dissenting from the denial of rehearing en banc:
I would hold that section 104 of the Antiterrorism and Effective Death Penalty Act (“AEDPA”),1 Pub.L. No. 104-132, 110 Stat. 1214 (codified in relevant part at 28 U.S.C. § 2254(d)(1)), violates the separation of powers doctrine and is unconstitutional. Section 2254(d)(1) constitutes a severe congressional incursion on the federal “judicial power,” which Article III of the Constitution vests wholly and exclusively in the federal courts. It does so in two principal ways: first, by prohibiting the federal courts from applying the ordinary *1262principles of stare decisis in deciding habe-as cases involving prisoners held in state custody, thereby interfering with the federal courts’ normal adjudicatory process; and second, by requiring federal courts to give effect to incorrect state rulings that, in the federal courts’ independent judgment, violate the Constitution. Such a congressional breach of the federal judiciary’s integrity and independence, of its duty to maintain the supremacy of the Constitution, and, indeed, of the constitutional structure itself, should not go unchecked by this court. For this reason, and because I believe that this is the type of case an en banc court should hear,2 I dissent from the court’s unfortunate decision to let the panel decision become the law of the circuit.3
I.
Section 2254(d)(1) prescribes an unconstitutional standard of review for federal courts’ adjudication of habeas petitions alleging that the petitioner’s state custody is “in violation of the Constitution or laws or treaties of the United States.” 28 U.S.C. § 2254(a). It provides:
(d) An application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted with respect to any claim that was adjudicated on the merits in State court proceedings unless the adjudication of the claim—
(1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States;
§ 2254(d)(1).
The statute departs from pre-AEDPA habeas jurisprudence in two key ways. First, whereas previously federal courts could rely on the jurisprudence of the courts of appeals to determine whether the state court decision was in violation of federal law, after AEDPA they are limited to clearly established law “as determined by the Supreme Court.” Id.; see Williams v. Taylor, 529 U.S. 362, 412, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (“[Section] 2254(d)(1) restricts the source of clearly established law to this Court’s jurisprudence.”).4 Second, federal courts *1263may no longer grant habeas petitioners relief in cases in which the state court judgment clearly violated federal law unless the state court’s erroneous ruling was also “objectively unreasonable.” Williams, 529 U.S. at 409, 120 S.Ct. 1495; see also id. at 410, 120 S.Ct. 1495 (“[A]n unreasonable application of federal law is different from an incorrect application of federal law.”) (emphasis in original); id. at 411, 120 S.Ct. 1495 (noting that, prior to AEDPA, “a state court’s incorrect legal determination ha[d] [never] been allowed to stand because it was reasonable. We have always held that federal courts, even on habeas, have an independent obligation to say what the law is.” (quoting Wright v. West, 505 U.S. 277, 305, 112 S.Ct. 2482, 120 L.Ed.2d 225 (1992) (O’Connor, J., concurring))) (emphasis and alteration in original) (internal quotation marks omitted). These changes impose a severe restriction on ha-beas petitioners’ ability to secure federal relief from state detention that violates the Constitution. More important for our purposes, they represent a fundamental breach of the separation of powers — an unconstitutional intrusion by Congress into the federal judiciary’s independent and exclusive duty to “say what the law is.” Marburg v. Madison, 5 U.S. (1 Cranch) 137, 177, 2 L.Ed. 60 (1803).
II.
