*960Under California's second degree felony-murder law, petitioner Gregory White's guilt depended upon an abstract legal issue that had nothing to do with his actions. The trial court had to adjudicate whether California's offense of manufacturing methamphetamine-in general, not in White's case in particular-was "inherently *691dangerous to human life" such that it qualified as a felony murder predicate. If so, the fact that White's co-conspirator died from burns incurred during the manufacture meant White was guilty of not just the drug crime, but of murdering his accomplice.
Had our Legislature listed methamphetamine manufacture among the crimes that can serve as a predicate for first degree felony murder, White would be guilty of murder with no claim that the crime of conviction was unconstitutionally vague. The Legislature would have provided notice to the public and adequately guided the courts. But there is no statutory list of predicate crimes for second degree felony murder. A defendant such as White may find out whether his crime qualifies after he committed it, when a court determines whether the crime, taken in the abstract, fits the amorphous inherent-dangerousness-to-life standard.
When it decided Johnson v. United States (2015) --- U.S. ----, 135 S.Ct. 2551, 192 L.Ed.2d 569, the United States Supreme Court held that a law is unconstitutionally vague if it "require[s] courts to assess the hypothetical risk posed by an abstract generic version of the offense." ( Welch v. United States (2016) --- U.S. ----, 136 S.Ct. 1257, 1264, 194 L.Ed.2d 387.) This was a "new rule" of constitutional law. ( Id. at p. 1264.)
Following Johnson , our Supreme Court set an order to show cause in our court for the Attorney General to show why White "is not entitled to reversal of his second degree felony murder conviction because the reasoning set forth in Johnson ... renders the California second-degree [felony] murder rule unconstitutionally vague." Our Supreme Court cited Professor Evan Tsen Lee's law review article, Why California's Second-Degree Felony-Murder Rule Is Now Void for Vagueness (2015) 43 Hastings Const. L.Q. 1, UC Hastings Research Paper No. 158. Thereafter, in a different case, a three-judge panel of the United States Court of Appeals for the Ninth Circuit issued a reasoned decision explaining that, for purposes of allowing a defendant to file a second federal habeas corpus petition, the defendant had advanced "a plausible position" that California's second degree felony-murder statute was unconstitutional under Johnson . ( Henry v. Spearman (9th Cir. 2018) 899 F.3d 703, 711.)
I conclude that under Johnson , California's second degree felony-murder law is unconstitutionally vague because it requires courts to assess the hypothetical risk posed by an abstract generic version of the offense.
*961I.
JOHNSON 'S TWO-PRONGED VAGUENESS TEST
The Due Process Clause of the Fifth Amendment prohibits the government from taking away life, liberty, or property based on a criminal law that is "so vague that it fails to give ordinary people fair notice of the conduct it punishes, or so standardless that it invites arbitrary enforcement." ( Johnson , supra , 135 S.Ct. at p. 2556, citing Kolender v. Lawson (1983) 461 U.S. 352, 357-358, 103 S.Ct. 1855, 75 L.Ed.2d 903.) Because all statutory language leaves some need for interpretation, what is "fair" notice and "so" standardless is a matter of degree. In Johnson , the Court identified a type of criminal law-no doubt rare-that was sufficiently unfair and standardless that the Constitution cannot permit it.
Johnson addressed one clause in the definition of "violent felony" found in the federal Armed Career Criminal Act (ACCA), 18 U.S.C. § 924(e)(2)(B). Known *692as the "residual clause," that clause used the following italicized language to identify particular prior convictions that would significantly increase a defendant's punishment: those for 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 ." ( Johnson , supra , 135 S.Ct. at pp. 2555-2556.)
The Court held that "[t]wo features of the residual clause conspire to make it unconstitutionally vague." ( Johnson , supra , 135 S.Ct. at p. 2557.) The first feature was that evaluation of the prior crime depended upon an abstract inquiry about the nature of the crime, a framework known as the "categorical approach." ( Ibid. ) That approach tied the assessment of a crime's riskiness to "a judicially imagined 'ordinary case' of a crime," not to how a particular offender committed it. ( Ibid . ) In determining whether a crime carried a sufficient risk of danger to qualify under the residual clause, a court was to imagine the "ordinary case" of the crime and then evaluate its risk. ( Ibid. ) Johnson 's problem with the ordinary case construct was that it required an "abstract inquiry" rather than one based on the defendant's actual facts: "The residual clause ... 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 both in nature and degree of effect,' this abstract inquiry offers significantly less predictability than one '[t]hat deals with the actual, not with an imaginary condition other than the facts.' " ( Id. at p. 2561 ; see also id. at p. 2563 [prior Supreme Court opinions rejected vagueness challenges to the residual clause based on the imprecision of the *962phrase " 'serious potential risk' " without reaching "the uncertainty introduced by the need to evaluate the riskiness of an abstract ordinary case of a crime"].)
The second feature contributing to the vagueness of the residual clause was the "uncertainty about how much risk" was needed to qualify a crime as a violent felony under the unspecific statutory standard of " 'serious potential risk.' " ( Johnson , supra , 135 S.Ct. at p. 2558.) That standard was not vague on its own, but it was problematic when measuring the abstraction of an "ordinary case": "It is one thing to apply an imprecise 'serious potential risk' standard to real-world facts; it is quite another to apply it to a judge-imagined abstraction." ( Ibid. ; see also id. at p. 2561 ["we 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, Johnson 's vagueness test does not affect statutes that occasionally generate a difficult legal issue about whether a defendant's crime falls within their ambit. Johnson likewise does not affect statutes that apply an indefinite standard such as "dangerousness" to a defendant's actual conduct. (See Sessions v. Dimaya (2018) --- U.S. ----, 138 S.Ct. 1204, 1214, 200 L.Ed.2d 549 ( Dimaya ) ["Many perfectly constitutional statutes use imprecise terms like 'serious potential risk' ... or 'substantial risk.' "].) In all, Johnson made clear that its holding that the residual clause was unconstitutionally vague resulted from both features it discussed. That is, a criminal statute is constitutionally void for vagueness if it employs a combination of (1) an abstract judicial construct that is (2) evaluated by an uncertain standard. As the Court put it, "[b]y 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 produces more unpredictability and arbitrariness *693than the Due Process Clause tolerates." ( Johnson , supra , 135 S.Ct. at p. 2558.)1 *963II.
