dissenting.
As the Court’s opinion acknowledges, this case is “another in a series,” ante, at 4, 180 L. Ed. 2d, at 67. More specifically, it is an attempt to clarify, for the fourth time since 2007, what distinguishes “violent felonies” under the residual clause of the Armed Career Criminal Act (ACCA), 18 U.S.C. § 924(e)(2)(B)(ii), from other crimes. See James v. United States, 550 U.S. 192, 127 S. Ct. 1586, 167 L. Ed. 2d 532 (2007); Begay v. United States, 553 U.S. 137, 128 S. Ct. 1581, 170 L. Ed. 2d 490 (2008); Chambers v. United States, 555 U.S. 122, 129 S. Ct. 687, 172 L. Ed. 2d 484 (2009). We try to include an ACCA residual-clause case in about every second or third volume of the United States Reports.
As was perhaps predictable, instead of producing a clarification of the Delphic residual clause, today’s opinion produces a fourth ad hoc judgment that will sow further confusion. Insanity, it has been said, is doing the same thing over and over again, but expecting different results. Four times is enough. We should admit that ACCA’s residual provision is a drafting failure and declare it void for vagueness. See Kolender v. Lawson, 461 U.S. 352, 357, 103 S. Ct. 1855, 75 L. Ed. 2d 903 (1983).
I
ACCA defines “violent felony,” in relevant part, as “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). Many years of prison hinge on whether a crime falls within this definition. A felon convicted of possessing a firearm who has three prior violent-*83felony convictions faces a 15-year mandatory minimum sentence and the possibility of life imprisonment. See § 924(e)(1); see United States v. Harrison, 558 F.3d 1280, 1282, n. 1 (CA11 2009). Without those prior convictions, he
[564 U.S. 29]
would face a much lesser sentence, which could not possibly exceed 10 years. See § 924(a)(2).
Vehicular flight is a violent felony only if it falls within ACCA’s residual clause; that is, if it “involves conduct that presents a serious potential risk of physical injury to another.” § 924(e)(2)(B)(ii). Today’s opinion says, or initially seems to say, that an offense qualifies as a violent felony if its elements, in the typical case, create a degree of risk “ ‘comparable to that posed by its closest analog among the enumerated offenses.'" Ante, at 8, 180 L. Ed. 2d, at 70. That is a quotation from the Court’s opinion in the first of our residual-clause trilogy, James, 550 U.S., at 203, 127 S. Ct. 1586, 167 L. Ed. 2d 532. I did not join that opinion because I thought it should suffice if the elements created a degree of risk comparable to the least risky of the enumerated offenses, whether or not it was the closest analog. See id., at 230, 127 S. Ct. 1586, 167 L. Ed. 2d 532 (Scalia, J., dissenting). The problem with applying the James standard to the present case is that the elements of vehicular flight under Indiana law are not analogous to any of the four enumerated offenses. See Ind. Code § 35-44-3-3 (2004). Nor is it apparent which of the enumerated offenses most closely resembles, for example, statutory rape, see United States v. Daye, 571 F.3d 225, 228-236 (CA2 2009); possession of a sawed-off shotgun, see United States v. Upton, 512 F.3d 394, 403-405 (CA7 2008); or a failure to report to prison, see Chambers, supra. I predicted this inadequacy of the “closest analog” test in my James dissent. See 550 U.S., at 215, 127 S. Ct. 1586, 167 L. Ed. 2d 532.
But as it turns out, the Court’s inability to identify an analog makes no difference to the outcome of the present case. For today’s opinion introduces the James standard with the words “[f]or instance,” ante, at 8, 180 L. Ed. 2d, at 70. It is (according to the Court) merely one example of how the enumerated crimes (burglary, arson, extortion, and crimes using explosives) “provide guidance.” Ibid. And the opinion then proceeds to obtain guidance from the risky-as-the-least-risky test that I suggested (but the Court rejected) in James—finding vehicular
[564 U.S. 30]
flight at least as risky as both arson and burglary. See ante, at 6-9, 180 L. Ed. 2d, at 68-71.
