Robert Sawyer was sentenced to death by a Louisiana jury on September 19, 1980 for the brutal slaying of Frances Arwood. Today we decide his appeal from the denial by a United States District Court of his petition for writ of habeas corpus. We have elsewhere recorded the long history of Sawyer’s efforts to overturn his conviction.2 Sawyer’s attack has now boiled *1276down to three arguments. First, he argues that his court-appointed trial counsel was ineffective in certain respects. Second, and closely related to the first, he argues that his conviction should be set aside because his appointed counsel had not been licensed for five years as required by La. Code Crim.P. art. 512. Finally, he argues that in closing argument the prosecutor misled the jury about its role in capital sentencing as condemned by Caldwell v. Mississippi, 472 U.S. 320, 105 S.Ct. 2633, 86 L.Ed.2d 231 (1985).
A panel of this court rejected Sawyer’s contentions, dividing over the Caldwell issue, and we took the case en banc. We reject Sawyer's first two contentions for the reasons stated by the panel, affirm the district court’s denial of Sawyer’s petition for relief from his conviction, and turn to the difficult question of whether Sawyer is entitled to a new sentencing hearing because the state misled the jury about the jury’s responsibility in deciding whether Sawyer should be executed.
Part I summarizes the facts. In Part II we sketch the constitutional principles that frame our inquiry. We next in Part III address the statutory overlay to the constitutional issues, as presented by the Supreme Court’s recent decision in Teague v. Lane, — U.S. -, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). Because we conclude that we cannot apply Teague without first defining the scope of Caldwell, we turn back in Part IV to the substantive constitutional questions. We endorse a version of Sawyer’s construction of Caldwell. We find in Part V, however, that Caldwell so defined is a new rule within the meaning of Teague, and that Caldwell does not fit within either of Teague’s two exceptions. Sawyer’s Caldwell argument is therefore Teague-bsxred. The prosecutorial argument complained of will thus vitiate Sawyer’s death sentence only if Sawyer can prevail under the earlier rule of Donnelly v. DeChristoforo, 416 U.S. 637, 94 S.Ct. 1868, 40 L.Ed.2d 431 (1974). In Part VI we conclude that Sawyer has no Donnelly claim. We therefore affirm denial of Sawyer’s petition to vacate his sentence.
I
Caldwell addressed constitutional issues that arise when a prosecutor misleads a capital jury about its responsibility for the sentencing decision. The prosecutor’s argument creates a possibility that the jury will decide between life and death without an appropriate sense of grave responsibility. Sawyer contends that Caldwell mandates a new sentencing trial any time a prosecutor taints the proceeding with a Caldwell-type argument, unless the argument had “no effect” upon the jury. Louisiana, however, says that a Caldwell-type prosecutorial argument will not generate constitutional grounds for reversal unless the argument rendered the sentencing phase “fundamentally unfair” to the defendant. Louisiana would have us focus upon effective prejudice to the defendant, rather than effective dilution of the jury’s sense of responsibility. The case turns upon this disagreement.
Sawyer’s Caldwell claim arises out of remarks which the prosecutor made in his closing argument during the trial’s sentencing phase. The details of the prosecu-torial remarks are important to Sawyer’s argument. We therefore repeat those remarks here. The prosecutor told the jury:
The law provides that if you find one of these circumstances then what you are doing as a juror, you yourself will not be sentencing Robert Sawyer to the electric chair. What you are saying to this Court, to the people of this Parish, to any appellate court, the Supreme Court of this State, the Supreme Court possibly of the United States, that you the people as a fact finding body from all the facts and evidence you have heard in relationship to this man’s conduct are of the opinion that there are aggravating circumstances as defined by the statute, by the State Legislature that this is a type of crime that deserves that penalty. It is merely a recommendation so try as he may, if Mr. Weidner tells you that each and every one of you I hope can live with your conscience and try and play upon your emotions, you cannot deny, it is a difficult decision. No *1277one likes to make those [sic] type of decision but you have to realize if but for this man’s actions, but for the type of life that he has decided to live, if of his own free choosing, I wouldn’t be here presenting evidence and making argument to you. You wouldn’t have to make the decision [emphasis supplied].
The prosecutor drew the jury’s attention to the brutal nature of the crime for which Sawyer stood convicted. The prosecutor then returned to the theme of the jury’s responsibility, saying:
There is really not a whole lot that can be said at this point in time that hasn’t already been said and done. The decision is in your hands. You are the people that are going to take the initial step and only the initial step and all you are saying to this court, to the people of this Parish, to this man, to all the judges that are going to review this case after this day, is that you the people do not agree and will not tolerate an individual to commit such a heinous and atrocious crime to degrade such a fellow human being without the authority and impact of the law of Louisiana. All you are saying is that this man from his actions could be prosecuted to the fullest extent of the law. No more and no less (emphasis supplied).
After arguing that a death penalty was justified in Sawyer’s case, the prosecutor struck the theme of jury responsibility again, telling the jury that their mistakes could be corrected by later decision-makers:
It’s all your3 doing. Don’t feel otherwise. Don’t feel like you are the one, because it is very easy for defense lawyers to try and make each and every one of you feel like you are pulling the switch. That is not so. It is not so and if you are wrong in your decision believe me, believe me there will be others who will be behind you to either agree with you or to say you are wrong so I ask that you do have the courage of your convictions (emphasis supplied).
II
The problem of Caldwell error touches upon three of the Constitution’s grandest themes. Two of these are obvious. The problem implicates federalism, because the state asserts a power to decide for itself questions of criminal procedure. Caldwell analysis also concerns individual rights, since the defendant contends that diminishing a capital jury’s sense of responsibility subjects him to cruel and unusual punishment. The third theme is perhaps less obvious, but no less important to understanding the issues raised by a Caldwell claim. Caldwell touches the principle of popular self-government, because the direct expression of popular sentiment through juries remains an important aspect of the people’s participation in the government, and a crucial check upon the state’s authority to define the limits of crime and punishment.
The jury seems always to be at the center of the judicial struggle with the death penalty. This should not be surprising. Differences over the role of the jury reflect differences over the wisdom of the penalty itself. The legislative judgment specifying execution as the punishment appropriate to certain crimes embodies a confidence both about the moral principles of the community and about the capacity of the criminal justice system to resolve factual disputes. Coupled with that confidence must be an equal certitude that the jury will be able to bring the community's principles to bear, and so judge blame and guilt accurately in the individual case.
In McGautha v. California, 402 U.S. 183, 91 S.Ct. 1454, 28 L.Ed.2d 711 (1971), Justice Harlan summarized how history had given expression to this deep link between the death penalty and the jury. Justice Harlan explained that legislatures “to meet the problem of jury nullification ... did not try, as before, to refine further the definition of capital homicides. Instead, they adopted the method of forthrightly granting juries the discretion which they *1278had been exercising in fact.” Id. at 199, 91 S.Ct. at 1463. Justice Harlan observed that the Court had earlier concluded that “one of the most important functions any jury can perform in making such a selection is to maintain a link between contemporary and community values and the penal system — a link without which the determination of punishment could hardly reflect the evolving standards of decency that mark the progress of a maturing society.” Id. at 202, 91 S.Ct. at 1464, quoting Witherspoon v. Illinois, 391 U.S. 510, 519 n. 15, 88 S.Ct. 1770, 1775 n. 15, 20 L.Ed.2d 776 (1968).
We have long recognized that decisions that depend essentially upon inarticulable judgment and common sense intuition are prime candidates for jury decision. Indeed, we refer to these judgments as “blackbox decisions.” The sentencing decision in capital cases is born out of an inherent and unique mixture of anger, judgment and retribution, and requires a determination whether certain acts are so beyond the pale of community standards as to warrant the execution of their author. This decision to punish by death is a paradigmatic “black-box” call. To say that the decision can at best only be guided, not determined, by a judicial instruction or lawyers’ argument underscores the decision’s irreducible discretionary core.
A commitment to jury resolution of these blackbox decisions reflects a commitment to submit these issues to an active exercise of practical judgment, rather than to the reified precision of legal analysis. But the jury, of course, checks not only legalism but the government more generally. It protects from punishment those defendants who are innocent in the judgment of their peers. For both these reasons, the right to trial by jury has long been cherished within our legal tradition. Blackstone commended juries as an “admirable criterion of truth, and most important guardian both of public and private liberty.” W. Blackstone, 4 Commentaries 407. The Constitution expressly secures the right to jury trial. It is, then, neither coincidental nor surprising that the jury’s integrity should be so aggressively protected in capital cases, when the stakes are so high.
