In this case we consider whether the district court properly granted a Petition for Writ of Habeas Corpus filed by Calvin Jerold Burdine based on state habeas court findings that Burdine's court-appointed attorney slept repeatedly throughout the guilt-innocence phase of his 1984 capital murder trial. The district court concluded "sleeping counsel is equivalent to no counsel at all" and granted relief pursuant to 28 U.S.C. § 2254. A divided panel of this Court reversed, holding that (1) the district court's presumption of prejudice for purposes of ineffective assistance constituted a new rule of law from which Burdine could not benefit under Teague's nonretroactivity doctrine, and (2) the circumstances of Burdine's representation did not require a presumption of prejudice to ensure the fairness of Burdine's capital murder trial. See Burdine v. Johnson, 231 F.3d 950 (5th Cir. 2000).

As an en banc court, we AFFIRM the judgment of the district court.1 The Supreme Court has long recognized that "a trial is unfair if the accused is denied counsel at a critical stage of his trial." United States v. Cronic, 466 U.S. 648, 659 (1984). When a state court finds on the basis of credible evidence that defense counsel repeatedly slept as evidence was being introduced against a defendant, that defendant has been denied counsel at a critical stage of his trial. In such circumstances, the Supreme Court's Sixth Amendment jurisprudence compels the presumption that counsel's unconsciousness prejudiced the defendant.

In January 1984, after a trial that included 12 hours and 51 minutes of total time before the jury over a period of six days, a Harris County, Texas jury convicted petitioner Burdine of capital murder in connection with the death of W.T. "Dub" Wise. Wise was killed in April 1983 during the course of a robbery committed by Douglas McCreight and Burdine. After the jury affirmatively answered the two special issues, the state trial court assessed punishment as death by legal injection in accordance with Texas law. See Tex. Penal Code Ann. § 19.03(a)(2). The Texas Court of Criminal Appeals affirmed Burdine's conviction and sentence on direct appeal. See Burdine v. Texas, 719 S.W.2d 309 (Tex. Crim. App. 1986). Throughout his trial and direct appeal, Burdine's court-appointed counsel was Joe F. Cannon of Houston.

Burdine's initial state application for a writ of habeas corpus was denied on June 29, 1994. Burdine filed a second application in December 1994. In relation to that application, the state habeas court conducted an evidentiary hearing during which Burdine called eight witnesses, including three jurors from the capital murder trial and the clerk of the court in which the trial was held. These four neutral witnesses, which the state habeas court found highly credible, testified that Cannon repeatedly dozed or slept as the State questioned witnesses and presented evidence supporting its case against Burdine.

Daniel Strickland, the foreman of the jury, recalled seeing Cannon doze or nod off between two and five times while the prosecuting attorney questioned witnesses. Myra Davis remembered being struck by the spectacle of Cannon's sleeping on the second day of trial, the same day that trial judge Joseph Guarino had chastised her for tardiness. According to Davis, Cannon "would nod his head down on his chest" with his eyes closed during the questioning of witnesses. "I was thinking to myself, you know look at him and [Judge Guarino] calls me out [for tardiness] in front of all these people, . . . and look at what that man is doing." Like Davis, Craig Engelhardt related that Cannon "would nod his head down, bob it, with eyes closed during all this." Engelhardt recalled Cannon sleeping as many as ten times during the trial, at one point for "a good probably at least 10 minutes" as the prosecution questioned a witness.

The testimony of Rose Berry, the deputy clerk assigned to the trial court that conducted Burdine's trial, confirmed the jurors' recollections. Berry recalled "lots of incidents" of Cannon sleeping during the trial. Though Berry could not specify a proportion of the trial in which Cannon slept, she did "know that he fell asleep and that he was asleep for long periods of time during the questioning of witnesses." According to the state habeas court, Berry was "the most compelling witness" in the proceeding not only because of her neutrality, but also because she was not required to pay attention to witnesses or the prosecutor and thus had a better opportunity to observe Cannon's conduct.

Other witnesses at the hearing, including Judge Joseph Guarino, prosecutor Ned Morris, and Carolyn Bonnin, a juror, testified that they had not noticed Cannon asleep during the trial. The prosecutor's testimony was challenged by James Pillow, the court coordinator of the trial court at the time of Burdine's trial. Pillow recalled having a conversation with the prosecutor, in which the prosecutor questioned Cannon's competency to represent capital defendants and suggested that Cannon not be appointed counsel in future capital cases. Neither the prosecutor nor Judge Guarino recalled ever discussing this issue, but Pillow noted that Cannon was not appointed by Judge Guarino to represent capital defendants after Burdine's trial. Cannon himself testified he had a "habit" of closing his eyes and tilting his head forward while concentrating, but that he never slept during Burdine's trial. The state habeas court pointed out the inconsistency between Cannon's testimony and the descriptions of the four neutral witnesses that saw Cannon's head bobbing. Moreover, Cannon's testimony as to his concentration habit was impeached by Philip Scardino, an attorney who worked with Cannon on a different capital case. While Scardino did not recall Cannon concentrating with his eyes closed, he did observe Cannon dozing during the voir dire of witnesses.

On April 3, 1995, the state habeas court entered comprehensive findings of fact and conclusions of law. After detailing the evidence presented during the evidentiary hearing, the court entered "a finding that defense counsel dozed and actually fell asleep during portions of [Burdine's] trial on the merits, in particular the guilt-innocence phase when the State's solo prosecutor, was questioning witnesses and presenting evidence." Based on evidence that "defense counsel repeatedly dozed and/or actually slept during substantial portions of [Burdine's] capital murder trial so that defense counsel was, in effect, absent[,]" the habeas court concluded that a showing of prejudice in accordance with Strickland v. Washington, 466 U.S. 668, 104 S.Ct 2052 (1984),was not required.2 Accordingly, the court recommended that habeas relief be granted on Burdine's claim of ineffective assistance of counsel. In a one-page, unsigned opinion, the Texas Court of Criminal Appeals agreed that "the trial court's findings of fact [regarding the sleeping of trial counsel] are supported by the record." The court nevertheless concluded that Burdine was not entitled to relief because "he failed to discharge his burden of proof under Strickland v. Washington, 446 [sic; 466] U.S. 669 (1984)." Ex Parte Burdine, Writ No. 16,725-06 (Tex. Crim. App. April 6, 1995).

Burdine then filed an application for a writ of habeas corpus in the federal district court for the Southern District of Texas pursuant to 28 U.S.C. § 2254. That court determined, on the basis of the factual findings made by the state habeas court and accepted by the Court of Criminal Appeals, that Cannon's unconsciousness during Burdine's capital murder trial amounted to constructive denial of counsel for substantial periods of that trial. See Burdine v. Johnson, 66 F.Supp.2d 854, 866 (S.D. Tex. 1999). Consequently, the district court determined that prejudice should be presumed in accordance with the Supreme Court's analysis inStrickland and United States v. Cronic, 466 U.S. 648, 104 S.Ct. 2039 (1984). See id. The State now appeals from this determination.

This federal habeas proceeding turns on the effect of state court findings that counsel repeatedly slept "during portions of [Burdine's] trial on the merits, in particular during the guilt-innocence phase when the State's solo prosecutor was questioning witnesses and presenting evidence." Although the Texas Court of Criminal Appeals rejected Burdine's habeas application, it found that the record supported the habeas court's findings of fact. In this appeal, the State concedes that we are bound by the habeas court's findings of fact. Specifically, the State "does not dispute that [counsel] dozed and actually fell asleep intermittently during Burdine's capital murder trial." The State maintains that habeas relief is nevertheless inappropriate for two reasons: (1) the district court's presumption of prejudice on the facts of this case amounts to a "new rule" that Teague v. Lane, 489 U.S. 288, 109 S.Ct. 1060 (1989), bars Burdine from raising in this collateral proceeding, and (2) the facts of Burdine's case do not warrant a presumption of prejudice because Burdine's counsel slept during indeterminate periods of what otherwise amounted to an adversarial trial.

The State's arguments fail to address the fundamental unfairness in Burdine's capital murder trial created by the consistent unconsciousness of his counsel. It is well established that a defendant "requires the guiding hand of counsel at every step in the proceedings against him." Powell v. Alabama, 287 U.S. 45, 69, 53 S.Ct. 55, 64 (1932). Moreover, both the Supreme Court and this Court have recognized that the absence of counsel at critical stages of a defendant's trial undermines the fairness of the proceeding and therefore requires a presumption that the defendant was prejudiced by such deficiency. See United States v. Cronic, 466 U.S. 648, 659 (1984); United States v. Russell, 205 F.3d 768, 770-71 (5th Cir. 2000). Applying this longstanding principle, we conclude that a defendant's Sixth Amendment right to counsel is violated when that defendant's counsel is repeatedly unconscious through not insubstantial portions of the defendant's capital murder trial. Under such circumstances, Cronic requires that we presume that the Sixth Amendment violation prejudiced the defendant.

A. Burdine Does Not Seek the Benefit of a "New Rule"

The State first argues that Burdine's claim creates a new rule of law barred by the non-retroactivity principle of Teague. When applying Teague to determine whether Burdine is eligible to habeas relief we follow three steps: (1) we determine when Burdine's conviction and sentence became final, (2) we "survey the legal landscape as it then existed to determine whether a state court considering [Burdine]'s claim at the time his conviction became final would have felt compelled by existing precedent to conclude that the rule he seeks was required by the Constitution[,]" and (3) if Burdine seeks the benefit of a new rule, we must decide whether the rule falls within one of the narrow exceptions to the non-retroactivity principle. Caspari v. Bohlen, 510 U.S. 383, 390, 114 S.Ct. 948, 953 (1994); see also Fisher v. Texas, 169 F.3d 295, 305 (5th Cir. 1999). Whether Teague's non-retroactivity rule precludes Burdine from benefitting from the presumption of prejudice he asserts is a question of law, we therefore engage in this three-part analysis de novo. See United States v. Shunk, 113 F.3d 31, 34 (5th Cir. 1997).

All parties agree that Burdine's conviction became final in 1987, when the Supreme Court denied certiorari. See Caspari, 510 U.S. at 390, 114 S.Ct. at 953. Thus, we begin our analysis with the second issue: whether a state court in 1987 would have felt compelled by Supreme Court precedent to conclude that the Sixth Amendment required a presumption of prejudice when a defendant's counsel slept repeatedly during the defendant's capital murder trial as evidence was being presented by the State. Because application of Sixth Amendment principles firmly established by 1987 compel such a presumption of prejudice on the facts of this case, we conclude that the rule Burdine seeks to benefit from is not new, and hence not barred by Teague.

In Teague, a plurality of the Supreme Court espoused Justice Harlan's view of retroactivity that a new rule of law would not be applied on collateral review to cases that became final prior to the announcement of the new rule. Teague, 489 U.S. at 310, 109 S.Ct. at 1075. Since Teague, the Court has clarified that this principle of non-retroactivity "serves to ensure that gradual developments in the law over which reasonable jurists may disagree are not later used to upset the finality of state convictions valid when entered." Sawyer v. Smith, 110 S.Ct. 2822, 2828 (1990). Moreover, the rule reflects the limited purpose of federal habeas corpus "to ensure that state convictions comply with the federal law in existence at the time the conviction became final, and not provide a mechanism for the continuing reexamination of final judgments based upon later emerging legal doctrine." Id.

Applying the non-retroactivity principle in a way that balances the need for finality of convictions against the need to enforce minimal constitutional protections has proven challenging. Even in Teague, the plurality opinion acknowledged that the task of determining whether a case announces a new rule is often difficult. For that reason, the plurality expressly did not "attempt to define the spectrum of what may or may not constitute a new rule" for purposes of retroactivity. Teague, 489 U.S. at 301, 109 S.Ct. at 1070; see also Mackey v. United States, 401 U.S. 667, 695 (Harlan, J., concurring in judgments and dissenting in part) (recognizing "the inevitable difficulties that will arise in attempting to determine whether a particular decision has really announced a "new" rule at all or whether it has simply applied a well-established constitutional principle to govern a case which is closely analogous to those which have been previously considered in the prior case law."). The Teague plurality did, however, offer the following general guidelines:

"[A] case announces a new rule when it breaks new ground or imposes a heretofore new obligation on the States or the Federal Government. To put it differently, a case announces a new rule if the result was not dictated by precedent existing at the time the petitioner's conviction became final."

Id. (emphasis in original). A majority of the Court employed these guidelines shortly after Teague in Penry v. Lynaugh, 492 U.S. 316, 109 S.Ct. 2934 (1989). Indeed, only in Penry did a majority of the Court adopt Teague's non-retroactivity doctrine. See id. Thus, the analysis in Penry is instructive with respect to the distinction between a new rule under Teague and an application of established principles to a case that is analogous to prior precedent.

Penry claimed that his Eighth Amendment rights were violated because the jury in his capital murder trial was unable to fully consider and give effect to mitigating evidence of his mental retardation and childhood abuse when answering Texas' three statutory special issues at sentencing. Penry did not facially challenge the Texas death penalty statute. Instead, Penry claimed that, "on the facts of [his] case, the jury was unable to fully consider and give effect to the mitigating evidence . . . in answering the three special issues." 492 U.S. at 315, 109 S.Ct. at 2945 (emphasis added). The State argued that Penry's asserted rule amounted to an extension of established principles and consequently was barred by Teague. The Court was thus faced with determining whether the rule asserted by Penry was an application of established principles or a new rule of law.

Prior to 1986, when Penry's conviction and sentence became final, the Supreme Court had established that "in capital cases the fundamental respect for humanity underlying the Eighth Amendment requires consideration of the character and record of the individual offender and the circumstances of the particular offense as a constitutionally indispensable part of the process of inflicting the penalty of death." See Penry, 492 U.S. at 316, 109 S.Ct. at 2945 (quoting Woodson v. North Carolina, 428 U.S. 280, 304, 96 S.Ct. 2978, 2991 (1976)). The Court had stressed this fundamental principle in upholding Texas' capital punishment statute against a facial Eighth Amendment challenge in Jurek v. Texas, 428 U.S. 262, 96 S.Ct. 2950 (1976). In Jurek, the Court concluded that Texas' sentencing scheme satisfied the Eighth Amendment provided that sentencing juries were allowed to consider any mitigating circumstances relevant to a specific case. Jurek, 428 U.S. at 272, 96 S.Ct. at 2956. Supreme Court decisions subsequent to Jurek and Woodson, but prior to Penry's conviction becoming final, reaffirmed the need for an individualized assessment of the appropriateness of the death penalty under the Eighth Amendment. See Lockett v. Ohio, 438 U.S. 586, 98 S.Ct. 2954 (1978); Eddings v. Oklahoma, 455 U.S. 104, 102 S.Ct. 869 (1982) (both concluding that a sentencer cannot be precluded from considering and giving effect to relevant mitigating circumstances when determining the appropriateness of the death penalty in a particular case). At the same time, however, no Supreme Court decision prior to Penry had commanded courts to instruct juries how to consider specific mitigating evidence in a particular case. Instead, the case law established fundamental Eighth Amendment principles for application in analogous cases.