But what about the test that determined the outcome in our second case in this “series”—the “purposeful, violent, and aggressive” test of Begay? Fear not. That incompatible variation has been neither overlooked nor renounced in today’s tutti-frutti opinion. “In many cases,” we are told, it “will be redundant with the inquiry into risk.” Ante, at 13, 180 L. Ed. 2d, at 73. That seems to be the case here—though why, and when it will not be the case, are not entirely clear. The Court’s accusation that Sykes “overreads the opinions of this Court,” ante, at 12, 180 L. Ed. 2d, at 72, apparently applies to his interpretation of Begay’s “purposeful, violent, and aggressive” test, which the Court now suggests applies only “to strict-liability, negligence, and recklessness crimes,” ante, at 13, 180 L. Ed. 2d, at 73. But that makes no sense. If the test excluded only those unintentional crimes, it would be recast as the “purposeful” test, since the last *84two adjectives (“violent, and aggressive”) would do no work. For that reason, perhaps, all 11 Circuits that have addressed Begay “overrea[d]” it just as Sykes does*—and as does the Government, see Brief for United States 8.
The only case that is not brought forward in today’s opinion to represent yet another test is the third and most recent in the trilogy, Chambers, 555 U.S. 122, 129 S. Ct. 687, 172 L. Ed. 2d 484—which applied both the risky-as-the-least-risky test and the “purposeful, violent, and aggressive” test to reach the conclusion that failure to
[564 U.S. 31]
report for periodic incarceration was not a crime of violence under ACCA. But today’s opinion does cite Chambers for another point: Whereas James rejected the risky-as-the-least-risky approach because, among other reasons, no “hard statistics” on riskiness “have been called to our attention,” 550 U.S., at 210, 127 S. Ct. 1586, 167 L. Ed. 2d 532; and whereas Begay made no mention of statistics; Chambers explained (as today’s opinion points out) that “statistical evidence sometimes ‘helps provide a conclusive . . . answer’ concerning the risks that crimes present,” ante, at 10, 180 L. Ed. 2d, at 71 (quoting Chambers, supra, at 129, 129 S. Ct. 687, 172 L. Ed. 2d 484). Today’s opinion then outdoes Chambers in the volume of statistics that it spews forth—statistics compiled by the International Association of Chiefs of Police concerning injuries attributable to police pursuits, ante, at 11, 180 L. Ed. 2d, at 71; statistics from the Department of Justice concerning injuries attributable to burglaries, ibid.; statistics from the U. S. Fire Administration concerning injuries attributable to fires, ibid., and (by reference to Justice Thomas’s concurrence) statistics from the National Center for Statistics & Analysis, the Pennsylvania State Police Bureau of Research, the FBI Law Enforcement Bulletin and several articles published elsewhere concerning injuries attributable to police pursuits, ante, at 10, 180 L. Ed. 2d, at 71 (citing ante, at 19-20, 180 L. Ed. 2d, at 77-78 (Thomas, J., concurring in judgment)).
