DIANE P. WOOD, Circuit Judge.
Anthony Gray-Bey filed a petition for a writ of habeas corpus in the United States District Court for the Eastern District of Arkansas. The district court construed the petition as a successive petition filed without proper authorization from- the court of appeals and transferred the motion to this court to be considered as an application for leave to file a successive motion under 28 U.S.C. § 2255. In an Order dated October 7, 1999, this court dismissed the application without prejudice under Circuit Rule 22.2. Gray-Bey has filed another application with additional supplemental materials.
The governing statute calls for this court to act on Gray-Bey’s application within 30 days of its filing, in this case January 7, 2000. See 28 U.S.C. § 2244(b)(3). The initial question before us is whether this court has the power under any circumstances to extend the time for final disposition of the application. We agree with our sister circuits that such power exists, and that the 30-day period may be extended for those few cases which require reasoned adjudication and cannot be resolved within the statutory period. See, e.g., United States v. Barrett, 178 F.3d 34, 42 n. 2 (1st Cir.1999) (stating 30-day time limit for court of appeals granting or denying authorization to file second or successive habeas corpus petition is “precatory, not mandatory”), quoting Rodriguez v. Superintendent, Bay State Correctional Ctr., 139 F.3d 270, 272 (1st Cir. 1998); In re Siggers, 132 F.3d 333, 336 (6th Cir.1997) (reading the language of *868§ 2244(b)(3) as “hortatory or advisory rather than mandatory”); Galtieri v. United States, 128 F.3d 33, 36-37 (2d Cir.1997) (ruling that § 2244(b)(3) must be applied with “flexibility” and concluding that courts should not forego “reasoned adjudication” in the small number of cases that cannot be resolved within 30 days); In re Vial, 115 F.3d 1192, 1194 n. 3 (4th Cir. 1997) (noting that the court exceeded the 30-day limit but concluding that the importance of the issue justified the delay). In a small number of extraordinary cases, the courts cannot perform their assigned judicial function under the Constitution without a more thorough exploration of the legal arguments than is possible in the statutory period. The alternative — uninformed and arbitrary grants or denials of applications — is unacceptable in a system that strives always to operate under the rule of law.
Our dissenting colleague believes that such arbitrariness has been commanded by Congress, because it used the word “shall” in 28 U.S.C. § 2244(b)(3). With ah due respect, we believe that this reading fails to take into account the inherent equitable powers of the federal courts — powers that have been recognized by our sister circuits in their own acknowledgments both that the 30-day rule applies in the overwhelming majority of cases, but that the court retains the power to override it when compelling circumstances demand that action. This is not the extraordinary and lawless conclusion that our colleague claims it is. To the contrary, it reflects a reconciliation between the commands of legislation and the exigencies of judicial decisionmaking
that is well grounded in the law. For example, the Supreme Court has consistently taken the approach we and our sister circuits have adopted for § 2244(b)(3) in the cases establishing the abstention doctrine. As is the case with § 2244(b)(3), the statute in which Congress confers general federal jurisdiction on the lower federal courts is worded in mandatory language. See 28 U.S.C. § 1331 (“The district courts shall have original jurisdiction of all civil actions arising under the Constitution, laws, or treaties of the United States.”)(emphasis added). That phrasing might seem to leave the federal courts no choice but to decide questions within their jurisdiction. Nevertheless, the Court has recognized that in certain cases it is wiser to decline deciding the merits of the case, either for a period of time or altogether. See Railroad Comm’n v. Pullman Co., 312 U.S. 496, 500, 61 S.Ct. 643, 85 L.Ed. 971 (1941) (“The resources of equity are equal to an adjustment that will avoid the waste of a tentative decision as well as the friction of a premature constitutional adjudication.”); Burford v. Sun Oil Co., 319 U.S. 315, 318, 63 S.Ct. 1098, 87 L.Ed. 1424 (1943) (considering abstention to be “a matter of equitable discretion”); Younger v. Harris, 401 U.S. 37, 44, 91 S.Ct. 746, 27 L.Ed.2d 669 (1971) (invoking comity as a reason to restrain federal courts acting inequity from enjoining most pending state criminal proceedings); Pennzoil Co. v. Texaco Inc., 481 U.S. 1, 10, 107 S.Ct. 1519, 95 L.Ed.2d 1 (1987) (requiring federal court abstention in civil proceedings to avoid intrusion into the Texas judicial sys*869tem). Moreover, the use of “shall” in the Constitution is not always or necessarily understood to be mandatory. For example, Article III, § 2 states that “[t]he judicial Power shall extend” to various things, including federal questions and controversies between citizens of different states (emphasis added). However, federal jurisdiction does not extend to the full limits of the grant in Article III, despite the potential mandate of the word “shall.” Instead, Congress has imposed an amount in controversy requirement which limits the number of cases that can be heard in diversity jurisdiction, see 28 U.S.C. § 1332. And federal question jurisdiction as we know it today was not granted until 1875 with the passage of the Judiciary Act. See Act of March 3,1875, 18 Stat. 470.
