We granted rehearing in banc in three unrelated cases to consider issues concerning the lawfulness of the exclusion of the public from a criminal trial during the testimony of an undercover police officer. Each case presents primarily two specific issues: (1) whether the prosecution established a sufficient justification for courtroom closure to override the defendant's usual right to a public trial, and (2) whether a trial judge, acting upon a request for courtroom closure, is required to consider sua sponte alternatives to closure during the testimony of one witness. These issues arise on appeals in three cases that considered petitions for habeas corpus filed by prisoners challenging state court convictions. In No. 95-2463, Steven Ayala appeals from the June 26, 1995, decision of the District Court for the Southern District of New York (John F. Keenan, Judge) denying his petition. In No. 95-2626, Peter J. Lacy, Superintendent of Bare Hill Correctional Facility, appeals from the June 19, 1995, judgment of the District Court for the Southern District of New York (Shira Scheindlin, Judge) granting the petition of Charles Okonkwo. In No. 95-2801, Howard Pearson appeals from the October 26, 1995, judgment of the District Court for the Southern District of New York (Peter K. Leisure, Judge), denying his petition.

We conclude that in all three cases the prosecution sufficiently justified the courtroom closure, and that a trial judge, having already considered closure during the testimony of one witness as an alternative to complete closure, is not required to consider sua sponte further alternatives to closure but needs to consider only further alternatives suggested by the parties. We therefore affirm in No. 95-2463 (Ayala) and No. 95-2801 (Pearson) and reverse in No. 95-2626 (Okonkwo).

All three petitioners were convicted in separate trials in New York Supreme Court of selling drugs. Their convictions were based on so-called "buy and bust" activities of undercover police officers. The officer, posing as a narcotics user, purchases the drugs, and the seller is arrested shortly thereafter. The details of the offenses are set forth in the panel opinions. See Ayala v. Speckard, 89 F.3d 91, 92 (2d Cir.) ("Ayala I "), modified on denial of rehearing, 102 F.3d 649 (2d Cir.1996) ("Ayala II "); Okonkwo v. Lacy, 104 F.3d 21, 22 (2d Cir.1997) ("Okonkwo "); Pearson v. James, 105 F.3d 828, 829 (2d Cir.1997) ("Pearson "). We detail separately the circumstances concerning the courtroom closure in each case.

In No. 95-2463 (Ayala), the State moved to close the courtroom to spectators during the testimony of Detective Willie Dotson, the undercover officer who purchased drugs from Ayala. At a hearing before the state court trial judge, Dotson testified that he had been making undercover drug purchases for two years and expected to continue doing so for approximately six months in the 41st Precinct of New York City, the precinct to which he was assigned. See Ayala I, 89 F.3d at 92-93. He also testified that he had been working in the specific area where Ayala's sale occurred for the past month and knew that he would be returning there after his trial testimony. Transcript of State Court July 15, 1991, Hearing ("Ayala Transcript") at 42. He identified that location as "the area around 1006 Intervale Avenue." Id. at 46. Dotson asserted that if he was recognized by people in the courtroom and then returned to "this particular area," he would "fear for [his] life." Ayala I, 89 F.3d at 93. He also testified that, on prior occasions while he was working undercover, people on the street had indicated to others that he was a police officer. He also admitted that he testified about the need for courtroom closure in every case in which his purchases occasioned a trial. The trial judge credited Dotson's testimony, Ayala Transcript at 50-52, and ordered the courtroom cleared of spectators during Dotson's trial testimony. The judge made clear that he would be "troubled by the testimony of an officer who simply makes this application in every case regardless of any circumstances which appl[y] specifically to the particular case in question," id. at 50, and relied primarily on the fact that Officer Dotson would be returning to the precise location in which he had arrested the defendant, id. at 51.

In No. 95-2626 (Okonkwo), the State moved at an in camera hearing to close the courtroom to spectators during the testimony of John Swift, the undercover officer who had purchased drugs from Okonkwo. He was a member of the Manhattan South Tactical Narcotics Team, with responsibility to investigate street sales of narcotics south of 59th Street in Manhattan. Swift testified that undercover officers follow the practice of "maintenance," returning to the location where they had previously made drug purchases. Transcript of State Court February 5-6, 1990, Hearing ("Okonkwo Transcript") at 4-5. He said that he expected to engage in "maintenance" in the area where he had purchased drugs from Okonkwo, id. at 7, which he described as "Cooper Square," id. at 6.

Swift testified that his life would be in danger if his identity as an undercover officer was "exposed to the community in the area where [he] would operate." Okonkwo, 104 F.3d at 23. The trial judge found that Swift continues to work in the "general area" where the crime occurred, and that, if his identity became known, his life or at least his continuing undercover activity would be jeopardized. He therefore ordered the courtroom closed to spectators during Swift's trial testimony. Okonkwo Transcript at 9.

In No. 95-2801 (Pearson), the State moved to close the courtroom to spectators during the trial testimony of Denise DiBenedetto, the undercover officer who purchased drugs from Pearson. Like the officer who purchased from Okonkwo, she was assigned to the Manhattan South Tactical Narcotics Team. DiBenedetto testified that she had been working as an undercover officer for 15 months and was currently active in the area of "West 42nd Street and Eighth Avenue" in Manhattan. Transcript of State Court September 7, 1990, Hearing ("Pearson Transcript") at 20. She said that she had been working in the described area 25 times in the past 30 days and was continuing to work there on ongoing investigations.

When asked what could happen if her identity was revealed to the public, she answered, "Okay my cover could be blown and I could get killed." Pearson Transcript at 22. The trial judge found that DiBenedetto was still active in the area where she bought drugs from Pearson and that she had reason to fear retaliation if her identity was disclosed. Responding to the defendant's contention that courtroom closure during the testimony of undercover officers would be "automatic" if permitted in Pearson's case, the trial judge disagreed, pointing out that in the last hearing he had held on a similar claim, he had denied closure. In Pearson's case, the judge ordered the courtroom closed to spectators during DiBenedetto's trial testimony.

Ayala. Ayala's conviction was affirmed, People v. Ayala, 202 A.D.2d 262, 608 N.Y.S.2d 642 (1st Dep't 1994), and leave to appeal to the New York Court of Appeals was denied, People v. Ayala, 83 N.Y.2d 908, 614 N.Y.S.2d 390, 637 N.E.2d 281 (1994). His petition for a writ of habeas corpus to challenge his conviction on the ground of a denial of his constitutional right to a public trial was denied by Judge Keenan.

A panel of this Court, consisting of Judges Cardamone, Altimari, and Parker, reversed. Ayala I, 89 F.3d at 92. Writing for the panel, Judge Altimari relied on two grounds. First, he ruled that, although the State had an overriding interest in protecting both "the safety, as well as the confidentiality" of an undercover officer, the State had failed to show "a substantial probability" that this interest "would likely have been prejudiced" if the officer had testified in open court. Id. at 95. He pointed out that the officer had testified that there was nothing special about Ayala's trial that made him fearful, and had not suggested that anyone in a position to "blow" his cover was likely to be in the courtroom when he testified. Id. The opinion appeared to rule that an inadequate showing of likely prejudice had been made with respect to the State's interest in either the undercover officer's "safety" or "cover." Id.

Second, Judge Altimari ruled that a trial judge has an obligation sua sponte to consider alternatives to courtroom closure, and that the failure to do so in Ayala's case was a further reason for ruling the closure to have violated the defendant's right to a public trial. Id. at 96-97.

The State petitioned for panel rehearing, contending both that the panel's rulings should be reconsidered and that the requirement of a trial judge's sua sponte consideration of alternatives to closure was a "new rule" that could not be applied retroactively in a habeas petition in light of Teague v. Lane, 489 U.S. 288, 299-310, 109 S. Ct. 1060, 1069-93, 103 L. Ed. 2d 334 (1989) (plurality opinion), and Penry v. Lynaugh, 492 U.S. 302, 313-14, 109 S. Ct. 2934, 2943-45, 106 L. Ed. 2d 256 (1989) (adopting plurality opinion in Teague ). In a per curiam opinion, the Ayala panel first rejected the Teague argument on the ground that the State had waived the argument by failing to assert it in opposition to Ayala's appeal from the denial of his habeas petition. Ayala II, 102 F.3d at 651. The panel pointed out that Ayala's appellate brief to the panel had urged both of the grounds on which the panel ultimately relied, thereby affording the State an opportunity to contend that agreement with either ground, especially the requirement of sua sponte consideration of alternatives, would be a new rule. Id.

Second, the panel ruled that, even if the Teague argument was not waived, no new rule had been announced since both grounds of the panel's original opinion relied on "a straightforward application of the constitutional standards enunciated" in prior Supreme Court decisions. Id. at 652. However, with respect to the first ground--substantial probability of prejudice to an overriding state interest, the rehearing opinion distinguished between the State's interest in the safety of the undercover officer and the State's interest in minimizing the risk of compromising the officer's effectiveness. The panel restated its view that the State's showing was insufficient with respect to the safety interest, but asserted that the initial panel opinion "did not address" whether the State had made the requisite showing concerning compromising the officer's effectiveness. Id. Finally, the rehearing panel restated its view that the trial judge must sua sponte consider alternatives to courtroom closure. Id. at 652-54. Ultimately, rehearing was denied solely for lack of sua sponte consideration of alternatives. Id. at 654.

