Bowe v. United States

JUSTICE GORSUCH, with whom JUSTICE THOMAS and JUSTICE ALITO join, and with whom JUSTICE BARRETT joins as to Part I, dissenting.¶

Under the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), courts of appeals serve an important gatekeeping function. When a state prisoner seeks to challenge his conviction or sentence in federal court for the first time, he normally proceeds directly to district court under 28 U. S. C. §2254. But before a state prisoner may bring a second or successive attack in district court, AEDPA requires him to obtain a certificate from a panel of the relevant federal court of appeals. And under §2244, an appellate panel may certify a case to proceed only subject to certain demanding rules.¶

Two are relevant here. First, an appellate panel may not authorize do-over claims—claims the prisoner already “presented in a prior” federal challenge. §2244(b)(1). Second, neither a state prisoner nor state authorities may file a petition for rehearing or seek “a writ of certiorari” to contest an appellate panel’s decision granting or denying certification. §2244(b)(3)(E).¶

AEDPA contains parallel directions for federal prisoners. When a federal prisoner seeks to challenge his conviction or sentence collaterally for the first time, he usually proceeds directly to district court under §2255. But before he can pursue a second or successive challenge, it “must be certified as provided in section 2244 by a panel of the appropriate court of appeals.” §2255(h). Through that crossreference, then, federal prisoners seeking to bring second or successive collateral attacks are generally subject to the same rules as their state counterparts.¶

That brings us to Michael Bowe. After shooting two guards while trying to rob an armored vehicle in 2008, he pleaded guilty to three federal crimes. Since then, he has pursued five nearly identical postconviction challenges. In each, Mr. Bowe has asserted that one of the statutes under which he was convicted, 18 U. S. C. §924(c), is unconstitutional under our void-for-vagueness doctrine. And in all but the first of his challenges, Mr. Bowe has centered his argument around the same case: United States v. Davis , 588 U. S. 445 (2019).¶

When Mr. Bowe sought permission to bring his latest collateral attack, the Eleventh Circuit demurred. Because he had pursued his Davis claim three times before, the court held that his latest filing represented an impermissible doover claim barred by §2244(b)(1). Now, Mr. Bowe asks us to reverse the Eleventh Circuit’s decision. Yes, he concedes, AEDPA bars state prisoners from pursuing do-over claims. But, he insists, federal prisoners are different. They (and they alone) are free to bring the same claim in district court over and over again.¶

Before we can take up that question, however, we must answer another. Under §2244(b)(3)(E), Mr. Bowe acknowledges, this Court lacks certiorari jurisdiction to review an appellate panel’s decision denying a state prisoner leave to file a second or successive collateral challenge. But here again, Mr. Bowe argues federal prisoners are different. He submits that they (and, once more, they alone) are free to challenge adverse certification decisions on writ of certiorari.¶

Because I do not believe that either of Mr. Bowe’s pleas for federal-prisoner exceptionalism can be squared with AEDPA’s text, I respectfully dissent.¶

I¶

I understand my colleagues’ eagerness to reach the merits of this case. As they observe, the courts of appeals are divided on the question whether federal prisoners like Mr. Bowe may pursue do-over claims. Ante, at 5. Doubtless, too, it is often this Court’s job to resolve circuit splits and bring nationwide uniformity to the application of federal law.¶

But it is also one of our paramount responsibilities to respect the limits on our jurisdiction. And, to my eyes, AEDPA precludes us from exercising certiorari jurisdiction here. Section 2244 instructs that “the grant or denial of an authorization by a court of appeals to file a second or successive” collateral attack “shall not be the subject of a petition . . . for a writ of certiorari.” §2244(b)(3)(E). Everyone agrees this provision precludes us from exercising certiorari jurisdiction to review an appellate panel’s decision granting or denying certification to state prisoners. And because §2255(h) tells us that the same rules apply to federal prisoners, exactly the same outcome follows for federal prisoners like Mr. Bowe.¶

Importantly, though, that does not mean we lack for ways to resolve the circuit split my colleagues worry over. AEDPA may prevent us from exercising certiorari jurisdiction under §1254(1). But AEDPA not only leaves untouched our power to entertain original habeas petitions filed directly in this Court. Felker v. Turpin , 518 U. S. 651, 660661 (1996). AEDPA also leaves untouched our power to accept certified “question[s] of law” from the federal courts of appeals and provide them with “binding instructions” on how to answer those questions under §1254(2), a provision found right next door to the one governing our certiorari jurisdiction.¶

Accustomed as we are to proceeding by way of certiorari under §1254(1), it’s true that we have rarely accepted certified questions under §1254(2). But a rare procedure befits rare circumstances. And this case involves a rare circumstance indeed—a circuit split that would otherwise stagger on forever. Rather than ignore §2244’s express constraint on our certiorari jurisdiction, then, we might have simply signaled to the courts of appeals our willingness to accept certification in this or a similar case. Cf. United States v. Rice , 327 U. S. 742, 746–747 (1946) (accepting certification in the uncommon situation of a “circuit court of appeals . . . equally divided on” questions of law).¶

A¶

To appreciate why we lack certiorari jurisdiction, it helps to start with a bird’s eye view of how AEDPA addresses second or successive collateral attacks.¶