“[T]he doctrine of separation of powers ... is at the heart of our Constitution.” Buckley v. Valeo, 424 U.S. 1, 119, 96 S.Ct. 612, 46 L.Ed.2d 659 (1976). The Framers considered the division of governmental power into separate departments “a vital check against tyranny,” id. at 121, 96 S.Ct. 612, a “self-executing safeguard against the encroachment or aggrandizement of one branch at the expense of the other,” id. at 122, 96 S.Ct. 612. While the Constitution contains checks and balances to temper all three branches, the Framers were particularly wary of the tendency of the legislature to usurp the other branches’ power — especially that of the judiciary. See Met. Wash. Airports Auth. v. Citizens for the Abatement of Aircraft Noise, Inc., 501 U.S. 252, 273-74, 111 S.Ct. 2298, 115 L.Ed.2d 236 (1991) (“[I]t is against the enterprising ambition of this department, that the people ought to indulge all their jealousy and exhaust all their precautions .... Its constitutional powers being at once more extensive and less susceptible of precise limits, it can with the greater facility, mask under complicated and indirect measures, the encroachments which it makes on the co-ordinate departments.” (quoting The Federalist No. 18, at 332-34 (J. Cooke ed.1961)) (internal quotation mark omitted; alteration in original)); see also Plant v. Spendthrift Farm, Inc., 514 U.S. 211, 221, 115 S.Ct. 1447, 131 L.Ed.2d 328 (1995) (“Th[e] sense of a sharp necessity to separate the legislative from the judicial power[ ] ... triumphed among the Framers of the new Federal Constitution.”). As such, the Constitution vests “the judicial power” fully and unconditionally in the judicial department. U.S. Const. Art. Ill § 1.
Over the course of our nation’s history, the Supreme Court has developed the contours of this “judicial power,” setting the boundaries between Congress and the judiciary essential to maintaining Article Ill’s “vital check against tyranny.” In Marburg v. Madison, the Court forcefully described the primary function of the federal judiciary: “It is emphatically the *1264province and duty of the judicial department to say what the law is.” 5 U.S. (1 Cranch) at 177 (emphasis added). That is, the federal courts are charged with interpreting the Constitution and ensuring that the statutory law is consistent with it — a determination that is, in the words of Chief Justice John Marshall, the “very essence of judicial duty.” Id. at 178.
Because the duty to “say what the law is” is vested entirely and exclusively in the judicial branch, the Court has made plain that Congress may not interfere with the federal courts’ independent process of adjudication and interpretation.5 In United States v. Klein, 80 U.S. (13 Wall.) 128, 20 L.Ed. 519 (1871), the Court struck down a federal statute that required it to interpret a presidential pardon as conclusive evidence that the claimant had aided the South during the Civil War and was therefore not entitled to recover for property confiscated by the government during that war. The Court held that this statute “passed the limit which separates the legislative from the judicial power,” id. at 147, by prescribing a “rule of decision,” id. at 146, which limited the Court’s independent adjudication. In doing so, the Court acknowledged that Congress could have denied it jurisdiction over the appeal entirely, pursuant to its authority to make “exceptions” and “regulations” to the Court’s appellate jurisdiction. Id. at 145; see also U.S. Const. Art. Ill § 2. Once it had granted jurisdiction, however, Congress could not then dictate to the Court how to exercise its “judicial power.” Klein, 80 U.S. (13 Wall.) at 147.
This crucial distinction — between withholding jurisdiction altogether and granting jurisdiction but restricting its full and independent exercise — has been reiterated time and again since Klein. See, e.g., Gutierrez de Martinez v. Lamagno, 515 U.S. 417, 430, 115 S.Ct. 2227, 132 L.Ed.2d 375 (1995) (“Congress may be free to establish a compensation scheme that operates without court participation.... But that is a matter quite different from instructing a court automatically to enter a judgment pursuant to a decision the court has no authority to evaluate.”); Yakus v. United States, 321 U.S. 414, 468, 64 S.Ct. 660, 88 L.Ed. 834 (1944) (Rutledge, J., dissenting) (“It is one thing for Congress to withhold jurisdiction. It is entirely another to confer it and direct that it be exercised in a manner inconsistent with constitutional requirements .... [Wlhenever the judicial power is called into play, it is responsible *1265directly to the fundamental law and no other authority can intervene to force or authorize the judicial body to disregard it.”); Henry P. Monaghan, Marbury and the Administrative State, 83 Colum. L.Rev. 1, 11 (1983) (“There is no halfway position in constitutional cases; so long as it is directed to decide the case, an Article III court cannot be ‘jurisdictionally’ shut off from full consideration of the substantive constitutional issues.... ”). Professors James Liebman and William Ryan have described the distinction as one between the “quantitative” power to confer jurisdiction and the “qualitative” power to determine how courts may exercise the “judicial power.” James S. Liebman & William F. Ryan, “Some Effectual Power”: The Quantity and Quality of Decision-making that Article III and the Supremacy Clause Demand of the Federal Courts, 98 Colum. L.Rev. 696, 773-75 (1998). The Court has consistently permitted Congress to exercise its quantitative power over the federal courts’ jurisdiction, including in the habeas context. See, e.g., Felker v. Turpin, 518 U.S. 651, 664, 116 S.Ct. 2333, 135 L.Ed.2d 827 (1996) (upholding AEDPA’s provision restricting the ability of federal courts on habeas to consider successive petitions). It has never, however, allowed Congress to tinker with the qualitative aspects of the adjudicative process. See Wright, 505 U.S. at 305, 112 S.Ct. 2482 (O’Connor, J., concurring) (“[A] state court’s incorrect legal determination has [njever been allowed to stand because it was reasonable. We have always held that federal courts, even on habeas, have an independent obligation to say what the law is.”).