Okonkwo. Okonkwo's conviction was affirmed, People v. Okonkwo, 176 A.D.2d 163, 574 N.Y.S.2d 186 (1st Dep't 1991), and leave to appeal to the New York Court of Appeals was denied, People v. Okonkwo, 79 N.Y.2d 862, 580 N.Y.S.2d 733, 588 N.E.2d 768 (1992). On his petition for a writ of habeas corpus to challenge his conviction on the ground of a denial of his constitutional right to a public trial, Judge Scheindlin concluded that the State had not shown an interest sufficient to warrant courtroom closure and that the state trial judge's findings were insufficient for such closure. She ruled that the writ should issue unless the state trial court determined, upon adequate findings after an evidentiary hearing, that the closure was warranted. See Okonkwo, 104 F.3d at 22. She also ruled that the petitioner's argument concerning lack of sua sponte consideration of alternatives had been procedurally defaulted for lack of explicit presentation to the state courts.

A panel of this Court, consisting of Judges Miner, Walker, and Leval, modified the District Court's ruling and affirmed, as modified. The panel first ruled, in agreement with Ayala I, that the undercover officer's "generalized concern for his safety" was insufficient to warrant closure. Okonkwo, 104 F.3d at 25. The panel next ruled that the District Court had failed to consider the State's interest in maintaining the effectiveness of the undercover officer. The panel explicitly identified that interest as a sufficient state interest for purposes of courtroom closure, id., and appeared to agree with the state trial and appellate courts that substantial probability of prejudice to that interest had been shown, id. Finally, the panel ruled that Okonkwo's argument concerning sua sponte consideration of alternatives had not been defaulted, and concluded, "constrained by Ayala," id. at 26, that lack of such consideration violated Okonkwo's right to a public trial. The panel also ruled that the Ayala violation required the petitioner's release in the absence of a retrial, rather than the renewed state court hearing that the District Court had ordered. Id.

Pearson. Pearson's conviction was affirmed by the Appellate Division, People v. Pearson, 186 A.D.2d 61, 588 N.Y.S.2d 146 (1st Dep't 1992), and by the Court of Appeals, People v. Martinez, 82 N.Y.2d 436, 604 N.Y.S.2d 932, 624 N.E.2d 1027 (1993). On his petition for a writ of habeas corpus to challenge his conviction on the ground of a denial of his constitutional right to a public trial, Judge Leisure concluded that the State had shown sufficient risk of prejudice to an overriding interest in the undercover officer's safety and cover and that sua sponte consideration of alternatives was not required of the trial judge. He therefore denied the petition. See Pearson, 105 F.3d at 830.

A panel of this Court, consisting of Judges Newman, Jacobs, and Cabranes, reversed. Id. at 831. The opinion of then-Chief Judge Newman concluded that the State had "adequately supported an overriding interest in protecting the undercover officer from exposure of her identity." Id. at 830. The opinion further ruled that the trial judge's failure to consider alternatives sua sponte violated the requirement set forth in Ayala I and Ayala II, and therefore reversed the denial of habeas relief. Judges Jacobs and Cabranes wrote a concurring opinion to express their disagreement with the Ayala requirement, though accepting its precedential force. Pearson, 105 F.3d at 831 (concurring opinion).

In banc rehearing. Rehearing in banc was sought and granted in Okonkwo and Pearson. Though our mandate had issued following the decision in Ayala II, we also voted to recall our mandate in that case and rehear it along with Okonkwo and Pearson. We did so because rehearing of the latter two appeals would inevitably involve reconsideration of the rulings of the panel that decided Ayala I and Ayala II, and that panel included two senior judges who could participate in the in banc (even to a limited extent) only if it included rehearing of a decision by a panel of which they were members, see 28 U.S.C. § 46(c)(1). We invited and received supplemental briefs.

The State respondents raise an issue preliminary to our consideration of the merits of the habeas petitions in all three cases. They contend that the petitioners are urging the application of a "new rule" within the meaning of Teague, 489 U.S. at 301, 315, 109 S. Ct. at 1070, 1077-78 (plurality opinion), which generally may not be applied retroactively in a habeas corpus proceeding to a state court conviction that has become final. See Penry, 492 U.S. at 313-14, 109 S.Ct. at 2943-45 (adopting the view of the Teague plurality opinion). Specifically, the respondents contend that we would be announcing a "new rule" if we accepted the petitioners' claims that the prosecution had failed to demonstrate a sufficient state interest to warrant courtroom closure during the testimony of the undercover officers, and that the trial judges had erred by not considering alternatives to closure sua sponte.

In Teague, the Supreme Court was asked to apply to the petit jury the "fair cross section" requirement applicable to selection of a jury venire. See Taylor v. Louisiana, 419 U.S. 522, 95 S. Ct. 692, 42 L. Ed. 2d 690 (1975). The Court ruled that before deciding the merits of that contention, it "should" ask whether such a rule would be applied retroactively, Teague, 489 U.S. at 300-01, 109 S.Ct. at 1069-70 (plurality opinion), a result that could occur only if the rule advocated by the habeas petitioner was "dictated by precedent existing at the time the defendant's conviction became final," id. at 301, 109 S.Ct. at 1070 (emphasis in original), and, if not, whether either of two exceptions to the Teague retroactivity principle applies, id. at 311, 109 S.Ct. at 1075 (exception for new rule that places primary conduct beyond proper law-making authority or requires observance of procedures implicit in concept of ordered liberty).

Where a court is asked to apply retroactively a rule announced prior to the filing of a habeas petition but after the petitioner's conviction has become final, it will normally be obliged to make a Teague analysis and determine whether the rule sought to be applied is "new." That was the context in which the Supreme Court recently decided O'Dell v. Netherland, --- U.S. ----, 117 S. Ct. 1969, 138 L. Ed. 2d 351 (1997). The Court had previously decided that a defendant facing the death penalty must be allowed to respond to a prosecutor's claim of future dangerousness by informing the jury that he will not be eligible for parole. See Simmons v. South Carolina, 512 U.S. 154, 114 S. Ct. 2187, 129 L. Ed. 2d 133 (1994). The habeas petitioner in O'Dell sought to have Simmons applied retroactively. Since the applicable rule had already been announced, the only issue was the rule's retroactivity. See Butler v. McKellar, 494 U.S. 407, 415, 110 S. Ct. 1212, 1217-18, 108 L. Ed. 2d 347 (1990) (determining that rule previously announced in Arizona v. Roberson, 486 U.S. 675, 108 S. Ct. 2093, 100 L. Ed. 2d 704 (1988), concerning effect of request for counsel in separate investigation, was new); Penry, 492 U.S. at 319-28, 109 S.Ct. at 2947-52 (determining that rule previously announced in Eddings v. Oklahoma, 455 U.S. 104, 102 S. Ct. 869, 71 L. Ed. 2d 1 (1982), and Lockett v. Ohio, 438 U.S. 586, 98 S. Ct. 2954, 57 L. Ed. 2d 973 (1978), concerning consideration of mitigating circumstances in death penalty sentencing, was not new). Our pending cases are procedurally more like Teague itself, since petitioners ask us to announce rules that the respondents regard as "new" and at the same time to apply such rules to habeas corpus petitions. This procedural distinction between Teague and O'Dell does not lessen the significance of Teague; it simply creates for the habeas court an opportunity, absent in the O'Dell context, to determine whether special circumstances exist that make it appropriate to consider and reject a claim on the merits, rather than limiting consideration to whether the claim, if sustained, would constitute a new rule.

In the special circumstances of these appeals, we do not believe that Teague precludes our consideration and rejection of the merits of the petitioners' claims. See Collins v. Youngblood, 497 U.S. 37, 40-41, 110 S. Ct. 2715, 2718, 111 L. Ed. 2d 30 (1990) ("Generally speaking, '[r]etroactivity is properly treated as a threshold question ....' ") (quoting Teague, 489 U.S. at 300, 109 S.Ct. at 1070 (emphasis added)). The Ayala panel has ruled that the state interests advanced to warrant closure during the testimony of an undercover officer were not sufficiently supported to warrant the closure that occurred and that a trial judge has an obligation sua sponte to consider alternatives even to closure during the testimony of one witness. If this in banc court were now to decide only that the rules announced in Ayala I and Ayala II are new, without deciding that they do not correctly interpret the Constitution, we would create considerable uncertainty for the district courts of this Circuit, and perhaps for the state courts as well. Though the in banc court could vacate the decisions of the Ayala panel, removing their precedential force, that course would leave uncertainty as to whether the decisions of that panel would again be made by the three judges of the Ayala panel in subsequent cases, and by other members of this Court. If that uncertainty can be dispelled by rejecting, not merely vacating, the decisions of the Ayala panel, district judges facing closure claims in the future will be usefully guided.

The present context also includes the unusual circumstance that the two principal issues raised in the three pending cases have both been recently decided by the New York Court of Appeals contrary to the decisions of the Ayala panel. See People v. Ramos, 90 N.Y.2d 490, 662 N.Y.S.2d 739, 685 N.E.2d 492 (1997). The interest in comity, which underlies Teague, 489 U.S. at 308, 109 S.Ct. at 1074, is plainly served by eliminating the tension between the panel's rulings as to Ayala and the Court of Appeals' rulings as to Ramos. Moreover, since our rulings on the merits leave the challenged convictions undisturbed, we do not impair the State's interest in finality, which also underlies Teague, id. We therefore turn to the merits of these appeals.