AEDPA seeks “to advance the finality of criminal convictions.” Mayle v. Felix , 545 U. S. 644, 662 (2005). It does so, in part, by imposing “stric[t] limit[s]” on the ability of federal and state prisoners to bring multiple postconviction attacks in federal court. See Jones v. Hendrix , 599 U. S. 465, 476 (2023). So, for example, a federal prisoner’s second or successive collateral attack “must be certified . . . to contain” a showing of either new facts or new law. §2255(h). The new-facts route requires the prisoner to show that he possesses newly discovered evidence that, if proved, would establish his innocence clearly and convincingly. See §2255(h)(1). The new-law route requires him to argue that his conviction or sentence is unlawful in light of a new rule of constitutional law that this Court has made retroactively applicable to cases on collateral review. See §2255(h)(2). Similar rules apply to state prisoners under §2244(b)(2). 1¶

Even beyond those constraints, AEDPA imposes others. A federal prisoner’s second or successive §2255 attack not only must “contain” either new facts or new law; it also “must be certified as provided in section 2244.” §2255(h). Everyone agrees this cross-reference directs us to §2244(b)(3). That paragraph provides these instructions:¶

“(A) Before a second or successive application permitted by this section is filed in the district court, the applicant shall move in the appropriate court of appeals for an order authorizing the district court to consider the application.¶

“(B) A motion in the court of appeals . . . shall be determined by a three-judge panel of the court of appeals.¶

“(C) The court of appeals may authorize the filing of a second or successive application only if it determines that the application makes a prima facie showing that the application satisfies the requirements of this subsection.¶

“(D) The court of appeals shall grant or deny the authorization . . . not later than 30 days after the filing of the motion.¶

“(E) The grant or denial of an authorization . . . shall not be appealable and shall not be the subject of a petition for rehearing or for a writ of certiorari.”¶

With this overview in hand, our lack of certiorari jurisdiction quickly becomes clear. Section 2244(b)(3)(E) prohibits a state prisoner from invoking our certiorari jurisdiction to challenge an order denying him leave to file a second or successive collateral attack. Indeed, this Court has already held as much. Felker , 518 U. S., at 658–659. And §2255(h)¶

------¶

1 Put aside for the moment AEDPA’s additional instructions about doover claims—they become relevant only when we come to the merits. See Part II, infra .¶

tells us that a federal prisoner seeking to pursue a second or successive collateral attack must obtain certification “as provided in section 2244” for state prisoners. Accordingly, the same bar on certiorari applies.¶

That commonsense conclusion finds unanimous support in the courts of appeals. Just as §2244(b)(3)(E) bars certiorari petitions, it also bars petitions for panel or en banc rehearing. Since AEDPA’s adoption in 1996, a great many federal prisoners have sought rehearing on the theory that this rule applies only to state prisoners, not them. Yet every circuit to consider that approach has rejected it, recognizing instead that subparagraph (E) applies to federal prisoners no less than their state counterparts, and thus prohibits them from petitioning for panel or en banc rehearing. See, e . g ., In re Clark , 837 F. 3d 1080, 1082–1083 (CA10 2016) (collecting cases); In re Baptiste , 828 F. 3d 1337, 1340 (CA11 2016); Págan-San Miguel v. United States , 736 F. 3d 44, 46, n. 1 (CA1 2013) ( per curiam ); In re Sonshine , 132 F. 3d 1133, 1134 (CA6 1997).¶

Our own precedent carries the same message. In Castro v. United States , 540 U. S. 375 (2003), the government argued that subparagraph (E) precluded this Court from considering a federal prisoner’s challenge to an appellate panel decision denying him leave to file a second or successive attack. At the same time, the government contended, the terms of subparagraph (E) leave it free to challenge many appellate panel decisions granting a federal prisoner permission to proceed. Id ., at 380–381. The Court rejected both submissions. First, after conducting a painstakingly detailed analysis of the filing history of the particular prisoner before it, the Court concluded that his challenge was not in fact second or successive. So subparagraph (E)which applies only to second or successive challengessimply was not implicated in the case at hand. See id ., at 379–380. Second, the Court rejected the government’s “anomal[ou]s” suggestion that subparagraph (E) might allow it, but not a prisoner, to challenge an adverse appellate panel decision regarding a second or successive filing. Id ., at 380–381.¶

Neither holding makes sense unless subparagraph (E) applies to federal prisoners pursuing second or successive collateral attacks. Why take the trouble to show this federal prisoner wasn’t pursuing a second or successive attack subject to subparagraph (E)—and why bother rejecting the government’s lopsided argument about how subparagraph (E) works—if subparagraph (E) has no bearing whatsoever on federal prisoners in the first place? While Castro may not have announced a “holding” that subparagraph (E) applies to federal prisoners pursuing second or successive collateral attacks, ante, at 9, n. 3, all of its work was surely grounded on that premise.¶

Our decision in Hohn v. United States , 524 U. S. 236 (1998), is equally telling. That case concerned 28 U. S. C. §2253(c), which requires a federal prisoner to obtain a certificate of appealability (COA) before he may appeal a district court’s adverse decision in a §2255 proceeding. The jurisdictional question before the Court was whether a court of appeals’ denial of a COA is a “case” within the meaning of §1254(1) and thus reviewable on certiorari. 524 U. S., at 241. Concluding that it is, the Court leaned heavily on a comparison between §2253(c) and §2244(b)(3)(E). The latter provision, the Court said, imposes a “clear limit” on our jurisdiction. Id ., at 250. And, the Court reasoned, the existence of that “clear limit,” coupled with the absence of comparable language in §2253(c), suggested that federal prisoners may seek certiorari to challenge an adverse COA decision. See id. , at 249–250. Much as in Castro , the Court in Hohn may not have expressly reached a holding about §2244(b)(3)(E)’s meaning. But its work only makes sense if that provision “clear[ly]” precludes certiorari jurisdiction for federal prisoners.¶

