dissenting.
In today’s opinion, the court does not answer the one question that should be before it: “Would some reasonable juror believe that Paul House committed first degree murder and should be subject to the death penalty?” See Schlup v. Delo, 513 U.S. 298, 329, 115 S.Ct. 851, 130 L.Ed.2d 808 (1995). A negative conclusion must be supported by “new reliable evidence — whether it be exculpatory scientific evidence, trustworthy eyewitness accounts, or critical physical evidence — that was not presented at trial.” Id. at 324, 115 S.Ct. *780851. Since the court does not decide that question today (though it certainly strongly hints at an answer), I have no affirmative case to rebut. As I will show below, if that question were squarely decided today, one would have to truncate drastically the class of “reasonable jurors” to produce a negative answer.
Instead, the court gives the petitioner what he is seeking, a delay in execution, by a tactic that, if valid and sanctioned by the Supreme Court, would allow any court of appeals to duck its responsibility to decide cases, in favor of returning them to state courts through a process that is without precedent.
The court’s opinion attempts to find a precedent for such certification in cases such as Stewart v. Smith, 534 U.S. 157, 122 S.Ct. 1143, 151 L.Ed.2d 592 (2001) and Zant v. Stephens, 462 U.S. 862, 870, 103 S.Ct. 2733, 77 L.Ed.2d 235 (1983). These cases are not really on point; they involved situations in which the Supreme Court sought to understand more precisely the import of a challenged state law that had been interpreted by federal courts of appeals. In our case, the certified questions ask to have light shed on an issue already before us, whether “there [is] no state avenue open to process such a claim [of actual innocence].” See Schlup, 513 U.S. at 314 n. 28, 115 S.Ct. 851 (quoting Herrera v. Collins, 506 U.S. 390, 417, 113 S.Ct. 853, 122 L.Ed.2d 203 (1988) (O’Connor, J., concurring)). Ibid. Apparently a federal court would not be warranted in granting habeas relief, even making all the other necessary assumptions, if such an avenue were open. However, those cases in no way open the door for a court of appeals, having squarely before it a claim (ineffective assistance of counsel) that is fully exhausted in the state courts, to ask a state court for a rumination on this question.
If the Tennessee Supreme Court answered this question, what would the result be? If it firmly states that there is no state avenue open, then we simply return to our current position, many months or years down the road, with our court forced at that point to confront the question that is actually before it now. On the other hand, if it declares that there is an avenue open to process such a claim, then I would presume that habeas should be denied on that basis alone, again many months down the road.
This would presumably raise the odd (and unseemly) spectacle of this court denying a habeas petition and the prisoner proceeding toward execution, with only the possibility (which he would have always had anyway) of some last-minute intervention by the state courts. See Workman v. State, 41 S.W.3d 100 (Tenn.2001).
In reality, the court’s decision this day seems primarily to be a fishing expedition, undertaken in the hope that the Tennessee Supreme Court will take us off the hook by re-assuming control of the case, under the majority’s gentle guidance, or in the hope that we will, at a minimum, inject a significant additional amount of delay. Because I believe that we have squarely presented before us a question that we can answer, I respectfully dissent.
I
Judge Merritt has written an able argument on petitioner’s behalf. That argument might convince some, or even most, reasonable jurors that Paul House is actually innocent or should not be convicted. It would not, however, be so compelling that it is more likely than not that no reasonable juror would convict and thus, under established law, it should not avail Mr. House. The court focuses on three new items that would allegedly be so powerful *781that the Schlup standard would be met. I will deal with each briefly.
1. The semen on the victim’s clothing.
At trial, truthful evidence was presented showing that there were spots of semen on the victim’s nightgown and underclothing (though none on or in her body) that were consistent with the physical characteristics of defendant House, though no claim was made that this proved identity. Nor, contrary to the majority’s claim, did the State affirmatively contend that he had raped her. At most, as the quotation at page 771 of the majority opinion shows, the state argued that tricking a woman out of bed in the middle of the night and getting her down into a wooden ravine was consistent with “trying to get her to do something that she ... would [not] want to do with Mr. House.... ” Defense counsel noted at trial that the semen was equally consistent with the physical characteristics of the victim’s husband who was, even then, an alternate suspect under the defense’s theory.