Supreme Court briefs are an inappropriate place to develop the key facts in a case. We normally give parties more robust protection, leaving important factual questions to district courts and juries aided by expert witnesses and the procedural protections of discovery. See Fed. Rules Crim. Proc. 16(a)(1)(F), (G); Fed. Rules Evid. 702-703, 705. An adversarial process in the trial courts can identify flaws in the methodology of the studies that the parties put forward; here, we accept the studies’ findings on faith, without examining their methodology at all. The Court does not examine, for example, whether the police-pursuit data on which it relies is a representative sample of all vehicular flights. The data may be skewed toward the rare and riskier forms
[564 U.S. 32]
of flight. See post, at 40, n. 4, 180 L. Ed. 2d, at 90 (Kagan, J., dissenting). We also have *85no way of knowing how many injuries reported in that data would have occurred even absent pursuit, by a driver who was driving recklessly even before the police gave chase. Similar questions undermine confidence in the burglary and arson data the Court cites. For example, the Court relies on a U. S. Fire Administration dataset to conclude that 3.3 injuries occur per 100 arsons. See ante, at 11, 180 L. Ed. 2d, at 72. But a 2001 report from the same U. S. Fire Administration suggests that roughly one injury occurs per 100 arsons. See Arson in the United States, Vol. 1 Topical Fire Research Series, No. 8, pp. 1-2 (rev. Dec. 2001), online at http: //www. usfa. dhs. gov/downloads/ pdf/tfrs/vli8-508.pdf (as visited May 27, 2011, and available in Clerk of Court’s case file). The Court does not reveal why it chose one dataset over another. In sum, our statistical analysis in ACCA cases is untested judicial factfinding masquerading as statutory interpretation. Most of the statistics on which the Court relies today come from Government-funded studies, and did not make an appearance in this litigation until the Government’s merits brief to this Court. See Brief for Petitioner 17; see also Chambers, supra, at 128-129, 129 S. Ct. 687, 172 L. Ed. 2d 484 (demonstrating that the same was true in that case).
But the more fundamental problem with the Court’s use of statistics is that, far from eliminating the vagueness of the residual clause, it increases the vagueness. Vagueness, of course, must be measured ex ante— before the Court gives definitive meaning to a statutory provision, not after. Nothing is vague once the Court decrees precisely what it means. And is it seriously to be expected that the average citizen would be familiar with the sundry statistical studies showing (if they are to be believed) that this-or-that crime is more likely to lead to physical injury than what sundry statistical studies (if they are to be believed) show to be the case for burglary, arson, extortion, or use of explosives? To ask the
[564 U.S. 33]
question is to answer it. A few words, then, about unconstitutional vagueness.
II
When I dissented from the Court’s judgment in James, I said that the residual clause’s “shoddy draftsmanship” put courts to a difficult choice:
“They can (1) apply the ACCA enhancement to virtually all predicate offenses, . . . ; (2) apply it case by case in its pristine abstraction, finding it applicable whenever the particular sentencing judge (or the particular reviewing panel) believes there is a ‘serious potential risk of physical injury to another’ (whatever that means); (3) try to figure out a coherent way of interpreting the statute so that it applies in a relatively predictable and ad-ministrable fashion to a smaller subset of crimes; or (4) recognize the statute for the drafting failure it is and hold it void for vagueness....” 550 U.S., at 229-230, 127 S. Ct. 1586, 167 L. Ed. 2d 532.
My dissent “tried to implement,” id., at 230, 127 S. Ct. 1586, 167 L. Ed. 2d 532, the third option; and the Court, I believed, had chosen the second. “Today’s opinion,” I wrote, “permits an unintelligible criminal statute to survive uncorrected, unguided, and unexplained.” Id., at 230-231, 127 S. Ct. 1586, 167 L. Ed. 2d 532.
My assessment has not been changed by the Court’s later decisions in the ACCA “series.” Today’s opinion, *86which adds to the “closest analog” test (James) the “purposeful, violent, and aggressive” test (Begay), and even the risky-as-the-least-risky test that I had proposed as the exclusive criterion, has not made the statute’s application clear and predictable. And all of them together—or even the risky-as-the-least-risky test alone, I am now convinced—never will. The residual-clause series will be endless, and we will be doing ad hoc application of ACCA to the vast variety of state criminal offenses until the cows come home.
[564 U.S. 34]
That does not violate the Constitution. What does violate the Constitution is approving the enforcement of a sentencing statute that does not “give a person of ordinary intelligence fair notice” of its reach, United States v. Batchelder, 442 U.S. 114, 123, 99 S. Ct. 2198, 60 L. Ed. 2d 755 (1979) (internal quotation marks omitted), and that permits, indeed invites, arbitrary enforcement, see Kolender, 461 U.S., at 357, 103 S. Ct. 1855, 75 L. Ed. 2d 903. The Court’s ever-evolving interpretation of the residual clause will keep defendants and judges guessing for years to come. The reality is that the phrase “otherwise involves conduct that presents a serious potential risk of physical injury to another” does not clearly define the crimes that will subject defendants to the greatly increased ACCA penalties. It is not the job of this Court to impose a clarity which the text itself does not honestly contain. And even if that were our job, the further reality is that we have by now demonstrated our inability to accomplish the task.
