concurring.
It is hard for a court to invalidate a provision that does not even apply to the case at hand. It is harder still to invalidate a whole gaggle of provisions that actually support the result reached in the case. Yet that, according to the rhetoric in support of the State’s petition for rehearing, is the feat we have managed here. I think the rhetoric is seriously overblown.
We are told, in support of rehearing, that our decision effectively invalidates every rape-shield law in this circuit. That argument cannot be squared with the text of those laws. The core of any rape-shield law is its proscription against evidence of past sexual activity by the victim. But every one of those laws — or at least every one we are said to have vaporized here— contains an exception for evidence of the victim’s prior sexual activity with the defendant. And that is exactly the kind of evidence that we hold should have been admitted in this case. Michigan’s statute excepts from its proscription “[ejvidence of the victim’s past sexual conduct with the actor.” Mich. Comp. Laws § 750.520j(l)(a). Ohio’s statute does the same. See Ohio Rev.Code § 2907.02(D) (excepting evidence of “the victim’s past sexual activity with the offender”). So does the Tennessee rule. See Tenn. R. Evid. 412(c)(3) (allowing admission of evidence of “sexual behavior ... with the accused, on the issue of consent”). The federal rule affirmatively provides that “evidence of specific instances of sexual behavior” between the alleged victim and defendant “is admissible,” if offered to prove consent and otherwise admissible under the rules. Fed.R.Evid. 412(b)(1)(B) *290(emphasis added). Kentucky’s rule is identical to the federal one. See Ky. R. Evid. 412(b)(1)(B). I am of course mindful that we apply only constitutional rules, not statutory ones, on habeas review. But rhetorically speaking, our decision applies these laws, rather than abrogates them.
Equally misdirected is the claim that we have trampled upon the policies that animate these laws. As Judge Norris’s opinion carefully explains, we hold that, under the rather extraordinary facts presented here, certain evidence of the complainant’s prior consensual sex with the defendant was indispensable to his ability to present a complete defense at trial. The argument now is that, in deeming that evidence indispensable, we have indulged in outdated inferences whose eradication was a principal aim of these laws. But the argument again is let down by the laws. Every one of these laws is supported by significant state interests. (The extent to which those interests are implicated in a particular case, as discussed below, is another matter.) And yet, notwithstanding those important interests, every one of these laws contains an exception for evidence of consensual sex with the defendant. These laws must infer something very important about such evidence; and they do so especially in cases — like this one — where consent itself is the issue. The inference is that, in some (and perhaps most) cases, evidence of past consensual sex with the defendant is highly relevant to the issue of consent in the incident giving rise to the charge. We merely conclude that this is such a case.
Notable as well is that the caselaw in Michigan, and the terms of the federal and Tennessee rules, expressly wave off the rape-shield proscription as to “evidence the exclusion of which would violate the constitutional rights of the defendant.” Fed.R.Evid. 412(b)(1)(C); see also People v. Hacked, 421 Mich. 338, 365 N.W.2d 120, 124 (1984) (recognizing that, in some situations, admission of evidence relating to the victim’s sexual conduct “may be required to preserve a defendant’s constitutional right to confrontation”); People v. Adair, 452 Mich. 473, 550 N.W.2d 505, 511 (1996) (rejecting a construction of Michigan’s rape-shield statute that would have “run[ ] the risk of violating a defendant’s Sixth Amendment constitutional right to confrontation”); Tenn. R. Evid. 412(c)(1). We do not, of course, look to these rules in determining whether there was a violation of clearly established federal constitutional law. But these provisions do show that the rules’ drafters knew very well that they were operating close to the constitutional line. That the rules’ application might cross it, on occasion, should surprise no one.
I also think the rehearing petition overstates the breadth of our holding in this case. We do not hold that the Supreme Court’s decision in Crane v. Kentucky, 476 U.S. 683, 106 S.Ct. 2142, 90 L.Ed.2d 636 (1986), clearly establishes a rape defendant’s constitutional right to admit evidence of the victim’s prior consensual activity with the defendant (much less with persons other than the defendant). What we do say, as a starting point for the analysis, is that Crane clearly establishes that “the Constitution guarantees criminal defendants ‘a meaningful opportunity to present a complete defense.’ ” Id. at 690, 106 S.Ct. 2142 (quoting California v. Trombetta, 467 U.S. 479, 485, 104 S.Ct. 2528, 81 L.Ed.2d 413 (1984)). The State of Ohio, as amicus, says this rule is stated at too high a level of generality to support habeas relief here. That point is a serious one; the more general a rule, the more difficult it is to say that the state court unreasonably applied the rule to a particular set of facts. With generality often comes discretion. But I think the facts of *291Crane and those of this case converge at a level closer to the ground than the State realizes.