This is not the only situation in which a reading of a statute in isolation from the rest of the law might lead one to think, erroneously, that either the statute or the Constitution itself has been violated. To take but two examples, Atchison, Topeka & Santa Fe Ry. Co. v. Wichita Bd. of Trade, 412 U.S. 800, 819-20, 93 S.Ct. 2367, 37 L.Ed.2d 350 (1973) (plurality opinion), recognized a limited power in the courts to suspend rates pending review of a final order of the Interstate Commerce Commission, in part because there was no “provision in the relevant statutes depriving federal courts of their general equitable power to preserve the status quo to avoid irreparable harm pending review.” Similarly, in the Regional Rail Reorganization Act Cases, 419 U.S. 102, 128-29, 95 S.Ct. 335, 42 L.Ed.2d 320 (1974), and Dames & Moore v. Regan, 453 U.S. 654, 689, 101 S.Ct. 2972, 69 L.Ed.2d 918 (1981), the Court saved the constitutionality of statutes (the International Emergency Economic Powers Act, 91 Stat. 1626, 50 U.S.C. §§ 1701-06 (1976 ed. Supp. Ill) in Dames & Moore and the Regional Rail Reorganization Act Cases, 45 U.S.C. § 701 et seq. (1970 ed. Supp. Ill)) against takings challenges by noting that Congress had not taken the independent step of repealing the Tucker Act, 28 U.S.C. § 1491. The Court felt free to do so even though the laws in question made no mention of the Tucker Act. Just so here: Congress imposed the 30-day time limit in § 2244(b)(3), but it did not repeal the All Writs Act, 28 U.S.C. § 1651; Congress thus recognized that the courts retain the power to take extraordinary steps when they are needed. We note too that well established canons of statutory construction support this position. Repeals by implication are disfavored, see TVA v. Hill, 437 U.S. 153, 190, 98 S.Ct. 2279, 57 L.Ed.2d 117 (1978); see also Argentine Republic v. Amerada Hess Shipping Corp., 488 U.S. 428, 429, 109 S.Ct. 683, 102 L.Ed.2d 818 (1989), and so there is every reason to assume that the background laws conferring powers on the federal courts remain in full force. Furthermore, courts must if they can interpret statutes to avoid constitutional problems. See, e.g., NLRB v. Catholic Bishop of Chicago, 440 U.S. 490, 506-07, 99 S.Ct. 1313, 59 L.Ed.2d 533 (1979) (“[I]n the absence of a clear expression of Congress’ intent to bring teachers in church-operated schools within the jurisdiction of the Board, we decline to construe the [National Labor Relations Act] in a manner that could in turn call upon the Court to resolve difficult and sensitive questions arising out of the guarantees of the First Amendment religion clauses.”). While, as the dissent point's out, this court concluded that such a saving interpretation was not possible in French v. Duckworth, 178 F.3d 437 (7th Cir.), cert. granted, — U.S.-, 120 S.Ct. 578, 145 L.Ed.2d 481 (1999), the administrative time limit at issue here is quite different from the substantive command to terminate injunctions that we considered in French. We see no reason to deviate from the unanimous conclusion of our sister circuits and assume a rigidity in the 30-day time period that Congress may not have intended.
Indeed, there is good reason to conclude that Congress affirmatively recognized that extraordinary action may be neces*870sary in certain cases, which can be found in another of the statutes governing habe-as corpus. Our dissenting colleague is correct that § 2244(b)(3) is seemingly clear in its mandate; however, he does not adequately take into account the implications of § 2266, which also provides for limitations periods for the federal courts to make decisions on applications and motions for relief in capital cases. 28 U.S.C. § 2266. Section 2266 provides that a court may consider several factors in deciding whether to delay the disposition of an application for a writ of habeas corpus, including, “Whether the case is so unusual or so complex, due to the number of defendants, the nature of the prosecution, or the existence of novel questions of fact or law, that it is unreasonable to expect adequate briefing within the time limitations established by subparagraph(A).” 28 U.S.C. §' 2266(a)(l)(C)(ii)(II). Here, our decision to extend the 30-day time limitation is based precisely on these considerations. Our dissenting colleague correctly points out that the regime for capital cases is spelled out in greater detail than is the procedure for successive petitions. We, however, draw no negative inference from the lack of symmetry between the two statutes regarding the question of permissible delays. What § 2266 demonstrates is that Congress was not willing to risk a miscarriage of justice in order to achieve timeliness at all costs. Congress inserted the time limitations to avoid delays due to perceived judicial foot dragging and “general congestion of the court’s calendar,” 28 U.S.C. § 2266(a)(l)(C)(iii), not to mandate hasty and unreasoned judicial decision making. The fact that Congress omitted detailed language regarding permissible delays in § 2244(b)(3) does not indicate that it threw all caution to the wind in every set of circumstances falling outside § 2266.