The Supreme Court concluded that despite the absence of a specific holding requiring the instruction sought by Penry, the rule Penry sought to benefit from was dictated by the Eighth Amendment principles espoused and enforced in the Court's prior cases. According to the Court, it was firmly established at the time of Penry's conviction that a sentencer in Texas had to consider any mitigating evidence specific to the circumstances of Penry and his crime. Penry, 492 U.S. at 317, 109 S.Ct at 2946. The rule that Penry sought - a requirement that the jury be instructed specifically what mitigating evidence it should consider and how it should consider that evidence when answering Texas' special issues - was not "new" for the purposes of Teague because it represented a specific application of general Eighth Amendment principles outlined in prior analogous cases. Though Penry's claim for relief did require the State to issue specific instructions that it previously had not issued, the rule which controlled Penry's case was nevertheless not new. Moreover, the rule did not impose a new obligation on Texas, it simply required that Texas fulfill its obligation, expressed in Jurek, to ensure "that the special issues [of the Texas capital punishment statute] would be interpreted broadly enough to permit the sentencer to consider all of the relevant mitigating evidence a defendant might present in imposing sentence." Id.

Penry's recognition that the application of established general procedural principles in an analogous context is not a new rule barred by Teague remains the law today. See Bousley v. United States, 523 U.S. 614, 620, 118 S.Ct. 1604, 1610 (1998) (rejecting the argument that the petitioner's claim that his guilty plea was not knowing and intelligent was barred by Teague in part because "[t]here is surely nothing new about this principle . . . ."); see also Wright v. West, 505 U.S. 277, 304, 112 S.Ct. 2482, 2497 (1992) (O'Connor, J., concurring) ("If a proffered factual distinction between the case under consideration and pre-existing precedent does not change the force with which the precedent's underlying principle applies, the distinction is not meaningful, and any deviation from precedent is not reasonable."). Just as Penry sought an application of Eighth Amendment principles well-established at the time his conviction became final, Burdine now seeks the application of Sixth Amendment principles that were well-established at the time that his conviction became final. Just asTeague did not prevent Penry from receiving the benefit of established Eighth Amendment protections, it does not prevent Burdine from receiving the benefit of established Sixth Amendment protections.

At the time that Burdine's conviction became final in 1987, it was well established in the legal landscape that defendants have the Sixth Amendment right to effective assistance of counsel at every critical stage of the proceedings against them. See Powell,287 U.S. at 69, 53 S.Ct. at 64.3 The purpose of this Sixth Amendment guarantee was and "is to ensure that a defendant has the assistance necessary to justify reliance on the outcome of the proceeding." Strickland v. Washington, 466 U.S. 668, 689, 104 S.Ct. 2052, 2067 (1984); Cronic, 466 U.S. at 658, 104 S.Ct at 2046 ("[T]he right to the effective assistance of counsel is recognized not for its own sake, but because of the effect it has on the ability of the accused to receive a fair trial"). Because the Sixth Amendment serves solely to ensure a fair and reliable trial, "any deficiencies in counsel's performance must be prejudicial to the defense in order to constitute ineffective assistance under the Constitution. Id. In Cronic, however, the Court recognized that some egregious circumstances "are so likely to prejudice the accused that the cost of litigating their effect in a particular trial is unjustified." Cronic, 466 U.S. at 658, 104 S.Ct at 2046. Both in Cronic and in Strickland, the Supreme Court recognized that the absence or denial of counsel at a critical stage of a criminal proceeding represents one of the egregious circumstances that requires the presumption of prejudice. See Cronic, 466 U.S. at 659, 104 S.Ct. 2047; Strickland, 466 U.S. at 692, 104 S.Ct. at 2067.4 Burdine seeks an application of this rule to the facts of his case. He argues that he was repeatedly without counsel throughout the most critical part of his capital murder trial: the guilt-innocence phase. Because he was without counsel, Burdine argues that we should presume prejudice in accordance with Cronic and Strickland. We agree with Burdine that the rule he seeks to benefit from is neither new, nor should it have been surprising to the State of Texas at the time of Burdine's conviction in 1987.

The State concedes that Cronic calls for the presumption of prejudice when, during a critical stage of trial, counsel is either (1) totally absent, or (2) present but prevented from providing effective assistance. See Cronic 466 U.S. at 659 n.25, 104 S.Ct. 2047 n.25. The State argues that applying this rule to the facts of Burdine's case expands Cronic's holding and therefore creates a "new rule" barred by Teague. Specifically, the State maintains that (1) Cronic calls for a presumption of prejudice relating to absence of counsel only when state action causes such absence, and (2) any absence by Burdine's attorney was not proven to have taken place during a "critical stage" of Burdine's trial, as such term was understood by the Court in Cronic. We disagree with the State's excessively narrow reading of Cronic.

Initially, we note that the State's proposed state action requirement does not flow from the language of Cronic. Cronicrecognized that because our system of justice deems essential the assistance of counsel, "a trial is unfair if the accused is denied counsel at a critical stage of his trial." Id. In a footnote following this sentence, the Court explained that presumption of prejudice was appropriate "when counsel was either totally absent, or prevented from assisting the accused during a critical stage of the proceeding." Though the term "prevented from assisting the accused" suggests the existence of some indeterminate external force, no inference of a state action requirement is possible from the Court's language discussing the appropriateness of a presumption when counsel is "totally absent."5 Later in Cronic the Court more directly dispelled the State's proposed state action requirement when it dismissed the idea that the cause of a Sixth Amendment deficiency should control whether a presumption of prejudice was warranted. The Court explained:

"The fact that the accused can attribute a deficiency in his representation to a source external to trial counsel does not make it any more or less likely that he received the type of trial envisioned by the Sixth Amendment, nor does it justify reversal of his conviction absent an actual effect on the trial process or the likelihood of such an effect."

Cronic, 466 U.S. at 662 n.31, 104 S.Ct. at 2048 n.31. We conclude that the Sixth Amendment principle animating Cronic's presumption of prejudice is the fundamental idea that a defendant must have the actual assistance of counsel at every critical stage of a criminal proceeding for the court's reliance on the fairness of that proceeding to be justified. The Court in Cronic was not concerned with the cause of counsel's absence, but rather the effect of such absence on the fairness of the criminal proceeding.

Our recent discussion of Cronic in United States v. Russell, 205 F.3d 768 (5th Cir. 2000), supports this interpretation and rejects the State's excessively narrow reading. On appeal from the denial of his section 2255 motion, Russell urged this Court to presume prejudice under Cronic based on the absence of counsel. Russell, along with 16 co-defendants, was on trial for conspiracy to possess drugs and conspiracy to launder money. Several days into the trial, Russell's counsel fell ill and was absent for two days of trial as evidence was being presented against Russell's co-conspirators. Though an attorney for one of Russell's codefendants represented to the trial court that he had Russell's permission to act as counsel during the absence of Russell's own attorney, we concluded that it was unclear from the record whether the district court had accepted counsel's attempt to represent the petitioner. Russell, 205 F.3d at 769-71. Despite the absence of a clear waiver of counsel, the trial court allowed the trial to continue after instructing the government not to present evidence directly relating to Russell while his counsel was absent.

While we refused to adopt a per se rule that the taking of any evidence at trial in the absence of counsel warrants a presumption of prejudice, we did recognize Cronic's emphasis on the need to have counsel at every critical stage of a trial to ensure its fairness and reliability. Id. (quoting Cronic, 466 U.S. at 658, 104 S.Ct. at 2047). In characterizing Cronic's holding, we did not require a showing that the state was responsible for the absence of counsel. Instead, we interpreted Cronic as focusing on the overall fairness of the proceeding, and specifically on whether the absence of counsel was at a critical stage of the trial. We noted that while Cronic did not provide substantial guidance with respect to what parts of a trial are "critical," the following guidelines could be distilled:

First, there must be a denial of such significance that it makes the adversary process itself unreliable. [Cronic, 466 U.S. at 659, 104 S.Ct. at 2047]. Second, the Cronic court makes clear that "only when surrounding circumstances justify a presumption of ineffectiveness can a Sixth Amendment claim be sufficient without inquiry into counsel's actual performance at trial."

205 F.3d at 771 (quoting Cronic, 466 U.S. at 662, 104 S.Ct. at 2048). Applying these principles to Russell's case, we concluded that the adversary process in Russell's trial was unreliable because Russell's counsel was not "present to keep the taint of conspiracy from spreading to the client." Id. at 772. On this basis, we held that counsel's absence was at a critical stage and presumed prejudice.6

As reflected by our discussion in Russell, Cronic presumes prejudice based on the absence of counsel when such absence threatens the overall fairness of a trial. While state responsibility for counsel's absence may be relevant in examining the fairness of a trial, state action is not and has never been a prerequisite for invoking Cronic to presume prejudice. ReadingCronic to impose such a prerequisite would require shifting the opinion's emphasis from the fairness and reliability of criminal proceedings to the culpability of a state in distorting the adversarial process. For that reason, we reject the State's attempt to cast Burdine's argument as a "new rule" by imposing a state action requirement onto Cronic's principle that Sixth Amendment prejudice is presumed when a defendant demonstrates the absence of counsel at a critical stage of his criminal proceeding.

The State also attempts to characterize Burdine's argument as a new rule by limiting the meaning of "critical stage" as that term is used in Cronic and certain cases cited in Cronic. Initially, the State argues that the Supreme Court intended "the Sixth Amendment concept of 'critical stage'" to refer "not to the trial itself, but rather to phases of a criminal proceeding other thanthe trial." (emphasis in original). We quickly dispense with this argument. All of the Supreme Court cases that the State cites as supporting its proposition assume that the presentation of evidence against a defendant is a critical stage of a criminal proceeding. See, e.g. Brewer v. Williams, 430 U.S. 387, 398, 97 S.Ct. 1232, 1239 (1977). The cases cited by the State simply extend the concept of the trial as a critical stage to other discrete periods; in so extending the concept, they do not question the fact that the trial itself remains a critical stage of any criminal proceeding. Maine v. Moulton, 474 U.S. 169, 170, 106 S.Ct. 477, 484 (1985) (noting that the Supreme Court has not limited the right to assistance of counsel to participation at trial, but has not abrogated that right during the trial itself).

The State next argues that because Burdine cannot demonstrate precisely when Cannon slept during his trial, he cannot prove that Cannon slept during critical stages of his criminal proceeding. In this regard, the State asks more of Burdine than the Supreme Court or this Court has ever asked of a defendant attempting to show the absence of counsel during a critical stage of trial. To justify a particular stage as "critical," the Court has not required the defendant to explain how having counsel would have altered the outcome of his specific case. Rather, the Court has looked to whether "the substantial rights of a defendant may be affected" during that type of proceeding. United States v. Taylor, 933 F.2d 307, 312 (5th Cir. 1991) (citing Mempha v. Rhay, 389 U.S. 128, 134, 88 S.Ct. 254, 256 (1967); Gideon v. Wainright, 372 U.S. 335, 342-43, 88 S.Ct. 792, 795-96 (1963)); see also United States v. Gouveia, 467 U.S. 180, 189, 104 S.Ct. 2292, 2298 (1984) (suggesting that a proceeding is critical when the accused is confronted by the legal procedural system or the expertise of a state adversary). Thus, in Russell, this Court was satisfied by Russell's showing that evidence was being adduced by the State against his co-conspirators while Russell's counsel was absent. See Russell, 205 F.3d at 772. We did not require Russell to demonstrate that the evidence adduced against his co-defendants did in fact have an adverse impact on his own fortune or that the presence of his attorney would have improved his chances of an acquittal.7 Such a standard would require that the defendant, in effect, prove prejudice in order to receive a presumption of prejudice. That was not the standard announced in Cronic. Therefore, to the extent that the State maintains that application of the term "critical stage" to the facts of Burdine's case would constitute a new rule, we dismiss the State's argument. Burdine has alleged and the state court findings support the fact that Burdine's counsel was unconscious, and hence absent, repeatedly throughout the guilt-innocence phase of Burdine's trial as evidence was being produced against Burdine. That this stage of Burdine's trial was "critical" was well established in 1987 and is well established today.

In sum, we conclude that Burdine seeks the benefit of a rule well-established at the time that his conviction became final: when a defendant does not have counsel at every critical stage of a criminal proceeding, the court must presume that such egregious deficiency prejudiced the fairness of the trial. Because Burdine does not seek the benefit of a new rule, we need not discuss the various exceptions to Teague. Instead, we turn to whether the merits of Burdine's case warrant the application of this longstanding rule.

B. Is Presumption of Prejudice Appropriate in Burdine's Case?

The State purports to accept the state trial court's findings that defense counsel slept during substantial portions of Burdine's trial. Nonetheless, the State painstakingly conducts a page-by-page analysis of the trial record in an apparent attempt to demonstrate that counsel was awake during significant portions of the trial.8 Yet, once we have accepted as presumptively correct the state court's finding that counsel slept "during portions of [Burdine's] trial on the merits, in particular during the guilt-innocence phase when the State's solo prosecutor was questioning witnesses and presenting evidence," there is no need to attempt to further scrutinize the record. See Javor v. United States, 724 F.2d 831, 834 (9th Cir. 1984) (holding that "[w]hen a defendant's attorney is asleep during a substantial portion of his trial, the defendant has not received the legal assistance necessary to defend his interests at trial" and thus, prejudice must be presumed).

The factual findings made during Burdine's state habeas proceedings demonstrate that Burdine's counsel was repeatedly asleep, and hence unconscious, as witnesses adverse to Burdine were examined and other evidence against Burdine was introduced. This unconsciousness extended through a not insubstantial portion of the 12 hour and 51 minute trial. Unconscious counsel equates to no counsel at all. Unconscious counsel does not analyze, object, listen or in any way exercise judgment on behalf of a client. As recognized by the Second Circuit, "the buried assumption in ourStrickland cases is that counsel is present and conscious to exercise judgment, calculation and instinct, for better or worse. But that is an assumption we cannot make when counsel is unconscious at critical times." Tippins v. Walker, 77 F.3d 682, 687 (2d Cir. 1996).9 When we have no basis for assuming that counsel exercised judgment on behalf of his client during critical stages of trial, we have insufficient basis for trusting the fairness of that trial and consequently must presume prejudice.

The State suggests that because Cannon was physically present in the courtroom, his dozing constituted a form of performance that should be subjected to prejudice analysis. The State maintains that it is impossible to distinguish between sleeping counsel and other impairments that nevertheless have been subjected to prejudice analysis. We disagree. An unconscious attorney does not, indeed cannot, perform at all. This fact distinguishes the sleeping lawyer from the drunk or drugged one. Even the intoxicated attorney exercises judgment, though perhaps impaired, on behalf of his client at all times during a trial. Yet, the attorney that is unconscious during critical stages of a trial is simply not capable of exercising judgment. The unconscious attorney is in fact no different from an attorney that is physically absent from trial since both are equally unable to exercise judgment on behalf of their clients. Such absence of counsel at a critical stage of a proceeding makes the adversary process unreliable, and thus a presumption of prejudice is warranted pursuant to Cronic.