Really, the idea that §2244(b)(3)(E) applies to federal prisoners and bars them from seeking certiorari has been uncontroversial since AEDPA’s adoption three decades ago. So much so that, until recently, even the author of today’s majority opinion accepted it. In another case involving Mr. Bowe (yes, the same Mr. Bowe), our colleague urged the Court to consider accepting an original habeas petition to resolve the very same do-over circuit split now before us. In re Bowe , 601 U. S. ___ (2024) (SOTOMAYOR, J., statement respecting denial of habeas corpus). Invoking that unusual procedure, our colleague insisted, was necessary precisely because “§2244(b)(3)(E)’s bar on petitioning for review” would preclude Mr. Bowe from invoking our certiorari jurisdiction. Id ., at ___ (slip op., at 2). That was right then, and it is right still.¶

B¶

How does Mr. Bowe respond? He does not dispute that all the rules laid out in §§2244(b)(3)(A) through (E) apply to state prisoners. As a result, he acknowledges, neither state prisoners nor state officials may challenge an appellate panel’s certification decision on certiorari. Accordingly, there are only two ways to resolve circuit splits about §2244 in the state-prisoner context: original habeas writs and certification under §1254(2).¶

Even when it comes to federal prisoners, Mr. Bowe cedes much ground. He admits that §2255(h) subjects federal prisoners to the same rules that apply to state prisoners in §§2244(b)(3)(A), (B), (C), and (D). Brief for Petitioner 32. He concedes, too, that every court of appeals to consider the issue has held that subparagraph (E) also applies to federal prisoners. Tr. of Oral Arg. 15. Still, Mr. Bowe insists, he’s discovered something all those courts (and this one) have missed for the past 30 years: Subparagraph (E)—and it alone—doesn’t apply to federal prisoners.¶

His theory runs this way. Section 2255(h) instructs that a federal prisoner’s second or successive attack “must be certified as provided in section 2244.” That language, Mr. Bowe submits, sweeps in only those parts of §2244 that address how an appellate panel goes about certifying a second or successive filing. And it does not sweep in subparagraph (E) because—unlike all of §2244(b)(3)’s other subparagraphs—it does not “provide” for certification.¶

It may be a wonderfully convenient theory for Mr. Bowe, but no court has adopted it for good reason. Just look again at the block quotation of statutory text above. Each and every one of §2244(b)(3)’s subparagraphs “provide[s]” for how a second or successive collateral attack “must be certified” by a “panel of the appropriate court of appeals.” §2255(h). Subparagraph (A) routes the certification decision to a court of appeals. Subparagraph (B) directs that the decision must be made by a three-judge panel. Subparagraph (C) sets forth the standard the panel must use when deciding whether to certify. Subparagraph (D) specifies the timing for a decision. And subparagraph (E) provides that panel certification is the exclusive avenue for certification decisions (without foreclosing the panel from asking this Court, via §1254(2), for “instructions” on “question[s] of law” along the way). What begins with an application to a three-judge panel under subparagraph (A) ends with its disposition by that same panel under subparagraph (E).¶

Notice, too, the hash Mr. Bowe’s interpretation makes of the rest of the statute. Subparagraph (B) mandates that a certification motion “shall be determined by a three-judge panel of the court of appeals.” What would be the point of that direction if, as Mr. Bowe suggests, a prisoner could petition a court of appeals (en banc) or this Court (on certiorari) to certify a second or successive collateral attack instead? Plainly, subparagraph (E) works in tandem with subparagraph (B) to ensure that the certification decision comes from, and only from, a three-judge panel. Ignoring (E) undermines (B).¶

Ignoring (E) also makes a nonsense of subparagraph (D)’s instruction requiring a panel to grant or deny certification within 30 days. That mandate is perfectly sensible if, as subparagraph (E) provides, the decision of a three-judge panel is final. But it is wholly insensible if, as Mr. Bowe supposes, any prisoner can seek panel rehearing, then en banc rehearing, and certiorari too. The entire point of AEDPA and its certification process was to “reduce delays,” Rhines v. Weber , 544 U. S. 269, 276 (2005) (internal quotation marks omitted), and “advance the finality of criminal convictions,” Mayle , 545 U. S., at 662. Yet on Mr. Bowe’s reading, the certification process alone would typically take not 30 days but many multiples of that. 2¶

C¶

Mr. Bowe’s line of argument, understandably, causes my colleagues some discomfort. Thankfully, they are unwilling to say—as Mr. Bowe does—that every court of appeals to have reached the issue has erred in holding that subparagraph (E) bars federal prisoners from seeking panel or en banc rehearing. Ante, at 16. To avoid calling those sound decisions into question, the majority pivots in a different direction. On its theory, even if subparagraph (E) bars federal prisoners from pursuing petitions for panel and en banc rehearing, the statute does not bar them from seeking certiorari.¶