In our view, Gray-Bey’s application presents several legal issues which have yet to be resolved by this circuit. As these issues are important and recurring, we have concluded that the issues presented in this case should not be decided without the benefit of full briefing and adversarial presentation. We recognize that Congress reformed habeas corpus in order to ensure that cases moved swiftly toward resolution without unnecessary delay. However, to meet Congress’s goal, it is imperative that the courts develop clear procedures that will apply to all of these cases. To do so takes time. Briefly delaying Gray-Bey’s case — and the few others presenting similarly complicated and knotty questions— will allow us to ensure that the rules governing habeas develop properly and will facilitate future expeditious treatment of these cases. Taking the extra time to allow for full presentation of the issues will make the eventual decision more helpful to future litigants, the district courts, and future panels of this court. Handling all of these cases on an expedited basis risks leaving everyone involved in the dark.
As for the merits of the petition, while we certainly respect the views that our dissenting colleague has offered, the depth of discussion his treatment required simply underscores the fact that these are serious points that deserve an open, adversary presentation. We therefore decline the implicit invitation either to agree or disagree with his conclusions or the route he uses to reach them until after both sides have been provided the opportunity to brief their positions fully and present them to this panel at oral argument. We hereby order the Clerk of the Court to appoint counsel to represent petitioner Gray-Bey and instruct counsel to address the following issues in their briefs, in addition to any other points that require attention in counsel’s professional judgment:
1. Given that Gray-Bey raised his claim under Bailey v. United States, 516 U.S. 137, 116 S.Ct. 501, 133 L.Ed.2d 472 (1995), in his prior § 2255 motion before this court but not before the district court, has his Bailey claim already been “presented” for purposes of § 2244(b)(1)?
*8712. Does 18 U.S.C. § 924(c) prohibit the conduct for which Gray-Bey was convicted? If so,does Gray-Bey’s Bailey claim rest on a new rule of constitutional law, thereby qualifying Gray-Bey for relief under § 2244(b)(2)?
3. What is the effect of the Arkansas District Court’s decision refusing to consider Grey-Bey’s petition for relief under § 2241?
4. Does this court’s decision in In re Davenport, 147 F.3d 605 (7th Cir. 1998), require Gray-Bey to file his petition under § 2241, and would such a § 2241 petition be properly construed as nonsuccessive?
Recognizing our duty to respect the statutory command for speed, we also instruct the Clerk’s office to expedite this case as follows. Oral argument shall be set for the week of February 21, 2000. Petitioner’s brief shall be due on January 25, 2000, and respondent’s answer on February 11, 2000. Petitioner’s reply shall be due one week later, on February 18, 2000.
EASTERBROOK, Circuit Judge,
dissenting.
Anthony Gray-Bey is serving a sentence of 256 months’ incarceration for conspiring to distribute cocaine, possessing cocaine with intent to distribute, using a telephone to facilitate his drug business, and using a firearm during and in relation to drug trafficking. His convictions and sentences have been affirmed, see United States v. Goines, 988 F.2d 750 (7th Cir.1993), and a collateral attack under 28 U.S.C. § 2255 was unsuccessful. Gray-Bey v. United States, 156 F.3d 733 (7th Cir.1998). On December 8, 1999, Gray-Bey filed an application for leave to commence a second proceeding for collateral relief. Congress has set a deadline for action: “The court of appeals shall grant or deny the authorization to file a second or successive application not later than 30 days after the filing of the motion.” 28 U.S.C. § 2244(b)(3)(D), applied to federal prisoners by 28 U.S.C. § 2255 ¶ 8. That time expires today, January 7, 2000. But my colleagues do not “grant or deny” the requested authorization. Instead they issue an order directing the Clerk to appoint counsel for Gray-Bey and specifying four issues that briefs should address. We lack authority to depart from an Act of Congress, so I record my disagreement.
