Dissenting:
I respectfully dissent from the majority’s summary refusal to grant a stay of execution and to address Jose Ceja’s claim that to execute him now, after 23 years of incarceration, would violate the Eighth Amendment’s prohibition against the infliction of cruel and unusual punishment. Accordingly, I would grant a certificate of appealability, see 28 U.S.C. § 2253(c)(1) (West 1997), and address the merits of his claims.
I speak first to the circumstances of the man whose claims the majority declines to review. Ceja has been incarcerated on death row since December 19, 1974 — twenty-three (23) years. He has spent more time on death row than any other inmate in Arizona, possibly more than any other inmate in the United States. Ceja has spent more than half of his life on death row, entering at the age of 19. He is now 42. At the start, he was an irresponsible, street-tough teenager without a high school degree. He is now a middle-aged man with a GED and several college courses to his credit who has held employment as a porter on death row and as a law clerk in the law library at the prison.
For twenty-three years, Ceja has suffered the anxiety of impending death and the greatly restricted activity allowed death row inmates. During that time, Ceja has had an execution date set at least five times: February 8, 1978; September 24, 1980; May 11, 1983; December 19, 1984; and January 21, 1998. For 23 years, Ceja has lived in solitary confinement, much of it in the typical death row cell on Cell Block 6 at the Arizona State Prison in Florence. Those cells are little more than a 7’ x 10’ windowless concrete box with a metal sink and toilet and a concrete slab for a bed. Activity outside that cell is typically limited to 3 one-to-two hour periods per week in which the inmate may shower or exercise. Visitations and phone privileges are much more limited than those for the general prison population. Many of a death row inmate’s neighbors are deeply disturbed men responsible for some of the most notorious murders in Arizona.
If Ceja is executed, his de facto sentence will be 23 years of solitary confinement in the most horrible portion of the prison — death row — followed by execution. There has never been such a sentence imposed in this country — or any other, to my knowledge. Neither Arizona nor any other state would ever enact a law calling for such a punishment.1
The sentencing judge who initially decided that the death sentence was the appropriate punishment now unequivocally states that executing Jose Ceja now after 23 years of incarceration on death row is too harsh a *1370punishment for his crimes. The reasons for imposing the death sentence — as stated by the sentencer himself — are no longer served by execution.
Ceja’s petition raises both jurisdictional and Eighth Amendment claims that are of the highest importance, claims that have never been addressed by a panel of our court. When a prisoner facing execution brings important and meritorious claims before us, we have an obligation to grant a stay so that we can address the merits of those claims. See Barefoot v. Estelle, 463 U.S. 880, 893, 103 S.Ct. 3383, 3394-95, 77 L.Ed.2d 1090 (1983). This is especially so in the present case. Two members of the Supreme Court have said that Ceja’s execution-related claim may have substantial merit, and they have invited the courts of appeals to explore the issue and give it a full hearing on the merits. See Lackey v. Texas, 514 U.S. 1045, 115 S.Ct. 1421, 131 L.Ed.2d 304 (1995) (Stevens, J., joined by Breyer, J., respecting the denial of certiorari). Their conclusion grows out of the Court’s long-standing insistence, most eloquently articulated by Justice White, that the imposition of the death penalty must serve legitimate and substantial penological goals in order to survive Eighth Amendment scrutiny.
At the moment that [a proposed execution] ceases realistically to further these purposes [of deterrence and the coherent expression of moral outrage], the emerging question is whether its imposition in such circumstances would violate the Eighth Amendment. It is my view that it would, for its imposition would then be the pointless and needless extinction of life with only marginal contributions to any discernible social or public purposes. A penalty with such negligible returns to the State would be patently excessive and cruel and unusual punishment violative of the Eighth Amendment.
Furman v. Georgia, 408 U.S. 238, 312-13, 92 S.Ct. 2726, 2764, 33 L.Ed.2d 346 (1972) (White, J., concurring). The panel’s summary dismissal of the vital question that Ceja has raised is wholly unsupportable. We should accept the Justices’ invitation in Lackey and give Ceja’s claim the hearing it deserves.
The Supreme Court of the United States has now denied Ceja’s request for a writ of certiorari to the decision by the Supreme Court of Arizona to deny the Lackey claim that Ceja brought before that court. That denial enhances rather than lessens our obligation to hear Ceja’s claims on the merits.
I.