Of course, the Court has since rejected McGautha’s teaching that “[t]o identify before the fact those characteristics which call for the death penalty ... in language which can be fairly understood and applied by the sentencing authority, appear to be tasks which are beyond present human ability.” 402 U.S. at 204, 91 S.Ct. at 1466. The Court has demanded that states guide the jury’s discretion. The Court has also permitted states to take some power away from the jury. But the jury’s sense of gravity, and the responsible discretion it fosters, remain crucial to post-McGautha sentencing schemes. Caldwell articulates a constitutional protection against state conduct that diminishes the jury’s perception of its awesome responsibility.
In this sense, Caldwell itself is but the trace of a more comprehensive rule, one that might have trusted jury discretion to protect individual rights and express the scope of state power. The Court’s post-McGautha jurisprudence has instead sought to secure individual rights by limiting jury discretion, and has deferred to the states’ own restrictions upon jury power. The Constitution, after all, permits the people to speak through state law as well as through juries. Federalism, no less than jury participation, ties local penalties to local sentiment and local judgment.
Nonetheless, it is necessary to perceive the larger theme in order to understand its trace within the composition that remains. Caldwell stands in part for the continuing vigor of the ideals articulated by Justice Harlan in McGautha. Caldwell treats jury discretion within a framework that recognizes both federal and state limits upon the jury’s power. But it is the larger whole behind the trace which accounts for Caldwell’s peculiar nexus to the constitutional mix of individual autonomy, federalism, and populism.
Indeed, this reflection of McGautha’s ideals in Caldwell forms the lynchpin of Sawyer’s argument here, and was the fulcrum for the argument that divided our panel. Only if Caldwell harkens back to the high esteem which McGautha accorded jury discretion can Caldwell impose, as *1279Sawyer would have it, considerably more stringent restrictions than its Due Process Clause precursor, Donnelly v. DeChristoforo, 416 U.S. 637, 94 S.Ct. 1868, 40 L.Ed.2d 431 (1974). Donnelly subjected prosecutorial argument to a generalized “fundamental fairness” standard, which would benefit Sawyer only were he able to show actual prejudice from the argument complained of. Sawyer’s principal argument presupposes that the Eighth Amendment, as interpreted by Caldwell, puts a particular premium upon responsible jury discretion in a proceeding that fixes punishment at life or death. It is that premium which would, on Sawyer’s argument, distinguish Caldwell from Donnelly. The existence of that premium in turn assumes that a jury’s deliberation may be even more crucial at the punishment phase than it is in choosing between guilt and innocence. That assumption makes sense only if, as Justice Harlan argued in McGautha, the jury’s capacity to express moral sentiment directly is peculiarly essential to questions of capital blameworthiness.
Because Sawyer’s claim comes before us by way of a habeas petition, not by direct appeal, we view the delicate constitutional mix through a similarly complex statutory overlay. The law of the habeas writ balances the vindication of constitutional rights against the state’s constitutionally legitimate interest in maintaining a criminal justice system capable of producing final convictions. The Supreme Court refined anew this balance in Teague v. Lane, — U.S. -, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989). Teague’s rule precludes habe-as petitioners from seeking to overturn their convictions on the basis of rules new by comparison with the date their convictions became final. This statutory balance provides, however, exceptions for constitutional claims of a certain character. It may therefore wrap back around the constitutional issues, and so, in Sawyer’s case, back around the questions about jury responsibility in capital cases. Yet a plurality, at least, of the Teague Court regarded the Teague retroactivity inquiry as a preemptive threshold to constitutional analysis. 109 S.Ct. at 1069. Accord, Penry v. Ly-naugh, — U.S. -, 109 S.Ct. 2934, 2944, 106 L.Ed.2d 256 (1989) (applying Teague as threshold barrier to constitutional analysis). Because Teague may present a threshold barrier to fuller consideration of Sawyer’s constitutional claims, we begin our analysis with that case.
Ill
The Supreme Court did not decide Teag-ue until after the en banc court heard oral argument in this case. At our request the parties have filed briefs regarding Teag-ue ’s applicability to Sawyer’s petition.
Teague adopts much of what Justice Harlan long advocated as the correct view of federal habeas. Under Teague a federal habeas petitioner attacking a final state conviction may rely only upon the law in effect when his conviction became final. There are two exceptions. First, the petitioner may rely upon a new rule if it would place “certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe.” Id., 109 S.Ct. at 1073 (quoting Mackey v. United States, 401 U.S. 667, 692, 91 S.Ct. 1160, 1175, 28 L.Ed.2d 404 (1971) (Harlan, J., concurring in part and dissenting in part)). The Court has since declared that this first exception also applies to rules which exempt certain persons entirely from capital punishment. Penry, 109 S.Ct. at 2955. Second, the petitioner may rely on a new rule requiring the observance of “those procedures that ... are ‘implicit in the concept of ordered liberty’ ” Teague, 109 S.Ct. at 1073, quoting Mackey, 401 U.S. at 693, 91 S.Ct. at 1180 (opinion of Harlan, J.) (inside quote from Palko v. Connecticut, 302 U.S. 319, 325, 58 S.Ct. 149, 152, 82 L.Ed. 288 (1937) (Cardozo, J.)).
A majority of the Teague Court fully subscribed to this restriction on the use of federal habeas to attack final state court convictions. Teague left much of the restriction’s content in doubt, although some of that ambiguity was removed by the Court’s later decision in Penry v. Lynaugh, 109 S.Ct. 2934, 2944 (opinion of O’Connor, J., for the Court). In Teague itself, four justices concluded, in an opinion by Justice O’Connor, that the second proviso, drawn *1280from Cardozo’s incorporation formulation, should be modified to limit its scope “to those new procedures without which the likelihood of an accurate conviction is seriously diminished.” 109 S.Ct. at 1076-77. The remaining justices filed four separate opinions: Justice White concurred separately, as did Justice Stevens; Justice Blackmun joined part of Justice Stevens’s opinion, and added a brief writing of his own; and Justices Brennan and Marshall dissented.
Teague was not a capital case, and the plurality disclaimed any decision regarding its application to an effort by a state prisoner to overturn his death sentence. Justice Stevens joined Justice O’Connor’s opinion insofar as it adopted Justice Harlan’s restrictions on federal habeas. He dissented, however, from the plurality’s insistence that “the only procedural errors deserving correction on collateral review are those that undermine ‘an accurate determination of innocence or guilt’_” Id. at 1081. He suggested that “a touchstone of factual innocence would provide little guidance in certain important types of cases, such as those challenging the constitutionality of capital sentencing hearings.” Id. Justice Stevens noted that Justice Harlan’s interest in making convictions final was “an interest that is wholly inapplicable to the capital sentencing context.” Id. at 1081 n. 3. Justice O’Connor’s plurality opinion replied that because Teague was not himself under a death sentence, the Court need not express any opinion “as to how the retroac-tivity approach we adopt today is to be applied in the capital sentencing context. We do, however, disagree with Justice Stevens’s suggestion.... As we have often stated, a criminal judgment necessarily includes the sentence imposed upon the defendant.” Id. at 1077 n. 3.4
Note three did not gain majority support, since Justice White neither joined it nor otherwise mentioned Teague’s application to death cases. Justice Brennan’s dissenting opinion, joined by Justice Marshall, assumes that the plurality would apply the new limits to death cases, and observes that “the plurality’s new rule apparently would not prevent capital defendants ... from raising Eighth Amendment, due process, and equal protection challenges to capital sentencing procedures on habeas corpus.” Id. at 1089 n. 5.
The Penry decision settled Teague’s application to death cases. In Part II-A of her opinion for a fractured Court, Justice O’Connor, joined by the Chief Justice and Justices White, Scalia, and Kennedy, held that Teague did apply to capital cases. The plurality simply observed that the finality concerns underlying the Teague doctrine hold equally well in capital cases, and offered no further analysis. The four remaining Justices dissented from the relevant portion of Justice O’Connor’s opinion.
It remains unclear, however, whether Teague necessarily operates as a threshold barrier preempting full analysis of the constitutional claims asserted. The Teague plurality clearly thought that a Teague bar would preempt discussion of the constitutional merits. 109 S.Ct. at 1069-70, 1077. However, Justices Stevens and Blackmun, who joined the plurality to constitute a majority in favor of Justice Harlan’s approach to retroactivity, expressly rejected the plurality’s position on this matter. Justice Stevens, joined by Justice Blackmun, contended that the Court should proceed by “first determining whether the trial process violated any of the petitioner’s constitutional rights and then deciding whether the petitioner is entitled to relief.” Justice Stevens went on to observe that, absent a precise formulation of the rule in question, it may be difficult to determine whether the rule is in fact “new” at all. Id. at 1079-80 & n. 2. Finally, Justice White once again declined to join the relevant portion of the plurality opinion, leaving unclear his own position on the relation between the constitutional and Teague issues.