JOHNSON 'S TWO PRONGS APPLIED TO SECOND DEGREE FELONY MURDER
Under Johnson , then, a statute fails to provide ordinary people fair notice of what is criminal when it requires courts to apply an indefinite standard to an abstract construction of a statute that is not tied to their own conduct. This holding condemns few laws, but, in my view, one of them is California second degree felony murder.2
First degree felony murder is defined as a killing committed during the perpetration or attempt to perpetrate any of several specified felonies. ( Pen. Code, § 189.) There is nothing unconstitutionally vague about that statute. There is no statutory list of specified felonies, however, for second degree felony murder. In our state, that crime has been judicially defined as " 'an unlawful killing in the course of the commission of a felony that is inherently dangerous to human life but is not included among the felonies enumerated in section 189.' " ( People v. Chun (2009) 45 Cal.4th 1172, 1182, 91 Cal.Rptr.3d 106, 203 P.3d 425 ( Chun ) [quoting People v. Robertson (2004) 34 Cal. 4th 156, 164, 17 Cal.Rptr.3d 604, 95 P.3d 872 ].) The idea is that when a defendant perpetrates a " 'felony inherently dangerous to human life,' " the requisite malice for murder is imputed to him when death results. ( Chun , supra , at p. 1184, 91 Cal.Rptr.3d 106, 203 P.3d 425.) As applied by our courts, second degree felony murder carries the same two-pronged flaw as the residual clause at issue in Johnson .
A. Abstract Determination
As with the "residual clause" found unconstitutionally vague in Johnson , courts determine whether a felony is "inherently dangerous to human life" categorically for the felony, independent of the facts of any particular offense. " 'In determining whether a felony is inherently dangerous [under the second degree felony-murder rule], the court looks to the elements of the felony in the abstract , "not the 'particular' facts of the case," i.e., not to the defendant's specific conduct.' [Citation.] That is, we determine whether the felony 'by its very nature ... cannot be committed without creating a substantial risk that someone will be killed ....' [Citations.]" ( People v. Howard (2005) 34 Cal.4th 1129, 1135, 23 Cal.Rptr.3d 306, 104 P.3d 107 ( Howard ).)
*694*964For this law, the "very nature" of the felony that a court is to consider does not turn on whether the ordinary case of the felony is dangerous to human life, but rather on whether the felony "possibly could be committed without creating such peril." ( People v. Burroughs (1984) 35 Cal.3d 824, 830, 833, 201 Cal.Rptr. 319, 678 P.2d 894 ( Burroughs ); see People v. Satchell (1971) 6 Cal.3d 28, 40, 98 Cal.Rptr. 33, 489 P.2d 1361 ( Satchell ).)
Thus, the courts are charged with abstractly constructing and then evaluating the possible means of committing the offense to determine whether any such means is not sufficiently dangerous. This approach was applied in Howard , supra , 34 Cal.4th 1129, 23 Cal.Rptr.3d 306, 104 P.3d 107, the most recent case where our Supreme Court adjudicated whether a particular crime was inherently dangerous for purposes of second degree felony murder. Howard addressed whether "driving with a willful or wanton disregard for the safety of persons or property while fleeing from a pursuing police officer," in violation of Vehicle Code section 2800.2, was an inherently dangerous felony. ( Howard , supra , at p. 1132, 23 Cal.Rptr.3d 306, 104 P.3d 107.) That felony certainly seems dangerous ordinarily : on the facts of Howard , the defendant fled pursuing officers in a sport utility vehicle at speeds of up to 90 miles per hour, running stop signs and a red light, and driving on the wrong side of a road, until he collided with another vehicle and killed the driver. ( Ibid . ) Indeed, the felony even appears dangerous by its terms, as it expressly requires as an element the "willful or wanton disregard for the safety of persons or property." ( Ibid. )
Our Supreme Court noted, however, that under the statute, the " 'willful or wanton disregard' " element can be met by "any flight from an officer during which the motorist commits three traffic violations that are assigned a 'point count.' " ( Howard , supra , 34 Cal.4th at p. 1137, 23 Cal.Rptr.3d 306, 104 P.3d 107.) Such traffic violations include, among others, ones that are relatively nonhazardous: driving with a suspended license, driving slightly more than 55 miles per hour when there is no higher posted speed limit, and making a right turn without signaling for 100 feet before turning. ( Id. at pp. 1137-1138, 23 Cal.Rptr.3d 306, 104 P.3d 107.) The possibility of such a nonhazardous violation was enough for the Supreme Court to conclude "a violation of [Vehicle Code] section 2800.2 is not, in the abstract, inherently dangerous to human life." ( Id. at pp. 1138-1139, 23 Cal.Rptr.3d 306, 104 P.3d 107.) Thus, because the Supreme Court was able to identify a way that the offense possibly could be committed without a high probability of death, the crime could not serve as a predicate for second degree felony murder.
The Court reasoned similarly in considering the crime of practicing medicine without a license under conditions creating a risk of great bodily harm, serious physical or mental illness, or death. ( Burroughs , supra , 35 Cal.3d at p. 833, 201 Cal.Rptr. 319, 678 P.2d 894.) The Court held that this felony was not inherently dangerous to human life, principally because the crime could be committed, for example, *965with the victim suffering great bodily injury by a broken bone, in a manner that is not life-threatening. ( Id. at p. 831, 201 Cal.Rptr. 319, 678 P.2d 894 ; see also People v. Henderson (1977) 19 Cal.3d 86, 94-95, 137 Cal.Rptr. 1, 560 P.2d 1180 [felony false imprisonment dangerous when accomplished by "violence" or "menace" but not inherently dangerous to life because it also may be accomplished by "fraud or deceit"]; People v. Caffero (1989) 207 Cal.App.3d 678, 683, 255 Cal.Rptr. 22 [felony child abuse not inherently dangerous to life because "[i]f the statute may be violated *695by conduct which does not endanger human life, it is not inherently dangerous to human life."].)
Our Supreme Court's cases not only indicate that the hypothetical "possibility" of committing the offense without danger to life need not be the "ordinary case," but the cases have not required that the possibility be common at all. For instance, the Court has reasoned that possession of a sawed-off shotgun is not inherently dangerous because the possessor may have that firearm as an heirloom, curio, or keepsake rather than as a weapon. (See Satchell , supra , 6 Cal.3d at p. 42, 98 Cal.Rptr. 33, 489 P.2d 1361 [statute "makes no distinction between the innocent 'collector' and the hardened criminal"].) That logic did not depend on how common it is for a sawed-off shotgun to be kept as an heirloom, only that it is possible to do so. (Cf. District of Columbia v. Heller (2008) 554 U.S. 570, 625, 128 S.Ct. 2783, 171 L.Ed.2d 637 [short-barreled shotguns are "not typically possessed by law-abiding citizens for lawful purposes"].)