II. The Applicable Standards for Courtroom Closure

The Sixth Amendment guarantees every person accused in a criminal prosecution the right to a "public" trial. U.S. Const. amend. 6. That basic right has a long and distinguished history. See In re Oliver, 333 U.S. 257, 266-73, 68 S. Ct. 499, 504-08, 92 L. Ed. 682 (1948). It applies not only to the evidence phase of a criminal trial, see Gannett Co. v. DePasquale, 443 U.S. 368, 379, 99 S. Ct. 2898, 2905, 61 L. Ed. 2d 608 (1979), but also to other adversary proceedings, such as a pretrial suppression hearing, see Waller v. Georgia, 467 U.S. 39, 43, 104 S. Ct. 2210, 2213-14, 81 L. Ed. 2d 31 (1984). The explicit Sixth Amendment right of the accused is complemented by an implicit, "qualified" First Amendment right of the press and the public of access to a criminal trial. See Press-Enterprise Co. v. Superior Court, 478 U.S. 1, 9, 106 S. Ct. 2735, 2740-41, 92 L. Ed. 2d 1 (1986) ("Press-Enterprise II ") (access to transcript of preliminary hearing); Press-Enterprise Co. v. Superior Court, 464 U.S. 501, 104 S. Ct. 819, 78 L. Ed. 2d 629 (1984) ("Press-Enterprise I ") (access to jury selection); Globe Newspaper Co. v. Superior Court, 457 U.S. 596, 102 S. Ct. 2613, 73 L. Ed. 2d 248 (1982) (access to trial and pretrial hearings); Richmond Newspapers, Inc. v. Virginia, 448 U.S. 555, 100 S. Ct. 2814, 65 L. Ed. 2d 973 (1980) (access to trial).

However, closure of a criminal trial courtroom may constitutionally occur under limited circumstances. The strict standards for closure were first enunciated by the Supreme Court, with varying formulations, in cases considering the First Amendment access rights of the press and the public. Thus, in Richmond Newspapers, closure was permitted only upon a showing of "an overriding interest articulated in findings." Id. at 581, 100 S.Ct. at 2829 (plurality opinion of Burger, C.J.). In Globe Newspaper, the Court said that closure to "inhibit the disclosure of sensitive information" required a showing that denial of public access "is necessitated by a compelling governmental interest, and is narrowly tailored to serve that interest." 457 U.S. at 606-07, 102 S.Ct. at 2620. In Press-Enterprise I, the Court said, "The presumption of openness may be overcome only by an overriding interest based on findings that closure is essential to preserve higher values and is narrowly tailored to serve that interest." 464 U.S. at 510, 104 S.Ct. at 824.

These rigorous standards were explicitly applied to limitations on a defendant's Sixth Amendment right to a public trial in Waller, 467 U.S. at 47, 104 S.Ct. at 2216 ("In sum, we hold that under the Sixth Amendment any closure of a suppression hearing over the objections of the accused must meet the tests set out in Press-Enterprise [I] and its predecessors."). Waller reformulated the standards for courtroom closure into a four-factor test:

the party seeking to close the hearing must advance an overriding interest that is likely to be prejudiced,

the closure must be no broader than necessary to protect that interest,

the trial court must consider reasonable alternatives to closing the proceeding, and

it must make findings adequate to support the closure.

Id. at 48, 104 S.Ct. at 2216.

Shortly after Waller, in the context of closure of a preliminary hearing, ordered to avoid the adverse effects of pretrial publicity, the Court further refined the first factor to require "a substantial probability that the defendant's right to a fair trial will be prejudiced by publicity that closure would prevent," Press-Enterprise II, 478 U.S. at 14, 106 S.Ct. at 2743 (emphasis added), and repeated the third factor's requirement of a finding demonstrating that "reasonable alternatives to closure cannot adequately protect the defendant's fair trial rights," id.

A. First Waller Factor--the Interest Justifying Closure.

As noted, the Supreme Court has used various formulations to describe the gravity of the interest that will justify courtroom closure, as well as the degree of certainty that the asserted interest will be harmed. The interest has been described as "overriding," Waller, 467 U.S. at 48, 104 S. Ct. at 2216-17; Richmond Newspapers, 448 U.S. at 581, 100 S. Ct. at 2829-30, "compelling," Globe Newspaper Co., 457 U.S. at 607, 102 S.Ct. at 2620, and "cause shown that outweighs the value of openness," Press-Enterprise I, 464 U.S. at 509, 104 S.Ct. at 823. The Court has said that the asserted interest must be shown to be "likely to be prejudiced," Waller, 467 U.S. at 48, 104 S. Ct. at 2216, and has more recently required "a substantial probability" that the interest will be prejudiced, Press-Enterprise II, 478 U.S. at 14, 106 S.Ct. at 2743.

Our Court has endeavored to reconcile the various formulations in the Supreme Court decisions by relating the gravity of the interest asserted to the degree of closure requested. Thus, in Woods v. Kuhlmann, 977 F.2d 74, 76 (2d Cir.1992), we ruled that only a "substantial reason" was required to justify exclusion from the courtroom of members of the defendant's family during a witness's testimony, after the witness reported that some of the family members had threatened her. See Guzman v. Scully, 80 F.3d 772, 775 (2d Cir.1996) (applying "substantial reason" standard for partial closure). Other circuits have taken the same approach. See United States v. Sherlock, 962 F.2d 1349, 1357 (9th Cir.1992) (as amended) ("substantial reason" for partial closure); Nieto v. Sullivan, 879 F.2d 743, 753 (10th Cir.1989) (same); Douglas v. Wainwright, 739 F.2d 531, 533 (11th Cir.1984) (same). "The burden on the movant [for closure] to show prejudice increases the more extensive the closure sought." United States v. Doe, 63 F.3d 121, 129 (2d Cir.1995).

It may be doubted whether trial judges can make meaningful distinctions between "compelling" and "overriding" interests or can distinguish between whether such interests are "likely to be prejudiced" or whether there is a "substantial probability of" prejudice. We believe the sensible course is for the trial judge to recognize that open trials are strongly favored, to require persuasive evidence of serious risk to an important interest in ordering any closure, and to realize that the more extensive is the closure requested, the greater must be the gravity of the required interest and the likelihood of risk to that interest. After all, a word like "overriding" is really not a calibrated measure of the gravity of an interest; it reflects a conclusion that a particular interest asserted, together with the likelihood of risk to that interest, is sufficient to justify the degree of closure sought.

B. Third Waller Factor--Consideration of Alternatives.

The Ayala panel ruled that the third Waller factor, requiring consideration of reasonable alternatives to closure, imposed on a trial judge an obligation to consider alternatives sua sponte. In considering this issue, we gain some understanding of the Supreme Court's requirement by first recalling that the Court said that the standards enunciated in prior First Amendment cases apply to the Sixth Amendment right at issue in Waller, and then by examining what the Court meant by "alternatives" in those First Amendment cases. In Press-Enterprise I, after noting that the trial judge had "closed an incredible six weeks of voir dire without considering alternatives to closure," 464 U.S. at 513, 104 S.Ct. at 825, the Court suggested two possibilities: (1) "[t]hose parts of the transcript reasonably entitled to privacy could have been sealed without such a sweeping order," id., and (2) the trial judge might have "disclose[d] the substance of the sensitive answers while preserving the anonymity of the jurors involved," id. The import of the Court's concern appears to be that the trial judge erred in not considering alternatives to complete closure. See id. ("[T]here was also a failure to consider alternatives to closure and to total suppression of the transcript."). Similarly, in Press-Enterprise II, the Court faulted the California Supreme Court for its failure "to consider whether alternatives short of complete closure would have protected the interests of the accused." 478 U.S. at 14, 106 S. Ct. at 2743 (emphasis added).

Some, but not all, members of the majority are of the view that Waller appears to indicate that alternatives to complete closure are what the Court required trial judges to consider sua sponte when First Amendment closure standards are applied in the Sixth Amendment context. After noting that the trial judge had not made findings to justify "closure of the entire hearing," 467 U.S. at 48, 104 S. Ct. at 2216, the Court pointed out that the trial judge had not considered the alternatives of (1) "directing the government to provide more detail about its need for closure, in camera if necessary," id., and (2) "closing only those parts of the hearing that jeopardized the interests advanced," id. at 48-49, 104 S.Ct. at 2217.

Since none of the alternatives that the Supreme Court identified in Press-Enterprise I, Press-Enterprise II, or Waller had been suggested by any of the parties, it is arguable that the Court expects trial courts to consider lesser alternatives sua sponte only before taking the extreme step of closing an entire proceeding. That is the conclusion recently reached by the New York Court of Appeals. See Ramos, 90 N.Y.2d at 500-06, 662 N.Y.S.2d at 744-48, 685 N.E.2d at 497-501.

Moreover, even if the Supreme Court has imposed an obligation upon a trial judge to give sua sponte consideration to alternatives to complete courtroom closure, the Court has never held that a criminal case defendant who has not requested a more limited alternative has a right to a new trial, just because the trial judge failed to consider this or other alternatives sua sponte. The First Amendment cases did not require a new trial for any criminal defendant, and Waller, which ordered only the partial relief of a new suppression hearing, did so on review of the conviction of a defendant who had objected to the complete closure of the hearing that had occurred in that case. Until the decision of the Ayala panel, no case of which we are aware had ever reversed a criminal conviction because the trial judge failed to consider an alternative to courtroom closure (whether complete or partial) that had not been requested by the defendant.1 There is certainly no automatic basis for doing so. Though the right of the public and the press to attend a criminal proceeding absent circumstances justifying closure is of undoubted importance, the reversal of a criminal conviction for a trial judge's failure to consider an alternative not requested by a defendant is arguably too high a price to pay to protect that right.