I necessarily must first address the jurisdiction of our court to hear the Lackey claim that Ceja has raised. The Antiterrorism and Effective Death Penalty Act of 1996 (“AED-PA”) has drastically curtailed the ability of prisoners to obtain relief in the federal courts from allegedly unconstitutional convictions or sentences. Most significantly for present purposes, the Act places strict limitations on the claims that a prisoner can raise in a second or successive petition. The Act provides:
(2) A claim presented in a second or successive habeas corpus application under section 2254 that was not presented in a prior application shall be dismissed unless—
(A) the applicant shows that the claim relies on a new rule of constitutional law, made retroactive to cases on collateral review by the Supreme Court, that was previously unavailable; or
(B)(i) the factual predicate for the claim could not have been discovered previously through the exercise of due diligence; and
(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.
28 U.S.C. § 2244(b)(2). Thus, the thrust of the Act is that a prisoner who seeks to take advantage of the Great Writ must enumerate all the grounds on which he believes himself deserving of such relief in his first petition in federal court. If he fails to do so, he will be *1371barred from pleading his claims later on unless he can make out a convincing claim of actual innocence or the Supreme Court articulates a new, retroactive rule of constitutional law. While the new restrictions may be harsh, they still provide claimants with the opportunity to have all their grounds for relief heard in federal court.
We have recognized, however, that there is a small class of claims for which the AEDPA, at least on its face, appears not to leave open even this small window of opportunity: claims involving the circumstances of a prisoner’s execution. Such claims do not become ripe for adjudication until the state has issued a warrant for the prisoner’s execution. As we explained in Poland v. Stewart, 117 F.3d 1094 (9th Cir.1997), “Where there is no danger of imminent and certain injury to a party, an issue has not matured sufficiently to warrant judicial intervention.” Id. at 1104 (citations omitted). In the case of an Eighth Amendment claim regarding the circumstances of an execution, as opposed to a death-row inmate’s conviction or confinement, when the inmate “does not currently face any risk of execution [under the contested circumstances, he] will face no hardship or immediate or certain danger if [the courts] do not review his Eighth Amendment claim.” Id.
We first encountered the problem that such claims face under the AEDPA in Martinez-Villareal v. Stewart, 118 F.3d 628, 634-35 (9th Cir.1997), cert. granted, — U.S. -, 118 S.Ct. 294, 139 L.Ed.2d 226 (1997). The petitioner in Martinez-Villareal, a death-row inmate from Arizona, claimed that he was not competent to be executed and that carrying out his execution would violate the Eighth Amendment as interpreted by the Supreme Court in Ford v. Wainwright, 477 U.S. 399, 106 S.Ct. 2695, 91 L.Ed.2d 335 (1986). Martinez-Villareal had raised his claim in his first habeas petition, where it was dismissed on the ground that it was not yet ripe for adjudication, See id. at 629-30. Upon the issuance of a warrant for his execution, Martinez-Villareal attempted to reassert his as-yet-unadjudicated claim of incompetence in a second petition. This time, however, he faced the apparently insurmountable hurdle of 28 U.S.C. § 2244(b)(2). Indeed, as we recognized, this was a problem that every prisoner in Martinez-Villareal’s position would necessarily face.
Even if the State had issued its warrant of execution prior to our consideration of Martinez-Villareal’s first habeas petition, his competency claim would still be premature. In Lonchar v. Thomas, 517 U.S. 314, 320-21, 116 S.Ct. 1293, 1297, 134 L.Ed.2d 440 (1996), the Supreme Court instructed that ‘if the district court cannot dismiss the [first habeas] petition on the merits before the scheduled execution, it is obligated to address the merits and must issue a stay to prevent the case from becoming moot.’ Once this stay has been issued, the execution is not imminent, and the competency claim becomes premature. Accordingly, a competency claim cannot be asserted in a first habeas petition.
Martinez-Villareal, 118 F.3d at 630. The apparent result was that the AEDPA categorically prohibited federal courts from ever hearing claims of incompetence to be executed in a petition for a writ of habeas corpus. Such a result would have raised serious questions under Article I of the Constitution, which expressly provides that “The Privilege of the Writ of Habeas Corpus shall not be suspended_” U.S. Const., Art. I, § 9, cl. 2.2
*1372In order to prevent this potential constitutional conflict, we concluded that the AED-PA’s restrictions should be read not to apply to claims, like Martinez-Villareal’s, which, by definition, could not be resolved in a first habeas petition. We had previously determined that dismissals of petitions containing claims for which state court remedies had not yet been exhausted were “without prejudice” under the AEDPA, such that § 2244 could not be read to apply to the refiling of those petitions following proper exhaustion of state remedies. See In re Turner, 101 F.3d 1323, 1323 (9th Cir.1996). In Martinez-Villareal, we extended Turner’s holding to encompass claims dismissed from a first habeas petition on ripeness grounds, as well. See Martinez-Villareal, 118 F.3d at 632-34. This reading, we found, was consistent with Congress’ intent to secure to prisoners an opportunity (if only a single opportunity) to have their federal claims for relief adjudicated in federal court.
