« Your Turn, Retributivists | Main | Let’s Give Them Somethin’ To Talk About—A Tax Fix »

Monday, November 16, 2015

How an Obscure SCOTUS Procedure Can Solve AEDPA's Retroactivity Catch-22 (and a Growing Circuit Split)

Thanks to Montgomery v. Louisiana, the retroactivity of new Supreme Court decisions is already an important part of the Court's current Term. But as I explain in the post that follows, a new application pending before the Justices, In re Butler, raises a far more important retroactivity question, one that is already the subject of a 5-3 (and growing) circuit split, one that has an ever-shortening clock, and, most significantly, one that may only be definitively answerable if the Court does something it hasn't done in 90 years--issue an "original" writ of habeas corpus.

To unpack this dense but significant topic, Part I flags the origins of the problem--the restrictions on second-or-successive applications for post-conviction relief in the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), and the Supreme Court's fractured 2001 interpretation of those provisions in Tyler v. Cain. Part II turns to the current circuit split, which involves whether the Court's June decision in Johnson v. United States, which invalidated a provision of the Armed Career Criminals Act (ACCA), can be retroactively enforced in second-or-successive petitions. Part III explains why that circuit split can't be resolved by the Supreme Court via certiorari--and why, instead, the best way for the Court to take up the Johnson question is through an "original" writ of habeas corpus in a case like Butler. Finally, Part IV argues that the Court should use Butler not just to answer the Johnson question, but also to resolve the debate over Tyler, lest this exact same scenario repeat itself after the next Johnson-like ruling.

I.  AEDPA and Tyler v. Cain

Of all of AEDPA's restrictions on post-conviction relief, perhaps none are more sweeping than the limits on "second-or-successive" petitions filed in federal courts by state or federal prisoners. As relevant here, AEDPA requires petitioners in such cases to first get permission to file such a claim from the Court of Appeals, which may only "certify" the claim if, as relevant here, it relies upon "a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable."

In other words, unless the claim is based upon newly discovered evidence, second-or-successive petitions can only go forward when they rest upon new Supreme Court decisions that, under Teague v. Lane, may be enforced retroactively. (Under Teague, new "substantive" rules may be retroactively enforced, whereas new "procedural" rules may not, unless they are "watershed" rules of criminal procedure). That part is clear (or, at least, well understood). What's less clear is the meaning of the word "made" in the emphasized language above: Must the Supreme Court expressly state that the particular new rule in question is retroactive, or is it enough that the retroactivity of the rule obviously follows from--and is effectively settled by--other existing Supreme Court retroactivity precedents? 

In Tyler v. Cain, the Justices considered this very question, and ruled, 5-4, that "a new rule is not 'made retroactive to cases on collateral review' unless the Supreme Court holds it to be retroactive." But Justice O'Connor, whose vote was necessary to the result, opened the door to a slightly broader interpretation in her concurrence. As she wrote,

a single case that expressly holds a rule to be retroactive is not a sine qua non for the satisfaction of this statutory provision. This Court instead may “ma[k]e” a new rule retroactive through multiple holdings that logically dictate the retroactivity of the new rule. . . . [I]f we hold in Case One that a particular type of rule applies retroactively to cases on collateral review and hold in Case Two that a given rule is of that particular type, then it necessarily follows that the given rule applies retroactively to cases on collateral review. In such circumstances, we can be said to have “made” the given rule retroactive to cases on collateral review.

In the 14 years since Tyler, the lower courts have generally followed Justice O'Connor's concurrence, rather than Justice Thomas's majority opinion. The debate, instead, has focused on whether particular rulings qualify as a "Case Two." The latest battleground on this question involves the Court's June decision in Johnson.

