "if a statutory list is drafted to offer 'illustrative examples,' then it includes only a crime's means of commission."57 The government argues the converse, apparently claiming that statutes describing anything but illustrative examples are automatically divisible. This is not the holding of Mathis , nor is it logically compelled by what the Mathis Court did hold. The presence of an illustrative list of statutory examples may settle the question in one *527direction, but the absence of such a list is not dispositive in the other.
The government casts its gaze farther afield, pointing to other statutory features unmentioned by the Mathis Court but that it nonetheless urges suggest divisibility. It would have us read significance into the facts that, for instance, "[e]ach subsection [of Texas Penal Code § 30.02(a) ] is separated by the word 'or,' " and that "each subsection requires 'different and separate acts to commit' the offense enumerated in that subsection." The extent to which features like this bear on the divisibility question is unclear.58 The first point involves a legislative drafting decision of uncertain significance in this context, while the second verges on circularity: disjunctively phrased offenses, by their very nature, involve different kinds of conduct or mens rea requirements.59 Disjunction means difference. The government may mean that the relevant subsections of Texas Penal Code § 30.02(a) are so different that they ought not be read as different ways of committing a single, indivisible offense, but its argument comes bereft of reasoning and it fails to explain just how different is too different. In fact, a plurality of the Supreme Court has already expressed grave doubt about the ability of a court to examine the factual differences between statutory alternatives and label them elements or means through sheer force of reason.60
The arguments along these lines sum to the assertion that the Texas burglary statute does not fit the government's conception of what an indivisible statute looks like. But the Court has given us a test to apply, and that test is not a Rorschach. We are bound to examine how a state treats its own statute using the materials that the Court said speak with sufficient certainty on the matter. For this reason, we decline to hold that these structural statutory features are sufficient to resolve the question of divisibility when they point in the opposite direction of sources that the Mathis Court did say were relevant-state decisions on the subject of jury unanimity.61
*528Next, the government points to several state double jeopardy cases involving Texas's burglary statute. According to the government, because these decisions reach different outcomes on the question of double jeopardy depending on the statutory alternative charged, the statute must be divisible. The government's argument, however, shares the same flaw as its previous arguments: the Supreme Court did not list double jeopardy cases when it outlined sources of state law that could answer the question of a statute's divisibility with sufficient certainty.
And for good reason. As an initial matter, different states apply their own tests for enforcing their own double jeopardy rules, and therefore simply tracking double jeopardy cases would mean using a different test for divisibility based on the rules of the underlying state.62 None of the sources that the Mathis Court actually pointed to have this flickering quality.63 Further, the Fourth Circuit rejected basically the same double jeopardy argument in United States v. Cabrera-Umanzor , in the course of holding that a Maryland child abuse statute is indivisible.64 It explained that statutory distinctions made by state courts in a double jeopardy analysis do not automatically inform the divisibility analysis.65 The Mathis Court, in turn, cited Cabrera-Umanzor as an example of a federal court properly performing the divisibility inquiry.66
There is another, more conceptual reason why the double jeopardy cases provided by the government shed little light on divisibility. Texas state courts have adopted the Blockburger test for double jeopardy, which asks courts to determine the facts that must be proven under different statutory alternatives.67 When statutory *529alternatives require proof of different facts, they lead to different outcomes under the Blockburger test.68 This means that the Texas courts' inquiry bottoms out in an examination of the factual differences between statutory alternatives in a disjunctively worded statute. But again, all experience suggests that factual differences alone do not cast enough light to answer the divisibility with the needed certainty.69 Alternative means and alternative elements both necessarily entail factual differences; the decisive question for the purpose of divisibility analysis is not whether factual differences exist, but what legal effect accompanies those factual differences.70
In light of Texas case law, we hold that Texas Penal Code §§ 30.02(a)(1) and (a)(3) are not distinct offenses, but are rather separate means of committing one burglary offense. To the extent that it is inconsistent with this holding, we also overrule our earlier decision in United States v. Uribe .71
III.
Before considering whether Texas Penal Code §§ 30.02(a)(1) and (a)(3) correspond to the Court's generic definition of burglary, we step back to consider the purpose and function of generic burglary. In Taylor , when it first interpreted the scope of burglary encoded in the ACCA, the Supreme Court did not read the statute's definition as being pegged to the labels deployed by the various states.72 It expressly refused to do so, holding that the ACCA's version of burglary charts a fixed category of conduct independent of state labels, in order to preserve the virtues of uniformity and fairness in sentencing.73
This decision rested on the clear premise that different portions of state definitions would not fall within the generic definition's scope, a reality that the Taylor Court acknowledged. But the Taylor Court was not animated by the purpose of maximizing the number of states that fall within or without the ACCA's ambit.74 It was rather engaged in implementing Congress's intent from the sources it deemed appropriate, and with a burglary definition in service of predictability in sentencing. The idea was to ensure that similar conduct was similarly treated in the enhancement of federal sentences.
The Taylor Court's approach was cautious; even after choosing to deploy a generic definition, it could have outlined that definition more broadly. But to do so would increase the risk of sweeping in criminal conduct of disparate character. If the federal definition were slackened too much, a defendant who broke into a building to escape the cold and only once inside decided to pilfer a jacket could be subject to the same enhancement as a defendant who planned an elaborate theft of that same building.75 Or a defendant who broke into the unoccupied cab portion of a pickup truck could be subject to the same enhancement as a defendant who broke into *530an occupied family house.76 Our reading of the ACCA's scope is against the backdrop of the important congressional goal of treating like conduct alike. The Taylor Court clearly recognized this goal when it read the ACCA as containing a narrower scope than it might have, well aware of its significant sentencing force and its potential for unintended sentencing disparity.77
Nor does the Taylor Court's approach disserve states that opt to extend their burglary definitions broadly. States remain free to define and punish burglary however they like-they can prescribe sentences for their nongeneric burglary statutes that compensate for the ACCA's inapplicability. They can define different offense degrees or tinker with their statutes' divisibility structures to carve out suitably generic forms.78 Or states can ignore the existence of the ACCA, mindful that it is a federal statute that memorialized Congress's preferred definition of burglary at the time it was enacted. However states ultimately choose to respond, clarity in defining the reach of the ACCA's generic definition enables legislatures to accurately consider federal policy in deciding how to shape their own.79
In the hands of the fifty states with their myriad local concerns, the scope of burglary at the state level was a dynamic target when the ACCA was passed and it continues to be one today.80 It is for Congress, however, to alter the federal definition if and when it deems appropriate.81 These principles inform the question of whether a particular state provision qualifies as generic burglary.