As I discuss at greater length below, Ceja’s claim that it would serve no legitimate penological purpose to carry out his execution under the present circumstances is one that only became ripe for adjudication upon the issuance of the present death warrant. If we read the AEDPA to foreclose Ceja from bringing his claim in a successive petition, we would provoke the same conflict with the Suspension Clause that we sought to avoid in Martinez-Villareal. I therefore conclude that the saving construction that we gave to the AEDPA in Martinez-Villareal must be extended to reach Ceja’s execution-related claims.
Martinez-Villareal had actually asserted his claim of incompetence to be executed in his first habeas petition and seen that claim dismissed as unripe. See Martinez-Villareal, 118 F.3d at 629-30; Martinez-Villareal v. Lewis, 80 F.3d 1301, 1309 n. 1 (9th Cir.1996). We seemingly adopted this step in Martinez-Villareal’s march toward relief as a necessary one for future claimants seeking to enjoy the benefit of its holding. See Martinez-Villareal, 118 F.3d at 634 (“Under our holding, a competency claim must be raised in a first habeas petition_”). It is apparent, however, that this filing requirement cannot apply to all execution-related claims. Had Martinez-Villareal become incompetent only after he filed his first petition, for example, then imposing this requirement would not have saved the AEDPA from its potential conflict with the Suspension Clause. If we are to preserve the consistency of its analytical foundations, we must read Martinez-Villareal to impose its filing requirement only upon those unripe claims for which the factual predicate actually exists at the time of the first petition, and could, through the exercise of due diligence, have been discovered, at the time of the prisoner’s first filing.
To adopt any other rule would produce absurd results. If a state decided to adopt the practice of breaking prisoners on the rack as its preferred method of execution, we could not impute to Congress an intention to permit access to the Great Writ only to those death-row inmates who were fortunate enough not yet to have filed a first petition at the time this torturous execution method was adopted. This would provoke the very conflict with the Suspension Clause that our opinion in Martinez-Villareal sought to avoid.
When Ceja first requested collateral review of his sentence in 1983, the factual predicate did not exist for the execution-related claim for which he now seeks relief. At that point, he had been on death row for eight and a half years, a long but hardly remarkable tenure and certainly not enough to raise a colorable Lackey claim. Thus, the Martinez-Villareal filing requirement does not apply to Ceja’s claim.
II.
In finding that the Eighth Amendment does not categorically prohibit the state from imposing the ultimate sanction upon our most serious offenders, the Supreme Court has repeatedly articulated an important qual*1373ification: the imposition of the death penalty (rather than life imprisonment) upon a serious offender must serve some legitimate pe-nological end that could not otherwise be accomplished. If “the punishment serves no penal purpose more effectively than a less severe punishment,” Furman v. Georgia, 408 U.S. 238, 280, 92 S.Ct. 2726, 2747, 33 L.Ed.2d 346 (1972) (Brennan, J., concurring), then it is unnecessarily excessive within the meaning of the Punishments Clause. The Court has defined the penological justifications that can support a legitimate application of the death penalty. Those justifications are twofold: “retribution and deterrence of capital crimes by prospective offenders.” Gregg v. Georgia, 428 U.S. 153, 183, 96 S.Ct. 2909, 2929-30, 49 L.Ed.2d 859 (1976) (plurality opinion).
When a proposed execution “ceases realistically to further these purposes ..., its imposition would then be the pointless and needless extinction of life with only marginal contributions to any discernible social or public purposes ... [and] would be patently excessive and cruel and unusual punishment violative of the Eighth Amendment.” Furman, 408 U.S. at 312-13, 92 S.Ct. at 2764 (White, J., concurring). In no uncertain terms, the Court has said that when the State proposes to kill a human being as a form of punishment for a criminal act, “the sanction cannot be so totally without penological justification that it results in the gratuitous infliction of suffering.” Gregg, 428 U.S. at 183, 96 S.Ct. at 2929.