II.  Johnson and the Circuit Split

In Johnson, as noted above, an 8-1 Court struck down the so-called "residual clause" of the ACCA on the ground that it is impermissibly vague. As a result, not only are sentences based upon the residual clause no longer valid, but individuals who have already served what would otherwise be the statutory maximum (10 years) are presumably entitled to release, since there is no longer any positive authority for their continuing incarceration. Although there's therefore little question that Johnson falls on the "substantive" side of the Teague line, there's also nothing in Justice Scalia's opinion for the Court that says as much--and that therefore clarifies, per Justice O'Connor's Tyler concurrence, that Johnson's new rule is "of that particular type." In other words, Johnson may or may not be a "Case Two," depending upon whether the Court has to expressly say that it's a substantive rule, or whether it's enough that, based upon prior decisions, it's clear that its rule is substantive.

That's the issue on which lower courts have divided. As of this writing, five circuits (the First, Second, Seventh, Eighth, and Ninth) have authorized second-or-successive claims based upon Johnson, with the First, and Seventh providing detailed opinions explaining that, in their view, it's sufficiently clear from existing Supreme Court precedent that Johnson's rule is substantive, and is therefore an example of Justice O'Connor's "Case Two." Three circuits (the Fifth, Tenth, and Eleventh--over a dissent) have held to the contrary--reading Justice O'Connor's concurrence to require some explicit recognition by the Court that a new rule is "of that particular type," i.e., substantive for purposes of retroactive enforcement.

This circuit split is deeply problematic in two respects: First, it creates massive inequities as between federal prisoners convicted under ACCA's residual clause in the five circuits that have allowed second-or-successive Johnson claims and the three that haven't--with those in the latter category now in prison pursuant to convictions that, at the very least, should trigger resentencing, if not (for those who have served 10 years) outright release. Second, because AEPDA imposes a rigid one-year statute of limitations on second-or-successive claims, those serving potentially unlawful sentences have a closing window within which to obtain relief based upon Johnson. Under AEDPA, such claims must be filed by June 26, 2016--the one-year anniversary of Johnson itself. Thus, the circuit split needs to be resolved by the end of this Supreme Court Term--if not sooner.

Usually, of course, there's an easy way to resolve a circuit split like this one; the Court just grants certiorari to review one (or more) of the circuit-level decisions. Here, however, that's not possible: AEDPA itself takes away the Court's certiorari jurisdiction in cases in which the Court of Appeals denies certification (which insulates the Fifth, Tenth, and Eleventh Circuit decisions from review), and the party that lost in the other five circuits that granted certificates--the federal government--hasn't sought certiorari, ostensibly because it agrees that Johnson can be enforced retroactively in second-or-successive cases. 

Thus, AEDPA, TylerJohnson, and the government's litigating position have produced something of a perfect storm--where there's a major circuit split, and no immediately obvious way for the Supreme Court to resolve it. Enter In re Butler, a petition for an "original" writ of habeas corpus from the Supreme Court.

III.  The Supreme Court's "Original" Habeas Jurisdiction

There are few topics in Federal Courts in which there's a bigger disconnect between academic interest and real-world significance than the Supreme Court's so-called "original" habeas jurisdiction -- "'original' in the sense of being filed in the first instance in [the Supreme] Court, but nonetheless for constitutional purposes an exercise of [the] Court's appellate (rather than original) jurisdiction." Ever since Ex parte McCardle, the Court has alluded to "original" habeas writs as a crucial constitutional backstop -- "an unorthodox but sometimes necessary means of exercising review in situations where other avenues for relief are either practically or formally unavailable." Thus, the theoretical availability of original habeas has allowed the Court to sidestep the grave constitutional questions that would otherwise have arisen from various efforts to strip the Court's appellate jurisdiction in habeas cases, including in McCardle itself, and, more recently, Felker v. Turpin.

In Felker itself, the Court was also dealing with AEDPA's limits on its certiorari jurisdiction over second-or-successive petitions, and had no trouble recognizing its power to issue "original" writs of habeas corpus, which AEDPA left untouched, as an available remedy in appropriate cases--and one that obviated constitutional objections to AEDPA under the Exceptions Clause of Article III. As Justice Souter warned in his concurrence, though, "if it should later turn out that statutory avenues other than certiorari for reviewing a gatekeeping determination were closed, the question whether the statute exceeded Congress's Exceptions Clause power would be open." And, in an eerily prescient coda, he added, "The question could arise if the courts of appeals adopted divergent interpretations of the gatekeeper standard," i.e., exactly what has happened after Johnson.