IV.
Because Texas Penal Code §§ 30.02(a)(1) and (a)(3) are indivisible, we must use the categorical approach to examine the viability of Herrold's two burglary *531convictions under the ACCA. Under the vanilla version of the categorical approach, if either Texas Penal Code § 30.02(a)(1) or (a)(3) is broader than generic burglary, then neither of Herrold's two burglary convictions may serve as the basis of an ACCA sentence enhancement. We begin by evaluating the scope of Texas Penal Code § 30.02(a)(3).
A.
Subsection 30.02(a)(3) of Texas's burglary statute proscribes entry into a building or habitation followed by commission or attempted commission of a felony, theft, or assault.82 This formulation renders the provision broader than generic burglary, and it does so for lack of a sufficiently tailored intent requirement. The ACCA's definition of generic burglary requires "unlawful or unprivileged entry into, or remaining in, a building or structure, with intent to commit a crime ."83 Both the Supreme Court's language and its sources suggest that this constitutes a contemporaneity requirement: to be guilty of generic burglary, a defendant must have the intent to commit a crime when he enters or remains in the building or structure.84 Subsection 30.02(a)(3) contains no textual requirement that a defendant's intent to commit a crime contemporaneously accompany a defendant's unauthorized entry. And we have repeatedly held that because of this fact, it is broader than the ACCA's generic definition.85
The government disagrees. Relying mostly on out-of-circuit precedent, it argues that despite the fact that Texas Penal Code § 30.02(a)(3) only expressly speaks of unauthorized entry,86 the "remaining in" portion of the ACCA's generic burglary definition can save it. According to the reading the government would have us *532adopt, this is so because "someone who enters a building or structure, and while inside, commits or attempts to commit a felony will necessarily have remained inside the building or structure to do so."87 This reading is made available only by a broad understanding of the Supreme Court's reference to "remaining in" in Taylor . Rather than referring to "a discrete event that occurs at the moment when a perpetrator, who at one point was lawfully present, exceeds his license and overstays his welcome,"88 this reading of "remaining in" would define it as a continuous state that begins immediately after unauthorized entrance and lasts until departure.
The breadth of the government's reading is clear. The Taylor Court spoke of "unlawful or unprivileged entry into, or remaining in" with the requisite intent as if they were alternative possible acts.89 Yet the net effect of the government's linguistic move puts entry almost entirely out of focus; because all entry is followed by its version of remaining in, and because the remaining in lasts until departure, almost every instance of entry would automatically involve remaining in. For this same reason-and in combination with the accompanying removal of a contemporaneity requirement-statutes that seem to speak only of unlawful entry counterintuitively correspond instead to generic remaining in.
The more natural way of reading the Supreme Court's reference to "remaining in" in its generic burglary definition-and the way we have chosen to read it in the past90 -would retain the distinction between the two outlined categories of conduct. Under that reading, the "remaining in" language captures burglars who initially have a license to enter a particular location but who remain there once that license expires in order to commit a crime. Generic burglary would require these defendants to possess the intent to commit a crime while remaining in this narrower sense-that is, at the moment they exceed their license in order to commit a crime.91
In addition to ensuring that the two types of conduct function as true alternatives, this interpretation has the support of the sources that the Taylor Court relied on in crafting its generic burglary definition. After the Taylor Court articulated the elements of generic burglary, it directly cited only the then-current edition of the influential LaFave and Scott criminal law treatise. In that treatise, LaFave and Scott address the remaining in alternative, explaining that the language's purpose is to capture defendants who lawfully enter a location and then remain, once their license to be there is lost, in order to commit a crime.92 Indeed, the treatise's sole *533example of this type of burglary describes "a bank customer who hides in the bank until it closes and then takes the bank's money."93
LaFave and Scott directly allude to Texas Penal Code § 30.02(a)(3) in this discussion. They opine that Texas enacted § 30.02(a)(3) in order to avoid potential problems of proof "concerning whether the defendant's intent was formed before or after the unlawful entry or remaining."94 From this, we can gather that LaFave and Scott understand "remaining in" in the narrow sense. To speak of problems of proof associated with possible intent formation "after the unlawful ... remaining"95 would be incoherent otherwise-the only way intent can form after "remaining" in the broad sense would be if it formed after the defendant totally left the premises. LaFave and Scott also describe the very statute in this case- Texas Penal Code § 30.02(a)(3) -as an "alternative" to the ordinary "unlawful entry or remaining" forms of burglary, borne out of problems of proof associated with those conventional categories of conduct.96 Thus, the sole source that the Taylor Court directly cited for its generic burglary definition both describes "remaining in" narrowly and distinguishes it from Texas Penal Code § 30.02(a)(3).
The Taylor Court also mentions the Model Penal Code in its analysis, but the cited edition does not include any "remaining in" language at all.97 To the extent the Model Penal Code drafters do discuss the existence of "remaining in" language in other burglary statutes, they are in accord with LaFave and Scott about the genre of bad actors whom that language was meant to reach: those who are initially licensed to be on a property but who exceed their license in order to commit a crime.98
Finally, the Taylor Court noted that its "generic sense" of the offense would have been recognized as burglary by most states at the time Taylor was decided.99 But not all states used "remaining in" language in their burglary statutes-LaFave and Scott list twenty-five in their treatise.100 The states that did include the language at the relevant time appear to have been split in how they understood its scope.