------¶

2 Mr. Bowe suggests this is no problem because subparagraph (E) allows courts of appeals to order rehearing sua sponte . And having discovered porosity in this otherwise solid subparagraph, Mr. Bowe wonders why we need to worry about finality at all. But even if AEDPA tolerates sua sponte rehearing, that is no threat to finality because sua sponte rehearing is rare, and judges do not call for it needlessly. The same is true of certification under §1254(2), for that matter. These modest, rarely employed, and judicially initiated procedures are hardly camels’ noses poking under AEDPA’s tent.¶

There is an obvious reason Mr. Bowe did not press that eccentric theory before us: It cannot be squared with the statutory text. Section 2255 instructs that a federal prisoner’s second or successive attack “must be certified as provided in section 2244.” And what subparagraph (E) “provide[s]” is a package deal. The statute bars rehearing petitions and certiorari petitions in a single breath: It provides that a panel’s certification decision “shall not be the subject of a petition for rehearing or for a writ of certiorari.” §2244(b)(3)(E). The notion that all of subparagraph (E) might apply to federal prisoners minus the six words “or for a writ of certiorari” is utterly implausible as a textual matter.¶

One might think this problem would cause the majority to hesitate. Instead, my colleagues plow on, conjuring up a clear-statement rule to support their course. Yes, the majority admits, a fair reading of §2255(h)’s cross-reference to §2244 “might include” §2244(b)(3)(E)’s certiorari bar. Ante, at 19. But, my colleagues insist, Congress “must speak clearly” when it wishes to withdraw our certiorari jurisdiction, ante , at 8, and the language Congress employed here is not “clear” enough to be worth following , ante , at 12–14.¶

There’s no avoiding the odd contours of the majority’s clear-statement rule. Apparently, Congress may restrict and channel collateral review in the lower courts, as it has done in subparagraphs (A) through (D). Congress may also bar petitions for panel or en banc rehearing, as it has done in subparagraph (E), cutting off the usual method by which a court of appeals “reconcile[s] its internal difficulties.” Wisniewski v. United States , 353 U. S. 901, 902 (1957) ( per curiam ). Seemingly, Congress can do all that without having to satisfy any clear-statement rule. The only work the majority’s rule performs concerns certiorari petitionsand the only effect that rule has is to pencil out subparagraph (E)’s six words “or for a writ of certiorari” in cases involving federal prisoners. All while leaving those same six words in force when it comes to state prisoners.¶

That gerrymandered rule may suit Mr. Bowe’s present needs and appeal to our vanity. But it is pointless. Even if a clear statement were required here, Congress gave us one. For all the reasons explored above, AEDPA’s certiorari bar imposes, in Hohn ’s words, a “clear limit” on our jurisdiction. See Part I–A, supra . Nor, again, is there any need to pretend otherwise just to address the circuit split Mr. Bowe identifies, for we can just as easily reach it through certification under §1254(2) as we can through certiorari under §1254(1). Indeed, the Eleventh Circuit refused to invoke §1254(2) only because it believed we would deny certification—a belief we could easily dispel in this case simply by saying it isn’t so. See App. 78–79.¶

Worse still, the majority’s bespoke certiorari-for-federalprisoners-clear-statement rule lacks any grounding in law. Just ask yourself: If not from thin air, where could such a rule possibly come from?¶

Perhaps the most obvious candidate would be the Constitution. Sometimes, after all, when faced with two competing statutory interpretations this Court invokes the canon of constitutional avoidance to choose the one that avoids, rather than implicates, a constitutional problem. See, e . g ., Clark v. Martinez , 543 U. S. 371, 381–382 (2005). But the majority does not rest its clear-statement rule on that doctrine. Nor could it. This Court has already expressly rejected the notion that §2244(b)(3)’s limitations violate the Suspension Clause or the Exceptions Clause, the two constitutional provisions most obviously relevant to cases like this one. See Felker , 518 U. S., at 661–664.¶

That should come as no surprise either. The writ of habeas corpus traditionally addressed cases of executive detention without trial. Accordingly, a federal prisoner was ineligible for habeas corpus if he had been convicted by a court of competent jurisdiction. See id. , at 663; Brown v.¶

Davenport , 596 U. S. 118, 128–129 (2022). It was not until the 20th century that this Court began allowing federal prisoners to mount collateral attacks on jurisdictionally sound convictions. Cf. Glasgow v. Moyer , 225 U. S. 420, 429 (1912) (“The principle of the cases is the simple one that if a court has jurisdiction of the case the writ of habeas corpus cannot be employed to re-try the issues, whether of law, constitutional or other, or of fact”). Put simply, for much of our Nation’s history, there was no collateral review of the vast majority of convictions, much less a right to certiorari in such cases.¶

With constitutional avoidance off the table, the majority advances an unbriefed theory to prop up its novel clearstatement rule. Citing Boechler v. Commissioner , 596 U. S. 199 (2022), the majority observes that this Court generally reads statutorily prescribed procedural requirements as forfeitable claim-processing rules rather than unwavering jurisdictional mandates unless Congress “clearly” indicates otherwise. Ante , at 11. And, the majority insists, it is doing something similar here. See ibid . Hardly. No one disputes that AEDPA’s certiorari bar at issue here is jurisdictional. So Boechler not only does not “contro[l] this case,” ante , at 11, n. 5, it has nothing to say about this case at all.¶

