Sessions v. Dimaya, 584 U.S. ___ (2018)

NOTICE: This opinion is subject to formal revision before publication in the preliminary print of the United States Reports. Readers are requested to notify the Reporter of Decisions, Supreme Court of the United States, Washington, D. C. 20543, of any typographical or other formal errors, in order that corrections may be made before the preliminary print goes to press.

SUPREME COURT OF THE UNITED STATES

_________________

No. 15–1498

_________________

JEFFERSON B. SESSIONS, III, ATTORNEY GENERAL, PETITIONER v. JAMES GARCIA DIMAYA

on writ of certiorari to the united states court of appeals for the ninth circuit

[April 17, 2018]

Justice Kagan announced the judgment of the Court and delivered the opinion of the Court with respect to Parts I, III, IV–B, and V, and an opinion with respect to Parts II and IV–A, in which Justice Ginsburg, Justice Breyer, and Justice Sotomayor join.

Three Terms ago, in Johnson v. United States, this Court held that part of a federal law’s definition of “violent felony” was impermissibly vague. See 576 U. S. ___ (2015). The question in this case is whether a similarly worded clause in a statute’s definition of “crime of violence” suffers from the same constitutional defect. Adhering to our analysis in Johnson, we hold that it does.

I

The Immigration and Nationality Act (INA) renders deportable any alien convicted of an “aggravated felony” after entering the United States. 8 U. S. C. §1227(a)(2)(A)(iii). Such an alien is also ineligible for cancellation of removal, a form of discretionary relief allowing some deportable aliens to remain in the country. See §§1229b(a)(3), (b)(1)(C). Accordingly, removal is a virtual certainty for an alien found to have an aggravated felony conviction, no matter how long he has previously resided here.

The INA defines “aggravated felony” by listing numerous offenses and types of offenses, often with cross-references to federal criminal statutes. §1101(a)(43); see Luna Torres v. Lynch, 578 U. S. ___, ___ (2016) (slip op., at 2). According to one item on that long list, an aggravated felony includes “a crime of violence (as defined in section 16 of title 18 . . . ) for which the term of imprisonment [is] at least one year.” §1101(a)(43)(F). The specified statute, 18 U. S. C. §16, provides the federal criminal code’s definition of “crime of violence.” Its two parts, often known as the elements clause and the residual clause, cover:

“(a) an offense that has as an element the use, attempted use, or threatened use of physical force against the person or property of another, or

“(b) any other offense that is a felony and that, by its nature, involves a substantial risk that physical force against the person or property of another may be used in the course of committing the offense.”

Section 16(b), the residual clause, is the part of the statute at issue in this case.

To decide whether a person’s conviction “falls within the ambit” of that clause, courts use a distinctive form of what we have called the categorical approach. Leocal v. Ashcroft, 543 U. S. 1, 7 (2004) . The question, we have explained, is not whether “the particular facts” underlying a conviction posed the substantial risk that §16(b) demands. Ibid. Neither is the question whether the statutory elements of a crime require (or entail) the creation of such a risk in each case that the crime covers.[1] The §16(b) inquiry instead turns on the “nature of the offense” gener- ally speaking. Ibid. (referring to §16(b)’s “by its nature” language). More precisely, §16(b) requires a court to ask whether “the ordinary case” of an offense poses the requisite risk. James v. United States, 550 U. S. 192, 208 (2007) ; see infra, at 7.

In the case before us, Immigration Judges employed that analysis to conclude that respondent James Dimaya is deportable as an aggravated felon. A native of the Philippines, Dimaya has resided lawfully in the United States since 1992. But he has not always acted lawfully during that time. Twice, Dimaya was convicted of first-degree burglary under California law. See Cal. Penal Code Ann. §§459, 460(a). Following his second offense, the Government initiated a removal proceeding against him. Both an Immigration Judge and the Board of Immigration Appeals held that California first-degree burglary is a “crime of violence” under §16(b). “[B]y its nature,” the Board reasoned, the offense “carries a substantial risk of the use of force.” App. to Pet. for Cert. 46a. Dimaya sought review in the Court of Appeals for the Ninth Circuit.

While his appeal was pending, this Court held unconstitutional part of the definition of “violent felony” in the Armed Career Criminal Act (ACCA), 18 U. S. C. §924(e). ACCA prescribes a 15-year mandatory minimum sentence if a person convicted of being a felon in possession of a firearm has three prior convictions for a “violent felony.” §924(e)(1). The definition of that statutory term goes as follows:

“any crime punishable by imprisonment for a term exceeding one year . . . that—

“(i) has as an element the use, attempted use, or threatened use of physical force against the person of another; or

“(ii) is burglary, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.” §924(e)(2)(B) (emphasis added).

The italicized portion of that definition (like the similar language of §16(b)) came to be known as the statute’s residual clause. In Johnson v. United States, the Court declared that clause “void for vagueness” under the Fifth Amendment’s Due Process Clause. 576 U. S., at ___–___ (slip op., at 13–14).

Relying on Johnson, the Ninth Circuit held that §16(b), as incorporated into the INA, was also unconstitutionally vague, and accordingly ruled in Dimaya’s favor. See Dimaya v. Lynch, 803 F. 3d 1110, 1120 (2015). Two other Circuits reached the same conclusion, but a third distinguished ACCA’s residual clause from §16’s.[2] We granted certiorari to resolve the conflict. Lynch v. Dimaya, 579 U. S. ___ (2016).

II

“The prohibition of vagueness in criminal statutes,” our decision in Johnson explained, is an “essential” of due process, required by both “ordinary notions of fair play and the settled rules of law.” 576 U. S., at ___ (slip op., at 4) (quoting Connally v. General Constr. Co., 269 U. S. 385, 391 (1926) ). The void-for-vagueness doctrine, as we have called it, guarantees that ordinary people have “fair notice” of the conduct a statute proscribes. Papachristou v. Jacksonville, 405 U. S. 156, 162 (1972) . And the doctrine guards against arbitrary or discriminatory law enforcement by insisting that a statute provide standards to govern the actions of police officers, prosecutors, juries, and judges. See Kolender v. Lawson, 461 U. S. 352, 357 –358 (1983). In that sense, the doctrine is a corollary of the separation of powers—requiring that Congress, rather than the executive or judicial branch, define what conduct is sanctionable and what is not. Cf. id., at 358, n. 7 (“[I]f the legislature could set a net large enough to catch all possible offenders, and leave it to the courts to step inside and say who could be rightfully detained, [it would] substitute the judicial for the legislative department” (internal quotation marks omitted)).

The Government argues that a less searching form of the void-for-vagueness doctrine applies here than in Johnson because this is not a criminal case. See Brief for Petitioner 13–15. As the Government notes, this Court has stated that “[t]he degree of vagueness that the Constitution [allows] depends in part on the nature of the enactment”: In particular, the Court has “expressed greater tolerance of enactments with civil rather than criminal penalties because the consequences of imprecision are qualitatively less severe.” Hoffman Estates v. Flipside, Hoffman Estates, Inc., 455 U. S. 489, 498 –499 (1982). The removal of an alien is a civil matter. See Arizona v. United States, 567 U. S. 387, 396 (2012) . Hence, the Govern- ment claims, the need for clarity is not so strong; even a law too vague to support a conviction or sentence may be good enough to sustain a deportation order. See Brief for Petitioner 25–26.

But this Court’s precedent forecloses that argument, because we long ago held that the most exacting vagueness standard should apply in removal cases. In Jordan v. De George, we considered whether a provision of immigration law making an alien deportable if convicted of a “crime involving moral turpitude” was “sufficiently definite.” 341 U. S. 223, 229 (1951) . That provision, we noted, “is not a criminal statute” (as §16(b) actually is). Id., at 231; supra, at 1–2. Still, we chose to test (and ultimately uphold) it “under the established criteria of the ‘void for vagueness’ doctrine” applicable to criminal laws. 341 U. S., at 231. That approach was demanded, we explained, “in view of the grave nature of deportation,” ibid.—a “drastic measure,” often amounting to lifelong “banishment or exile,” ibid. (quoting Fong Haw Tan v. Phelan, 333 U. S. 6, 10 (1948) ).

Nothing in the ensuing years calls that reasoning into question. To the contrary, this Court has reiterated that deportation is “a particularly severe penalty,” which may be of greater concern to a convicted alien than “any potential jail sentence.” Jae Lee v. United States, 582 U. S. ___, ___ (2017) (slip op., at 11) (quoting Padilla v. Kentucky, 559 U. S. 356, 365, 368 (2010) ). And we have observed that as federal immigration law increasingly hinged deportation orders on prior convictions, removal proceedings became ever more “intimately related to the criminal process.” Chaidez v. United States, 568 U. S. 342, 352 (2013) (quoting Padilla, 559 U. S., at 365). What follows, as Jordan recognized, is the use of the same standard in the two settings.

For that reason, the Government cannot take refuge in a more permissive form of the void-for-vagueness doctrine than the one Johnson employed. To salvage §16’s residual clause, even for use in immigration hearings, the Gov- ernment must instead persuade us that it is materially clearer than its now-invalidated ACCA counterpart. That is the issue we next address, as guided by Johnson’s analysis.

III

Johnson is a straightforward decision, with equally straightforward application here. Its principal section begins as follows: “Two features of [ACCA’s] residual clause conspire to make it unconstitutionally vague.” 576 U. S., at ___ (slip op., at 5). The opinion then identifies each of those features and explains how their joinder produced “hopeless indeterminacy,” inconsistent with due process. Id., at ___ (slip op., at 7). And with that reasoning, Johnson effectively resolved the case now before us. For §16’s residual clause has the same two features as ACCA’s, combined in the same constitutionally problem- atic way. Consider those two, just as Johnson described them:

“In the first place,” Johnson explained, ACCA’s residual clause created “grave uncertainty about how to estimate the risk posed by a crime” because it “tie[d] the judicial assessment of risk” to a hypothesis about the crime’s “ordinary case.” Id., at ___ (slip op., at 5). Under the clause, a court focused on neither the “real-world facts” nor the bare “statutory elements” of an offense. Ibid. Instead, a court was supposed to “imagine” an “idealized ordinary case of the crime”—or otherwise put, the court had to identify the “kind of conduct the ‘ordinary case’ of a crime involves.” Ibid. But how, Johnson asked, should a court figure that out? By using a “statistical analysis of the state reporter? A survey? Expert evidence? Google? Gut instinct?” Ibid. (internal quotation marks omitted). ACCA provided no guidance, rendering judicial accounts of the “ordinary case” wholly “speculative.” Ibid. Johnson gave as its prime example the crime of attempted bur- glary. One judge, contemplating the “ordinary case,” would imagine the “violent encounter” apt to ensue when a “would-be burglar [was] spotted by a police officer [or] private security guard.” Id., at ___–___ (slip op., at 5–6). Another judge would conclude that “any confrontation” was more “likely to consist of [an observer’s] yelling ‘Who’s there?’ . . . and the burglar’s running away.” Id., at ___ (slip op., at 6). But how could either judge really know? “The residual clause,” Johnson summarized, “offer[ed] no reliable way” to discern what the ordinary version of any offense looked like. Ibid. And without that, no one could tell how much risk the offense generally posed.

Compounding that first uncertainty, Johnson continued, was a second: ACCA’s residual clause left unclear what threshold level of risk made any given crime a “violent felony.” See ibid. The Court emphasized that this feature alone would not have violated the void-for-vagueness doctrine: Many perfectly constitutional statutes use imprecise terms like “serious potential risk” (as in ACCA’s residual clause) or “substantial risk” (as in §16’s). The problem came from layering such a standard on top of the requisite “ordinary case” inquiry. As the Court explained:

“[W]e do not doubt the constitutionality of laws that call for the application of a qualitative standard such as ‘substantial risk’ to real-world conduct; the law is full of instances where a man’s fate depends on his estimating rightly . . . some matter of degree[.] The residual clause, however, requires application of the ‘serious potential risk’ standard to an idealized ordinary case of the crime. Because the elements necessary to determine the imaginary ideal are uncertain[,] this abstract inquiry offers significantly less predictability than one that deals with the actual . . . facts.” Id., at ___ (slip op., at 12) (some internal quotation marks, citations, and alterations omitted).

So much less predictability, in fact, that ACCA’s residual clause could not pass constitutional muster. As the Court again put the point, in the punch line of its decision: “By combining indeterminacy about how to measure the risk posed by a crime with indeterminacy about how much risk it takes for the crime to qualify as a violent felony, the residual clause” violates the guarantee of due process. Id., at ___ (slip op., at 6).[3]

Section 16’s residual clause violates that promise in just the same way. To begin where Johnson did, §16(b) also calls for a court to identify a crime’s “ordinary case” in order to measure the crime’s risk. The Government explicitly acknowledges that point here. See Brief for Petitioner 11 (“Section 16(b), like [ACCA’s] residual clause, requires a court to assess the risk posed by the ordinary case of a particular offense”). And indeed, the Government’s briefing in Johnson warned us about that likeness, observing that §16(b) would be “equally susceptible to [an] objection” that focused on the problems of positing a crime’s ordinary case. Supp. Brief for Respondent, O. T. 2014, No. 13–7120, pp. 22–23. Nothing in §16(b) helps courts to perform that task, just as nothing in ACCA did. We can as well repeat here what we asked in Johnson: How does one go about divining the conduct entailed in a crime’s ordinary case? Statistical analyses? Surveys? Experts? Google? Gut instinct? See Johnson, 576 U. S., at ___ (slip op., at 5); supra, at 7; post, at 16–17 (Gorsuch, J., concurring in part and concurring in judgment). And we can as well reiterate Johnson’s example: In the ordinary case of attempted burglary, is the would-be culprit spotted and confronted, or scared off by a yell? See post, at 16 (opinion of Gorsuch, J.) (offering other knotty examples). Once again, the questions have no good answers; the “ordinary case” remains, as Johnson described it, an excessively “speculative,” essentially inscrutable thing. 576 U. S., at ___ (slip op., at 5); accord post, at 27 (Thomas, J., dissenting).[4]

And §16(b) also possesses the second fatal feature of ACCA’s residual clause: uncertainty about the level of risk that makes a crime “violent.” In ACCA, that threshold was “serious potential risk”; in §16(b), it is “substantial risk.” See supra, at 2, 4. But the Government does not argue that the latter formulation is any more determinate than the former, and for good reason. As The Chief Justice’s valiant attempt to do so shows, that would be slicing the baloney mighty thin. See post, at 5–6 (dissenting opinion). And indeed, Johnson as much as equated the two phrases: Return to the block quote above, and note how Johnson—as though anticipating this case—refers to them interchangeably, as alike examples of imprecise “qualitative standard[s].” See supra, at 8; 576 U. S., at ___ (slip op., at 12). Once again, the point is not that such a non-numeric standard is alone problematic: In Johnson’s words, “we do not doubt” the constitutionality of applying §16(b)’s “substantial risk [standard] to real-world conduct.” Id., at ___ (slip op., at 12) (internal quotation marks omitted). The difficulty comes, in §16’s residual clause just as in ACCA’s, from applying such a standard to “a judge-imagined abstraction”—i.e., “an idealized ordinary case of the crime.” Id., at ___, ___ (slip op., at 6, 12). It is then that the standard ceases to work in a way consistent with due process.

In sum, §16(b) has the same “[t]wo features” that “conspire[d] to make [ACCA’s residual clause] unconstitutionally vague.” Id., at ___ (slip op., at 5). It too “requires a court to picture the kind of conduct that the crime involves in ‘the ordinary case,’ and to judge whether that abstraction presents” some not-well-specified-yet-sufficiently-large degree of risk. Id., at ___ (slip op., at 4). The result is that §16(b) produces, just as ACCA’s residual clause did, “more unpredictability and arbitrariness than the Due Process Clause tolerates.” Id., at ___ (slip op., at 6).

IV

The Government and dissents offer two fundamentally different accounts of how §16(b) can escape unscathed from our decision in Johnson. Justice Thomas accepts that the ordinary-case inquiry makes §16(b) “impossible to apply.” Post, at 27. His solution is to overthrow our historic understanding of the statute: We should now read §16(b), he says, to ask about the risk posed by a particular defendant’s particular conduct. In contrast, the Government, joined by The Chief Justice, accepts that §16(b), as long interpreted, demands a categorical approach, rather than a case-specific one. They argue only that “distinctive textual features” of §16’s residual clause make applying it “more predictable” than its ACCA counter- part. Brief for Petitioner 28, 29. We disagree with both arguments.