Justice Souter's warning has proven prophetic; although the Court has received a number of serious, significant original habeas applications in the 19 years since Felker, it has yet to grant a single one, even in a 1999 retroactivity case in which the federal government agreed that original habeas was warranted on remarkably similar facts [the issue in that case was eventually resolved--against retroactivity--in Tyler]. Some of these petitions have come in high-profile capital cases, such as those of Troy Davis and Warren Lee Hill, where original habeas was the only way to prevent potentially unconstitutional executions. But whereas those cases may present more emotionally stirring narratives, the Johnson retroactivity issue is, in some ways, a cleaner vehicle for an original writ, since (1) the question before the Court isn't a "merits" question, but a simple retroactivivty question; and (2) the relevant statutes specifically contemplate that the Supreme Court, and not the lower courts (or state courts), will provide the definitive answer to that question. At the very least, if the Court wasn't going to grant in cases like Davis and Hill, and if it's not going to use original habeas to resolve disputes like the Johnson retroactivity issue, then original habeas really is a historical relic--and the constitutional questions Justice Souter worried about might finally have to be confronted.

IV.  Using Original Habeas to Solve the Problem

If you're still reading, hopefully I've convinced you by now that the Court should grant an original writ of habeas corpus to resolve the Johnson retroactivity issue (or, at the very least, should set the matter for full briefing and argument). There's at least one other pending original application raising the same question, but what makes Butler so attractive is the sentencing issue--because he has already served 10 years, he's entitled to outright release if Johnson is enforceable through a second-or-successive petition, meaning that the Court could simply grant habeas relief and be done. But should the Court do more than just grant the writ? Briefly, let me sketch out two further steps the Court can take--and then explain why, in my view, one is clearly better than the other:

A.  Hold that Johnson is Retroactive

Beyond simply granting the writ in Butler's case (which would leave other cases unsettled), the easiest way out, which would take about a paragraph, would be to expressly hold that Johnson is a "substantive" rule under Teague, and to therefore "ma[k]e" it retroactively enforceable in second-or-successive petitions under AEDPA. Such a ruling would then allow prisoners in the Fifth, Tenth, and Eleventh Circuits to obtain relief--including resentencing and, in cases like Butler, outright release. That wouldn't resolve the circuit split; it would simply moot it, since there would no longer be any question over whether the Supreme Court had "made" Johnson retroactive.

B.  Resolve the Circuit Split Over the Meaning of Tyler

The shortcoming of that approach is that, while it would moot the circuit split over Johnson, it wouldn't resolve the cause of the circuit split--i.e., lingering disagreement over the meaning of Justice O'Connor's Tyler concurrence. Thus, to avoid this exact scenario from arising again, the Court could use an original writ in a case like Butler to clarify who has the better of Tyler--the circuits that interpret it liberally to allow retroactive enforcement whenever it is sufficiently clear that a new rule is substantive, or the circuits that interpret it narrowly to require the Supreme Court to specifically say that a new rule is substantive. I have my own views on how the Justices should answer that question (both on the merits and to spare them from having to take pointless follow-on cases after each new rule is announced). But however this question is resolved, it should be clear that settling it in the context of Johnson will have a salutary effect for future litigation. 

*                              *                              *

In a paper I wrote in 2011, I argued that there's actually a value in preserving the obscurity of the Supreme Court's original habeas jurisdiction--and that, if original writs became common, they'd lose their utility as a safety valve, since Congress would presumably also think to take away that authority as part of future jurisdiction-stripping initiatives. But there's a difference between elusive remedies and illusory ones. For two decades, we have labored under the fiction that AEDPA's gatekeeper provisions don't raise serious constitutional problems entirely because of this safety valve. If, as a result of disuse, it turns out that the safety valve is sealed shut, then we can no longer dodge those constitutional questions. Thus, although we may be in the midst of a perfect storm for retroactivity, a case like Butler may actually be the perfect vehicle for the Justices to remind themselves about their original habeas authority--and, in the process, to issue an opinion that dramatically reduces the need for such relief in future retroactivity cases.