To lift up just one example, New York's "remaining in" statute appears to have been particularly influential.101 We know *534that by the time Taylor was decided, New York's highest court had squarely considered and rejected the broad reading of "remaining in" now urged by the government.102 Indeed, the New York Court of Appeals recognized that this reading would go too far in sweeping different types of conduct into the ambit of burglary: "A defendant who simply trespasses with no intent to commit a crime inside a building does not possess the more culpable mental state that justifies punishment as a burglar."103 Just so; as we have observed in the past, "teenagers who unlawfully enter a house only to party, and only later decide to commit a crime, are not common burglars."104
Not only does the broad version of "remaining in" involve a less culpable mental state on the part of the defendant, it also likely presents less danger to victims. Indeed, the Taylor Court's analysis was partially based on the premise that "[t]he fact that an offender enters a building to commit a crime often creates the possibility of a violent confrontation."105 Scenarios in which a defendant trespasses but does not intend to commit a crime must engender less risk of confrontation than ones in which he enters just to commit a crime . The broad reading urged by the government leads to the conflation of this type of conduct with generic burglary, however, undercutting Congress's goal of treating like conduct alike for the purposes of the ACCA's sentence enhancement and expanding a harsh sentencing enhancement beyond its natural reach.106 Further, in light of the lack of consensus that existed at the time Taylor was decided,107 and that *535apparently persists today,108 the narrower reading is more consistent with the Supreme Court's apparent view that its burglary definition would have obtained in most states.109
The government points out that its reading of Taylor 's"remaining in" language finds support in decisions issued by the Fourth and Sixth Circuits. They are not persuasive. In United States v. Bonilla , the Fourth Circuit considered the Texas *536burglary statute at issue here, while in United States v. Priddy , the Sixth Circuit considered a similar Tennessee burglary provision. In Bonilla , a divided panel concluded that subsection 30.02(a)(3) is generic burglary because "a defendant convicted under section (a)(3) necessarily developed the intent to commit the crime while remaining in the building, if he did not have it at the moment he entered."110 Similarly, in Priddy , the Sixth Circuit saw the Tennessee burglary as "a 'remaining-in' variant of generic burglary because someone who enters a building or structure and, while inside, commits or attempts to commit a felony will necessarily have remained inside the building or structure to do so."111 With due respect, these statements do not answer, but rather beg, the question of the meaning of the phrase "remaining in."
On the other hand, the most recent treatment of the question by the Eighth Circuit considered an expansive interpretation of "remaining in" before deciding to take the opposite tack. In the relevant case, United States v.McArthur , the Eighth Circuit held that a materially identical Minnesota burglary statute is nongeneric because "remaining in," for the purposes of generic burglary, is "a discrete event that occurs at the moment when a perpetrator, who at one point was lawfully present, exceeds his license and overstays his welcome."112 The Eighth Circuit recognized that holding otherwise would "would render the 'unlawful entry' element of generic burglary superfluous, because every unlawful entry with intent would become 'remaining in' with intent as soon as the perpetrator enters."113
We decline to retreat from our previous holding that Texas Penal Code § 30.02(a)(3) -Texas's burglary offense allowing for entry and subsequent intent formation-is broader than generic burglary.
B.
Following our initial decision that Texas Penal Code § 30.02(a)(3) is not generic, we have, in an effort to cabin fanciful hypothetical readings, issued United States v. Castillo-Rivera .114 That decision requires criminal defendants to establish "a realistic probability" that courts will apply a state statute in a posited nongeneric way before a court may hold that it fails the categorical approach.115 We may look to state court decisions to satisfy this requirement. Texas courts have repeatedly held that under Texas Penal Code § 30.02(a)(3), a defendant can form the intent to commit a crime after an unauthorized entry.116 For this *537reason, and under Castillo-Rivera , there is nothing speculative about the reach of Texas Penal Code § 30.02(a)(3). Because Texas Penal Code § 30.02(a)(3) is plainly broader than generic burglary, and because Texas Penal Code §§ 30.02(a)(1) and (a)(3) are indivisible, neither of Herrold's two convictions under the Texas burglary statute may serve as the predicates of a sentence enhancement under the ACCA.
V.
Herrold argues that even if Texas Penal Code §§ 30.02(a)(1) and (a)(3)were divisible, he would still not satisfy the requirements for a sentence enhancement under the ACCA. This is so, according to him, because one of his ACCA-predicate convictions was for burglary of a habitation under Texas Penal Code § 30.02(a)(1). There are powerful arguments on both sides of the question; we think it important to describe them in full in order to explain why we ultimately choose not to decide the question of whether the definition of "habitation" applicable in Texas Penal Code § 30.02(a)(1) makes it broader than generic burglary.
A.
Texas Penal Code § 30.02(a)(1) dictates that a defendant commits burglary if he "enters a habitation, or a building (or any portion of a building) not then open to the public, with intent to commit a felony, theft, or an assault."117 "Habitation," in turn, is defined as "a structure or vehicle that is adapted for the overnight accommodation of persons ," including subportions thereof.118 It is unclear whether this burglary provision's application to "vehicle[s]" "adapted for the overnight accommodation of persons" renders it broader than the federal, generic definition of burglary.
As a starting point, we know that the generic definition of burglary outlined by the Taylor Court extends only to the burglary of "building[s] or other structure[s]," and we know that this category generally excludes vehicles.119 Indeed, we have the Supreme Court's own language on the subject. In the decisions it has issued after Taylor , the Supreme Court has had occasion to consider whether several other state burglary statutes fit within Taylor 's generic definition. In holding that these statutes are broader than generic burglary, the Court has suggested that vehicles ordinarily fall outside the scope of generic burglary.
Thus, in Shepard v. United States , the Court considered the ACCA viability of a Massachusetts burglary statute that extended to unlawful entry into "a building, ship, vessel or vehicle."120 The Court said that "[t]he [ACCA] makes burglary a violent felony only if committed in a building or enclosed space ... , not in a boat or motor vehicle ."121 More recently, in Mathis , the Court considered an Iowa statute extending the scope of burglary to "any building, structure, [or] land, water, or air vehicle ... adapted for overnight accommodation of persons, or occupied by persons for the purpose of carrying on business or other activity, or for the storage or safekeeping of anything of value."122 The Mathis Court held that this definition exceeded *538the scope of generic burglary, and, as in Shepard , it used language to suggest that vehicles are outside of that scope: "Iowa's statute, by contrast, reaches a broader range of places: 'any building, structure, [or] land, water, or air vehicle. ' "123 The Court paid no attention to the limiting characteristics imposed by the Iowa statute-the requirement that any vehicle be "adapted for overnight accommodation of persons, or occupied by persons for the purpose of carrying on business or other activity, or for the storage or safekeeping of anything of value." Instead, the Court flatly said that the Iowa statute is overbroad because it reaches "land, water, or air vehicle[s] ," full stop. The natural implication of the Court's repeated language across these cases is that vehicles should generally be treated as falling outside the scope of generic burglary.124
On the question of whether narrower subcategories of vehicles such as RVs and motor homes are generic, the picture gets decidedly blurrier. On one hand, we have the legislative history of the ACCA that the Taylor Court found relevant. While the ACCA itself offers no textual definition of burglary, the ACCA's predecessor statute did, and it extended only to buildings.125
*539The definition was dropped when the statute was updated into its current form, but the Taylor Court explained that "[t]he legislative history as a whole suggests that the deletion of the 1984 definition of burglary may have been an inadvertent casualty of a complex drafting process," and it concluded that "there is there simply is no plausible alternative that Congress could have had in mind."126 As a result, the Court described Taylor 's generic burglary definition as "practically identical" to the one deleted from the statute.127
We also have the sources that the Taylor Court relied on in crafting its generic definition. As explained before, the sole source directly cited by the Taylor Court for its generic burglary formulation is LaFave and Scott. On the same page of the treatise edition that the Supreme Court cited for its proposition that generic burglary must occur within "a building or other structure," the authors explain that some state burglary statutes go farther. They write that, in contrast to statutes limited to "buildings" and "structures," some statutes "extend to still other places, such as all or some types of vehicles."128 And among the statutes listed as extending to "still other places" is the very Texas burglary of a habitation provision at issue in this case.129 From this, we can conclude that LaFave and Scott did not consider a vehicle adapted for overnight accommodation to count as "a building or other structure"-the locational category that the Taylor Court adopted for its definition.