2. Muncey “confessions.”
At the habeas hearing, two women testified that Hubert Muncey confessed to killing his wife. These are statements made many years after the offense, by two women who knew the petitioner, that tend to exculpate him. The women were found not to be credible by the federal district judge. They were available to the defense at the time of trial, but either were not investigated, or were not deemed of sufficient use to call to testify. Finally, contrary to the implication at page 5 of the court’s opinion, Artie Lawson did not testify at trial, but only at the habeas hearing. While these witnesses might convince some jurors (unlike the federal district judge), a juror need not be unreasonable to consider these statements to be last-minute fabrications in support of the defendant.
3.Blood evidence.
This is a convoluted scientific argument on behalf of an implication of a vast conspiracy on the part of Tennessee law enforcement officials. Simply put, the theory is that House never had any of the victim’s blood on his jeans while the jeans were in his possession. Instead, blood from the victim’s autopsy was smeared on the jeans in an elaborate charade designed to simulate the blood having been, present when the jeans were found. This theory is based on a highly contested interpretation of abbreviations in the original scientific report, and on disputed interpretations of the amount and location of blood from certain vials after they had been transported. Again, in my view, some reasonable jurors would be likely to believe the direct testimony and evidence in preference to a disputed scenario requiring the perjurious cooperation of numerous government officials.
In addition, some reasonable jurors could (and at least one would) be persuaded by the strong circumstantial evidence presented at trial. Ms. Turner’s testimony regarding House’s absence during the crucial time surrounding the murder and his return in a disheveled condition is consistent with his committing the murder, and inconsistent with his defense theory. Finally, some reasonable juror, in assessing the strength of the new evidence, would agree with the district judge’s finding that House’s direct testimony, presented for the first time at the habeas hearing, lacked credibility.
II
The Schlup Standard.
The appropriate standard for this court to employ in judging whether the evidence *782of House’s “actual innocence” is so strong that we should consider his claim of ineffective assistance of counsel, regardless of his procedural default, is set out in Schlup v. Delo.1 While various forms of words are used at different points in the opinion, the clearest statement appears at page 329, where Justice Stevens states that this extraordinary step is permitted only if we judge that “it is more likely than not that no reasonable juror would have convicted” the petitioner. Schlup, 513 U.S. at 329, 115 S.Ct. 851 (emphasis added) (internal quotation marks omitted). In making this judgment, Schlup “allows the reviewing tribunal also to consider the probative force of. relevant evidence that was either excluded or unavailable at trial.” See Schlup, 513 U.S. at 327-28, 115 S.Ct. 851, quoted in Lott v. Coyle, 261 F.3d 594, 621 (6th Cir.2001) (Cole and Merritt, JJ.). As Justice Stevens indicates, this is not the same as the standard in Jackson v., Virginia, 443 U.S. 307, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979). Schlup, 513 U.S. at 330, 115 S.Ct. 851. This judgment is not limited to the evidence at the original trial, and it focuses on what “would” happen, rather than what constitutionally “could” happen.
Even so, this standard is still one that requires us conscientiously to consider the entire range of “reasonable” jurors. In America, today, there are approximately 200 million persons eligible to serve as jurors. Even with the numbers of statutory and other disqualifications, one would trust that the large majority of these persons are in fact reasonable. For petitioner to fail to meet the Schlup standard, we need only assess that one of those potential jurors has a 50 + % likelihood of convicting, considering all the new evidence. As the review of the evidence above shows, that is not a difficult judgment to make. That judgment becomes even easier when the impact of the defendant’s own incredible testimony at the habeas hearing is taken into account, as it must be.
However, the difficulty for a defendant seeking to satisfy this standard is even greater, for the standard does not require that any individual reasonable juror actually have a 50 + % probability of voting to convict. The standard requires only that we believe it to be more likely than not that not a single one of the aggregation of reasonable jurors who might have some inclination to vote for conviction would in fact do so. I will explain that difference.
If each person in a group of 100 reasonable jurors had a 10% likelihood of voting to convict the defendant, the outcome of the Schlup assessment would depend almost entirely on our belief about the independence of the jurors’ judgments. By that, I mean that if each juror’s opinion is independent of that of every other juror, it is almost certain that one out of the 100 jurors would convict.2 If we believed that every juror thought exactly like every other one — that is, that their opinions had no independence at all, then there would be only a 10% chance of conviction by any juror in the whole group, and the Schlup standard would be met.