As in Russell, we decline to adopt a per se rule that any dozing by defense counsel during trial merits a presumption of prejudice. Our holding, that the repeated unconsciousness of Burdine's counsel through not insubstantial portions of the critical guilt-innocence phase of Burdine's capital murder trial warrants a presumption of prejudice, is limited to the egregious

facts found by the state habeas court in this case.10

Based on the state court's findings that have been accepted by all as presumptively correct, we affirm the district court's grant of federal habeas corpus relief and vacate Burdine's capital murder conviction. The State is free to retry Burdine for capital murder.11

AFFIRMED.

PATRICK E. HIGGINBOTHAM, Circuit Judge, joined by Chief Judge CAROLYN DINEEN KING, W. EUGENE DAVIS, and JACQUES L. WIENER, Circuit Judges, concurring:

I concur fully in the excellent opinion of the Court, but write separately to explain my preferred path. As for the merits of Burdine's claim, he had no lawyer for not insignificant amounts of time as the government presented its case. This is surely a denial of the constitutional right to counsel. The more difficult issue is the threshold Teagueaquestion of whether this court has the power to grant the relief Burdine requests on habeas review.

* Teague v. Lane1 announced that a federal court reviewing a habeas petition cannot apply a "new rule" of law in granting relief to the prisoner.2 In other words, a federal habeas petitioner cannot rely on a rule of federal constitutional law that did not exist at the time his conviction became final. The rule of Teagueis no legal technicality. Rather, it is a recognition of the important, but sensitive, role of habeas corpus in our federal system. Teague is an integral component of the structure by which constitutional questions arising out of state criminal convictions are reviewed by federal courts.

Originating in the Habeas Corpus Act of 1867, in tandem with the Reconstruction Amendments, the modern writ of habeas corpus has operated as a vital safeguard of the federal constitutional rights of persons convicted in state courts.3 It has a history bound up in the expansion of federal supervision over the States and the genesis of modern civil rights, and in particular the movement toward racial equality. Habeas corpus originally served only to ensure the release of persons imprisoned without legal process, but its reach expanded slowly until the landmark decision of Brown v. Allen.4 Only then in 1953 did federal habeas afford relief from a state conviction based on constitutional error.

In practical effect, Brown replaced direct review in the Supreme Court of state convictions by enlisting the lower federal courts in the task of reviewing claims of constitutional deprivation ensuing from state criminal convictions. In the same stroke, Brown expanded the availability to state prisoners of a remedy for constitutional violations suffered during their prosecution. This expansion of the writ, however, threatened the finality of criminal convictions. Unlike appellate review, habeas review was not bound by time limits,5 and changes in the law could render convictions that were valid when decided invalid under current law. In such cases, no longer would the Reconstruction-era concern with States flouting federal law be relevant--instead, state-court convictions that complied with federal law could be challenged years later based on developments in federal law that the state courts could not have anticipated.

Following Brown, the Supreme Court struggled with the problem of habeas courts applying recent federal decisions to old convictions. It first attempted to give only prospective effect to its newly found constitutional rights,6 an effort that drew sharp criticism.7 Urged on by Justice Harlan, the Court finally abandoned the effort at prospective effect8 and turned to the scope of the habeas remedy in Teague.9 By forbidding a federal court from applying legal rules that did not exist at the time the prisoner's conviction became final, Teague allows federal courts to provide review of state-court convictions, in a fashion akin to appellate review, without interfering with convictions in state courts that acted in compliance with federal law. So it is thatTeague is a rich and powerful discipline for the wielding of federal power. More to the point, it is not an equitable doctrine relaxing or drawing taut for cases perceived as deserving of adjustment--in either direction. At the least the doctrine has matured to this relative fixity.

The Teagueaprinciple, though easy to state, can pose difficult questions in its application. The majority opinion relies primarily on Penry v. Lynaugh,10 properly so. But the trail of cases since Penryahas left it near the outer limits of the Court's willingness to conclude that a proposed rule is not a new rule.11 For me, the import of Penry is better understood in light of Sawyer v. Butler,12 a case more representative of the Teagueajurisprudence since Penry. In Sawyer, this court en banc addressed the claim of a habeas petitioner that the prosecutor at his trial violated the rule announced in Caldwell v. Mississippi,13 which had been decided by the Supreme Court after Sawyer's conviction had become final. Caldwell had held that a prosecutor's statements to the sentencing jury in a capital case that diminish the jury's sense of responsibility in its sentencing role require reversal of the sentence of death.14 We ruled that Caldwell announced a new rule, one that did not exist at the time Sawyer's conviction became final.15 Thus, our court was barred from applying Caldwell to Sawyer's case.

In concluding that Caldwellaannounced a new rule, we pointed to two salient facts: First, Caldwell was the first case to conclude that the Eighth Amendment provided the basis for overturning a death sentence because of prosecutorial statements.16 Second, Caldwell eliminated the requirement present in older due process cases that the defendant must show that the prosecutor's statements rendered the trial fundamentally unfair.17 Instead,Caldwell created a presumption of fundamental unfairness that did not exist before. Thus, Caldwell lowered the threshold for finding reversible error.18 Sawyerastands for the crucial principle that a rule is a new rule when its articulation changes the elements or the burdens of proof a prisoner must satisfy to prove a constitutional violation.

Sawyer also explained Penry. We observed that, at base,Penryawas not about whether a new rule or an old rule applied; instead, "Penry involved the consistent application of an established constitutional rule to, in essence, changes in the facts."19 This distinction between cases presenting new rules and cases presenting new facts is central to the functioning of Teague. State courts cannot and need not divine the future of federal constitutional law; they need only follow the rules extant at the time of the defendant's conviction and appeal. Thus, Teagueprohibits federal courts from judging state-court convictions against standards developed after those convictions became final. On the other hand, every court--state or federal--has the duty to faithfully apply legal rules to the distinct facts of each case. Teague does not bar a federal court from reviewing the application of an old rule to new facts in state court; as we explained inSawyer, Penry said as much.20

Since Sawyer, the Supreme Court has further articulated the approach to law and fact in the Teague analysis:

Teague and our subsequent decisions interpreting it require a federal court to answer an initial question, and in some cases a second. First, it must be determined whether the decision relied upon announced a new rule. If the answer is yes and neither exception applies, the decision is not available to the petitioner. If, however, the decision did not announce a new rule, it is necessary to inquire whether granting the relief sought would create a new rule because the prior decision is applied in a novel setting, thereby extending the precedent.21

I believe this is the appropriate framework for analyzing Burdine's claim.

Our decision today relies on no new rule. Although the principal dissent takes issue with this conclusion, the only rule being applied is forty years old. The so-called Cronicarule dates back to the 1961 decision Hamilton v. Alabama.22 In Hamilton, the Supreme Court held that absence of counsel at an arraignment in Alabama was per se reversible. The Court concluded that "[a]rraignment under Alabama law is a critical stage in a criminal proceeding."23 Because arraignment was a critical stage of the proceeding and the prisoner was denied counsel at arraignment, the Court "[did] not stop to determine whether prejudice resulted," but reversed his conviction.24 Thus the Cronic rule--that (1) denial of counsel at (2) a critical stage of proceedings mandates reversal25--was established twenty-six years before Burdine's conviction became final. Most important, the Court in Hamiltonmade clear that determination of what was a "critical stage" rested upon the facts of each case. It acknowledged that arraignments in other jurisdictions may not be critical to the defense.26 What mattered to the Court was that the facts of the case justified the conclusion that counsel was denied at a critical stage.

Since Hamilton, the Supreme Court has applied the same rule to different facts. In each case, the Court did not reformulate the rule, but applied the pre-existing rule to a different set of facts. In White v. Maryland,27 the Court deemed a preliminary hearing to be a critical stage. The Court looked to the facts of the case to determine whether "rights are preserved or lost" and concluded that even if normally a preliminary hearing is not a critical stage in Maryland, in this case a guilty plea was entered.28 Although the defendant later reversed his plea, the initial plea of guilty was entered into evidence at trial.29 In other cases, the Supreme Court has found denial of counsel to require reversal even when the denial affected only a single trial decision,30 a portion of the testimony,31 or closing arguments before a judge.32

Likewise, the element of "absence of counsel" has not been sliced to a succession of fine legal rules, but an assessment of the facts and realities of the individual case. The Supreme Court has held that the physical presence of counsel does not prevent his "absence" for purposes of the Cronic rule.33 To the contrary "absence" means simply that the defendant was without counsel.

What is clear from this line of Supreme Court cases is that the applicable rule in this case is not new. Absence of counsel at a critical stage of trial renders the trial unfair and requires reversal. What is also clear is that what constitutes a "critical stage of the proceeding" and even "absence of counsel" depends on an assessment of the facts of each case. Thus, the Teague issue in this case reduces to whether, in applying an old rule to the facts of this case, the facts are so "novel" that we in effect do not apply the rule, but create a new rule with broader scope or greater bite.34

The facts of this case do not test the boundaries of theCronicarule. That sleeping counsel is absent counsel is elementary. Burdine's slumbering counsel presents us with a new factual situation, hopefully rare, but not a difficult question of the application of law to fact.35 The novelty of this case stems not from the implausibility of applying the rule to these facts, but from the stunning image of an attorney sleeping in the courtroom while his client is on trial for his life. We are not asked to stretch to conclude that counsel was absent in every relevant sense.

So, too, there is nothing new about concluding that the facts of this case demonstrate that the taking of evidence against Burdine is a critical phase of the trial. No more recently than 1963 did the Supreme Court describe a critical stage of the proceedings as a point at which "rights are preserved or lost."36 Failure to object to the admission of evidence waives any subsequent assertion of error. A lawyer's absence during substantial portions of testimony cripples his ability to cross-examine the witnesses and impairs his ability to present the defense case and jury arguments.37

Surely the presentation of the evidence of guilt is a critical phase. Nor is it an answer that Burdine "freely and voluntarily confessed to his crime" (and hence that his lawyer slept didn't matter).38 This ignores both the record in this case and the reality that the effort to persuade a jury not to vote for death often runs, as here, throughout the guilt phase of the trial. The phrasing of the questions, their sequence and rhythm set tone and paint a picture. They become the platform for presenting the penalty case and final argument. The search for the precise evidence that came in as Burdine's counsel slept rests upon a view of trial dynamics and reality that confounds my forty years in the courtroom. With respect to my colleagues, that is not the way it works, and for the same reasons it is not the law. We presume prejudice because experience tells us that an occurrence presents both a high probability of prejudice and a difficulty of "proving it" in any finite sense. The law speaks of presumption not to supply a missing ingredient, but rather to recognize its inevitable presence.39 Right to counsel at critical stages is only an example of this principle. We simply will not put a person on trial for his life in the absence of counsel.

Indeed, a lawyer asleep in the courtroom is more harmful than one who is physically absent. A message is sent to the jury when a defense counsel sleeps, sometimes as long as ten minutes, the prosecutor continues to present evidence, the judge does nothing (says he didn't see it)--all the officers of the court pay it no mind. This is just a "slow plea": going through the motions is the message. That is what happened. It will not do to dance away from the facts. They were found by the state court and come to this court unchallenged by the State.

The principal dissent relies heavily on a notion introduced byButler v. McKellar40 that a habeas court cannot apply a rule if at the time the prisoner's conviction became final the application of the rule was "debatable among reasonable jurists."41 The dissent then cites the divided panel opinion in this case and the Texas courts' divergent opinions in Burdine's state habeas case as evidence that the application of the Cronic rule was "debatable among reasonable jurists."42 It is true that Butler implies that one can count heads to determine whether a rule is new or not.43 It is also true that the Supreme Court has since abandoned this approach in judging reasonableness.

In Stringer v. Black,44 the Supreme Court rejected the argument that the prisoner's requested rule was new because the Fifth Circuit had held to the contrary before the Supreme Court announced the rule. The Supreme Court stated, "The purpose of the new rule doctrine is to validate reasonable interpretations of existing precedents. Reasonableness, in this as in many other contexts, is an objective standard, and the ultimate decision whether [the rule] was dictated by precedent is based on an objective reading of the relevant cases."45 The Supreme Court tersely concluded that the decisions of two unanimous panels of the Fifth Circuit had been unreasonable.46

Cases since Butlerahave rejected contentions by States that federal habeas courts should further defer to state rulings of law. The suggestion in Wright v. West47 that federal courts defer to the state courts' application of law was made by a minority of the Court and expressly rejected by a majority of the Justices to consider the proposal.48 Two terms ago, a majority of the Supreme Court reiterated its conviction that a federal court must make an independent judgment in applying an old rule to a set of facts.49

We must make an independent determination of the application of Cronic to the facts of this case. I am convinced that precedent dictated the conclusion that Burdine's counsel was absent during a critical stage of the trial and that a contrary conclusion would be unreasonable, viewed objectively.

E. GRADY JOLLY, Circuit Judge, joined by JERRY E. SMITH, Circuit Judge, dissenting:

Because the record in this case makes clear that Burdine is plainly guilty of capital murder beyond a reasonable doubt; because Burdine voluntarily confessed to his crime; because, even though Burdine was fully aware that his counsel had slept at points during the trial, he repeatedly heaped post-trial compliments on his counsel for his performance at trial and continually has expressed confidence in his counsel after trial; because the record fairly establishes that Burdine's counsel actually provided competent representation throughout the course of the trial; because there is no suggestion in the record that Burdine suffered any prejudice on account of counsel's alleged sleeping, that is, there is no suggestion that the outcome in this case would have been any different on account of the allegations now made; because Burdine waited eleven years before he ever raised the "sleeping lawyer" claim; because there is no evidence in the record that shows that counsel's sleeping occurred at a critical stage in the trial, and because the now silent Burdine apparently could have offered testimony on this point but has chosen not to do so; and finally, because I am led to believe by these facts that the "sleeping lawyer" claim is in large part a diverting tactic to create the impression of a miscarriage of justice in a case in which substantial justice has been done, I respectfully dissent from the granting of habeas relief on the basis of the "sleeping lawyer" claim. I would remand for consideration of his other claims to determine whether any have such merit to afford Burdine federal habeas relief.

RHESA HAWKINS BARKSDALE, Circuit Judge, joined by EDITH H. JONES, JERRY E. SMITH, and EMILIO M. GARZA, Circuit Judges, dissenting:

"Bad facts make bad law." This is just such a case. The "bad facts" -- the deplorable sleeping by Calvin Jerold Burdine's court-appointed trial counsel, Joe Frank Cannon -- have, I fear, driven the majority to make "bad law". This is vividly demonstrated by the anomaly of the majority's stating that, for presumed-prejudice purposes, the entire guilt phase of a capital murder trial is not a "critical stage" (one of its bases for maintaining that, in granting presumed-prejudice, it has not established a "new rule" for Teague-bar purposes), while, in a special concurrence, four judges in that majority nevertheless maintain it is such a stage.1

The majority is not alone in its abhorrence at the spectacle of Cannon sleeping during a capital murder trial; but, our decision must not be influenced, much less dictated, by this. In focusing so narrowly and intently on Cannon's sleeping, the majority has lost sight of the reasons for the Sixth Amendment's requiring effective assistance of counsel in a criminal proceeding: adversarial testing of the prosecution's case and reliability of the result. Two of the key cases that shaped these contours make that plain.