In the pending cases, however, we need not decide whether a sua sponte obligation to consider alternatives to complete closure exists because the trial judges in these cases took the far lesser step of closing the courtroom only during the testimony of one witness, albeit an important one. Whether or not a sua sponte obligation exists to consider alternatives to complete closure, we see nothing in the First Amendment cases or in Waller to indicate that once a trial judge has determined that limited closure is warranted as an alternative to complete closure, the judge must sua sponte consider further alternatives to the alternative deemed appropriate. At that point, it becomes the obligation of the party objecting to the trial court's proposal to urge consideration of any further alternatives that might avoid the need for even a limited closure. This too is the conclusion reached by the New York Court of Appeals. See Ramos, 90 N.Y.2d at 500-05, 662 N.Y.S.2d at 744-48, 685 N.E.2d at 497-501.

The pending cases well illustrate the hazard of obliging a trial judge, who has already considered the alternative of partial closure during the testimony of one witness, to have a further obligation sua sponte to consider alternatives to the alternative. The petitioners suggest that among the further alternatives that the judge should have considered were disguising the undercover officer or placing a screen between the witness and the courtroom spectators. Yet refinements such as these encounter substantial objections. Disguising the witness risks lessening the jury's opportunity to observe the witness's demeanor and assess credibility, and a screen risks implying to the jury that the family or friends of the defendant in attendance are likely to be dangerous. Even if Waller requires a trial judge to consider alternatives to complete closure, we do not believe that the Supreme Court wanted trial judges selecting the alternative of limited closure to consider further alternatives that themselves pose substantial risks to a fair trial for the defendant. Of course, if some further alternative is suggested by the defendant or the prosecution, the trial judge should give it consideration.

III. Application of Closure Standards to the Pending Cases

A. First Waller Factor. In each of the three pending cases, the state court trial judges adequately determined that courtroom closure was warranted during the testimony of the undercover officer. The officers in all three cases testified that they were continuing their undercover work and would soon be returning in an undercover capacity to the same areas where the defendants had been arrested. These areas were described with particularity--"the area around 1006 Intervale Avenue" (Ayala ), "Cooper Square" (Okonkwo ), and "West 42nd Street and Eighth Avenue" (Pearson ).

The state interest in maintaining the continued effectiveness of an undercover officer is an extremely substantial interest, and the trial judge in each case was amply justified in concluding that this interest would be seriously prejudiced by requiring the officer to testify in an open courtroom. There is no requirement that the prosecution must prove that particular individuals likely to attend the trial will disclose the officer's identity. Of course, the defendant himself has an opportunity to observe the officer (a second opportunity, if the defendant is guilty), and might communicate a description of the officer to others, particularly if the defendant is at liberty pending trial. The defendant's right of presence at his trial requires accepting that risk, but the right to a public trial does not require the further risk that the officer's identity will become known through observation by members of the public who might enter the courtroom and see the officer testifying. The gravity of the state interest in protecting the secrecy of the officer's identity from casual observers and the likelihood that this interest will be prejudiced by the officer's testifying in open court are both sufficiently substantial to justify the limited closure of the courtroom during the officer's testimony. The closure is limited not only because it lasts only for the testimony of one witness, albeit an important witness, but also because there is no limitation at all on the right of the public or the press to examine the transcript of the officer's testimony. Since the state interest in maintaining the secrecy of the undercover officer's identity warranted the limited closure, we need not consider the respondents' additional point that the closure was also justified by the risk to the officer's safety.

B. Third Waller Factor. In all three cases, the trial judge closed the courtroom, with justification, only during the testimony of one witness, and did so without limiting access to the transcript of the officer's testimony. No additional alternatives were suggested by any party, and the trial judges had no obligation to consider additional alternatives sua sponte. We note that none of the defendants requested that family members be permitted to remain in the courtroom, a request that would have required careful consideration by the trial judge. See Vidal v. Williams, 31 F.3d 67, 68-69 (2d Cir.1994) (writ of habeas corpus granted because of unwarranted exclusion of defendant's family members); see also Guzman, 80 F.3d at 774-76 (writ of habeas corpus granted because, among other things, court failed to inquire into alleged exclusion of defendant's family members).

C. Second and Fourth Waller Factors. There is no substantial issue raised as to the second and fourth factors. The limited closure occurring only during the testimony of the undercover officer was "no broader than necessary," Waller, 467 U.S. at 48, 104 S. Ct. at 2216, to protect the interest in preserving the secrecy of the undercover officer's identity, and the trial judges in each case made findings "adequate to support the closure," id.Conclusion

For these reasons, in No. 95-2463 (Ayala) and No. 95-2801 (Pearson) we affirm the judgment of the District Court; in No. 95-2626 (Okonkwo) we reverse the judgment of the District Court and remand with directions to enter judgment dismissing the petition. We also vacate the decisions in Ayala I and Ayala II.

The majority's decision to reach the merits of this case despite the rule of Teague v. Lane, 489 U.S. 288, 300, 109 S. Ct. 1060, 1069-70, 103 L. Ed. 2d 334 (1989) (plurality opinion), is expressly limited to the very "special" and "unusual" circumstances of this case. Maj. Op. at 67-68. The majority notes that its decision to bypass Teague 's non-retroactivity principle is premised on the combination of (1) the uncertainty that would be created in this Circuit by vacating the panel opinion in Ayala v. Speckard, 102 F.3d 649, 651 (2d Cir.1996)(per curiam), in banc without discussing the merits, and (2) an existing tension concerning the merits between the federal and state courts of New York. Maj. Op. at 68-69. Thus, the majority's treatment of Teague is not particularly troubling from a precedential standpoint because I think we will not soon see another in banc case within such a unique context.

Nevertheless, I would not reach the merits of petitioners' claim that the Sixth Amendment mandates sua sponte consideration by the trial court of alternatives to a partial closure. This claim invites a "new" rule of constitutional law within the meaning of Teague, as recently illuminated by the Supreme Court's decision in O'Dell v. Netherland, --- U.S. ----, 117 S. Ct. 1969, 138 L. Ed. 2d 351 (1997)--issued after the panel opinions in these cases, and indeed, after the oral argument of the in banc appeals.

The majority holds that a trial court need not consider "alternatives to the alternative" of closing the courtroom for a single witness whose testimony is made public in the transcript, Maj. Op. at 71. Without quarreling with the majority's opinion on the merits--indeed, I would join it without hesitation if this case were before us on direct appeal and if the Teague rule therefore was not implicated--it undeniably addresses the proposal by the petitioners of a new constitutional rule that the Sixth Amendment public trial right of a defendant requires a trial court sua sponte to consider alternatives to partial closure. In my view, once the court determines that petitioners have proposed a new rule of constitutional law that, if adopted, could not be applied retroactively, the court should dismiss petitioners' claim at this threshold stage without considering the merits. Teague, 489 U.S. at 300, 109 S.Ct. at 1069-70.

"Retroactivity is properly treated as a threshold question, for, once a new rule is applied to the defendant in the case announcing the rule, evenhanded justice requires that it be applied retroactively to all who are similarly situated." Id. Thus, "[b]efore a state prisoner may upset his state conviction or sentence on federal collateral review, he must demonstrate as a threshold matter that the court-made rule of which he seeks the benefit is not 'new.' " O'Dell, at ----, 117 S. Ct. at 1973 (emphasis added).

Strong concerns for comity and finality caution against reaching the merits of a proposed rule unnecessarily on collateral review.1 See Teague, 489 U.S. at 308, 109 S.Ct. at 1074.

The costs imposed upon the State[s] by retroactive application of new rules of constitutional law on habeas corpus ... generally far outweigh the benefits of this application. In many ways the application of new rules to cases on collateral review may be more intrusive than the enjoining of criminal prosecutions, for it continually forces the States to marshal resources in order to keep in prison defendants whose trials and appeals conformed to then-existing constitutional standards. Furthermore, ... [s]tate courts are understandably frustrated when they faithfully apply existing constitutional law only to have a federal court discover, during a [habeas] proceeding, new constitutional commands.

Id. at 310, 109 S.Ct. at 1075 (internal quotation marks and citations omitted). The interests of comity and finality are so strong, that a court may consider the issue sua sponte, even where no party raises the issue. See, e.g., id. at 300, 109 S. Ct. at 1069-70 (noting that retroactivity issue was not raised except in an amicus brief). In Ayala, 102 F.3d at 651, a panel of this court concluded that the Teague issue had been waived. However, as in Teague, I believe that the circumstances of this case justify our sua sponte consideration of the issue regardless of whether the state has waived the argument.

In these cases, I easily conclude that the rule sought by petitioners--namely, that a district court has a sua sponte obligation to consider alternatives to a partial closure--is "new" for purposes of the Teague inquiry. "[A] case announces a new rule when it breaks new ground or imposes a new obligation on the States or the Federal Government." Teague, 489 U.S. at 301, 109 S.Ct. at 1070. Stated another way, "a case announces a new rule if the result was not dictated by precedent existing at the time the defendant's conviction became final." Id. The Supreme Court has recently explained that a rule is "new" for purposes of Teague "unless it can be said that a state court, at the time the conviction or sentence became final, would have acted objectively unreasonably by not extending the relief later sought in federal court." O'Dell, at ----, 117 S. Ct. at 1973. In short, "Teague asks state court judges to judge reasonably, not presciently." Id. at ----, 117 S.Ct. at 1978. Thus, in general, where a proposed rule is "susceptible to debate among reasonable minds," id. at ----, 117 S.Ct. at 1975 (internal quotation marks omitted), it is new.