A

The essentials of Justice Thomas’s position go as follows. Section 16(b), he says, cannot have one meaning, but could have one of two others. See post, at 27. The provision cannot demand an inquiry merely into the elements of a crime, because that is the province of §16(a). See supra, at 2 (setting out §16(a)’s text). But that still leaves a pair of options: the categorical, ordinary-case approach and the “underlying-conduct approach,” which asks about the specific way in which a defendant committed a crime. Post, at 25. According to Justice Thomas, each option is textually viable (although he gives a slight nod to the latter based on §16(b)’s use of the word “involves”). See post, at 24–26. What tips the scales is that only one—the conduct approach—is at all “workable.” Post, at 27. The difficulties of the ordinary-case inquiry, Justice Thomas rightly observes, underlie this Court’s view that §16(b) is too vague. So abandon that inquiry, Justice Thomas urges. After all, he reasons, it is the Court’s “plain duty,” under the constitutional avoidance canon, to adopt any reasonable construction of a statute that escapes constitutional problems. Post, at 28–29 (quoting United States ex rel. Attorney General v. Delaware & Hudson Co., 213 U. S. 366, 407 (1909) ).

For anyone who has read Johnson, that argument will ring a bell. The dissent there issued the same invitation, based on much the same reasoning, to jettison the categorical approach in residual-clause cases. 576 U. S., at ___–___ (slip op., at 9–13) (opinion of Alito, J.). The Court declined to do so. It first noted that the Government had not asked us to switch to a fact-based inquiry. It then observed that the Court “had good reasons” for originally adopting the categorical approach, based partly on ACCA’s text (which, by the way, uses the word “involves” identi- cally) and partly on the “utter impracticability” of the alter- native. Id., at ___ (slip op., at 13) (majority opinion). “The only plausible interpretation” of ACCA’s residual clause, we concluded, “requires use of the categorical approach”—even if that approach could not in the end satisfy constitutional standards. Ibid. (internal quotation marks and alteration omitted).

The same is true here—except more so. To begin where Johnson did, the Government once again “has not asked us to abandon the categorical approach in residual-clause cases.” Ibid. To the contrary, and as already noted, the Government has conceded at every step the correctness of that statutory construction. See supra, at 9. And this time, the Government’s decision is even more noteworthy than before—precisely because the Johnson dissent laid out the opposite view, presenting it in prepackaged form for the Government to take off the shelf and use in the §16(b) context. Of course, we are not foreclosed from going down Justice Thomas’s path just because the Government has not done so. But we find it significant that the Government cannot bring itself to say that the fact-based approach Justice Thomas proposes is a tenable interpretation of §16’s residual clause.

Perhaps one reason for the Government’s reluctance is that such an approach would generate its own constitutional questions. As Justice Thomas relates, post, at 22, 28, this Court adopted the categorical approach in part to “avoid[ ] the Sixth Amendment concerns that would arise from sentencing courts’ making findings of fact that properly belong to juries.” Descamps v. United States, 570 U. S. 254, 267 (2013) . Justice Thomas thinks that issue need not detain us here because “the right of trial by jury ha[s] no application in a removal proceeding.” Post, at 28 (internal quotation marks omitted). But although this particular case involves removal, §16(b) is a criminal statute, with criminal sentencing consequences. See supra, at 2. And this Court has held (it could hardly have done otherwise) that “we must interpret the statute con- sistently, whether we encounter its application in a criminal or noncriminal context.” Leocal, 543 U. S., at 12, n. 8. So Justice Thomas’s suggestion would merely ping-pong us from one constitutional issue to another. And that means the avoidance canon cannot serve, as he would like, as the interpretive tie breaker.

In any event, §16(b)’s text creates no draw: Best read, it demands a categorical approach. Our decisions have consistently understood language in the residual clauses of both ACCA and §16 to refer to “the statute of conviction, not to the facts of each defendant’s conduct.” Taylor v. United States, 495 U. S. 575, 601 (1990) ; see Leocal, 543 U. S., at 7 (Section 16 “directs our focus to the ‘offense’ of conviction . . . rather than to the particular facts”). Simple references to a “conviction,” “felony,” or “offense,” we have stated, are “read naturally” to denote the “crime as generally committed.” Nijhawan v. Holder, 557 U. S. 29, 34 (2009) ; see Leocal, 543 U. S., at 7; Johnson, 576 U. S., at ___ (slip op., at 13). And the words “by its nature” in §16(b) make that meaning all the clearer. The statute, recall, directs courts to consider whether an offense, by its nature, poses the requisite risk of force. An offense’s “nature” means its “normal and characteristic quality.” Webster’s Third New International Dictionary 1507 (2002). So §16(b) tells courts to figure out what an offense normally—or, as we have repeatedly said, “ordinarily”—entails, not what happened to occur on one occasion. And the same conclusion follows if we pay attention to language that is missing from §16(b). As we have observed in the ACCA context, the absence of terms alluding to a crime’s circumstances, or its commission, makes a fact-based interpretation an uncomfortable fit. See Descamps, 570 U. S., at 267. If Congress had wanted judges to look into a felon’s actual conduct, “it presumably would have said so; other statutes, in other contexts, speak in just that way.” Id., at 267–268.[5] The upshot of all this textual evidence is that §16’s residual clause—like ACCA’s, except still more plainly—has no “plausible” fact-based reading. Johnson, 576 U. S., at ___ (slip op., at 13).

And finally, the “utter impracticability”—and associated inequities—of such an interpretation is as great in the one statute as in the other. Ibid. This Court has often described the daunting difficulties of accurately “reconstruct[ing],” often many years later, “the conduct underlying [a] conviction.” Ibid.; Descamps, 570 U. S., at 270; Taylor, 495 U. S., at 601–602. According to Justice Thomas, we need not worry here because immigration judges have some special factfinding talent, or at least experience, that would mitigate the risk of error attaching to that endeavor in federal courts. See post, at 30. But we cannot see putting so much weight on the superior factfinding prowess of (notoriously overburdened) immigration judges. And as we have said before, §16(b) is a criminal statute with applications outside the immigration context. See supra, at 2, 13. Once again, then, we have no ground for discovering a novel interpretation of §16(b) that would remove us from the dictates of Johnson.

B

Agreeing that is so, the Government (joined by The Chief Justice) takes a narrower path to the same desired result. It points to three textual discrepancies between ACCA’s residual clause and §16(b), and argues that they make §16(b) significantly easier to apply. But each turns out to be the proverbial distinction without a difference. None relates to the pair of features—the ordinary-case inquiry and a hazy risk threshold—that Johnson found to produce impermissible vagueness. And none otherwise affects the determinacy of the statutory inquiry into whether a prior conviction is for a violent crime. That is why, contrary to the Government’s final argument, the experience of applying both statutes has generated confusion and division among lower courts.

1

The Government first—and foremost—relies on §16(b)’s express requirement (absent from ACCA) that the risk arise from acts taken “in the course of committing the offense.” Brief for Petitioner 31. (The Chief Justice’s dissent echoes much of this argument. See post, at 6–7.) Because of that “temporal restriction,” a court applying §16(b) may not “consider risks arising after” the offense’s commission is over. Ibid. In the Government’s view, §16(b)’s text thereby demands a “significantly more focused inquiry” than did ACCA’s residual clause. Id., at 32.

To assess that claim, start with the meaning of §16(b)’s “in the course of” language. That phrase, understood in the normal way, includes the conduct occurring throughout a crime’s commission—not just the conduct sufficient to satisfy the offense’s formal elements. The Government agrees with that construction, explaining that the words “in the course of” sweep in everything that happens while a crime continues. See Tr. of Oral Arg. 57–58 (Oct. 2, 2017) (illustrating that idea with reference to conspiracy, burglary, kidnapping, and escape from prison). So, for example, conspiracy may be a crime of violence under §16(b) because of the risk of force while the conspiracy is ongoing (i.e., “in the course of” the conspiracy); it is irrelevant that conspiracy’s elements are met as soon as the participants have made an agreement. See ibid.; United States v. Doe, 49 F. 3d 859, 866 (CA2 1995). Similarly, and closer to home, burglary may be a crime of violence under §16(b) because of the prospects of an encounter while the burglar remains in a building (i.e., “in the course of” the burglary); it does not matter that the elements of the crime are met at the precise moment of his entry. See Tr. of Oral Arg. 57–58 (Oct. 2, 2017); James, 550 U. S., at 203. In other words, a court applying §16(b) gets to consider everything that is likely to take place for as long as a crime is being committed.

Because that is so, §16(b)’s “in the course of” language does little to narrow or focus the statutory inquiry. All that the phrase excludes is a court’s ability to consider the risk that force will be used after the crime has entirely concluded—so, for example, after the conspiracy has dissolved or the burglar has left the building. We can construct law-school-type hypotheticals fitting that fact pattern—say, a burglar who constructs a booby trap that later knocks out the homeowner. But such imaginative forays cannot realistically affect a court’s view of the ordinary case of a crime, which is all that matters under the statute. See supra, at 2–3, 7. In the ordinary case, the riskiness of a crime arises from events occurring during its commission, not events occurring later. So with or without §16(b)’s explicit temporal language, a court applying the section would do the same thing—ask what usually happens when a crime goes down.

And that is just what courts did when applying ACCA’s residual clause—and for the same reason. True, that clause lacked an express temporal limit. But not a single one of this Court’s ACCA decisions turned on conduct that might occur after a crime’s commission; instead, each hinged on the risk arising from events that could happen while the crime was ongoing. See, e.g., Sykes v. United States, 564 U. S. 1, 10 (2011) (assessing the risks attached to the “confrontations that initiate and terminate” vehicle flight, along with “intervening” events); Chambers v. United States, 555 U. S. 122, 128 (2009) (rejecting the Government’s argument that violent incidents “occur[ring] long after” a person unlawfully failed to report to prison rendered that crime a violent felony). Nor could those decisions have done otherwise, given the statute’s concern with the ordinary (rather than the outlandish) case. Once again, the riskiness of a crime in the ordinary case depends on the acts taken during—not after—its commission. Thus, the analyses under ACCA’s residual clause and §16(b) coincide.

The upshot is that the phrase “in the course of” makes no difference as to either outcome or clarity. Every offense that could have fallen within ACCA’s residual clause might equally fall within §16(b). And the difficulty of deciding whether it does so remains just as intractable. Indeed, we cannot think of a single federal crime whose treatment becomes more obvious under §16(b) than under ACCA because of the words “in the course of.”[6] The phrase, then, cannot cure the statutory indeterminacy Johnson described.

Second, the Government (and again, The Chief Justice’s dissent, see post, at 6) observes that §16(b) focuses on the risk of “physical force” whereas ACCA’s residual clause asked about the risk of “physical injury.” The §16(b) inquiry, the Government says, “trains solely” on the conduct typically involved in a crime. Brief for Petitioner 36. By contrast, the Government continues, ACCA’s residual clause required a second inquiry: After describing the ordinary criminal’s conduct, a court had to “speculate about a chain of causation that could possibly result in a victim’s injury.” Ibid. The Government’s conclusion is that the §16(b) inquiry is “more specific.” Ibid.

But once more, we struggle to see how that statutory distinction would matter. To begin with, the first of the Government’s two steps—defining the conduct in the ordinary case—is almost always the difficult part. Once that is accomplished, the assessment of consequences tends to follow as a matter of course. So, for example, if a crime is likely enough to lead to a shooting, it will also be likely enough to lead to an injury. And still more important, §16(b) involves two steps as well—and essentially the same ones. In interpreting statutes like §16(b), this Court has made clear that “physical force” means “force capable of causing physical pain or injury.” Johnson v. United States, 559 U. S. 133, 140 (2010) (defining the term for purposes of deciding what counts as a “violent” crime). So under §16(b) too, a court must not only identify the conduct typically involved in a crime, but also gauge its potential consequences. Or said a bit differently, evaluating the risk of “physical force” itself entails considering the risk of “physical injury.” For those reasons, the force/injury distinction is unlikely to affect a court’s analysis of whether a crime qualifies as violent. All the same crimes might—or, then again, might not—satisfy both requirements. Accordingly, this variance in wording cannot make ACCA’s residual clause vague and §16(b) not.

Third, the Government briefly notes that §16(b), unlike ACCA’s residual clause, is not preceded by a “confusing list of exemplar crimes.” Brief for Petitioner 38. (The Chief Justice’s dissent reiterates this argument, with some additional references to our caselaw. See post, at 10–12.) Here, the Government is referring to the offenses ACCA designated as violent felonies independently of the residual clause (i.e., burglary, arson, extortion, and use of explosives). See supra, at 4. According to the Government, those crimes provided “contradictory and opaque indications” of what non-specified offenses should also count as violent. Brief for Petitioner 38. Because §16(b) lacks any such enumerated crimes, the Government concludes, it avoids the vagueness of ACCA’s residual clause.

We readily accept a part of that argument. This Court for several years looked to ACCA’s listed crimes for help in giving the residual clause meaning. See, e.g., Begay v. United States, 553 U. S. 137, 142 (2008) ; James, 550 U. S., at 203. But to no avail. As the Government relates (and Johnson explained), the enumerated crimes were themselves too varied to provide such assistance. See Brief for Petitioner 38–40; 576 U. S., at ___ (slip op., at 12). Trying to reconcile them with each other, and then compare them to whatever unlisted crime was at issue, drove many a judge a little batty. And more to the point, the endeavor failed to bring any certainty to the residual clause’s application. See Brief for Petitioner 38–40.

But the Government’s conclusion does not follow. To say that ACCA’s listed crimes failed to resolve the residual clause’s vagueness is hardly to say they caused the problem. Had they done so, Johnson would not have needed to strike down the clause. It could simply have instructed courts to give up on trying to interpret the clause by reference to the enumerated offenses. (Contrary to The Chief Justice’s suggestion, see post, at 12, discarding an interpretive tool once it is found not to actually aid in interpretation hardly “expand[s]” the scope of a statute.) That Johnson went so much further—invalidating a statutory provision rather than construing it independently of another—demonstrates that the list of crimes was not the culprit. And indeed, Johnson explicitly said as much. As described earlier, Johnson found the residual clause’s vagueness to reside in just “two” of its features: the ordinary-case requirement and a fuzzy risk standard. See 576 U. S., at ___–___ (slip op., at 5–6); supra, at 7–8. Strip away the enumerated crimes—as Congress did in §16(b)—and those dual flaws yet remain. And ditto the textual indeterminacy that flows from them.

2

Faced with the two clauses’ linguistic similarity, the Government relies significantly on an argument rooted in judicial experience. Our opinion in Johnson, the Government notes, spoke of the longstanding “trouble” that this Court and others had in “making sense of [ACCA’s] residual clause.” 576 U. S., at ___ (slip op., at 9); see Brief for Petitioner 45. According to the Government, §16(b) has not produced “comparable difficulties.” Id., at 46. Lower courts, the Government claims, have divided less often about the provision’s meaning, and as a result this Court granted certiorari on “only a single Section 16(b) case” before this one. Ibid.[7] “The most likely explanation,” the Government concludes, is that “Section 16(b) is clearer” than its ACCA counterpart. Id., at 47.

But in fact, a host of issues respecting §16(b)’s application to specific crimes divide the federal appellate courts. Does car burglary qualify as a violent felony under §16(b)? Some courts say yes, another says no.[8] What of statutory rape? Once again, the Circuits part ways.[9] How about evading arrest? The decisions point in different directions.[10] Residential trespass? The same is true.[11] Those examples do not exhaust the current catalogue of Circuit conflicts concerning §16(b)’s application. See Brief for National Immigration Project of the National Lawyers Guild et al. as Amici Curiae 7–18 (citing divided appellate decisions as to the unauthorized use of a vehicle, firearms possession, and abduction). And that roster would just expand with time, mainly because, as Johnson explained, precious few crimes (of the thousands that fill the statute books) have an obvious, non-speculative—and therefore undisputed—“ordinary case.” See 576 U. S., at ___–___ (slip op., at 5–6).

Nor does this Court’s prior handling of §16(b) cases support the Government’s argument. To be sure, we have heard oral argument in only two cases arising from §16(b) (including this one), as compared with five involving ACCA’s residual clause (including Johnson).[12] But while some of those ACCA suits were pending before us, we received a number of petitions for certiorari presenting related issues in the §16(b) context. And after issuing the relevant ACCA decisions, we vacated the judgments in those §16(b) cases and remanded them for further consideration.[13] That we disposed of the ACCA and §16(b) petitions in that order, rather than its opposite, provides no reason to disregard the indeterminacy that §16(b) shares with ACCA’s residual clause.