The weight of federal case law seems to support the conclusion that the federal generic definition of burglary may not extend to any vehicles, even the narrower subset circumscribed by the Texas burglary of a habitation provision. Almost every federal court that has found itself in the position to consider similar burglary statutes has concluded that the inclusion of any vehicles renders a state burglary provision nongeneric.130 Almost all of the cases that the government cites to the contrary have been overruled131 or pre-dated Shepard and Mathis .132
*540The government appropriately recognizes that vehicles are generally excluded but, on the other hand, it asks that we draw the generic definition's line for "building[s] or other structure[s]" to include vehicles that double as "dwellings" or "mobile habitations." It points to several sources that it argues support its choice to read the definition in this way. The government directs us, for instance, to the Model Penal Code's burglary definition relied upon by the Taylor Court. That definition extends to "occupied structures," which is defined to include "vehicle[s] ... adapted for overnight accommodation" and others.133
The government also argues that all conduct that would have been considered burglary for the purposes of the common law must also be burglary for the purposes of the ACCA. Because "mobile habitations" such as motor homes and RVs would have been valid common law burglary sites,134 the argument goes, they must also be valid generic burglary sites; the former is just a subset of the latter.135
Finally, the government presents a list of state statutes in effect at the time Taylor was decided. Fixing on the Taylor Court's statement that the ACCA's generic definition of burglary corresponds to "the generic sense in which the term [was then] used in the criminal codes of most States," it argues that our reading cannot be correct because it would render too many Taylor -contemporaneous burglary statutes nongeneric. Indeed, according to the government, "the protection of mobile dwellings was part of the vast majority of state codes when Congress enacted the ACCA."
There are several problems with at least this final line of argument.136 First, the character of the state statutes belies the very limitation the government argues *541it supports; the "vast majority" of state statutes that expressly considered vehicles seem to have either extended to all vehicles137 or extended to some subset of vehicles broader than dwellings and habitations.138 Thus, the government's argument proves too much.139 If its approach were correct, it would make no sense to draw the line at vehicles-cum -dwellings-the tallying would require some larger subcategory of vehicles to count as viable locations for generic burglary. And this would make the Supreme Court's own articulations of the definition of generic burglary and seemingly categorical disavowals of vehicles somewhat bizarre in context. We also do not read Taylor to mean that any feature of a burglary provision in effect in more than half of the states when Taylor was decided must ipso facto be part of the federal generic definition.140 The Taylor Court seemingly well understood that its generic definition could be underinclusive: "[a]lthough the exact formulations vary, the generic, contemporary meaning of burglary contains at least the following elements ...."141 Put another way, nowhere in Taylor did the Court characterize its definition of generic burglary as the maximum common denominator among then-contemporaneous state burglary statutes. It opted to be more conservative, relying on a set of discrete sources it deemed useful and distilling the set of characteristics it deemed appropriate. Taylor offers no invitation to reset the Court's work.
B.
As we need not decide the question of whether Texas Penal Code § 30.02(a)(1) is nongeneric, for the reason that the powerful arguments we have described lie on both sides of it, it is not immediately clear where the Texas burglary of a habitation provision falls. We welcome any additional guidance from the Court.142
VI.
To summarize, the burglary provisions encoded in Texas Penal Code §§ 30.02(a)(1) and (3) are indivisible. Texas Penal Code § 30.02(a)(3) is nongeneric because it criminalizes entry and subsequent intent formation rather than entry with intent to commit a crime. For these reasons, Herrold's ACCA sentence enhancement cannot stand. We VACATE and REMAND
*542to the district court to resentence him in accordance with our decision today.
HAYNES, Circuit Judge, joined by JOLLY, JONES, CLEMENT, OWEN, ELROD, and SOUTHWICK, Circuit Judges, dissenting:
The majority opinion upends years of well-settled law. Just over a year ago, this court confirmed that Texas Penal Code § 30.02(a) is a divisible statute, and the Supreme Court denied certiorari. United States v. Uribe , 838 F.3d 667 (5th Cir. 2016), cert. denied , --- U.S. ----, 137 S.Ct. 1359, 197 L.Ed.2d 542 (2017). The effect of the majority opinion, in addition to unsettling established precedent, is to render all burglary convictions in the second-most populous state in the country nullities as far as the ACCA is concerned. That is no small thing. In just a single year, Texans reported 152,444 burglaries, all of which now escape the ACCA's reach. See TEX. DEP'T PUB. SAFETY, CRIME IN TEXAS 2015 6 (2015), http://www.dps.texas.gov/crimereports/15/citCh2.pdf. From this misguided determination, I respectfully dissent.
As a general matter, we are all in agreement, as the majority opinion describes, that the quest in cases such as this one is to determine: (1) what are the elements of generic burglary, and (2) does the Texas statute match those elements? Mathis v. United States , --- U.S. ----, 136 S.Ct. 2243, 2248, 195 L.Ed.2d 604 (2016). If part of the statute does match and part does not, we end up in the divisibility quagmire addressed at length in the majority opinion. But if all parts of the statute match the elements for generic burglary, then the conviction "counts" under the ACCA, regardless of any divisibility issues. I conclude that the latter is true here and, therefore, I respectfully disagree about the necessity of deciding the divisibility of Texas Penal Code § 30.02(a).