Although I agree with the dissenting opinion's general approach to "new rule" analysis, I take issue with its conclusion. Under the standards of Teague and O'Dell, the rule sought by petitioners--that a trial court must sua sponte consider alternatives to partial closure--is plainly a new rule. Although in Waller v. Georgia, 467 U.S. 39, 47, 104 S. Ct. 2210, 2215-16, 81 L. Ed. 2d 31 (1984), the Supreme Court adopted the First Amendment four-part test for closure into the context of the Sixth Amendment, including a requirement that the court consider alternatives to closure, the Court was not directly faced with the question whether a trial court must consider alternatives to closure sua sponte. 467 U.S. at 48, 104 S. Ct. at 2216-17. And the nature of the public trial right in the Sixth Amendment context differs significantly from the public trial right that derives from the First Amendment. In the latter context, there is a risk that the only parties present--the prosecutor and the defendant--may agree that closure is proper, leaving the public's interest unrepresented unless the trial court assumes the responsibility of protecting that interest. In the context of the Sixth Amendment, however, the principal owner of the right--the defendant--is in the courtroom and can defend his own constitutional rights without the aid of the trial judge. The dissent is assuredly correct when it points out that "the public, judges, and the justice system" share an interest with the defendant in public trials. Dis. Op. at 76. However, even if sua sponte consideration of alternatives to closure is required to protect the interests of the public, it is far from clear to me why the district court's failure to engage in sua sponte consideration should inure to the benefit of a defendant who did not request (and may well not have wanted) any alternative, compare Ayala v. Speckard, 102 F.3d 649 (2d Cir.1996) (per curiam) with People v. Ramos, 90 N.Y.2d 490, 662 N.Y.S.2d 739, 685 N.E.2d 492 (1997) and People v. Martinez, 82 N.Y.2d 436, 604 N.Y.S.2d 932, 624 N.E.2d 1027 (1993) and Pearson v. James, 105 F.3d 828, 831 (2d Cir.1997) (concurring opinion). Indeed, Justice Powell, who authored the Court's opinion in Waller, expressed in his concurring opinion in Gannett Co. v. DePasquale, 443 U.S. 368, 401, 99 S. Ct. 2898, 2916-17, 61 L. Ed. 2d 608 (1979) (Powell, J., concurring), that, even in the context of the First Amendment, the party opposing the closure bears the burden of suggesting alternatives to complete closure.2 The history of these cases suggests that there is at the very least considerable room for reasonable minds to differ as to whether sua sponte consideration of alternatives to closure is required. It cannot be said that the state courts were objectively unreasonable in refusing to grant the relief requested on post-conviction review; the rule sought is therefore "new," and may not be applied retroactively.

I cannot agree with the implications in the majority opinion that the application of the Teague "new rule" doctrine is somehow affected by whether a reviewing court is asked both to announce a new rule and to apply it retroactively. Maj. Op. at 67-68. While it is true that sometimes a court is asked to apply a previously announced constitutional rule retroactively and therefore must only determine whether applying the rule retroactively would implicate comity and finality concerns, such cases are not the only ones in which the Teague "new rule" issue must be addressed as a threshold matter. Instead, "the question whether a decision [announcing a new rule should] be given prospective or retroactive effect should be faced at the time of [that] decision." Teague, 489 U.S. at 300, 109 S.Ct. at 1070 (internal quotations and citations omitted).

The majority also justifies its decision not to engage in the Teague inquiry on the "special circumstances" of these cases. Maj. Op. at 68. These special circumstances, according to the majority, arise from the "considerable uncertainty" resulting from the panel decisions. Maj. Op. at 68. While it may be true that a decision by the in banc panel on the merits of the proposed constitutional rule would "usefully guide[ ]" the lower courts, id., I do not agree that these special circumstances excuse us from engaging in the "new rule" inquiry as a threshold matter. In Teague itself, the Court declined to determine whether the fair cross section requirement applied to the petit jury despite the fact that the District Court for the Northern District of Illinois had rejected the claim, a three-judge panel of the Seventh Circuit had reversed, a majority of the judges on the Seventh Circuit voted to hear the case en banc and vacated the panel's decision, and finally the en banc panel rejected petitioner's fair cross section claim, holding that the fair cross section requirement was limited to the jury venire, with one judge dissenting. 489 U.S. at 294, 109 S.Ct. at 1066. Widespread actual confusion did not justify a merits ruling in Teague; the mere potential for confusion about the standards for closing courtrooms does not justify reaching the merits here. The fact that the same issue was recently decided by the New York Court of Appeals weighs against reaching the merits. See People v. Ramos, 90 N.Y.2d 490, 662 N.Y.S.2d 739, 685 N.E.2d 492 (1997). Indeed, it highlights the comity and finality concerns motivating the "new" rule inquiry.

Because I conclude the petitioners' proposed rule is "new" and thus could not be applied on collateral review, I would not reach the merits of the question of sua sponte consideration of alternatives to closure.

I concur in the judgment.

PARKER, Circuit Judge, dissenting.

Contrary to the majority, I believe that the Supreme Court's four-prong test set forth in Waller v. Georgia, 467 U.S. 39, 48, 104 S. Ct. 2210, 2216-17, 81 L. Ed. 2d 31 (1984), requires a trial judge to consider sua sponte alternatives to courtroom closure in a case where alternatives are not suggested by a party otherwise objecting to closure. This interpretation obeys the mandatory language of the Supreme Court's third Waller factor that "the trial court must consider reasonable alternatives to closing the proceeding," id. (emphasis added), and fulfills the requirement of the second Waller factor that "the closure must be no broader than necessary to protect [the] interest [of the party seeking closure]...." Id. (emphasis added).

Based on this 1984 holding in the Waller case, the rule requiring a judge to sua sponte consider reasonable alternatives to closure is hardly a "new" rule, as defined in Teague v. Lane, 489 U.S. 288, 299-310, 109 S. Ct. 1060, 1069-75, 103 L. Ed. 2d 334 (1989) (plurality opinion), which would otherwise preclude this Court from applying such a rule to these habeas petitioners in this case retroactively.1 See Penry v. Lynaugh, 492 U.S. 302, 313-14, 109 S. Ct. 2934, 2943-45, 106 L. Ed. 2d 256 (1989) (adopting plurality opinion in Teague ).

Petitioners' claims require us to decide whether the four prongs of the Waller test apply to a criminal trial in which the government requests the complete closure of the courtroom during the testimony of its key witness--an undercover police officer--in a "buy-and-bust" case. I believe that the trial courts in these three cases did not meet all four Waller prongs because each trial judge merely rubber-stamped the government's request for courtroom closure over the defendants' objections, and failed to consider sua sponte alternatives to closure so as to be no broader than necessary to protect the government's purported interest. Therefore, the trial courts ran afoul of Waller--and violated the petitioners' Sixth Amendment rights--when they failed to consider alternatives to closing the courtroom during the entire testimony of the undercover officers.

For these reasons, I dissent from the judgment. I would reach the same result as was reached previously in these appeals heard by separate panels of this Court, vacating the convictions and remanding for new trials within a reasonable time.

Before a criminal proceeding may be closed to the press and public, the plain language of Waller requires the trial court to consider, among other things, whether there are "reasonable alternatives to closing the proceeding." 467 U.S. at 48, 104 S. Ct. at 2216. In our opinion following the rehearing in Ayala, we held that Waller requires a trial judge to consider sua sponte less drastic alternatives where neither party has suggested any alternatives. Ayala v. Speckard, 102 F.3d 649, 654 (2d Cir.1996) (per curiam) ("Ayala II ").

In Waller, the Supreme Court specifically addressed the trial court's failure to consider reasonable alternatives. 467 U.S. at 48-49, 104 S.Ct. at 2216-17. The Court provided examples of reasonable alternatives that the trial court could have considered (e.g., providing more details about the need for closure, closing only certain parts of the hearing). Id. Nothing in Waller, however, suggests that the defendant in that case--who opposed closure--made any of these suggestions, other than offering a general objection to closure.

Undoubtedly, " '[t]he requirement of a public trial is for the benefit of the accused,' " Waller, 467 U.S. at 46, 104 S.Ct. at 2215 (quoting Gannett Co. v. DePasquale, 443 U.S. 368, 380, 99 S. Ct. 2898, 2906, 61 L. Ed. 2d 608 (1979)). The Sixth Amendment guarantees as much. But, the public, judges, and the justice system have equally substantial interests in public trials which closure offends. See Waller, 467 U.S. at 46, 104 S.Ct. at 2215; see also Press-Enterprise Co. v. Superior Ct. of Calif., 464 U.S. 501, 508, 104 S. Ct. 819, 823, 78 L. Ed. 2d 629 (1984) ("Press-Enterprise I "); In re Oliver, 333 U.S. 257, 271, 68 S. Ct. 499, 506-07, 92 L. Ed. 682 (1948). In fact, there is no set formula on how to allocate "the 'right' to openness as between the accused and the public, or whether it is a component inherent in the system benefitting both...." Press-Enterprise I, 464 U.S. at 508, 104 S.Ct. at 823. With regard to the constitutional guarantee of the public right to trial, the First and Sixth Amendments are inextricably linked, serving the same ends within our political system. The whole notion of trial being "public" implies that people other than the defendant must have a stake in this right in order to give it any meaning. Hence, the Sixth Amendment cannot be solely for the benefit of the defendant even though he may be the most direct beneficiary. This strong Sixth Amendment requirement of a public trial in criminal cases emphasizes that publicity is "of critical importance to our type of government in which the citizenry is the final judge of the proper conduct of public business." Cox Broadcasting Corp. v. Cohn, 420 U.S. 469, 495, 95 S. Ct. 1029, 1046, 43 L. Ed. 2d 328 (1975). Also, the First Amendment has been held to permit the press and public to intervene in all stages of a criminal proceeding in order to prevent closure of the courtroom. See, e.g., Press-Enterprise v. Superior Ct. of Calif., 478 U.S. 1, 10, 106 S. Ct. 2735, 2741, 92 L. Ed. 2d 1 (1986) ("Press-Enterprise II ").