Our second degree felony-murder law differs from the residual clause in Johnson because our analysis requires judges to cogitate on various possible means of committing the offense rather than on the ordinary case of the offense. In my view, this distinction is not meaningful to constitutional vagueness, as each inquiry requires "a judge-imagined abstraction" ( Johnson , supra , 135 S.Ct. at p. 2561 ) that is untethered to the conduct of the particular defendant. This is the domain of legal interpretation by lawyers and judges, not of fair notice to ordinary people. (See, e.g., Howard , 34 Cal.4th at p. 1141, 23 Cal.Rptr.3d 306, 104 P.3d 107 (conc. & dis. opn. of Brown, J.) [disagreeing with majority on ground that "if any offense should easily qualify as inherently dangerous, Vehicle Code section 2800.2 certainly would"], pp. 1141-1148, 23 Cal.Rptr.3d 306, 104 P.3d 107 (dis. opn. of Baxter, J.) [same].)
B. Indefinite Standard
For a felony to be inherently dangerous to human life, our law requires that the abstract version of the offense involved " 'a high probability that death will result.' " ( Chun , supra , 45 Cal.4th at p. 1207, 91 Cal.Rptr.3d 106, 203 P.3d 425 [quoting People v. Hansen (1994) 9 Cal.4th 300, 309, 36 Cal.Rptr.2d 609, 885 P.2d 1022, overruled on another ground in Chun , supra , at p. 1199, 91 Cal.Rptr.3d 106, 203 P.3d 425 ];
*966People v. Patterson (1989) 49 Cal.3d 615, 627, 262 Cal.Rptr. 195, 778 P.2d 549 ( Patterson ) [establishing " 'high probability' " standard because a "less stringent standard would inappropriately expand the scope of the second degree felony-murder rule"].)
This "high probability" test is no less ill-defined than the "serious potential risk" standard found in the residual clause in Johnson . The problem with such a "fuzzy risk standard" ( Dimaya , supra , 138 S.Ct. at p. 1221 ) is that it allows for a wide range of arguable results in evaluating the danger of any particular abstract of a felony, giving inadequate notice of which crimes are to be punished and creating unpredictable determinations by our courts. In Patterson , for instance, our Supreme Court established the "high probability" test but did not decide whether the offense at issue, furnishing cocaine, met that standard, sending the case to the trial court to evaluate evidence on the matter. ( Patterson , supra , 49 Cal.3d at pp. 618, 625, 262 Cal.Rptr. 195, 778 P.2d 549.) If the answer to that question was not knowable by the Supreme Court, it was not something that the ordinary person could be expected to know. A remand for an evidentiary hearing about the facts of a defendant's crime is standard appellate *696fare. More unusual, if not unique to California second degree felony murder, is a remand for a court to take evidence, wholly divorced from the defendant's case, to determine whether an offense is criminal.
III.
METHAMPHETAMINE MANUFACTURE AS AN ILLUSTRATION OF INDETERMINACY
Petitioner White's offense, manufacturing methamphetamine, provides a good illustration of the indeterminacy of our state's second degree felony-murder inquiry. The majority concludes that "manufacturing methamphetamine in an unprofessional laboratory is inherently dangerous to human life." (Maj. opn., ante , at pp. 684-85.) I do not necessarily disagree with this statement, which makes the application of second degree felony murder to methamphetamine manufacture seem to be an easy case. But I am not convinced that the majority is answering the right question.
Our Supreme Court's "possible means" approach to second degree felony murder requires us to consider not the means by which defendant White's cohort actually committed the offense-cooking the substance in a cramped unprofessional laboratory-but whether the offense, taken in the abstract, possibly could be committed in a way that is not inherently dangerous to human life. I see two ways that this felony illustrates Johnson 's point about the malleability of an abstract felony inquiry.
*967First, it is not clear why we are to assume the case of an unprofessional laboratory. The offense of manufacturing methamphetamine could be accomplished in a well-ventilated, professional laboratory.3 It could be executed by a trained chemist.4 In fact, our record indicates that the manufacture would be dramatically safer in such circumstances. The lead law enforcement agent in this case testified that, as part of his training, he had manufactured methamphetamine in controlled settings on five occasions, each time in a certified lab. Presumably, he was not risking his life in doing so. (See People v. James (1998) 62 Cal.App.4th 244, 264, 74 Cal.Rptr.2d 7 [recounting expert testimony that experienced cooks can manufacture methamphetamine relatively safely].) Manufacture by trained chemists in a laboratory, to be sure, is not the ordinary case of methamphetamine manufacturing in California. But we are not charged with evaluating the ordinary case, but whether the offense is, by its nature , dangerous to human life. If Howard held driving with willful or wanton disregard for safety to be not inherently dangerous due to the possibility that a driver might be driving in that manner without threatening lives, on what basis do we disregard the possibility of a relatively safe manufacture here?
Along these lines, and perhaps more realistic for most criminals than manufacture in a laboratory, methamphetamine could be manufactured in an isolated location such as a field or empty lot-putting no nonparticipants at risk-with those involved in manufacture wearing protective, fire-resistant gear that would drastically reduce the risk of death, even if a fire occurred. I cannot see a basis for us to *697ignore such a possibility in considering the inherent dangerous of the offense, unless, contrary to cases such as Howard , Burroughs , and Satchell , we are to consider the ordinary case rather than whether the offense possibly can be committed without a high probability of death. Is the only reason not to consider the "empty lot" scenario that no one has provided testimony that it is possible? Does the law change if, in the next case, the defendant is not someone like Gregory White but instead like Walter White?5 The uncertainty of this type of hypothetical inquiry is precisely what animated Johnson .