But analyzing the first question also requires a bit of a step back. Why are we asking what "generic burglary" is in the first place? It is not a law school exam hypothetical but, rather, an attempt to give effect to Congress's use of the term "burglary" in the ACCA. See Mathis , 136 S.Ct. at 2252 (explaining that the first of three reasons for the approach employed by the Court is effectuating the intent of Congress). Since the Supreme Court first implemented the categorical approach to the ACCA, it has defined "burglary" as "the generic sense in which the term is now used in the criminal code of most States." Taylor v. United States , 495 U.S. 575, 598, 110 S.Ct. 2143, 109 L.Ed.2d 607 (1990). Using that measuring stick-and no Supreme Court case suggests we should not-this case becomes much easier.
Both past and present state statutes indicate § 30.02(a) is generic burglary. At the time the ACCA was amended to include the new definition of burglary, 41 states (covering 84% of the population) defined burglary to reach crimes committed in vehicles used or adapted for overnight habitation (some of which involve generic vehicles which I recognize the Court has clearly excluded from ACCA consideration).1 That pattern continues today, *543with 41 states (covering more than 85% of the population) defining burglary to reach such crimes.2 Similarly, as the Supreme Court has recognized, at the time of the ACCA's passage numerous states protected individuals from burglaries committed by "remaining in" a structure. See id. My tally is more than half the states at the time of the ACCA amendment3 and 30 *544today.4 Texas's § 30.02(a)(2) and (a)(3) fit firmly within the ambit of the "remaining in" statutes that constitute generic burglary.
None of the above matters, of course, if clear Supreme Court precedent binds us to the outcome described in the majority opinion. Our role as a lower court is to faithfully apply the law as interpreted by the Supreme Court. However, I conclude that the majority opinion goes awry in deciding that § 30.02(a)(3) is not "generic burglary." I also conclude that defining "habitation" to include vehicles adapted for overnight accommodation does not remove this subsection from the class of "generic burglary." Accordingly, Herrold's convictions should count for ACCA purposes.
I begin with § 30.02(a)(3). We have longstanding precedent holding that this subsection is not "generic burglary." See United States v. Emeary , 794 F.3d 526 (5th Cir. 2015) ; United States v. Castaneda , 740 F.3d 169 (5th Cir. 2013) (per curiam); United States v. Constante , 544 F.3d 584 (5th Cir. 2008) (per curiam). However, since the majority of the en banc court has determined to reassess precedent concerning § 30.02(a), we can and should reassess this particular precedent as well.
Subsection (a)(3) provides: "(a) A person commits an offense if, without the effective consent of the owner, the person: ... (3) enters a building or habitation and commits or attempts to commit a felony, theft, or an assault." Thus, (a)(3) requires unprivileged entry into the building or habitation, as required for "generic burglary." Herrold argues, however, that (a)(3) differs from "generic burglary" because it does not require the intent to commit the "felony, theft, or assault" to have been formed before or at the time of the unprivileged entry. Our court agreed with this overall argument in United States v. Herrera-Montes , 490 F.3d 390, 392 (5th Cir. 2007) (analyzing *545Tenn. Code Ann. § 39-14-402 ), and in Constante we applied it to (a)(3), see 544 F.3d at 587.
As subsequent decisions from other circuits have demonstrated, the analysis of Constante wholly overlooks that unlawfully "remaining in" a building with intent to commit a crime also qualifies as "generic burglary." United States v. Priddy , 808 F.3d 676, 684-85 (6th Cir. 2015), abrogated on other grounds by United States v. Stitt , 860 F.3d 854 (6th Cir. 2017) (en banc), petition for cert. filed , (U.S. Nov. 24, 2017) (No.17-765) (analyzing the same Tennessee burglary statute as Herrera-Montes and coming to a different result); United States v. Bonilla , 687 F.3d 188, 193-94 (4th Cir. 2012) ; see also United States v. Reina-Rodriguez , 468 F.3d 1147, 1155-56 (9th Cir. 2006), overruled on other grounds by United States v. Grisel , 488 F.3d 844, 851 n.5 (9th Cir. 2007) (en banc). Bonilla explained that excluding statutes such as (a)(3) is based upon a "too rigid" reading of Taylor "given that a defendant convicted under [§] (a)(3) necessarily developed the intent to commit the crime while remaining in the building, if he did not have it at the moment he entered." 687 F.3d at 194.
In Taylor , the Court determined that the restrictive common-law definition of burglary could not have been what Congress intended when it deleted a definition of burglary from the ACCA. 495 U.S. at 593-95, 110 S.Ct. 2143. The Court reasoned that many states had moved beyond the common-law definition, and "construing 'burglary' to mean common-law burglary would come close to nullifying that term's effect in the statute, because few of the crimes now generally recognized as burglaries would fall within the common-law definition." Id. at 594, 110 S.Ct. 2143. Instead, the Court explained that "generic burglary" contains "at least the following elements: an unlawful or unprivileged entry into, or remaining in, a building or other structure, with intent to commit a crime." Id. at 598, 110 S.Ct. 2143 (emphasis added). In light of the Court's express rejection of the common-law definition, and the criminal codes of nearly half the states at the time, the Taylor definition plainly does not require intent to commit an additional crime at the time of entry, as at common law.
In adopting this generic definition, the Court recognized that "exact formulations" of the elements may vary among the states, and so for ACCA purposes, a state statute need only correspond "in substance to the generic meaning of burglary." Id. at 598-99, 110 S.Ct. 2143. Taylor is therefore not concerned with definitional technicalities but, rather, with substantively enforcing Congress's policy of singling out a property crime that bears "inherent potential for harm to persons." Id. at 588, 110 S.Ct. 2143. Indeed, the omission of a definition for burglary following the 1986 ACCA amendments suggests "that Congress did not wish to specify an exact formulation that an offense must meet in order to count as 'burglary' for enhancement purposes." Id. at 598-99, 110 S.Ct. 2143.