On this point, Penry leaves the matter unclear. A majority did join a portion of Justice O’Connor’s opinion which characterized Teague as a rule to be applied “as a *1281threshold matter,” 109 S.Ct. at 2944 (Part II-A). Indeed, in Part IV-A all nine Justices joined a portion of the opinion which included a reference to Teague as a threshold test. Id. at 2952. We must take care, however, not to overstate the significance of these votes. Thus, although Justice Stevens joined Part IV-A of Justice O’Con-nor’s opinion, he reiterated in a separate concurrence his view that the constitutional rule should be articulated before Teague is applied. The threshold character of the Teague bar was not the primary topic of Part II-A or Part IV-A, and it would be unwise to assume that each Justice joining those parts intended that Teague function as a threshold barrier in every case where it applied.
More importantly, however, Justice O’Connor’s own opinion mixed the Teague inquiry with the constitutional questions. In order to decide that Penry’s requested rule was dictated by precedent, and so not new, she had to decide precisely the substantive question which divided the Justices five-to-four over Part III of her opinion: that is, the question of whether Penry’s proposed rule was the best possible interpretation — let alone the interpretation “dictated by” — Supreme Court precedent. 109 S.Ct. at 2944-46 (Part II — B). Likewise, Justice Scalia, dissenting in part and joined by the Chief Justice, Justice White and Justice Kennedy, observed that “[t]he merits of the mitigation issue, and the question of whether, in raising it on habeas, petitioner seeks application of a ‘new rule’ within the meaning of Teague, are obviously interrelated.” 109 S.Ct. at 2964.
The relationships that led to a mixing of the Teague issues and the constitutional issues in Penry become all the more powerful when a petitioner attempts not to establish a new rule, but to rely, as Sawyer would like to, upon a rule that is new by comparison to his own conviction yet is well established by the time of his habeas petition. In such a case, a court may have to reach the constitutional questions even to define what the petitioner complains of — in Sawyer’s case, for example, “Caldwell error.” Moreover, the court does not risk the awkward outcome of establishing a new rule in a case where it has no application. See Teague, 109 S.Ct. at 1077-78. The rule relied upon — for example, the rule governing Caldwell error — exists by the time the Teague issues arise in connection with a particular prisoner’s petition.
Indeed, Sawyer’s argument illustrates the difficulties that may arise from an attempt to separate Teague analysis from the substance of the constitutional claims raised. Whether Caldwell is a new rule, and whether Caldwell is a rule “implicit in the concept of ordered liberty” that implicates factual innocence, both depend in part upon what Caldwell means, and, more specifically, upon the relation between Caldwell and Donnelly. This dependence is made unmistakably clear by Louisiana’s briefing of the Teague issue, which suggests that Teague is no bar to Sawyer’s Caldwell claim precisely because Sawyer is wrong about the relation between Caldwell and Donnelly. If the Supreme Court had made clear that Teague necessarily bars an inquiry into the merits of the petitioner’s constitutional claims, we would perhaps have to resolve the Teague issues by a conditional discussion of Teague’s application to what Sawyer says Caldwell might mean. Such a conjectural analysis of possible rules would, however, entail considerable awkwardness, do nothing to clarify the substantive law, and defeat rather than serve judicial economy — which would be the ostensible goal of any version of Teag-ue that preempted some constitutional inquiries.
We thus choose to address the merits of Sawyer’s interpretation of Caldwell before applying Teague to Caldwell. We do not mean, however, by adopting this strategy to suggest that Teague never bars inquiry into the constitutional merits of a petitioner’s claim. It remains possible that an application of Teague to a conjectural rule may be appropriate in cases where the Teague issues do not turn, as they do here, upon a highly precise specification of the rule in question. We leave that issue for a case in which it is properly presented, and turn to the merits of the constitutional arguments.
*1282IV
At a general level, Caldwell’s import is clear. Regardless of whether the Court moves toward or away from the McGautha acceptance of juror discretion, the sentencing jury must continue to feel the weight of responsibility so long as it has responsibility. Lifting the sense of responsibility frustrates the core contribution of the jury and the cardinal justification for its role. For the jury to see itself as advisory when it is not, or to be comforted by a belief that its decision will not have effect unless others make the same decision, is a frustration of the essence of the jury function. It is not surprising then that jury arguments calculated to have that effect have long been condemned by numerous jurisdictions. See Caldwell, 105 S.Ct. at 2642 nn. 4 & 5. See also Mello, Taking Caldwell v. Mississippi Seriously, 30 B.C.L.Rev. 283, 305-308 & nn. 100-114 (1989). The decision of the Court in Caldwell reflects this reality, insight born more of experience than of empirical study or abstract exposition.
In no way is the importance of Caldwell error diminished by the possibility that a state may dispense with the jury’s sentencing power in capital cases. See Spaziano v. Florida, 468 U.S. 447, 104 S.Ct. 3154, 82 L.Ed.2d 340 (1984). The evil of Caldwell-type prosecutorial arguments is not that they divest juries of their responsibility, but rather that they distort the jury’s understanding of a power which it in fact retains. The decision-maker empowered to choose between life and death must not be relieved of the gravity attending that choice. Whether a judge or jury decides the sentence, the responsibility to decide must remain adjoined to the power to decide. It would, of course, be less likely that a prosecutor could mislead a judge, whose own knowledge of the law should overcome any misleading argument. But a judge who misunderstands the sentencing decision in a capital case creates a Constitutional defect no less significant than a jury which misunderstands its decision. Cf. Hickerson v. Maggio, 691 F.2d 792, 794-95 (5th Cir.1982).
The argument between Sawyer and Louisiana does not draw into question these general observations. Sawyer contends that Caldwell, recognizing the unique role of the jury in capital sentencing, imposes an especially stringent procedural safeguard by requiring that the defendant receive a new sentencing hearing if the prosecutor’s argument had any effect on the jury’s perception of its own responsibility. Louisiana concedes the impropriety of pros-ecutorial argument that misleads the jury as to its role, but contends that the sentencing phase is marred by a constitutional defect only if the prosecutorial argument rendered it “fundamentally unfair.” According to Louisiana, Caldwell did not establish a “no effect” test for constitutional error, but simply applied Donnelly's “fundamental fairness” test to the facts of a sentencing hearing. On this argument, Caldwell extends Donnelly to punishment proceedings without altering Donnelly’s rule by any reaffirmation of McGautha’s reflections upon jury responsibility.
It is this argument which brought the case before the en banc court. To resolve it, we must consider Caldwell in some detail. We begin with the facts.
Caldwell killed the owner of a grocery store in the course of a robbery. His lawyers’ plea for mercy at the sentencing phase of his capital murder trial rested on his poverty, troubled youth, and character evidence. His lawyers argued
[Ejvery life is precious and as long as there’s life in the soul of a person, there is hope. There is hope, but life is one thing and death is final. So I implore you to think deeply about this matter. It is his life or death—the decision you’re going to have to make, and I implore you to exercise your prerogative to spare the life of Bobby Caldwell.... I’m sure [the prosecutor is] going to say to you that Bobby Caldwell is not a merciful person, but I say unto you he is a human being. That he has a life that rests in your hands. You can give him life or you can give him death. It’s going to be your decision. I don’t know what else I can say to you but we live in a society where we are taught that an eye for an eye is not the solution.... You are the judges and you will have to decide his fate. It is *1283an awesome responsibility, I know — an awesome responsibility.
Caldwell, 105 S.Ct. at 2637. The argument triggered the following exchanges:
“ASSISTANT DISTRICT ATTORNEY: Ladies and gentlemen, I intend to be brief. I’m in complete disagreement with the approach the defense has taken. I don’t think its fair. I think it’s unfair. I think the lawyers know better. Now, they would have you believe that you’re going to kill this man and they know— they know that your decision is not the final decision. My God, how unfair can you be? Your job is reviewable. They know it. Yet they ...
“COUNSEL FOR DEFENDANT: Your Honor, I’m going to object to this statement. It’s out of order.
“ASSISTANT DISTRICT ATTORNEY: Your Honor, throughout their argument, they said this panel was going to kill this man. I think that’s terribly unfair.
“THE COURT: Alright, go on and make the full expression so the Jury will not be confused. I think it proper that the jury realizes that it is reviewable automatically as the death penalty commands. I think that information is now needed by the Jury so they will not be confused. “ASSISTANT DISTRICT ATTORNEY: Throughout their remarks, they attempted to give you the opposite, sparing the truth. They said ‘Thou shalt not kill.’ If that applies to him, it applies to you, insinuating that your decision is the final decision and that they’re gonna take Bobby Caldwell out in the front of this Courthouse in moments and string him up and that is terribly, terribly unfair. For they know, as I know, and as Judge Baker has told you, that the decision you render is automatically reviewable by the Supreme Court. Automatically, and I think it’s unfair and I don’t mind telling them so.”