Section 2244(b)(3)(D) is explicit: “The court of appeals shall grant or deny the authorization to file a second or successive application not later than 30 days after the filing of the motion.” “Shall,” not “should” or “endeavor to” or “make progress toward”; this statute does not leave wriggle room. The court shall grant or deny the application within 30 days. In this construction “shall” means “must”. Context may show that a particular “shall” is synonymous with “may,” see Gutierrez de Martinez v. Lamagno, 515 U.S. 417, 432 n.9, 115 S.Ct. 2227, 132 L.Ed.2d 375 (1995), but replacing “shall” with “may” in § 2244(b)(3)(D) makes the passage vapid. In its context, the “shall” of § 2244(b)(3)(D) is a mandate to judges, not a grant of permission or a first-party expression of futurity (“I shall go to the opera tonight”) that is subject to alteration. How could Congress have been clearer? If “shall grant or deny ... not later than 30 days” is not mandatory, what language could be mandatory? Would it help if the word “must” had been used instead? That word has fewer senses, see Bryan A. Garner, A Dictionary of Modem Legal Usage 939-42 (2d ed.1995), but it is not statutory ambiguity that undergirds the majority’s disposition.
Other circuits, whose decisions the majority cites, don’t fret about shadings among “must,” “shall,” and “should.” They simply balk at deadlines. In re Vial, 115 F.3d 1192 (4th Cir.1997) (en banc), the first of the decisions to disregard § 2244(b)(3)(D), offers this defense, which I quote in full: “[W]e exceeded the 30-day time limitation established by 28 U.S.C. § 2244(b)(3)(D) for decisions on requests for permission to institute a second or *872successive § 2255 proceeding. We are convinced, however, that the importance of the issue presented justified the delay.” 115 F.3d at 1194 n. 3. This “explanation” is shocking; the court ignores the text and asserts a right to violate a statute. The Constitution adopts a different hierarchy; statutes are superior to judges’ views about wise policy. See Bank of Nova Scotia v. United States, 487 U.S. 250, 255, 108 S.Ct. 2369,101 L.Ed.2d 228 (1988); United States v. Payner, 447 U.S. 727, 736-37,100 S.Ct. 2439, 65 L.Ed.2d 468 (1980). The second circuit takes the same general approach as the fourth, with a more extended effort to justify elevating judicial views over legislative ones. Galtieri v. United States, 128 F.3d 33, 36-37 (2d Cir.1997). Like the fourth circuit, Galtieri confesses that it is not interpreting § 2244(b)(3)(D); the court instead claims a common-law power to disregard it in pursuit of better adjudication. Yet whether lengthier adjudication is “better” is the very question to which § 2244(b)(3)(D) speaks. Like many recent decisions of the Supreme Court, § 2244(b)(3)(D) instantiates the principle that dispatch has value. Both Congress and the Justices want to reduce the period during which the validity of a conviction is open to question. Galtieri observes that more time leads to fewer substantive errors, which may well be true; but how to reconcile the quest for accuracy with other competing objectives is a legislative task, traditionally implemented through devices such as statutes of limitations and outer periods for action. Congress has concluded that, once a prisoner has had a direct appeal and one full collateral attack, further proceedings must be abbreviated. That decision is entitled to our respect and obedience, whether we approve it or not.
Only the sixth circuit has offered a reason compatible with legislative supremacy under the Constitution. In re Siggers, 132 F.3d 333, 336 (6th Cir.1997), holds that § 2244(b)(3)(D) is unenforceable because it does not specify the consequence of delay. The court believed that statutes are directory rather than mandatory when no consequence is attached to disobedience, and if this is so then the 30 days indeed may be exceeded — for the meaning of all legal texts depends on a background of interpretive principles. Congress could have said something like “if 30 days pass without action, then the application is denied automatically.” Because it did not do this, Siggers holds, a court may exceed the time limit with impunity. The first circuit agrees with Siggers, see Rodriguez v. Superintendent, 139 F.3d 270, 272 (1st Cir. 1998), reiterated in dictum in United States v. Barrett, 178 F.3d 34, 42 n. 2 (1st Cir.1999), but I do not, because I do not think that “no stated consequence means non-mandatory” is a background norm of interpretation in federal law.