And of course, this Court’s experience in deciding ACCA cases only supports the conclusion that §16(b) is too vague. For that record reveals that a statute with all the same hallmarks as §16(b) could not be applied with the predictability the Constitution demands. See id., at ___–___ (slip op., at 6–9); supra, at 6–9. The Government would condemn us to repeat the past—to rerun the old ACCA tape, as though we remembered nothing from its first showing. But why should we disregard a lesson so hard learned? “Insanity,” Justice Scalia wrote in the last ACCA residual clause case before Johnson, “is doing the same thing over and over again, but expecting different results.” Sykes, 564 U. S., at 28 (dissenting opinion). We abandoned that lunatic practice in Johnson and see no reason to start it again.

V

Johnson tells us how to resolve this case. That decision held that “[t]wo features of [ACCA’s] residual clause conspire[d] to make it unconstitutionally vague.” 576 U. S., at ___ (slip op., at 5). Because the clause had both an ordinary-case requirement and an ill-defined risk threshold, it necessarily “devolv[ed] into guesswork and intuition,” invited arbitrary enforcement, and failed to provide fair notice. Id., at ___ (slip op., at 8). Section 16(b) possesses the exact same two features. And none of the minor linguistic disparities in the statutes makes any real difference. So just like ACCA’s residual clause, §16(b) “produces more unpredictability and arbitrariness than the Due Process Clause tolerates.” Id., at ___ (slip op., at 6). We accordingly affirm the judgment of the Court of Appeals.

It is so ordered.

SUPREME COURT OF THE UNITED STATES

_________________

No. 15–1498

_________________

JEFFERSON B. SESSIONS, III, ATTORNEY GENERAL, PETITIONER v. JAMES GARCIA DIMAYA

on writ of certiorari to the united states court of appeals for the ninth circuit

[April 17, 2018]

Justice Gorsuch, concurring in part and concurring in the judgment.

Vague laws invite arbitrary power. Before the Revolution, the crime of treason in English law was so capa- ciously construed that the mere expression of disfavored opinions could invite transportation or death. The founders cited the crown’s abuse of “pretended” crimes like this as one of their reasons for revolution. See Declaration of Independence ¶21. Today’s vague laws may not be as invidious, but they can invite the exercise of arbitrary power all the same—by leaving the people in the dark about what the law demands and allowing prosecutors and courts to make it up.

The law before us today is such a law. Before holding a lawful permanent resident alien like James Dimaya subject to removal for having committed a crime, the Immigration and Nationality Act requires a judge to determine that the ordinary case of the alien’s crime of conviction involves a substantial risk that physical force may be used. But what does that mean? Just take the crime at issue in this case, California burglary, which applies to everyone from armed home intruders to door-to-door salesmen peddling shady products. How, on that vast spectrum, is anyone supposed to locate the ordinary case and say whether it includes a substantial risk of physical force? The truth is, no one knows. The law’s silence leaves judges to their intuitions and the people to their fate. In my judgment, the Constitution demands more.

*

I begin with a foundational question. Writing for the Court in Johnson v. United States, 576 U. S. ___ (2015), Justice Scalia held the residual clause of the Armed Career Criminal Act void for vagueness because it invited “more unpredictability and arbitrariness” than the Constitution allows. Id., at ___ (slip op., at 6). Because the residual clause in the statute now before us uses almost exactly the same language as the residual clause in Johnson, respect for precedent alone would seem to suggest that both clauses should suffer the same judgment.

But first in Johnson and now again today Justice Thomas has questioned whether our vagueness doctrine can fairly claim roots in the Constitution as originally understood. See, e.g., post, at 2–6 (dissenting opinion); Johnson, supra, at ___–___ (opinion concurring in judgment) (slip op., at 6–18). For its part, the Court has yet to offer a reply. I believe our colleague’s challenge is a serious and thoughtful one that merits careful attention. At day’s end, though, it is a challenge to which I find my- self unable to subscribe. Respectfully, I am persuaded in- stead that void for vagueness doctrine, at least properly conceived, serves as a faithful expression of ancient due process and separation of powers principles the framers recognized as vital to ordered liberty under our Constitution.

Consider first the doctrine’s due process underpinnings. The Fifth and Fourteenth Amendments guarantee that “life, liberty, or property” may not be taken “without due process of law.” That means the government generally may not deprive a person of those rights without affording him the benefit of (at least) those “customary procedures to which freemen were entitled by the old law of England.” Pacific Mut. Life Ins. Co. v. Haslip, 499 U. S. 1, 28 (1991) (Scalia, J., concurring in judgment) (internal quotation marks omitted). Admittedly, some have suggested that the Due Process Clause does less work than this, allowing the government to deprive people of their liberty through whatever procedures (or lack of them) the government’s current laws may tolerate. Post, at 3, n. 1 (opinion of Thomas, J.) (collecting authorities). But in my view the weight of the historical evidence shows that the clause sought to ensure that the people’s rights are never any less secure against governmental invasion than they were at common law. Lord Coke took this view of the English due process guarantee. 1 E. Coke, The Second Part of the Institutes of the Laws of England 50 (1797). John Rutledge, our second Chief Justice, explained that Coke’s teachings were carefully studied and widely adopted by the framers, becoming “ ‘almost the foundations of our law.’ ” Klopfer v. North Carolina, 386 U. S. 213, 225 (1967) . And many more students of the Constitution besides—from Justice Story to Justice Scalia—have agreed that this view best represents the original understanding of our own Due Process Clause. See, e.g., Murray’s Lessee v. Hoboken Land & Improvement Co., 18 How. 272, 277 (1856); 3 J. Story, Commentaries on the Constitution of the United States §1783, p. 661 (1833); Pacific Mut., supra, at 28–29 (opinion of Scalia, J.); Eberle, Procedural Due Process: The Original Understanding, 4 Const. Comment. 339, 341 (1987).

Perhaps the most basic of due process’s customary protections is the demand of fair notice. See Connally v. General Constr. Co., 269 U. S. 385, 391 (1926) ; see also Note, Textualism as Fair Notice, 123 Harv. L. Rev. 542, 543 (2009) (“From the inception of Western culture, fair notice has been recognized as an essential element of the rule of law”). Criminal indictments at common law had to provide “precise and sufficient certainty” about the charges involved. 4 W. Blackstone, Commentaries on the Laws of England 301 (1769) (Blackstone). Unless an “offence [was] set forth with clearness and certainty,” the indictment risked being held void in court. Id., at 302 (emphasis deleted); 2 W. Hawkins, Pleas of the Crown, ch. 25, §§99, 100, pp. 244–245 (2d ed. 1726) (“[I]t seems to have been anciently the common practice, where an indictment appeared to be [in]sufficient, either for its uncertainty or the want of proper legal words, not to put the defendant to answer it”).

The same held true in civil cases affecting a person’s life, liberty, or property. A civil suit began by obtaining a writ—a detailed and specific form of action asking for particular relief. Bellia, Article III and the Cause of Action, 89 Iowa L. Rev. 777, 784–786 (2004); Subrin, How Equity Conquered Common Law: The Federal Rules of Civil Procedure in Historical Perspective, 135 U. Pa. L. Rev. 909, 914–915 (1987). Because the various civil writs were clearly defined, English subjects served with one would know with particularity what legal requirement they were alleged to have violated and, accordingly, what would be at issue in court. Id., at 917; Moffitt, Pleadings in the Age of Settlement, 80 Ind. L. J. 727, 731 (2005). And a writ risked being held defective if it didn’t provide fair notice. Goldington v. Bassingburn, Y. B. Trin. 3 Edw. II, f. 27b (1310) (explaining that it was “the law of the land” that “no one [could] be taken by surprise” by having to “answer in court for what [one] has not been warned to answer”).

The requirement of fair notice applied to statutes too. Blackstone illustrated the point with a case involving a statute that made “stealing sheep, or other cattle” a fel- ony. 1 Blackstone 88 (emphasis deleted). Because the term “cattle” embraced a good deal more then than it does now (including wild animals, no less), the court held the statute failed to provide adequate notice about what it did and did not cover—and so the court treated the term “cattle” as a nullity. Ibid. All of which, Blackstone added, had the salutary effect of inducing the legislature to reenter the field and make itself clear by passing a new law extending the statute to “bulls, cows, oxen,” and more “by name.” Ibid.

This tradition of courts refusing to apply vague statutes finds parallels in early American practice as well. In The Enterprise, 8 F. Cas. 732 (No. 4,499) (CC NY 1810), for example, Justice Livingston found that a statute setting the circumstances in which a ship may enter a port during an embargo was too vague to be applied, concluding that “the court had better pass” the statutory terms by “as unintelligible and useless” rather than “put on them, at great uncertainty, a very harsh signification, and one which the legislature may never have designed.” Id., at 735. In United States v. Sharp, 27 F. Cas. 1041 (No. 16,264) (CC Pa. 1815), Justice Washington confronted a statute which prohibited seamen from making a “revolt.” Id., at 1043. But he was unable to determine the meaning of this provision “by any authority . . . either in the common, admiralty, or civil law.” Ibid. As a result, he declined to “recommend to the jury, to find the prisoners guilty of making, or endeavouring to make a revolt, however strong the evidence may be.” Ibid.[1]

Nor was the concern with vague laws confined to the most serious offenses like capital crimes. Courts refused to apply vague laws in criminal cases involving relatively modest penalties. See, e.g., McJunkins v. State, 10 Ind. 140, 145 (1858). They applied the doctrine in civil cases too. See, e.g., Drake v. Drake, 15 N. C. 110, 115 (1833); Commonwealth v. Bank of Pennsylvania, 3 Watts & Serg. 173, 177 (Pa. 1842). As one court put it, “all laws” “ought to be expressed in such a manner as that its meaning may be unambiguous, and in such language as may be readily understood by those upon whom it is to operate.” McConvill v. Mayor and Aldermen of Jersey City, 39 N. J. L. 38, 42 (1876). “ ‘It is impossible . . . to dissent from the doctrine of Lord Coke, that acts of parliament ought to be plainly and clearly, and not cunningly and darkly penned, especially in penal matters.’ ” Id., at 42–43.

These early cases, admittedly, often spoke in terms of construing vague laws strictly rather than declaring them void. See, e.g., post, at 4–5 (opinion of Thomas, J.); Johnson, 576 U. S., at ___–___ (opinion of Thomas, J.) (slip op., at 8–10). But in substance void the law is often exactly what these courts did: rather than try to construe or interpret the statute before them, judges frequently held the law simply too vague to apply. Blackstone, for example, did not suggest the court in his illustration should have given a narrowing construction to the term “cattle,” but argued against giving it any effect at all. 1 Blackstone 88; see also Scalia, Assorted Canards of Contemporary Legal Analysis, 40 Case W. Res. L. Rev. 581, 582 (1989) (“I doubt . . . that any modern court would go to the lengths described by Blackstone in its application of the rule that penal statutes are to be strictly construed”); Note, Indefinite Criteria of Definiteness in Statutes, 45 Harv. L. Rev. 160, n. 3 (1931) (explaining that “since strict construction, in effect, nullified ambiguous provisions, it was but a short step to declaring them void ab initio”); supra, at 5, n. 1 (state courts holding vague statutory terms “void” or “null”).

What history suggests, the structure of the Constitution confirms. Many of the Constitution’s other provisions presuppose and depend on the existence of reasonably clear laws. Take the Fourth Amendment’s requirement that arrest warrants must be supported by probable cause, and consider what would be left of that requirement if the alleged crime had no meaningful boundaries. Or take the Sixth Amendment’s mandate that a defendant must be informed of the accusations against him and allowed to bring witnesses in his defense, and consider what use those rights would be if the charged crime was so vague the defendant couldn’t tell what he’s alleged to have done and what sort of witnesses he might need to rebut that charge. Without an assurance that the laws supply fair notice, so much else of the Constitution risks becoming only a “parchment barrie[r]” against arbitrary power. The Federalist No. 48, p. 308 (C. Rossiter ed. 1961) (J. Madison).

Although today’s vagueness doctrine owes much to the guarantee of fair notice embodied in the Due Process Clause, it would be a mistake to overlook the doctrine’s equal debt to the separation of powers. The Constitution assigns “[a]ll legislative Powers” in our federal government to Congress. Art. I, §1. It is for the people, through their elected representatives, to choose the rules that will govern their future conduct. See The Federalist No. 78, at 465 (A. Hamilton) (“The legislature . . . prescribes the rules by which the duties and rights of every citizen are to be regulated”). Meanwhile, the Constitution assigns to judges the “judicial Power” to decide “Cases” and “Controversies.” Art. III, §2. That power does not license judges to craft new laws to govern future conduct, but only to “discer[n] the course prescribed by law” as it currently exists and to “follow it” in resolving disputes between the people over past events. Osborn v. Bank of United States, 9 Wheat. 738, 866 (1824).

From this division of duties, it comes clear that legislators may not “abdicate their responsibilities for setting the standards of the criminal law,” Smith v. Goguen, 415 U. S. 566, 575 (1974) , by leaving to judges the power to decide “the various crimes includable in [a] vague phrase,” Jordan v. De George, 341 U. S. 223, 242 (1951) (Jackson, J., dissenting). For “if the legislature could set a net large enough to catch all possible offenders, and leave it to the courts to step inside and say who could be rightfully detained, and who should be set at large[,] [t]his would, to some extent, substitute the judicial for the legislative department of government.” Kolender v. Lawson, 461 U. S. 352, 358 , n. 7 (1983) (internal quotation marks omitted). Nor is the worry only that vague laws risk allowing judges to assume legislative power. Vague laws also threaten to transfer legislative power to police and prosecutors, leaving to them the job of shaping a vague statute’s contours through their enforcement decisions. See Grayned v. City of Rockford, 408 U. S. 104, 108 –109 (1972) (“A vague law impermissibly delegates basic policy matters to policemen, judges, and juries for resolution on an ad hoc and subjective basis”).

These structural worries are more than just formal ones. Under the Constitution, the adoption of new laws restricting liberty is supposed to be a hard business, the product of an open and public debate among a large and diverse number of elected representatives. Allowing the legislature to hand off the job of lawmaking risks substituting this design for one where legislation is made easy, with a mere handful of unelected judges and prosecutors free to “condem[n] all that [they] personally disapprove and for no better reason than [they] disapprove it.” Jordan, supra, at 242 (Jackson, J., dissenting). Nor do judges and prosecutors act in the open and accountable forum of a legislature, but in the comparatively obscure confines of cases and controversies. See, e.g., A. Bickel, The Least Dangerous Branch: The Supreme Court at the Bar of Politics 151 (1962) (“A vague statute delegates to administrators, prosecutors, juries, and judges the authority of ad hoc decision, which is in its nature difficult if not impossible to hold to account, because of its narrow impact”). For just these reasons, Hamilton warned, while “liberty can have nothing to fear from the judiciary alone,” it has “every thing to fear from” the union of the judicial and legislative powers. The Federalist No. 78, at 466. No doubt, too, for reasons like these this Court has held “that the more important aspect of vagueness doctrine ‘is not actual notice, but . . . the requirement that a legislature establish minimal guidelines to govern law enforcement’ ” and keep the separate branches within their proper spheres. Kolender, supra, at 358 (quoting Goguen, supra, at 575 (emphasis added)).

*

Persuaded that vagueness doctrine enjoys a secure footing in the original understanding of the Constitution, the next question I confront concerns the standard of review. What degree of imprecision should this Court tolerate in a statute before declaring it unconstitutionally vague? For its part, the government argues that where (as here) a person faces only civil, not criminal, consequences from a statute’s operation, we should declare the law unconstitutional only if it is “unintelligible.” But in the criminal context this Court has generally insisted that the law must afford “ordinary people . . . fair notice of the conduct it punishes.” Johnson, 576 U. S., at ___ (slip op., at 3). And I cannot see how the Due Process Clause might often require any less than that in the civil context either. Fair notice of the law’s demands, as we’ve seen, is “the first essential of due process.” Connally, 269 U. S., at 391. And as we’ve seen, too, the Constitution sought to preserve a common law tradition that usually aimed to ensure fair notice before any deprivation of life, liberty, or property could take place, whether under the banner of the criminal or the civil law. See supra, at 2–7.