The majority’s invocation of Boechler is not just inapt, it is unfortunate. Two years ago, in an obscure administrative law case, this Court expressly refused to treat Boechler as a permission slip for creating any manner of new interpretive presumptions that may happen to suit our tastes but that do not “approximat[e] rea[l]” statutory meaning. Compare Loper Bright Enterprises v. Raimondo , 603 U. S. 369, 455, n. 1 (2024) (KAGAN, J., dissenting) (arguing that, in light of cases like Boechler , we may as well accept other interpretive fictions like Chevron ), with 603 U. S., at 399 (majority opinion) (holding otherwise). Yet that is exactly how the majority treats Boechler today.¶

The picture does not improve from there. The Boechler clear-statement rule operates only in a particular context and exists only for particular reasons. Treating too many rules as jurisdictional rather than as claim-processing rules, this Court has said, would “result in the waste of judicial resources and may unfairly prejudice litigants.” Henderson v. Shinseki , 562 U. S. 428, 434 (2011). Yet neither of those rationales applies here. If anything, it is the majority’s new rule that promises to waste judicial resources by undermining finality. And there is no colorable argument that Congress acts unfairly by preventing both prisoners and governments from appealing adverse certification decisions. Far from respecting Boechler , then, the majority more nearly defies it.¶

With Boechler a dry hole, the majority finds itself forced to prospect elsewhere for legal authority. Its next stop is Castro . Ante, at 8–10. But the majority’s quote mining of that decision only winds up highlighting just how unprecedented its course today really is.¶

In Castro , remember, the government advocated for a lopsided construction of §2244(b)(3)(E). On the government’s view, the statute permitted it to seek certiorari to challenge a circuit decision granting certification, but precluded a federal prisoner from seeking certiorari to challenge a circuit decision denying certification. 540 U. S., at 380. As we have seen, the Court rejected that submission for two reasons. First, the prisoner there did not seek to pursue a second or successive attack subject to subparagraph (E). Second, the government’s reading would have “close[d] our doors to a class of habeas petitioners seeking review” while leaving those same doors open to the government. Id ., at 381. Expanding on the second rationale, the Court said it would not endorse such an “anomal[ous]” reading of subparagraph (E) “without any clear indication that such was Congress’ intent.” Id ., at 380–381. Doubtless, this commentary reflected an intuition that Congress would not lightly treat similarly situated litigants differently. But none of that speaks to the situation before us, where the construction of subparagraph (E) at issue is fair and square: no certiorari review of certification decisions, no matter who won below.¶

If anything, and as we have seen, Castro does more to hurt than help the majority’s cause. See Part I–A, supra . Throughout that decision, the Court took as given that subparagraph (E) does bar federal prisoners from seeking certiorari to challenge adverse certification decisions regarding their second or successive filings. Why (again) would Castro take the trouble to explain that the challenge there wasn’t a second or successive one if subparagraph (E) doesn’t apply to federal prisoners anyway? And why would Castro worry about the anomalous possibility that only the government may seek certiorari when, in fact, federal prisoners can too?¶

The majority offers no colorable answers to these questions because none can be had. Just like Hohn and our colleague’s opinion two years ago in In re Bowe , Castro fully appreciated that subparagraph (E) imposes a “ clear limit ” on our certiorari jurisdiction, one applicable to federal and state prisoners alike. Hohn , 524 U. S., at 250 (emphasis added); see also In re Bowe , 601 U. S., at ___ (slip op., at 2) (acknowledging that “§2244(b)(3)(E)’s bar on petitioning for review” would preclude Mr. Bowe himself from seeking certiorari). 3¶

------¶

3 Attempting to wring more yet from Castro , the majority points to its citation of Utah v. Evans , 536 U. S. 452 (2002), for the “principle that we read limitations on our jurisdiction to review narrowly.” Ante, at 10 (internal quotation marks omitted). But Evans simply applied the anodyne presumption that federal agency action is amenable to some form of judicial review. See 536 U. S., at 463 (citing Webster v. Doe , 486 U. S. 592 (1988), and Bowen v. Michigan Academy of Family Physicians , 476 U. S. 667 (1986)). That presumption speaks not at all to the postconviction setting, let alone second or successive collateral attacks like this one. In this context, a litigant has already enjoyed extensive judicial review,¶

Unable to identify any plausible authority in existing law for its clear-statement rule, the majority is ultimately left to summon it from the ether. To be sure, my colleagues begin by pointing to the (unremarkable) principle that this Court does not read “statutory silence as implicitly modifying or limiting . . . jurisdiction that another statute specifically grants.” Hertz Corp. v. Friend , 559 U. S. 77, 83 (2010); Felker , 518 U. S., at 660–661. But from there, the majority engages in pure invention, devising a broad new legal rule that even express limitations on our certiorari jurisdiction are unenforceable absent a clear statement. See ante, at 8.¶

What a leap that is. This case is nothing like those in which we have historically declined to infer limitations on our certiorari jurisdiction based on statutory silence . All agree that §2255(h) expressly incorporates portions of §2244(b)(3). All agree that §2244(b)(3)(E) expressly bars “petition[s] for . . . writ[s] of certiorari.” The only point of disagreement concerns how those explicit provisions fit together. I believe they fit together in a way that deprives us of jurisdiction. But even if I am wrong about that, this is hardly a case of “silen[t] or ambiguous” text, Hertz , 559 U. S., at 83–84, nor one involving provisions that “mak[e] no mention of our authority,” Felker , 518 U. S., at 661. Truly, the majority’s clear-statement rule comes from nowhere at all.¶