The Fourth and Sixth Circuits have accordingly concluded that unlawful entry combined with an attempted or completed felony or theft therein qualifies as generic burglary under Taylor.5 Indeed, the only *546other federal circuit to determine whether a prior conviction under (a)(3) constitutes generic burglary has come to the opposite conclusion than this court has today. See Bonilla , 687 F.3d at 193. In doing so, the Fourth Circuit reasoned that because (a)(3) only applies where a defendant's presence in a building is unlawful, a completed or attempted felony therein necessarily requires intent to commit the felony either prior to unlawful entry or while unlawfully remaining in the building, which is all Taylor requires. Id. In other words, (a)(3) substantively contains the requisite intent element because to attempt or complete a crime requires intent to commit the crime. Similarly, in Priddy , the Sixth Circuit considered a Tennessee statute essentially identical to (a)(3) and found that it substantially corresponds to Taylor's definition of generic burglary. 808 F.3d at 684-85;see also United States v. Ferguson , 868 F.3d 514, 515-16 (6th Cir. 2017) (affirming the continued vitality of Priddy ). The Sixth Circuit reasoned that unlawful entry combined with an attempted or committed felony or theft therein is a " 'remaining-in' variant of generic burglary because someone who enters a building or structure and, while inside, commits or attempts to commit a felony will necessarily have remained inside the building or structure to do so." Priddy , 808 F.3d at 685. Even though the statute does not use the words "remaining in," it nonetheless contains that element because a person must remain in a building to commit a crime inside of it.
Bonilla , Priddy , and this case each illuminate an important aspect of § 30.02(a)(3) : It actually requires more than the minimum described by the Court in Taylor in that it requires an unlawful or unprivileged entry AND the actual commission or attempted commission of a crime; mere intent is not enough.6 There is nothing overbroad or overblown about considering as "generic burglary" an offense that involves an unlawful entry into a structure, plus the intent to commit a crime formed while remaining in the structure as evidenced by the actual commission or attempted commission of the crime. These are not mere irrelevancies a defendant would have no reason to challenge. Cf. Mathis , 136 S.Ct. at 2253 (explaining one of the reasons for an "elements-focus approach" is to avoid the unfairness to defendants who had no reason to dispute facts that were unnecessary to sustain the prior conviction). Thus, the "basic elements" of burglary as established in Taylor are present: 1) unlawful or unprivileged entry into, or remaining in, 2) a building or structure, 3) with intent to commit a crime-here as evidenced by the actual commission or attempted commission of the crime, not mere intent. Taylor , 495 U.S. at 598, 110 S.Ct. 2143. A contrary reading undercuts the very concept of "generic" burglary adopted in Taylor , where the Court said Congress aimed to prevent "offenders from invoking the arcane technicalities of the common-law definition of burglary to evade the [ACCA's] sentence-enhancement *547provision." Id. at 589, 110 S.Ct. 2143.
The majority opinion contends that defining "remaining in" broadly both "involve[s] a less culpable mental state on the part of the defendant" and "presents less danger to victims." Maj. Op. at 534. I respectfully disagree on both counts. The timing of when intent was formed implicates neither the culpability of the perpetrator nor the extent of danger to victims. If a perpetrator forms intent prior to entering a home but, once inside, discovers nothing worth taking, is he or she somehow less culpable or dangerous than a perpetrator who initially unlawfully7 enters without intent to commit an additional crime but, once inside, discovers something worth taking or, surprised by a resident in the home, commits an assault? The fact that (a)(3) requires commission or attempted commission of the crime implicates an even higher degree of culpability than one who commits burglary simply by forming the requisite intent prior to physical entry.
Consequently, because (a)(3) represents "generic burglary," its inclusion in § 30.02 does not render the statute overbroad, even assuming arguendo § 30.02(a) is indivisible.
This conclusion leads me to turn to an issue addressed, but not decided, in the majority opinion, which Herrold asserts-whether the definition of "habitation" is overbroad because it includes "a vehicle that is adapted for the overnight accommodation of persons." TEX. PENAL CODE § 30.01(1). The majority opinion ultimately does not decide the issue, noting there are "powerful arguments" on both sides of the debate. Maj. Op. at 540-41. However, because my outcome does not depend on the divisibility of § 30.02(a), I engage in such debate. Herrold appears to argue that a vehicle, regardless of purpose, is overbroad under § 30.02(a). This leaves open the potentially drastic outcome that generic burglary excludes all vehicles. Thus, I carefully consider the practical limitations and real-world applications of Texas's statute in analyzing whether a "vehicle adapted for overnight accommodation" is overbroad.
As an initial note, it is important to remember that Texas draws a distinction between burglary of vehicles that become "habitations" and ordinary "vehicles." See TEX. PENAL CODE §§ 30.01(3), 30.02, 30.04. Texas Penal Code § 30.04 criminalizes "burglary of vehicles," which a person violates when, "without the effective consent of the owner, he breaks into or enters a vehicle or any part of a vehicle with intent to commit any felony or theft." A "vehicle" is defined as "any device in, on, or by which any person or property is or may be propelled, moved, or drawn in the normal course of commerce or transportation, except such devices as are classified as 'habitation .' " TEX. PENAL CODE § 30.01(3) (emphasis added). Texas draws a clear line between ordinary "vehicles," which are prosecuted under § 30.04 and defined by § 30.01(3), and a "vehicle that is adapted for the overnight accommodation of persons," as defined under § 30.01(1) and prosecuted under § 30.02. Thus, a person who burglarizes an ordinary vehicle not adapted for overnight accommodation of persons cannot be prosecuted under § 30.02.
Despite these distinct statutes, Herrold argues that § 30.02(a) is prosecuted in Texas "to its full, non-generic extent." To find that application of a state statute is *548applied in a non-generic manner, we require "that a defendant must 'at least' point to an actual state case." United States v. Castillo-Rivera , 853 F.3d 218, 223 (5th Cir. 2017) (en banc) (citing Gonzales v. Duenas-Alvarez , 549 U.S. 183, 193, 127 S.Ct. 815, 166 L.Ed.2d 683 (2007) ). But "even pointing to [a case where a statute has been applied non-generically] may not be satisfactory." Id. Herrold brings to our attention an indictment, sentencing documents, and news articles related to a single case where charges were brought against multiple defendants under § 30.02 relating to mobile homes Herrold claims were "warehoused." Frankly, that Herrold searched high and low among hundreds of thousands of Texas burglary convictions over the years and could find only this example supports rather than contradicts the position that the statute is applied only generically. In any event, as the case involves a plea of guilty to the offense after indictment with little facts and no precedential opinion, this case is not an example of a non-generic application of § 30.02, even assuming arguendo that the "warehousing" point matters.8 To the extent Herrold argues other hypothetical scenarios will be non-generically treated, it is well-established that "clever hypotheticals" are not the basis upon which to judge a statute in question. Id. at 224. Stated simply, a Texas prosecutor bears the burden of proving that a "habitation" was burglarized; if insufficient or incredible evidence is put forward that a vehicle is a "habitation" as Texas defines it, the vehicle will not be treated as such. See Blankenship v. State , 780 S.W.2d 198, 209 (Tex. Crim. App. 1988) (en banc). Therefore, I focus on the non-hypothetical, practical applications of (a)(1) rather than implausible and unlikely "what ifs."