First principles aside, the government suggests that at least this Court’s precedents support adopting a less-than-fair-notice standard for civil cases. But even that much I do not see. This Court has already expressly held that a “stringent vagueness test” should apply to at least some civil laws—those abridging basic First Amendment freedoms. Hoffman Estates v. Flipside, Hoffman Estates, Inc., 455 U. S. 489, 499 (1982) . This Court has made clear, too, that due process protections against vague laws are “not to be avoided by the simple label a State chooses to fasten upon its conduct or its statute.” Giaccio v. Pennsylvania, 382 U. S. 399, 402 (1966) . So the happenstance that a law is found in the civil or criminal part of the statute books cannot be dispositive. To be sure, this Court has also said that what qualifies as fair notice depends “in part on the nature of the enactment.” Hoffman Estates, 455 U. S., at 498. And the Court has sometimes “expressed greater tolerance of enactments with civil rather than criminal penalties because the consequences of imprecision are qualitatively less severe.” Id., at 498–499. But to acknowledge these truisms does nothing to prove that civil laws must always be subject to the government’s emaci- ated form of review.

In fact, if the severity of the consequences counts when deciding the standard of review, shouldn’t we also take account of the fact that today’s civil laws regularly impose penalties far more severe than those found in many criminal statutes? Ours is a world filled with more and more civil laws bearing more and more extravagant punishments. Today’s “civil” penalties include confiscatory rather than compensatory fines, forfeiture provisions that allow homes to be taken, remedies that strip persons of their professional licenses and livelihoods, and the power to commit persons against their will indefinitely. Some of these penalties are routinely imposed and are routinely graver than those associated with misdemeanor crimes—and often harsher than the punishment for felonies. And not only are “punitive civil sanctions . . . rapidly expanding,” they are “sometimes more severely punitive than the parallel criminal sanctions for the same conduct.” Mann, Punitive Civil Sanctions: The Middleground Between Criminal and Civil Law, 101 Yale L. J. 1795, 1798 (1992) (emphasis added). Given all this, any suggestion that criminal cases warrant a heightened standard of review does more to persuade me that the criminal standard should be set above our precedent’s current threshold than to suggest the civil standard should be buried below it.

Retreating to a more modest line of argument, the government emphasizes that this case arises in the immigration context and so implicates matters of foreign relations where the Executive enjoys considerable constitutional authority. But to acknowledge that the President has broad authority to act in this general area supplies no justification for allowing judges to give content to an impermissibly vague law.

Alternatively still, Justice Thomas suggests that, at least at the time of the founding, aliens present in this country may not have been understood as possessing any rights under the Due Process Clause. For support, he points to the Alien Friends Act of 1798. An Act Concerning Aliens §1, 1Stat. 571; post, at 6–12 (opinion of Thomas, J.). But the Alien Friends Act—better known as the “Alien” part of the Alien and Sedition Acts—is one of the most notorious laws in our country’s history. It was understood as a temporary war measure, not one that the legislature would endorse in a time of tranquility. See, e.g., Fehlings, Storm on the Constitution: The First Deportation Law, 10 Tulsa J. Comp. & Int’l L. 63, 70–71 (2002). Yet even then it was widely condemned as unconstitutional by Madison and many others. It also went unenforced, may have cost the Federalist Party its existence, and lapsed a mere two years after its enactment. With this fuller view, it seems doubtful the Act tells us a great deal about aliens’ due process rights at the founding.[2]

Besides, none of this much matters. Whether Madison or his adversaries had the better of the debate over the constitutionality of the Alien Friends Act, Congress is surely free to extend existing forms of liberty to new classes of persons—liberty that the government may then take only after affording due process. See, e.g., Sandin v. Conner, 515 U. S. 472, 477 –478 (1995); Easterbrook, Substance and Due Process, 1982 S. Ct. Rev. 85, 88 (“If . . . the constitution, statute, or regulation creates a liberty or property interest, then the second step—determining ‘what process is due’—comes into play”). Madison made this very point, suggesting an alien’s admission in this country could in some circumstances be analogous to “the grant of land to an individual,” which “may be of favor not of right; but the moment the grant is made, the favor becomes a right, and must be forfeited before it can be taken away.” Madison’s Report 319. And, of course, that’s exactly what Congress eventually chose to do here. Decades ago, it enacted a law affording Mr. Dimaya lawful permanent residency in this country, extending to him a statutory liberty interest others traditionally have enjoyed to remain in and move about the country free from physical imprisonment and restraint. See Dimaya v. Lynch, 803 F. 3d 1110, 1111 (CA9 2015); 8 U. S. C. §§1101(20), 1255. No one suggests Congress had to enact statutes of this sort. And exactly what processes must attend the deprivation of a statutorily afforded liberty interest like this may pose serious and debatable questions. Cf. Murray’s Lessee, 18 How., at 277 (approving summary procedures in another context). But however summary those procedures might be, it’s hard to fathom why fair notice of the law—the most venerable of due process’s requirements—would not be among them. Connally, 269 U. S., at 391.[3]

Today, a plurality of the Court agrees that we should reject the government’s plea for a feeble standard of review, but for a different reason. Ante, at 5–6. My colleagues suggest the law before us should be assessed under the fair notice standard because of the special grav- ity of its civil deportation penalty. But, grave as that penalty may be, I cannot see why we would single it out for special treatment when (again) so many civil laws today impose so many similarly severe sanctions. Why, for example, would due process require Congress to speak more clearly when it seeks to deport a lawfully resident alien than when it wishes to subject a citizen to indefinite civil commitment, strip him of a business license essential to his family’s living, or confiscate his home? I can think of no good answer.

*

With the fair notice standard now in hand, all that remains is to ask how it applies to the case before us. And here at least the answer comes readily for me: to the extent it requires an “ordinary case” analysis, the portion of the Immigration and Nationality Act before us fails the fair notice test for the reasons Justice Scalia identified in Johnson and the Court recounts today.

Just like the statute in Johnson, the statute here instructs courts to impose special penalties on individuals previously “convicted of” a “crime of violence.” 8 U. S. C. §§1227(a)(2)(A)(iii), 1101(a)(43)(F). Just like the statute in Johnson, the statute here fails to specify which crimes qualify for that label. Instead, and again like the statute in Johnson, the statute here seems to require a judge to guess about the ordinary case of the crime of conviction and then guess whether a “substantial risk” of “physical force” attends its commission. 18 U. S. C. §16(b); Johnson, 576 U. S., at ___–___ (slip op., at 4–5). Johnson held that a law that asks so much of courts while offering them so little by way of guidance is unconstitutionally vague. And I do not see how we might reach a different judgment here.

Any lingering doubt is resolved for me by taking account of just some of the questions judges trying to apply the statute using an ordinary case analysis would have to confront. Does a conviction for witness tampering ordinarily involve a threat to the kneecaps or just the promise of a bribe? Does a conviction for kidnapping ordinarily involve throwing someone into a car trunk or a noncustodial parent picking up a child from daycare? These questions do not suggest obvious answers. Is the court supposed to hold evidentiary hearings to sort them out, entertaining experts with competing narratives and statistics, before deciding what the ordinary case of a given crime looks like and how much risk of violence it poses? What is the judge to do if there aren’t any reliable statistics available? Should (or must) the judge predict the effects of new technology on what qualifies as the ordinary case? After all, surely the risk of injury calculus for crimes like larceny can be expected to change as more thefts are committed by computer rather than by gunpoint. Or instead of requiring real evidence, does the statute mean to just leave it all to a judicial hunch? And on top of all that may be the most difficult question yet: at what level of generality is the inquiry supposed to take place? Is a court supposed to pass on the ordinary case of burglary in the relevant neighborhood or county, or should it focus on statewide or even national experience? How is a judge to know? How are the people to know?

The implacable fact is that this isn’t your everyday ambiguous statute. It leaves the people to guess about what the law demands—and leaves judges to make it up. You cannot discern answers to any of the questions this law begets by resorting to the traditional canons of statutory interpretation. No amount of staring at the statute’s text, structure, or history will yield a clue. Nor does the statute call for the application of some preexisting body of law familiar to the judicial power. The statute doesn’t even ask for application of common experience. Choice, pure and raw, is required. Will, not judgment, dictates the result.

*

Having said this much, it is important to acknowledge some limits on today’s holding too. I have proceeded on the premise that the Immigration and Nationality Act, as it incorporates §16(b) of the criminal code, commands courts to determine the risk of violence attending the ordinary case of conviction for a particular crime. I have done so because no party before us has argued for a different way to read these statutes in combination; because our precedent seemingly requires this approach; and because the government itself has conceded (repeatedly) that the law compels it. Johnson, supra, at ___ (slip op., at 13); Taylor v. United States, 495 U. S. 575, 600 (1990) ; Brief for Petitioner 11, 30, 32, 36, 40, 47 (conceding that an ordinary case analysis is required).

But any more than that I would not venture. In response to the problems engendered by the ordinary case analysis, Justice Thomas suggests that we should overlook the government’s concession about the propriety of that approach; reconsider our precedents endorsing it; and read the statute as requiring us to focus on the facts of the alien’s crime as committed rather than as the facts appear in the ordinary case of conviction. Post, at 20–32. But normally courts do not rescue parties from their concessions, maybe least of all concessions from a party as able to protect its interests as the federal government. And normally, too, the crucible of adversarial testing is crucial to sound judicial decisionmaking. We rely on it to “yield insights (or reveal pitfalls) we cannot muster guided only by our own lights.” Maslenjak v. United States, 582 U. S. ___, ___ (2017) (Gorsuch, J., concurring in part and concurring in judgment) (slip op., at 2).

While sometimes we may or even must forgo the adversarial process, I do not see the case for doing so today. Maybe especially because I am not sure Justice Thomas’s is the only available alternative reading of the statute we would have to consider, even if we did reject the government’s concession and wipe the precedential slate clean. We might also have to consider an interpretation that would have courts ask not whether the alien’s crime of conviction ordinarily involves a risk of physical force, or whether the defendant’s particular crime involved such a risk, but whether the defendant’s crime of conviction always does so. After all, the language before us requires a conviction for an “offense . . . that, by its nature, involves a substantial risk of physical force.” 18 U. S. C. §16(b) (emphasis added). Plausibly, anyway, the word “nature” might refer to an inevitable characteristic of the offense; one that would present itself automatically, whenever the statute is violated. See 10 Oxford English Dictionary 247 (2d ed. 1989). While I remain open to different arguments about our precedent and the proper reading of language like this, I would address them in another case, whether involving the INA or a different statute, where the parties have a chance to be heard and we might benefit from their learning.

It’s important to note the narrowness of our decision today in another respect too. Vagueness doctrine represents a procedural, not a substantive, demand. It does not forbid the legislature from acting toward any end it wishes, but only requires it to act with enough clarity that reasonable people can know what is required of them and judges can apply the law consistent with their limited office. Our history surely bears examples of the judicial misuse of the so-called “substantive component” of due process to dictate policy on matters that belonged to the people to decide. But concerns with substantive due process should not lead us to react by withdrawing an ancient procedural protection compelled by the original meaning of the Constitution.

Today’s decision sweeps narrowly in yet one more way. By any fair estimate, Congress has largely satisfied the procedural demand of fair notice even in the INA provision before us. The statute lists a number of specific crimes that can lead to a lawful resident’s removal—for example, murder, rape, and sexual abuse of a minor. 8 U. S. C. §1101(a)(43)(A). Our ruling today does not touch this list. We address only the statute’s “residual clause” where Congress ended its own list and asked us to begin writing our own. Just as Blackstone’s legislature passed a revised statute clarifying that “cattle” covers bulls and oxen, Congress remains free at any time to add more crimes to its list. It remains free, as well, to write a new residual clause that affords the fair notice lacking here. Congress might, for example, say that a conviction for any felony carrying a prison sentence of a specified length opens an alien to removal. Congress has done almost exactly this in other laws. See, e.g., 18 U. S. C. §922(g). What was done there could be done here.

But those laws are not this law. And while the statute before us doesn’t rise to the level of threatening death for “pretended offences” of treason, no one should be surprised that the Constitution looks unkindly on any law so vague that reasonable people cannot understand its terms and judges do not know where to begin in applying it. A government of laws and not of men can never tolerate that arbitrary power. And, in my judgment, that foundational principle dictates today’s result. Because I understand them to be consistent with what I have said here, I join Parts I, III, IV–B, and V of the Court’s opinion and concur in the judgment.

SUPREME COURT OF THE UNITED STATES

_________________

No. 15–1498

_________________

JEFFERSON B. SESSIONS, III, ATTORNEY GENERAL, PETITIONER v. JAMES GARCIA DIMAYA

on writ of certiorari to the united states court of appeals for the ninth circuit

[April 17, 2018]

Chief Justice Roberts, with whom Justice Kennedy, Justice Thomas, and Justice Alito join, dissenting.

In Johnson v. United States, we concluded that the residual clause of the Armed Career Criminal Act was unconstitutionally vague, given the “indeterminacy of the wide-ranging inquiry” it required. 576 U. S. ___, ___ (2015) (slip op., at 5). Today, the Court relies wholly on Johnson—but only some of Johnson—to strike down another provision, 18 U. S. C. §16(b). Because §16(b) does not give rise to the concerns that drove the Court’s decision in Johnson, I respectfully dissent.

I

The term “crime of violence” appears repeatedly throughout the Federal Criminal Code. Section 16 of Title 18 defines it to mean:

“(a) an offense that has as an element the use, attempted use, or threatened use of physical force against the person or property of another, or

“(b) any other offense that is a felony and that, by its nature, involves a substantial risk that physical force against the person or property of another may be used in the course of committing the offense.”

This definition of “crime of violence” is also incorporated in the definition of “aggravated felony” in the Immigration and Nationality Act. 8 U. S. C. §1101(a)(43)(F) (“aggravated felony” includes “a crime of violence (as defined in section 16 of title 18, but not including a purely political offense) for which the term of imprisonment [is] at least one year” (footnote omitted)). A conviction for an aggravated felony carries serious consequences under the immigration laws. It can serve as the basis for an alien’s removal from the United States, and can preclude cancellation of removal by the Attorney General. §§1227(a)(2)(A)(iii), 1229b(a)(3).

Those consequences came to pass in respondent James Dimaya’s case. An Immigration Judge and the Board of Immigration Appeals interpreted §16(b) to cover Dimaya’s two prior convictions for first-degree residential burglary under California law, subjecting him to removal. To stave off that result, Dimaya argued that the language of §16(b) was void for vagueness under the Due Process Clause of the Fifth Amendment.

The parties begin by disputing whether a criminal or more relaxed civil vagueness standard should apply in resolving Dimaya’s challenge. A plurality of the Court rejects the Government’s argument in favor of a civil standard, because of the “grave nature of deportation,” Jordan v. De George, 341 U. S. 223, 231 (1951) ; see ante, at 6 (plurality opinion); Justice Gorsuch does so for broader reasons, see ante, at 10–15 (Gorsuch, J., concurring in part and concurring in judgment). I see no need to resolve which standard applies, because I would hold that §16(b) is not unconstitutionally vague even under the standard applicable to criminal laws.

II

This is not our first encounter with §16(b). In Leocal v. Ashcroft, 543 U. S. 1 (2004) , we were asked to decide whether either subsection of §16 covers a particular category of state crimes, specifically DUI offenses involving no more than negligent conduct. 543 U. S., at 6. Far from finding §16(b) “hopeless[ly] indetermina[te],” Johnson, 576 U. S., at ___ (slip op., at 7), we considered the provision clear and unremarkable: “while §16(b) is broader than §16(a) in the sense that physical force need not actually be applied,” the provision “simply covers offenses that naturally involve a person acting in disregard of the risk that physical force might be used against another in committing an offense,” Leocal, 543 U. S., at 10–11. Applying that standard to the state offense at issue, we concluded—unanimously—that §16(b) “cannot be read to include [a] conviction for DUI causing serious bodily injury under Florida law.” Id., at 11.

Leocal thus provides a model for how courts should assess whether a particular crime “by its nature” involves a risk of the use of physical force. At the outset, our opinion set forth the elements of the Florida DUI statute, which made it a felony “for a person to operate a vehicle while under the influence and, ‘by reason of such operation, caus[e] . . . [s]erious bodily injury to another.’ ” 543 U. S., at 7. Our §16(b) analysis, in turn, focused on those specific elements in concluding that a Florida offender’s acts would not naturally give rise to the requisite risk of force “in the course of committing the offense.” Id., at 11. “In no ‘ordinary or natural’ sense,” we explained, “can it be said that a person risks having to ‘use’ physical force against another person in the course of operating a vehicle while intoxicated and causing injury.” Ibid.