In the end, neither Mr. Bowe’s textual arguments nor the majority’s atextual theory can obscure AEDPA’s message, one every court of appeals (and this Court) has understood since 1996. A litigant like Mr. Bowe may seek to pursue an original habeas petition in this Court. A court of appeals may certify questions to us. But §2244 bars state prisoners (and governments) from invoking our certiorari jurisdiction to challenge adverse certification decisions. Section 2255¶

------¶

including at least two opportunities to seek certiorari from us—first on direct review, and again in a first postconviction challenge.¶

extends that bar to federal prisoners. And because that bar applies here, we lack jurisdiction to decide this case. The majority’s conclusion otherwise is as improbable as it is novel. 4¶

II¶

Because the majority concludes that we have jurisdiction to reach the merits, I offer my views too. The question here, recall, concerns the Eleventh Circuit’s refusal to certify Mr. Bowe’s latest collateral attack. Like most other circuits, the Eleventh Circuit has held that §2244(b)(1) bars do-over claims by both state and federal prisoners. And that provision, the court reasoned, barred Mr. Bowe’s present challenge because he has already pursued the same challenge many times before.¶

------¶

JUSTICE JACKSON’s version of the argument proceeds differently. She admits that a dismissal may have the practical effect of a denial. Ante, at 2 (concurring opinion). But, she insists, that doesn’t hold true here because, by applying §2244(b)(1)’s do-over bar, the Eleventh Circuit applied the wrong “statutory criteria” to Mr. Bowe. Ante , at 4. There is just one problem. The statute bars review of any “grant or denial,” not any “grant or denial made after applying the correct statutory criteria.” JUSTICE JACKSON identifies nothing in the statute’s terms that might allow us to turn the Eleventh Circuit’s (supposed) merits error into a jurisdictionally decisive event. A denial is a denial. Recognizing the flaws of both versions of the argument, the majority rightly adopts neither.¶

4 In a final foray, Mr. Bowe and JUSTICE JACKSON contend that, even if subparagraph (E)’s certiorari bar generally applies to federal prisoners, it still does not apply in this particular case. Mr. Bowe’s version of the argument goes like this. Subparagraph (E) bars certiorari petitions from appellate orders that “grant or den[y]” certification to pursue a second or successive attack. And, Mr. Bowe submits, that language doesn’t apply to him because the Eleventh Circuit “dismissed” rather than “denied” his certification request. See App. 79. This argument fails, though, because “the label attached to an order is not dispositive”; its “ ‘practical effect’ ” is. Abbott v. Perez , 585 U. S. 579, 594 (2018). And here, the practical effect of the Eleventh Circuit’s “dismissal” order was to deny Mr. Bowe permission to pursue his second or successive attack.¶

In reversing the Eleventh Circuit, my colleagues do not question that §2244(b)(1) prohibits state prisoners’ do-over claims. Nor do my colleagues question that Mr. Bowe seeks to pursue a do-over claim in this case. Instead, they hold that federal prisoners like him are free to file in district court as many do-over claims as they please. Once more, I disagree because I see nothing in AEDPA’s text that supports that kind of special treatment for federal prisoners.¶

A¶

Begin by walking through the relevant statutory provisions. Section 2255 requires a federal prisoner to satisfy two essential requirements before he can pursue a second or successive challenge. First, his filing must “contain” either new facts or new law sufficient to satisfy §2255(h)(1) or (h)(2). Second, his filing “must be certified” by a court of appeals under the rules outlined in §2244. And to win certification under §2244, everyone accepts, a federal prisoner must satisfy at least §§2244(b)(3)(A), (B), (C), and (D). Ante , at 12–13. (With jurisdiction now out of the picture, we can set aside the dispute over (E).)¶

From there the right answer lies only a few steps away. Section 2244(b)(3)(C) states that “[t]he court of appeals may authorize the filing of a second or successive application only if it determines that the application makes a prima facie showing that the application satisfies the requirements of this subsection .” (Emphasis added.) No one questions that this provision applies to state and federal prisoners equally. Brief for Petitioner 32; Brief for United States 2526; ante, at 12–13. So by its terms, this provision means a federal prisoner must make “a prima facie showing” in a court of appeals that his second or successive filing satisfies the requirements of “this subsection”—namely, subsection (b) of §2244.¶

That tells us all we need to know. The very first provision of subsection (b) prohibits do-over claims. In the words of¶

§2244(b)(1), any claim presented in a prior application “shall be dismissed.” As a result, AEDPA no more tolerates repetitive litigation by federal prisoners than it does by state prisoners. And the Eleventh Circuit was right to deny Mr. Bowe leave to file again the same claim he had already pursued many times before.¶

What the statutory language teaches on this point, context confirms. Next to subsection (b)(1) sits subsection (b)(2). There, Congress addressed second or successive filings presenting new (as opposed to do-over) claims. As laid out in the margin, subsection (b)(2)’s rules for state prisoners’ new claims are similar to, but different from, §2255(h)’s rules for federal prisoners’ new claims. 5 And because §2255(h) applies particularly to federal prisoners, its specific requirements displace §2244(b)(2)’s more generally applicable ones. See Baltimore Nat. Bank v. State Tax Comm’n of Md. , 297 U. S. 209, 215 (1936). So when a court of appeals asks whether a federal prisoner has made a “prima facie showing” that his proposed filing “satisfies the requirements of this subsection” under §2244(b)(3)(C), the requirements of §§2255(h)(1)–(2) step into the shoes that §2244(b)(2) would otherwise fill. Tellingly, however, §2255(h) contains no such federal-prisoner-specific language when it comes to do-over claims. Instead, it leaves §2244(b)(1)’s generally applicable language in place. All of¶