We have, I recognize, upheld hopelessly vague criminal statutes in the past—indeed, in the recent past. See, e.g., Skilling v. United States, 561 U.S. 358, 130 S. Ct. 2896, 177 L. Ed. 2d 619 (2010). That is regrettable, see id., at 415, 130 S. Ct. 2896, 177 L. Ed. 2d 619 (Scalia, J., concurring in part and concurring in judgment). What sets ACCA apart from those statutes—and what confirms its incurable vagueness—is our repeated inability to craft a principled test out of the statutory text. We have demonstrated by our opinions that the clause is too vague to yield “an intelligible principle,” ante, at 15, 180 L. Ed. 2d, at 74, each attempt to ignore that reality producing a new regime that is less predictable and more arbitrary than the last. ACCA’s residual clause fails to speak with the clarity that criminal proscriptions require. See United States v. L. Cohen Grocery Co., 255 U.S. 81, 89-90, 41 S. Ct. 298, 65 L. Ed. 516 (1921).
The Court believes that the residual clause cannot be unconstitutionally vague because other criminal prohibitions also refer to the degree of risk posed by a defendant’s conduct. See ante, at 15-16, 180 L. Ed. 2d, at 74-75. Even apart from the fact that our
[564 U.S. 35]
opinions dealing with those statutes have not displayed the confusion evident in our four ACCA efforts, this is not the first time I have found the comparison unpersuasive:
“None of the provisions the Court cites ... is similar in the crucial relevant respect: None prefaces its judicially-to-be-determined requirement of risk of physical injury with the word ‘otherwise,’ preceded by four confusing examples that have little in common with respect to the supposedly defining characteristic. The 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 *87red’ assuredly does so.” James, supra, at 230, n. 7, 127 S. Ct. 1586, 167 L. Ed. 2d 532.
Of course even if the cited statutes were comparable, repetition of constitutional error does not produce constitutional truth.
We face a Congress that puts forth an ever-increasing volume of laws in general, and of criminal laws in particular. It should be no surprise that as the volume increases, so do the number of imprecise laws. And no surprise that our indulgence of impre-cisions that violate the Constitution encourages imprecisions that violate the Constitution. Fuzzy, leave-the-details-to-be-sorted-out-by-the-courts legislation is attractive to the Congressman who wants credit for addressing a national problem but does not have the time (or perhaps the votes) to grapple with the nitty-gritty. In the field of criminal law, at least, it is time to call a halt. I do not think it would be a radical step—indeed, I think it would be highly responsible—to limit ACCA to the named violent crimes. Congress can quickly add what it wishes. Because the majority prefers to let vagueness reign, I respectfully dissent.
[564 U.S. 36]
See United States v. Holloway, 630 F.3d 252, 260 (CA1 2011); United States v. Brown, 629 F.3d 290, 295-296 (CA2 2011) (per curiam); United States v. Lee, 612 F.3d 170, 196 (CA3 2010); United States v. Jenkins, 631 F.3d 680, 683 (CA4 2011); United States v. Harrimon, 568 F.3d 531, 534 (CA5 2009); United States v. Young, 580 F.3d 373, 377 (CA6 2009); United States v. Sonnenberg, 628 F.3d 361, 364 (CA7 2010); United States v. Boyce, 633 P.3d 708, 711 (CA8 2011); United States v. Terrell, 593 F.3d 1084, 1089-1091 (CA9 2010); United States v. Ford, 613 F.3d 1263, 1272-1273 (CA10 2010); United States v. Harrison, 558 F.3d 1280, 1295-1296 (CA11 2009).