Siggers’ approach implements Holmes’s bad-man theory that law’s meaning lies in the penalties for noncompliance. Holmes’s approach is a useful heuristic, but much of our law is based on a contrary premise: that rules are effective, and must be implemented in good faith, even if there is no stated penalty. See Kurowski v. Krajewski 848 F.2d 767, 774-75 (7th Cir.1988). The first amendment says “Congress shall make no law ... abridging the freedom of speech” (there’s that word again), and Art. I, § 2, cl. 3 specifies that a census “shall b’e made within three Years after the first Meeting of the Congress of the United States, and within every subsequent Term of ten Years” (a “shall” plus a deadline); no one believes that the Constitution’s failure to spell out what happens if Congress contravenes these rules means that Congress is entitled to contravene them! The United States “shall” publish “from time to time” a statement of accounts. Art. I, § 9, cl. 7. When holding that no one has standing to obtain judicial review of a claim that the United States has violated this rule, the Supreme Court emphasized that political actors nonetheless are obliged to follow it. United States v. Richardson, 418 U.S. 166, 179, 94 S.Ct. 2940, 41 L.Ed.2d 678 (1974). Claims under the Guaranty Clause, Art. IV § 4, are not justiciable, but *873the lack of enforcement does not imply that the political branches may ignore their obligations. Surely courts would not say that, because no statute attaches a penalty to judges’ failure to apply correct rules of law on collateral review, judges are free to disregard the entire Antiterrorism and Effective Death Penalty Act (of which § 2244 is a part). Yet that proposition is logically identical to Siggers’ treatment of § 2244(b)(3)(D).
The omission of a consequence from § 2244(b)(3)(D) means only that the courts must select a consequence in common-law fashion. Perhaps the right remedy is mandamus by a higher court. Perhaps it is the dismissal of the application, after the fashion of Carlisle v. United States, 517 U.S. 416, 116 S.Ct. 1460, 134 L.Ed.2d 613 (1996); United States v. Addonizio, 442 U.S. 178, 99 S.Ct. 2235, 60 L.Ed.2d 805 (1979); United States v. Kimberlin, 776 F.2d 1344 (7th Cir.1985); and Gaertner v. United States, 763 F.2d 787 (7th Cir. 1985) — though I could understand a holding that delay does not deprive the court of authority to grant the application eventually. Cf. Brock v. Pierce County, 476 U.S. 253, 106 S.Ct. 1834, 90 L.Ed.2d 248 (1986) (observing that the remedy for delay in administrative law need not be the elimination of the agency’s power to act, but not questioning the proposition that a statutory deadline is nonetheless binding). Perhaps the best response to the short time for action is a generous attitude toward the uncertainty that abbreviated de-cisionmaking produces (applications should be granted in close cases). Perhaps, to the contrary, courts should take a stingy attitude: a request for leave to conduct a third round of litigation could be denied unless within 30 days the court comes to believe that it has fair chance of success. All of these are possible, and judges reasonably could disagree about which to choose. But a deadline without an explicit penalty for delay is still a deadline, which must be followed, however difficult it may be to decide how to proceed.
Last year we encountered another statute, effective two days after § 2244, specifying a time limit. One part of the Prison Litigation Reform Act provides that prospective relief concerning the operation of prisons is automatically stayed 30 days after a party moves to modify the decree. 18 U.S.C. § 3626(e)(2). The sixth circuit held in Hadix v. Johnson, 144 F.3d 925, 944-46 (6th Cir.1998), that § 3626(e)(2) is subject to equitable alteration by federal courts. Addressing the same issue, we disagreed with Hadix, concluding instead that the language of § 3626(e)(2) is too clear to permit common-law modifications. French v. Duckworth, 178 F.3d 437, 441-43 (7th Cir.1999), cert. granted, — U.S. -, 120 S.Ct. 578, 145 L.Ed.2d 481 (1999). That seems to me equally true of § 2244(b)(3)(D). The panel in French went on to hold that § 3626(e)(2) is unconstitutional, a conclusion with which I disagreed, see 178 F.3d at 448-53 (dissenting from the denial of rehearing en banc). Right or wrong substantively, French shows the proper way to reason about a time limit for judicial action. We took § 3626(e)(2) seriously, even though that meant its doom.
Our failure to act on Gray-Bey’s application is more regrettable than the declaration of unconstitutionality in French, for the Supreme Court will have the last word on § 3626(e)(2). But “[t]he grant or denial of an authorization by a court of appeals to file a second or successive application shall not be appealable and shall not be the subject of a petition for rehearing or for a writ of certiorari.” 28 U.S.C. § 2244(b)(3)(E). Perhaps this language, unlike § 2244(b)(3)(D), does leave wriggle room: the “grant or denial” is not reviewable, but a refusal to issue a timely decision may be reviewed by mandamus or a petition for certiorari before judgment. Perhaps, even though the “grant or denial of an authorization” is not reviewable, any relief provided by the district court ultimately is subject to review of antecedent issues, including the question whether au*874thorization should have been granted. Perhaps, too, the Supreme Court’s decision in French may cast light on the propriety of my colleagues’ action. As a practical matter, however, a court’s failure to implement § 2244(b)(3)(D) in a given case is conclusive: the time it would take the Solicitor General or comparable state official to file a petition for mandamus, and the Supreme Court to issue a writ, may exceed the additional time before the court issues a decision.