The Court holds that the same provision we had no trouble applying in Leocal is in fact incapable of reasoned application. The sole justification for this turnabout is the resemblance between the language of §16(b) and the language of the residual clause of the Armed Career Criminal Act (ACCA) that was at issue in Johnson. The latter provision defined a “violent felony” to include “any crime punishable by imprisonment for a term exceeding one year . . . that . . . is burglary, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.” 18 U. S. C. §924(e)(2)(B)(ii) (emphasis added).

In Johnson, we concluded that the ACCA residual clause (the “or otherwise” language) gave rise to two forms of intractable uncertainty, which “conspire[d]” to render the provision unconstitutionally vague. 576 U. S., at ___ (slip op., at 5). First, the residual clause asked courts to gauge the “potential risk” of “physical injury” posed by the conduct involved in the crime. Ibid. That inquiry, we determined, entailed not only an evaluation of the “criminal’s behavior,” but also required courts to consider “how the idealized ordinary case of the crime subsequently plays out.” Ibid. Second, the residual clause obligated courts to compare that risk to an indeterminate standard—one that was inextricably linked to the provision’s four enumerated crimes, which presented differing kinds and degrees of risk. Id., at ___ (slip op., at 6). This murky confluence of features, each of which “may [have been] tolerable in isolation,” together “ma[de] a task for us which at best could be only guesswork.” Id., at ___ (slip op., at 10).

Section 16(b) does not present the same ambiguities. The two provisions do correspond to some extent. Under our decisions, both ask the sentencing court to consider whether a particular offense, defined without regard to the facts of the conviction, poses a specified risk. And, relevant to both statutes, we have explained that in deciding whether statutory elements inherently produce a risk, a court must take into account how those elements will ordinarily be fulfilled. See James v. United States, 550 U. S. 192, 208 (2007) (this categorical inquiry asks “whether the conduct encompassed by the elements of the offense, in the ordinary case, presents” the requisite risk).[1] In the Court’s view, that effectively resolves this case. But the Court too readily dismisses the significant textual distinctions between §16(b) and the ACCA residual clause. See also ante, at 2 (opinion of Gorsuch, J.). Those differences undermine the conclusion that §16(b) shares each of the “dual flaws” of that clause. Ante, at 21 (majority opinion).

To begin, §16(b) yields far less uncertainty “about how to estimate the risk posed by a crime.” Johnson, 576 U. S., at ___ (slip op., at 5). There are three material differences between §16(b) and the ACCA residual clause in this respect. First, the ACCA clause directed the reader to consider whether the offender’s conduct presented a “potential risk” of injury. Forced to give meaning to that befuddling choice of phrase—which layered one indeterminate term on top of another—we understood the word “potential” to signify that “Congress intended to encompass possibilities even more contingent or remote than “a simple ‘risk.’ ” James, 550 U. S., at 207–208. As we explained in Johnson, that made for a “speculative” inquiry “detached from statutory elements.” 576 U. S., at ___ (slip op., at 5). In other words, the offense elements could not constrain the risk inquiry in the manner they do here. See Leocal, 543 U. S., at 11. The “serious potential risk” standard also forced courts to assess in an expansive way the “collateral consequences” of the perpetrator’s acts. For example, courts had to take into account the concern that others might cause injury in attempting to apprehend the offender. See Sykes v. United States, 564 U. S. 1, 8 –9 (2011). Section 16(b), on the other hand, asks about “risk” alone, a familiar concept of everyday life. It therefore calls for a commonsense inquiry that does not compel a court to venture beyond the offense elements to consider contingent and remote possibilities.

Second, §16(b) focuses exclusively on the risk that the offender will “use[ ]” “physical force” “against” another person or another person’s property. Thus, unlike the ACCA residual clause, “§16(b) plainly does not encompass all offenses which create a ‘substantial risk’ that injury will result from a person’s conduct.” Leocal, 543 U. S., at 10, n. 7 (emphasis added). The point is not that an inquiry into the risk of “physical force” is markedly more determinate than an inquiry into the risk of “physical injury.” But see ante, at 19–20. The difference is that §16(b) asks about the risk that the offender himself will actively employ force against person or property. That language does not sweep in all instances in which the offender’s acts, or another person’s reaction, might result in unintended or negligent harm.

Third, §16(b) has a temporal limit that the ACCA residual clause lacked: The “substantial risk” of force must arise “in the course of committing the offense.” Properly interpreted, this means the statute requires a substantial risk that the perpetrator will use force while carrying out the crime. See Leocal, 543 U. S., at 10 (“The reckless disregard in §16 relates . . . to the risk that the use of physical force against another might be required in committing a crime.”). The provision thereby excludes more attenuated harms that might arise following the completion of the crime. The ACCA residual clause, by contrast, contained no similar language restricting its scope. And the absence of such a limit, coupled with the reference to “potential” risks, gave courts free rein to classify an offense as a violent felony based on injuries that might occur after the offense was over and done. See, e.g., United States v. Benton, 639 F. 3d 723, 732 (CA6 2011) (finding that “solicitation to commit aggravated assault” qualified under the ACCA residual clause on the theory that the solicited individual might subsequently carry out the requested act).

Why does any of this matter? Because it mattered in Johnson. More precisely, the expansive language in the ACCA residual clause contributed to our determination that the clause gave rise to “grave uncertainty about how to estimate the risk posed by a crime.” 576 U. S., at ___ (slip op., at 5). “Critically,” we said—a word that tends to mean something—“picturing the criminal’s behavior is not enough.” Ibid. (emphasis added). Instead, measuring “potential risk” “seemingly require[d] the judge to imagine how the idealized ordinary case of the crime subsequently plays out.” Ibid. (emphasis added). Not so here. In applying §16(b), considering “the criminal’s behavior” is enough.

Those three distinctions—the unadorned reference to “risk,” the focus on the offender’s own active employment of force, and the “in the course of committing” limitation—also mean that many hard cases under ACCA are easier under §16(b). Take the firearm possession crime from Johnson itself, which had as its constituent elements (1) unlawfully (2) possessing (3) a short-barreled shotgun. None of those elements, “by its nature,” carries “a substantial risk” that the possessor will use force against another “in the course of committing the offense.” Nothing inherent in the act of firearm possession, even when it is unlawful, gives rise to a substantial risk that the owner will then shoot someone. See United States v. Serafin, 562 F. 3d 1105, 1113 (CA10 2009) (recognizing that “Leocal instructs [a court] to focus not on whether possession will likely result in violence, but instead whether one possessing an unregistered weapon necessarily risks the need to employ force to commit possession”).[2] Yet short-barreled shotgun possession presented a closer question under the ACCA residual clause, because the “serious potential risk” language seemingly directed us to consider “the circumstances and conduct that ordinarily attend the offense,” in addition to the offense itself. Johnson, 576 U. S., at ___ (Alito, J., dissenting) (slip op., at 17); see id., at ___–___ (slip op., at 19–20) (reasoning that the crime must qualify because “a person who chooses to break the law and risk the heavy criminal penalty incurred by possessing a notoriously dangerous weapon is [likely] to use that weapon in violent ways”).

Failure to report to a penal institution, the subject of Chambers v. United States, 555 U. S. 122 (2009) , is another crime “whose treatment becomes more obvious under §16(b) than under ACCA,” ante, at 18. In Chambers, the Government argued that the requisite risk of injury arises not necessarily at the time the offender fails to report to prison, but instead later, when an officer attempts to recapture the fugitive. 555 U. S., at 128. The majority is correct that we ultimately “reject[ed]” the Government’s contention. Ante, at 18. But we did so after “assum[ing] for argument’s sake” its premise—that is, “the relevance of violence that may occur long after an offender fails to report.” 555 U. S., at 128; see id., at 129 (looking at 160 cases of “failure to report” and observing that “none at all involved violence . . . during the commission of the offense itself, [nor] during the offender’s later apprehension”). The “in the course of committing the offense” language in §16(b) helpfully forecloses that debate.

DUI offenses are yet another example. Because §16(b) asks about the risk that the offender will “use[ ]” “physical force,” we readily concluded in Leocal that the subsection does not cover offenses where the danger arises from the offender’s negligent or accidental conduct, including drunk driving. 543 U. S., at 11. Applying the ACCA residual clause proved more trying. When asked to decide whether the clause covered drunk driving offenses, a majority of the Court concluded that the answer was no. Begay v. United States, 553 U. S. 137 (2008) . Our decision was based, however, on the inference that the clause must cover only “purposeful, ‘violent,’ and ‘aggressive’ conduct”—a test derived not from the “conduct that presents a serious potential risk of physical injury” language, but instead by reference to (what we guessed to be) the unifying characteristics of the enumerated offenses. Id., at 144–145. Four Members of the Court criticized that test, see id., at 150–153 (Scalia, J., concurring in judgment); id., at 158–160, 162–163 (Alito, J., dissenting), though they themselves disagreed about whether DUIs were covered, see id., at 153–154 (opinion of Scalia, J.); id., at 156–158 (opinion of Alito, J.). And the Court distanced itself from the Begay requirement only a few years later when confronting the crime of vehicular flight. See Sykes, 564 U. S., at 12–13; Johnson, 576 U. S., at ___–___ (slip op., at 8–9).

Which brings me to the second part of the Court’s analysis: its objection that §16(b), like the ACCA residual clause, leaves “uncertainty about the level of risk that makes a crime ‘violent.’ ” Ante, at 10. The “substantial risk” standard in §16(b) is significantly less confusing because it is not tied to a disjointed list of paradigm offenses. Recall that the ACCA provision defined a “violent felony” to include a crime that “is burglary, arson, or extortion, involves use of explosives, or otherwise involves conduct that presents a serious potential risk of physical injury to another.” 18 U. S. C. §924(e)(2)(B)(ii) (emphasis added). As our Court recognized early on, that “otherwise” told the reader to understand the “serious potential risk of physical injury” standard by way of the four enumerated crimes. James, 550 U. S., at 203. But how, exactly? That question dogged our residual clause cases for years, until we said no más in Johnson.

In our first foray, James, we resolved the case by asking whether the risk posed by the crime of attempted burglary was “comparable to that posed by its closest analog among the enumerated offenses,” which was completed burglary. 550 U. S., at 203. While that rule “[took] care of attempted burglary,” it “offer[ed] no help at all with respect to the vast majority of offenses, which have no apparent analog among the enumerated crimes.” Johnson, 576 U. S., at ___ (slip op., at 7). The James dissent, for its part, would have determined the requisite degree of risk from the least dangerous of the enumerated crimes, and compared the offense to that. 550 U. S., at 218–219 (opinion of Scalia, J.). But that approach also proved to be harder than it sounded. See id., at 219–227.

After James came Begay, in which we concluded that the enumerated offenses served as an independent limitation on the kind of crime that could qualify. 553 U. S., at 142; see Chambers, 555 U. S., at 128 (applying the Begay standard). As discussed, that test was short lived (though we did not purport to wholly repudiate it). See Sykes, 564 U. S., at 13. Finally, in Sykes—our penultimate residual clause case—we acknowledged the prior use of the closest-analog test in James, but instead focused on whether the risk posed by vehicular flight was “similar in degree of danger” to the listed offenses of arson and burglary. 564 U. S., at 8–10. As a result, Justice Scalia’s dissent characterized the Sykes majority as applying the test from his prior dissent in James, not James itself. See 564 U. S., at 29–30, 33. This series of precedents laid bare our “repeated inability to craft a principled test out of the statutory text,” id., at 34 (opinion of Scalia, J.), as the Court ultimately acknowledged in Johnson, 576 U. S., at ___ (slip op., at 7).

The enumerated offenses, and our Court’s failed attempts to make sense of them, were essential to Johnson’s conclusion that the residual clause “leaves uncertainty about how much risk it takes for a crime to qualify as a violent felony.” Id., at ___ (slip op., at 6). As Johnson explained, the issue was not that the statute employed a fuzzy standard. That kind of thing appears in the statute books all the time. Id., at ___, ___ (slip op., at 6, 12). In the majority’s retelling today, the difficulty inhered solely in the fact that the statute paired such a standard with the ordinary case inquiry. See ante, at 8, 10–11, 21. But that account sidesteps much of Johnson’s reasoning. See 576 U. S., at ___–___, ___, ___–___, ___ (slip op., at 4–5, 6, 7–9, 12). Our opinion emphasized that the word “otherwise” “force[d]” courts to interpret the amorphous standard “in light of” the four enumerated crimes, which are “not much more similar to one another in kind than in degree of risk posed.” Id., at ___, ___ (slip op., at 6, 8). Or, as Johnson put it more vividly, “[t]he phrase ‘shades of red,’ standing alone, does not generate confusion or unpredictability; but the phrase ‘fire-engine red, light pink, maroon, navy blue, or colors that otherwise involve shades of red’ assuredly does so.” Id., at ___ (slip op., at 12). Indeed, the author of Johnson had previously, and repeatedly, described this feature of the residual clause as the “crucial . . . respect” in which the law was problematic. See James, 550 U. S., at 230, n. 7 (opinion of Scalia, J.); Sykes, 564 U. S., at 35 (opinion of Scalia, J.).

With §16(b), by contrast, a court need simply consider the meaning of the word “substantial”—a word our Court has interpreted and applied innumerable times across a wide variety of contexts.[3] The court does not need to give that familiar word content by reference to four different offenses with varying amounts and kinds of risk.

In its effort to recast a considerable portion of Johnson as dicta, the majority speculates that if the enumerated offenses had truly mattered to the outcome, the Court would have told lower courts to “give up on trying to interpret the clause by reference to” those offenses, rather than striking down the provision entirely. Ante, at 21. No litigant in Johnson suggested that solution, which is not surprising. Such judicial redrafting could have expanded the reach of the criminal provision—surely a job for Congress alone.

In any event, I doubt the majority’s proposal would have done the trick. And that is because the result in Johnson did not follow from the presence of one frustrating textual feature or another. Quite the opposite: The decision emphasized that it was the “sum” of the “uncertainties” in the ACCA residual clause, confirmed by years of experience, that “convince[d]” us the provision was beyond salvage. Johnson, 576 U. S., at ___ (slip op., at 10). Those failings do not characterize the provision at issue here.

III

The more constrained inquiry required under §16(b)—which asks only whether the offense elements naturally carry with them a risk that the offender will use force in committing the offense—does not itself engender “grave uncertainty about how to estimate the risk posed by a crime.” And the provision’s use of a commonplace substantial risk standard—one not tied to a list of crimes that lack a unifying feature—does not give rise to intolerable “uncertainty about how much risk it takes for a crime to qualify.” That should be enough to reject Dimaya’s facial vagueness challenge.[4]

Because I would rely on those distinctions to uphold §16(b), the Court reproaches me for not giving sufficient weight to a “core insight” of Johnson. Ante, at 10, n. 4; see ante, at 15 (opinion of Gorsuch, J.) (arguing that §16(b) runs afoul of Johnson “to the extent [§16(b)] requires an ‘ordinary case’ analysis”). But the fact that the ACCA residual clause required the ordinary case approach was not itself sufficient to doom the law. We instead took pains to clarify that our opinion should not be read to impart such an absolute rule. See Johnson, 576 U. S., at ___ (slip op., at 10). I would adhere to that careful holding and not reflexively extend the decision to a different statute whose reach is, on the whole, far more clear.

The Court does the opposite, and the ramifications of that decision are significant. First, of course, today’s holding invalidates a provision of the Immigration and Nationality Act—part of the definition of “aggravated felony”—on which the Government relies to “ensure that dangerous criminal aliens are removed from the United States.” Brief for United States 54. Contrary to the Court’s back-of-the-envelope assessment, see ante, at 23, n. 12, the Government explains that the definition is “critical” for “numerous” immigration provisions. Brief for United States 12.

In addition, §16 serves as the universal definition of “crime of violence” for all of Title 18 of the United States Code. Its language is incorporated into many procedural and substantive provisions of criminal law, including provisions concerning racketeering, money laundering, domestic violence, using a child to commit a violent crime, and distributing information about the making or use of explosives. See 18 U. S. C. §§25(a)(1), 842(p)(2), 1952(a), 1956(c)(7)(B)(ii), 1959(a)(4), 2261(a), 3561(b). Of special concern, §16 is replicated in the definition of “crime of violence” applicable to §924(c), which prohibits using or carrying a firearm “during and in relation to any crime of violence,” or possessing a firearm “in furtherance of any such crime.” §§924(c)(1)(A), (c)(3). Though I express no view on whether §924(c) can be distinguished from the provision we consider here, the Court’s holding calls into question convictions under what the Government warns us is an “oft-prosecuted offense.” Brief for United States 12.

Because Johnson does not compel today’s result, I respectfully dissent.