A necessary implication of the constitutional prohibition on congressional tinkering with the manner of judicial decision-making is that Congress may not eliminate wholesale any tool of judicial reasoning from the judicial toolbox. Courts in common law systems have developed a panoply of tools to guide them in the interpretive process, among them inferring rules from text or structure, reasoning from analogy, and applying rules from precedent. Cf Philip Bobbitt, Constitutional Interpretation 12-22 (1991). No one tool takes precedence in any situation, and in any given case several of the tools may work in tension with one another. A critical aspect of judicial reasoning is thus determining which tool most appropriately governs a particular case — in other words, which tool affords the most effective process of judicial analysis. Such a determination is inherently bound up with the power to “say what the law is” and cannot, therefore, be assumed by any person or institution outside Article III.
One tool in the judicial toolbox — stare decisis — bears particular Article III significance. The Supreme Court has held that Congress may neither require federal courts to reopen final judgments, Plaut, 514 U.S. 211, 115 S.Ct. 1447, 131 L.Ed.2d 328, nor substitute its own interpretation of a constitutional right for that already articulated by the Supreme Court, City of Boerne v. Flores, 521 U.S. 507, 117 S.Ct. 2157, 138 L.Ed.2d 624 (1997). These cases together “suggest a vision of Article III courts that requires Congress not only to leave their final judgments intact, but also to treat (and allow other courts to treat) their decisions as having ordinary stare decisis effect.” Vicki Jackson, Introduction: Congressional Control of Jurisdiction and the Future of the Federal Courts — Opposition, Agreement, and Hierarchy, 86 Geo. L.J. 2445, 2469-70 (1998) (emphasis added). The central importance of stare decisis to the Article III “judicial power” is well-founded; as the Supreme Court has explained, stare decisis is “the means by which we ensure that the law will not merely change erratically, but will develop in a principled and intelligible fashion,” a process that “permits society to *1266presume that bedrock principles are founded in the law rather than in the proclivities of individuals, and thereby contributes to the integrity of our constitutional system of government, both in appearance and in fact.” Vasquez v. Hillery, 474 U.S. 254, 265-66, 106 S.Ct. 617, 88 L.Ed.2d 598 (1986). Because the “Court’s power lies ... in its legitimacy,” Planned Parenthood of S. Pa. v. Casey, 505 U.S. 833, 865, 112 S.Ct. 2791, 120 L.Ed.2d 674 (1992), the legitimating function of stare decisis makes it an especially critical tool of judicial reasoning, warranting even greater insulation from congressional control.6
In addition to prescribing the federal courts’ exclusive and independent authority to “say what the law is,” Marbury established another fundamental function of the Article III “judicial power”: the duty to maintain the supremacy of federal law and, above all, the Constitution. See Liebman & Ryan, supra, at 770-71. In Mar-bury, the Court was asked to apply a federal statute that it believed, in its independent judgment, violated the Constitution — it gave the Court original jurisdiction over a class of cases not among those explicitly enumerated in Article III. In holding the statute unconstitutional, Chief Justice Marshall made absolutely plain that the courts may not be forced to give effect to a law that conflicts with the Constitution. He queried, “If an act of the legislature, repugnant to the constitution, is void, does it, notwithstanding its invalidity, bind the courts, and oblige them to give it effect?” Marbury, 5 U.S. (1 Cranch) at 177. Marshall’s answer could not have been clearer — to require the federal courts to give effect to an unconstitutional law would be “an absurdity too gross to be insisted on.” Id.
