concurring.
I agree with the majority's view that Marion County’s now-extinct system for assigning trial judges did not violate Tyson’s right to a fair trial and that the trial court did not err in refusing Tyson’s proffered jury instructions. I write separately because I believe the trial court’s exclusion of witnesses as a sanction for Tyson’s breach of the court’s discovery order violated his Sixth Amendment rights; an error, however, that in the habeas corpus context of this case must be deemed harmless.
The Sixth Amendment provides that “[i]n all criminal prosecutions, the accused shall enjoy the right ... to have compulsory process for obtaining witnesses in his favor.” U.S. Const. amend. VI. “Few rights are more fundamental than that of an accused to present witnesses in his own defense.” Chambers v. Mississippi, 410 U.S. 284, 302, 93 S.Ct. 1038, 1049, 35 L.Ed.2d 297 (1973). This right is not absolute, however, and must yield to the “enforcement of firm, though not always inflexible rules relating to the identification and presentation of evidence.” Taylor v. Illinois, 484 U.S. 400, 411, 108 S.Ct. 646, 654, 98 L.Ed.2d 798 (1988). In Taylor, the Supreme Court held that a trial court can in certain circumstances exclude witnesses as a sanction for a discovery breach without violating the constitution. Taylor mandates that in determining whether to decree such a sanction, a court must exercise its discretion by balancing a defendant’s “fundamental right” against “countervailing public interests,” including:
the integrity of the adversary process, which depends both on the presentation of reliable evidence and the rejection of unreliable evidence, the interest in the fair and efficient administration of justice, and the potential prejudice to the truth-determining function of the trial process....
Id. at 414-15, 108 S.Ct. at 656. The Court stated that it was neither “necessary nor appropriate to draft a comprehensive set of standards to guide the exercise of discretion in every possible case.” Id. at 414, 108 S.Ct. at 656.
The Taylor Court upheld the exclusion of an alibi witness of whom the defense had known but had failed to disclose in response to the prosecution’s pre-trial discovery request. The Court noted that the defense had acted willfully and in bad faith in not disclosing the witness until after the prosecution’s two primary witnesses had testified. In addition, the Court identified a “sufficiently strong inference that “witnesses are being found that weren’t really there,’ to justify the sanction of preclusion.” Id. at 417, 108 S.Ct. at 657. “Regardless of whether prejudice to the prosecution could have been avoided [by a less severe sanction], it [was] plain that the case fit[ ] into the category of willful misconduct in which the severest sanction [was] appropriate.” Id; see also Eckert v. Tansy, 936 F.2d 444 (9th Cir.1991); Chappee v. Vose, 843 F.2d 25 (1st Cir.1988).
The First Circuit has subsequently found constitutional error in a trial court’s exclusion of an alibi witness as a sanction for a discovery violation. Bowling v. Vose, 3 F.3d 559 (1st Cir.1993), cert. denied, — U.S. —, 114 S.Ct. 1236, 127 L.Ed.2d 580 (1994). In Bowling, the defense divulged the name of an alibi witness after the close of the prosecution’s case and five days after a prosecution witness testified on cross-examination to an earlier starting time than the prosecution had originally estimated for the arson fire. In applying Taylor, the court first noted that the defense had not acted in bad faith because even though it had previously known about the witness, it could not have known of her importance until after it had cross-examined the prosecution’s witness. The court thus viewed defense counsel’s conduct as “negligence at worst.” The court concluded that the exclusion of exculpatory evidence *450would distort, rather than further, the truth-seeking function of the trial, and held that the non-willful nature of the violation allayed fears that the testimony had been fabricated. Finally, the court stated that because the evidence consisted of only one witness, whose testimony was not of a technical nature, a continuance could have sufficed, even though the prosecution would have been prejudiced by the introduction of a new defense theory. Id. at 562.
Taylor indicates that willful misconduct and clearly unreliable testimony constitutionally support the dramatic sanction of witness exclusion. Bowling, on the other hand, teaches that a trial court cannot exclude reliable testimony as a sanction for negligent conduct that would pose minimal prejudice to the prosecution. Turning to the instant ease, it is necessary to determine whether the trial court abused its discretion by ordering witness exclusion as a sanction for defense counsel’s discovery violation.