The benchmark for judging any claim of ineffectiveness must be whether counsel's conduct so undermined the proper functioning of the adversarial process that the trial cannot be relied on as having produced a just result.

Strickland v. Washington, 466 U.S. 668, 686 (1984) (emphasis added).

The right to the effective assistance of counsel is ... the right of the accused to require the prosecution's case to survive the crucible of meaningful adversarial testing. When a true adversarial criminal trial has been conducted -- even if defense counsel may have made demonstrable errors -- the kind of testing envisioned by the Sixth Amendment has occurred.

....

[T]he right to the effective assistance of counsel is recognized not for its own sake, but because of the effect it has on the ability of the accused to receive a fair trial. Absent some effect of challenged conduct on the reliability of the trial process, the Sixth Amendment guarantee isgenerally not implicated.

United States v. Cronic, 466 U.S. 648, 656, 658 (1984) (emphasis added; footnote omitted). This being a capital murder case does not alter this. See Strickland, 466 U.S. at 686.

The majority only pays lip service to these factors, Maj. Op.at 15-16; it avoids applying them to this case. For example, it does not even mention Burdine's confession and Cannon's repeated efforts to keep it from the jury. Nor does it mention Burdine's testimony in which he admitted both robbing the victim and being present at his murder. The prosecution's case was more than tested; the result, more than reliable. Cannon's sleeping does not change that.

The extra-judicially, recently revealed evidence withheld by Burdine concerning his nudging Cannon during trial when he slept is very relevant to Burdine's presumed-prejudice-due-to-Cannon's-sleeping claim. (This new evidence was confirmed by Burdine's counsel at en banc oral argument.) The majority, however, does not mention it, except, in response to this dissent, summarily stating in a footnote that, in essence, Burdine's knowledge at trial about Cannon's sleeping, the resulting action Burdine took (and did not take) at trial and post-judgment, and this evidence-withholding do not matter. Maj. Op. at 28 n.10. The special concurrence does not mention the subject. But the withheld evidence colors this entire appeal; we cannot disregard it. Moreover, this twelfth-hour revelation transforms this presumed-prejudice claim into one totally different from that for which our court granted en banc review. On this basis alone, we should reverse and remand. At the very least, we should remand for the district court to develop this evidence, and its implications regarding the presumed-prejudice claim.

In short, this appeal, this presumed-prejudice claim, is just not as simple, just not as cut-and-dried, as the majority and, especially, the special concurrence would have it. Our court does not write on a clean slate; we must deal with long-established precedent designed to accommodate the strong competing interests at play when presumed-prejudice is claimed. To resolve this difficult and emotional claim, we must go back to first principles. I regret greatly that our court has not done so. I will.

Accordingly, in addition to pointing out the general overall errors in the majority's analysis (part I., 5-17), and discussing the underlying proceedings, including Cannon's efforts to keep out Burdine's confession (part III., 23-38), this dissent goes into the requisite detail to cover the sub-issues raised by Burdine's presumed-prejudice claim:

* Whether Burdine's evidence-withholding affects, if not forecloses, his claim (part II., 17-23);

* Whether, on the facts at hand (including the impossibility of determining when Cannon slept in conjunction with what was then taking place at trial), to grant presumed-prejudice is to retroactively apply a "new" rule, in violation of precedent barring such procedure (part IV.A.1., 43-73);

* Whether, even if allowing presumed-prejudice for Cannon's sleeping is a new rule, it nevertheless meets one of the exceptions to barring its retroactive application (part IV.A.2., 73-81); and

* Whether, even if allowing presumed-prejudice for Cannon's sleeping is not a new rule, Burdine, on the record at hand, satisfies the elements for that rule (part IV.B., 81-102).

Before addressing the underlying facts and usual issues involved for presumed-prejudice vel non, we must address the unique issues surrounding Burdine's knowing during trial about Cannon's sleeping, but not raising it as an issue until 11 years later, and even then withholding evidence about it. The subject bears on Burdine's presumed-prejudice claim; on the conduct of his habeas counsel, Robert Lee McGlasson, II, who elected not to present (withheld) this evidence; and on the integrity of this proceeding and this court. See part II., infra.

This aside, in reviewing a solemn state judgment, and although it denies doing so, the majority creates a "new" rule for presumed-prejudice and applies it retroactively, contrary to binding precedent. Under this new rule, the requisite prejudice for an ineffective-assistance claim is to be presumed because of the "repeated unconsciousness of [Cannon] through not insubstantial portions of the critical guilt-innocence phase of Burdine's capital murder trial". Maj. Op. at 28. But, the majority seeks to immediately wipe away the new rule it has just labored mightily to confect by holding this rule "is limited to the egregious facts found by the state habeas court". Id. Truly, this raises result-driven jurisprudence to a new level.

Under Supreme Court and our precedent, the majority's "new" rule cannot be applied retroactively to this habeas claim. But, even if the rule is not "new", it cannot be applied to this case because, in the light of the state habeas court factual findings (state-findings), and contrary to the majority's characterization of them, Cannon was not "repeatedly unconscious through not insubstantial portions of the ... trial". Id. at 8.

Overarching all of this are three actions by the majority which turn the basis for presumed-prejudice on its head. That doctrine is designed for instances of deficient attorney-performance that are so obvious and so easy to identify, and where resulting prejudice is so likely, that examination of the record for prejudice vel non is not worth the cost of doing so.

First, the majority allows presumed-prejudice, even though the claim based on such (in its words) "egregious facts" was not raised until a second state habeas application, 11 years after the trial. If these facts were so egregious, the claim would have been more than obvious to Burdine during trial and, most especially, in the light of his recent withheld-evidence admission.

Second, the majority does not just allow, it rewards, this evidence-withholding, about which our en banc court inquired, sua sponte, at oral argument. This admission is not only cause for rejecting presumed-prejudice but also, on remand, for requiring an evidentiary hearing concerning the withheld-evidence and this possibly improper tactic by Burdine's habeas counsel.

Third, contrary to the procedure established for the narrow circumstances and resulting limited instances in which a court is to award presumed-prejudice, the majority has had to examine the record, shepherd the state-findings favorable to its position, turn a blind eye to those unfavorable (including the withheld-evidence), and make unwarranted inferences about those facts in order to, with the greatest effort, shoehorn this case into its new -- momentarily lived -- rule.

This is precisely how the presumed-prejudice doctrine is not supposed to work. In short, what the majority has done with its new rule flies in the face of the principles underlying both nonretroactivity for federal habeas concerning state prisoners and implementation of presumed-prejudice.

The solemn state judgment of Burdine's guilt is not before us. Instead, we are reviewing a subsequent, equally solemn state judgment that he received the assistance of counsel necessary for a fair trial. The sole issue at hand is whether prejudice resulting from ineffective-assistance must be proved by Burdine, as is the usual case, or, instead and as is very rare, is to be presumed. Presumed-prejudice vel non is a profoundly important issue; it touches on compelling interests of finality and comity. It must be decided by applying binding precedent.

Three Supreme Court decisions, two of which were rendered on the same day and are quoted from earlier, provide the primary guidance for our review: United States v. Cronic and Strickland v. Washington, 466 U.S. 648 and 668, respectively (1984); and Teague v. Lane, 489 U.S. 288 (1989). First, proving ineffective-assistance violative of the Sixth Amendment ordinarily requires showing both that counsel rendered deficient performance, and that there is a reasonable probability that, but for that deficient performance, the result of the proceeding would have been different. Strickland, 466 U.S. at 687, 694. Second, in certain narrow circumstances (including denial of counsel at a "critical stage" of the proceeding) where prejudice is so likely that case-by-case inquiry is not worth the cost, prejudice will be presumed. Id. at 692; Cronic, 466 U.S. at 658-59. Third, "new" rules of criminal procedure will not be applied retroactively on collateral review unless certain narrow exceptions apply; "a case announces a new rule if the result was not dictated by precedent existing at the time the defendant's conviction became final". Teague, 489 U.S. at 301 (plurality) (emphasis in original). The special concurrence assists in this respect: by underscoring that implementing Teague is neither a "legal technicality", Sp. Con. at 1, nor "an equitable doctrine", id. at 3.

We are not a state habeas court; we cannot make factual findings. The key binding/controlling state-finding is that Cannon

dozed and actually fell asleep during portionsof [Burdine's] trial on the merits, in particular during the guilt-innocence phase when the State's solo prosecutor[] was questioning witnesses and presenting evidence.

Ex parte Burdine, No. 379,444-B, at 13 (183d Dist. Ct. Harris County, Tex., 4 April 1995) (emphasis added). Concerning presumed-prejudice, this is the only state-finding that even approaches being specific. But, of utmost importance, and contrary to the majority's rule ( again, applied only to this case), there is no state-finding that Cannon was "repeatedly unconscious" during "substantial" portions of the trial. Likewise, there are no state-findings as to:

* When Cannon "dozed" as opposed to "slept";

* How long he slept, individually and collectively;

* How many times he slept;

* How deeply he slept;

* What happened while he slept, including which witness(es) was(were) testifying or other evidence was being presented; and

* When the sleeping occurred -- which day(s), or whether during the morning or afternoon.

Moreover, the state habeas trial court did not discredit testimony by the trial judge and prosecutor that they did not observe Cannon sleeping. Because Burdine waited 11 years to raise the claim, memories have, of course, faded, making it impossible to determine what evidence was being presented while Cannon slept. To make matters worse, Burdine withheld critical evidence on this point.

In any event, the majority's rule is based on two factual premises not found by the state habeas court: (1) Cannon was "repeatedly unconscious", (2) for "not insubstantial" portions of trial. Maj. Op. at 8, 26. To overcome what should be an insurmountable obstacle for habeas review, the majority posits that, although "the state habeas court used slightly different language" in describing the sleep-episodes, "each variation reflects that Cannon slept on multiple occasions during the guilt-innocence phase of Burdine's trial". Id. at 6 n.2. But, none of the various ways in which the state habeas court described Cannon's "dozing" and/or "sleeping" justifies the majority's claim that those findings "support the fact that Burdine's counsel was unconscious, and hence absent, repeatedly throughout the guilt-innocence phase of Burdine's trial as evidence was being produced against Burdine". Id. at 25.

In a finding separate from the controlling finding quoted earlier (Cannon "dozed and actually fell asleep during portions of [Burdine's] trial"), the state habeas trial court stated it did "not discredit the testimony of [the prosecutor] and [the trial judge that they did not see Cannon sleeping] in [its] finding that [Cannon] repeatedly dozed and or actually slept at trial". Ex parte Burdine, No. 379,444-B, at 14. Regarding Cannon's inattentiveness, however, we do not know from these findings the difference between "dozing" and "sleeping". (The different forms of inattention usually will be referred to collectively as "sleep".)

Likewise, the state habeas trial court made no finding that Cannon's dozing or sleeping reached the level of "unconsciousness".2 Moreover, the testimony of the witnesses at the state habeas evidentiary hearing -- describing Cannon as "dozing", "nodding", "bobbing his head", and "asleep" -- do not support the majority's assumption that Cannon was, as a result, "repeatedly unconscious". As the Second Circuit recognized inTippins v. Walker, 77 F.3d 682, 689 (2d Cir. 1996), "consciousness and sleep form a continuum, and ... there are states of drowsiness that come over everyone from time to time during a working day, or during a trial". Instead, as discussed in note 2, supra, the majority, lacking both evidentiary and legal support, has made its own factual finding that the dozing and/or sleeping "repeatedly" reached "unconsciousness". This it cannot do.

Even assuming arguendo Cannon was "unconscious" each time he slept, the majority does not define "not insubstantial". Does it intend for substantiality to be judged by the length of sleep, or is it to be based on the significance of the evidence being presented while counsel slept and its impact on the defense? Seeid. at 685 ("The word 'substantial' ... is unhelpful. It can refer to the length of time counsel slept, or the proportion of the proceedings missed, or the significance of those proceedings."). In the light of the majority's stated refusal to adopt a per serule that the entire trial is a "critical stage", and the impossibility, on this record, of determining when Cannon slept, the majority apparently has chosen the former -- length of sleep-time. Yet there is no quantitative state-finding upon which to base the majority's conclusion that Cannon was "repeatedly unconscious through not insubstantial portions" of the trial.3

The witnesses' testimony at the state habeas evidentiary hearing was not consistent with regard to whether Cannon slept, much less how many times he did so, when, and for how long. In the light of those inconsistencies, the lack of a state-finding quantifying the frequency or length of Cannon's dozing or sleeping is quite understandable.4

Is "not insubstantial" the same as "substantial"? Of course, "substantial" has many uses in the legal context.5 In discussing whether a stage of a criminal proceeding is "critical", the majority states the Supreme Court has considered whether "the substantial rights of a defendant may be affected". Maj. Op. at 23. Black's Law Dictionary defines "substantial right" as "[a]n essential right that potentially affects the outcome of a lawsuit and is capable of legal enforcement and protection, as distinguished from a mere technical or procedural right". Black's Law Dictionary 1324 (7th ed. 1999).6 But, in holding that Cannon's being "repeatedly unconscious for not insubstantial" portions of the guilt-innocence phase of a capital murder trial warrants presuming prejudice, the majority does not attempt to determine whether the evidence presented while the sleeping occurred affected Burdine's "substantial rights". It cannot do so on this record.

Perhaps the majority views "substantiality" as a continuum, in which there is some middle ground which is neither "substantial" nor "insubstantial". In any event, the majority gives no guidance to federal habeas courts, which may well in the future consider similar claims, regarding how to determine whether sleeping is "not insubstantial", when, as in this case, there is no state-finding of substantiality (quantitative or qualitative).

Despite the majority's attempt to limit its rule solely "to the egregious facts found by the state habeas court", Maj. Op. at 28, its rule will not be applied just in this case. The majority can limit the holding to this record; but otherwise, the rule must be shaped so that it can be applied -- as it may well be -- in future cases. This rule, however, will result in uncertainty and undermine accuracy. For example, how many minutes of sleeping, or how many nods or head bobs will trigger presumed-prejudice? Moreover, allowing presumed-prejudice under these circumstances will encourage defendants not to bring observed sleeping by their counsel to the attention of the court during trial and not to raise the claim on direct appeal, which undermines the strong interest in finality recognized in Teague and its progeny. Finally, the rule imposes a new obligation on the States in our circuit, by requiring trial judges and prosecutors to closely and unceasingly monitor defense counsel throughout trial to ensure defense counsel is awake. If counsel closes his eyes even momentarily, the trial judge or prosecutor had best stop the trial and inquire, "Are you awake?" Nothing in Cronic comes close to dictating such a result.