SUPREME COURT OF THE UNITED STATES

_________________

No. 15–1498

_________________

JEFFERSON B. SESSIONS, III, ATTORNEY GENERAL, PETITIONER v. JAMES GARCIA DIMAYA

on writ of certiorari to the united states court of appeals for the ninth circuit

[April 17, 2018]

Justice Thomas, with whom Justice Kennedy and Justice Alito join as to Parts I–C–2, II–A–1, and II–B, dissenting.

I agree with The Chief Justice that 18 U. S. C. §16(b), as incorporated by the Immigration and Nationality Act (INA), is not unconstitutionally vague. Section 16(b) lacks many of the features that caused this Court to invalidate the residual clause of the Armed Career Criminal Act (ACCA) in Johnson v. United States, 576 U. S. ___ (2015). ACCA’s residual clause—a provision that this Court had applied four times before Johnson—was not unconstitutionally vague either. See id., at ___ (Thomas, J., concurring in judgment) (slip op., at 1); id., at ___–___ (Alito, J., dissenting) (slip op., at 13–17). But if the Court insists on adhering to Johnson, it should at least take Johnson at its word that the residual clause was vague due to the “ ‘sum’ ” of its specific features. Id., at ___ (majority opinion) (slip op., at 10). By ignoring this limitation, the Court jettisons Johnson’s assurance that its holding would not jeopardize “dozens of federal and state criminal laws.” Id., at ___ (slip op., at 12).

While The Chief Justice persuasively explains why respondent cannot prevail under our precedents, I write separately to make two additional points. First, I continue to doubt that our practice of striking down statutes as unconstitutionally vague is consistent with the original meaning of the Due Process Clause. See id., at ___–___ (opinion of Thomas, J.) (slip op., at 7–18). Second, if the Court thinks that §16(b) is unconstitutionally vague because of the “categorical approach,” see ante, at 6–11, then the Court should abandon that approach—not insist on reading it into statutes and then strike them down. Accordingly, I respectfully dissent.

I

I continue to harbor doubts about whether the vagueness doctrine can be squared with the original meaning of the Due Process Clause—and those doubts are only amplified in the removal context. I am also skeptical that the vagueness doctrine can be justified as a way to prevent delegations of core legislative power in this context. But I need not resolve these questions because, if the vagueness doctrine has any basis in the Due Process Clause, it must be limited to cases in which the statute is unconstitutionally vague as applied to the person challenging it. That is not the case for respondent, whose prior convictions for first-degree residential burglary in California fall comfortably within the scope of §16(b).

A

The Fifth Amendment’s Due Process Clause provides that no person shall be “deprived of life, liberty, or prop- erty, without due process of law.” Section 16(b), as incorpo-rated by the INA, cannot violate this Clause unless the following propositions are true: The Due Process Clause requires federal statutes to provide certain minimal procedures, the vagueness doctrine is one of those procedures, and the vagueness doctrine applies to statutes governing the removal of aliens. Although I need not resolve any of these propositions today, each one is questionable. I will address them in turn.

1

First, the vagueness doctrine is not legitimate unless the “law of the land” view of due process is incorrect. Under that view, due process “require[s] only that our Government . . . proceed . . . according to written constitutional and statutory provision[s] before depriving someone of life, liberty, or property.” Nelson v. Colorado, 581 U. S. ___, ___, n. 1 (2017) (Thomas, J., dissenting) (slip op., at 2, n. 1) (internal quotation marks omitted). More than a half century after the founding, the Court rejected this view of due process in Murray’s Lessee v. Hoboken Land & Improvement Co., 18 How. 272 (1856). See id., at 276 (holding that the Due Process Clause “is a restraint on the legislative as well as on the executive and judicial powers of the government”). But the textual and historical support for the law-of-the-land view is not insubstantial.[1]

2

Even under Murray’s Lessee, the vagueness doctrine is legitimate only if it is a “settled usag[e] and mod[e] of proceeding existing in the common and statute law of England, before the emigration of our ancestors.” Id., at 277. That proposition is dubious. Until the end of the 19th century, “there is little indication that anyone . . . believed that courts had the power under the Due Process Claus[e] to nullify statutes on [vagueness] ground[s].” Johnson, supra, at ___ (opinion of Thomas, J.) (slip op., at

11). That is not because Americans were unfamiliar with vague laws. Rather, early American courts, like their English predecessors, addressed vague laws through statutory construction instead of constitutional law. See Note, Void for Vagueness: An Escape From Statutory Interpretation, 23 Ind. L. J. 272, 274–279 (1948). They invoked the rule of lenity and declined to apply vague penal statutes on a case-by-case basis. See Johnson, 576 U. S., at ___–___ (opinion of Thomas, J.) (slip op., at 7–10); e.g., ante, at 5–6, and n. 1 (Gorsuch, J., concurring in part and concurring in judgment) (collecting cases).[2] The modern vagueness doctrine, which claims the judicial author- ity to “strike down” vague legislation on its face, did not emerge until the turn of the 20th century. See Johnson, 576 U. S., at ___–___ (opinion of Thomas, J.) (slip op., at 11–13).

The difference between the traditional rule of lenity and

the modern vagueness doctrine is not merely semantic. Most obviously, lenity is a tool of statutory construction, which means States can abrogate it—and many have. Hall, Strict or Liberal Construction of Penal Statutes, 48 Harv. L. Rev. 748, 752–754 (1935); see also Scalia, Assorted Canards of Contemporary Legal Analysis, 40 Case W. Res. L. Rev. 581, 583 (1989) (“Arizona, by the way, seems to have preserved a fair and free society without adopting the rule that criminal statutes are to be strictly construed” (citing Ariz. Rev. Stat. §1-211C (1989))). The vagueness doctrine, by contrast, is a rule of constitutional law that States cannot alter or abolish. Lenity, moreover, applies only to “penal” statutes, 1 Blackstone, Commentaries on the Laws of England 88 (1765), but the vagueness doctrine extends to all regulations of individual conduct, both penal and nonpenal, Johnson, 576 U. S., at ___ (opinion of Thomas, J.) (slip op., at 6); see also Note, Indefinite Criteria of Definiteness in Statutes, 45 Harv. L. Rev. 160, 163 (1931) (explaining that the modern vagueness doctrine was not merely an “extension of the rule of strict construction of penal statutes” because it “expressly include[s] civil statutes within its scope,” reflecting a “regrettable disregard” for legislatures).[3] In short, early American courts were not applying the modern vagueness doctrine by another name. They were engaged in a fundamentally different enterprise.

Tellingly, the modern vagueness doctrine emerged at a time when this Court was actively interpreting the Due

Process Clause to strike down democratically enacted laws—first in the name of the “liberty of contract,” then in the name of the “right to privacy.” See Johnson, 576 U. S., at ___–___ (opinion of Thomas, J.) (slip op., at 13–16). That the vagueness doctrine “develop[ed] on the federal level concurrently with the growth of the tool of substantive due process” does not seem like a coincidence. Note, 23 Ind. L. J., at 278. Like substantive due process, the vagueness doctrine provides courts with “open-ended authority to oversee [legislative] choices.” Kolender v. Lawson, 461 U. S. 352, 374 (1983) (White, J., dissenting). This Court, for example, has used the vagueness doctrine to invalidate antiloitering laws, even though those laws predate the Declaration of Independence. See Johnson, supra, at ___ (opinion of Thomas, J.) (slip op., at 7) (discussing Chicago v. Morales, 527 U. S. 41 (1999) ).

This Court also has a bad habit of invoking the Due Process Clause to constitutionalize rules that were traditionally left to the democratic process. See, e.g., Williams v. Pennsylvania, 579 U. S. ___ (2016); BMW of North America, Inc. v. Gore, 517 U. S. 559 (1996) ; Foucha v. Louisiana, 504 U. S. 71 (1992) ; cf. Montgomery v. Louisiana, 577 U. S. ___ (2016). If vagueness is another example of this practice, then that is all the more reason to doubt its legitimacy.

3

Even assuming the Due Process Clause prohibits vague laws, this prohibition might not apply to laws governing the removal of aliens. Cf. Johnson, 576 U. S., at ___, n. 7 (opinion of Thomas, J.) (slip op., at 17, n. 7) (stressing the need for specificity when assessing alleged due process rights). The Founders were familiar with English law, where “ ‘the only question that ha[d] ever been made in regard to the power to expel aliens [was] whether it could be exercised by the King without the consent of Parliament.’ ” Demore v. Kim, 538 U. S. 510, 538 (2003) (O’Connor, J., concurring in part and concurring in judgment) (quoting Fong Yue Ting v. United States, 149 U. S. 698, 709 (1893) ). And, in this country, the notion that the Due Process Clause governed the removal of aliens was not announced until the 20th century.

Less than a decade after the ratification of the Bill of Rights, the founding generation had an extensive debate about the relationship between the Constitution and federal removal statutes. In 1798, the Fifth Congress enacted the Alien Acts. One of those Acts, the Alien Friends Act, gave the President unfettered discretion to expel any aliens “he shall judge dangerous to the peace and safety of the United States, or shall have reasonable grounds to suspect are concerned in any treasonable or secret machinations against the government thereof.” An Act Concerning Aliens §1, 1Stat. 571. This statute was modeled after the Aliens Act 1793 in England, which similarly gave the King unfettered discretion to expel aliens as he “shall think necessary for the publick Secur- ity.” 33 Geo. III, ch. 4, §18, in 39 Eng. Stat. at Large 16. Both the Fifth Congress and the States thoroughly de- bated the Alien Friends Act. Virginia and Kentucky enacted resolutions (anonymously drafted by Madison and Jefferson) opposing the Act, while 10 States enacted counter-resolutions condemning the views of Virginia and Kentucky. See Fehlings, Storm on the Constitution: The First Deportation Law, 10 Tulsa J. Comp. & Int’l L. 63, 85, 103 (2002).

The Jeffersonian Democratic-Republicans, who viewed the Alien Friends Act as a threat to their party and the institution of slavery,[4] raised a number of constitutional objections. Some of the Jeffersonians argued that the Alien Friends Act violated the Fifth Amendment’s Due Process Clause. They complained that the Act failed to provide aliens with all the accouterments of a criminal trial. See, e.g., Kentucky Resolutions ¶6, in 4 The Debates in the Several Conventions on the Adoption of the Federal Constitution 541–542 (J. Elliot ed. 1836) (Elliot’s Debates); 8 Annals of Cong. 1982–1983 (1798) (statement of Rep. Gallatin); Madison’s Report on the Virginia Resolutions (Jan. 7, 1800), in 6 Writings of James Madison 361–362 (G. Hunt ed. 1906) (Madison’s Report).[5]

The Federalists gave two primary responses to this due process argument. First, the Federalists argued that the rights of aliens were governed by the law of nations, not the Constitution. See, e.g., Randolph, Debate on Virginia Resolutions, in The Virginia Report of 1799–1800, pp. 34–35 (1850) (Virginia Debates) (statement of George K. Taylor) (arguing that aliens “were not a party to the [Constitution]” and that “cases between the government and

aliens . . . arise under the law of nations”); id., at 100 (statement of William Cowan) (identifying the source of rights “as to citizens, the Constitution; as to aliens, the law of nations”); A. Addison, A Charge to the Grand Juries of the County Courts of the Fifth Circuit of the State of Pennsylvania 18 (1799) (Charge to the Grand Juries) (“[T]he Constitution leaves aliens, as in other countries, to the protection of the general principles of the law of nations”); Answer to the Resolutions of the State of Kentucky, Oct. 29, 1799, in 4 Records of the Governor and Council of the State of Vermont 528 (1876) (denying “that aliens had any rights among us, except what they derived from the law of nations, and rights of hospitality”). The law of nations imposed no enforceable limits on a nation’s power to remove aliens. See, e.g., 1 E. de Vattel, Law of Nations, §§230–231, pp. 108–109 (J. Chitty et al. transl. and ed. 1883).

Second, the Federalists responded that the expulsion of aliens “did not touch life, liberty, or property.” Virginia Debates 34. The founding generation understood the phrase “life, liberty, or property” to refer to a relatively narrow set of core private rights that did not depend on the will of the government. See Wellness Int’l Network, Ltd. v. Sharif, 575 U. S. ___, ___–___ (2015) (Thomas, J., dissenting) (slip op., at 9–10); Nelson, Adjudication in the Political Branches, 107 Colum. L. Rev. 559, 566–568 (2007) (Nelson). Quasi-private rights—“privileges” or “franchises” bestowed by the government on individuals—did not qualify and could be taken away without judicial process. See B&B Hardware, Inc. v. Hargis Industries, Inc., 575 U. S. ___, ___ (2015) (Thomas, J., dissenting) (slip op., at 12); Nelson 567–569. The Federalists argued that an alien’s right to reside in this country was one such privilege. See, e.g., Virginia Debates 34 (arguing that “ordering away an alien . . . was not a matter of right, but of favour,” which did not require a jury trial); Report of the Select Committee of the House of Representatives, Made to the House of Representatives on Feb. 21, 1799, 9 Annals of Cong. 2987 (1799) (stating that aliens “remain in the country . . . merely as matter of favor and permission” and can be removed at any time without a criminal trial); Charge to the Grand Juries 11–13 (similar). According to the Minority Address of the Virginia Legislature (anonymously drafted by John Marshall), “[T]he right of remaining in our country is vested in no alien; he enters and remains by the courtesy of the sovereign power, and that courtesy may at pleasure be withdrawn” without judicial process. Address of the Minority in the Virginia Legislature to the People of that State 9–10 (1799) (Virginia Minority Address). Unlike “a grant of land,” the “[a]dmission of an alien to residence . . . is revocable, like a permission.” A. Addison, Analysis of the Report of the Committee of the Virginia Assembly 23 (1800). Removing a resident alien from the country did not affect “life, lib- erty, or property,” the Federalists argued, until the alien became a naturalized citizen. See id., at 23–24; Charge to the Grand Juries 11–13. That the alien’s permanent residence was conferred by statute would not have made a difference. See Nelson 571, 580–582; Teva Pharmaceuticals USA, Inc. v. Sandoz, Inc., 574 U. S. ___, ___, n. 2 (2015) (Thomas, J., dissenting) (slip op., at 9, n. 2).

After the Alien Friends Act lapsed in 1800, Congress did not enact another removal statute for nearly a century. The States enacted their own removal statutes during this period, see G. Neuman, Strangers to the Constitution 19–43 (1996), and I am aware of no decision questioning the legality of these statutes under State due-process or law-of-the-land provisions. Beginning in the late 19th century, the Federal Government reinserted itself into the regulation of immigration. When this Court was presented with constitutional challenges to Congress’ removal laws, it initially rejected them for many of the same reasons that Marshall and the Federalists had cited in defense of the Alien Friends Act. Although the Court rejected the Federalists’ argument that resident aliens do not enjoy constitutional rights, see Wong Wing v. United States, 163 U. S. 228, 238 (1896) , it agreed that civil deportation statutes do not implicate “life, liberty, or property,” see, e.g., Harisi- ades v. Shaughnessy, 342 U. S. 580, 584 –585 (1952) (“[T]hat admission for permanent residence confers a ‘vested right’ on the alien [is] not founded in precedents of this Court”); United States ex rel. Turner v. Williams, 194 U. S. 279, 290 (1904) (“[T]he deportation of an alien who is found to be here in violation of law is not a deprivation of liberty without due process of law”); Fong Yue Ting, 149 U. S., at 730 (“[Deportation] is but a method of enforcing the return to his own country of an alien who has not complied with [statutory] conditions . . . . He has not, therefore, been deprived of life, liberty, or property without due process of law”); id., at 713–715 (similar). Consistent with this understanding, “federal immigration laws from 1891 until 1952 made no express provision for judicial review.” Demore, 538 U. S., at 538 (opinion of O’Connor, J.).

It was not until the 20th century that this Court held that nonpenal removal statutes could violate the Due Process Clause. See Wong Yang Sung v. McGrath, 339 U. S. 33, 49 (1950) . That ruling opened the door for the Court to apply the then-nascent vagueness doctrine to immigration statutes. But the Court upheld vague standards in immigration laws that it likely would not have tolerated in criminal statutes. See, e.g., Boutilier v. INS, 387 U. S. 118, 122 (1967) (“ ‘psychopathic personality’ ”); Jordan v. De George, 341 U. S. 223, 232 (1951) (“ ‘crime involving moral turpitude’ ”); cf. Mahler, supra, at 40 (“ ‘undesirable residents’ ”). Until today, this Court has never held that an immigration statute is unconstitutionally vague.