As an initial matter, I agree with the majority that the trial court did not abuse its discretion in holding that its discovery order required disclosure as soon as possible. The Indiana Court of Appeals interpreted this to mean Friday evening after an attorney from defense counsel’s office had interviewed the women and discussed their information with the trial team. Tyson v. Indiana, 619 N.E.2d 276, 283 (Ind.App.1993) (“The trial court reasonably concluded that, at that time, the State should have been notified of the existence of the women and their potential testimony.”), cert. denied, — U.S. —, 114 S.Ct. 1216, 127 L.Ed.2d 562 (1994). Therefore, the defense violated the discovery order by fading to apprise the prosecution and court of the new witnesses until Sunday and Monday respectively. The trial court did not find either that the defense had acted willfully or in bad faith or that it should have discovered these witnesses at an earlier date.
A primary ground given by the trial court for excluding the witnesses was their potentially prejudicial effect on the government’s case. It has never been held that the defense violated the order for the purpose of achieving a tactical advantage. See Taylor, 484 U.S. at 417, 108 S.Ct. at 657. The defense’s violation may still have given Tyson such an advantage because the defense alone knew of the potential testimony of these witnesses and perhaps used that information in cross-examining several of the prosecution’s witnesses. Yet my review of the trial record indicates that the state courts may have overestimated the advantage gained by the defense. The state courts appeared to accept that the defense knew about the testimony when it cross-examined the victim, the hotel personnel, and the limousine driver. The defense actually cross-examined most of these witnesses, including the victim and hotel personnel, before the investigating attorney had interviewed the newly discovered witnesses. Whether the investigator could or should have interviewed these witnesses earlier is not at issue here. Therefore, the defense could only have cross-examined witnesses on Saturday, including the limousine driver, with any unfair knowledge. Moreover, even if the discovery of the witnesses was a surprise, the basis of their testimony was not. Tyson had already testified before the grand jury that he and the victim had kissed in the limousine. Cf. Chappee, 843 F.2d at 31 (“Once the Commonwealth had been cradle-slung into a false sense of security, and [their witness] subjected to cross without fair warning of the nature of the defense, [the defendant’s] chances of success ... were (unfairly) maximized.”) (emphasis added). All of this appears to suggest that the prejudice to be suffered by the prosecution and the advantage to be gained by the defense had the witnesses testified may not have been as extensive as the state courts believed.
The prosecution also argued that the defense’s discovery breach had prejudiced it in three other ways. First, the State asserted that the introduction of these witnesses would disturb the rhythm of its case by requiring a continuance of several days, if not a week, while it prepared to examine them and reevaluated its case-in-chief. Second, it contended that this delay would have adversely affected the sequestered jury. Third, the prosecution maintained that the use of these witnesses would have required it to recall certain witnesses on rebuttal, a po*451tentially less effective method of examination. The trial court seemed to accept some or all of these arguments, stating that “the State has, in fact, been prejudiced in this matter ... [and the prejudice] cannot be cured by a recess or a continuance because of the fact that we are in the middle and almost completed with the State’s presentation of the evidence based on what they thought the evidence was to show.” Cf. United States v. Johnson, 970 F.2d 907, 912 (D.C.Cir.1992) (noting that the alternative sanction of a continuance would clearly have “disrupted the trial schedule and harmed the government, which had brought in witnesses from all over the country,” and a second alternative of a delay after the prosecution had rested would also have been “burdensome as it would have required holding a jury over during the delay”).
This conclusion of prejudice overlooks that the prosecution would have faced most of the delay and disruption even without the' defense’s discovery violation because it, like the defense, would still have been caught off guard by the arrival of these witnesses. Even if the defense had apprised the prosecution and the court of the existence of these witnesses on Friday evening, the prosecution would have had to attempt to find witnesses to impeach them and would have had to prepare its case accordingly. In determining whether the trial court abused its discretion in holding that the defense’s conduct, and the effects of its actions warranted the sanction of exclusion, the defense should not be held totally accountable for such unavoidable consequences.