Because, as a federal habeas appellate court, we do not engage in fact-finding, we cannot do as the majority has done and find Cannon was "repeatedly unconscious through not insubstantial portions" of trial. Indeed, the solemn state judgment under review rejected the state habeas trial court's recommended conclusion of law that simply repeated Burdine's "allegation" that Cannon "repeatedly dozed and/or actually slept during substantial portions" of trial. See Ex parte Burdine, No. 16,725-06, at 1, 901 S.W.2d 456 (Tex. Crim. App.), cert. denied, 515 U.S. 1107 (1995);Ex parte Burdine, No. 379,444-B, at 18-19. In rejecting/disavowing that conclusion, the Texas Court of Criminal Appeals held: Burdine must prove actual prejudice under the Strickland two-prong test; and he failed to do so. Ex parte Burdine, No. 16,725-06, at 1. As discussed infra, it may well be that, in citing Strickland, the Court of Criminal Appeals was citing the portion discussing the narrow circumstances for presumed-prejudice. In any event, it rejected the recommended conclusion that Burdine had established such circumstances. Id.

In sum, the majority has turned its back on the ratio decidendi for the Supreme Court cases that must inform our analysis: Teague, decided 12 years ago; and Cronic and Strickland, decided 17 years ago. Each serves a strong interest. The fact that this is a capital murder case does not change that. But, the fact that Burdine waited 11 years to assert the claim, and then withheld crucial evidence, most certainly should guide our analysis, because these tactics strike at the very goals Teague was designed to foster and protect: finality and comity.

Therefore, I must respectfully dissent. I would hold that, under the circumstances of this case, prejudice must be proved. Accordingly, I would remand on that and the myriad ineffective assistance and other issues Burdine raised in his federal habeas application, which the district court did not address. See notes 18-19, infra. It may well be that, on remand, Burdine could, inter alia, satisfy the Strickland two-prong test for ineffective assistance and be accorded a new trial on that basis.

Notwithstanding Burdine's sitting beside Cannon throughout trial, the record contains no affidavit or testimony by Burdine regarding Cannon's sleeping. But, at oral argument before our en banc court, Burdine's habeas counsel admitted he withheld evidence that, at times, Burdine nudged Cannon during trial to awaken him. For the first time in this lengthy state and federal habeas process (since 1987), this crucial evidence has come to light. Had it been timely presented, it could have had a profound impact, certainly on presumed-prejudice vel non. Perhaps, this evidence would have easily pinpointed the portions of the trial during which Cannon slept (such easy identification being an essential element for presumed-prejudice). At the very least, it would have assisted in developing the record on that issue.

This sea change for this extremely belated presumed-prejudice claim began when, on 28 October 2000, in an interview the day after the panel opinion was rendered, Burdine's counsel, for the first time, claimed Burdine kept trying to awaken Cannon during trial.7 When asked at enbanc oral argument why that crucial evidence hadnot been presented in support of Burdine's claim, his counsel responded that Burdine was entitled to choose what evidence to present.

In general, that is true. But, the evidence has been placed in front of us.8 In effect, through this extra-judicial admission, Burdine's habeas counsel has supplemented the record, albeit in a most unusual way. For our court to disregard it is to fail to do our duty.

Now that we have this evidence, it goes without saying that Burdine cannot have it both ways. Knowing what we now know, we cannot allow Burdine, on the one hand, to have withheld during the state habeas proceeding such evidence which might have pinpointed when Cannon was sleeping, while, on the other hand, continuing to claim presumed-prejudice based primarily upon one extremely broad state-finding, that, in turn, was based upon extremely non-specific evidence. In other words, knowing what we now know, Burdine cannotbe allowed to be sheltered by the very uncertainty that assists, ifnot causes, the majority to presume prejudice. The majority allows him to do so. Again, it does not even mention this tactic, other than, in response to this dissent, stating it has no bearing on its presumed-prejudice analysis. Maj. Op. at 28 n.10.9

Because a habeas proceeding is civil in nature, Burdine had the burden of proving his claims by a preponderance of the evidence, including that Cannon slept during a critical stage of trial. See Browder v. Dir., Dep't. of Corr. of Ill., 434 U.S. 257, 269 (1978); Walker v. Johnston, 312 U.S. 275, 286 (1941); Irving v. Breazeale, 400 F.2d 231, 236 (5th Cir. 1968). In the light of our now knowing this key evidence, which Burdine elected not to present, we should, at the very least, employ the uncalled-witness rule. This new evidence about nudging Cannon, which Burdine did not present to the state habeas court, permits a negative inference. See, e.g., Streber v. Comm'r of Internal Revenue, 138 F.3d 216, 221 (5th Cir. 1998) (court may draw negative inference from party's failure to produce witness "'whose testimony would elucidate the transaction'" (quoting Graves v. United States, 150 U.S. 118, 121 (1893)); Gumbs v. Int'l Harvester, Inc., 718 F.2d 88, 96 (3d Cir. 1983) (party's unexplained failure or refusal to produce evidence that would tend to throw light on issues authorizes inference that such evidence would be unfavorable to that party).

Although, as an appellate court, we do not find facts, we can (here, sua sponte) conclude as a matter of law that, had the state habeas court been aware of this withheld-evidence, it would have drawn a negative inference. Cf. Albiar v. State, 739 S.W.2d 360, 362-63 (Tex. Crim. App. 1987) (en banc) (in criminal case, prosecutor may comment in closing argument on defendant's failure "to call a competent and material witness, when it is shown that such witness was available to testify on behalf of the defendant, but was not called by the defendant to testify"; "failure to produce available evidence justifies an inference that it would be unfavorable to the defendant"); Interest of P.A.O., M.P.O., & S.L.O., No. 08-98-00436-CV, 2001

WL 175620, at *13 (Tex. App. -- El Paso 22 Feb. 2001) (unpublished) (in proceeding for termination of parental rights, "jury could draw whatever inference was reasonable under the circumstances [because] ... 'Fifth Amendment does not forbid adverse inferences against parties to civil actions when they refuse to testify in response to probative evidence offered against them'" (quoting Baxter v. Palmigiano, 425 U.S. 308, 318 (1976))).10 If for no other reason, we must do so in order to protect the integrity of our court and this proceeding. Cf. Int'l Union, United Auto., Aerospace & Agric. Implement Workers of Am. (UAW) v. Nat'l Labor Relations Bd., 459 F.2d 1329, 1339 (D.C. Cir. 1972) ("the adverse inference rule plays a vital role in protecting the integrity of the administrative process in cases where a subpoena is ignored").

Accordingly, because Burdine's habeas counsel chose to present this crucial evidence only extra-judicially, and to do so only after completion of the proceedings in the state habeas court and in the federal district court, we are justified in concluding that the state habeas court would have made the following inference adverse to Burdine: had Burdine testified at the state habeas evidentiary hearing, he would have pinpointed the sleeping episodes as having occurred during the presentation of uncontested evidence, for which no response or other action would have been required by Burdine's trial counsel, Cannon. Obviously, this adverse inference would be fatal to his claim that the sleeping occurred at a "critical stage". Therefore, on this basis alone, we should reject presumed-prejudice.11 If not, we should remand to the district court to develop this evidence and its bearing on the presumed-prejudice claim. At the very least, the evidence-withholding admission colors Burdine's presumed-prejudice claim.

The underlying murder was committed 18 years ago. Since then, in addition to the trial (1984) and direct appeal (1986), there have been two state habeas applications (1987, supplemented in 1988 and August 1994, and December 1994) and the pending federal application (1995). The majority ignores the underlying facts (especially Burdine's result-dictating confession and Cannon's attempts to keep it out) and, for the most part, the prior proceedings. They must, however, be examined in order to conduct the analyses mandated for the presumed-prejudice claim and theTeague-bar vel non.

The opinion of the Texas Court of Criminal Appeals, affirming the conviction and sentence on direct appeal, details well the facts underlying Burdine's capital murder conviction. The issue at hand compels repeating that recitation.

On April 20, 1983, the body of the victim, Wise, was discovered lying face down in the north bedroom of his trailer. Wise's hands and legs were bound with cord, and his mouth was gagged. There was a stab wound on Wise's back and blood in the shoulder area and hair.

The State established through competent medical testimony that the cause of Wise's death was two stab wounds to the back. Wise's scalp was lacerated; his mouth was gagged with socks and a pillowcase. The force of the stab wounds was sufficient to break Wise's rib. One of the knife wounds appeared to have been caused by the knife offered in evidence by the State.

The police determined that several items were missing from Wise's trailer: [among other things,] a television, ... handgun, automatic bank teller card, ... and items of clothing. The serial number on the handgun was entered into the National Crime Information Center computer. The gun was described as a Smith and Wesson revolver, gold- and nickel-plated with pearl handles.

[Burdine] gave an extrajudicial confession to the murder. He also testified at trial, where he limited his participation in the killing to that of an accomplice to the aggravated robbery of Wise.

Viewed in the light most favorable to the prosecution, the evidence showed that [Burdine] and Wise met in November of 1982. The two men had a homosexual relationship which continued for approximately three and a half months while [Burdine] was living with Wise. Wise, a night supervisor at Statewide Security Service, obtained a job at the security company for [Burdine].

Eventually, [Burdine] and Wise quarreled about the manner in which Wise handled [Burdine]'s earnings. [Burdine] testifiedthat Wise asked him to move from the trailer after [Burdine] refused to prostitute himself for Wise. [Burdine] moved out, and approximately two weeks later he resigned his job at the security company. According to [Burdine], Wise subsequently "put a contract out on him."

[Burdine] then met Douglas McCreight, a homosexual male, who did not know Wise. On April 18, 1983, [Burdine] and McCreight decided to go to Wise's trailer in order to get money from him. The money was to be obtained either voluntarily or through robbery. [Burdine] warned McCreight not to try anything "funny" with Wise in his bedroom, because Wise kept a gun there.

Soon after they entered the trailer, McCreight asked to use the bathroom. [Burdine] and Wise remained in the living room. When McCreight returned to the living room, he was wearing a pair of gloves and carrying Wise's gun and a large hunting knife. McCreight then ordered Wise to lie on the floor. McCreight removed Wise's glasses, and [Burdine] directed McCreight to take the cord from the telephone; the cord was used to bind Wise's wrists. [Burdine] told McCreight that something was needed to keep Wise quiet, because he would "squeal like a pig in a slaughterhouse."

[Burdine] retrieved a pair of socks, which McCreight stuffed in Wise's mouth, and a section of sheet, which [Burdine] used to tie the gag in place. At this point, [Burdine] and McCreight made Wise move to another part of the trailer, where they would be less likely to be seen by a neighbor. [Burdine] and McCreight began to stack items in the living room by the front door so that they could take them later.

[Burdine] and McCreight then decided that "something had to be done" with Wise because he could identify [Burdine]. McCreight cut the electrical cord of a clock radio and bound Wise's legs with it. [Burdine] and McCreight then unsuccessfully attempted to smother Wise to death. They placed Wise face down on the bed with his face on a pillow. McCreight held a pillow over Wise's head, and [Burdine] held Wise's feet. Wise thrashed around on the bed so much that McCreight was unable to smother him, and Wise sat up on the bed, whimpering and crying. After further discussion, [Burdine] directed McCreight to hit the top of Wise's head with a lead-filled police sap. McCreight struck Wise several times; Wise bled profusely and lay still.

McCreight and [Burdine] then left the trailer, taking the stolen items with them. They discussed Wise and again decided that something needed to be done so that he could not identify [Burdine]. After re-entering the bedroom, McCreight made the sign of the cross and then stabbed Wise in the back. [Burdine] then told McCreight, "What the hell, hand me the knife," and [Burdine] also stabbed Wise in the back.

[Burdine] and McCreight left the trailer and drove to Austin. While in Austin, [Burdine] pawned a television set and obtained money from different automatic teller machines using Wise's bank card. [Burdine] and McCreight proceeded from Austin to California. After they arrived in Eureka, California, they pawned Wise's gun for thirty dollars. Within thirty minutes of this transaction, both men were arrested at a local gas station.

After hearing from the Eureka, California police department that Wise's gun had been recovered, Detectives G.T. Neely and R.W. Holland, both Houston police officers, traveled to California on April 28 [ten days after the murder]. They met McCreight and [Burdine] at the local courthouse. At [Burdine]'s initial appearance before a magistrate, he was given Miranda warnings.

The officers then conducted separate interviews of McCreight and [Burdine]. [Burdine] was again given Miranda warnings by Detective Neely. [Burdine] then gave the officers an oral statement and consented to the search of the pickup truck which he had been driving at the time of his arrest.

In the back of [Burdine]'s truck, the officers found a large hunting knife and some property which had been taken from Wise's trailer, including a suitcase, articles of clothing, some eight-track tapes, and several pieces of jewelry. Pursuant to [Burdine]'s statement, Wise's television and ring were recovered in Austin.

At trial, [Burdine] testified that only McCreight stabbed Wise. However, [Burdine] admitted that he had anticipated some violence when he and McCreight went to Wise's trailer. He also admitted that he told McCreight where Wise kept his gun, that he warned McCreight that Wise would "scream like a stuck hog," and that he told McCreight to gag Wise and to take Wise to the back of the trailer if he were going "to do anything." Further, [Burdine] admitted that he took Wise's property, that he used Wise's automatic bank teller card to obtain money in Austin, and that he pawned the television in Austin.

Burdine v. State, 719 S.W.2d 309, 312-14 (Tex. Crim. App. 1986) (emphasis added; footnote omitted), cert. denied, 480 U.S. 940 (1987).

At trial, Cannon's theory of defense was: McCreight instigated the murder; Burdine, who was recovering from lung-removal surgery, was too weak to have participated in the stabbing (Cannon had Burdine exhibit his surgical scars to the jury); and Wise had taken advantage of Burdine, stealing his money, harassing and threatening him, including putting out contracts to physically harm him, and attempting to force him to prostitute himself.

An examination of the state court record reveals that, despite Cannon's sleeping during unidentified portions of the trial, there was a meaningful adversarial testing of the State's case. Cannon filed pre-trial motions, including for discovery, inspection, and production of evidence, such as Wise's criminal and employment records and photo albums of nude boys allegedly found in Wise's residence after the murder. At trial, Cannon cross-examined the State's witnesses, made objections, and presented witnesses on behalf of Burdine. Most importantly, Cannon vigorously contested the admissibility of Burdine's confession -- obviously, the key evidence of his guilt. See Pyles v. Johnson, 136 F.3d 986, 996 (5th Cir.) (defendant's "confession was probably the most probative and damaging evidence that could be admitted against him" (internal quotation marks, citation, and brackets omitted)), cert. denied, 524 U.S. 933 (1998).

During parts of the first and second days of trial, a hearing was held on Burdine's motion to suppress that confession. At that hearing, Burdine testified: he asked for a lawyer prior to making the statement in California, but was told by Houston detectives he would not need an attorney because one would be appointed for him on his return to Texas; and, he twice told the detectives he did not participate in stabbing Wise, but they forced him to confess, telling him he could not return to Texas until he admitted his involvement in the murder. Cannon argued that the confession was not voluntary and should not be admitted. The motion was overruled.

When Burdine subsequently testified at the guilt-innocence phase of trial, he admitted his participation in the robbery, but denied stabbing Wise. Regarding his confession, his testimony was consistent with his suppression hearing testimony. And, the jury was instructed on voluntariness.