Although the majority cites Vial, Sig-gers, Galtieñ, and Rodñguez, it does not endorse their reasoning (as opposed to their'results). Instead it offers a broader proposition, which amounts to the conclusion that Congress just can’t set deadlines for litigation — not because the Constitution liberates judges from time limits (the conclusion of French) but because judges frequently exercise equitable powers concerning many statutes that include the word “shall.” I grant that occasionally the word “shall” is non-mandatory when read in context, but I do not think that cases such as Younger v. Harris, 401 U.S. 37, 91 S.Ct. 746, 27 L.Edüd 669 (1971), and Burford v. Sun Oil Co., 319 U.S. 315, 63 S.Ct. 1098, 87 L.Ed. 1424 (1943), which deal with comity among judicial systems, offer much help on the question whether a statute saying that a court “shall” do something “not later than 30 days” permits the court to take more than 30 days. Time limits traditionally have been strictly enforced in federal law. E.g., Browder v. Director, Department of Corrections, 434 U.S. 257, 98 S.Ct. 556, 54 L.Edüd 521 (1978); United States v. Locke, 471 U.S. 84, 105 S.Ct. 1785, 85 L.Ed.2d 64 (1985); Lampf, Pleva, Lipkind, Prupis & Petigrow u Gilbertson, 501 U.S. 350, 111 S.Ct. 2773, 115 L.Edüd 321 (1991); Carlisle; Addonizio.
Congress did not repeal the All Writs Act, 28 U.S.C. § 1651, when enacting § 2244(b)(3)(D), a fact that my colleagues deem significant, but no one thinks that § 1651 is a license to ignore statutory rules. Pennsylvania Bureau of Correction v. United States Marshals Service, 474 U.S. 34, 106 S.Ct. 355, 88 L.Ed.2d 189 (1985), holds that § 1651 does not permit a court to disregard 28 U.S.C. § 2243. I cannot imagine why judges would have more power to modify § 2244 than they do to modify § 2243. Section 2266, another part of the aedpa, also is illuminating: that statute allows judges to extend deadlines when certain conditions are satisfied; § 2244(b)(3)(D) does not. Ordinary principles of interpretation lead to the conclusion that the circumstances justifying an extension under § 2266 do not justify extra time under § 2244(b). Otherwise what’s the point of the differences between the two statutes?
In the end, the majority’s approach rests on the proposition that federal judges have discretion to depart from federal statutes for good reasons — and that judges, rather than the political branches, define which reasons are “good.” Few propositions could be more subversive of the rule of law. Pennsylvania Bureau of Correction is one among many cases denying that federal courts have any such power. The equitable-discretion approach taken in Radix (and by the Solicitor General in French) is far more circumscribed and cannot be reworked to fit § 2244(b)(3)(D). It is that courts may issue injunctions preserving the status quo unless Congress cancels that authority expressly. Although § 3626(e)(2) says that the existing injunction is automatically stayed once 30 days have run, it does not preclude judges from issuing another, identical injunction under their general equitable powers pending further litigation. That seems to me too much an evasion of § 3626(e)(2) to be sound. (Likewise, the maneuver in Triestman v. United States, 124 F.3d 361, 366-67 (2d Cir.1997), of denying the application pro forma and immediately granting rehearing in order to afford more time for decision is a transparent evasion of § 2244(b)(3)(D) because the court’s order does not reflect an actual decision as opposed to a bookkeeping entry. Galtieñ *875has the virtue of candor, and the second circuit apparently ceased using Triest-man’s approach after deciding Galtieri.)
Right or wrong, reliance on equitable discretion to preserve the status quo offers no support for my colleagues’ action today. There is no longstanding “general equitable power” to authorize applications for second or successive collateral attacks— the norm is that one is sufficient — and there is no status quo to “preserve” by equitable relief; § 2244(b) is a novelty that must be understood on its own terms. Moreover, the Solicitor General has stressed in French that equitable discretion to extend the 30-day period depends on a strong substantive claim for ongoing relief. My colleagues do not conclude that Gray-Bey has a strong claim that is likely to prevail on further consideration. His claim is weak and readily may be decided now.