Posted by Steve Vladeck on November 16, 2015 at 10:54 AM in Civil Procedure, Constitutional thoughts, Criminal Law, Steve Vladeck | Permalink


You seem to take it as given that a decision invalidating part of a criminal statute, on procedural due process grounds, must be deemed substantive. That is quite a leap: the SCOTUS didn't hold that armed career criminals can't ever be given enhanced sentences- only that Congress must speak more clearly than it did. So I don't place the rule in Johnson on par with the rule in say, Roper, which categorically prohibits capital punishment for the mentally retarded.

Posted by: Da Man | Nov 16, 2015 5:13:19 PM

Granted that Johnson isn't Roper. But I don't think there's any question that it's substantive under Teague, Bousley, and Summerlin, and neither do the lower courts or the federal government. Indeed, in _first_ 2255 motions, it's already being enforced retroactively without any real pushback. Of course, the Court could say on the merits that it _isn't_ substantive, but that seems very unlikely--and would, in any event, be an important explanation of why everyone is wrong...

Posted by: Steve Vladeck | Nov 16, 2015 7:49:46 PM

One tricky aspect of this would be deciding whether to take up an ACCA case or a guidelines career offender case. The latter would implicate another burgeoning circuit split (or possibly two) about whether vagueness doctrine applies to the guidelines at all, and whether guidelines error is ever cognizable on habeas. Albeit it would be possible for those issues to come to the Court outside the second or successive context, but just taking an ACCA second or successive case would leave them unresolved.

Posted by: Jay | Nov 16, 2015 9:04:55 PM

About your last comment, I believe you're misreading the Fifth Circuit's decision (at least), which very clearly holds that Johnson isn't substantive. As they creatively see it, Johnson merely held that the government can't impose ACCA's enhanced sentences under a certain procedure - the faulty notice given by the vague residual clause. As to the other circuits on that side of the split, the Tenth Circuit says it isn't for them to decide, on a successive motion, whether Johnson is substantive, and expresses no view; the Eleventh Circuit says in dicta that it's substantive, but then says (bafflingly) that it's not the sort of substantive rule that the Court has specifically held post-Teague is substantive. So there's actually a circuit split on whether Johnson's substantive, and the two circuits that specifically rely on the held-by-the-Court requirement to deny relief do so for different reasons; the Tenth says the Court has to hold Johnson substantive, while the Eleventh says that the Court must pronounce some sub-category of substantive rules into which Johnson falls substantive.

Posted by: Asher Steinberg | Nov 16, 2015 9:08:27 PM

Asher: Fair enough re: the Fifth Circuit (although that's nuts, in my view). But as I hope is clear, the importance of original habeas in a case like Butler doesn't depend upon the _answer_ to whether Johnson is substantive. Either way, there's a circuit split for which there's no remedy other than extraordinary relief from the Supreme Court. I'd be awfully surprised if the Court says Johnson isn't substantive, but that would be an important pushback against the five circuits that have granted post-Johnson certificates...

Posted by: Steve Vladeck | Nov 16, 2015 9:17:35 PM

Your point is clear, but actually, I don't know if you're right that extraordinary relief is needed, given the Fifth Circuit's ruling. Because couldn't there be, in the Fifth Circuit, a first petition arguing Johnson's retroactivity, and wouldn't the Fifth Circuit be bound by Judge Higginbotham's opinion to say Johnson is a procedural, non-watershed rule (or to deny a COA on the ground that Johnson's retroactivity is settled there), and wouldn't the Court have certiorari jurisdiction from that decision or that denial of a COA, on a first petition? I believe the answer to all of these questions is yes. If that's right, then the Court has certiorari jurisdiction to decide Johnson's retroactivity, very may well get a case out of the Fifth Circuit to decide just that, probably would grant to correct the Fifth Circuit's "nut[ty]" view (though I'm not sure it's that nutty), and, in doing so, would take care, going forward, of Johnson's retroactivity for second/successive purposes.