Thus, for more than a century after the founding, it was, at best, unclear whether federal removal statutes could violate the Due Process Clause. And until today, this Court had never deemed a federal removal statute void for vagueness. Given this history, it is difficult to conclude that a ban on vague removal statutes is a “settled usag[e] and mod[e] of proceeding existing in the common and statute law of England, before the emigration of our ancestors” protected by the Fifth Amendment’s Due Process Clause. Murray’s Lessee, 18 How., at 277.

B

Instead of a longstanding procedure under Murray’s Lessee, perhaps the vagueness doctrine is really a way to enforce the separation of powers—specifically, the doctrine of nondelegation. See Chapman & McConnell, Due Process as Separation of Powers, 121 Yale L. J. 1672, 1806 (2012) (“Vague statutes have the effect of delegating lawmaking authority to the executive”). Madison raised a similar objection to the Alien Friends Act, arguing that its expansive language effectively allowed the President to exercise legislative (and judicial) power. See Madison’s Report 369–371. And this Court’s precedents have occasionally described the vagueness doctrine in terms of nondelegation. See, e.g., Grayned v. City of Rockford, 408 U. S. 104, 108 –109 (1972) (“A vague law impermissibly delegates basic policy matters”). But they have not been consistent on this front. See, e.g., Aptheker v. Secretary of State, 378 U. S. 500, 516 (1964) (“ ‘The objectionable qual- ity of vagueness . . . does not depend upon . . . unchanneled delegation of legislative powers’ ”); Maynard v. Cartwright, 486 U. S. 356, 361 (1988) (“Objections to vagueness under the Due Process Clause rest on the lack of notice”).

I agree that the Constitution prohibits Congress from delegating core legislative power to another branch. See Department of Transportation v. Association of American Railroads, 575 U. S. ___, ___ (2015) (AAR) (Thomas, J., concurring in judgment) (slip op., at 3) (“Congress improperly ‘delegates’ legislative power when it authorizes an entity other than itself to make a determination that requires an exercise of legislative power”); accord, Whitman v. American Trucking Assns., Inc., 531 U. S. 457, 487 (2001) (Thomas, J., concurring). But I locate that principle in the Vesting Clauses of Articles I, II, and III—not in the Due Process Clause. AAR, supra, at ___–___ (opinion of Thomas, J.) (slip op., at 2–3); see also Hampton v. Mow Sun Wong, 426 U. S. 88, 123 (1976) (Rehnquist, J., dissenting) (“[T]hat there was an improper delegation of authority . . . has not previously been thought to depend upon the procedural requirements of the Due Process Clause”). In my view, impermissible delegations of legislative power violate this principle, not just delegations that deprive individuals of “life, liberty, or property,” Amdt. 5.

Respondent does not argue that §16(b), as incorporated by the INA, is an impermissible delegation of power. See Brief for Respondent 50 (stating that “there is no delegation question” in this case). I would not reach that question here, because this case can be resolved on narrower grounds. See Part I–C, infra. But at first blush, it is not at all obvious that the nondelegation doctrine would jus- tify wholesale invalidation of §16(b).

If §16(b) delegates power in this context, it delegates power primarily to the Executive Branch entities that administer the INA—namely, the Attorney General, immigration judges, and the Board of Immigration Appeals (BIA). But Congress does not “delegate” when it merely authorizes the Executive Branch to exercise a power that it already has. See AAR, supra, at ___ (opinion of Thomas, J.) (slip op., at 3). And there is some founding-era evidence that “the executive Power,” Art. II, §1, includes the power to deport aliens.

Blackstone—one of the political philosophers whose writings on executive power were “most familiar to the Framers,” Prakash & Ramsey, The Executive Power Over Foreign Affairs, 111 Yale L. J. 231, 253 (2001)—described the power to deport aliens as executive and located it with the King. Alien friends, Blackstone explained, are “liable to be sent home whenever the king sees occasion.” 1 Commentaries on the Laws of England 252 (1765). When our Constitution was ratified, moreover, “[e]minent English judges, sitting in the Judicial Committee of the Privy Council, ha[d] gone very far in supporting the . . . expulsion, by the executive authority of a colony, of aliens.” Demore, 538 U. S., at 538 (opinion of O’Connor, J.) (quoting Fong Yue Ting, 149 U. S., at 709). Some of the Federalists defending the Alien Friends Act similarly argued that the President had the power to remove aliens. See, e.g., Virginia Debates 35 (statement of George K. Taylor) (arguing that the power to remove aliens is “most properly entrusted” with the President, since “[h]e, by the Constitution, was bound to execute the laws” and is “the executive officer, with whom all persons and bodies whatever were accustomed to communicate”); Virginia Minority Address 9 (arguing that the removal of aliens “is a measure of general safety, in its nature political and not forensic, the execution of which is properly trusted to the department which represents the nation in all its interior relations”); Charge to the Grand Juries 29–30 (“As a measure of national defence, this discretion, of expulsion or indulgence, seems properly vested in the branch of the government peculiarly charged with the direction of the executive powers, and of our foreign relations. There is in it a mixture of external policy, and of the law of nations, that justifies this disposition”). More recently, this Court recognized that “[r]emoval decisions” implicate “our customary policy of deference to the President in matters of foreign affairs” because they touch on “our relations with foreign powers and require consideration of changing political and economic circumstances.” Jama v. Immigration and Customs Enforcement, 543 U. S. 335, 348 (2005) (internal quotation marks omitted). Taken together, this evidence makes it difficult to confidently conclude that the INA, through §16(b), delegates core legislative power to the Executive.

Instead of the Executive, perhaps §16(b) impermissibly delegates power to the Judiciary, since the Courts of Appeals often review the BIA’s application of §16(b). I assume that, at some point, a statute could be so devoid of content that a court tasked with interpreting it “would simply be making up a law—that is, exercising legislative power.” Lawson, Delegation and Original Meaning, 88 Va. L. Rev. 327, 339 (2002); see id., at 339–340 (providing examples such as a gibberish-filled statute or a statute that requires “ ‘goodness and niceness’ ”). But I am not confident that our modern vagueness doctrine—which focuses on whether regulations of individual conduct provide “fair warning,” are “clearly defined,” and do not encourage “arbitrary and discriminatory enforcement,” Grayned, 408 U. S., at 108; Kolender, 461 U. S., at 357—accurately demarcates the line between legislative and judicial power. The Founders understood that the interpretation of legal texts, even vague ones, remained an exercise of core judicial power. See Perez v. Mortgage Bankers Assn., 575 U. S. ___, ___–___ (2015) (Thomas, J., concurring in judgment) (slip op., at 8–9); Hamburger, The Constitution’s Accommodation of Social Change, 88 Mich. L. Rev. 239, 303–310 (1989). Courts were expected to clarify the meaning of such texts over time as they applied their terms to specific cases. See id., at 309–310; Nelson, Originalism and Interpretive Conventions, 70 U. Chi. L. Rev. 519, 526 (2003). Although early American courts declined to apply vague or unintelligible statutes as appropriate in individual cases, they did not wholesale invalidate them as unconstitutional delegations of legislative power. See Johnson, 576 U. S., at ___–___, and n. 3 (opinion of Thomas, J.) (slip op., at 10–11, and n. 3).

C

1

I need not resolve these historical questions today, as this case can be decided on narrower grounds. If the vagueness doctrine has any basis in the original meaning of the Due Process Clause, it must be limited to case-by-case challenges to particular applications of a statute. That is what early American courts did when they applied the rule of lenity. See id., at ___ (slip op., at 10). And that is how early American courts addressed constitutional challenges to statutes more generally. See ibid. (“[T]here is good evidence that [antebellum] courts . . . understood judicial review to consist ‘of a refusal to give a statute effect as operative law in resolving a case,’ a notion quite distinct from our modern practice of ‘ “strik[ing] down” legislation’ ” (quoting Walsh, Partial Unconstitutionality, 85 N. Y. U. L. Rev. 738, 756 (2010)).

2

This Court’s precedents likewise recognize that, outside the First Amendment context, a challenger must prove that the statute is vague as applied to him. See Holder v. Humanitarian Law Project, 561 U. S. 1, 18 –19 (2010); United States v. Williams, 553 U. S. 285, 304 (2008) ; Maynard, 486 U. S., at 361; Hoffman Estates v. Flipside, Hoffman Estates, Inc., 455 U. S. 489, 495 , and n. 7 (1982) (collecting cases). Johnson did not overrule these precedents. While Johnson weakened the principle that a facial challenge requires a statute to be vague “in all applications,” 576 U. S., at ___ (slip op., at 11) (emphasis added), it did not address whether a statute must be vague as applied to the person challenging it. That question did not arise because the Court concluded that ACCA’s residual clause was vague as applied to the crime at issue there: unlawful possession of a short-barreled shotgun. See id., at ___ (slip op., at 9).

In my view, §16(b) is not vague as applied to respondent. When respondent committed his burglaries in 2007 and 2009, he was “sufficiently forewarned . . . that the statutory consequence . . . is deportation.” De George, 341 U. S., at 232. At the time, courts had “unanimous[ly]” concluded that residential burglary is a crime of violence, and not “a single opinion . . . ha[d] held that [it] is not.” United States v. M. C. E., 232 F. 3d 1252, 1255–1256 (CA9 2000); see also United States v. Davis, 881 F. 2d 973, 976 (CA11 1989) (explaining that treating residential burglary as a crime of violence was “[i]n accord with common law tradition and the settled law of the federal circuits”). Residential burglary “ha[d] been considered a violent offense for hundreds of years . . . because of the potential for mayhem if burglar encounters resident.” United States v. Pinto, 875 F. 2d 143, 144 (CA7 1989). The Model Penal Code had recognized that risk, see ALI, Model Penal Code §221.1, Comment 3(c), p. 75 (1980); the Sentencing Commission had recognized that risk; see United States Sentencing Commission, Guidelines Manual §4B1.2(a)(2) (Nov. 2006); and this Court had repeatedly recognized that risk, see, e.g., James v. United States, 550 U. S. 192, 203 (2007) ; Taylor v. United States, 495 U. S. 575, 588 (1990) . In Leocal v. Ashcroft, 543 U. S. 1 (2004) , this Court unanimously agreed that burglary is the “classic example” of a crime of violence under §16(b), because it “involves a substantial risk that the burglar will use force against a victim in completing the crime.” Id., at 10.

That same risk is present with respect to respondent’s statute of conviction—first-degree residential burglary, Cal. Penal Code Ann. §§459, 460(a) (West 1999). The California Supreme Court has explained that the State’s burglary laws recognize “the dangers to personal safety created by the usual burglary situation.” People v. Davis, 18 Cal. 4th 712, 721, 958 P. 2d 1083, 1089 (1998) (emphasis added). “ ‘[T]he fact that a building is used as a home . . . increases such danger,’ ” which is why California elevates residential burglary to a first-degree offense. People v. Rodriguez, 122 Cal. App. 4th 121, 133, 18 Cal. Rptr. 3d 550, 558 (2004); see also People v. Wilson, 208 Cal. App. 3d 611, 615, 256 Cal. Rptr. 422, 425 (1989) (“[T]he higher degree . . . is intended to prevent those situations which are most dangerous, most likely to cause personal injury” (emphasis deleted)). Although unlawful entry is not an element of the offense, courts “unanimous[ly]” agree that the offense still involves a substantial risk of physical force. United States v. Avila, 770 F. 3d 1100, 1106 (CA4 2014); accord, United States v. Maldonado, 696 F. 3d 1095, 1102, 1104 (CA10 2012); United States v. Scanlan, 667 F. 3d 896, 900 (CA7 2012); United States v. Echeverria–Gomez, 627 F. 3d 971, 976 (CA5 2010); United States v. Becker, 919 F. 2d 568, 573 (CA9 1990). First-degree residential burglary requires entry into an inhabited dwelling, with the intent to commit a felony, against the will of the homeowner—the key elements that create the risk of violence. See United States v. Park, 649 F. 3d 1175, 1178–1180 (CA9 2011); Avila, supra, at 1106–1107; Becker, supra, at 571, n. 5. As this Court has explained, “[t]he main risk of burglary arises not from the simple physical act of wrongfully entering onto another’s property, but rather from the possibility of a face-to-face confrontation between the burglar and a third party.” James, supra, at 203.

Drawing on Johnson and the decision below, the Court suggests that residential burglary might not be a crime of violence because “ ‘only about seven percent of burglaries actually involve violence.’ ” Ante, at 9, n. 3 (citing Dimaya v. Lynch, 803 F. 3d 1110, 1116, n. 7 (CA9 2015)); see Bureau of Justice Statistics, S. Catalano, National Crime Victimization Survey: Victimization During Household Burglary 1 (Sept. 2010), https://www.bjs.gov/content/pub/ pdf/vdhb.pdf (as last visited Apr. 13, 2018). But this statistic—which measures actual violence against a member of the household, see id., at 1, 12—is woefully underinclusive. It excludes other potential victims besides household members—for example, “a police officer, or a bystande[r] who comes to investigate,” James, supra, at 203. And §16(b) requires only a risk of physical force, not actual physical force, and that risk would seem to be present whenever someone is home during the burglary. Further, Johnson is not conclusive because, unlike ACCA’s residual clause, §16(b) covers offenses that involve a substantial risk of physical force “against the person or property of another.” (Emphasis added.) Surely the ordinary case of residential burglary involves at least one of these risks. According to the statistics referenced by the Court, most burglaries involve either a forcible entry (e.g., breaking a window or slashing a door screen), an attempted forcible entry, or an unlawful entry when someone is home. See Bureau of Justice Statistics, supra, at 2 (Table 1). Thus, under any metric, respondent’s convictions for first-degree residential burglary are crimes of violence under §16(b).

3

Finally, if facial vagueness challenges are ever appropriate, I adhere to my view that a law is not facially vague “ ‘[i]f any fool would know that a particular category of conduct would be within the reach of the statute, if there is an unmistakable core that a reasonable person would know is forbidden by the law.’ ” Morales, 527 U. S., at 112 (Thomas, J., dissenting) (quoting Kolender, 461 U. S., at 370–371 (White, J., dissenting)). The residual clause of ACCA had such a core. See Johnson, 576 U. S., at ___ (slip op., at 10); id., at ___–___ (Alito, J., dissenting) (slip op., at 14–15). And §16(b) has an even wider core, as The Chief Justice explains. Thus, the Court should not have invalidated §16(b), either on its face or as applied to respondent.

II

Even taking the vagueness doctrine and Johnson at face value, I disagree with the Court’s decision to invalidate §16(b). The sole reason that the Court deems §16(b) unconstitutionally vague is because it reads the statute as incorporating the categorical approach—specifically, the “ordinary case” approach from ACCA’s residual clause. Although the Court mentions “[t]wo features” of §16(b) that make it vague—the ordinary-case approach and an imprecise risk standard—the Court admits that the second feature is problematic only in combination with the first. Ante, at 8. Without the ordinary-case approach, the Court “do[es] not doubt” the constitutionality of §16(b). Ante, at 10.

But if the categorical approach renders §16(b) unconstitutionally vague, then constitutional avoidance requires us to make a reasonable effort to avoid that interpretation. And a reasonable alternative interpretation is available: Instead of asking whether the ordinary case of an alien’s offense presents a substantial risk of physical force, courts should ask whether the alien’s actual underlying conduct presents a substantial risk of physical force. I will briefly discuss the origins of the categorical approach and then explain why the Court should abandon it for §16(b).

A

1

The categorical approach originated with Justice Blackmun’s opinion for the Court in Taylor v. United States, 495 U. S. 575 (1990) . The question in Taylor was whether ACCA’s reference to “burglary” meant burglary as defined by state law or burglary in the generic sense. After “devoting 10 pages of [its] opinion to legislative history,” id., at 603 (Scalia, J., concurring in part and concurring in judgment), and finding that Congress had made “an inadvertent casualty in [the] complex drafting process,” id., at 589–590 (majority opinion), the Court concluded that ACCA referred to burglary in the generic sense, id., at 598. The Court then addressed how the Government would prove that a defendant was convicted of generic burglary, as opposed to another offense. Id., at 599–602. Taylor rejected the notion that the Government could introduce evidence about the “particular facts” of the defendant’s underlying crime. Id., at 600. Instead, the Court adopted a “categorical approach,” which focused primarily on the “statutory definition of the prior offense.” Id., at 602.