Id. at 2637-38. A divided Mississippi Supreme Court affirmed and the Supreme Court granted certiorari. Speaking for the Court, Justice Marshall concluded that “it is constitutionally impermissible to rest a death sentence on a determination made by a sentencer who has been led to believe that the responsibility for determining the appropriateness of the defendant’s death rests elsewhere.” Caldwell, 105 S.Ct. at 2639. He explained that the court’s post-Furman review of state procedures “has taken as a given that capital sentencers would view their task as the serious one of determining whether a specific human being should die at the hands of the State.” Id. at 2640. He then found “specific reasons to fear substantial unreliability as well as bias in favor of death sentences when there are state-induced suggestions that the sentencing jury may shift its sense of responsibility to an appellate court.” Id.
The State proposed three reasons why the prosecutor’s argument should not upset the death sentence. The State argued that under California v. Ramos, 463 U.S. 992, 1001-06, 103 S.Ct. 3446, 3453-56, 77 L.Ed.2d 1171 (1983), it was free to instruct juries in capital cases about appellate processes. In part IV(a) of the Caldwell opinion, joined only by Justice Brennan, Justice Blackmun and Justice Stevens, Justice Marshall rejected this argument. He concluded that, unlike in Ramos, the argument in Caldwell was not relevant to a valid state penological interest and was misleading. In the Caldwell plurality’s view, appellate review was simply not relevant to the juror’s task of determining an appropriate sentence. For that reason, the prosecutor’s argument that the jurors should view themselves as only taking a preliminary step in the sentencing determination served no valid state interest. Justice O’Connor’s concurring opinion agreed, but refused to read Ramos “to imply that the giving of nonmisleading and accurate information regarding the jury’s role ... is irrelevant to the sentencing decision.” Id., 105 S.Ct. at 2646 (O’Connor, J., concurring; emphasis in original). In her view the prosecutor’s argument was impermissible because it misled “in a manner that diminished the jury’s sense of responsibility.” Id.
The Court next rejected the state’s contention that the prosecutor’s argument was a reasonable response to defense counsel’s argument. The Court observed that the prosecutor’s reference to appellate review did not respond to defense counsel’s suggestion that a sentence of life would be *1284without parole, nor to the defense’s religious theme and plea for mercy.
Finally, and most importantly for our purposes, the Court rejected the State’s contention that in any event the effect of the prosecutor’s argument should be measured by the standard of Donnelly v. De-Christoforo, which would judge improper prosecutorial arguments to vitiate a sentencing proceeding only if they rendered the proceedings fundamentally unfair. The Court distinguished Donnelly on two grounds. First, the Court pointed out that in Donnelly the trial court gave a strong curative instruction to the jury, while in Caldwell the judge not only gave no correcting instruction but “stated to the jury that the remarks were proper.” Id., 105 S.Ct. at 2645. Second, in Donnelly the remarks were ambiguous and not focused pointedly upon “ ‘the principal concern’ of our jurisprudence concerning the death penalty, the ‘procedure by which the State imposes the death sentence.’ ” Id. (quoting California v. Ramos, 463 U.S. at 999, 103 S.Ct. at 3452).
Justice Rehnquist, joined by Justice White, dissented, contending that when the argument was placed in its full trial setting it “fell far short of telling the jury that it would not be responsible for imposing the death penalty.” 105 S.Ct. at 2649 (Rehnquist, J., dissenting). Rather, “the thrust of the prosecutor’s argument was that the jury was not solely responsible for petitioner’s sentence.” Id. at 2650 (emphasis in original). He observed that under Ramos there was nothing wrong with telling a jury that its decision is subject to appellate review, and that the prosecutor did not mislead the jury by suggesting that its decision would be subject to de novo review.
The division between the Caldwell majority and the dissenting Justices, like the division between Sawyer’s argument and Louisiana’s argument, turns in significant part upon the fate of Donnelly’s “fundamental fairness” formula in capital sentencing proceedings. As we shall see, the effect upon a death sentence of Caldwell error and the nature of the inquiry into whether it exists, including the record sources to be examined, are entwined parts of its very definition. That is, what a reviewing court is to look for and how it is to set about judging its effect upon a criminal conviction is part of the definition of Caldwell error. Much of the argument here is over the ingredients of the prohibition.
Sawyer, as we have said, argues that Caldwell modifies Donnelly by mixing in traces of the regard for jury decision-making so powerfully articulated in McGautha. Sawyer argues that the prosecutor’s argument at the sentencing phase of his trial misled the jury regarding its role. In particular, he contends that the argument unambiguously told the jury that its role was only to recommend punishment and that others would check their decision, an argument even more pointed than in Caldwell. Sawyer maintains that such an argument effectively renders a proceeding fundamentally unfair by definition, and that the standard of Donnelly is therefore inapplicable because superfluous. It follows, he argues, that he is entitled to a new sentencing hearing before a jury properly aware of its responsibility. According to Sawyer, neither a contemporaneous objection nor participation by the trial judge are prerequisites to a Caldwell claim. Caldwell mandates a new sentencing hearing so long as the court reviewing Caldwell error “cannot say that [the prosecutor’s statements] had no effect on the sentencing decision.” Caldwell, 472 U.S. at 328-29, 105 S.Ct. at 2639-40. Sawyer says that in Kirkpatrick v. Blackburn, 777 F.2d 272, 289-90 (5th Cir.1985), this court declared that “the no effect test applies to the state’s effort to minimize the jury’s sense of responsibility, not to every other improper argument.” He maintains that the Supreme Court in Darden v. Wainwright adopted this court's position holding Caldwell applicable in any case where the prosecutor “mislead[s] the jury as to its role in the sentencing process in a way that allows the jury to feel less responsible than it should for the sentencing decision.” Darden, 477 U.S. 168, 184 n. 15, 106 S.Ct. 2464, 2473 n. 15, 91 L.Ed.2d 144 (1986).
As already mentioned, Louisiana contends, in essence, that Caldwell merely ap*1285plies Donnelly to a case where the combination of prosecutorial and judicial action at a sensitive moment rendered the proceedings especially unfair to the defendant. Louisiana argues that no new sentencing hearing should be ordered unless we find both that there was Caldwell error and that it rendered the trial fundamentally unfair. Pointing to Darden v. Wainwright, the State argues that Sawyer must show prosecutorial misconduct that “so infected the trial as to deny due process.” See Darden, 106 S.Ct. at 2472 (quoting Donnelly v. DeChristoforo). Louisiana argues that the due process standard is applicable because the prosecutor’s argument, when stripped of non-misleading statements, was not as clear and focused as in Caldwell. Louisiana stresses the absence both of any objection by defense counsel and of any signal from the trial judge that might have endorsed the prose-cutorial misstatement.
We agree with Sawyer that Caldwell must be read in light of McGautha. The state cannot resist a conclusion that it improperly diminished a jury’s sense of responsibility in its sentencing role with the argument that a jury with such diminished responsibility nonetheless did not render the proceedings fundamentally unfair. See, e.g., Coleman v. Brown, 802 F.2d 1227, 1238-41 (10th Cir.1986), cert. denied, 482 U.S. 909, 107 S.Ct. 2491, 96 L.Ed.2d 383 (1987); see also Campbell v. Kincheloe, 829 F.2d 1453, 1460-61 (9th Cir.1987), cert. denied, — U.S. -, 109 S.Ct. 380, 102 L.Ed.2d 369 (1988); Dutton v. Brown, 812 F.2d 593, 596-97 (10th Cir.1987) (en banc), cert. denied, 484 U.S. 836, 108 S.Ct. 116, 98 L.Ed.2d 74 (1987); Mann v. Dugger, 844 F.2d 1446, 1457-58 (11th Cir.1988) (en banc). Cf. Hopkinson v. Shillinger, 866 F.2d 1185, 1226-33 (10th Cir.1989); id. at 1233-38 (Logan, J., dissenting). Once it is accepted that a death sentence by a jury with such a diminished sense of responsibility is “fundamentally incompatible with the Eighth Amendment requirement that the jury make an individualized decision that death is the appropriate punishment in a specific case”—and the Supreme Court has told us precisely that, see Darden, 106 S.Ct. at 2473 n. 15,—it is apparent that, as Sawyer contends, the Donnelly issue of fundamental fairness is subsumed in the threshold question of whether there was Caldwell error.
If the state has misled the jury in the manner condemned by Caldwell, it can be no answer that the culprit was the prosecutor and not the judge. With either source, the error is the same. Although in Caldwell there was an objection and a potent affirmation of the misleading argument by the trial judge, the relevance of these events was to the question of whether the jury was actually misled. In other words, the absence of objection and trial judge participation are highly relevant to the question of whether a jury was misled, but their absence is not determinative as a matter of law of the question of whether the state misled the jury. We do not read the Court’s opinion in Darden to the contrary.