The Supreme Court in discussing "automobiles" in Taylor or generic "vehicles" in the Iowa statute in Mathis was not faced with and did not address the question of whether, for purposes of determining what "generic burglary" involves, Congress would have intended to exclude mobile homes or similar vehicles adapted for overnight use. Rather, Taylor expressed concern about generic burglary encompassing crimes such as "shoplifting and theft of goods from a 'locked' but unoccupied automobile," which were not clearly violent felonies, and subjecting citizens of different states to different sentencing enhancement requirements under the ACCA. 495 U.S. at 591, 110 S.Ct. 2143 (citing CAL. PENAL CODE ANN . § 459 (1990) ). Therefore, the Court determined the three elements of generic burglary, described above, to standardize the definition of generic burglary. Id. at 598, 110 S.Ct. 2143. The Court never expressly considered a vehicle that is not only used as a home but particularly adapted for that use and, therefore, did not foreclose debate on the issue.
An understanding of Taylor is critical to resolving this issue. That being said, the term "vehicle" does not appear in the ACCA and only becomes an issue as the statute was interpreted by Taylor and applied to state statutes.9 We do not read *549cases like statutes,10 and therefore, we take "vehicle adapted for overnight accommodation" to mean "the interpretation that best fits within" Taylor 's framework. See Vance v. Ball State Univ. , 570 U.S. 421, 133 S.Ct. 2434, 2446, 186 L.Ed.2d 565 (2013) ; Stitt , 860 F.3d at 881 (Sutton, J., dissenting). Herrold focuses on the use of the term "vehicles," arguing that in Taylor , the Court concluded that "vehicles" are outside the definition of the generic burglary, so, he says, that's it. The Government, on the other hand, points out that the Texas statute distinguishes between "vehicles" and "habitations" and that the latter-defined to encompass brick and mortar as well as mobile homes-is in keeping with the majority of state statutes protecting structures. The Government provided an appendix describing at least 25 states where, at the time of the ACCA's enactment, structural burglary would have included vehicles expressly adapted for overnight accommodation of persons, like the Texas statute.11 Combining those statutes with statutes that include vehicles broadly (which would thus be considered non-generic for ACCA purposes), occupied vehicles would have been included in the burglary statutes of at least 43 states.12 As noted earlier, Taylor explicitly stated that what Congress "meant by 'burglary' [is] the generic sense in which the term is now used in the criminal codes of most States." 495 U.S. at 598, 110 S.Ct. 2143. Taylor also repeatedly spoke of a "building or structure," capturing the idea that the location of the burglary could be a "structure" that was not a "building." That idea captures well the "vehicle adapted for overnight accommodation of persons," which Texas includes within its definition of a habitation, as distinct from "automobiles," which are not included.
The Taylor Court's understanding of Congress's intent when enacting the ACCA further supports the conclusion that burglary of a "vehicle adapted for overnight accommodation" is generic burglary. The Court noted that Congress did not limit ACCA predicate offense burglaries to those that may be especially dangerous, as "Congress apparently thought that all burglaries serious enough to be punishable by imprisonment for more than a year" were potentially violent and "likely to be committed by career criminals."
*550Taylor , 495 U.S. at 588, 110 S.Ct. 2143. Congress included burglary "because of its inherent potential for harm to persons." Id. (emphasis added). A person would likely be present where the person is living, irrespective of whether that is a traditional home or a "vehicle adapted for overnight accommodation." Any other understanding could lead to anomalies, such as a sentencing enhancement for burglarizing an unoccupied building, but no sentencing enhancement if an occupied mobile home is burglarized. This would be inconsistent with Congress's intent to protect individuals from harm. Again, there will be some structures of any kind that are unoccupied, but it is the potential for harm that the Taylor court addressed; the burglar may have no way to know whether the particular structure is currently occupied so including both occupied and unoccupied structures in the definition makes sense.
Further, Congress desired to prevent criminals from "invoking the arcane technicalities of the common-law definition of burglary to evade the sentence-enhancement provision." Id. at 589, 110 S.Ct. 2143. Would excluding a dwelling on the basis of whether it has (or, at some time, had) wheels not be invoking one of those very "arcane technicalities"? Taylor drew the line at the potential presence of people, not wheels.13 To say a traditional home is protected by ACCA enhancements whereas a mobile home is not simply does not comport with Congress's intent and Taylor 's reasoning.
In determining the "contemporary meaning of burglary," the Government notes that the Taylor Court relied on Model Penal Code provisions that explicitly included "vehicles adapted for overnight accommodation" as an ACCA predicate crime. See id. at 598 n.8, 110 S.Ct. 2143. At that time, the Model Penal Code stated that "[a] person is guilty of burglary if he enters a building or occupied structure, or separately secured or occupied portion thereof, with purpose to commit a crime therein." Id. (quoting MODEL PENAL CODE § 221.1 ( AM. LAW INST . 1980) ). The Model Penal Code defined an "occupied structure" as "any structure, vehicle , or place adapted for overnight accommodation of persons , or for carrying on business therein, whether or not a person is actually present." MODEL PENAL CODE § 221.0 ( AM. LAW INST . 1980) (emphasis added); see also § 221.1 cmt. 3 at 73. Notably, this definition mirrors the language in the Texas burglary statute, and numerous other states' burglary statutes. See TEX. PENAL CODE ANN . § 30.01. The identity of definitions shows that the Taylor Court understood the exact language at issue today to constitute generic burglary, and Herrold's argument would narrow Taylor and the Model Penal Code definition on which it based its holding.