At the penalty phase, outside the presence of the jury and against Cannon's advice, Burdine declined to testify. Immediately thereafter, in the presence of the jury, Cannon asked Burdine if he wished "to take the stand and plead for [his] life". Before being interrupted by the trial judge, Burdine responded to Cannon: "No, sir, they didn't listen to me the first time, I don't see -- ".

At the conclusion of the penalty phase, Cannon asked Burdine if he wanted him (Cannon) to handle the appeal or whether he wanted the court to appoint someone else. Burdine replied: "Your Honor, with the court's permission, I would like to have Mr. Joe Cannon represent me".

On direct appeal, Burdine, represented by Cannon, raised 17 points of error, including the admission of his confession. Burdine, 719 S.W.2d at 312.12 Regarding his confession, Burdine claimed: it was obtained in violation of his right to counsel; it was induced by police trickery; and the officers' promise to return him to Houston quickly if he confessed rendered it involuntary. Id. at 317.

In October 1986, the Texas Court of Criminal Appeals affirmed the conviction and sentence. Id. at 309. As for the confession, the court held that the record supported the trial court's findings that Burdine waived his right to counsel, received no promises, and was in good physical condition at that time. Id. at 318. The court also held: even if Burdine's confession were disregarded, the evidence was sufficient to support a conviction under the law of the parties. Id. at 315. The Supreme Court denied certiorari in March 1987. Burdine v. Texas, 480 U.S. 940 (1987).

Represented by new counsel, Burdine filed his first state habeas application in July 1987, presenting approximately ten claims, including that his custodial statements were admitted in violation of the Constitution and that he was denied effective- assistance at trial and on direct appeal.13 For the ineffective-assistance claim, Burdine presented approximately ten bases, including that Cannon was ineffective in cross-examination of witnesses, in arguments, and in preparation and presentation of the defense.14 And, in March 1988, Burdine filed a supplemental state habeas application, adding several claims.15

That September, a special master appointed by the state habeas trial court conducted a three-day evidentiary hearing, at which Cannon, among others, was called as a witness by Burdine and testified regarding his theory of defense and his investigation of possible mitigating evidence.16 Two years later, in October 1990, the court-appointed master made proposed findings of fact and conclusions of law and recommended vacating Burdine's sentence, based on the prosecutor's comments about homosexuality during closing argument, as well as on Cannon's performance at the penalty phase. With regard to the latter, the special master recommended presumed-prejudice under Cronic; alternatively, that Burdine had established prejudice under the Strickland two-prong test. But, nearly four years later, in June 1994, the state habeas trial court recommended that the Texas Court of Criminal Appeals deny relief. Ex parte Burdine, No. 379,444-A (183d Dist. Ct. Harris County, Texas, 29 June 1994). That same month, Burdine filed a second supplemental state habeas application, claiming, under the Eighth and Fourteenth Amendments, the State had forfeited its right to execute him, because of various forms of post-conviction torture.

That December (1994), the Court of Criminal Appeals denied relief. Ex parte Burdine, No. 16,725-02 (Tex. Crim. App. 12 Dec. 1994). Four days later, the state habeas trial court issued supplemental findings of fact and conclusions of law, recommending that the State had not forfeited its right to execute Burdine.

Burdine filed a second state habeas application later that month, nearly 11 years after trial. Among other claims, he asserted for the first time that Cannon dozed and/or slept repeatedly at trial. The application states that the factual basis for that claim was not known to counsel until 27 December 1994, when the jury foreman informed counsel's investigator that Cannon slept during portions of Burdine's trial.17

Regarding Burdine's Cannon-slept-claim, the state trial court conducted an evidentiary hearing in February 1995. At that hearing, described in detail infra, Burdine, as noted, did not testify, nor did he submit an affidavit, concerning the claimed sleeping. As also noted, his counsel withheld the evidence, recently revealed, that Burdine nudged Cannon during trial. Burdine presented the testimony of three jurors, the prosecutor, the trial judge, the court clerk, the trial judge's court coordinator (who had testified in the first state habeas evidentiary hearing), and an attorney who had served as co-counsel with Cannon on another capital murder case. The State presented the testimony of Cannon and one juror.

That April, the state habeas trial court recommended granting relief, finding, inter alia: Cannon "dozed and actually fell asleep during portions of [Burdine's] trial on the merits, in particular during the guilt-innocence phase when the State's solo prosecutor[] was questioning witnesses and presenting evidence". Ex parte Burdine, No. 379,444-B, at 13. Its recommended Conclusion of Law number 1 provided Burdine had

established per se ineffective assistance of counsel based on the allegation that [Cannon] repeatedly dozed and/or actually slept during substantial portions of [Burdine's] capital murder trial so that [Cannon] was, in effect, absent and that such conduct by [Cannon] is inherently prejudicial and thus no showing of prejudice is necessary.

Id. at 18-19 (emphasis added).

But, as noted supra and of critical importance here, later that month the Court of Criminal Appeals expressly rejected that recommended conclusion, ruling instead: although the trial court's factual findings were supported by the record, Burdine was not entitled to relief because he had failed to meet his burden of proof under Strickland. Ex parte Burdine, No. 16,725-06, at 1, 901 S.W.2d 456. The Supreme Court denied certiorari on 30 May 1995. Burdine v. Texas, 515 U.S. 1107 (1995).

Meanwhile, in April 1995, Burdine sought federal habeas relief. He presented ten claims, including ineffective-assistance.18 Burdine's ineffective-assistance claim, in addition to asserting presumed-prejudice because Cannon slept during trial, listed 14 other bases.19

In September 1999, without addressing the remaining claims,the district court granted relief on the presumed-prejudice claim. Burdine v. Johnson, 66 F. Supp. 2d 854 (S.D. Tex. 1999). It adopted the rule of Javor v. United States, 724 F.2d 831, 834 (9th Cir. 1984), decided before Cronic, that prejudice must be presumed if counsel slept for a "substantial portion" of trial. Burdine, 66 F. Supp. 2d at 862. To determine what constituted a "substantial portion", however, the district court applied the analysis announced in Tippins, 77 F.3d 682, decided in 1996, long after completion of Burdine's trial, appeals, and state habeas proceedings: "(1) did counsel sleep for repeated and/or prolonged lapses; (2) was counsel actually unconscious; and (3) were the defendant's interests at stake while counsel was asleep". Burdine, 66 F. Supp. 2d at 863-64. In applying the first element of theTippins analysis, the district court stated:

The state [habeas trial] court concluded [(notfound)] that Cannon slept for numerous periods of time and the sleeping was "substantial." These findings of fact were explicitly adopted by the Texas Court of Criminal Appeals. This Court finds, pursuant to the presumption of correctness standard, that Cannon slept on numerous occasions throughout Burdine's criminal trial and for substantial periods of time.

Id. at 865 (emphasis added).

The district court failed to note there was no state-finding that the sleeping occurred during "substantial" portions of trial. It also overlooked the Court of Criminal Appeals' express rejectionof the recommended conclusion of law, which simply described Burdine's "allegation". Ex parte Burdine, No. 379,444-B, at 18-19;Ex parte Burdine, No. 16,725-06, at 1.

Moreover, the district court did not conduct a Teagueanalysis, despite the State's having raised Teague as a bar to Burdine's sleeping-counsel-claim. See Goeke v. Branch, 514 U.S. 115, 117 (1995) ("Although a court need not entertain [a Teague] defense if the State has not raised it, a court must apply it if it was raised by the State." (citations omitted)).

Presumed-prejudice should be rejected because of the withheld-evidence tactic employed by Burdine's counsel, which our en banc court noted sua sponte and inquired about at argument. In the alternative, the claim still fails.

Burdine filed his federal habeas application (1995) prior to enactment of the Anti-Terrorism and Effective Death Penalty Act of 1996 (AEDPA). Therefore, AEDPA's standards for reviewing the state court's decision are not applicable. E.g., Perillo v. Johnson, 205 F.3d 775, 793 (5th Cir. 2000). As a result, we apply the pre-AEDPA standards. For applying such standards to ineffective-assistance claims, questions of deficient performance and prejudice are legalconclusions reviewed de novo. Moore v. Johnson, 194 F.3d 586, 603-04 (5th Cir. 1999). Likewise, whether Teague precludes Burdine from benefitting from the claimed prejudice-presumption is reviewedde novo. See United States v. Shunk, 113 F.3d 31, 34 (5th Cir. 1997) (§ 2255).

On the other hand, for habeas review, "[t]he state court's subsidiary findings of specific historical facts and state court credibility determinations are ... entitled to a presumption of correctness under [pre-AEDPA] § 2254(d)". Moore, 194 F.3d at 604. Therefore, as the State concedes, we are bound by the key state-finding that Cannon "dozed and actually fell asleep during portions of [Burdine's] trial on the merits, in particular during the guilt-innocence phase when the State's solo prosecutor[] was questioning witnesses and presenting evidence".

Generally, we presume effective assistance of counsel; Burdine has the burden of overcoming that presumption. See Cronic, 466 U.S. at 658. Only in extremely narrow circumstances will prejudice be presumed. E.g., Craker v. McCotter, 805 F.2d 538, 542 (5th Cir. 1986). "'The essence of an ineffective assistance claim is that counsel's unprofessional errors so upset the adversarial balance between defense and prosecution that the trial was rendered unfair and the verdict rendered suspect.'" Goodwin v. Johnson, 132 F.3d 162, 172 (5th Cir. 1998) (quoting Kimmelman v. Morrison, 477 U.S. 365, 374 (1986)). Restated, "[t]he benchmark for judging any claim of ineffectiveness must be whether counsel's conduct so underminedthe proper functioning of the adversarial process that the trial cannot be relied on as having produced a just result". Strickland, 466 U.S. at 686.

Teague's nonretroactivity principle "prevents a federal court from granting habeas relief to a state prisoner based on a rule announced after his conviction and sentence became final", Caspari v. Bohlen, 510 U.S. 383, 389 (1994) (emphasis in original), unless certain narrow exceptions (two announced by the Supreme Court and a third recently created by our court) apply. The majority holds its rule is not a proscribed "new rule" under Teague. I respectfully disagree.

Habeas corpus is "to afford relief to those whom society has 'grievously wronged'". Brecht v. Abrahamson, 507 U.S. 619, 637(1993) (defining harmless error standard applicable in habeas cases (citation omitted)). Concomitantly, granting a new trial -- especially based on a presumption -- has serious consequences.

Retrying defendants whose convictions are set aside ... imposes significant social costs, including the expenditure of additional time and resources for all the parties involved,the erosion of memory and dispersion of witnesses that accompany the passage of time and make obtaining convictions on retrial more difficult, and the frustration of society's interest in the prompt administration of justice.

Id. (emphasis added; internal quotation marks and citation omitted). The Teague nonretroactivity doctrine "validates reasonable, good-faith interpretations of existing precedents made by state courts, and thus effectuates the States' interest in the finality of criminal convictions and fosters comity between federal and state courts". Gilmore v. Taylor, 508 U.S. 333, 340 (1993) (internal quotation marks and citation omitted); see also Lockhart v. Fretwell, 506 U.S. 364, 372 (1993) (Teague nonretroactivity rule "was motivated by a respect for the States' strong interest in the finality of criminal convictions, and the recognition that a State should not be penalized for relying on the constitutional standards that prevailed at the time the original proceedings took place" (internal quotation marks and citation omitted)).

Teague serves these interests by "validat[ing] reasonable, good-faith interpretations of existing precedents made by state courts even though they are shown to be contrary to later decisions". Butler v. McKellar, 494 U.S. 407, 414 (1990).

In many ways the application of new rules to cases on collateral review may be more intrusive than the enjoining of [state] criminal prosecutions, for it continuallyforces the States to marshal resources in order to keep in prison defendants whose trials and appeals conformed to then-existing constitutional standards. Furthermore, ... state 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.

Teague, 489 U.S. at 310 (plurality) (emphasis in original; internal quotation marks, citations, and brackets omitted).

The Teague doctrine recognizes that "[a]pplication of constitutional rules not in existence at the time a conviction became final seriously undermines the principle of finality which is essential to the operation of our criminal justice system. Without finality, the criminal law is deprived of much of its deterrent effect". Id. at 309 (plurality); see also id. ("'No one, not criminal defendants, not the judicial system, not society as a whole is benefited by a judgment providing that a man shall tentatively go to jail today, but tomorrow and every day thereafter his continued incarceration shall be subject to fresh litigation.'" (quoting Mackey v. United States, 401 U.S. 667, 691 (1971) (Harlan, J., concurring in judgments in part and dissenting in part))).20

"In determining whether a state prisoner is entitled to habeas relief, a federal court should apply Teague by proceeding in three steps." Caspari, 510 U.S. at 390.

First, we must determine when [Burdine's] conviction and sentence became final forTeague purposes.... Second, we must "survey the legal landscape as it then existed and determine whether a state court considering the defendant's claim at the time his conviction became final would have felt compelled by existing precedent to conclude that the rule he seeks was required by the Constitution."... Third, if we determine that [Burdine] seeks the benefit of a new rule, we must consider whether "that rule falls within ... the ... narrow exceptions to the nonretroactivity principle".

Fisher v. Texas, 169 F.3d 295, 305 (5th Cir. 1999) (quoting Caspari, 510 U.S. at 390).

Burdine's conviction became final in 1987, when the Supreme Court denied certiorari. Therefore, the second step of the Teagueanalysis concerns whether, based on the "legal landscape" in 1987, Burdine seeks a "new rule". "[I]n general, a case announces a 'new rule' when it breaks new ground or imposes a new obligation on the States or the Federal Government". Butler, 494 U.S. at 412 (emphasis added). Restated, "a decision announces a new rule if the result was not dictated by precedent existing at the time the defendant's conviction became final". Id. (emphasis in original; internal quotation marks and citations omitted). We must ask whether, in 1987, Texas courts "would have felt compelled by existing precedent to conclude that the [presumed-prejudice] rule [Burdine] seeks was required by the" Sixth Amendment. Fisher, 169 F.3d at 305.

Obviously, the Teague inquiry is more difficult in cases in which the "decision is reached by an extension of the reasoning of previous cases". Butler, 494 U.S. at 412-13. "Courts frequently view their decisions as being 'controlled' or 'governed' by prior opinions even when aware of reasonable contrary conclusions reached by other courts". Id. at 415. But, when the new rule vel nondetermination is "susceptible to debate among reasonable minds", the rule is "new". Id.21

At issue in Butler was whether a new rule was established by Arizona v. Roberson, 486 U.S. 675 (1988) (Fifth Amendment bars police-initiated interrogation following suspect's request for counsel in context of separate investigation). The explanation inButler of why Roberson announced a new rule is pertinent here:

In Roberson, ... the Court found Edwards [v. Arizona, concerning waiver vel non of right to counsel during police interrogation,] controlling but acknowledged a significant difference of opinion on the part of several lower courts that had considered the question previously. That the outcome in Roberson was susceptible to debate among reasonable minds is evidenced further by the differing positions taken by the judges of the Courts of Appeals for the Fourth and Seventh Circuits .... It would not have been an illogical or even a grudging application of Edwards to decide that it did not extend to the facts ofRoberson. We hold, therefore, that Robersonannounced a new rule.