“To establish a Caldwell violation, a defendant necessarily must show that the remarks to the jury improperly described the role assigned to the jury by local law.” Dugger v. Adams, — U.S. -, 109 S.Ct. 1211, 1215, 103 L.Ed,2d 435 (1989). In short, a prosecutor’s statements to the jury accurately describing its role will not support a Caldwell claim. At the same time, a statement can be literally true but quite misleading by failing, for example, to disclose information essential to make what was said not misleading. Indeed much of our law of fraud under the Securities Act rests on just such a reality. See 17 C.F.R. 240.10b-5(b).
It is suggested that, in spite of these considerations, a willingness to find Caldwell error from unobjected to argument by a prosecutor unwisely creates an incentive for defense counsel to not object. After all, an objection may lead to a curative instruction and any appellate point is not lost by remaining silent. The questionable validity of the assumed incentives aside, these concerns as well as the other values that lie behind our usual insistence that error be preserved are not unique to Caldwell error. The essence of the doctrine of plain error is that a loss of funda*1286mental rights outweighs the values behind rules insisting upon an objection. More to the point, the decision to entertain claimed constitutional error without a contemporaneous objection belongs in the first instance to the state, when as here, we review a state court conviction. A state may insist upon a contemporaneous objection. And, ordinarily, a federal habeas court is bound by that decision and cannot reach claims of error found by the state to have been waived. Dugger v. Adams, 109 S.Ct. at 1215. In short, whether to insist upon a contemporaneous objection as a matter of orderliness, as distinguished from the question of whether an objection is an element of the constitutional claim itself, is a matter for the state court.
It is suggested that even if the Caldwell issue must be addressed because the state reached its merits, a contemporaneous objection is an element of a Caldwell claim. We have concluded that a timely objection is an essential element of a claim of racial discrimination in the exercise of preemptory challenges under Batson v. Kentucky, 476 U.S. 79, 106 S.Ct. 1712, 90 L.Ed.2d 69 (1986). Jones v. Butler, 864 F.2d 348, 369 (5th Cir.1988) (on petition for rehearing). But the constitutional rule in Batson rests on a change in the requirement of proof from that of Swain v. Alabama, 380 U.S. 202, 85 S.Ct. 824, 13 L.Ed.2d 759 (1965) (insisting upon proof of a pattern of discrimination by prosecutors in cases) to the case specific procedures of Batson. Teague, 109 S.Ct. at 1066. Bat-son assures an objecting defendant that a prosecutor striking black veniremen will articulate non-racial reasons for its decisions. An objection is plainly central to a Batson claim. Caldwell, by contrast, rests on “the assumption that a capital sentencing jury recognizes the gravity of its task and proceeds with the appropriate awareness of its ‘truly awesome responsibility.’ ” Caldwell, 105 S.Ct. at 2646. It instructs that if the State seeks “to minimize the jury’s sense of responsibility for determining the appropriateness of death,” and “we cannot say that this effort had no effect on the sentencing decision,” then “that decision does not meet the standard of reliability that the Eighth Amendment requires.” Id. In Caldwell, unlike in Batson, the constitutional defect — if it exists — is observable and measurable by a reviewing court even absent any objection. We reject the suggested analogy between these two very different doctrines.
In sum, we reject Louisiana’s proffered definition of Caldwell. We do so after noting that its core is diminishing the responsibility of the jury by misdescribing its role under state law and after rejecting the suggestion that its elements include showings of fundamental unfairness, a contemporaneous objection or trial court participation.
Continuing our definition of Caldwell error, we turn to the question of what an appellate court looks to in gauging the state’s conduct, and quickly find that the nature of the prohibition takes us a long way toward the answer. What has been communicated to the jury by the state cannot be disentangled from the total trial scene, and thus that is our terrain. While the prosecutor’s argument will often be the natural point of departure, we must turn to the opposing argument and then to instructions of the court, both in its formal charge and in any rulings on objections. The initial focus will be upon the close of the sentencing hearing, yet inquiry may proceed not only to the guilt phase but to jury selection as well. In short, a trial cannot be cabined into distinct segments. As the Supreme Court phrased it: “not only is the challenged instruction but one of many such instructions, but the process of instruction itself is but one of several components of the trial which may result in the judgment of conviction.” Cupp v. Naughten, 414 U.S. 141, 147, 94 S.Ct. 396, 400, 38 L.Ed.2d 368 (1973). We conclude that the inquiry is whether under all facts and circumstances, including the entire trial record, the state has misled the jury regarding its role under state law to believe that the responsibility for determining the appropriateness of defendant’s death rests elsewhere.
While this is inevitably a case-by-case inquiry with a broad terrain to be surveyed, there are a number of events that *1287obviously may loom large and quickly focus the inquiry. First, the trial judge is an extraordinarly puissant figure. A direct and uncorrected misstatement to the jury that misleads the jury regarding its role will be difficult to salvage. For example, Caldwell error was found by the Eleventh Circuit when a trial judge told the jury that he was the ultimate determinant of whether the defendant was sentenced to death. The Circuit reached this conclusion even though the jury’s role under Florida law is advisory. Adams v. Wainwright, 804 F.2d 1526, 1532-33 (1986), modified on denial of rehearing, 816 F.2d 1493 (1987), rev’d on other grounds, Dugger v. Adams, — U.S. -, 109 S.Ct. 1211, 103 L.Ed.2d 435 (1989). Second, the absence of objection by competent counsel may suggest that the argument as it played in the courtroom was less pointed than it now reads in the transcript. Third, the argument may take on a different hue when read as a reply to opposing counsel. Fourth, the court may have mitigated the effect of counsel’s argument by instructing the jury that the judge is the sole source of the law and that the lawyer’s arguments are not evidence. Fifth, veniremen often receive extensive instruction during voir dire. These instructions, as well as the questions and advices of counsel, are also relevant. Finally, through the course of trial the judge may give detailed instructions to the jury about its role. Such familiar instructions are part of the message to the jury and all must be considered. We list these lines of inquiry to explain the scope of inquiry that may be required in review of asserted Caldwell error, without suggesting that the list is exhaustive. By definition, it is not and cannot be. Indeed, in some cases the presence or absence of error will be readily determinable solely on the basis of the prosecutor’s argument and the trial judge’s treatment of it.
V
A
Sawyer’s conviction was final at least by 1984 when the Supreme Court denied his petition for certiorari. See Sawyer v. Louisiana, 466 U.S. 931, 104 S.Ct. 1719, 80 L.Ed.2d 191 (1984). Because Sawyer wishes to rely on the Court’s later decision in Caldwell, he must grapple with the limitation of Teague. Sawyer first argues that Teague does not bar his argument because Caldwell did not announce a new rule, so that the prosecutor’s argument was constitutionally infirm measured by the law in place in 1984 when his conviction became final.
The Supreme Court’s decision in Penry, left the definition of a “new rule” in some doubt. Justice O’Connor reiterated her statement, first presented in Teague, that a case “announces a new rule when it breaks new ground or imposes a new obligation on the States or the Federal Government, [or,] to put it differently ... if the result was not dictated by precedent.” Penry, 109 S.Ct. at 2944 (quoting Teague, 109 S.Ct. at 1070 (plurality opinion). Yet Justice O’Con-nor’s application of this standard led Justice Scalia, joined by three colleagues, to contend that the Court had only given “lip-service” to the Teague standard. Penry, 109 S.Ct. at 2964 (opinion of Scalia, J., dissenting; Part II). Justice Scalia said that “it challenges the imagination to think that today’s result is ‘dictated’ by our prior cases.” Id. at 2965. He went on to say that “[i]f Teague does not apply to a claimed ‘inherency’ as vague and debatable as that in the present ease, then it applies only to habeas requests for plain overruling,” and went so far as to remark that “[it] is rare that a principle of law as significant as that in Teague is adopted and gutted in the same term.” Id. at 2965.
Justice Scalia’s comments are especially significant because he speaks on behalf of all three Justices who joined Justice O’Con-nor’s plurality opinion in Teague, and on behalf of Justice White as well. Yet, Justice Brennan, in his separate Penry opinion, apparently does not agree with Justice Scalia that Teague has been gutted. Justice Brennan reiterates his contention, first made in his dissent from Teague itself, that the Teague rule is an “unprecedented curtailment of the reach of the Great Writ,” and accuses the majority of compounding its errors by extending Teague to death eases.
*1288Indeed, Justice O’Connor’s application in Penry of Teague’s, “new rule” formula may well have turned upon facts which she thought unique to Penry’s claims. In Justice O’Connor’s view, Penry sought only to compel Texas “to fulfill the assurance upon which [Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976)] was based: namely, that the special issues would be interpreted broadly enough to permit the sentencer to consider all of the relevant mitigating evidence a defendant might present in imposing sentence.” 109 S.Ct. at 2945. Penry’s claim rested on the clearly established and specific Constitutional rule that “a State could not, consistent with the Eighth and Fourteenth Amendments, prevent the sentencer from considering and giving effect to evidence relevant to the defendant’s background or character or to the offense that mitigates against imposing the death penalty.” Justice O’Connor concluded that the path from Jurek to Penry involved the consistent application of an established constitutional rule to, in essence, changes in the facts.