Thus, under these circumstances, the outcome of the Schlup assessment could turn on whether we believe there is any independence, or difference, in the thought *783processes of reasonable jurors. I believe that the experience of virtually every lawyer who has participated in a jury trial is that the thought processes of jurors have a considerable degree of independence. The reason that one juror may vote to convict (or to acquit) may differ significantly from that of other jurors. When that is the case, we have the following result. If any significant number of reasonable jurors have some likelihood of conviction, there is a very high probability that at least one such juror will in fact convict, given almost any reasonable degree of independence.3 Thus, an honest application of the Schlup standard means that a prisoner can meet it only if a judge can conscientiously assert that every reasonable juror is almost certain to vote to acquit. This case does not meet that standard.
In summary, in order to justify overturning a state court judgment, the Schlup standard “does not merely require a showing that a reasonable doubt exists in light of the new evidence, but rather that no reasonable juror would have found the defendant guilty.” Schlup, 513 U.S. at 329, 115 S.Ct. 851. Footnote 4 of the court’s opinion expresses the court’s belief that House’s sentence must be overturned if “even 50 of 100 ... jurors would now seriously doubt the persuasiveness of the state’s case.” This statement is refreshingly candid, because it expresses what a careful reader might perceive to be the standard of many jurists in death penalty cases. However, it is as clear as the English language can make it that this is not the standard stated by Justice Stevens in the quotation above. The footnote clearly supports overturning a sentence in a situation (quite likely the situation in our case), where half of reasonable jurors would not, as the footnote says, “seriously doubt the persuasiveness of the state’s case.” In such a situation, I find' it impossible to believe that a decision-maker, conscientiously reading the words of the Schlup decision, could determine that “no reasonable juror would have found the defendant guilty.” '
As a federal appellate court reviewing a federal habeas court’s review of a state court’s, determination, and attempting to consider (as required by Schlup) what “any reasonable juror” (and thus all reasonable jurors) would do, we are inevitably making hypothetical judgments. That statement is as true of the majority and special dissenting opinions as it is of mine. But we must do that in order to be faithful to Schlup. Otherwise we are simply doing what some observers may think is done regularly in such cases — substituting our own judgment of what the evidence shows for the opinion of the jury, and of all jurors.
My mathematical analysis at pages 782-83 is not the main line, of my analysis. As clearly stated at page 782, a claim such as House’s fails whenever the reviewing judge can state a belief that some single reasonable juror would in fact still convict, taking *784into account all the new evidence (see page 782).
The mathematical analysis that so excites the court is simply the portion that goes beyond the statement above to indicate the reality of the meaning of the words that Justice Stevens penned. The standard is indeed a very difficult one to meet; it is not designed simply to allow judges to substitute their own judgment for that of some, or even most, jurors. As Schlup states: “It is not the district court’s independent judgment as to whether reasonable doubt exists that the standard addresses.” 513 U.S. at 329, 115 S.Ct. 851.
That is all that my analysis attempts to explicate.
The fundamental error of footnote 4 is boldly stated when it complains that the dissent believes “that House should be executed” [more accurately, that his petition should be rejected on a proper application of the law] because “at least 1 hypothetical hold out juror would disagree” with the majority’s implied opinion in this case. Any complaint with this statement simply betrays an unwillingness to face up to the exact language in Schlup. That opinion specifically says that “the standard requires the ... court to make a probabilistic determination about what reasonable, properly instructed jurors would do.” Schlup, 513 U.S. at 319, 115 S.Ct. 851 (emphasis added). Thus, the Supreme Court specifically requires us to make the type of determination that appears to upset the court. Further, the complaint in footnote 4 simply refuses to confront the fact that if one juror, acting reasonably, would have voted to “find him guilty beyond a reasonable doubt” then it is defini-tionally impossible to make the necessary determination, in the words of Justice Stevens, that “no juror, acting reasonably, would have voted to find him guilty beyond a reasonable doubt.” Ibid, (emphasis added). The fact that the court may consider such an argument to be “hypothetical,” “medieval scholastic” or “remarkable” does not alter the words and holding in the Supreme Court’s opinion.
Ill
A careful reader will discern that a number of the significant criticisms of the dissent that are leveled in section III of the court’s opinion do not correspond with what is actually written in the dissent. At page 778, the court’s opinion states that I am applying “the old standard of the sufficiency of the evidence at trial.” In fact, at page 781, I consider the question in light of all the evidence, old and new, as I clearly state at page 782. Further a careful reader will find that I make no reference to the “any reasonable juror could convict” standard of Jackson v. Virginia, but rather rely solely on the holdings in Schlup.