The Court extended this supremacy-maintaining principle to the federal courts’ review of state court decisions in Martin v. Hunter’s Lessee, 14 U.S. (1 Wheat.) 304, 4 L.Ed. 97 (1816), and Cohens v. Virginia, 19 U.S. (6 Wheat.) 264, 5 L.Ed. 257 (1821). The Court explained in those cases that, although the Constitution itself did not grant the Supreme Court appellate jurisdiction over state court proceedings, once Congress had done so via the Judiciary Act of 1789, the Court had the authority and the responsibility to ensure that state court rulings conformed to the dictates of the Constitution. Once a federal court was charged with reviewing state court judgments, it could not, consistent with its duty to maintain the supremacy of the Constitution, uphold state court rulings that it concluded violated that supreme law. The federal courts’ duty to “say what the law is,” in other words, entailed a corresponding duty not to effectuate any law, whether made by Congress or by state courts, that ran counter to their own independent interpretation of the Constitution’s strictures.
*1267In sum, the essential functions of the Article III “judicial power,” as delineated by the Supreme Court over the past two centuries, are as follows: (1) The federal judiciary has the ultimate authority to interpret the Constitution. (2) Congress need not grant the federal courts jurisdiction over any particular class of cases, but once it has done so, it cannot interfere with the manner in which the federal courts interpret law and decide cases. (3) The federal judiciary is duty-bound to maintain the supremacy of the Constitution, and thus a federal court cannot be required to give effect to any law — be it a federal statute or a state court decision— that, in the court’s independent judgment, violates the Constitution.
III.
Section 2254(d)(1) violates these fundamental attributes of the “judicial power,” granted exclusively to the federal judiciary in Article III, in two principal ways. First, in dictating that federal courts may grant habeas relief only in cases in which the state proceeding violated clearly established law “as determined by the Supreme Court,” § 2254(d)(1) prescribes a “rule of decision” — i.e., a limitation on the federal courts’ independent adjudicatory process— akin to the statute invalidated in Klein. The rule of decision prescribed in AEDPA, however, is more egregious than that of Klein because it effectively suspends the ordinary doctrine of stare decisis in the federal courts’ adjudication of habeas cases. Second, § 2254(d)(1) violates Article III by requiring federal courts to defer to, and thereby give effect to, decisions of state courts that the federal courts interpret as violating the Constitution. Both congressionally imposed rules pose a serious threat to the integrity of our nation’s separation of powers; neither can be tolerated.
A.