Ceja claims that the ability of the state to further the ends of retribution and deterrence has been drastically diminished as a result of the extraordinary period of time that has elapsed since the date of his conviction. Justices Stevens and Breyer have confirmed that his claim has merit, writing, “It is arguable that neither ground retains any force for prisoners who have spent some 17 years under a sentence of death.” Lackey, 514 U.S. at 1045, 115 S.Ct. at 1421 (emphasis added). Certainly, Ceja has raised a color-able claim for commutation. We should grant a stay in order to give his claim the consideration it deserves.
A.
In saying that “retribution” can serve as a legitimate basis'for imposing the death penalty, the Supreme Court has recognized an important distinction between the two concepts that the term might be held to embrace: the expression of moral outrage by a community; and the exaction of blood vengeance. I believe that the Court’s treatment of the subject of retribution definitively demonstrates that the exaction of blood vengeance is not a legitimate basis for the imposition of the death penalty. If the State is to justify the execution of Ceja at this late date by adverting to its interest in “retribution,” it must defend the proposition that killing Ceja will serve as an effective expression of the community's moral outrage, despite the extraordinary length of time that has elapsed since Ceja’s conviction. I do not believe that the State will be able to satisfy this burden. In any event, it should be required to do so before this court. I am convinced that Ceja’s challenge is sufficiently substantial that we must grant a stay in order to give that challenge full consideration.
1.
The model of “retribution” that the Court has embraced as a legitimate basis for imposition of the death penalty is the “expression of society’s moral outrage at particularly offensive conduct.” Gregg, 428 U.S. at 183, 96 S.Ct. at 2930. At base, this model of retribution looks ultimately to the stability and coherence of the community as its justification. It asserts that, on rare occasions, a community may need to give voice, and dramatically so, to the injury that it feels at the transgression of its most preciously guarded norms. The Gregg Court characterized this sentiment as “the community’s belief that certain crimes are themselves so grievous an affront to humanity that the only adequate response may be the penalty of death.” Id. at 184, 96 S.Ct. at 2930. This vision of retribution seeks to secure some form of moral and emotional closure to the community, to allow it to move beyond the shock and pain that are occasioned by the most heinous and ugly of crimes. It also finds its justification partly in the fear that communities will resort to *1374vigilantism if they lose their faith in the willingness of the State to respond adequately to such crimes. As Justice Stewart wrote, “When people begin to believe that organized society is unwilling or unable to impose upon criminal offenders the punishment they ‘deserve,’ then there are sown the seeds of anarchy — of self-help, vigilante justice, and lynch law.” Furman, 408 U.S. at 308, 92 S.Ct. at 2761 (Stewart, J., concurring). To the extent that capital punishment is a necessary and indispensable tool in service of these ends, a powerful argument can be made for its application, as the Court recognized in Gregg.
But the Court also recognized in Gregg that allowing the search for retribution to guide a State’s hand in deciding where the axe shall fall is both distasteful and dangerous. See Gregg, 428 U.S. at 182-85, 96 S.Ct. at 2929-31. Indeed, the Court has explained, on a number of occasions, that “[rjetribution is no longer the dominant objective of the criminal law.” Williams v. New York, 337 U.S. 241, 248, 69 S.Ct. 1079, 1084, 93 L.Ed. 1337 (1949); Gregg, 428 U.S. at 183, 96 S.Ct. at 2929-30. Thus, it is not enough, in seeking to justify an application of the death penalty, simply to point to the details of a terrible crime and assert a desire for “retribution.” Even accepting the legitimacy of expressions of moral outrage as a justification for the death penalty as a general proposition, whenever the State proposes to carry out an execution in a particular instance, “the sanction imposed cannot be so totally without penological justification that it results in the gratuitous infliction of suffering.” Id. at 183, 96 S.Ct. at 2929.
The Court has explained that the death penalty cannot serve as a legitimate expression of a community’s moral outrage unless there is some coherent relationship between the standards that are employed in administering that sanction and the community’s desire for moral and emotional closure. As Justice Brennan has said, “If capital crimes require the punishment of death in order to provide moral reinforcement for the basic values of the community, those values can only be undermined when death is ... rarely inflicted upon criminals who commit the crimes.” Furman, 408 U.S. at 303, 92 S.Ct. at 2759 (Brennan, J., concurring). In other words, the severe limitations that the Court has imposed on the application of the death penalty since its decisions in Furman and Gregg are not simply the product of some abstract desire to mete out punishments in a fair and even-handed manner. Those limitations are necessary if the continued application of the death penalty is to be effective at accomplishing the narrow penological goals that serve as its only legitimate justification.