The second way that this crime illustrates Johnson 's concerns is that the crime of manufacturing methamphetamine "encompasses the initial and *968intermediate steps carried out to process a controlled substance." ( People v. Coria (1999) 21 Cal.4th 868, 874, 89 Cal.Rptr.2d 650, 985 P.2d 970 ; People v. Hard (2003) 112 Cal.App.4th 272, 279, 5 Cal.Rptr.3d 107 ["The scope of the prohibition in [Health & Saf. Code] section 11379.6 is broad because the production of methamphetamine is an incremental, not instantaneous process, often conducted in a piecemeal fashion to avoid detection. [Citation.] Because the police can interrupt production at any stage, the language of [Health & Saf. Code] section 11379.6 should be construed broadly to ensure that a manufacturer will not receive any benefit from police intervention early in the process."].) That means that the statute is violated even by conduct that is not the part of the manufacturing process that might involve an explosion. A defendant could, for example, violate the statute by bringing a large amount of a precursor chemical (such as pseudoephedrine) to a place where he expected methamphetamine to be manufactured, only to have the manufacturing plan interrupted by an undercover officer. (Compare, e.g., People v. Lancellotti (1993) 19 Cal.App.4th 809, 811, 23 Cal.Rptr.2d 640 [conviction for manufacturing methamphetamine upheld where the defendant owned a "boxed" laboratory in a storage locker, even though the locker did not contain a piece of equipment and a chemical agent which were necessary for the final step of the manufacturing process].) It is not clear to me why the majority does not use, as an abstract version of the felony, such ways of committing it where there is little chance of explosion.
Finally, on top of these uncertainties about the proper abstract version of the felony of manufacturing methamphetamine, the uncertainty of the second prong of the inherent dangerous inquiry completes the constitutional vagueness problem. We must find that there is a "high probability" that death will result from the abstract version of the felony, yet we have no guidance as to what a high probability is. That uncertain threshold alone causes unpredictability. But the data that is used as evidence underlying any such probability determination-which depends upon what litigants in a particular case choose to introduce-compounds the uncertainty.
In that regard, I share with the majority the view that, when courts consider whether a probability is "high" for these purposes, it is preferable that they act with "scientific precision" in a manner that is "evidence-based." (Maj. opn., ante , at p. 686.) The alternative is that judges simply use their subjective instincts. (See *698People v. Nichols (1970) 3 Cal.3d 150, 158 & fn. 3, 89 Cal.Rptr. 721, 474 P.2d 673 [stating without explanation that offense of burning a motor vehicle under Penal Code former section 449a was inherently dangerous, though the offense contained no requirement that the vehicle be occupied and the statute criminalized burning various other items, such as a stack of hay].) *969The statistical evidence here, however, comes down to weakly sourced testimony from the case agent that about one in five methamphetamine manufacturing crimes results in an explosion. (Maj. opn., ante , at p. 684.) Thus, in the instant case, we have no evidence of how often death results from methamphetamine manufacture. There surely is some causal connection between the frequency of explosions and the frequency of death. But it is only our gut instincts (or our subjective experiences) that determine whether this connection results a high probability, whatever level that is. Simply to illustrate the type of additional statistical evidence that might instead by considered by a court, I note that a study available on the Federal Centers for Disease Control and Prevention website evaluated 1,325 "meth-related chemical incidents" that were reported in five states (not California) over the 12 years covering 2001-2012, and found that in 87 (or 7 percent of the incidents) persons were injured, and there were a total of two deaths-one person, who might have been a meth cook, was found dead in a laboratory; the second death was of a law enforcement officer.6
I do not know whether this study is the best one to rely on if statistical evidence of deaths during methamphetamine were evaluated. If this study were considered, I do not know whether two deaths out of thousands of methamphetamine manufacturers should constitute evidence of a "high probability" in our circumstances. I also do not know how one could ever come up with a mortality statistic that reflects an abstract version of felony that is the least hazardous possible way of committing it, rather than reflecting "all" methamphetamine manufactures or the "ordinary" case.
I do know that the uncertainty of these matters is the type of uncertainty identified by the Supreme Court in Johnson , which explained how no judicial approach could prevent "the risk comparison required by the residual clause from devolving into guesswork and intuition." ( Johnson , supra , 135 S.Ct. at p. 2559.) These types of policy determinations are normally the domain of the Legislature. There is no statutory vagueness problem when our Legislature considers evidence and opts to add an offense, such as methamphetamine manufacture, to the list of offenses that can be predicates for first degree felony murder. One can find the legislative determination either wise or misguided, but, regardless, the statute is not vague. If a statute defines the crime, citizens have notice of what is criminal, and the courts have a reasonably specific penal law to apply.
But the situation is different where a court is charged with applying a loose legal standard to decide what crimes are felony murder predicates, based on *970the expert testimony or statistical evidence that a particular party thinks to introduce in a particular case. The judicial process is ideal for resolving individual disputes. Expert testimony is typically used to determine whether a particular person committed the crime. Here, in contrast, the majority is using expert testimony to determine, *699in essence, whether a matter should be criminalized. The logical consequence of such caselaw policymaking is that the criminal law could change if the next party does a better job of marshaling the real-world evidence. This possibility underscores the serious constitutional vagueness problem. Remarkably, in our prior decision on methamphetamine manufacture as a predicate, this court even acknowledged that new expert testimony offered in a subsequent case can change the status of a predicate offense. (See People v. James , supra , 62 Cal.App.4th at p. 263, 74 Cal.Rptr.2d 7 ["Once a published appellate opinion holds a felony is (or is not) inherently dangerous, that precedent is controlling, unless and until a litigant makes an offer of proof that technological changes have changed the status of the felony "], italics added.) Thus, apparently not only could a defendant be convicted of second degree felony murder where there was no law expressly including his crime at the time of his offense, but he could even be convicted if the law was squarely on his side, but the prosecution put on a better showing of statistical evidence and expert testimony in the next case. I believe this possibility is obviously anathema to the due process concept of fair notice, and it is one illustration of the particular problem with having courts take evidence in individual cases to determine what crimes qualify as second degree felony-murder predicates.
IV.