Section 2254(d)(l)’s restriction of habeas relief only to prisoners whose state proceedings violated clearly established law that was “determined by the Supreme Court” prevents federal courts from exercising their independent judgment to “say what the law is.” It is, for that reason alone, unconstitutional. Congress might be able, if it so chose, to deny at least the lower federal courts jurisdiction over habe-as corpus cases arising from state convictions. U.S. Const. Art. Ill § 2; but see Jordan Steiker, Incorporating the Suspension Clause: Is There a Constitutional Right to Federal Habeas Corpus for State Prisoners?, 92 Mich. L.Rev. 862, 911 (1994) (arguing that “the Suspension Clause, viewed through the lens of the Fourteenth Amendment, affords state prisoners a constitutional right to federal review of constitutional claims in the lower federal courts”). Once it has granted such jurisdiction, however — as it has via § 2254(a) — Congress may not, consistent with the constitutional separation of powers, dictate to the federal courts how to adjudicate those cases. Section 2254(d)(1) does exactly that by requiring federal courts to ignore not only the precedent of other circuits, which may be persuasive to the case at hand, but also the normally binding precedent of their own circuits in determining whether a habeas petitioner is “in custody in violation of the Constitution or laws or treaties of the United States.” § 2254(a). In other words, § 2254(d)(1) prohibits the application of the normal rules of stare decisis, a central tool of judicial decision-making, by federal courts exercising habeas jurisdiction.7
*1268Not only does this suspension of the usual principles of stare decisis violate the clear dictate of Marbury and Klein that Congress not tinker with the qualitative aspects of Article III adjudication, it turns the notion of judicial hierarchy on its head. Under § 2254(d)(1), federal district courts are required to ignore the clear and binding precedents of the circuit courts above them if those precedents have not also been clearly established by the Supreme Court. This is true even where the circuit precedent is longstanding, and even where every other circuit has established the identical interpretation of the constitutional principle. Such a situation demonstrates the practical absurdity of § 2254(d)(l)’s limitation of precedent to Supreme Court holdings. The Supreme Court’s appellate review is discretionary; its purpose is to clarify the law and to resolve conflicts among the circuit courts. Therefore, the Supreme Court is unlikely to grant review to establish a constitutional principle that is already firmly ingrained, or upon which the lower courts agree. The perverse result is that some of the most universal constitutional principles might never become “clearly established law, as determined by the Supreme Court,” and thus their violations may never warrant habeas relief under § 2254(d)(1).8 Congress’s grant of habeas jurisdiction to the federal courts in such cases is nothing but a hollow promise.
B.
Section 2254(d)(1) violates Article III yet further by requiring the federal courts to give effect to state court rulings that in the federal courts’ independent judgment violate the Constitution. It does so in two ways. First, as explained above, a federal court may not grant habeas relief under AEDPA unless the Supreme Court has clearly articulated the governing constitutional rule. Thus, a federal court is required under AEDPA to uphold a state court proceeding that it determines violated the Constitution if that determination is based upon circuit, and not Supreme Court, precedent.
Second, even where the federal court’s determination of a constitutional violation is based upon clearly established Supreme Court precedent, the court is required, nevertheless, to uphold the state court’s unlawful action if the state court’s ruling was an “objectively reasonable” application of the Supreme Court’s constitutional interpretation. In other words, federal courts must give effect to state courts’ incorrect applications of federal constitutional law, as long as those applications are not “objectively unreasonable,” in other words, if a state court judge could reasonably have made the constitutional error. This mandate, which directs the federal courts to ignore constitutional violations simply because a reasonable state court judge could have reached the erroneous conclusion involved, flies in the face of Marbury and later cases establishing the federal courts’ crucial duty to maintain the supremacy of the Constitution above all other law.
IV.
The Crater panel attempts to sidestep these arguments by characterizing § 2254(d)(1) as merely a “choice of law rule,” Crater, 491 F.3d at 1127 (quoting Green v. French, 143 F.3d 865, 874 (4th Cir.1998), cert. denied 525 U.S. 1090, 119 *1269S.Ct. 844, 142 L.Ed.2d 698 (1999), overruled on other grounds by Williams, 529 U.S. 362, 120 S.Ct. 1495, 146 L.Ed.2d 389), which “does not restrict the federal courts’ power to interpret the law, but only sets standards for what state court errors of law require federal habeas relief,” Crater, 491 F.3d at 1128. Certainly it is within Congress’s power to restrict the federal courts’ ability to grant relief, the panel reasons; such a provision is no different from, say, a statute of limitations or a rule of procedure — one of Congress’s undisputed “quantitative” controls over habeas jurisdiction. This reframing of § 2254(d)(1) — as affecting merely the court’s ability to grant a remedy, and not its power to determine the underlying right — is the reddest of herrings. The panel’s characterization misapprehends the very nature of the writ of habeas corpus, the right to which is inextricably bound with — indeed, synonymous with — its remedy.