------¶

“(i) the factual predicate for the claim could not have been discovered previously through the exercise of due diligence; and¶

5 Section 2244(b)(2)(B)’s new-facts provision requires that¶

“(ii) the facts underlying the claim, if proven and viewed in light of the evidence as a whole, would be sufficient to establish by clear and convincing evidence that, but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” In contrast, §2255(h)(1)’s new-facts provision requires¶

“newly discovered evidence that, if proven and viewed in light of the evidence as a whole, would be sufficient to establish by clear and convincing evidence that no reasonable factfinder would have found the movant guilty of the offense.”¶

which suggests that §2244(b)(1) applies no less to federal prisoners than to state prisoners—just as the text says. 6¶

Another contextual clue comes from the 30-day clock. Just as with §2244(b)(3)(C), no one questions that §2244(b)(3)(D) applies to both federal and state prisoners. Under that provision, a federal court of appeals has just 30 days to issue a certification decision. Complying with that tight deadline is possible if courts of appeals can dispose of do-over claims without further inquiry. But complying is a good deal harder if panels must ask whether each and every proposed collateral attack involves new evidence of innocence or a retroactively applicable rule of constitutional law. In this case, for example, it took me five minutes to appreciate that Mr. Bowe has already advanced, many times before, the same claim he proposes to pursue here. I do not know, nor would it be so easy to determine, whether Mr. Bowe’s latest filing makes a compelling showing of new facts or new law. Contra, ante, at 23 (majority erroneously suggesting that determining whether a litigant “has raised a claim before” is no easier than determining whether he has identified compelling new facts or law).¶

Statutory structure and history, too, have something to say. Before AEDPA, separate provisions addressed do-over claims by state prisoners and federal prisoners. Back then, §2244 permitted (but did not require) a court to deny relief to a state prisoner who “present[ed] no new ground not heretofore presented and determined” in a prior¶

------¶

6 The majority claims my understanding of how §2244(b)(2) and §2255(h) interact renders §2255(h)(2) “mere surplusage.” Ante, at 2223. But that claim is easily falsified. If §2255(h)(2) really were “mere surplusage,” then it could be struck out of the statute with no change. So, then, imagine a version of §2255(h) without subsection (h)(2). That imaginary provision would require every “second or successive motion . . . to contain . . . newly discovered evidence” capable of satisfying subsection (h)(1)’s new-facts requirement. That is very different from the real §2255(h), which allows a prisoner to show either new facts, §2255(h)(1), or new law, §2255(h)(2).¶

application. 28 U. S. C. §2244(a) (1988 ed.); 62 Stat. 965966. Similarly, §2255 said that a court “shall not be required to entertain a second or successive motion for similar relief ” by a federal prisoner. See 28 U. S. C. §2255 (1988 ed.). In AEDPA, Congress amended both provisions. Now, §2255(h) channels second or successive §2255 motions through the §2244 rules applicable to state prisoners. And AEDPA added to §2244 a new mandatory (no longer permissive) bar against do-over claims, flatly directing that they “shall be dismissed.” §2244(b)(1); 110 Stat. 12201221.¶

Exactly nothing in this history or AEDPA’s structure suggests Congress sought to loosen do-over restrictions for anyone—let alone completely do away with those restrictions for federal prisoners. To the contrary, the statute’s history and structure suggest, as we have long observed, that AEDPA imposed new and “strict[er] limit[s]” on postconviction relief. Jones , 599 U. S., at 476. All of which indicates, once again, that Congress meant what it said in §2244(b)(3)(C): The “requirements of this subsection” apply to federal and state prisoners alike—and those requirements include subsection (b)(1)’s bar on do-over claims.¶

B¶

Seeking some way to justify a rule prohibiting state prisoners from pursuing do-over claims—but also allowing federal prisoners to do just that—Mr. Bowe and the majority offer a grab bag of unpersuasive theories.¶

First, the majority observes that §2244(a) specifically references “section 2255,” the provision governing federal prisoners’ collateral attacks, while §2244(b)(1) contains no such reference. Because “we presume differences in language . . . convey differences in meaning,” the majority submits, it must follow that §2244(b)(1)’s bar on do-over claims “applies only to state prisoners.” Ante , at 20 (internal quotation marks omitted). But that cannot be right.¶

Section 2244 mentions “section 2255” only one time, in §2244(a). If its absence elsewhere mattered, then none of the rules in §§2244(b)(3)(A) through (D) would apply to federal prisoners. Yet not even the majority or Mr. Bowe thinks that. To the contrary, both concede that the whole point of §2255(h)’s cross-reference to §2244 is to ensure federal prisoners, just like their state counterparts, must follow each of those provisions. Ante, at 12–13; Brief for Petitioner 32.¶

Second, and in a similar vein, Mr. Bowe stresses that subsection (b)(1) prohibits do-over claims “in a second or successive habeas corpus application under section 2254 .” (Emphasis added.) Seeking to capitalize on the italicized language, Mr. Bowe notes that §2254 governs postconviction “applications” by state prisoners, while federal prisoners proceed by way of §2255 “motions.” Given that difference in nomenclature, Mr. Bowe contends, subsection (b)(1) must apply only to state prisoners because that provision addresses “applications” and not “motions.” The majority briefly nods in the same direction. See ante, at 20.¶