The charges made at page 778 that “the test used by our dissenting colleagues [is] the sufficiency of the evidence in the trial record” and “whether there is evidence that could have supported a jury’s decision to convict, regardless of the new evidence” are nowhere held, supported, nor implied in the dissent.
A careful reading of Schlup also belies the statement at page 778 that “the question is ... ‘if they [the new statements] ... are true’ whether a juror .... would vote to convict.” ’ (quoting Schlup, 513 U.S. at 331, 115 S.Ct. 851). That language is excerpted from the portion of Schlup that specifically makes that statement in the context of facts that “the Court of Appeals assumed [were true] for the purpose of applying its understanding of the Sawyer standard.” Ibid. In other words, Justice Stevens was saying that if the reviewing *785court believed the new evidence to be true (while the Justice recognized that “those new statements may, of course, be unreliable”) then the Schlup standard would be met. He certainly did not say that a reviewing court was required to find rehable any and all new evidence proferred.
Finally, it should be noted that this dissent does not state, contrary to page 779, that “all the questions before us” are fully exhausted. As is clearly stated at page 780 “ineffective assistance of counsel,” that is fully exhausted.
IV
The Aggravating Factors.
The court’s opinion also tries to use the allegedly new evidence to undermine the three aggravating factors found by the jury-
With regard to the first aggravating factor, our court’s statement (page 773) that the primary evidence for the existence of “torture or serious physical abuse” was the evidence of rape is belied both by the state’s failure specifically to argue that point, and by the extensive evidence of the physical beating and strangulation of the victim.
The second aggravating factor is that House committed the crime in the course of an attempted rape or kidnaping. Kid-naping includes “unlawfully ... inveigling [or] enticing away another with the intent of causing him to be secretly confined or imprisoned against his will.” Surely some reasonable juror would find that, if House committed the offense as described in the testimony, he enticed the victim away in order to confine her against her will. Not every reasonable juror would be likely to believe that he tricked her out of the house in order to discuss epistemology, and a sudden affray over Foucault caused him to kill her in a blind rage.
The third aggravating factor is the simplest: that House had previously been convicted of a felony “involving the use or threat of violence to a person.” This item was amply proved by Utah state records showing House’s conviction for aggravated sexual assault. The only error was that evidence was also introduced (truthfully) that he had been sentenced to five years to life for that offense, and was on parole. All that the Tennessee Supreme Court found was that the latter factual information was irrelevant. It certainly did not find that there was any prejudice by the introduction of such evidence. Since House was at liberty when the crime was committed, he must have either served out his sentence, escaped, or been on parole. Of the three, being on parole is the most benign interpretation of the evidence. Nor was the length of the sentence for this crime particularly prejudicial. Thus, the third aggravating factor has not been undermined to any significant degree.
V
I turn now to the specific questions that the court’s opinion certifies to the Tennessee Supreme Court. It is frequently said, in response to a rhetorical question, that there are “smuggled assumptions” within a question — the unstated premises upon which the question rests. In this case, however, the assumptions have not been smuggled — they have been cleared through customs with all duties paid.
The first question specifically states that “a defendant whose sole remaining aggravating factor was introduced in an erroneous manner” is involved. (Emphasis added.) Of course, the destruction of the other factors is the very element that should be decided, not assumed. Indeed, as discussed above, the remaining aggravating factors are at most weakened, but *786not removed, even if the defendant’s evidence is believed. Of course, if his evidence is believed, he didn’t commit the crime at all, so the aggravating factors should not even be considered. If, on the other hand, he did commit the crime, then the strength of the evidence that purports to undermine the aggravating factors is itself demolished or greatly weakened.
The second question appears to be a demand that the Tennessee Supreme Court re-entertain a petition for post-conviction relief. It asks “in light of the new DNA evidence” (which has never been presented to the Tennessee courts), and in light of “the error in the presentation of the third aggravating factor” (a matter fully considered by the Tennessee Supreme Court), what is the state of Tennessee law? This is essentially asking the Tennessee court to render an advisory opinion based on a record that its processes and fact finders have not considered.