Because of these disagreements about the meaning of the Teague test, the Court’s opinions in Teague and Penry do not immediately yield a clearly articulable definition of a “new rule.” We must interpret what Justice O’Connor has said by reference to the purposes served by the Teague rule. To undertake that inquiry, we first turn to the complex of concerns now accommodated within federal habeas jurisprudence.
A federal court’s role in a habeas attack on a state court conviction is only to review for errors of constitutional magnitude. The Constitution commands us to defer to federalism, and so recognizes that the solemn judgment of a state’s highest court enjoys a presumption of validity, which may be overcome only for failure to abide by the Constitution itself. The role that remains for federal courts is by no means modest. To the contrary, viewed over the full span of history, it is rather an extraordinary reach for superintending power. Indeed, the first legislation empowering federal courts to issue a writ for state custody did not come until the Habeas Act of 1867. Until the Court’s decision in Brown v. Allen, 344 U.S. 443, 73 S.Ct. 397, 97 L.Ed. 469 (1953), “federal courts would never consider the merits of a constitutional claim raised on habeas if the petitioner had a fair opportunity to raise his arguments in the original proceeding....”5 Seen in this light, casting our role as that of a constitutional backstop is hardly a retrenchment, and Teague’s reach for finality is modest indeed.
Teague, whether applied to a capital sentence or to a more ordinary case, is by no means a return to the law that preceded Brown v. Allen, if indeed it is a turn in that direction at all. Teague rather reflects a distinct and basic judgment that, putting aside the cases falling within its two provisos, there is no fundamental unfairness inherent in refusing to wield federal power to upset state court convictions and sentences of death arrived at in complete conformity to constitutional standards in place when the convictions became final. Due regard for the constitutional structure of federalism, and the protection it accords to state government, counsels the opposite — that only preservation of constitutional principles justifies the intrusion.
The Teague judgment about the federal role acknowledges that neither finality nor federalism will condone constitutional acquiescence in the conviction of persons factually innocent of the crime charged. Our efforts to reduce the risk of convicting an innocent person are evidenced by myriad procedural safeguards and by high requirements of proof. These restrictions reflect a commitment to accurate outcomes so firm that we consciously increase the chance of acquitting guilty persons to reduce the chance of convicting the innocent. It is not surprising, then, that the Supreme Court is fairly unanimous in its view that a state court prisoner can rely upon a fundamental constitutional rule implicating factual innocence even though *1289that rule was not announced until after his conviction became final.
It might nonetheless be contended that the Court’s “factual innocence” proviso is not enough to vindicate the rights of prisoners, and that capital cases show particularly well various considerations that compel a narrow formulation of Teague’s “new rule” element. One reasoning along these lines might point to the inherent finality of the death penalty, and contend that the benefit of every announced constitutional rule should be given to a prisoner facing this extreme penalty. One might likewise argue that in habeas petitions challenging a death sentence but not the underlying conviction, the state need not fear that it will have to relitigate issues of innocence and guilt on the basis of stale evidence, and so run the risk of freeing a criminal who would have been convicted by a fair and timely trial. Finally, continuing to reason against finality interests on the basis of concerns unique to death cases, one might argue that in such cases there is no danger that the state’s efforts at rehabilitation will lose their focus because of the habeas process; that habeas petitioners succeed more frequently in capital cases than in other cases; and that other factors, external to the habeas system, are responsible for delays in the execution of state prisoners.
Yet unless we suppose a perfectly stable constitutional jurisprudence, it is unclear how finality could ever be achieved if these arguments are accepted at full reach. As the Court made clear in Penry, the order of magnitude of punishment is not relevant to Teague’s support of finality so long as we except rules implicating factual innocence. The “death is different” argument in this context is little more than an argument against the validity of the punishment itself. As an argument directed to the purposes of Teague — the matter now before us — it fails.
Of course, the penalty is different from all others in many respects. We recognize that it is the extreme of punishments when we reserve the punishment for the most extreme of crimes, as we do under our present law. Death sentences, which by their nature aim at retribution or deterrence and not at rehabilitation, obviously do implicate different state purposes than do terms of incarceration. But that the interests are different does not imply that they are less deserving of federal deference, or that comity concerns are any less important. A state policy predicated upon the certainty of exact retribution, no less than a state policy predicated upon incarceration in a facility designed in part to rehabilitate, suffers when the prospect of punishment is confused by a series of collateral federal attacks.
Indeed, much that is unique about the law controlling death cases is in fact a powerful testament to the need for the finality-serving rules of Teague. The constitutionally secured rules announced for death cases by the Supreme Court since McGautha v. California, 402 U.S. 183, 91 S.Ct. 1454, 28 L.Ed.2d 711 (1971), have come in such number and with such rapidity that the entire jurisprudence is fairly described as being in a state of flux. During the ten year period ending with the final day of the Supreme Court’s 1988 term, it granted plenary review in sixty-seven cases and at least thirty-five of those can, with little dissent, be described as presenting issues of substantial reach. The destabilizing impact of such a sea-change in controlling law presents problems of administration unique to death cases. In the 1986 term alone, the Supreme Court acted on eighty requests for stay of execution. This undermines the argument that Teague has no application to death cases.
Nor is there anything inhumane in an insistence that a death-sentenced state prisoner confine his attack upon that sentence to the rules in effect when his conviction became final. So long as nothing new implicates the petitioner’s factual innocence, we, confronted with the need for sureness of punishment as contrasted with the never ending uncertainty and serendipitous state of a nigh open set of rules, see little to persuade us that respect for human dignity counsels against application of finality rules.
*1290In light of the powerful reasons that justify the Teague doctrine, we see no cause to limit its application to the rare or extraordinary case. When a rule is indeed dictated by precedent — a word Justice O’Connor took care to emphasize in Penry as she did in Teague — then a state can reasonably be asked to anticipate its articulation, and enforcing the rule in a habeas proceeding will not intrude upon the state’s legitimate interest in the finality of convictions. Otherwise, however, Teague must bar the rule’s application. We do not, despite Justice Scalia’s strong words in dissent, read Penry to the contrary. Instead, we believe that Justice O’Connor regarded Penry as a special case, one simply reapplying the rule of Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950, 49 L.Ed.2d 929 (1976) to a unique development in state law. See Penry, 109 S.Ct. at 2945 (Opinion of O’Con-nor, J., Part II-B, discussing Jurek). Justice O’Connor honored the language of the Teague opinion, and we must assume she intended to honor its spirit as well.
B
Sawyer correctly observes that many state courts, including Louisiana, had before Caldwell developed common law rules forbidding misleading jury argument about the importance of the jury’s decision. See, e.g., Pait v. State, 112 So.2d 380, 383-84 (Fla.1959); Blackwell v. State, 76 Fla. 124, 79 So. 731, 731, 735-36 (1918); Wiley v. State, 449 So.2d 756, 762 (Miss.1984); State v. Jones, 296 N.C. 495, 251 S.E.2d 425, 427-29 (1979); State v. Gilbert, 273 S.C. 690, 258 S.E.2d 890, 894 (1979); Hawes v. State, 240 Ga. 327, 240 S.E.2d 833, 839 (1977); People v. Morse, 60 Cal.2d 631, 36 Cal.Rptr. 201, 211-12, 388 P.2d 33, 43-44 (1964); State v. Mount, 30 N.J. 195, 152 A.2d 343, 351-52 (1959); People v. Johnson, 284 N.Y. 182, 30 N.E.2d 465, 467 (1940). For example, the Louisiana Supreme Court ordered a new sentencing hearing in a 1982 capital case when, without objection, the prosecutor argued to the jury that the “buck” started with them and did not end there, and that “everything” will more than likely be reviewed by every appeals court in the United States. State v. Willie, 410 So.2d 1019, 1034-35 (La. 1982). But it does not necessarily follow from the circumstance that Louisiana law forbade the argument made by Sawyer’s prosecutor when it was made, that the Supreme Court did not announce a “new” rule in Caldwell.
Sawyer’s argument fails to deal with Teague’s, explicit offer of Ford v. Wainwright, 477 U.S. 399, 106 S.Ct. 2595, 91 L.Ed.2d 335 (1986), as an example of a decision that created a “new rule.” Ford held that the Eighth Amendment prohibits states from inflicting the penalty of death on an insane prisoner. Execution of the insane was prohibited at common law. Indeed, the Ford Court observed that “[t]he bar against executing a prisoner who has lost his sanity bears impressive historical credentials ...” 477 U.S. at 406, 106 S.Ct. at 2600. Moreover, twenty-six states of the forty-one with a death penalty had “statutes explicitly requiring the suspension of the execution of a prisoner who meets the legal test for incompetence.” Id. at 408, n. 2, 106 S.Ct. at 2601, n. 2. Sawyer’s reliance upon Caldwell’s common law roots and its relationship to the rules of several states, including Louisiana, cannot be squared with the Supreme Court’s view that it created a new rule in Ford v. Wainwright.