The circumstances surrounding Ceja’s proposed execution raise serious questions as to the possible effectiveness of his execution in serving those narrowly defined goals, the same types of questions that the Court wrestled with in Furman and in Gregg. In those cases, the Court recognized that the ability of an execution to provide moral and emotional closure to a shocked community diminished as the connection between crime and punishment became more attenuated and more arbitrary. Ceja’s claim is no different. It is far from clear, on an intuitive level, what healing or stabilizing effect Ceja’s execution will have twenty-three years after the date of his original conviction. Does this mean that Ceja’s community (through the voice of the sentencing judge) was wrong in thinking, when it imposed the death penalty, that executing Ceja could serve a vital and necessary function in that community? Within the moral and ethical framework that Gregg has established, no, perhaps not. But we simply cannot uncritically assume that the interests that legitimized that initial decision will continue to be vital and strong, no matter how many decades elapse before Ceja’s execution is proposed to be carried out. There is no statute of limitations on the Eighth Amendment. At the very least, after such an extraordinary length of time, the State must demonstrate that its interest in securing stability and harmony to the community through the exaction of this most extreme form of retribution continues to have affect and force.
2.
The unshakable foundation of our Eighth Amendment jurisprudence, particularly in *1375the context of the death penalty, is that the most severe punishments that the State seeks to impose must be assessed in light of our evolving standards of fairness and decency in a civilized society. See Gardner v. Florida, 430 U.S. 349, 357, 97 S.Ct. 1197, 1204, 51 L.Ed.2d 393 (1977); Gregg, 428 U.S. at 171-73, 96 S.Ct. at 2924-25; Witherspoon v. Illinois, 391 U.S. 510, 519-20 n. 15, 88 S.Ct. 1770, 1775-76 n. 15, 20 L.Ed.2d 776 (1968). The Court has held that, even in light of this imperative, a vision of retribution that seeks to give voice to moral outrage is not “inconsistent with our respect for the dignity of men [and women].” Gregg, 428 U.S. at 183, 96 S.Ct. at 2929-30. But a vision of retribution that seeks to exact blood vengeance enjoys no such respect or deference. So defined, “Retaliation, vengeance, and retribution have been roundly condemned as intolerable aspirations for a government in a free society.” Furman, 408 U.S. at 343, 92 S.Ct. at 2779 (Marshall, J., concurring). “The ‘cruel and unusual’ language limits the avenues through which vengeance can be channeled. Were this not so, the language would be empty and a return to the rack and other tortures would be possible in a given ease.” Id. at 345, 92 S.Ct. at 2780.
If we were to enfranchise a primitive, eye-for-an-eye model of blood vengeance as a legitimate and sufficient justification for selecting and imposing punishment in our society, I can see no principled limits on the power of the State to impose extreme and arbitrary suffering. It is the purpose of the Punishments Clause to guard against this result. The death penalty can be justified, when it can be justified, only to the extent that it is necessary to serve vital and legitimate penological goals. I believe it to be firmly established that a bare desire to exact blood vengeance from the perpetrator of a crime, harbored and nursed along over the course of years and decades, cannot satisfy that requirement. Any other conclusion would truly signal a return to the anarchy and vigilantism that Justice Stewart has warned us against so eloquently. See Furman, 408 U.S. at 308, 92 S.Ct. at 2761-62.
B.
There has always been substantial debate as to whether the death penalty has any measurable deterrent effect over the imposition of a life sentence, and the Supreme Court, in upholding particular applications of the death penalty, has never purported to resolve that debate. Rather, the Court has expressed considerable ambivalence. “Although some of the studies [we have reviewed] suggest that the death penalty may not function as a significantly greater deterrent than lesser penalties, there is no convincing empirical evidence either supporting or refuting this view.” Gregg, 428 U.S. at 185, 96 S.Ct. at 2931. The Court’s willingness to uphold particular applications of the death penalty despite this ambivalence stems from its deference to the institutional function of state legislatures and the lack of hard evidence in refutation.