Butler, 494 U.S. at 415 (internal quotation marks and citations omitted).

Although the majority does not mention it, the Supreme Court and our court frequently have applied this "debatable among reasonable jurists" standard in determining whether a rule is "new".22 See, e.g., Caspari, 510 U.S. at 393 ("a reasonable jurist reviewing our precedents at the time respondent's conviction and sentence became final would not have considered the application of the Double Jeopardy Clause to a noncapital sentencing proceeding to be dictated by our precedents"); Graham v. Collins, 506 U.S. 461, 476 (1993) ("The result in a given case is not dictated by precedent if it is susceptible to debate among reasonable minds, or, put differently, if reasonable jurists may disagree" (internal quotation marks and citations omitted)); Matthew v. Johnson, 201 F.3d 353, 363 (5th Cir.) (in Teague analysis, "the reasonable views of state courts are entitled to consideration along with those of federal courts" (internal quotation marks and citation omitted)),cert. denied, 121 S. Ct. 291 (2000); Fisher, 169 F.3d at 305 ("reasonable jurists, considering the question in 1996, would nothave felt compelled by existing precedent to rule that religion-based peremptory challenges violate the Equal Protection Clause");Vega v. Johnson, 149 F.3d 354, 357 (5th Cir. 1998) ("If reasonable minds could differ on whether current law requires relief, we may not grant relief without creating a 'new rule' barred by Teague."),cert. denied, 525 U.S. 1119 (1999).23

As quoted earlier, the key component in determining whether a rule is new is whether a state court considering the claim at the time petitioner's conviction became final would have felt compelled by existing precedent to conclude that the rule sought by petitioner is required by the Constitution. Accordingly, we must determine whether the rule Burdine seeks is "dictated by ... precedent [existing at the time his conviction became final in 1987] -- whether, that is, the unlawfulness of [Burdine's] conviction was apparent to all reasonable jurists". Lambrix v. Singletary, 520 U.S. 518, 527-28 (1997) (emphasis in original). It is not enough, under Teague, that the rule is "a reasonable interpretation of prior law". Id. at 538. Instead, the relevant inquiry is "whether no other interpretation was reasonable". Id. (emphasis in original); see also Graham, 506 U.S. at 477 (same). "Unless reasonable jurists hearing [Burdine's] claim at the time his conviction became final [in 1987] would have felt compelled by existing precedent to rule in his favor [on his presumed-prejudice claim], we are barred from doing so now." Fisher, 169 F.3d at 305 (internal quotation marks, brackets, and citation omitted).

Whether Cronic dictates presumed-prejudice when, as in this case, counsel sleeps during unidentifiable portions of a capital murder trial, but otherwise provides meaningful assistance to his client, is certainly susceptible to debate among reasonable jurists, as reflected by the majority and dissenting opinions of our court's panel, 231 F.3d 950, vacated, 234 F.3d 1339 (5th Cir. 2000), and, most certainly, by the previously referenced opinions of the state habeas trial court and the Court of Criminal Appeals.

The state habeas trial court's recommended conclusion was that Burdine had established presumed-prejudice. Ex Parte Burdine, No. 379,444-B, at 18-19. In 1995, however, a majority of the Court of Criminal Appeals rejected that conclusion; it held Burdine was not entitled to relief "because he has failed to discharge his burden of proof under Strickland". Ex Parte Burdine, No. 16,725-06, at 1, 901 S.W.2d 456. Three justices dissented, noting that, in Javor, 724 F.2d 831, the Ninth Circuit had found a Sixth Amendment violation under similar circumstances; the dissent stated: "The issue presented in this case has never been addressed by the United States Supreme Court nor by this court". Ex parte Burdine, 901 S.W.2d at 458. As stated earlier, the majority opinion by the Court of Criminal Appeals does not even refer to presumed-prejudice, although perhaps its citation to Strickland was intended to include not only the prejudice analysis as part of the two-prong test, but also Strickland's discussion of presumed-prejudice. In any event, the recommended presumed-prejudice conclusion was rejected.

A survey of the legal landscape as it existed when Burdine's conviction became final in 1987 demonstrates that the rule fashioned now by the majority is not dictated by such precedent. Restated, Texas courts, considering Burdine's claim in 1987, would not have felt compelled to presume prejudice because of Cannon's sleeping during unidentified portions of trial.

As the majority notes, the Supreme Court held, in 1932, that a capital defendant has a constitutional right to "the guiding hand of counsel at every step in the proceedings against him". Powell v. Alabama, 287 U.S. 45, 69 (1932). In this regard, the Court held subsequently that showing prejudice was not necessary when the defendant was denied counsel at arraignment, a critical stage of the proceedings, because certain defenses were lost if not then pleaded. Hamilton v. Alabama, 368 U.S. 52, 53-55 (1961).24 Similarly, a defendant was denied assistance of counsel when the trial judge, pursuant to state statute, denied defense counsel the opportunity to be heard in summation at a bench trial, despite the fact there was no way to know whether argument might have affected the outcome. Herring v. New York, 422 U.S. 853, 864-65 (1975). And, shortly thereafter, the Court reversed a decision that defendant's failure to claim prejudice was fatal to his Sixth Amendment claim, and then concluded that a court order preventing him from consulting with his counsel during a 17-hour overnight recess between defendant's direct and cross-examination deprived him of assistance of counsel. Geders v. United States, 425 U.S. 80, 82, 91 (1976).

As discussed, the quite well-known, and quite often applied, standards for ineffective-assistance were established in 1984 inStrickland and Cronic. Under Strickland's two-prong test, "the defendant must show that counsel's performance was deficient" -- "counsel made errors so serious [he] was not functioning as the 'counsel' guaranteed the defendant by the Sixth Amendment". Strickland, 466 U.S. at 687 (emphasis added). In addition, he "must show that the deficient performance prejudiced the defense" -- "counsel's errors were so serious as to deprive the defendant of a fair trial, a trial whose result is reliable". Id. (emphasis added). For the prejudice prong, "[t]he defendant must show that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different". Id. at 694 (emphasis added). "A reasonable probability is a probability sufficient to undermine confidence in the outcome." Id.

Strickland, which concerned sentencing for a capital murder case, observed, however: "In certain Sixth Amendment contexts, prejudice is presumed". Id. at 692. Such contexts were described as "[a]ctual or constructive denial of the assistance of counselaltogether" and "various kinds of state interference with counsel's assistance". Id. (emphasis added). "Prejudice in these circumstances is so likely that case-by-case inquiry into prejudice is not worth the cost." Id. (emphasis added). "[S]uch circumstances involve impairments of the Sixth Amendment right that are easy to identify ... and [in those instances where] the prosecution is directly responsible, easy for the government to prevent". Id. (emphasis added).25 "[A] similar, though more limited, presumption of prejudice" applies "when counsel is burdened by an actual conflict of interest". Id.

Cronic, decided the same day as Strickland, held presumed-prejudice unwarranted under the circumstances of that case (for complex mail fraud prosecution, young lawyer with real estate practice appointed to represent defendant and allowed only 25 days for pretrial preparation). Cronic, 466 U.S. at 666. But, as it did in Strickland, the Court observed: "There are ... circumstances ... so likely to prejudice the accused that the cost of litigating their effect in a particular case is unjustified". Id. at 658. Such circumstances include: (1) "the complete denial of counsel", id. at 659 (emphasis added); (2) where "counselentirely fails to subject the prosecution's case to meaningful adversarial testing", id. (emphasis added); (3) "when [as also discussed in note 24, supra,] counsel was either totally absent, or prevented from assisting the accused during a critical stage of the proceeding", id. at 659 n.25 (citing, inter alia, Geders, Herring, and Hamilton; emphasis added); and (4) "when counsel labors under an actual conflict of interest", id. at 662 n.31. But, "[a]part from circumstances of that magnitude, ... there is generally no basis for finding a Sixth Amendment violation unless the accused can show how specific errors of counsel undermined the reliability of the finding of guilt". Id. at 659 n.26.26

The majority holds that its rule is dictated by the third circumstance described in Cronic: counsel is absent during acritical stage. It holds there is such absence when counsel "is repeatedly unconscious during not insubstantial portions" of the guilt-innocence phase of a capital murder trial. In support of its holding its rule is not new, but instead merely an application to an analogous case of the general Sixth Amendment principles established in Cronic, the majority, Maj. Op. at 11-15, erroneously relies on Penry v. Lynaugh, 492 U.S. 302 (1989), regarding Penry's claim that the Texas death penalty statute prevented the jury from giving any mitigating effect to the evidence of his mental retardation and abuse as a child. That claim, however, did not require the creation of a new rule, because, prior to Penry'sconviction becoming final, established precedent required the State to allow the jury to give effect to mitigating evidence in making its sentencing decision. See Saffle v. Parks, 494 U.S. 484, 491-92 (1990). Nothing in Penry supports the majority's extension ofCronic's critical-stage rule to the facts of Burdine's case -- even the facts found improperly by the majority.

Burdine's case is not analogous to the circumstances for whichCronic found the presumption appropriate. The cases cited inCronic as examples of counsel's being either absent or prevented from assisting the accused during a critical stage (Geders,Herring, and Hamilton) did not involve circumstances even remotely analogous to Burdine's. In fact, in each, the government was responsible for the denial of counsel. See Geders, 425 U.S. at 82 (court order); Herring, 422 U.S. at 864-65 (state statute);Hamilton, 368 U.S. at 53 (no counsel appointed for arraignment). Obviously, the State of Texas was not responsible for Cannon's sleeping. Indeed, the state habeas trial court credited the testimony of the trial judge and prosecutor that they did not observe him doing so.

Of course, as Cronic also noted, "[t]he fact that the accused can attribute a deficiency in his representation to a source external to trial counsel[, such as the State,] does not make it any more or less likely that he received the type of trial envisioned by the Sixth Amendment, nor does it justify reversal of his conviction absent an actual effect on the trial process or the likelihood of such an effect". Cronic, 466 U.S. at 662 n.31.27 Nevertheless, that the denial of counsel in Geders, Herring, andHamilton was government-instigated serves to distinguish them from the claimed denial in this case. And, obviously, because the State was not responsible for Cannon's sleeping, such conduct was not "easy for [it] to prevent". Strickland, 466 U.S. at 692.

Moreover, the stage of the proceeding at which counsel was denied in Geders, Herring, and Hamilton was easily identifiable. See Geders, 425 U.S. at 82 (overnight recess); Herring, 422 U.S. at 864-65 (closing argument); Hamilton, 368 U.S. at 53 (arraignment). In contrast, it is impossible to determine when Cannon slept. Accordingly, Burdine's claim does not "involve impairments of the Sixth Amendment right that are easy to identify". Strickland, 466 U.S. at 692.

Although Burdine contends Javor supports his claim's being encompassed by the critical stage circumstance described in Cronic, the majority, Maj. Op. at 26-27, does not apply Javor's rule that presumed-prejudice is appropriate when counsel sleeps through a substantial portion of trial. Javor, 724 F.2d at 833. Instead, it has created a different rule: presumed-prejudice is appropriate when counsel is "repeatedly unconscious through not insubstantial portions".

When Burdine's conviction became final in 1987, Texas courts would not have felt compelled to apply Javor, decided in 1984. Cf.Magouirk v. Phillips, 144 F.3d 348, 361 (5th Cir. 1998) ("state courts are not bound by Fifth Circuit precedent when making a determination of federal law"). Moreover, neither Cronic nor the majority opinion in Strickland cited Javor as an example of the denial of counsel at a critical stage. Instead, Javor was cited only in a concurring opinion in Strickland. 466 U.S. at 703 n.2 (Brennan, J., concurring). Obviously, Javor being so cited makes clear that the Supreme Court was quite well aware of Javor and the claim that sleeping counsel justifies presumed-prejudice. Javor's not being cited in the Cronic and Strickland majority opinions as an example of presumed-prejudice is strong medicine indeed. It is, at the very least, an indication the Court had not decided presumed-prejudice was applicable in such a situation and amply demonstrates that, for Burdine's claim, the majority's application of presumed-prejudice is a "new" rule, one not dictated by Cronic.28

Moreover, as noted, Javor (presumed-prejudice if slept through substantial portion) does not dictate the majority's rule (presumed-prejudice if "repeatedly unconscious through not insubstantial portions"). There was evidence in Javor that: defense counsel slept during a substantial portion of trial; counsel failed to participate when evidence against the defendant was being presented; counsel stated to counsel for a co-defendant that he had missed some of the testimony; other counsel often "nudged" and "kicked" counsel to awaken him; and the trial judge was concerned about counsel's inattentiveness. Javor, 724 F.2d at 833-34.

For Burdine, the district court applied Javor's "substantial portion" rule, modified by the rule announced in Tippins, decided in 1996: prejudice must be presumed if, while the defendant's interests were at stake, counsel slept for repeated or prolonged lapses and was actually unconscious. Burdine, 66 F. Supp. 2d at 863-64. As discussed, the majority does not adopt that rule either. In any event, in 1987, Texas courts obviously would not have felt compelled to apply the substantiality analysis fromTippins, which was not decided until 1996, long after the conclusion of Burdine's trial, direct appeal, and state habeas proceedings.

It is not even clear Tippins applied presumed-prejudice. Itstates: "Under these circumstances, where the adversary nature of the proceeding was subject to repeated suspensions [because of counsel's unconsciousness] there is little difference between saying that prejudice will be presumed and saying that prejudice has been demonstrated". Tippins, 77 F.3d at 687. The Tippinscourt concluded: "Tippins suffered prejudice, by presumption or otherwise, if his counsel was repeatedly unconscious at trial for periods of time in which defendant's interests were at stake". Id. In most cases, in order to apply the Tippins analysis, examination of the trial record is necessary; otherwise, it would usually be difficult to determine whether the sleeping occurred while the defendant's interests were at stake. But, such record-examination is totally at odds with the rationale for presumed-prejudice (case-by-case inquiry not worth the cost of litigating prejudice vel nonbecause the Sixth Amendment violations are so easy to identify and prejudice is so likely to have occurred). See Strickland, 466 U.S. at 692.

Notwithstanding the majority's view that the Second Circuit has confirmed the rule and rationale in Tippins, it is not clear that the Second Circuit views Tippins as establishing a rule of presumed-prejudice.29 In any event, Tippins is distinguishable. Unlike in Burdine's case, the facts in Tippins regarding sleeping were undisputed. Tippins, 77 F.3d at 685. The trial judgetestified that Tippins' counsel "slept every day of the trial ... during testimony that was damaging and adverse to [Tippins'] interests". Tippins v. Walker, 889 F. Supp. 91, 92 (S.D.N.Y. 1995) (internal quotation marks and citation omitted), aff'd, 77 F.3d 682 (2d Cir. 1996). On one occasion when counsel was sleeping during testimony detrimental to Tippins, the trial judge removed the attorneys from the courtroom to admonish Tippins' counsel for sleeping. Id. The prosecutor also witnessed the sleeping, as did the court reporter, who testified she heard Tippins' counsel snoring several times. Id. And, a juror testified that counsel slept through approximately 65 percent of a critical prosecution witness' testimony. Id.; see also Tippins, 77 F.3d at 687-89. It was quite clear that Tippins' counsel slept while evidence harmful to Tippins' interests was being presented.