There was no federal jurisdiction over Sawyer’s present claim until this type of error gained a constitutional footing. Caldwell was certainly new in its conclusion that such arguments violated the Eighth Amendment. See Mello, 30 B.C.L. Rev. at 305. Of course, when Sawyer’s conviction became final, Donnelly would have authorized federal jurisdiction over a habeas petition attacking the sentence on due process grounds. Yet, as we have seen, Sawyer persuasively contends that Caldwell was more than a straightforward application of Donnelly to new facts. Our question is then whether these changes suffice to make Caldwell a new rule within the meaning of Teague.
The Teague court held that “a case announces a new rule when it breaks new ground or imposes a new obligation on the States or the Federal Government ... [t]o *1291put it differently, a case announces a new rule if the result was not dictated by precedent existing at the time the defendant’s conviction became final.” Teague, 109 S.Ct. at 1070 (emphasis in original). Accord, Penry, 109 S.Ct. at 2944. We have little difficulty in concluding that so measured, Caldwell’s greatly heightened intolerance of misleading jury argument is a new rule within the meaning of Teague. Its direct impact upon the finality of state convictions is illustrated by this ease.
Sawyer, however, nonetheless contends that this heightened standard for review of prosecutorial argument does not create a new rule. He has two arguments. First, Sawyer says that Louisiana not only condemned Caldwell-type prosecutorial argument, but did so under an Eighth Amendment standard identical to the Caldwell standard. For this proposition, Sawyer cites a string of Louisiana cases explaining that Louisiana’s death penalty procedures were designed to comply with the Supreme Court’s Eighth Amendment decisions, see, e.g., State v. Payton, 361 So.2d 866, 870-73 (La.1978), State v. Sonnier, 379 So.2d 1336, 1370 (La.1980), and State v. Willie, 410 So.2d 1019, 1032-33 (La.1982), and another string of Louisiana cases condemning Caldwell-type prosecutorial arguments, see, e.g., State v. Berry, 391 So.2d 406, 418 (La.1980), cert. denied, 451 U.S. 1010, 101 S.Ct. 2347, 68 L.Ed.2d 863 (1981); Willie, 410 So.2d at 1034-35, and State v. Robinson, 421 So.2d 229, 231-34 (La.1982).
Yet it is one thing to say that a state, inspired by earlier constitutional decisions, had anticipated Caldwell as a matter of state law, and a very different matter to say that the state had recognized a Caldwell-type rule as a constitutional restriction on its own power. In an effort to bring the Louisiana cases within the latter category, Sawyer relies heavily on the Louisiana Supreme Court’s recent post-Caldwell decision in State ex rel. Busby v. Butler, 538 So.2d 164, 173 (La.1988). There, the Louisiana Court said it need not consider a Caldwell claim on collateral attack after rejecting a similar state law challenge on direct review, for Caldwell “did not change our previous case law.” Again, however, this statement and the other remarks in Busby indicate only that the state law and Caldwell rules are coincident. The remarks do not show that Louisiana law condemned Caldwell argument because it regarded such argument as an Eighth Amendment violation. We therefore need not decide whether, if a rule is “new" as a matter of constitutional interpretation but not “new” in state interpretations of the federal Constitution, it is nonetheless “new” for purposes of the Teague bar upon collateral federal challenges to state convictions.
Sawyer next contends that this Circuit in Moore v. Blackburn, 774 F.2d 97, 98 (5th Cir.1985), cert. denied, 476 U.S. 1176, 106 S.Ct. 2904, 90 L.Ed.2d 990 (1986), has already decided that Caldwell is not a new rule. In Moore, we held that even if the Caldwell standard were separable from the Louisiana state standard for assessing prosecutorial argument, petitioner Moore should have anticipated in an earlier habeas petition the possibility of a distinct constitutional standard. We therefore held that Moore’s Caldwell argument was not “new” for purposes of the writ abuse doctrine, and stated that the doctrine would bar the argument. Sawyer’s attempt to rely on Moore must fail, for the meaning of “newness” differs in writ abuse cases from its meaning in Teague cases. In writ abuse cases, the key question is whether a particular argument is being made by attorneys: the argument is not “new” if it is being made, and so should be known to attorneys. The Supreme Court makes clear, however that a rule is new for purposes of Teague if it has not been accepted at the time the petitioner’s conviction became final. Teague, 109 S.Ct. at 1070. Moore thus cannot bear the freight Sawyer would put on it.
C
The Teague test allows two exceptions. Sawyer, however, cannot contend that the sentence imposed upon him was unlawful because the conduct for which he was charged is constitutionally privileged, or that he is among a class of persons protected against execution. He contends only that the state’s sentencing procedure was *1292unconstitutionally administered. Teague’s first exception, dealing with substantive limitations upon the criminal law-making authority, therefore does not apply.
We turn, then, to the plurality’s insistence in Teague that a new rule may be relied upon by a habeas petitioner if it both “requires the observance of those procedures that ... are implicit in the concept of ordered liberty” and “procedures without which the likelihood of an accurate conviction is seriously diminished.” Sawyer contends that we should not in his case apply the “accurate conviction” qualification to Harlan’s proviso. First, Sawyer argues that the qualification is only a plurality view. The Penry opinions did not discuss the “fundamental to ordered liberty” proviso, or the “actual innocence” qualification to it. However, Justice White’s joinder in Justice Scalia’s dissent, and in Part II-A of Justice O’Connor’s opinion (which references the Teague plurality’s formulation of the exceptions), strongly suggests that Justice White has adopted the position of the Teag-ue plurality. In any event, our short answer is that pending further direction from the Supreme Court, and in particular the full view of Justice White, we should follow the course set by the plurality as best we can.
Second, Sawyer argues that confining use of new rules to those implicating factual innocence has no relevance to a jury’s decision to impose a death and not a life sentence. We are not persuaded. A habeas petitioner may not escape this limitation on use of a new rule by confining his attack to the jury’s decision to impose a death rather than life sentence. Rather, such a petitioner must show that the new rule insists on procedures without which the correctness of the jury’s decision to punish by death rather than by life imprisonment is seriously diminished.
We thus accept the plurality’s formulation of the proviso, and turn to its application. Our task is made difficult by the newness of the amalgam of the second proviso as well as its uncertain precedential footing. Justice O’Connor’s opinion for the plurality insisted that it is “unlikely that many such components of basic due process have yet to emerge.” 109 S.Ct. at 1077. Justice O’Connor went on to say that these components are “best illustrated by recalling the classic grounds for the issuance of a writ of habeas corpus — that the proceeding was dominated by mob violence; that the prosecutor knowingly made use of perjured testimony; or that the conviction was based on a confession extorted from the defendant by brutal methods” (citations omitted). The new rule contended for in Teague was an extension to petit juries of the fair cross section requirement of a jury venire. The plurality opinion concluded that such a new rule “would not be a ‘bedrock procedural element’ that would be retroactively applied under the second exception_” Id. at 1077.
While the Court has made plain that it expects to encounter few newly discovered bedrock procedural rules, it is not clear how Caldwell, with its condemnation of a particular type of jury argument, fits into the Teague scheme. This difficulty stems in part from uncertainty about Teague’s, standard for sorting the bedrock from the host of other rules calculated to enhance the efficiency and fairness of a trial. We can immediately put aside rules that only enhance as distinguished from rules essential to fundamental fairness. This first cut is informed by developed principles of incorporation doctrine that leave the states free of all but the core assurances, variously expressed as rejecting “tail with the hide” and “jot-for-jot” incorporation. See e.g., Duncan v. Louisiana, 391 U.S. 145, 181, 88 S.Ct. 1444, 1465, 20 L.Ed.2d 491 (1968) (Harlan, J., dissenting). For example, the Fourteenth Amendment requires Louisiana to provide Sawyer a jury and a fundamentally fair trial. Louisiana has wide latitude in its choice of procedures for doing so and few procedures are so essential as to be required by the Fourteenth Amendment. This distinction is reflected in our willingness to find errors to be harmless and our refusal to grant relief absent a demonstration not only that the rule was violated but also that its violation rendered a trial fundamentally unfair.