In sum, we cannot say that the judgment of the Georgia Legislature that capital punishment may be necessary in some cases is clearly wrong. Considerations of federalism, as well as respect for the ability of a legislature to evaluate, in terms of its particular State, the moral consensus concerning the death penalty and its social utility as a sanction, require us to conclude, in the absence of more convincing evidence, that the infliction of death as a punishment for murder is not without justification and thus is not unconstitutionally severe.
Gregg, 428 U.S. at 187, 96 S.Ct. at 2931-32 (emphasis added). The question that the Court deliberately left open in Gregg was what might constitute “more convincing evidence” that a particular application of the death penalty is, in fact, “without [penological] justification.”
It is Ceja’s claim that the imposition of an extraordinary delay between the time of conviction and the time that an execution is proposed to be carried out provides the “convincing evidence” that the Gregg Court envisioned in holding open the possibility that some applications of the death penalty would violate the Eighth Amendment. This claim finds considerable support in the Court’s own discussion of the capacity of the death penal*1376ty to function as a deterrent. As Justice Brennan wrote in Furman, “Proponents of this argument [that the death penalty can be justified by its deterrent effect] necessarily admit that its validity depends upon the existence of a system in which the punishment of death is invariably and swiftly imposed.” Furman, 408 U.S. at 302, 92 S.Ct. at 2758 (Brennan, J., concurring). Justice Marshall echoed this sentiment: “For capital punishment to deter anybody it ... must ... follow swiftly upon completion of the offense.” Id. at 354 n. 124, 92 S.Ct. at 2785 n. 124 (Marshall, J., concurring).
Gregg expressly left open the possibility that there might be circumstances in which a future claimant could demonstrate that carrying out his proposed execution, rather than imposing a life sentence, could not possibly have any deterrent effect upon the commission of future criminal acts. If there is any category of prisoners who can make out a substantial claim on this score, it is those prisoners whom the State proposes to execute decades after they have been convicted. As Justice Stevens said, “the additional deterrent effect from an actual execution now, on the one hand, as compared to 17 years on death row followed by the prisoner’s continued incarceration for life, on the other, seems minimal.” Lackey, 514 U.S. at 1046, 115 S.Ct. at 1421-22 (Stevens, J., respecting the denial of certiorari). If the State believes that there are real and legitimate deterrence interests that will be served by carrying out Ceja’s execution at this date, then it can make its ease on that score before our court. But there can be no doubt that Ceja has brought a substantial challenge, of precisely the kind that the Court envisioned in Gregg, to the continuing sufficiency of deterrence as a justification for carrying out his execution.
III.
Our court’s en banc decision in McKenzie v. Day, 57 F.3d 1461 (9th Cir.1995) (en banc), does not speak at all to the claim that Ceja has raised. That opinion addressed an entirely different issue: whether it constitutes cruel and unusual punishment to confine an individual on death row for extremely prolonged periods of time. The analysis that we brought to bear in denying relief to the claimant in McKenzie has no application to the claim that Ceja has raised. Ceja’s claim has never been thoughtfully considered by a panel of our court.
A.
There are two types of Eighth Amendment claim that a prisoner might assert following a decades-long tenure on death row. First, the prisoner might argue that confining a person on death row for such an extended period of time itself constitutes cruel and unusual punishment. The core of this claim is the assertion that having a death sentence hanging over one’s head subjects one to extraordinary psychological duress, as well as the extreme physical and social restrictions that inhere in life on death row, and that it constitutes cruel and unusual punishment to impose such conditions of stress upon a death row inmate for a period of decades. See Lackey, 514 U.S. at 1045-46, 115 S.Ct. at 1421-22. Second, the prisoner might argue that, after a sufficient amount of time has elapsed since the defendant’s conviction, the execution would violate the Eighth Amendment because the ability of the state to exact retribution and to deter other serious offenses by actually carrying out the defendant’s execution is drastically diminished. As discussed above, the core of this claim is that the penological justification for carrying out an execution disappears when a period of decades has elapsed between the conviction and the proposed execution date, and that actually executing the defendant under such circumstances is an inherently excessive punishment that no longer serves any legitimate purpose. See id.; Furman v. Georgia, 408 U.S. 238, 312, 92 S.Ct. 2726, 2763-64, 33 L.Ed.2d 346 (1972) (White, J., concurring).