Crane requires consideration of two factors in determining whether the exclusion of evidence denies the defendant a meaningful right to present a complete defense. The first is the extent to which the evidence was “central to the defendant’s claim of innocence.” Id. The second is the extent to which its exclusion was supported by a “valid state justification[.]” Id. In Crane, the defendant was convicted of murder. He was 16 years old at the time of the crime. There was “no physical evidence to link him” to the murder. Id. at 691, 106 S.Ct. 2142. The State’s evidence of guilt was primarily Crane’s own confession. Crane sought to discredit the confession with testimony that “he had been detained in a windowless room for a protracted period of time, that he had been surrounded by as many as six police officers during the interrogation, that he had repeatedly requested and been denied permission to telephone his mother, and that he had been badgered into making a false confession.” Id. at 685, 106 S.Ct. 2142. That evidence, the Supreme Court said, was “highly relevant” to the reliability and credibility of the confession, which again was the State’s primary evidence of guilt. Id. at 691, 106 S.Ct. 2142. And the Court saw no justification for excluding the evidence under the circumstances presented there. The Court did not “pass on the strength or merits” of Crane’s defense. Id. But it held, unanimously, that the exclusion of Crane’s testimony violated his right to present a complete defense.
The analysis flows in the same channels here. First, the excluded evidence was “central to the defendant’s claim of innocence.” Id. at 690, 106 S.Ct. 2142. Gagne was convicted of forcing Clark to engage in an outlandish sexual encounter with himself and another man. The events making up that encounter were largely undisputed; what was disputed was whether Clark consented. There was little if any physical evidence of non-consent. The State’s primary evidence of guilt, rather, was Clark’s own testimony that she did not consent. Gagne sought to discredit that testimony with evidence that — only one month before the subject incident — Clark had voluntarily engaged in three-way sex with Gagne and another man, and that she had suggested the same thing on another occasion. That evidence would have included testimony not only from Gagne, but also from Ruben Bermudez, the other participant in the three-way the month before, who according to Gagne’s lawyer was prepared to testify at trial. See 1/2/01 Hearing Tr. 18-19. But the trial court excluded all of that testimony. That exclusion was affirmed by the state court of appeals, which held-remarkably-that the excluded testimony was not even relevant to the case.
That the excluded evidence was highly relevant can be inferred, as discussed above, from the rape-shield statutes themselves. Moreover, as Judge Norris explains, in both this case and Crane the excluded evidence, if credited, answers a question that, if left unanswered, leads directly to conviction. See Maj. Op. at 288-89.
The State says the trial court’s admission of evidence of another sexual encounter — the so-called Tony’s Lounge incident, which involved five people — -made the three-way evidence only marginal, rather than central, to Gagne’s defense. But the prosecution itself did a credible job of explaining, in closing argument, why the Tony’s Lounge evidence did very little to answer the question described above. I think that Judge Norris’s opinion convincingly finishes the job.
*292What is significant is the effect of the Tony’s Lounge evidence upon the State’s interest in excluding the Gagne and Bermudez three-way testimony. This, in my view, is where the bottom falls out of the State’s argument on appeal. As an initial matter, this case only weakly implicates the interests protected by Michigan’s rape-shield statute, since the statute’s terms did not even bar the excluded testimony, but instead left its admission to the discretion of the Ingham County Circuit Judge. See Mich. Comp. Laws § 750.520j (l)(a). And it is hard to see what was left of those interests, such as they were in this case, given the evidence that was admitted at trial. That evidence included that, during the Tony’s Lounge incident, Clark had engaged in oral sex with Swathwood shortly after intercourse with Gagne, and that she had engaged in consensual oral sex with Gagne minutes before the very incident for which he was convicted (and moreover that she had drunk a pint of vodka and nine or so beers and smoked crack in the hours before the incident). I entirely agree that Michigan’s rape-shield law protects important state interests in the vast majority of cases in which it is implicated. But I submit that, under the circumstances of this trial, there was virtually nothing left of those interests to protect.
So here, as in Crane, the prosecution’s case hinged on a single account of what happened during the underlying incident. In each case, the nature of that account — a confession there, and the outlandish conduct alleged to have been coerced here— tended to make the account especially persuasive to a jury. In each case, the trial court excluded highly relevant evidence that, if credited, could have undermined the account upon which the State relied. And in each case the State’s interest in excluding that evidence was negligible.
These parallels emerge, admittedly, only after close analysis. But I think they are clear enough to render the state court’s decision unreasonable within the meaning of the habeas statute. (It bears mention that two Justices of the Michigan Supreme Court voted to reverse the decision summarily, and that Justice Markman voted to hear the case. See 469 Mich. 982, 673 N.W.2d 755 (Mich.2003) (unpublished table disposition).) The exclusion of the three-way evidence clearly violated Gagne’s Sixth Amendment and procedural due process rights. He should have been able to present that evidence to a jury before ultimately being sentenced to a prison term of up to 45 years.
My purpose in reciting this evidence, however, is not to explain what Judge Norris has already explained in his nuanced opinion. And I readily admit that reasonable people, including the respected Chief Judge of our court, can disagree with my conclusion in this case. Nor do I mean to “pass on the strength or merits” of Gagne’s defense. Crane, 476 U.S. at 691, 106 S.Ct. 2142. I think those merits should have been left to the jury; and otherwise I do not hold a torch for anyone involved in the underlying events of this case.
My purpose, instead, is to emphasize the limited nature of our holding in this case. See also Maj. Op. at 288-89. With respect, I do not think that holding has nearly the jurisprudential consequences that Michigan and the amici States seem to think it has.
For these reasons, I continue to concur fully in Judge Norris’s opinion.