Caldwell manifestly implicates two principles that would be fundamental in the sense required by Teague’s second proviso. *1293The first is Donnelly’s restriction requiring that a proceeding not be “fundamentally unfair” to the defendant. The second is the more expansive regard for jury discretion suggested by McGautha, a regard trimmed back, as we have mentioned, by the Court’s later interpretations of the Eighth Amendment. Were Sawyer seeking to rely on either of these principles as new rules, his argument would be compelling. Yet Donnelly’s principle is not new by comparison to Sawyer’s conviction, and McGautha’s general themes do not constitute a rule at all. What Sawyer seeks to rely upon is Caldwell’s modification of Donnelly in light of the ideals discussed in McGautha. That modification is not itself so fundamental as to be “implicit in the concept of ordered liberty.” After all, the only defendants who need to rely on Caldwell rather than Donnelly are those who must concede that the prosecutorial argument in their case was not so harmful as to render their sentencing trial “fundamentally unfair.”
A recent decision of the Supreme Court supplies additional guidance for our inquiry. In Dugger v. Adams, the Court decided whether Florida’s procedural default rule barred Adams’s Caldwell claim. To resolve that issue, the Court had to determine whether a “fundamental miscarriage of justice” would result if the procedural default rule were permitted to defeat Adams’s Caldwell claim. The Court held that no such miscarriage of justice would arise.
In reaching its conclusion, the Adams Court wrote as follows:
The dissent “assumes arguendo ” that a fundamental miscarriage of justice results whenever “there is a substantial claim that the constitutional violation undermined the accuracy of the sentencing decision.” ... According to the dissent, since “the very essence of a Caldwell claim is that the accuracy of the sentencing determination has been unconstitutionally undermined,” ... the standard for showing a fundamental miscarriage of justice is necessarily satisfied. We reject this overbroad view. Demonstrating that an error is by its nature the kind of error that might have affected the accuracy of a death sentence is far from demonstrating that an individual defendant probably is “actually innocent” of the sentence he or she received. The approach taken by the dissent would turn the case in which an error results in a fundamental miscarriage of justice, the “extraordinary case,” ... into an all too ordinary one.
109 S.Ct. at 1217-18 & n. 6.
Adams, of course, does not directly control Teague’s application to a Caldwell claim. Adams applies a “fundamental miscarriage of justice” standard to determine whether a Caldwell claim might fit within an exception to the procedural default rule. Teague applies an “implicit in the concept of ordered liberty” and “implicating factual innocence” standard to determine whether a Caldwell claim might fit within an exception to the doctrine barring habeas petitioners from relying on new rules. The verbal formulae are different, and their application thus might differ, too. Moreover, the Adams miscarriage standard requires scrutiny of the facts of a particular case, while the Teague ordered liberty standard looks to the character of the general rule asserted.
Nonetheless, we must take care not to exaggerate the substantive import of these semantic differences. Similar concerns underlie both the procedural default doctrine and the Teague doctrine prohibiting reliance upon new rules. Both doctrines recognize the importance of finality in criminal convictions. Both doctrines promote federal-state comity by requiring federal courts to defer to the integrity of state convictions. And both doctrines put a premium upon the obligation of defendants to raise all relevant arguments before their convictions become final. Indeed, in some respects Teague functions as a radical extension of the procedural default rule by forcing defendants to establish a new rule on direct appeal, rather than on collateral attack, if they wish to rely on such a rule. Because of the similarities between the two doctrines, it is difficult to see why a Caldwell violation should be sufficiently fundamental to require an exception to the “new rule” doctrine, but not so fundamental as *1294to require an exception to the procedural default doctrine.
Adams is also important for another reason: given the particular facts of Adams’s own case, the Court’s disposition of the case presupposes a judgment about the importance of Caldwell error to a sentencing determination. In Adams, as the Court noted, the trial judge “found an equal number of aggravating and mitigating circumstances.” The Court made clear that there was no fundamental miscarriage of justice even though Caldwell error goes to the accuracy of the sentencing procedure, and even though the case was a close one. In short, the mere possibility of a close case did not make the alleged error’s threat to accuracy sufficiently fundamental to warrant exemption from the procedural bar.
Sawyer’s Caldwell claim runs into comparable problems when analyzed in light of the second Teague proviso. Sawyer can argue at most that there would be a possibility, absent the alleged Caldwell violation, of a different outcome to the jury’s sentencing procedure. Yet, as we have already stated, the Court’s Teague opinion makes quite clear that not every procedural rule affecting the accuracy of a trial will fit within the “ordered liberty” proviso. To hold otherwise would be to cling to “jot-for-jot” or “tail-with-the-hide” incorporation, and to make the “extraordinary case into the ordinary one.” Instead, the examples listed by the Teague Court — trial by mob rule, use of perjured testimony, or the extraction of confessions through brutal torture — either so distort the judicial process as to leave one with the impression that there has been no judicial determination at all, or else skew the actual evidence crucial to the trier of fact’s disposition of the case. Here the jury did have an opportunity, even if procedurally flawed, to contemplate and review the relevant evidence. Sawyer’s Caldwell claim has neither the overwhelming influence upon accuracy nor the intimate connection with factual innocence demanded by the second Teague proviso.
Our extended exposition of the nature of Caldwell error reinforces the inferences we draw from the Supreme Court’s decision in Adams. Caldwell error does indeed implicate core aspects of the sentencing procedure. As such, it implicates both the integrity of that procedure and the accuracy of the determination in any particular case. Yet to say that accuracy is implicated is not to say that the defendant is necessarily prejudiced. In fact, Caldwell’s deference to the fundamental character of the jury’s role manifests itself precisely in its refusal to require actual prejudice to the defendant. Caldwell views prosecutorial argument as a basis for reversal if, when viewed within the context of the whole, it had an effect upon the jury’s perception of its role in the sentencing proceeding. It is, of course, unusual to presume the existence of reversible error, on the basis of the prosecutor’s comments, absent any showing of prejudice. This presumption is an important one, and, we would hope, will contribute to the increased integrity and accuracy of criminal procedure in this sensitive area. But none of this makes Caldwell so fundamental, or so connected with factual innocence, as to fit within Teague’s second proviso.
VI
Of course, if Sawyer were able to show actual prejudice, he would be able to proceed under the more general fundamental fairness standard of Donnelly v. De-Christoforo. Yet Sawyer has not contended that such prejudice exists here, and we, after a thorough review of the record, can find none.
We have covered considerable ground about the content of the Caldwell rule. Yet, the dissent falls silent on this set of issues, perhaps because the posture of the case does not require that we apply Caldwell. It is then only on narrow, but crucial grounds, that our opinion is engaged, and to assist its focus we conclude with one observation in reply to the dissenting view.
Judicial tradition demands that new rules find their trace in older ones. This search is near the core of discipline that distinguishes judges from other decision makers. Judges excel at the task. Such artisans possess a very important tool — a gauge of generality. It is no surprise then that *1295Teague’s effort to limit federal habeas by asking whether a rule is new invites those who resist the restriction to reach for their tool kit. At a sufficient level of abstraction there are no new rules. The judicial artisan can start by asserting that the old rule is that a trial must be fundamentally fair, that a defendant is entitled to procedural due process. Stated this generally there have been few if any new rules for the trial of criminal cases.
The dissent does precisely this, resting its assertions on little more than that the old rule is the prohibition of unfair jury argument. Its old law is unnarrowed by any definition of its reach and force such as whether it treats such state conduct as inherently destructive of required fairness or insists on demonstrated prejudice in a given case. Louisiana rejected Sawyer’s claim and Sawyer has no federal habeas claim without Caldwell. Nonetheless, we are told that Caldwell was dictated by precedent, that it broke no new ground and that it imposed no new obligation on the states. We are asked to believe that Caldwell simply applied well established constitutional principles. If to the uninitiated this is dissemblance, unhappily it is to the cognoscenti business as usual. By adroit use of the generality gauge our able dissenting colleagues can breathe superficial credibility into the fable that Caldwell broke no new ground, imposed no new obligation on the states and was dictated by precedent. With all deference, there is afoot here no more than a resistance to the principles of finality adopted by Teague. We should not play such sophistical games. The issue here is whether the federal judiciary will take hold of the open ended character of the habeas remedy it has created. We are persuaded that little or nothing is left of Teague’s promise if the dissent’s view is accepted. We think that this artisan’s destruction of so recent a decision by the Supreme Court should be rejected and we do so. Ultimately only Teague’s authors can tell us if they meant what they said or if they have changed their minds.
For these reasons, we find that Teague bars Sawyer from pursuing his Caldwell claim. We affirm the district court’s decision denying Sawyer’s petition for a writ of habeas corpus.
. Sawyer v. Butler, 848 F.2d 582 (5th Cir.1988).
. This word was likely recorded inaccurately by the stenographer. From context, it is clear that the prosecutor said, “It’s all you're doing." The two phrases sound identical, but their meanings are nearly opposite.
. Justice Blackmun joined Justice Stevens's reservations about Teague's applicability to death cases.
. See Mackey v. United States, 401 U.S. 667, 684, 91 S.Ct. 1160, 1175, 28 L.Ed.2d 404 (1971) (Har-Ian, J., concurring).