The late arrival of these witnesses in this well-publicized ease could also lead to doubt about the reliability of their testimony. The Supreme Court has said that it is reasonable “to presume that there is something suspect about a defense witness who is not identified until after the 11th hour has passed.” Taylor, 484 U.S. at 414, 108 S.Ct. at 655. Such a presumption is not appropriate here. These witnesses surprised the defense with their initial contact. Indeed, an attorney not associated with either party in the case but a relative of one of the proposed witnesses twice advised these women not to come forward with their information and to avoid becoming involved in the case. In these circumstances, one cannot easily presume that the testimony was fabricated. The prosecution’s contentions that the women may have seen Tyson with his girlfriend on a different night, or that they could not really have seen what they said they did, were more properly matters for the State’s cross-examination and a jury determination. Because these witnesses were apparently relevant and not clearly unreliable, their exclusion potentially hindered, rather than furthered, the truth-seeking purpose of the trial.
In light of the discussion above, did the facts of this case warrant the sanction of witness exclusion? The Supreme Court did not hold in Taylor that a court could order witness preclusion as a sanction for every discovery violation. Michigan v. Lucas, 500 U.S. 145, 152, 111 S.Ct. 1743, 1748, 114 L.Ed.2d 205 (1991). Instead, the exclusion sanction should be limited to the “most extreme situations,” Taylor, 484 U.S. at 417 n. 23, 108 S.Ct. at 657 n. 23 where the defense’s conduct and the detrimental effects of that conduct outweigh the defendant’s constitutional rights. In my judgment, this case did not present such circumstances. The defense, in violating the discovery order, acted neither willfully nor in bad faith and cannot be said to have seriously endangered the integrity of the judicial system. The testimony was not clearly unreliable and did not present a new defense theory. It is true that the use of these witnesses might have required a continuance in a trial with a sequestered jury and might have required the prosecution to bring back certain of its witnesses on rebuttal. However, the prejudice and delay the State would have suffered in these circumstances cannot be completely attributed to the defense. Balancing these factors, I would hold that Tyson’s constitutional rights outweighed the seriousness and effects the defense’s discovery violation. Where, as here, there is no willful misconduct by the defense, limited prejudice to the State, and potentially material testimony, witnesses should not be excluded as a sanction for a *452discovery violation consistent with the Sixth Amendment.1
A finding of constitutional error does not end the inquiry, however, for habeas relief is not granted unless the error had a ‘“substantial and injurious effect or influence in determining the jury’s verdict.’ ” Brecht v. Abrahamson, — U.S. —, —, 113 S.Ct. 1710, 1721, 123 L.Ed.2d 353 (1993) (quoting Kotteakos v. United States, 328 U.S. 750, 776, 66 S.Ct. 1239, 1253, 90 L.Ed. 1557 (1946)); see also O’Neal v. McAninch, — U.S. —, 115 S.Ct. 992, 130 L.Ed.2d 947 (1995) (“When a federal judge in a habeas proceeding is in grave doubt about whether a trial error of federal law had a substantial and injurious effect or influence in determining the jury’s verdict, that error is not harmless.”) (internal quotations omitted). As the majority indicates, Brecht is applied even if the state courts failed to employ the more stringent “harmless beyond a reasonable doubt” standard of Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 828, 17 L.Ed.2d 705 (1967).