This argument fails for much the same reason as its predecessor. If the absence of the word “motion” in §2244 made all the difference, then none of §§2244(b)(3)(A) through (D)’s rules would apply to federal prisoners because each of those provisions speaks only of “applications.” Yet, no one believes that. Instead, everyone appreciates, §2255 routes federal prisoners’ “motions” through all those provisions. In each, then, the word “applications” is best read as interchangeable with “motions,” just as most circuits have understood. Cf. Magwood v. Patterson , 561 U. S. 320, 324, n. 1 (2010) (“Although 28 U. S. C. §2244(b) refers to a habeas ‘application,’ we use the word ‘petition’ interchangeably”).¶

Third, Mr. Bowe replies that there’s something else special about subsection (b)(1). Unlike §§2244(b)(3)(A) through (D), he says, subsection (b)(1) doesn’t just speak of “applications,” but “application[s] under section 2254. ”¶

(Emphasis added.) And that more specific reference, Mr. Bowe insists, means subsection (b)(1) cannot apply to federal prisoners proceeding under §2255. But this argument doesn’t move the needle either. If (as it must be and as no one seriously disputes) the term “application” in §2244 is best read as embracing a federal §2255 “motion,” then the phrase “application under section 2254” in §2244 should likewise be understood to embrace a “motion under section 2255.” 7¶

Fourth, the majority suggests that §2255(h)’s cross-reference to §2244 captures only the latter’s “procedures” governing “how a panel certifies a second or successive filing.” Ante, at 21. And, the argument goes, §2244(b)(1)’s bar on do-over claims is no mere “procedure” but instead a substantive rule about what a filing may “contai[n].” Ibid. This theory fails too. Section 2255’s cross-reference does not speak of a procedure/substance distinction. Instead, §2255(h) directs courts to “certif[y]” motions “as provided in” §2244. Even the majority acknowledges that this language makes §2244(b)(3)(C) applicable to federal prisoners. Ante , at 12–13. And subparagraph (C) explicitly says “the requirements of this subsection”—which include §2244(b)(1)—must be satisfied before certification. It’s more than a little telling that the majority finds itself forced¶

------¶

7 That is not to say that subsection (b)(1)’s specific reference to §2254 applications, rather than to applications simpliciter , is superfluous. It has considerable work to do—just not the work Mr. Bowe supposes. By referencing §2254, the statute makes plain that the do-over bar applies only to applications (or motions, as the case may be) by prisoners seeking postconviction relief. That renders it inapplicable to habeas applications from state prisoners in preconviction executive detention, because those applications are filed under §2241 rather than §2254. And that distinction (unlike the one Mr. Bowe proposes between state and federal prisoners) makes sense. Convicted prisoners, by definition, have already received a great deal of judicial process, so a do-over bar makes sense as to them. Prisoners who haven’t been convicted, meanwhile, stand in very different shoes, so Congress decided to exempt them.¶

both to rewrite §2255(h)’s terms and to ignore §2244(b)(3)(C)’s directions.¶

Finally, Mr. Bowe retreats to the position that federal prisoners must be exempt from subsection (b)(1) because AEDPA was designed to treat state prisoners more harshly than federal prisoners. But, in fact, AEDPA often treats the two groups the same, and not just when it comes to second or successive filings. To take just one example, §2253(c) requires a prisoner to secure a “certificate of appealability” before appealing from a habeas or §2255 loss in federal district court. State and federal prisoners alike thus must satisfy that provision’s demands. See §2253(c); cf. §2266(a) (requiring identical treatment for certain state- and federalprisoner filings). And it should shock no one that Congress might proceed similarly here and exhibit an intolerance for do-over claims no matter who seeks to pursue them. 8¶

*¶

The majority today commits two serious mistakes. First, in its zeal to resolve a circuit split, the majority asserts jurisdiction where we have none. With statutory text and 30 years of case law in this Court and lower courts aligned¶

------¶

8 The government offers two additional theories for Mr. Bowe on the merits. But the majority pursues neither theory, and rightly so. For one, the government appears to suggest that subsection (b)(1) never applies during a court of appeals’ certification proceeding for any prisoner—state or federal. Only after certification, on this view, may a district court weed out do-over claims under subsection (b)(1). See Brief for United States 44–45. But, among other defects, the government’s view would imply that not only (b)(1) but also (b)(2) is inapplicable during certification. And, of course, that would render the whole certification process pointless. For another, the government suggests that, even if §2244(b)(1)’s do-over bar applies to federal prisoners, that rule does not apply in this particular case because Mr. Bowe never actually filed a previous collateral attack containing a Davis claim; he only sought leave to do so. Reply Brief 10–11. But we did not take this case to address that case-specific argument, nor did Mr. Bowe himself clearly advance it before us.¶

against it, the majority bases its claim of jurisdiction on a new and curiously contoured clear-statement rule that is as needless as it is unprecedented. Second, and compounding its jurisdictional overreach, the majority adopts an implausible view on the merits. AEDPA didn’t suddenly free federal prisoners from pre-existing restrictions on do-over claims. Just the opposite: It imposed additional “carefully crafted limits” on collateral attacks by federal and state prisoners alike. Jones , 599 U. S., at 482. And, quite unsurprisingly, one of those limits includes an unmistakable and absolute bar on do-over claims. Because neither of the majority’s innovations are compatible with Congress’s directions in AEDPA, I respectfully dissent.¶