Of course, they wouldn't clear up Tyler that way, but just from reading these opinions I'm somewhat unconvinced that there's really a sharp circuit split on what Tyler means. The basis for the Eleventh Circuit's disagreement with the Seventh and First, for example, turns out to be incredibly picayune (and off-base). They say that, while there is a case, Bousley, that holds that a decision which narrows a statute's scope by intepreting its terms is substantive, and while Summerlin repeats Bousley's holding, and while Johnson narrowed the ACCA's scope by interpreting its terms, and while you don't need a case that specifically holds Johnson is retroactive to satisfy Tyler, Bousley retroactivity can't count for second-or-successive purposes because Bousley happened to involve a pure question of statutory interpretation and isn't a rule of constitutional law as 2244 requires ("a new rule of constitutional law"). Except, of course, 2244 only says that the rule that applies retroactively must be constitutional, not that the case that makes it retroactive must be constitutional, or must itself involve the retroactivity of a constitutional rule. If the latter were the case, maybe nothing would be retroactive for second-or-successive purposes because we don't even know whether Teague is a constitutional rule.

Posted by: Asher Steinberg | Nov 16, 2015 10:17:51 PM

Asher: I certainly don't disagree that it's picayune.

As for the possibility of cert. from the Fifth Circuit's rejection on the merits of a first 2255 motion based upon Johnson, that's certainly _possible_, but (1) it assumes a Fifth Circuit panel will feel bound in a non-2255(h)(2) case by a 2255(h)(2) decision (I can imagine a world in which it wouldn't); and (2) in any event, and more importantly, it would have to get to the Court _and_ be decided by June 26, 2016, because 2255(f)(3) starts the one-year clock for second-or-successive motions on "the date on which the constitutional right asserted was _initially recognized_ by the Supreme Court, if the right has been _newly_ recognized by the Supreme Court and made retroactively applicable to cases on collateral review." In other words, the clock runs from Johnson, not a subsequent decision on cert. to the Fifth Circuit. In the absence of a Fifth Circuit decision that can get the issue to the Court between now and then, doesn't that only further justify original habeas?

Posted by: Steve Vladeck | Nov 16, 2015 10:26:06 PM

Oh, you're right about the clock problem, though I guess it's possible that the Fifth Circuit could deny a COA on a first petition tomorrow, or in enough time that the Court could grant cert and review the denial this term. They ought to be able to act pretty speedily on such a request for a COA, given Higginbotham's opinion. And so should the district courts below.

On the question of whether a Fifth Circuit panel would feel bound in a non-2255(h)(2) case by that opinion, the opinion barely even engages with the Tyler analysis; it seems to think that there's an antecedent question of whether the rule applies retroactively and decides that question. That's wrong (in the same way Saucier was - analytically at least - wrong), but it's how the opinion goes. Now, if the opinion had been written by a very conservative panel of, say, the Sixth or Ninth Circuits, I could imagine some other panel of those circuits being disingenuous enough to misread the opinion as really just doing Tyler, or brash enough to say that the bulk of the opinion's analysis is one huge dictum (which is arguable). But it's very unlikely that there's a Fifth Circuit panel out there that would do that, though conceivable.

Posted by: Asher Steinberg | Nov 16, 2015 11:23:34 PM

One of the biggest errors in the Fifth Circuit's opinion is to limit Johnson to holding that the residual clause simply failed to provide the "notice" that due process requires of criminal statutes. That is entirely incorrect. The panel forgets (or ignores) that vagueness doctrine has two independent prongs: notice and arbitrary enforcement--the latter of which having been repeatedly emphasized in SCOTUS opinions as the more important prong. In Johnson, the Court made two things crystal clear. First, that the residual clause violated BOTH prongs. And second, that the more constitutionally disturbing violation was the arbitrary and subjective inquiry required in order to apply the clause to any given defendant.