Although Taylor was interpreting one of ACCA’s enumerated offenses, this Court later extended the categorical approach to ACCA’s residual clause. See James, 550 U. S., at 208. That extension required some reworking. Because ACCA’s enumerated-offenses clause asks whether a prior conviction “is burglary, arson, or extortion,” 18 U. S. C. §924(e)(2)(B)(ii), Taylor instructed courts to focus on the definition of the underlying crime. The residual clause, by contrast, asks whether a prior conviction “involves conduct that presents a serious potential risk of physical injury to another.” §924(e)(2)(B)(ii). Thus, the Court held that the categorical approach for the residual clause asks “whether the conduct encompassed by the elements of the offense, in the ordinary case, presents a serious potential risk of injury to another.” James, supra, at 208 (emphasis added). This “ordinary case” approach allowed courts to apply the residual clause without inquiring into the individual facts of the defendant’s prior crime.

Taylor gave a few reasons why the categorical approach was the correct reading of ACCA, see 495 U. S., at 600–601, but the “heart of the decision” was the Court’s concern with limiting the amount of evidence that the parties could introduce at sentencing. Shepard v. United States, 544 U. S. 13, 23 (2005) . Specifically, the Court was worried about potential violations of the Sixth Amendment. If the parties could introduce evidence about the defendant’s underlying conduct, then sentencing proceedings might devolve into a full-blown minitrial, with factfinding by the judge instead of the jury. See id., at 24–26; Taylor, supra, at 601. While this Court’s decision in Almendarez-Torres v. United States, 523 U. S. 224 (1998) , allows judges to find facts about a defendant’s prior convictions, a full-blown minitrial would look “too much like” the kind of factfinding that the Sixth Amendment requires the jury to conduct. Shepard, 544 U. S., at 25. By construing ACCA to require a categorical approach, then, the Court was following “[t]he rule of reading statutes to avoid serious risks of unconstitutionality.” Ibid.

2

I disagreed with the Court’s decision to extend the categorical approach to ACCA’s residual clause. See James, 550 U. S., at 231–232 (dissenting opinion). The categorical approach was an “ ‘unnecessary exercise,’ ” I explained, because it created the same Sixth Amendment problem that it tried to avoid. Id., at 231. Absent waiver, a defendant has the right to have a jury find “every fact that is by law a basis for imposing or increasing punishment,” including the fact of a prior conviction. Apprendi v. New Jersey, 530 U. S. 466, 501 (2000) (Thomas, J., concurring). The exception recognized in Almendarez-Torres for prior convictions is an aberration, has been seriously undermined by subsequent precedents, and should be reconsidered. See Mathis v. United States, 579 U. S. ___, ___ (2016) (Thomas, J., concurring) (slip op., at 1); Shepard, supra, at 27–28 (Thomas, J., concurring in part and concurring in judgment). In my view, if the Government wants to enhance a defendant’s sentence based on his prior convictions, it must put those convictions in the indictment and prove them to a jury beyond a reasonable doubt.[6]

B

My objection aside, the ordinary-case approach soon created problems of its own. The Court’s attempt to avoid the Scylla of the Sixth Amendment steered it straight into the Charybdis of the Fifth. The ordinary-case approach that was created to honor the individual right to a jury is now, according to the Court, so vague that it deprives individuals of due process.

I see no good reason for the Court to persist in reading the ordinary-case approach into §16(b). The text of §16(b) does not mandate the ordinary-case approach, the concerns that led this Court to adopt it do not apply here, and there are no prudential reasons for retaining it. In my view, we should abandon the categorical approach for §16(b).

1

The text of §16(b) does not require a categorical approach. The INA declares an alien deportable if he is

“convicted of an aggravated felony” after he is admitted to the United States. 8 U. S. C. §1227(a)(2)(A)(iii). Aggravated felonies include “crime[s] of violence” as defined in §16. §1101(a)(43)(F). Section 16, in turn, defines crimes of violence as follows:

“(a) an offense that has as an element the use, attempted use, or threatened use of physical force against the person or property of another, or

“(b) any other offense that is a felony and that, by its nature, involves a substantial risk that physical force against the person or property of another may be used in the course of committing the offense.”

At first glance, §16(b) is not clear about the precise question it poses. On the one hand, the statute might refer to the metaphysical “nature” of the offense and ask whether it ordinarily involves a substantial risk of physical force. On the other hand, the statute might refer to the underlying facts of the offense that the offender committed; the words “by its nature,” “substantial risk,” and “may” would mean only that an offender who engages in risky conduct cannot benefit from the fortuitous fact that physical force was not actually used during his offense. The text can bear either interpretation. See Nijhawan v. Holder, 557 U. S. 29, 33 –34 (2009) (“[I]n ordinary speech words such as ‘crime,’ ‘felony,’ ‘offense,’ and the like sometimes refer to a generic crime . . . and sometimes refer to the specific acts in which an offender engaged on a specific occasion”). It is entirely natural to use words like “nature” and “offense” to refer to an offender’s actual underlying conduct.[7]

Although both interpretations are linguistically possible, several factors indicate that the underlying-conduct approach is the better one. To begin, §16(b) asks whether an offense “involves” a substantial risk of force. The word “involves” suggests that the offense must necessarily include a substantial risk of force. See The New Oxford Dictionary of English 962 (2001) (“include (something) as a necessary part or result”); Random House Dictionary of the English Language 1005 (2d ed. 1987) (“1. to include as

a necessary circumstance, condition, or consequence”); Oxford American Dictionary 349 (1980) (“1. to contain within itself, to make necessary as a condition or result”). That condition is always satisfied if the Government must prove that the alien’s underlying conduct involves a substantial risk of force, but it is not always satisfied if the Government need only prove that the “ordinary case” involves such a risk. See Johnson, 576 U. S., at ___ (Alito, J., dissenting) (slip op., at 12). Tellingly, the other aggravated felonies in the INA that use the word “involves” employ the underlying-conduct approach. See 8 U. S. C. §1101(a)(43)(M)(i) (“an offense that involves fraud or deceit in which the loss to the victim or victims exceeds $10,000”); §1101(h)(3) (“any crime of reckless driving or of driving while intoxicated or under the influence of alcohol or of prohibited substances if such crime involves personal injury to another”). As do the similarly worded provisions of the Comprehensive Crime Control Act of 1984, the bill that contained §16(b). See, e.g., 98Stat. 2059 (elevating the burden of proof for the release of “a person found not guilty only by reason of insanity of an offense involving bodily injury to, or serious damage to the property of, another person, or involving a substantial risk of such injury or damage”); id., at 2068 (establishing the sentence for drug offenses “involving” specific quantities and types of drugs); id., at 2137 (defining violent crimes in aid of racketeering to include “attempting or conspiring to commit a crime involving maiming, assault with a dangerous weapon, or assault resulting in serious bodily injury”).

A comparison of §16(b) and §16(a) further highlights why the former likely adopts an underlying-conduct approach. Section 16(a) covers offenses that have the use, attempted use, or threatened use of physical force “as an element.” Because §16(b) covers “other” offenses and is separated from §16(a) by the disjunctive word “or,” the natural inference is that §16(b) asks a different question. In other words, §16(b) must require immigration judges to look beyond the elements of an offense to determine whether it involves a substantial risk of physical force. But if the elements are insufficient, where else should immigration judges look to determine the riskiness of an offense? Two options are possible, only one of which is workable.

The first option is to consult the underlying facts of the alien’s crime and then assess its riskiness. This approach would provide a definitive answer in every case. And courts are already familiar with this kind of inquiry. Cf. Johnson, supra, at ___ (slip op., at 12) (noting that “dozens” of similarly worded laws ask courts to assess “the riskiness of conduct in which an individual defendant engages on a particular occasion”). Nothing suggests that Congress imposed a more limited inquiry when it enacted §16(b) in 1984. At the time, Congress had not yet enacted ACCA’s residual clause, this Court had not yet created the categorical approach, and this Court had not yet recognized a Sixth Amendment limit on judicial factfinding at sentencing, see Chambers v. United States, 555 U. S. 122, 132 (2009) (Alito, J., concurring in judgment).

The second option is to imagine the “ordinary case” of the alien’s crime and then assess the riskiness of that hypothetical offense. But the phrase “ordinary case” does not appear in the statute. And imagining the ordinary case, the Court reminds us, is “hopeless[ly] indetermina[te],” “wholly ‘speculative,’ ” and mere “guesswork.” Ante, at 7, 24 (quoting Johnson, supra, at ___–___ (slip op., at 5, 7)); see also Chambers, supra, at 133 (opinion of Alito, J.) (observing that the categorical approach is “nearly impossible to apply consistently”). Because courts disfavor interpretations that make a statute impossible to apply, see A. Scalia & B. Garner, Reading Law 63 (2012), this Court should reject the ordinary-case approach for §16(b) and adopt the underlying-facts approach instead. See Johnson, supra, at ___ (Alito, J., dissenting) (slip op., at 10) (“When another interpretation is ready at hand, why should we assume that Congress gave the clause a meaning that is impossible—or even, exceedingly difficult—to apply”).

2

That the categorical approach is not the better reading of §16(b) should not be surprising, since the categorical approach was never really about the best reading of the text. As explained, this Court adopted that approach to avoid a potential Sixth Amendment problem with sentencing judges conducting minitrials to determine a defendant’s past conduct. But even assuming the categorical approach solved this Sixth Amendment problem in criminal cases, no such problem arises in immigration cases. “[T]he provisions of the Constitution securing the right of trial by jury have no application” in a removal proceeding. Turner, 194 U. S., at 290. And, in criminal cases, the underlying-conduct approach would be perfectly constitutional if the Government included the defendant’s prior conduct in the indictment, tried it to a jury, and proved it beyond a reasonable doubt. See Johnson, 576 U. S., at ___ (Alito, J., dissenting) (slip op., at 12). Nothing in §16(b) prohibits the Government from proceeding this way, so the plurality is wrong to suggest that the underlying-conduct approach would necessarily “ping-pong us from one constitutional issue to another.” Ante, at 14.

If constitutional avoidance applies here at all, it requires us to reject the categorical approach for §16(b). According to the Court, the categorical approach is unconstitutionally vague. And, all agree that the underlying-conduct approach would not be. See Johnson, 576 U. S., at ___ (majority opinion) (slip op., at 12) (“[W]e do not doubt the constitutionality of laws that call for the application of a qualitative standard such as ‘substantial risk’ to real-world conduct”). Thus, if the underlying-conduct approach is a “reasonabl[e]” interpretation of §16(b), it is our “plain duty” to adopt it. United States ex rel. Attorney General v. Delaware & Hudson Co., 213 U. S. 366, 407 (1909) . And it is reasonable, as explained above.

In Johnson, the Court declined to adopt the underlying-conduct approach for ACCA’s residual clause. See 576 U. S., at ___–___ (slip op., at 12–13). The Court concluded that the categorical approach was the only reasonable reading of ACCA because the residual clause uses the word “convictions.” Id., at ___ (slip op., at 13). The Court also stressed the “utter impracticability of requiring a sentencing court to reconstruct, long after the original conviction, the conduct underlying that conviction.” Ibid.

Neither of these arguments is persuasive with respect to the INA. Moreover, this Court has already rejected them. In Nijhawan, this Court unanimously concluded that one of the aggravated felonies in the INA—“an offense that . . . involves fraud or deceit in which the loss to the victim or victims exceeds $10,000,” §1101(a)(43)(M)(i)—applies the underlying-conduct approach, not the categorical approach. 557 U. S., at 32. Although the INA also refers to “convict[ions],” §1227(a)(2)(A)(iii), the Court was not swayed by that argument. The word “convict[ion]” means only that the defendant’s underlying conduct must “ ‘be tied to the specific counts covered by the conviction,’ ” not “acquitted or dismissed counts or general conduct.” Id., at 42. As for the supposed practical problems with proving an alien’s prior conduct, the Court did not find that argument persuasive either. “[T]he ‘sole purpose’ of the ‘aggravated felony’ inquiry,” the Court explained, “ ‘is to ascertain the nature of a prior conviction; it is not an invitation to relitigate the conviction itself.’ ” Ibid. And because the INA places the burden on the Government to prove an alien’s conduct by clear and convincing evidence, §1229a(c)(3)(A), “uncertainties caused by the passage of time are likely to count in the alien’s favor,” id., at 42.

There are additional reasons why the practical problems identified in Johnson should not matter for §16(b)—even assuming they should have mattered for ACCA’s residual clause, see Lewis v. Chicago, 560 U. S. 205, 217 (2010) (“[I]t is not our task to assess the consequences of each approach and adopt the one that produces the least mischief. Our charge is to give effect to the law Congress enacted”). In a removal proceeding, any difficulties with identifying an alien’s past conduct will fall on immigration judges, not federal courts. But those judges are already accustomed to finding facts about the conduct underlying an alien’s prior convictions, since some of the INA’s aggravated felonies employ the underlying-conduct approach. The BIA has instructed immigration judges to determine such conduct based on “any evidence admissible in re- moval proceedings,” not just the elements of the offense or the record of conviction. See Matter of Babaisakov, 24 I. & N. Dec. 306, 307 (2007). No one has submitted any evidence that the BIA’s approach has been “utter[ly] impracticab[le]” or “daunting[ly] difficul[t]” in practice. Ante, at 15. And even if it were, “how much time the agency wants to devote to the resolution of particular issues is . . . a question for the agency itself.” Ali v. Mukasey, 521 F. 3d 737, 741 (CA7 2008). Hypothetical burdens on the BIA should not influence how this Court interprets §16(b).

In short, we should not blithely assume that the reasons why this Court adopted the categorical approach for ACCA’s residual clause also apply to the INA’s list of aggravated felonies. As Nijhawan explained, “the ‘aggravated felony’ statute, unlike ACCA, contains some language that refers to generic crimes and some language that almost certainly refers to the specific circumstances in which a crime was committed.” 557 U. S., at 38. “The question” in each case is “to which category [the aggra- vated felony] belongs.” Ibid. As I have explained, §16(b) belongs in the underlying-conduct category. Because that is the better reading of §16(b)’s text—or at least a reasonable reading—the Court should have adopted it here.

3

I see no prudential reason for maintaining the categorical approach for §16(b). The Court notes that the Government “explicitly acknowledges” that §16(b) employs the categorical approach. Ante, at 9. But we cannot permit the Government’s concessions to dictate how we interpret a statute, much less cause us to invalidate a statute enacted by a coordinate branch. See United States Nat. Bank of Ore. v. Independent Ins. Agents of America, Inc., 508 U. S. 439, 446 –447 (1993); Young v. United States, 315 U. S. 257, 258 –259 (1942). This Court’s “traditional practice” is to “refus[e] to decide constitutional questions” when other grounds of decision are available, “whether or not they have been properly raised before us by the parties.” Neese v. Southern R. Co., 350 U. S. 77, 78 (1955) (per curiam); see also Vermeule, Saving Constructions, 85 Geo. L. J. 1945, 1948–1949 (1997) (explaining that courts commonly “decide an antecedent statutory issue, even one waived by the parties, if its resolution could preclude a constitutional claim”). This Court has raised potential saving constructions “on our own motion” when they could avoid a ruling on constitutional vagueness grounds, even in cases where the Government was a party. United States v. L. Cohen Grocery Co., 255 U. S. 81, 88 (1921) . We should have followed that established practice here.

Nor should stare decisis prevent us from rejecting the categorical approach for §16(b). This Court has never held that §16(b) incorporates the ordinary-case approach. Although Leocal held that §16(b) incorporates a version of the categorical approach, the Court must not feel bound by that decision, as it largely overrules it today. See ante, at 22, n. 7. Surely the Court cannot credibly invoke stare decisis to defend the categorical approach—the same approach it says only a “lunatic” would continue to apply. Ante, at 24. If the Court views the categorical approach that way—the same way Johnson viewed it—then it must also agree that “[s]tanding by [the categorical approach] would undermine, rather than promote, the goals that stare decisis is meant to serve.” 576 U. S., at ___ (slip op., at 15). That is especially true if the Court’s decision leads to the invalidation of scores of similarly worded state and federal statutes, which seems even more likely after today than it did after Johnson. Instead of adhering to an interpretation that it thinks unconstitutional and then using that interpretation to strike down another statute, the Court should have taken this opportunity to abandon the categorical approach for §16(b) once and for all.

*  *  *

The Court’s decision today is triply flawed. It unnecessarily extends our incorrect decision in Johnson. It uses a constitutional doctrine with dubious origins to invalidate yet another statute (while calling into question countless more). And it does all this in the name of a statutory interpretation that we should have discarded long ago. Because I cannot follow the Court down any of these rabbit holes, I respectfully dissent.