Subsequent Supreme Court decisions have not contradicted this understanding. In Shepard v. United States , 544 U.S. 13, 125 S.Ct. 1254, 161 L.Ed.2d 205 (2005), the Court addressed a Massachusetts burglary statute that included vehicles and vessels in general. Id. at 15-16, 125 S.Ct. 1254. The Shepard Court was principally faced with determining the permissible documents to be used to narrow a statute of conviction following a guilty plea, and therefore was not presented with, and did not address, the narrow subset of "vehicles adapted for overnight accommodation." Id. at 26, 125 S.Ct. 1254. Indeed, the Massachusetts *551statutes said nothing about "overnight accommodation." See MASS. GEN. LAWS ANN ., ch. 266, §§ 16, 18 (2000). Meanwhile, in Mathis , the Court analyzed an Iowa burglary statute that included two prongs, one of which criminalized, inter alia, burglary of any "land, water, or air vehicle," and the second which focused on its use-"overnight accommodation, business or other activity, or the storage or safekeeping of anything of value." See State v. Dixon , 826 N.W.2d 516, 2012 WL 6193877 *2, 2012 Iowa App. LEXIS 1043 *6 (Iowa App. 2012) (not designated for publication) (citing State v. Pace , 602 N.W.2d 764, 769 (Iowa 1999) ); see also State v. Rooney , 862 N.W.2d 367, 376-78 (Iowa 2015) (discussing the two prongs). Because it concluded that statute was indivisible, it did not have to determine whether a vehicle adapted for overnight use as an accommodation by itself would qualify, as the Iowa statute also included vehicles used for storage and, thus, encompassed more than generic burglary.14 See Mathis , 136 S.Ct. at 2250 (emphasis omitted).
Because the Supreme Court's precedents do not answer the question directly, we are left to analyze whether burglary of a "vehicle adapted for overnight accommodation" in a state distinguishing such burglaries from those of regular vehicles is more like "generic burglary" of a habitation, which is an ACCA burglary, or more like a burglary of a regular vehicle, which is not.
Our sister circuits have divided on this issue while analyzing the versions of their statutes in effect at the time of the case. The Tenth Circuit has directly assessed the Texas burglary statute at issue here, holding that it encompasses only generic burglary. United States v. Spring , 80 F.3d 1450, 1461-62 (10th Cir. 1996) (noting that Texas's statute was "not analogous to the theft of an automobile or to the other property crimes whose relative lack of severity the Taylor Court (and presumably, Congress) meant to exclude from its generic definition" (quoting United States v. Sweeten , 933 F.2d 765, 771 (9th Cir. 1991), overruled by Grisel , 488 F.3d at 851 n.5 (en banc) ). Most recently, the Seventh Circuit construed the Illinois residential burglary statute to determine that the inclusion of burglary of a "mobile home [or] trailer ... in which at the time of the alleged offense the owners or occupants actually reside" did not preclude the statute from being considered generic burglary. Smith v. United States , 877 F.3d 720, 722, 724 (7th Cir. 2017). Regarding a mobile home, the court noted that, under Illinois law, a "mobile home" is nothing more than a "prefabricated house," easily dismissing the argument that a mobile home is not a "building or structure." Id. at 722-23. Although including the word "trailer" was a closer call, the court looked to the purposes of Taylor to hold that the Illinois residential burglary statute defined generic burglary, despite the fact that it included "[t]railers used as dwellings." Id. at 724-25 ("We think it unlikely that the Justices set out in Taylor to adopt a definition of generic burglary that is satisfied by no more than a handful of states-if by any. Statutes should be read to have consequences rather than to set the stage for semantic exercises.").
While other circuits have held that statutes with language akin to "vehicle adapted for overnight accommodation" do not encompass generic burglary, this determination has not been without debate and *552dissent. See, e.g. , Grisel , 488 F.3d at 849-51 (holding that the Oregon burglary statute was broader than generic burglary, based upon the assumption, questioned by the dissent, that "in the criminal codes of most states, the term 'building or structure' does not encompass objects that could be described loosely as structures but that are either not designed for occupancy or not intended for use in one place"). Some of these circuits did not entertain much, if any, debate on the issue. See, e.g. , United States v. Sims , 854 F.3d 1037, 1040 (8th Cir. 2017), petition for cert. filed , (U.S. Nov. 24, 2017) (No. 17-766); United States v. Lamb , 847 F.3d 928, 931 (8th Cir. 2017), petition for cert. filed , (U.S. July 10, 2017) (No. 17-5152); United States v. Gundy , 842 F.3d 1156, 1165 (11th Cir. 2016), cert. denied , --- U.S. ----, 138 S.Ct. 66, 199 L.Ed.2d 47 (2017) ; United States v. White , 836 F.3d 437, 445-46 (4th Cir. 2016).
An excellent example of the debate associated with this issue is Stitt . In Stitt , the court concluded that Taylor proscribed "all things mobile or transitory" from generic burglary. 860 F.3d at 859. Judge Sutton, writing for himself and five other judges in dissent, disagreed with this characterization of Taylor . Id. at 876 (Sutton, J., dissenting). Judge Sutton replied that the "no-vehicles-or-tents rule implies that every state's basic burglary statute is non-generic ," essentially "render[ing] generic burglary a null set." Id. at 880-81. He argued that this result is not required; "we should give the Court and Congress more credit" than understanding Taylor and the ACCA to mandate an essentially toothless statute. Id. at 881. As Judge Sutton so aptly put it, "[i]t's a strange genus that doesn't include any species." Id. at 880.
Lacking a clear consensus, we are thus brought back to our analysis of Taylor , mindful that we need not leave common sense at the door. Both Congress's and Taylor 's intent seem clear-to protect the public from career criminals that commit or have committed potentially violent felonies. Even setting aside the statutes that (a) are likely considered overbroad due to the inclusion of routine vehicles or (b) are potentially divisible, 25 states' statutes include provisions protecting vehicles adapted or used for habitation.15 The number mushrooms when you add back in the potentially divisible statutes (7 states16 ) and the statutes already overbroad due to the inclusion of vehicles, or a state court's reading of the statute in a way that is overbroad (9 states17 ). This is not, of course, a binding declaration as to whether those statutes are non-generic or divisible; additional analysis would have to be done. But that so many states' statutes would be in question ought to give us pause. We should not impute to Congress such a jarring outcome in the absence of a clear requirement under the law to do so. Careful consideration of Supreme Court precedent plus common sense dictate that this cannot be the result.
Accordingly, I would affirm, and I respectfully dissent from the court's determination not to do so.