His initial problem — one that does not make my colleagues’ list of four issues to be briefed — is time. Section 2255 ¶ 6 provides:
A 1-year period of limitation shall apply to a motion under this section. The limitation period shall run from the latest of—
(1) the date on which the judgment of conviction becomes final;
(2) the date on which the impediment to making a motion created by governmental action in violation of the Constitution or laws of the United States is removed, if the movant was prevented from making a motion by such governmental action;
(3) the date on which the right asserted was initially recognized by the Supreme Court, if that right has been newly recognized by the Supreme Court and made retroactively applicable to cases on collateral review; or
(4) the date on which the facts supporting the claim or claims presented could have been discovered through the exercise of due diligence.
Only subsection (3) might authorize Gray-Bey’s application in December 1999, more than six years after his conviction was affirmed (and more than three years after Congress added the one-year period of limitations to the law of collateral attack). But the “right” that Gray-Bey wants to assert “was initially recognized by the Supreme Court” on December 6, 1995, in Bailey v. United States, 516 U.S. 137, 116 S.Ct. 501, 133 L.Ed.2d 472 (1995), more than four years before Gray-Bey filed his application. Far from being unavailable to Gray-Bey until now, Bailey came in time for Gray-Bey to present it on his first collateral attack. Gray-Bey lost not because Bailey is prospective only, but because he neglected to make the argument in the district court even though certiorari had been granted in Bailey on April 17, 1995, and the issue had been percolating for years. 156 F.3d at 742-MS.
Gray-Bey now points to Bousley v. United States, 523 U.S. 614, 118 S.Ct. 1604, 140 L.Ed.2d 828 (1998), which held that implementation of 18 U.S.C. § 924(c)(1) (the statute construed in Bailey) can lead to constitutional problems cognizable on collateral attack. Improper advice to a defendant pleading guilty could make the plea unintelligent, Bousley held; similarly, a trial record that lacks evidence adequate to establish all elements of the offense (as Bailey defines them) could support collateral relief to avoid imprisoning an innocent person. But these are not new principles of constitutional law. Davis v. United States, 417 U.S. 333, 94 S.Ct. 2298, 41 L.Ed.2d 109 (1974), establishes that actual innocence justifies collateral relief under § 2255, cf. Jackson v. Virginia, 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979), and the rule that involuntary guilty pleas may be set aside is even older. More than a year before Bousley, we applied these principles in collateral attacks based on Bailey. E.g., Stanback v. United States, 113 F.3d 651 (7th Cir.1997). Gray-Bey’s first collateral attack failed solely because he should have *876raised the issue sooner; that conclusion is incompatible with his current submission that 1997 or 1998 was too soon to invoke Bailey. Nothing that happened in 1998 restarts the time for a collateral attack. The initial principle of Bailey, which is the crucial time under § 2255 ¶ 6, dates from 1995, and the retroactivity rules predate Bousley.
The questions that my colleagues flag for counsel’s attention also have straightforward answers. Logically the initial question (though it is No. 4 in the majority’s list) is whether a § 2255 petition is necessary, or whether instead Gray-Bey may use 28 U.S.C. § 2241, to which the time limits and prior-authorization rules do not apply. The answer must be that § 2255 is the required route. Section 2241 is unavailable “unless it also appears that the remedy by motion is inadequate or ineffective to test the legality of his detention.” 28 U.S.C. § 2255 ¶ 5. Judicial emphasis must be on “test”: a § 2255 motion is not “inadequate or ineffective” merely because the petitioner loses. Nor do the changes made by the aedpa, which limit the number of § 2255 motions (and the time to file them) render § 2255 inadequate or ineffective to test the lawfulness of detention. No one is entitled to more than one collateral attack.
In re Davenport, 147 F.3d 605, 609 (7th Cir.1998), requires us to ask whether Gray-Bey’s initial § 2255 motion gave him “a reasonable opportunity to obtain a reliable judicial determination of the fundamental legality of his conviction and sentence.” In Davenport this court held that when the basis of the claim has not been legally established at the time of the first petition, the prisoner has not received a reasonable opportunity. The first petition filed by Sherman Nichols (one of two petitioners in Davenport) was resolved before the Supreme Court issued Bailey, and the law in this circuit was contrary to the way Bailey interpreted the word “use” in § 924(c)(1). Nichols therefore could not have obtained relief by a § 2255 petition. Gray-Bey, however, is differently situated.