When an individual in state custody seeks a writ of habeas corpus, the essence of his claim is not only that his constitutional (or other federal) rights have been violated. It is that his custody is in violation of those rights. See § 2254(a); Preiser v. Rodriguez, 411 U.S. 475, 484, 93 S.Ct. 1827, 36 L.Ed.2d 439 (1973) (“[T]he essence of habeas corpus is an attack by a person in custody upon the legality of that custody....”). The Supreme Court has unequivocally recognized the significance of this fact in holding that “when a state prisoner is challenging the very fact or duration of his physical imprisonment,” he may not bring his challenge under 42 U.S.C. § 1983, the statute that provides a cause of action for constitutional violations, but instead must file a petition for habeas corpus. Preiser, 411 U.S. at 500, 93 S.Ct. 1827. Moreover, the Court has made clear that this distinction is not based on the remedy sought. Thus, even when the prisoner seeks only damages for his constitutional violations, and not release from prison, he nevertheless may not bring the claim under § 1983 if a judgment in his favor “would necessarily imply the invalidity of his conviction or sentence.” Heck v. Humphrey, 512 U.S. 477, 487, 114 S.Ct. 2364, 129 L.Ed.2d 383 (1994).
The right at issue in habeas cases, in other words, is not the same right as that at issue in § 1983 cases. In § 1983 cases, it is the constitutional right itself that forms the basis of the claim. In habeas cases, in contrast, it is the right not to be held in custody in violation of the Constitution that lies at the core of the suit. This is no mere difference in semantics. Where a petitioner asserts the violation of a constitutional right — indeed, even one so fundamental as the right to be free from unreasonable searches and seizures — but the violation did not contribute to the custody of the petitioner, the Court has held that such a claim is not cognizable on habeas. See Stone v. Powell, 428 U.S. 465, 494, 96 S.Ct. 3037, 49 L.Ed.2d 1067 (1976) (holding that Fourth Amendment claims are not cognizable on habeas review because the Fourth Amendment exclusionary rule does not relate to the accuracy of the fact-finding process). When the custody itself is unlawful, however, the right to habeas applies. The panel’s characterization of § 2254(d)(1) as nothing but a “restriction on the scope of the ha-beas remedy,” Crater, 491 F.3d at 1127 (quoting Green, 143 F.3d at 875), is incorrect. Whereas a § 1983 suit challenges a past wrong — the constitutional violation— a habeas petition challenges the present detention that is based upon that past constitutional violation. Thus, when Congress directs a federal court to deny relief even when the court determines that the petitioner’s custody violates the Constitution or another federal law, it requires the federal court to be complicit in the ongoing violation of an individual’s right not to be “unlawfully subjected to physical re*1270straint.” Preiser, 411 U.S. at 486, 93 S.Ct. 1827.9 Put differently, § 2254(d)(1) mandates the “involuntary participation of the federal judiciary” in the perpetuation of the breach of habeas petitioners’ rights not to be detained in violation of the Constitution. Lawrence G. Sager, Klein’s First Principle: A Proposed Solution, 86 Geo. L.J. 2525, 2535 (1998). Separation of powers surely demands that Congress not so conscript the Third Branch.
Needless to say, § 2254(d)(1) is not, as the panel asserts, a mere “choice of law rule.” Crater, 491 F.3d at 1127 (quoting Green, 143 F.3d at 874). The law that Congress has “chosen” is clear in the very fact of its grant of habeas jurisdiction to the federal courts: the writ may be granted only to persons “in custody in violation of the Constitution or laws or treaties of the United States.” 28 U.S.C. § 2254(a) (emphasis added). AEDPA’s demand that federal courts disregard the full corpus of constitutional jurisprudence — -including both the precedents normally binding on them through stare deci-sis and the Constitution itself when the state courts got it wrong but their error was not unreasonable — and give effect to state court adjudications that, in the federal courts’ independent determination, violate the Constitution, makes a mockery of the careful boundaries between Congress and the courts that our Constitution’s Framers believed so essential to the prevention of tyranny.