The third question continues boldly to assert that the “evidence disproves the state’s theory that the defendant murdered the victim in the commission of a sexual assault.” It purports to ask the Tennessee Supreme Court to consider “malice aforethought,” but to do so “together with other newly discovered evidence of actual innocence.” Again, the theories are in contradiction. If House did not commit any crime, then malice aforethought is irrelevant. If he did commit the crime, albeit without depositing semen on the victim’s clothing, then it is extremely difficult to say that luring a person out of her bed and home in the middle of the night and murdering her in a way that results in her body being found “partially concealed in a brush pile at the bottom of a wooded embankment” 100 yards from her home (page 769, lines 3-5 of section I) does not permit some reasonable juror to find that there was malice aforethought. In particular, the statement from the court’s opinion quoted above presents the facts much more fairly than the court’s bland statement at page 11 that “[ajlthough House purportedly lured the victim out of her house, her body was found within yards of her doorstep.... ”
The majority’s basic problem is that if the evidence introduced at the habeas hearing does not compel a belief in House’s innocence of the crime of killing Ms. Muncey, then it does almost nothing to undermine the death sentence. Thus, the questions purporting to deal with the sentence independent of the crime seem to have no basis.
Finally, the first and third questions as stated at page 777 do not even purport to address the supposedly crucial question of whether a state avenue is open to process a claim. Instead, they ask, in terms that assume that all of the relevant material is procedurally properly before some Tennessee court, does “the law” (1) “require a new sentencing hearing” and (3) “require a new trial?”
Question 2 does at least appear to address some possibly relevant question by asking whether the Tennessee Supreme Court’s “review process now permit[s] it to remedy any error in the weighing process by the jury in light of newly discovered evidence?” However, beyond our invitation to the Supreme Court to “remedy any error ... by the jury,” the most the Tennessee Supreme Court could do would be to say that there are avenues for last-minute post-conviction petitions. This possibility has already been amply demonstrated in Workman v. State, 41 S.W.3d 100 (Tenn.2001). There, a “petition for a writ of error coram nobis” was filed in state court two days before a scheduled execution, based on alleged new evidence purportedly showing actual innocence. Id. at 104. The Tennessee Supreme Court *787swept aside all procedural obstacles and directed consideration on the merits. Id. at 103. Thus, there is clearly no basis for a remand to see if the Tennessee Supreme Court can do what it has already done in an analogous case.
VI
In short, the court’s opinion today is a procedural muddle, leading to far greater future muddles, to little practical effect other than to add an additional layer of delay.
Contrary to the statements in the court’s opinion, the question raised by this dissent is not whether this decision “will delay House’s execution.” (page 778) The question is whether it will unjustifiably delay reaching finality. I have no particular brief for this defendant’s execution, or his exoneration, other than as the evidence in this case leads me to make a decision based on existing law. I do have a brief for the value of reaching finality of decision, a brief that has been eloquently explained by the Supreme Court in McCleskey v. Zant, 499 U.S. 467, 491-92, 111 S.Ct. 1454, 113 L.Ed.2d 517 (1991). If the court is fully convinced that no avenues are available (contrary to Workman v. State), and if it is convinced, contrary to the above demonstration, that the Schlup standard can be met on this evidence, then it should simply say so. The mere invocation of the statement “better to be safe than sorry” (page 779) simply means that no decision should ever be made.
It is obvious that the court’s opinion is undergirded by an evaluation of the evidence that would argue that very few, if any, reasonable jurors would find House guilty beyond a reasonable doubt. If the court would clearly hold that now, at least we would have a straightforward legal decision, albeit one subject to challenge on further review. But in this case, we have undertaken a novel diversionary process, and added a new and unjustified procedure for avoiding finality. This decision could be invoked by any court whenever a claim of - actual innocence, however tenuous, is raised for the first time in federal court. I therefore respectfully dissent.
. This assumes, of course, that we have already determined that there is no state avenue open to consider that claim further. This is, of course, contrary to fact as shown by Workman v. State, 41 S.W.3d 100 (Tenn.2001). See infra at 785.
. The mathematics is such that the chance that no one of 100 independent jurors, each with a 10% probability of voting to convict, would convict, is .9 to the 100th power, which is less than 1 chance in 25,000.
. The mathematics is a bit complicated, but here is an example. If a group of 100 reasonable jurors each has a 10% likelihood of conviction, we can assess the degree of independence of each pair of jurors as follows:
1) If their opinions are wholly dependent, they will, of course, agree 100% of the time;
2) If their opinions are wholly independent, they would still agree 82% of the time. In our example, 81% of the time they would both acquit, 1% of the time they would both convict, and 18% of the time one would convict and one would acquit.
If their views are even a little bit independent, for example, if they now agree only 98% rather than 100% of the time, mathematics indicates that there is a very high likelihood, well beyond 50%, that at least one of the reasonable jurors would in fact vote to convict.