The majority does not dispute that the Sixth Amendment guarantee to a public trial is at least as strong as the First Amendment guarantee. In Press-Enterprise II, a First Amendment case, the Supreme Court reversed a courtroom closure in part because the trial court "failed to consider whether alternatives short of complete closure would have protected the interests of the accused." 478 U.S. at 14, 106 S. Ct. at 2743. Significantly, no one in that case suggested any alternatives because both parties sought closure. The trial court's failure to consider alternatives sua sponte, in part, resulted in a reversal of that court's decision. In other words, the Supreme Court realized that, when faced with requests for closure of proceedings, the trial court must consider alternatives sua sponte to protect its own interests in maintaining the appearance of fairness as well as those interests not capable of being represented by the parties at trial, such as the public and the press.2

Furthermore, Waller makes it clear that courtroom closure must be no broader than necessary. See 467 U.S. at 48, 104 S. Ct. at 2216. Implicit in that command is the requirement to consider alternatives less broad than complete closure. See United States v. Peters, 754 F.2d 753, 761 (7th Cir.1985)(overturning closure and exclusionary orders where trial court failed to "consider fully on the record other alternatives to closure, or to narrowly tailor its closure order").

Numerous alternatives to closure existed in the cases at bar. The court could have limited access to the courtroom; a screen could have been erected between the undercover officer and the audience; certain individuals, such as relatives, could have been permitted to remain in court; the officer could have worn a disguise. The trial court was also obligated by Waller to make findings as to why these alternatives would not protect the interest sought to be protected by closure. See 467 U.S. at 48, 104 S. Ct. at 2216. While ultimately the court may have rejected such alternatives as unnecessary, cumbersome, or even unfair to the defendant, it nonetheless had a duty to consider whether these alternatives might better protect the defendants' open trial rights. See, e.g., Peters, 754 F.2d at 761-62 (error for trial judge not to consider alternatives available to him under Waller ); see also Davis v. Reynolds, 890 F.2d 1105, 1110-12 (10th Cir.1989); Jones v. Henderson, 683 F. Supp. 917, 923-24 (E.D.N.Y.1988).

The Supreme Court has unambiguously held that court proceedings are presumed to be open and that "[t]he presumption of openness may be overcome only by an overriding interest based on findings that closure is essential to preserve higher values and is narrowly tailored to serve that interest." Press-Enterprise I, 464 U.S. at 510, 104 S. Ct. at 824. To give this presumption any meaning, once a party objects to closure, the court must--sua sponte if necessary--consider alternatives. Moreover, in light of the presumption, it is proper that the party seeking to deviate from the constitutional norm bears the burden of raising alternatives to full closure.

Furthermore, Waller does not hold that a court must consider alternatives to closure only where the party seeking closure suggests alternatives. Rather, Waller requires trial courts to consider alternatives in every case where closure is sought. Logically, this means that where the party seeking closure fails to suggest any alternatives, the court must do so sua sponte.

The majority acknowledges that Supreme Court precedents arguably indicate that trial courts are expected to "consider lesser alternatives sua sponte " prior to completely closing a proceeding. Ante at 71. Regarding these three cases before us on collateral appeal, however, the majority avoids applying Waller 's clear mandate to consider alternatives by creating an illogical and unprecedented distinction between what it terms as "complete" and "limited" closures. Ante at 71.

According to the majority, Waller at most requires sua sponte consideration only when there has been a "complete" closure of a proceeding, meaning that the press and public have been excluded from an entire criminal proceeding (e.g., an entire suppression hearing, an entire voir dire, an entire preliminary hearing, or presumably an entire trial). The majority would not apply the sua sponte requirement of Waller to so-called "limited" closures, where the courtroom is closed, not for an entire proceeding, but just for the testimony of one witness in a criminal trial. The majority rationalizes this distinction by stating that, in the case of "limited" closures, the trial court has already considered alternatives to "complete" closure, that is, the "limited" closure.

As a preliminary matter, I have found no federal case in this Circuit or beyond that creates what I regard as an artificial distinction between "entire" and "limited" closure when applying the third prong of Waller. More importantly, it is difficult to understand the logic of the distinction. In these buy-and-bust cases, there are typically two or three witnesses: the undercover detective who makes the buy and sometimes an arresting officer or a forensic lab chemist who confirms that the substance purchased was an illegal drug. Closure during the testimony of the undercover detective is closure of most of the trial. In contrast, in most cases a suppression hearing or other preliminary hearing constitutes a very minor (although often important) portion of the presentation of a criminal case. It makes far more sense to conclude that full closure occurs when all representatives of the public are excluded and partial closure occurs when some observers are permitted whether the exclusion is for the whole proceeding or only a portion thereof.

Further, the majority's distinction between "entire" and "limited" closure creates a misleading certitude for the application of Waller to courtroom closures. Using the majority's reasoning, whenever the government seeks courtroom closure during the testimony of an undercover witness who intends to continue his work in a buy-and-bust case, the trial courts may grant closure with no further discussion because this closure has been deemed "limited" by this Court. This amounts to a per se rule that violates the balancing of interests required by Waller and demonstrates the danger of not requiring the trial judge to consider sua sponte reasonable alternatives to closure on the record.

Moreover, this rationale fails to account for the true reason why, in some circumstances, "limited" closures may be allowed consistent with the Sixth Amendment: because other safeguards are in place to ensure that the defendant receives a fair trial. For example, when only specific individuals are excluded from the courtroom during the testimony of one witness, "an audience remains to ensure the fairness of the proceedings." United States v. Osborne, 68 F.3d 94, 98-99 (5th Cir.1995) (upholding exclusion of one of defendant's family members during the testimony of a minor victim). This form of limited closure "does not implicate the same secrecy and fairness concerns that a total closure does." Woods v. Kuhlmann, 977 F.2d 74, 76 (2d Cir.1992); accord Osborne, 68 F.3d at 99; United States v. Farmer, 32 F.3d 369, 371 (8th Cir.1994); United States v. Sherlock, 962 F.2d 1349, 1356-57 (9th Cir.1992); Nieto v. Sullivan, 879 F.2d 743, 753-54 (10th Cir.1989); Douglas v. Wainwright, 739 F.2d 531, 532-33 (11th Cir.1984). Therefore, in such cases, only a "substantial reason," as opposed to Waller 's "overriding interest," is required to justify closure. See Woods, 977 F.2d at 76; Osborne, 68 F.3d at 99; Farmer, 32 F.3d at 371; Sherlock, 962 F.2d at 1357; Nieto, 879 F.2d at 753; Douglas, 739 F.2d at 533.

What this analysis illustrates is that closure cases should not be categorized--as the majority suggests--on the basis of whether an entire proceeding or just part of a proceeding is closed. Rather, the relevant inquiry is whether the closure at issue excludes so much of the audience that the safeguards inherent in the public trial right are no longer in place. Once this occurs, the closure is "complete," and Waller 's four factors come into play. As I discussed in the preceding section, Waller requires the trial judge to consider, either sua sponte or at a party's suggestion, alternatives to closure to protect the defendant's Sixth Amendment rights.

In the three cases before us on rehearing, the closure that occurred may only be categorized as "complete." Upon the state's motion, and over the defendants' objections, the trial judge closed the courtroom to all members of the defendant's family, the press, and the public during the testimony of the undercover officer. In buy-and-bust cases, the prosecution invariably centers around this witness: the undercover officer who purchased the drugs provides the only testimony as to the defendant's identity as the seller. The only additional testimony is provided by the arresting officer and, in some cases, a police chemist.

The majority, however, claims that the trial judges all considered the government's request for closure as a reasonable alternative to closure of the entire trial from start to finish. Thus, in the majority's opinion, the trial judges need not consider "alternatives to the alternative" of completely closing the courtroom during the testimony of one witness. Ante at 71. This is an erroneous, post hoc construction of the events at these trials.

First, the government never asked for closure of the entire trial proceeding. As a consequence, the trial judge never considered closure during the testimony of one witness to be an alternative to that request which violated the third prong of Waller. 467 U.S. at 48, 104 S. Ct. at 2216.

Second, even crediting the majority's assumption that the trial judges in these cases considered reasonable alternatives as required under Waller, these judges still would have violated Waller 's fourth prong, which mandates that this consideration of alternatives be on the record for review upon appeal. Id. In none of these trials, however, did the judge preserve his supposed consideration of so-called "limited" closure on the record. And, a silent consideration is not enough when a trial judge is closing the courtroom to the public, impinging on important First and Sixth Amendment guarantees. See id.