THE MAJORITY OPINION
As laid out in the preceding sections, my view is that our second degree felony-murder law is unconstitutionally vague under Johnson because it has a defendant's guilt depend on a court's evaluation of a hypothetical risk posed by an abstract generic version of the offense. Today's majority opinion eschews arguing the wholly contrary view. Rather than offer a full-throated defense of California's second degree felony-murder rule, the majority emphasizes that it is holding only that "this record presented before us in this case" does not support a finding of constitutional vagueness. (Maj. opn., ante , at p. 675.)7
*971It is not clear that such an "as applied" approach is available for the type of constitutional vagueness identified by Johnson . There, the Supreme Court found it "[t]rue enough" that "there will be straightforward cases under the residual clause, because some crimes clearly pose a serious potential risk of physical injury" but nevertheless held the residual clause facially void. ( Johnson , supra , 135 S.Ct. at p. 2560 ; see id. at p. 2561 ["why should the existence of some clearly risky crimes save the residual clause?"]; id. at p. 2580 (dis. opn. of Alito, J.) [arguing that the Court erred in "concluding that the residual clause is facially void for vagueness"]; Henry v. Spearman (9th Cir. 2018) 899 F.3d 703, 709 [Supreme Court "struck down the residual clause in its entirety, even as to 'straightforward cases' "].) While other vagueness challenges, such as those to the breadth of a statute's wording, allow for as-applied challenges, it may be that Johnson error differs because it implicates the manner in which courts apply a criminal law to all defendants charged with it. (See *700People v. Superior Court (Apr. 16, 2019) 34 Cal.App.5th 376, ----, 246 Cal.Rptr.3d 128, 2019 WL 1615288 at *14 [suggesting that, while an "as applied" test remains for vagueness challenges, Johnson appeared to involve an "exceptional statute in which the key defect could be established without examining the statute as applied to the challenger's circumstances"].)
Regardless of whether the majority's "as applied" approach is even a viable possibility, it is important to note that the majority's view here leaves open void-for-vagueness challenges to our state's second degree felony-murder rule as applied to other offenses and other records. Under this view, our state is left not merely with cases where the parties argue about the result of applying the "high probability of death" standard categorically to particular crimes, but where they also contest whether that abstract inquiry reaches some undefined level of unacceptable vagueness to be constitutionally void on the particular record about the particular offense.
Putting the availability of the "as applied" approach aside, I disagree with three ways in which the majority opinion attempts to distinguish Johnson and find methamphetamine manufacture inherently dangerous.
A. The Alleged Structural Differences Between Our Law and the Residual Clause
In an attempt to distinguish California's second degree felony-murder rule from the federal ACCA residual clause struck down in Johnson , the majority alleges some differences between the two laws. (Maj. opn., ante , at pp. 680-81.) Though the majority ultimately states that "these differences do not form the basis of our opinion" (Id. at p. 681), I will nevertheless explain why these differences are not relevant to the vagueness inquiry.
*972First, the majority finds it significant that the residual clause is one part of a test that has "two branches," including enumerated crimes such as burglary and arson, and the residual clause, which evaluated unenumerated crimes under a "serious potential risk of physical injury to another" standard. (Maj. opn., ante , at p. 680.) The majority states that California's second degree felony-rule "had no such dichotomy." (Ibid. ) The lack of any such dichotomy, however, would not support the constitutionality of California's rule, because Johnson approved the constitutionality of the enumerated crimes clause. ( Johnson , supra , 135 S.Ct. at p. 2563.) That is, the fact that California's rule lacks a branch that is not vague, containing only the portion that is indeterminate, is not helpful to the law. In any event, California does have a branch of its felony murder law that enumerates crimes: first degree felony murder, in which the legislature listed crimes such as burglary and arson that provide a predicate for felony murder. Like the ACCA residual clause, the second degree felony-murder rule provides an indeterminate test to identify other unenumerated crimes. And, as with adding enumerated crimes to the valid clause that preceded the residual clause, our state's vagueness problem could be cured if the legislature provided a complete list of qualifying crimes, rather than leaving the status of any unlisted felonies to be determined by courts after charges are filed.
A further dimension to the majority's focus on the ACCA enumerated crimes clause has been held by the United States Supreme Court to be immaterial to a Johnson vagueness claim. The majority points out that the California law is not a residual clause defined in part by the enumerated list of crimes that precedes it. (Maj. opn., ante , pp. 680-81.) Johnson 's two-pronged vagueness analysis, however, *701does not turn on whether the law at issue combines a list of enumerated crimes with "catch-all" language for similar, unenumerated crimes. In Dimaya , supra , 138 S.Ct. 1204, the Supreme Court applied Johnson in holding that a clause defining the term "crime of violence" was unconstitutionally vague. ( Dimaya , supra , at p. 1211.) Neither the clause struck down in Dimaya nor its preceding language "specifically invoke[d] ... enumerated felonies" (maj. opn., ante , at p. 681). Instead, what mattered was whether the law in question "had both an ordinary-case requirement and an ill-defined risk threshold." ( Dimaya , supra , at p. 1223.) "Strip away the enumerated crimes," as Congress did in the clause in question, the Court stated, "and those dual flaws yet remain[,] [a]nd ditto the textual indeterminacy that flows from them." ( Id. at pp. 1221-1222.)8 Those dual flaws existed in the language struck down in Dimaya just as they do here. *973Apart from its focus on the enumerated crimes clause, the majority identifies another difference that is meaningless to a categorical inquiry. The majority notes that the "emphasis in the ACCA is on prior convictions ." (Maj. opn., ante , at p. 680; see id. at p. 681, fn. 10 [ACCA analyzed "prior convictions," whereas the California rule concerns "a current conviction that is its own end"] ). But when a test for qualifying an offense under a particular standard is categorical-that is, when it has nothing to do with the particular facts of the defendant's case-it does not matter whether a court is examining a charge or a conviction, past or present. The only question is whether the offense, analyzed in the abstract, qualifies under the applicable standard. Because both the ACCA residual clause and California second degree felony murder apply categorical tests, it does not matter to a vagueness inquiry whether prior convictions or current charges are at issue. There can be no distinction for that reason.
Finally, the majority notes that Johnson considered the impracticality of requiring a sentencing court to reconstruct old crimes, which (unlike with the California second degree felony-murder rule) could come from diverse jurisdictions with varying laws. (Maj. opn., ante , at pp. 680-81.) This has nothing to do with the issue before us. The majority is relying on a passage in Johnson that explained-in response to a change urged by dissenting Justices-why the Court was sticking to a categorical approach to the residual clause, rather than focusing on the defendant's particular conduct. (See Johnson , supra , 135 S.Ct. at p. 2562.) It does not matter to our inquiry today why either the ACCA or the California test apply a categorical approach that does not evaluate the particular defendant's conduct. All that matters is that both do so. That simple fact satisfies the first prong of Johnson's two-prong test and thus goes a long way to demonstrating that the inquiries implicate similar constitutional vagueness concerns.
