The Supreme Court heard oral argument in Welch v. United States last week, a case on whether the Court's decision on Johnson v. United States, which held that the residual clause of the Armed Career Criminal Act (ACCA) was void for vagueness, applies retroactively on collateral review. (Petitioner Welch and the United States agree that it does; an amicus appointed to defend the judgment below argued that it doesn't.) Whether it does or doesn't, I think, raises a difficult puzzle about why courts invalidate vague statutes that have clear applications.
The residual clause of the ACCA, if you're not familiar, provided that a felon convicted of possession of a firearm would receive a sentence with a mandatory minimum of 15 years, and a maximum of life, if he had three prior convictions for crimes which, inter alia, "involve conduct that presents a serious potential risk of injury to another." (Crimes counting towards this sentence enhancement also included burglary, explosives crimes, and arson, and prior to Johnson the Court had held that the risk posed by crimes captured by the residual clause must be similar to that posed by these listed offenses.) Largely because of the Court's (incorrect) insistence that the residual clause only captured crimes that "categorically" posed a serious potential risk of injury (that is, pre-Johnson the Court asked whether in an imagined ordinary case of a commission of a crime, that crime would pose a serious risk of injury, not whether the offender's actual conduct posed a serious risk of injury), the Court was unable to bring predictability to its or lower courts' interpretations of the residual clause, and abandoned the effort in Johnson, holding that the residual clause was void for vagueness.
Welch asks whether that holding is retroactive in habeas. For purposes relevant to Welch, a new rule announced by the Court applies retroactively to final convictions if it is substantive, but not if it is procedural. What does this mean? While the Court has, at times, suggested that substantive rules are those which interpret substantive constitutional guarantees, and that procedural rules are those which interpret procedural constitutional guarantees, it is tolerably clear from the Court's most fulsome majority opinion on the subject, linked to above, that what demarcates the two types of rules is their effects, not their sources. A substantive rule "alters the range of conduct or the class of persons that the law punishes," or, alters the range of conduct or the class of persons subject to a particular punishment. A procedural rule "do[es] not
produce a class of persons convicted of conduct the law does
not make criminal, but merely raise[s] the possibility that
someone convicted with use of the invalidated procedure might
have been acquitted otherwise."
Oral argument revealed broad consensus among the Justices that this characterization of the substantive/procedural divide is the right one. Even the Chief Justice commented, in questioning the amicus defending the judgment below, that a rule is only "properly categorized as procedural if there's some people who could be convicted . . . legitimately." On this rubric, Johnson's
invalidation of the residual clause, though an enforcement of the
Constitution's guarantee of fair notice (a procedural right), was a
substantive rule, because it produced a class of persons
whose sentences were illegitimate, i.e., who received a sentence that the law no longer provided.
The only issue on which any of the Justices appeared less than settled was whether there were, in fact, some persons who could be constitutionally sentenced under the residual clause. Suppose that the residual clause were constitutional as applied to cases where its application was clear. As Justice Alito's dissent in Johnson pretty persuasively argued, and the Johnson majority acknowledged, the residual clause's concept of a crime that involves conduct posing a serious risk of injury, while fuzzy-edged, does have many clear applications; crimes like attempted rape, attempted arson, or attempted kidnapping all obviously involve, in the typical case, conduct that presents a serious risk of injury. Felons in possession of a firearm with a sufficient number of prior convictions of crimes, like these, that obviously pose a serious risk of injury to others, would have been on fair notice that they would be subject to the residual clause's mandatory minimum. If they received fair notice, why should their sentences be invalid? If some subset of persons sentenced under the residual clause received valid sentences, Johnson would
then look like a procedural rule, one which merely raised a possibility
of a unconstitutional sentence in any particular case. Some people
sentenced under the residual clause would be the victims of a defective
procedure -- a statute that deprived them of fair notice -- but others
sentenced under the residual clause would not be the victims of a
defective procedure, because as applied to their prior convictions, the
residual clause did give fair notice.
The Chief Justice had a lengthy colloquy with the amicus along these lines, and eventually seemed to at least have doubts that the residual clause was unconstitutional in its clearest applications. It turns out, however, that the constitutionality of the residual clause in its clear applications was asked and answered in Johnson (which the Chief Justice joined) itself. In Johnson, Justice Alito argued, in dissent, that the residual clause wasn't vague in all its applications and therefore couldn't be facially void. Justice Scalia, who had previously made just this argument about facially invalidating vague statutes in dissent in Morales, replied for the Court that it had previously invalidated statutes as void for vagueness when those statutes had clear applications; for example, it had invalidated a statute that made it a crime for a grocer to charge an unreasonable price for food even though some prices would clearly be unreasonable. The Court ultimately held "that imposing an increased sentence under the residual clause . . . violates the Constitution's guarantee of due process," a holding which leaves no space for argument that residual-clause sentences may be constitutional in cases of prior offenses that clearly involve a serious risk of injury.
Johnson accurately characterized the Court's void-for-vagueness precedents; what sense, though, do those precedents, or Johnson, make? First, for avoidance of doubt, the Johnson Court expressly conceded it was "[t]rue enough" that "there will be straightforward cases under the residual clause, because some crimes clearly pose a serious potential risk of injury to another"; the Court only doubted how many of those straightforward cases there were. In one of these concededly straightforward cases, where is the due process violation, or if there isn't one, why should the defendant be able to assert the due process rights of others?