V.
It is our constitutional duty as Article III judges to “say what the law is” and to maintain the supremacy of our founding document. These duties demand that we declare AEDPA to be what it is — an unconstitutional infringement on the “judicial power,” which Article III grants entirely and exclusively to the federal courts. Crater v. Galaza does not live up to this fundamental responsibility and should not stand as this court’s statement on AED-PA’s constitutionality. For this reason, I respectfully dissent from this court’s denial of rehearing en banc.
. The statute's imposing title is somewhat of a misnomer. The provision held constitutional by the panel — section 2254(d)(1), the centerpiece of the statute's modification of federal habeas practice — has nothing to do with anti-terrorism and little to do with the death penalty. Rather, § 2254(d)(1) restricts the rights of all habeas petitioners detained in state custody, including those, as in this case, who have neither been sentenced to death nor convicted of an act of terrorism. The title was, however, politically appealing in the wake of the bombing of the Oklahoma federal building, on which event President Clinton relied as justification for the bill of which § 2254(d)(1) was a part. President Clinton’s Statement on Antiterrorism Bill Signing, 1996 WL 203049, *1 (Apr. 26, 1996).
. See Fed. R.App. P. 35(a)(2) (stating that an en banc rehearing is warranted if “the proceeding involves a question of exceptional importance”).
. The Supreme Court "has not squarely addressed [§ 2254(d)(l)]'s constitutional validity.” Crater v. Galaza, 491 F.3d 1119, 1129 (9th Cir.2007); see also Randy Hertz & James S. Liebman, Federal Habeas Corpus Practice and Procedure § 32.5 (5th ed.2001)[hereinafter "Hertz & Liebman”] (noting that the issue of § 2254(d)(l)'s constitutionality "was presented in the certiorari petition in Williams v. Taylor, but the Supreme Court denied certio-rari on that question, and neither of the two majority opinions addressed it”) (footnotes omitted). Four Justices, however, have recognized the constitutional problems with the provision and would interpret it in a manner that avoids those problems. See Williams v. Taylor, 529 U.S. 362, 375-79, 386-87, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000) (Stevens, J., concurring).
.Moreover, in addition to the requirement on the face of the statute that the law be clearly established "by the Supreme Court,” the Court has construed AEDPA as departing from its previous jurisprudence on this point in two other ways. First, whereas under Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989), a constitutional principle was considered “old” (i.e., clearly established) if it was recognized prior to the petitioner exhausting his direct appeals, under AEDPA a principle is clearly established only if it was recognized by the Supreme Court at the time of the petitioner’s conviction. See Williams, 529 U.S. at 412, 120 S.Ct. 1495; see also Hertz & Liebman, supra note 3, § 32.3. Second, the Court has interpreted the "clearly established" clause as referring "to the holdings, as opposed to the dicta,” of its decisions. Williams, 529 U.S. at 412, 120 S.Ct. 1495. *1263"In this respect, the ‘clearly established Federal law’ phrase bears only a slight connection to [the Court’s] Teague jurisprudence.” Id. Oddly, the following sentence of Justice O’Connor's opinion appears to contradict or at least raise questions as to the pronouncement just quoted. Whatever the rule may be, however, it appears that § 2254(d)(1) significantly narrows the Court’s previous standard.