B. The Claim That Real-World Evidence Cures Johnson's Ills
I also am not persuaded by the majority's view that Johnson 's constitutional vagueness problems are cured when courts rely "on real-world concrete evidence on *702the issue of inherent dangerousness." (Maj. opn., ante , at pp. 681-82.) That evidence here was expert testimony on the crime's dangerousness. (Id. at pp. 675-76, 677, 683-84.) The majority holds that once there is a "foundation of expert testimony" and "real-world scientific" evidence there is no need for a hypothesized or abstract determination of inherent dangerousness, so Johnson does not apply. (Maj. opn., ante , at p. 684.) *974The strength of the real-world evidence of dangerousness and the need for reliance on "a judge-imagined abstraction" of a felony ( Johnson , supra , 135 S.Ct. at p. 2558 ) are simply different matters. Our second degree felony-murder law requires a "judge-imagined abstraction" not because evidence is lacking, but because courts are prohibited from considering the particular facts of the case before it and must evaluate ways that the offense can be committed. (See Howard , supra , 34 Cal.4th at p. 1138, 23 Cal.Rptr.3d 306, 104 P.3d 107.) Thus, if a death occurs during a felony that is claimed to be a basis for second degree felony murder, judges must ponder in the abstract whether there is some way (or ways) of committing that felony that is sufficiently safe to not carry a "high probability" of death. This endeavor may be uncertain regardless of whether or not scientific evidence of dangerousness of the offense in general, or of its particular applications, is presented.
Over the years, for example, our Court of Appeal has reached differing conclusions as to whether felony child endangerment in violation of Penal Code section 273a is an inherently dangerous felony. (Compare People v. Shockley (1978) 79 Cal.App.3d 669, 677, 145 Cal.Rptr. 200 [affirming second degree felony-murder conviction based on a conception that it is inherently dangerous to life to allow a "child to be placed in such a situation that its person or health is endangered"] with People v. Lee (1991) 234 Cal.App.3d 1214, 1224, 286 Cal.Rptr. 117 [reversing second degree felony-murder conviction in part by hypothesizing an offense that "could include a broken arm or leg" but not jeopardize the life of the child].) This difference in judicial hypothesizing has nothing to do with evidence of the dangerousness of felony child endangerment. It instead has to do with the conceptions judges formulate about the possible ways of violating the statute and the level of abstraction at which the dangerousness of the offense is evaluated. That problem exists in this case because the result can turn on what abstract ways of committing the offense of methamphetamine manufacture a judge imagines. The vagueness problem resulting from "judge-imagined abstractions" applies even when, as here, the trial court relies on statistical or expert testimony in considering whether a felony is inherently dangerous to human life.
C. Reliance on People v. James in Lieu of the Abstract Felony Inquiry
Finally, the majority relies heavily on our James opinion, but I am doubtful that citing that opinion clears up the Johnson vagueness problem. James cannot easily be reconciled with the Supreme Court's opinion in Howard, decided several years after this court decided James . In Howard , our Supreme Court held that the possibility of committing a felony-driving with a willful or wanton disregard for the safety of persons or property while fleeing from a *975pursuing police officer-in a nonhazardous way was enough to render it "not, in the abstract, inherently dangerous to human life." ( Howard , supra , 34 Cal.4th at pp. 1138-1139, 23 Cal.Rptr.3d 306, 104 P.3d 107.) In James , two California Department of Justice experts testified that a skilled or experienced methamphetamine cooker could manufacture methamphetamine *703safely. ( People v. James , supra , 62 Cal.App.4th at p. 264, 74 Cal.Rptr.2d 7 [one expert testified that "an experienced methamphetamine 'cooker,' who knows what he or she is doing, can make methamphetamine using ephedrine 'in a relatively safe way' "; another testified that "methamphetamine can be manufactured 'relatively' safely, 'if the person is very skilled,' " and that "[s]ome 'cookers' have produced thousands of batches without an explosion"].) Under Howard 's approach, this testimony seems adequate to establish that, in the abstract, manufacturing methamphetamine is not inherently dangerous. The majority's approach ultimately finds it sufficient to tie explosions to methamphetamine manufacture in general. That approach would render the crime in Howard inherently dangerous: after all, many car accidents occur from reckless driving with disregard for safety. But the crime in Howard was held not inherently dangerous, even though a lot of accidents result from its commission. It was not inherently dangerous because "not all violations of [the felony] pose a danger to human life." ( Howard , supra , at p. 1139, 23 Cal.Rptr.3d 306, 104 P.3d 107.) It seems to me that the difference between the approaches in these two cases demonstrates the grave uncertainty resulting from requiring courts to analyze abstractions of crimes.
V.
PEOPLE v. FRANDSEN
While this appeal was under submission, a panel in another Court of Appeal district held in People v. Frandsen (Apr. 4, 2019) 33 Cal.App.5th 1126, 245 Cal.Rptr.3d 658, 2019 WL 1486863 ( Frandsen ) that the second degree felony-murder rule is not unconstitutionally vague. In attempting to steer California's entire rule around Johnson 's vagueness test, Frandsen takes a different tack than today's majority, which has limited its opinion to "this record presented ... in this case." (Maj. opn., ante , at p. 675.) The reasoning in Frandsen , however, turns on a misunderstanding of one phrase in Johnson .
Frandsen correctly recognizes that in assessing whether a crime is inherently dangerous to human life under the second degree felony-murder rule, a California court "looks to the elements of the felony in the abstract, not at the particular facts of the case. [Citations.]" ( Frandsen , supra , 33 Cal.App.5th at p. 1142, 245 Cal.Rptr.3d 658, 2019 WL 1486863 at *10 ). Consequently, Frandsen correctly construes the California cases that have held Johnson inapplicable where a crime requires the court to apply a statutory standard to "real-world facts" underlying the defendant's conviction. ( Frandsen , supra , at pp. 1142-44, 245 Cal.Rptr.3d 658, 2019 WL 1486863, at *10-11 [discussing *976People v. Ledesma (2017) 14 Cal.App.5th 830, 222 Cal.Rptr.3d 534 and People v. White (2016) 3 Cal.App.5th 433, 208 Cal.Rptr.3d 1 ].) A statute that applies a qualitative standard to a defendant's actual conduct-such as a prohibition on "reckless" driving-is not unconstitutionally vague under Johnson . ( Johnson , supra , 135 S.Ct. at p. 2561.) Such a law provides the defendant fair notice of the standard under which he will be tried and punished.
But both the ACCA residual clause and California's second degree felony-murder rule do not involve consideration of a defendant's actual conduct. Rather, they both involve a categorical approach that has nothing to do with the defendant; instead, a defendant's liability turns on a court's classifying, under an imprecise standard of risk, the entire crime that he committed. Frandsen goes wrong in its attempted manner of distinguishing the two categorical inquiries.