When one considers Johnson's actual holding and the Court's analysis of the statute, it becomes almost indefensible to argue that the rule it announced was not substantive. To be sure, the Johnson court did not hold that Congress cannot impose a mandatory minimum prison term based on a recidivist punishment provision. But what the Court did hold is that Congress cannot write such a provision in a way that requires sentencing judges to ignore the particular defendant's actual conduct in committing the predicate offense and instead determine whether the provision applies based on two subjective assessments of the universe of conduct that might (or might not) be involved when anyone commits that predicate offense: (1) what is the typical way this offense is likely to be committed; and (2) how much risk is inherent in that typical instance. That is a substantive rule that precludes Congress from enacting any statute that bases a mandatory minimum sentence on an inquiry that is necessarily arbitrary in each and every application.

The focus on "notice" in the cases deciding against retroactivity is, at best, either a misreading of Johnson itself or indicative of a larger misunderstanding of the vagueness doctrine in general.

Posted by: Evan | Nov 17, 2015 10:47:49 AM

"Granted that Johnson isn't Roper. But I don't think there's any question that it's substantive under Teague, Bousley, and Summerlin, and neither do the lower courts or the federal government."

I don't know that I would place enormous stock in the fact that the federal Government agrees that Johnson should apply retroactively. The decision does not make non-criminal what was formerly criminal; and it doesn't sayd that a certain class of offender's can't be convicted or punished. So what you are really arguing is that Teague's first exception includes any decision facially invalidating the statute that gave rise to the conviction or sentence--whatever the reason for the invalidation. I don't think that it's so clear cut.

Posted by: Da Man | Nov 17, 2015 1:51:22 PM

Well, Summerlin says that substantive rules include "decisions that narrow the scope of a criminal statute by interpreting its terms," which I'll stipulate isn't quite what Johnson's doing, but isn't an invalidation of a piece of a statute a fortiori substantive?

Posted by: Asher Steinberg | Nov 17, 2015 2:19:26 PM

Just a note that AEDPA takes away the Court's certiorari jurisdiction over both the denial and the grant of certification (which only strengthens your argument).

Posted by: James F. | Nov 17, 2015 2:31:57 PM

Then wouldn't there be tons of decisions retroactively applying SCOTUS decisions invalidating loitering statutes? I haven't found any.

Posted by: Da Man | Nov 17, 2015 3:43:01 PM

James -- True enough, but it doesn't take away jurisdiction over the subsequent 2255 motion, which can come up through the ordinary district court --> circuit court --> certiorari process. So the government could theoretically wait until it loses on the merits in a post-Johnson 2255(h)(2) motion, and seek direct review of _that_. It's because they won't that this is an issue.

Posted by: Steve Vladeck | Nov 17, 2015 4:25:17 PM

Da Man: No, unless there were tons of habeas petitions by convicted loiterers. But my sense is that not many people are in prison for loitering, and fewer are in prison for long enough to bother with habeas. Now, I do think you'll find that those loitering decisions have been applied on habeas review of convictions, postdating those decisions, under other statutes carrying longer sentences that prisoners argue are unconstitutionally vague, but to be fair to you that probably doesn't show much because I doubt those decisions announce new rules about vagueness generally.

Posted by: Asher Steinberg | Nov 17, 2015 5:27:12 PM

Asher -- Good point about brevity of confinement. Plus, a state statute invalidated on vagueness grounds would invariably lead to a state collateral attack (and probably a successful one at that), which would obviate the need for resort to federal habeas. If only the Supreme Court had invalidated 18 USC 1346 instead of narrowing it -- then we'd have the proper vehicle.

Posted by: Da Man | Nov 17, 2015 5:57:16 PM

Post a comment