The erroneous exclusion of cumulative or impeaching evidence usually constitutes harmless error. See, e.g., United States v. Martinez, 988 F.2d 685, 701-02 (7th Cir.) (exclusion of victim’s racist and violent nature, offered as support for defendant’s defense of provocation, if error, was harmless where other evidence of victim’s violence was already before the jury), cert. denied, — U.S. —, 114 S.Ct. 125, 126 L.Ed.2d 89 (1993); United States v. Petitjean, 883 F.2d 1341, 1348 (7th Cir.1989) (in extortion trial, any error in limiting cross-examination of victim as to amounts he owed defendant to show victim’s lack of credibility was harmless where five other witnesses testified to the victim’s poor reputation for truthfulness); Lange v. Young, 869 F.2d 1008, 1012 (7th Cir.) (any error in excluding physician’s testimony that defendant had previously been diagnosed as paranoid-schizophrenic was harmless where defendant’s medical records were admitted and three other doctors testified as to his present schizo-type personality disorder), cert. denied, 490 U.S. 1094, 109 S.Ct. 2440, 104 L.Ed.2d 996 (1989). But see Sandoval v. Acevedo, 996 F.2d 145, 149 (7th Cir.1993) (“[I]mpeachment evidence can be vital”). In the present case, all of the prior courts, as well as the majority here, have concluded that these witnesses would have presented cumulative impeachment evidence, and I concur. Clearly, this victim was otherwise impeached during the trial. Some witnesses testified that the victim made certain comments about the defendant during the day before the rape, which statements she denied making. Other witnesses testified that the victim told them certain versions of rape, most of them dissimilar from each other and from the account she presented. Even with all of this impeachment of the victim the jury chose to believe her version of the critical events in the hotel room. It is true that the excluded testimony would not have been cumulative in the sense of repeating what other witnesses had previously said or would say. However, these witnesses would have only given additional evidence contradicting the victim, in this case about a matter ultimately distinct from the issue of consent.
Moreover, the jury’s verdict was supported by other evidence. In addition to the jury’s assessment of Tyson’s testimony, the prosecution presented physical evidence of the victim’s vaginal abrasions consistent with sexual assault. See ante at 443. The State introduced evidence that the victim’s clothing was torn the critical evening and that the police found a sequin from that outfit in Tyson’s hotel room. Furthermore, the limousine driver testified to the victim’s agitated state after leaving Tyson’s hotel and the prosecution tendered evidence of Tyson’s hurried *453departure from Indianapolis shortly after the critical events. All of this serves to buttress the argument that the error in this case was harmless.
In O’Neal v. McAninch, the Supreme Court’s most recent pronouncement on habe-as harmless error review, the Court stated that when a judge has grave doubts as to whether an error substantially affected the jury’s verdict, he must find that it was not harmless. — U.S. —, 115 S.Ct. 992. In other words, when “the matter is so evenly balanced that he feels himself in virtual equipoise,” id., the judge must find in the defendant’s favor. Against the backdrop of this case, and recognizing the constraints imposed by habeas review, I do not possess “grave doubt” about the effect of this error. Although the excluded impeachment evidence was apparently relevant and arguably would have further addressed the victim’s credibility, strong impeachment along with significant other evidence corroborating the victim’s allegations had already been admitted. In light of this record, I cannot conclude that the excluded testimony would have had a substantial effect on the verdict.2
For the foregoing reasons, I concur.
. A trial court, of course, has the power to exclude evidence for reasons other than a violation of a discovery order, such as where newly discovered evidence is immaterial and its introduction would cause a serious disruption of the trial, the witness is deemed incompetent to testify, the evidence is hearsay, or there have been other infractions of the rules of evidence. See People v. Palomo, 35 F.3d 368, 374 (9th Cir.1994), cert. denied, — U.S. —, 115 S.Ct. 750, 130 L.Ed.2d 650 (1995). I cannot determine what the court would have done absent this specific discovery violation, and how I might view that decision, because here the court relied on the fact that this was a “substantial discovery violation,” and excluded the witnesses as a sanction for that breach.
. "Insofar as possible the habeas or appellate court shuns resolving credibility and weighing the evidence. Nevertheless, the Brecht-Kotteakos test for harmless error requires the habeas court to evaluate to some extent the probability of the outcome if the case were tried [with the excluded evidence].” Everette v. Roth, 37 F.3d 257, 262 (7th Cir.1994); Carter v. DeTella, 36 F.3d 1385, 1392 (7th Cir.1994); Cuevas v. Washington, 36 F.3d 612, 621 (7th Cir.1994).