Third, because the undercover officer's testimony provides the substance of the prosecution's case in buy-and-bust cases, closure of the courtroom during this witness's testimony compromises the interests that the Sixth Amendment was designed to protect. For example, the Sixth Amendment was designed to, among other things, safeguard the integrity of witnesses' testimony and the public's perception of the judicial process. See Waller, 467 U.S. at 46 & n. 4, 104 S. Ct. at 2215 & n. 4 (citing cases); Woods, 977 F.2d at 76; see also Estes v. Texas, 381 U.S. 532, 588, 85 S. Ct. 1628, 1662, 14 L. Ed. 2d 543 (1965) (Harlan, J., concurring) ("[T]he public-trial guarantee embodies a view of human nature, true as a general rule, that judges, lawyers, witnesses, and jurors will perform their respective functions more responsibly in an open court than in secret proceedings."). By closing the courtroom during the undercover officer's testimony, the trial judge has eliminated the pressure that an audience would assert on the officer to testify with complete honesty. Because the police officer is the prosecution's main witness, this lapse is significant. In addition, in closing the courtroom for the majority of the trial, the judge has also compromised the public appearance of fairness and perhaps jeopardized public confidence in the judicial system. Cf. Oliver, 333 U.S. at 270-71, 68 S.Ct. at 506-07; Press-Enterprise I, 464 U.S. at 508, 104 S.Ct. at 823 (citing Richmond Newspapers, Inc. v. Virginia, 448 U.S. 555, 569-71, 100 S. Ct. 2814, 2823-24, 65 L. Ed. 2d 973 (1980)).

In sum, I believe that the closure at issue in these cases excluded so much of the audience that the safeguards inherent in the public trial right were no longer in place. Therefore, the closure was "complete," and Waller 's four factors should have come into play. As previously stated, Waller requires the trial judge to consider, either sua sponte or at a party's suggestion, alternatives to closure. Because the trial judge in these cases did not consider alternatives, the closures violated the defendants' Sixth Amendment rights.

The Supreme Court has laid the bedrock constitutional principle that "[c]losed proceedings, although not absolutely precluded, must be rare and only for cause shown that outweighs the value of openness." Press-Enterprise I, 464 U.S. at 509, 104 S.Ct. at 823.3 In New York, however, courtroom closure during the testimony of an undercover officer in buy-and-bust cases is so common that trial closure has been the basis of appeal in numerous cases.4 Presumably, countless more were not appealed. At the state level, courtroom closure was upheld in all but five of these cases. See People v. Nieves, 90 N.Y.2d 426, 430, 683 N.E.2d 764, 660 N.Y.S.2d 858, 861 (1997); People v. Tolentino, 90 N.Y.2d 867, 869 684 N.E.2d 23, 661 N.Y.S.2d 593, 595 (1997); People v. Campos, 657 N.Y.S.2d 48, 49 (1st Dep't 1997); People v. Bobo, 653 N.Y.S.2d 617, 618 (2d Dept.1997); People v. Valenzuela, 234 A.D.2d 193, 194-95, 652 N.Y.S.2d 5, 6 (1st Dept.1996).5

As further evidence of the volume of the cases where closure is sought, the government during oral argument in this in banc appeal stated that hundreds or thousands of convictions would be overturned should this Court uphold the rule announced in Ayala.6 It is galling to my sense of fairness that courtroom closure is such a routine practice in New York buy-and-bust cases. Cf. Abraham G. Gerges, A Judicial Dilemma, 211 N.Y.L.J. 83:2 (1994) (describing trial judges "routinely" closing courtrooms during testimony of undercover officers in New York); see also John M. Leventhal, Public Trial: Keeping the Undercover "Undercover", 208 N.Y.L.J. 87:1 (1992) ("Although the courts [of New York] reject a per se exception for an undercover witness, a showing of almost any factor will justify closure.").

The wholesale "closure of courtrooms during the testimony of undercover police officers is strictly a New York phenomenon." Robin Zeidel, Note, Closing the Courtroom for Undercover Police Witnesses: New York Must Adopt a Consistent Standard, 4 J.L. & Pol'y 659, 663 (1996). In other jurisdictions where undercover officers may face the same dangers as those in New York, the reliance on courtroom closure has not developed on this scale. This is evidenced, in part, by the fact that very few other courts have addressed courtroom closure on a regular, on-going basis like New York.7

The practice of closing courtrooms in New York to protect undercover officers' identities has existed since at least the 1970s. See United States ex rel. Lloyd v. Vincent, 520 F.2d 1272, 1273 (2d Cir.1975) (upholding constitutionality of courtroom closure during testimony of two state undercover officers); People v. Hinton, 31 N.Y.2d 71, 73-74, 286 N.E.2d 265, 266-67, 334 N.Y.S.2d 885, 887-89 (1972) (ruling that testimony of undercover narcotics officer was exception to right to public trial). Recently, New York has stepped up its use of buy-and-bust operations. Clifford Krauss, Giuliani Sets New Policy to Spur Drug Arrests by Officers on Beats, N.Y. Times, Apr. 7, 1994, at A1 (reporting that New York City police will conduct "aggressive buy and bust operations"); Zeidel, supra, 4 J.L. & Pol'y at 679 n. 62 (citing New York City Police Department, Policy Strategy No. 3: Driving Dealers Out of New York 6 (1994)). For instance, from 1993 to 1994, misdemeanor narcotics arrests in New York City increased by almost fifty percent. Samson Mulugeta, Drug War's Smaller Battles, N.Y. Daily News, Apr. 9, 1995, at 1.

The testimony of undercover officers is central to obtaining convictions in buy-and-bust narcotics cases. See, e.g., Vincent, 520 F.2d at 1273; Hinton, 31 N.Y.2d at 74, 286 N.E.2d at 266-67, 334 N.Y.S.2d at 889. Usually, the undercover officer is the sole identifying witness called to the stand by the government. See People v. Boyd, 59 A.D.2d 558, 559, 397 N.Y.S.2d 150, 151 (2d Dep't 1977) (finding that only eye witness was undercover officer). Concededly, buy-and-bust operations put undercover officers at risk. Gerges, supra (finding that the open or partially closed courtroom "could prove deadly to some undercover officers").

Nonetheless, New York's adoption of a law enforcement practice that continually conflicts with defendants' and the public's constitutional rights is highly suspect. In addition, even after Waller was adopted as the law of New York, see People v. Kin Kan, 78 N.Y.2d 54, 57-58, 574 N.E.2d 1042, 1043-45, 571 N.Y.S.2d 436, 438 (1991), lower state courts have been inconsistent in their adherence to Waller as announcing a constitutional rule. See People v. Brown, 172 A.D.2d 844, 846, 569 N.Y.S.2d 208, (2d Dep't 1991) (upholding closure during undercover officer testimony without using Waller 's test); People v. White, 170 A.D.2d 629, 566 N.Y.S.2d 401 (2d Dep't) (using New York's older "compelling reasons" test superseded by Waller ). Thus, careful judicial scrutiny should be applied to these decisions.

I do not believe that this holding articulates a "new" rule of criminal procedure. I believe that the rule was laid out explicitly in Waller and other relevant precedent, all of which preceded the date upon which the petitioners' conviction became final and before their habeas petitions were filed. Therefore, I believe that Teague does not prevent us from applying that rule. See 489 U.S. at 310, 109 S. Ct. at 1075.8

In our per curiam opinion following rehearing in Ayala II, we specifically rejected the state's argument that requiring sua sponte consideration of alternatives to closure broke new constitutional ground. The Ayala panel reasoned, correctly, that Waller, and both Press-Enterprise I and Press-Enterprise II, were "existing precedent" during the relevant state court proceedings and that the initial Ayala opinion was "a straightforward application of the constitutional standards enunciated in those cases." 102 F.3d at 652.

In addition, neither the legal standards nor factual framework of the recent Supreme Court case of O'Dell v. Netherland, --- U.S. ----, 117 S. Ct. 1969, 138 L. Ed. 2d 351 (1997) alter our earlier determination that Ayala and its progeny do not announce a "new" constitutional rule. In O'Dell, the Supreme Court reiterated the Teague test in a three-step inquiry. First, the reviewing court determines the date upon which the defendant's conviction became final. Id. at ----, 117 S. Ct. at 1973 (citing Lambrix v. Singletary, --- U.S. ----, ----, 117 S. Ct. 1517, 1524, 137 L. Ed. 2d 771 (1997)). Second, the court considers whether the habeas petitioner relies upon a "new" rule in seeking to overturn his conviction. A rule is "new" if it was announced by a case that was decided after the petitioner's state conviction became final, although this is subject to a wide range of interpretations and applications by reviewing courts. See O'Dell, at ----, 117 S. Ct. at 1973 (citing cases that lay out slightly differing standards on how to decide whether a rule is "new").

A rule is not "new" if "it has simply applied a well-established constitutional principle to govern a case which is closely analogous to those which have be previously considered in the prior case law." Penry v. Lynaugh, 492 U.S. 302, 314, 109 S. Ct. 2934, 2944, 106 L. Ed. 2d 256 (1989) (quoting Mackey v. United States, 401 U.S. 667, 695, 91 S. Ct. 1160, 1181 (1971) (Harlan, J., concurring in judgments in part and dissenting in part)). If a "rule" is not deemed "new" but "old" (or simply "not new"), then the reviewing court must apply the rule to a petitioner's collateral appeal either as the relevant controlling precedent or under principles of retroactivity. See Penry, 492 U.S. at 315, 109 S.Ct. at 2945 (finding that retroactivity principles applied because Supreme Court decisions upon which petitioner relied were decided prior to time conviction became final) (citing Griffith v. Kentucky, 479 U.S. 314, 107 S. Ct. 708, 93 L. Ed. 2d 649 (1987)).

Third, if a reviewing court finds that the rule is indeed "new," the rule may nonetheless be applied retroactively if it fits one of two exceptions: (1) "[the rule] places certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe;' " or (2) the rule "requires the observance of 'those procedures that ... are "implicit in the concept of ordered liberty." ' " Teague, 489 U.S. at 307, 109 S.Ct. at 1073 (citations omitted); see also O'Dell, at ----, 117 S. Ct. at 1973.