Gray-Bey filed his first § 2255 petition before Bailey, but final decision came later. Had Gray-Bey properly preserved an argument about the meaning of “use” in § 924(c)(1), it would have been decided by this court, under correct principles of law, on his first § 2255 petition. Had we erred, the Supreme Court could have corrected us by certiorari. The reason Gray-Bey did not receive a decision was his own default. It is impossible to say that “the remedy by motion [under § 2255] is inadequate or ineffective to test the legality of his detention” when the only shortcoming is the petitioner’s omission of an issue. That omission is not a problem in the remedy; it is a problem in litigation strategy. It is, indeed, a fatal problem, for when Bousley held that Bailey can support collateral review, the Justices stressed that the legal arguments must have been presented to the district court — and they added that the-weight of existing precedent against a position does not justify omission. Bousley, 523 U.S. at 622-23, 118 S.Ct. 1604.
Next in logical sequence, if Gray-Bey’s application were timely, would be the question whether a successive petition may be justified under § 2255 ¶ 8(2) on the ground that Bailey is “a new rule of constitutional'law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable.” Bousley, which was decided while Gráy-Bey’s prior appeal was pending, holds that Bailey claims maybe raised on collateral review, and that Bailey may be applied retroactively to at least some cases (though perhaps Bousley himself was not a beneficiary, having forfeited his claim). There is a nice question about the scope of § 2255 ¶ 8(2): Does the successive petition have to assert reliance on a “new rule,” or is an old rule newly made retroactive enough? From Gray-Bey’s perspective, however, Bailey is not a “new rule”, as I have already observed; he argued it in 1997 *877when briefing his appeal. Had we rejected that claim on the ground that Bailey does not apply retroactively, then it would be tempting to say that Bousley reopens the door. But because Gray-Bey lost on forfeiture rather than the merits, one would think that the kind of new rule adequate to start a new period must be one that retroactively changes the law of forfeiture, rather than a decision making explicit that Bailey “applies.”
We need not wrestle this one to the ground, however, because Bousley does not make retroactive a new rule “of constitutional law”. Bailey resolved an issue of statutory interpretation, and Bousley observed that the proper understanding of § 924(c)(1) may lead to collateral relief to the extent that a statutory error gives rise to a constitutional flaw. In Bousley itself the potential flaw was an involuntary guilty plea (made involuntary by the fact that the defendant did not know the elements of the offense to which he was pleading). This is related to the approach our court earlier adopted for Bailey issues in Stanback. Similarly a statutory problem could give rise to a constitutional one if the record did not contain enough evidence to permit a rational trier of fact to find guilt beyond a reasonable doubt. As Bousley observed, however, defendants trying to take advantage of that possibility have a heavy burden, for they must show not only that the record fails to establish “use” but also that they did not “carry” the weapon, the other way to commit the crime. 523 U.S. at 624, 118 S.Ct. 1604. See also Muscarello v. United States, 524 U.S. 125, 118 S.Ct. 1911, 141 L.Ed.2d 111 (1998) (a drug dealer who does not actively “use” a gun in that crime nonetheless “carries” it if the weapon is moved about, even in the trunk of a car). We held accordingly in Buggs v. United States, 153 F.3d 439, 443-45 (7th Cir.1998), that deficiency of evidence can convert a statutory Bailey argument into a constitutional one, but only if the record really is deficient, and then only if the point has been properly preserved. (Note 4 of Buggs adds that Holm v. United States, 524 U.S. 236, 118 S.Ct. 1969, 141 L.Ed.2d 242 (1998), contributes nothing to the analysis of this question. The only question before the Court was whether a request for a certificate of appealability is a “case in” a court of appeals for certiorari purposes; the Court said nothing about the merits of the claim and instructed the eighth circuit to review them in the first instance in light of the Solicitor General’s position — which was functionally the same as the position we adopted in Buggs.)
Does Bousley make retroactive a “new rule of constitutional law” that the evidence must b sufficient to support the charge? Not at all; that rule has been around since Davis and Jackson. Gray-Bey’s second principal contention, that the jury instructions in his case did not anticipate Bailey, is not even a constitutional argument, for the reasons given in Young v. United States, 124 F.3d 794, 798-99 (7th Cir.1997). See also, e.g., Gilmore v. Taylor, 508 U.S. 333, 342, 113 S.Ct. 2112, 124 L.Ed.2d 306 (1993) (an error in stating the elements of the offense in jury instructions is not a constitutional defect). I therefore conclude that a successive petition cannot be authorized under § 2255 ¶ 8. Gray-Bey must live with the forfeiture in his prior petition, just as our published opinion concludes.