One answer is Henry Monaghan's famous "valid rule requirement," first advanced in a defense of First Amendment overbreadth doctrine - that people have a constitutional right not to be subjected to governmental sanctions under an invalid rule of law. Monaghan's arguments for, or really assertions of, the valid rule requirement have been the subject of what I view as fairly devastating criticism, but there is certainly some intuitive appeal to the claim that, e.g, after Citizens United someone who makes constitutionally proscribable inciting remarks can't be punished under a statute that bans all electioneering communications funded by certain sources, or that, after Heller, someone who owns a submachine gun can't be punished under a statute that bans possession of any gun. At a certain point, a statute that is broadly unconstitutional in most applications looks like it should be unconstitutional in all.
Thumbnail sketches of a few arguments for that intuition would go as follows. Perhaps a conviction grounded on, for example, the mere fact of gun possession improperly stigmatizes the defendant for engaging in protected conduct, or improperly expresses to the community that protected conduct is wrongful. Perhaps the trouble is simply that convictions under statutes like these don't require proof of unprotected conduct; whether or not it happened, a court didn't have to find it. Perhaps the reason that statutes like these can't merely be judicially narrowed to their constitutional applications, and upheld in such applications, is that we have no way of knowing whether the legislature would have regulated the constitutional applications had it realized that what it was doing was unconstitutional. (This is not, I think, just a severability concern, though it sounds like one.) Perhaps the problem with many, though not all, broadly unconstitutional statutes is their illicit purpose. Finally, we might agree with Matthew Adler that, while there are no personal rights against invalid rules, invalid rules are invalid rules and persons subject to them have the legal power to litigate their invalidity, whether or not their rights are violated. (To summarize a very complex argument, Adler argues that, for example, a flag-burner whose conduct violates an arson statute has no personal right against being punished for flag-burning when his underlying conduct is constitutionally proscribable, and that examples like these show that the deep structure of judicial review isn't about vindicating personal rights, but about invalidating invalid laws and using private litigants as a mere means to do so.)
The problem with facially invalidating vague statutes with non-vague applications is that none of these arguments seem to apply to vagueness. First, convictions for clear violations of broadly vague statutes don't improperly stigmatize any improper conduct. The unexceptionable message sent by an enhanced sentence for a crime that clearly poses a serious risk of injury is that crimes that clearly pose a serious risk of injury are bad, or warrant incapacitation.
Second, there's no epistemic objection to clear applications of vague statutes; when someone clearly violates a merely vague statute, we know, and the court has found, that he's engaged in some unprotected and punishable conduct.
Third, while narrowing a broadly unconstitutional statute to its constitutional core raises concerns of judicial legislation, narrowing a broadly vague statute to its clear applications doesn't raise the same concerns. A legislature that bans a large set of mostly protected speech acts may well not have banned the unprotected ones had it legislated constitutionally, because the reasons animating the former ban might not underwrite a much narrower piece of legislation (or be reasons on which Congress can constitutionally act). In cases of vagueness, though, a legislature that understood it was legislating vaguely would likely have merely legislated more clearly, and almost certainly would have captured, in a clearer statute, the conduct clearly captured by the vague statute. Congress's response to the invalidation of the residual clause is unlikely to be a rethinking of whether attempted rapes should enhance sentences; the clear applications of a vague statute are likely to be the very prototypes of what Congress intended to capture.
Fourth, while vague statutes, like any statutes, can be a product of illicit purposes, they're not especially likely to be, and there's certainly nothing illicit about wanting longer sentences for felons in possession of firearms with prior convictions for dangerous crimes. The problem in Johnson was solely Congress's failure to translate that reasonable policy goal into a clear legal command.
Fifth, as to an Adler-inspired argument for facial invalidation of vague statutes, even Adler concedes that vagueness challenges are a matter of personal rights - however proscribable the underlying conduct, a person convicted under a vague statute is deprived of his personal right to notice - which undercuts any Adlerian argument that people whose rights haven't been deprived by vague statutes can litigate their vagueness.
There are, I would concede, pragmatic arguments for the facial invalidation of vague statutes. These would simply be that, absent facial invalidation, courts are likely to underenforce fair-notice norms in as-applied vagueness challenges, and that the post-invalidation enactment of a statute that clearly captures what the legislature had vaguely described would create a vastly preferable state of affairs to a regime of as-applied vagueness challenges that both results (along with all manner of unpredictable results) in the underenforcement of fair notice and the underenforcement of the legislature's policies, to the extent that cases the legislature wanted to capture were captured only vaguely. I think it's probable that at least some members of the Court were thinking in these terms when they decided to facially invalidate the residual clause in Johnson.
Assessing the strength of these arguments, I think, requires some empiricism about the likelihood of a legislative response to invalidation; I'm not sure they have much force in a world where Congress is unlikely to respond. But even if we were confident about that, the pragmatic goods facial invalidation of vague statutes obtains do not require retroactivity, and may be canceled out by it. A legislative response to invalidation would enforce Congress's policy prospectively, but it will do nothing to remedy the retroactive de-enforcement of its policy. As far as retroactivity goes, the only pragmatic argument to be made is that, given the unavailability of retroactive application of procedural rules (which a rule inviting as-applied vagueness challenges probably would be*), falsely assuming that all convictions under a vague statute worked deprivations of fair notice produces more accurate, and thus, just results than denying retroactivity in all cases. That kind of argument, however, is in irreconcilable tension with the Court's decision to deny procedural rules retroactive effect in the first place, as retroactive application of procedural rules would be even more accurate than pretending those rules were violated in every applicable case.
* Note that this may be wrong. The substance/procedure divide may turn on formalism, not what the procedure of retroactive application would look like, or whether a rule generates possibilities or certainties of error, and to the extent a defendant is in jail for a crime which, as applied to him, the law does not punish (or is serving a sentence for conduct which, as applied to him, is not sentence-enhancing), the rule creating that as-applied invalidity could be deemed substantive.