Under the Teague doctrine, a rule is considered "new" on a case-by-case basis relative to each individual habeas petitioner. In all relevant precedent, "new" rules arise from specific judicial decisions, made by the Supreme Court and relied upon by the habeas petitioners to challenge the constitutionality of state proceedings. See, e.g., Teague, 489 U.S. at 294, 109 S. Ct. at 1066 (petitioner unsuccessfully relying on Batson v. Kentucky, 476 U.S. 79, 106 S. Ct. 1712, 90 L. Ed. 2d 69 (1986)); Penry, 492 U.S at 315, 109 S. Ct. at 2945 (petitioner successfully relying on Eddings v. Oklahoma, 455 U.S. 104, 102 S. Ct. 869, 71 L. Ed. 2d 1 (1982) and Lockett v. Ohio, 438 U.S. 586, 98 S. Ct. 2954, 57 L. Ed. 2d 973 (1978)); O'Dell, at ----, 117 S. Ct. at 1978 (petitioner unsuccessfully relying on Simmons v. South Carolina, 512 U.S. 154, 114 S. Ct. 2187, 129 L. Ed. 2d 133 (1994)). Thus, "new" rules have two aspects: one based on the date on which a court announced the relied-upon rule and the other based on the reviewing court's determination that the rule clearly applied to the petitioner's case before his conviction became final. The latter aspect gives rise to the "inevitable difficulties" of deeming a rule "new" or not. Penry, 492 U.S. at 314, 109 S.Ct. at 2944-45 (quoting Mackey, 401 U.S. at 695, 91 S.Ct. at 1181).

Not surprisingly, the tests used by the Supreme Court to determine whether a rule is "new" or not have varied somewhat each time a collateral appeal of this sort has reached the Supreme Court. In Teague, the Supreme Court stated 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] if the result was not dictated by precedent existing at the time that the defendant's conviction became final." 489 U.S. at 301, 109 S. Ct. at 1070 (citations omitted). Recently, the Court restated these principles strongly in O'Dell:

[W]e will not disturb a final state conviction or sentence unless it can be said that a state court, at the time the conviction or sentence became final, would have acted objectively unreasonably by not extending the relief later sought in federal court ... [and] "would have felt compelled by existing precedent to conclude that the rule ... was required by the Constitution."

--- U.S. at ----, 117 S. Ct. at 1973 (quoting Saffle v. Parks, 494 U.S. 484, 488, 110 S. Ct. 1257, 1260, 108 L. Ed. 2d 415 (1990)) (alteration and citation omitted).

The limitations on collateral review set out in O'Dell and Teague can only be understood in light of the purpose of federal collateral review--"to create an incentive for state courts to conduct their proceedings in a manner consistent with established constitutional standards." Wright v. West, 505 U.S. 277, 311, 112 S. Ct. 2482, 2500, 120 L. Ed. 2d 225 (1992) (Souter, J., concurring) (quotation marks and citations omitted).9 Thus, neither O'Dell nor Teague may be understood to mean that federal habeas courts must defer to state courts' interpretation of the United States Constitution and Supreme Court precedents, such as Waller here.10

Rather, O'Dell must be read to permit the federal court reviewing the state conviction or sentence to place itself in the position of a state court at or before the state court made the conviction final. From this point of view and this point in time, the federal court must then ask itself whether the state court "would have acted objectively unreasonably by not extending the relief later sought in federal court ... [and whether it] would have felt compelled by existing precedent to conclude that the rule ... was required by the Constitution." O'Dell, at ----, 117 S. Ct. at 1973 (quotation marks and citation omitted).

In my view, petitioners are entitled to habeas relief because they rely on a rule announced in Waller which was not "new" at the time that their convictions became final. The earliest of the petitioners' convictions to become final was in early 1992.11 Waller was decided in 1984 and Press-Enterprise II was decided in 1986. And, in 1991, the New York Court of Appeals incorporated the four Waller factors into state law in People v. Kan, 78 N.Y.2d at 57-58, 574 N.E.2d at 1043-45, 571 N.Y.S.2d at 438. Thus, at the time that petitioners' convictions became final, Waller supplied the rule of decision and, therefore, was not "new." See Penry, 492 U.S. at 315, 109 S.Ct. at 2945.

As discussed extensively in Parts I-III of this dissent, Waller mandates that a trial court consider sua sponte reasonable alternatives to complete courtroom closure. Hence, by 1992, when the first of the petitioners' conviction became final, requiring consideration of alternatives by the trial judge was a well-established constitutional principle undergirding the right to public trial. See, e.g., Davis v. Reynolds, 890 F.2d at 1112 (vacating and remanding the conviction of habeas petitioner because trial court failed to follow Waller prongs, including third, in violation of defendant's Sixth Amendment rights); see also In re Herald Co., 734 F.2d 93, 100 (2d Cir.1984) (Newman, J.) (holding, in the same year as Waller, that "the trial judge must consider alternatives and reach a reasoned conclusion that closure is a preferable course to follow to safeguard the interests at issue"); Jones v. Henderson, 809 F.2d 946, 951-52 (2d Cir.1987) (recognizing precedential force of Waller and remanding habeas petition to district court to determine if case represented change in law); United States v. Peters, 754 F.2d at 761-63 (vacating closure order because trial judge failed to consider alternatives available to him required by Waller ).12

In New York, the Waller test has been adopted to apply to cases, like these in banc appeals, involving the closure of a courtroom during the testimony of the government's key witness in a drug crime prosecution. See People v. Kan, 78 N.Y.2d at 57-58, 574 N.E.2d at 1043-45, 571 N.Y.S.2d at 438 (citing Waller and holding that the closure of a courtroom to all spectators, including defendant's family, during testimony of key cooperating witness violated defendant's right to public trial); People v. Martinez, 82 N.Y.2d 436, 442-43, 624 N.E.2d 1027, 1030-31, 604 N.Y.S.2d 932, 935-37 (1993) (citing Waller and holding, in part, that unparticularized testimony of undercover officer about his continuing activity in the Bronx was insufficient to warrant courtroom closure).

In Martinez, the Court of Appeals was faced with the question of whether the third prong of Waller required the trial court to consider reasonable alternatives to closure suggested by the defendant. 82 N.Y.2d at 443-44, 624 N.E.2d at 1031-32, 604 N.Y.S.2d at 936-37. This question was posed by the direct appeal of Pearson, a petitioner before this Court here. The Court of Appeals struck down Pearson's appeal, stating "[i]t is surely the better practice, and indeed may be the required practice, for a trial court to explicitly to consider whether closure is broader than necessary, to explore reasonable alternatives to closure and to make findings adequate to support closure" and affirming his conviction. Id., 82 N.Y.2d at 444, 624 N.E.2d at 1031, 604 N.Y.S.2d at 936 (citing Waller, 467 U.S. at 48, 104 S. Ct. at 2216) (emphasis added).

The ruling in Martinez, quoted above, leaves open the possibility for a federal court, like the panel in Ayala I, to overturn Pearson's and these other petitioner's convictions because the Court of Appeals baldly misstated the constitutional rule announced in Waller and erred in its application to Pearson. As I have stressed repeatedly herein, the Supreme Court in Waller did not hold that it "may be the required practice," to address its four prongs; rather, the Supreme Court held that it is the mandatory practice to address all of these four prongs so as to protect the Sixth Amendment constitutional rights of the defendant. See, e.g., Waller, 467 U.S. at 48, 104 S. Ct. at 2216 (prong three states that trial courts "must consider reasonable alternatives"). Surely, a federal court must overturn these petitioners' convictions on collateral appeal when the highest state court misapplies the very constitutional precedent relied upon incorrectly by that state court to affirm them.

This equivocation of the Court of Appeals in Martinez with respect to the Waller test is significant because the state's high court decision seems to be objectively unreasonable by not extending the relief later sought in federal court by these petitioners. See O'Dell, at ----, 117 S. Ct. at 1973. In addition, it is clear from the state and federal cases which existed at the time that petitioners' convictions became final that the Court of Appeals did feel compelled by existing precedent to conclude that the four prongs of Waller were required by the Constitution. See id. Nonetheless, the Court of Appeals misstated Waller 's requirements and, thus, applied its test concerning consideration of reasonable alternatives improperly to at least one, if not all, of these petitioners.13

It is of little or no import to the above analysis that the New York Court of Appeals has changed its thinking on this contentious aspect of Waller, recently holding that, where closure "is not facially overbroad," the party opposing closure bears the burden of suggesting reasonable alternatives to the court. See People v. [Robert] Ayala, 90 N.Y.2d 490, 685 N.E.2d 492, 662 N.Y.S.2d 739 (1997), cert. denied sub nom., Ayala v. New York, --- U.S. ----, 118 S. Ct. 574, 139 L. Ed. 2d 413 (1997). O'Dell cites disagreement among jurists as hard evidence of a rule's "newness" for Teague purposes. --- U.S. at ---- - ----, 117 S. Ct. at 1974-75. This disagreement refers, however, to discord that was in existence at the time that the petitioners' convictions became final, not disagreement that arose subsequent to the petitioners' collateral appeals. Id. at ----, 117 S.Ct. at 1975.

In conclusion, because the holding that I would reach in the three cases before us on rehearing is not "new" for Teague purposes, Waller may be applied to vacate all three petitioners' convictions. See Teague, 489 U.S. at 310, 109 S. Ct. at 1075; Penry, 492 U.S. at 329, 109 S.Ct. at 2952. Therefore, I dissent in the judgment affirming these petitioners' convictions. I would vacate their convictions and remand the case for retrial within a reasonable period.