*704Frandsen reads Johnson as containing an "[i]mplicit" holding that a categorization rule is not vague if it involves "consideration of the statutory elements of the crime ." ( Frandsen , supra , 33 Cal.App.5th at p. 1143, 245 Cal.Rptr.3d 658, 2019 WL 1486863 at *10.) It reasons that because the California second-degree felony murder rule involves "a statutory elements evaluation of risk," it is not an unconstitutionally vague law under Johnson . ( Frandsen , supra , at p. 1144, 245 Cal.Rptr.3d 658, 2019 WL 1486863, at *11 ).
This construction of Johnson is wrong. Johnson did not hold that consideration of the statutory elements of the crime immunizes an inquiry from unconstitutional vagueness. This is clear because the Supreme Court's test for applying the residual clause-which it articulated repeatedly, including in Johnson itself-required the consideration of the elements of the crime. (See Johnson , supra , 135 S.Ct. at p. 2564 ["To determine whether an offense falls within the residual clause, we consider 'whether the conduct encompassed by the elements of the offense , in the ordinary case, presents a serious potential risk of injury to another.' "], italics added; Dimaya , supra , 138 S.Ct. at p. 1253 ["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.' "], italics added, original italics omitted; James v. United States (2007) 550 U.S. 192, 202, 127 S.Ct. 1586, 167 L.Ed.2d 532 ["we consider whether the elements of the offense are of the type that would justify its inclusion within the residual provision"], original italics; Sykes v. United States (2011) 564 U.S. 1, 7, 131 S.Ct. 2267, 180 L.Ed.2d 60 [quoting preceding test from James v. United States ].) That is, in Johnson the Supreme Court invalided a test that involved the "consideration of the statutory elements of the crime." ( Frandsen , supra , 33 Cal.App.5th at p. 1143, 245 Cal.Rptr.3d 658, 2019 WL 1486863 at *10, italics omitted.) For this reason, there is no implicit holding of the *977type Frandsen discerns. The fact that the California second degree felony-murder inquiry involves consideration of the elements does not mean Johnson is inapplicable to it.
Frandsen 's view came from a misconstruction of a sentence in Johnson that stated that the residual clause inquiry was not tied "to real-world facts or statutory elements ." ( Frandsen , supra , 33 Cal.App.5th at p. 1143, 245 Cal.Rptr.3d 658, 2019 WL 1486863 at *10, citing Johnson , supra , 135 S.Ct. at p. 2557, original italics.) In context, Johnson 's reference to an inquiry tied to "statutory elements" means one that depends exclusively on the elements, leaving little or no room for judicial discretion. Two paragraphs before this sentence in Johnson , the Court explained why it was not invalidating another categorical test in the ACCA. That test, known as the elements clause, includes felonies that require force as an element. (See Welch v. United States , supra , 136 S.Ct. at p. 1261 ; 18 U.S.C. § 924(e)(2)(B)(i).) Johnson stated that with the residual clause, "[t]he court's task goes beyond deciding whether creation of risk is an element of the crime[,] ... unlike the part of the definition of a violent felony that asks whether the crime 'has as an element the use ... of physical force.' " ( Johnson , supra , 135 S.Ct. at p. 2557, italics omitted.) Thus, in distinguishing the residual clause from the elements clause, Johnson indicated that a law provides fair notice by stating that crimes containing a particular element (force) qualify for heightened punishment. This is the context for Johnson 's statement a few sentences later distinguishing the residual clause from laws tied to "statutory elements." (See also *705Welch v. United States , supra , 136 S.Ct. at p. 1261 [stating, after defining "elements clause" and "residual clause," that "[i]t is the residual clause that Johnson held to be vague and invalid"].) There is no example in Johnson of a satisfactory law that involves mere consideration of the elements. Especially when considered along with the fact that Johnson recognizes that the residual clause itself involved consideration of statutory elements, the sentence that Frandsen relies on cannot plausibly be read to mean that a legal test is protected from constitutional vagueness simply because it considers statutory elements.9 A test that depends exclusively on the elements, in contrast, is saved by the second prong of Johnson's vagueness test because it does not carry a "fuzzy risk standard" ( *978Dimaya , supra , 138 S.Ct. at p. 1221 ), even if it involves an abstract version of crimes and thus implicates the first prong of the test.
There is a difference between the residual clause inquiry and the second degree felony-murder one. But it is not whether statutory elements are "considered" as Frandsen stated. It is what the elements are considered for . With the residual clause, courts considered the elements to construct an ordinary case of the offense, which they then evaluated to see if it carried a serious potential risk of physical injury. For second degree felony murder, courts consider the elements to determine if they can construct a least harmful case in which the offense "possibly could be committed without creating" a high probability of death. ( Burroughs , supra , 35 Cal.3d at p. 830, 201 Cal.Rptr. 319, 678 P.2d 894.) A determination that our second degree felony-murder rule is not impermissibly vague must turn on a conclusion that our "possibly could be committed" inquiry provides fair notice of what is criminal, even though the residual clause "ordinary case" inquiry does not. As discussed earlier, I believe that both inquiries are indeterminate. Unlike the vast majority of criminal laws, both of these inquiries turn on the uncertainty of how judges will construe the defendant's offense in the abstract, against a qualitative risk standard.
In this regard, the principal case that Frandsen relies upon, People v. Hansen , supra , 9 Cal.4th 300, 36 Cal.Rptr.2d 609, 885 P.2d 1022 ( Hansen ), illustrates the unpredictability of California's "possibly could be committed" inquiry. Hansen held that discharging a firearm at an "inhabited dwelling house" ( Pen. Code, § 246 ) is inherently dangerous for the purposes of the second degree felony-murder rule. In holding that the crime is dangerous to human life, the court relied on the fact that people generally are in or around an inhabited building. ( Hansen , supra , at p. 310, 36 Cal.Rptr.2d 609, 885 P.2d 1022.) That is the ordinary case. But the test is whether the offense "possibly could be committed" without a high probability of danger to human life. Hansen recognized such a possibility: that "a defendant may be guilty of this felony even if, at the time of the shooting, the residents of the inhabited dwelling happen to be absent." ( *706Id. at pp. 324-325, 36 Cal.Rptr.2d 609, 885 P.2d 1022 ; see Pen. Code, § 246 [" 'inhabited' means currently being used for dwelling purposes, whether occupied or not"].) That means a defendant can commit the crime if, for example, he shoots at a cabin in the woods, knowing its occupant is away. With little explanation, however, Hansen rejected the unoccupied dwelling scenario as dispositive, viewing the offense at a high level of abstraction. Hansen 's analysis thus underscores the question: How can a litigant have fair notice as to which abstract possibilities will carry the day in court? In other cases, hypothetical possibilities have rendered the second degree felony-murder rule inapplicable. (See, e.g., People v. Lopez (1971) 6 Cal.3d 45, 51, 98 Cal.Rptr. 44, 489 P.2d 1372 [though prosecution argued that offense of escape was inherently dangerous because all escapes invite efforts at apprehension by officers, court rejected that argument by hypothesizing nonviolent *979escapes, such as one by a "committed inebriate who wanders off from a county road job in search of drink"].)