McKenzie addressed only the first of these two types of Lackey claims. McKenzie was a capital defendant who had been on death row in Montana for twenty years and had been sentenced and resentenced to death eight times during the course of his trial and post-conviction proceedings. In the petition that gave rise to our en banc decision, McKenzie claimed that “the state of Montana’s inordinate delay in carrying out his sentence con*1377stitutes cruel and unusual punishment, a claim similar to that raised in Texas by [the petitioner in the Lackey ease].” McKenzie, 57 F.3d at 1463. We declined to grant a writ. Our reasons for doing so were twofold. We found, first, that a prisoner’s excessive delay in bringing a claim of unconstitutionally cruel confinement could cause the prisoner to be barred from obtaining relief; and, second, that McKenzie had faded to make out a meritorious claim in any event. Our analysis in reaching these holdings has application only to the claim that prolonged confinement on death row can constitute cruel and unusual punishment. That analysis is inapplicable to the claim that, under some circumstances, there ceases to be any legitimate penological justification for carrying out an execution. McKenzie does not speak to this second type of Lackey claim at ah.
Most significantly, our conclusion that McKenzie was equitably barred from seeking a writ of habeas corpus rested on the assumption that McKenzie could have brought his claim much earlier in the post-conviction review process. See, e.g., McKenzie, 57 F.3d at 1465 (“Had McKenzie raised the Lackey claim any time during [the ten years prior to the issuance of the warrant for his execution], it could have been considered on the merits.... ”); id. (“McKenzie’s Lackey claim did arise long before he first raised it two-and-a-half weeks ago. As McKenzie himself points out, he has been on death row now for two decades. At the time his ease was last remanded to the district court in 1990, McKenzie had been on death row for 15 years, almost as long as Lackey himself. When we disposed of McKenzie’s claim in June 1994, he had been on death row for 19 years. The Lackey claim could have been raised and considered at either of those times, or anytime in between-”). We placed particular weight on the fact that McKenzie unnecessarily caused a death warrant to be stayed when he could have avoided that result by bringing his claim earlier. See id. (by bringing claim at eleventh hour, McKenzie provoked yet another stay of execution”); id. (McKenzie unnecessarily required court “to vacate a death warrant”); id. at 1468 (“McKenzie could and should have raised his Lackey claim at a time when it was capable of being resolved without staying a scheduled execution.”).
This analysis is inapposite to a claim that a particular, proposed execution can serve no legitimate penological end and is inherently excessive under the Eighth Amendment. As discussed above, a prisoner’s claim that carrying out his execution at a given time or in a given manner would violate the Eighth Amendment is, by definition, not ripe for adjudication until a death warrant has actually been issued. See Poland, 117 F.3d at 1104; Martinez-Villareal, 118 F.3d at 630. McKenzie’s entire analysis is based on the assumption that the claim it rejects became ripe for adjudication long before any warrant was issued. This analysis can apply only to a claim regarding the conditions of a prisoner’s confinement, and not the circumstances surrounding his proposed execution.3
That our opinion in McKenzie considered only the first type of Lackey claim, and not the second, is further demonstrated by the opinion’s discussion of the relief that might be appropriate to a claim of unconstitutionally cruel delay.
Finally, it is unclear to us whether, even if it were held that delay in the imposition of the death penalty constitutes cruel and unusual punishment, commutation of the death penalty will turn out to be the appropriate remedy.... [W]hatever anguish McKenzie has suffered is in the past and cannot be undone.
McKenzie, 57 F.3d at 1467. This concern obtains only where the prisoner’s claim is that delay in and of itself constitutes cruel and unusual punishment. Where a prisoner’s claim is that actually carrying out his execution would serve no legitimate penological ends, commutation is the appropriate remedy.
*1378B.
By the same token, McKenzie’s analysis on the merits is limited only to the claim of unconstitutionally cruel delay. We did not provide any independent analysis of the merits of McKenzie’s claim; rather, we adopted the analysis from a previously vacated opinion, Richmond v. Lewis, 948 F.2d 1473 (9th Cir.1990). This adopted analysis clearly addresses itself only to the claim that excessive time spent on death row might constitute cruel and unusual punishment. On three occasions in its short analysis, the Richmond court characterizes Richmond’s claim as one that the “accumulation of time a defendant spends on death row during the prosecution of his appeals can accrue into an independent constitutional violation.” Richmond, 948 F.2d at 1491. See also id. at 1492 (Richmond’s claim is that “the delay incurred during the prosecution of claims that fail on the merits could itself accrue into a substantive claim to the very relief that had been sought and properly denied in the first place”); id. (Richmond’s claim is that his “lengthy incarceration on death row during the pendency of his appeals [was] substantively and independently violative of the Constitution.”); McKenzie, 57 F.3d at 1494 (reproducing and adopting this language). A claim that “time ... spen[t] on death row ... can accrue into an independent constitutional violation,” or that a “lengthy incarceration on death row ... [is] substantively and independently vio-lative of the Constitution,” is a claim involving the alleged cruelty of the confinement itself, not the lack of penological justification for carrying out an execution decades after a defendant has been convicted.