. Article III does not, of course, prohibit the federal courts themselves from developing rules governing the manner in which they interpret law and adjudicate cases. The Supreme Court can, and has, prescribed rules binding federal courts to a particular method of decisionmaking. See, e.g., Saucier v. Katz, 533 U.S. 194, 200-01, 121 S.Ct. 2151, 150 L.Ed.2d 272 (2001) (holding that, in determining whether a state officer is entitled to qualified immunity for § 1983 purposes, courts may not consider whether the constitutional right was clearly established before determining first that a constitutional right was violated). This right of the federal courts to make rules governing their adjudicatory process is subject to one limitation: the courts may not restrict the exercise of the "judicial power” so greatly that they effectively abdicate their own jurisdiction. See New Orleans Pub. Serv., Inc. v. Council of the City of New Orleans, 491 U.S. 350, 358, 109 S.Ct. 2506, 105 L.Ed.2d 298 (1989) ("[T]he courts of the United States are bound to proceed to judgment and to afford redress to suitors before them in every case to which their jurisdiction extends. They cannot abdicate their authority or duty in any case in favor of another jurisdiction.” (quoting Chicot County v. Sherwood, 148 U.S. 529, 534, 13 S.Ct. 695, 37 L.Ed. 546 (1893) (citations and internal quotation marks omitted))); Colorado River Water Conservation Dist. v. United States, 424 U.S. 800, 817, 96 S.Ct. 1236, 47 L.Ed.2d 483 (1976) (noting that the federal courts have a "virtually unflagging obligation ... to exercise the jurisdiction given them”).
. The Crater panel quotes Casey for the proposition that "the rule of stare decisis is not an 'inexorable command,’ and certainly it is not such in every constitutional case.” Crater, 491 F.3d at 1126 n. 8 (quoting Casey, 505 U.S. at 854, 112 S.Ct. 2791) (internal quotation marks omitted). The panel's reliance on Casey is fundamentally misguided. It is one thing to say, as Casey does, that federal courts need not apply stare decisis under every circumstance. This statement is consistent with, and is indeed a natural inference from, the principle, described above, that a necessary aspect of judicial reasoning is determining which adjudicative tool is most appropriate in a particular case. It is quite another thing, however, to say that Congress may prohibit the application of the ordinary rules of stare decisis in all federal habeas cases. Such a statement cannot be reconciled with the constitutional requirement, elaborated in Mar-bury, that Article III courts remain free to use the full panoply of judicial tools to arrive at an independent adjudication of each case in which they are granted jurisdiction.
. I do not comment on the right of the federal courts themselves, and particularly the Supreme Court, to modify the rules of stare decisis. That is an entirely different matter that is in no way affected by the separation of powers principle we are considering here.
. Moreover, even if the Court were to refer to such longstanding, universal constitutional principles in an opinion, a mere reference would not constitute "clearly established law” under AEDPA. The Court has made clear that this term refers only to the holdings, and not to the dicta, of the Court. See Lockyer v. Andrade, 538 U.S. 63, 71, 123 S.Ct. 1166, 155 L.Ed.2d 144 (2003); Williams, 529 U.S. at 412, 120 S.Ct. 1495.
. For this reason, the argument that is often made (although not by the Crater panel)- — that § 2254(d)(1) is no different from Teague, in that both prohibit the federal courts from remedying a clear constitutional violation — is faulty on its face. Although Teague did prohibit federal courts from granting habeas relief for newly recognized constitutional violations, the justification most often offered for that result is that the Teague retroactivity rule did not require federal courts to be complicit in any ongoing violation of a habeas petitioner’s rights — because a petitioner’s custody is, by definition, not "unlawful” if no constitutional rule was recognized when that punishment was rendered. Section 2254(d)(1), on the other hand, requires federal courts to give effect to state court judgments rendered in violation of constitutional principles already established when the judgments were handed down, unless those principles were established by the Supreme Court and, even when they were, unless the state court’s application of them, though incorrect, was "objectively reasonable.” In that way, § 2254(d)(1), unlike Teague, violates Marbury's clear dictate that federal courts, charged with maintaining the supremacy of the Constitution, must not be required to give effect to laws — statutory or otherwise — that conflict with their independent interpretation of the Constitution's mandates.
Of course, there is another fundamental difference between § 2254(d)(1) and Teague. Teague is a judicial doctrine imposed on the federal judiciary by the highest court. AED-PA, in contrast, is a congressional mandate that dictates to the federal courts how to adjudicate cases on habeas review. Whatever constraints Teague placed on the federal courts are therefore of no relevance to the question of the constitutionality of § 2254(d)(1).