Johnson reviewed the case law concerning the residual clause and determined that judicial "experience" in applying that clause convinced it that the "uncertainties" were intolerable. ( Johnson , supra , 135 S.Ct. at p. 2560.) Hansen helps illustrate the same uncertainties in California's second degree felony-murder rule. As in Johnson , these differ in kind and degree from the uncertainties that affix to ordinary statutes. At the end of the day, I am not sure there can be a definitive answer to the question as to which abstract test-the residual clause in Johnson or California's second degree felony murder-rule-is more vague. Both inquiries require courts to assess hypothetical risks posed by an abstract consideration of offenses, so they are similarly malleable and unpredictable. The two inquiries cannot be cleanly distinguished in the manner that Frandsen attempts.
VI.
CONCLUSION
Our Supreme Court has not yet considered the second degree felony-murder law in light of the U.S. Supreme Court's new vagueness rule in Johnson . In a manner not previously articulated, Johnson 's two-pronged vagueness test delineates what makes our rule unconstitutionally uncertain. Because Johnson requires a finding of unconstitutional vagueness only when guilt depends on creating an abstraction of a crime divorced from the defendant's actual conduct, and then using an indefinite standard to evaluate the abstraction, it implicates only laws on the periphery of our Penal Code, if any others exist at all. Our courts surely will have few occasions where Johnson requires us to invalidate a statute. But Johnson could have been written about our second degree felony-murder law. That law embodies both inadequate notice to perpetrators and an indefinite standard for the courts. Under a straightforward application of Johnson , the law fails to comport with due process.
Importantly as well, apart from Johnson 's two-pronged test, the Supreme Court rejected the principle, indicated by some of its prior cases, that " 'a statute is void for vagueness only if it is vague in all its applications.' " (Johnson , 135 S.Ct. at p. 2561.) This principle has been applied in California as a federal constitutional requirement. (See, e.g., People v. Morgan (2007) 42 Cal.4th 593, 606, 67 Cal.Rptr.3d 753, 170 P.3d 129.) Before Johnson , the apparent need to prove across-the-board vagueness might have been an impediment to raising or applying a constitutional vagueness challenge to the second degree felony-murder law.
Johnson considered whether the residual clause was void for vagueness under the Due Process Clause of the Fifth Amendment (Johnson , supra , 135 S.Ct. at p. 2556 ), but the same analysis applies under the Due Process Clause of the Fourteenth Amendment and article I, section 7 of the California Constitution, both of which are implicated here. (See Welch v. United States , supra , 136 S.Ct. at pp. 1261-1262 ; People v. Heitzman (1994) 9 Cal.4th 189, 199, 37 Cal.Rptr.2d 236, 886 P.2d 1229.)
(See The San Diego Union-Tribune, SDSU grad indicted on charges he used campus lab to make drugs (June 30, 2005) < https://www.sandiegouniontribune.com/sdut-sdsu-grad-indicted-on-charges-he-used-campus-lab-2005jun30-story.html> [as of Apr. 25, 2019].)
(See The Modesto Bee, Ex-Ph.D student makes a deal in meth, theft case (Sept. 27, 2018) < https://www.modbee.com/news/local/crime/article3114437.html> [as of Apr. 25, 2019].)
(See Walter White (Breaking Bad )-Wikipedia, The Free Encyclopedia < https://en.wikipedia.org/wiki/Walter_White_(Breaking_Bad)> [as of Apr. 25, 2019].) The fictional White was a trained chemist who for a time manufactured methamphetamine in a professional laboratory.
Natalia Melnikova et al., Injuries from Methamphetamine-Related Chemical Incidents - Five States, 2001-2012 (Aug. 28, 2015) < https://www.cdc.gov/mmwr/preview/mmwrhtml/mm6433a4.htm> (as of Apr. 25, 2019).
See also majority opinion, ante , at page 671 ["the petition must be denied on this record"], p. 674 ["on this record, we do not find unconstitutional vagueness"], p. 686, fn. 16 ["[a]gain, our decision here is based on the record of this case"], p. 690 ["[t]he due process faults the United States Supreme Court found ... are not implicated here, on this record"], p. 690 ["consideration has been in the light of the record .... Given that record, we cannot find that he is entitled to a reversal of his conviction under Johnson ."]
Dimaya involved 18 U.S.C. § 16, the federal criminal code's definition of "crime of violence" that contained a residual clause the same as the ACCA's clause but no list of enumerated crimes. (Dimaya , supra , 138 S.Ct. at p. 1211.) The Court concluded that "Johnson tells us how to resolve this case" and that "just like ACCA's residual clause, § 16(b) 'produces more unpredictability and arbitrariness than the Due Process Clause tolerates.' " (Id. at p. 1223.)
Johnson also expressly did not invalidate the portion of the ACCA test found immediately before the residual clause that enumerates certain generic crimes, such as burglary, that are covered by the ACCA. (Johnson , supra , 135 S.Ct. at p. 2563.) That inquiry also depended exclusively on the elements of the defendant's offense, because "[t]o determine whether a past conviction is for one of those offenses, courts compare the elements of the crime of conviction with the elements of the 'generic' version of the listed offense." (Mathis v. United States (2016) --- U.S. ----, 136 S.Ct. 2243, 2247, 195 L.Ed.2d 604 ; see also Stokeling v. United States (2019) --- U.S. ----, 139 S.Ct. 544, 556, 202 L.Ed.2d 512 (dis. opn. of Sotomayor, J.) [discussing the valid elements and enumerated clauses, and the invalid residual clause].)