Every ease that the Richmond court cited in support of its conclusion dealt only with the cruelty of keeping a prisoner on death row (or otherwise confined under duress) for a period of many years. None of the cited opinions acknowledge or analyze the claim that, under some circumstances, carrying out a sentence of execution would have no legitimate penological justification. See Richmond, 948 F.2d at 1491 (citing Andrews v. Shulsen, 600 F.Supp. 408, 431 (D.Utah 1984), aff'd, 802 F.2d 1256 (10th Cir.1986); People v. Chessman, 52 Cal.2d 467, 498-99, 341 P.2d 679 (Cal.1959); Harrison v. United States, 392 U.S. 219, 221 n. 4, 88 S.Ct. 2008, 2009 n. 4, 20 L.Ed.2d 1047 (1968)). Richmond, like the panel opinion in McKenzie, simply does not address this second type of Lackey claim.4
CONCLUSION
For all the reasons I have elaborated, the panel should have granted a stay and accorded Ceja a hearing on the merits of its claims. I respectfully dissent from the panel’s summary refusal to do so.
. I rely heavily on Ceja’s petition to the United States Supreme Court for these facts.
. The Supreme Court has not yet articulated a rule for determining what might constitute a violation of the Suspension Clause. As we explained in Martinez-Villareal, however, the "saving construction” that the Court recently gave to another provision of the AEDPA in order to avoid reaching that question appears not to be available in the situation that this case presents.
In Felker [v. Turpin, 518 U.S. 651, -, 116 S.Ct. 2333, 2339, 135 L.Ed.2d 827 (1996)], the Court concluded that the "gatekeeping” system of the new Act, codified at 28 U.S.C. § 2244(b)(3), whereby a petitioner must obtain the permission of the court of appeals before filing a second or successive petition in the district court, “does not apply to our consideration of habeas petitions because it applies to applications 'filed in the district court.’ ” ... Unlike § 2244(b)(3), [however,] §§ 2244(b)(1) and (2) do not contain limiting language and refer generally to all second or successive peti-tions_ [This] suggests] that even the Supreme Court is bound by the new limitations on habeas review imposed by § 2244(b)(1) and *1372(2). If we are correct, then a state prisoner’s competency claim can never be heard by any federal court. The constitutional problem with this preclusion is patent.
Martinez-Villareal, 118 F.3d at 631-32.
. I do not suggest that I agree with the McKenzie court’s refusal to examine the merits of McKenzie’s claim. Because of the odd procedural posture of McKenzie, the dissents suggested that the second part of the majority’s opinion did not constitute a reasoned decision on the merits, and, indeed, amounted to nothing more than obiter dictum. See McKenzie, 57 F.3d at 1494-95 (Thompson, J., dissenting).
. There is only one sentence in the McKenzie opinion that could possibly be read to relate to the claim that Ceja has brought before us. In introducing its discussion of the Richmond case, the majority describes the claim it goes on to reject in the following terms.
[McKenzie] argues that to execute him after the great delay that has occurred between his conviction and date of execution (20 years), combined with the repeated resetting of his execution date (8 times), and the allegedly unconstitutional conditions of his confinement, amount to cruel and unusual punishment.
McKenzie, 57 F.3d at 1494 (emphasis added). The only coherent reading of this sentence is the most straightforward one. McKenzie claimed that the Eighth Amendment required commutation of his death sentence because of the extraordinary duress he had undergone. In rejecting that claim, we fully enumerated the sources of that duress: (1) the psychological torture that McKenzie claimed to have experienced as a result of the "great delay that ... occurred between his conviction and date of execution (20 years), combined with [2] the repeated resetting of his execution date (8 limes), and [3] the allegedly unconstitutional conditions of his confinement.” Id.
Every word of analysis and every citation that we adopted in McKenzie addressed only the claim that prolonged confinement on death row might itself constitute cruel and unusual punishment. Ceja’s claim that the death penalty might, under some circumstances, cease to serve any legitimate penological ends is entirely distinct from such a claim and is a question of first impression in this circuit.