Post-Cronic decisions demonstrate prejudice is to be presumedonly in very narrow circumstances, where the defendant receives no meaningful assistance of counsel. Recently, in Smith v. Robbins, 528 U.S. 259 (2000), the defendant claimed he had been denied effective assistance of appellate counsel because counsel requested leave to withdraw, supported by a brief and pursuant to a new California procedure which the defendant alleged failed to comply with Anders v. California, 386 U.S. 738 (1967) (establishing procedures for withdrawal of court-appointed appellate counsel for criminal defendant on direct appeal and for dismissal of appeal if there are no non-frivolous issues). The Court reversed the Ninth Circuit's judgment that the procedure used by counsel failed adequately to comply with the constitutional principles identified in Anders, but remanded for a determination of whether the appeal was frivolous or whether it warranted the filing of a merits-brief. Robbins, 528 U.S. at 283-85.

The Court instructed that, on remand, the defendant would be required to satisfy Strickland's two-prong test. Id. at 285. It explained: "where, as here, the defendant has received appellate counsel who has complied with a valid state procedure for determining whether the defendant's appeal is frivolous, and the State has not at any time left the defendant without counsel on appeal, there is no reason to presume that the defendant has been prejudiced". Id. at 286. Moreover, Robbins' claim did "not fall within any of the three categories of cases, described inStrickland, in which we presume prejudice rather than require a defendant to demonstrate it". Id. at 287. Those three categories were described as: "denial of counsel"; "various kinds of stateinterference with counsel's assistance"; and "when counsel is burdened by an actual conflict of interest". Id. (internal quotation marks and citations omitted). Therefore, the categories described in Cronic as (1) complete denial of counsel, (2) failure to subject the case to meaningful adversarial testing, and (3) totally absent during a critical stage are subsumed within the first category identified in Robbins: "denial of counsel".

The Robbins Court held the policies supporting the first two categories it described were inapplicable, because counsel's unreasonable choice of a procedure such as Anders or the new California procedure followed by Robbins' counsel, in lieu of filing a merits-brief, did not make prejudice "so likely that case-by-case inquiry into prejudice is not worth the cost". Id. (internal quotation marks and citation omitted). "Moreover, such an error by counsel is neither easy to identify (since it is necessary to evaluate a defendant's case in order to find the error) nor attributable to the prosecution." Id. at 287 n.15 (internal quotation marks and citation omitted).

Our court consistently has held likewise. For example, May v. Collins, 948 F.2d 162 (5th Cir. 1991), cert. denied, 502 U.S. 1046 (1992), rejected a presumed-prejudice claim that "the structure of the Texas sentencing statute so forced [May's] attorney's tactical decision on whether to present mitigating evidence as to result in a constructive denial" of counsel. Id. at 167. Our court noted that the Supreme "Court has found constructive denials of counsel only under a few limited circumstances". Id.

Goodwin, 132 F.3d 162, rejected a presumed-prejudice claim where appellate counsel failed to provide the appellate court with a suppression hearing transcript. "Cronic-type prejudice results in circumstances in which, although counsel is present, counsel's ineffectiveness is so egregious that the defendant is in effect denied any meaningful assistance of counsel at all. When the defendant receives at least some meaningful assistance, he must prove prejudice." Id. at 176-77 n.10 (citation omitted). Because Goodwin's counsel provided some meaningful assistance on appeal, "[t]he failure of Goodwin's appellate counsel to read two days of the trial record falls far short of establishing that any deficiency in his performance precluded meaningful appellate review entirely or in effect constituted no assistance of appellate counsel at all". Id. (citation omitted).

In Childress v. Johnson, 103 F.3d 1221 (5th Cir. 1997), prejudice was presumed where counsel was appointed merely to waive the defendant's right to a jury trial. But, our court emphasized: "constructive denial of counsel as described in Cronic affords onlya narrow exception to the requirement that prejudice be proved"; and "we have consistently distinguished shoddy representation fromno defense at all". Id. at 1229 (emphasis added). Thus, where "the defendant has received some meaningful assistance, it [is] necessary to prove prejudice". Id.

Jackson v. Johnson, 150 F.3d 520 (5th Cir. 1998), cert. denied, 526 U.S. 1041 (1999), rejected a presumed-prejudice claim where appellate counsel failed to include challenged videotape evidence as part of the record on appeal; Jackson could "point tono clearly established Federal law from the Supreme Court that says, in anything like his situation, prejudice is presumed". Id. at 524. Instead, "the constructive-denial claim is a very narrow exception to the Strickland prejudice requirement". Id. Because Jackson's claim involved "shoddy representation -- one essential error in the midst of otherwise adequate representation -- rather than total absence of counsel", Jackson had the burden of proving "the error complained of resulted in Strickland prejudice". Id. at 525 (footnote omitted).

As a final, and recent, example, in Gochicoa v. Johnson, 238 F.3d 278 (5th Cir. 2000), our court reversed the district court's holding of presumed-prejudice for the defendant's claim of constructive denial of counsel based on counsel's failure to object to inadmissible hearsay and to seek disclosure of an informant's identity. Id. at 283-84. "[P]rejudice is presumed ... only when the defendant demonstrates that counsel was not merely incompetent but inert, distinguishing shoddy representation from norepresentation at all. When the defendant complains of errors, omissions, or strategic blunders, prejudice is not presumed; bad lawyering, regardless of how bad, does not support the per sepresumption of prejudice." Id. at 284-85 (internal quotation marks and citation omitted). "When the defendant receives at least some meaningful assistance, he must prove prejudice in order to obtain relief for ineffective assistance of counsel." Id. at 285 (internal quotation marks and citation omitted). Because counsel had provided "some meaningful assistance", Cronic's constructive-denial test was not applicable. Id.30

As the above-discussed cases demonstrate, Cronic does not dictate the majority's rule -- far from it. More importantly, and at the very least, when Burdine's conviction became final in 1987, Texas courts would not have felt compelled to presume prejudice where Cannon slept during unidentified portions of trial, but otherwise provided some -- indeed, a great deal of -- meaningfulassistance to Burdine.

Again, the rationale for presumed-prejudice in cases involving the denial of counsel, actual or constructive, is: the circumstances involving the impairment of the right to counsel are easy to identify; prejudice is so likely that a case-by-case inquiry into prejudice vel non is not worth the cost; and, for instances of government interference, such circumstances are easy to prevent, because the prosecution is directly responsible. SeeRobbins, 528 U.S. at 287 & n.15; Strickland, 466 U.S. at 692;Cronic, 466 U.S. at 658. This rationale underscores the fact that, to presume prejudice in this case entails creating a new rule, because none of these circumstances are present.

As discussed in great detail infra, the record reflects periods of inactivity, but not necessarily sleep, by Cannon, bothwhen evidence harmful to Burdine's interests was being presented and when uncontested evidence (such as evidence of the robbery, which Burdine admitted committing) not harmful to his interests was being presented. Because the presumed-prejudice claim was not raised until 11 years after trial (as well as because he withheld evidence), it is impossible to identify accurately, on this record, whether those periods of inactivity reflect trial strategy or that, instead, Cannon was asleep.

This uncertainty is demonstrated by the following example. As part of his ineffective-assistance claim, Burdine asserted that Cannon's homophobic views created an unconstitutional conflict of interest, and that his homophobia adversely affected his performance at trial; Burdine also complained about Cannon's failure to object to the prosecutor's homophobic arguments and similar homophobic misconduct at trial. For example, to support his presumed-prejudice claim, he notes Cannon did not object when Burdine was asked "whether he 'voluntarily' remained in the 'homosexual lifestyle'" and whether, while engaging in homosexual sex, "he played the role of 'man' or 'woman'". But, part of Cannon's defense strategy was to portray Burdine as a victim of the murder victim, Wise, an older man who, in several ways, had taken great advantage of the much younger Burdine during their homosexual relationship (deposited Burdine's pay checks into his (Wise's) account; spent Burdine's money; attempted to persuade Burdine to prostitute himself). In short, such non-objection could well have been part of Cannon's trial strategy and not because he was asleep. On this record, we do not, and cannot, know.

But, if Burdine's contention that such questions are extremelyegregious and so objectionable is to be accepted, then it seems obvious Burdine, as well as the trial judge, would have looked to Cannon when the questions were asked, expecting an objection. If Cannon had been asleep, they would have noticed it. And, as he testified at the state habeas hearing, the trial judge would have done something about it.

As noted by the majority, it was claimed somewhat recently inUnited States v. Russell, 205 F.3d 768 (5th Cir. 2000), that the taking of any evidence at trial in the absence of counsel is prejudicial per se under Cronic. Our court stated it did "not so hold", declining to fashion such a rule. Id. at 771. (Teague was not addressed in Russell.) See also Vines v. United States, 28 F.3d 1123, 1128 (11th Cir. 1994) (rejecting defendant's contention that taking of evidence was necessarily critical stage of trial and refusing presumed-prejudice when no evidence directly inculpating defendant was presented while counsel temporarily absent).

As discussed in Russell, although "Cronic does not provide significant guidance on which parts of trial are considered 'critical'", 205 F.3d at 771, it does provide some guidance for determining whether counsel's absence is at such a stage:

First, there must be a denial of such significance that it makes the adversary process itself unreliable.... Second, theCronic court makes clear that "only when surrounding circumstances justify a presumption of ineffectiveness can a Sixth Amendment claim be sufficient without inquiry into counsel's actual performance at trial."

Id. (quoting Cronic, 466 U.S. at 662; emphasis in original); see also United States v. Taylor, 933 F.2d 307, 312 (5th Cir.) ("critical stages ... are those stages of the proceeding at which the substantial rights of a defendant may be affected"), cert. denied, 502 U.S. 883 (1991); United States ex rel. Thomas v. O'Leary, 856 F.2d 1011, 1014 (7th Cir. 1988) ("A critical stage is one where potential substantial prejudice to [a] defendant's rights inheres in the particular confrontation and where counsel's abilities can help avoid that prejudice.").

In Russell, our court concluded that Russell's counsel's actual absence from the courtroom (illness) was during a "critical stage": the Government presented evidence implicating several of his co-conspirators, although not directly implicating Russell. Russell, 205 F.3d at 770-72. Under such circumstances, "[t]he adversary process becomes unreliable when no attorney is present to keep the taint of conspiracy from spreading to the client". Id. at 772. But, unlike in Russell, where the evidence presented during counsel's absence was easily identifiable, we cannot determine from the trial or state habeas records what evidence was being presented, or other activity was taking place, while Cannon slept.

In rejecting the State's contention that Burdine cannot prove Cannon slept during a critical stage because it is impossible to determine when the sleeping occurred, the majority states it is notnecessary for the defendant to explain how having counsel would have altered the outcome. Maj. Op. at 23. (Obviously, this would be tantamount to requiring proving actual prejudice in order to receive presumed-prejudice.) As support, it cites Russell, noting that our court did not require Russell to demonstrate that the evidence adduced against his co-defendants during counsel's absence adversely impacted his defense or that the presence of his counsel would have improved his chance of acquittal. Maj. Op. at 24. The majority acknowledges that we do not know what specific evidence was being presented while Cannon slept; nevertheless, it notes -- pursuant to the state-finding -- that the evidence was being presented by the State against Burdine, and posits this "at the very least inferentially increased the taint of Burdine's guilt because he was the only defendant". Maj. Op. at 24 n.7 (emphasis in original).

But, the defendant in Russell, unlike Burdine, easily demonstrated that his counsel was actually absent at a criticalstage of the trial, when evidence was being presented against his co-conspirators which increased the taint of his guilt of conspiracy. Thus, there was no need in Russell for our court to speculate whether the evidence being presented in counsel's absence was potentially harmful to the defendant's interests. In short, our court did not require Russell to prove counsel's presence would have affected the outcome because it held prejudice must be presumed.

Burdine had the burden of proving Cannon was absent, by sleeping, during "critical stages" of trial. Because Burdine cannot demonstrate when Cannon slept (or, by withholding evidence, refused to attempt to do so), he has not shown it was at a "critical stage". The majority does not even discuss the facts of Burdine's case, much less the crucial point that, because Burdine admitted robbing Wise, the State's evidence of the robbery was uncontested by Burdine. Instead, the majority concludes that Burdine has established presumed-prejudice merely by demonstrating Cannon slept (characterized as "repeatedly unconscious") during some unidentified, "not insubstantial" portions of the guilt-innocence phase.

Obviously, I agree with the majority that a defendant need not prove actual prejudice in order to establish entitlement to presumed-prejudice. On the other hand, he must at least prove the existence of the circumstances warranting that presumption, i.e., the absence of counsel at a critical stage. See Triana v. United States, 205 F.3d 36, 43 (2d Cir.) ("Application of the per se rule requires proof of the relevant circumstance, not speculation that it might be true."), cert. denied, 121 S. Ct. 378 (2000). Such proof is lacking in this case.

The majority asserts that, as in Russell, it declines to adopt a per se rule of presumed-prejudice for "any dozing" by defense counsel during trial. Maj. Op. at 28. Nevertheless, despite its disclaimer, it has, in effect, adopted a rule that the entire guilt-innocence phase of a capital murder trial is a critical stage, warranting presumed-prejudice whenever counsel is "unconscious" during unidentified "not insubstantial" portions of it, irrespective of whether the evidence being presented while counsel slept was harmful to the defendant's interests, or whether counsel could have done anything to improve the defendant's circumstances had he been alert. Pursuant to the majority's rule, any sleeping by counsel during the guilt-innocence phase of a capital murder trial mandates presumed-prejudice. This is flatly inconsistent with our court's refusal to adopt a similar rule inRussell, and underscores that the majority's rule is "new" within the meaning of Teague.31

As discussed supra, because of Burdine's admission that he was guilty of robbing the victim and was present when the murder took place, much of the evidence presented by the State, such as photographs of the items taken during the robbery, and evidence that Burdine withdrew money from the victim's bank accounts at automatic teller machines following the murder, was not contested by Burdine. The sleeping may have taken place during the presentation of that evidence. We simply cannot tell from the record, because Burdine waited nearly 11 years to bring the claim, after memories had faded, making it impossible to identify when the sleeping occurred. Of course, this uncertainty is greatly exacerbated by Burdine's withholding evidence which, as stated, the majority does not discuss. By holding that prejudice must be presumed in these circumstances, the majority has overruled Russelland established a new rule.

Under these circumstances, and considering Teague's goals of finality and comity, an inquiry in this case into actual prejudicevel non is more than worth the cost in determining that question. And, of special importance, because neither the prosecutor nor the trial judge was aware of Cannon's sleeping, it could not have been easily prevented by the State. Accordingly, Burdine seeks, and the majority applies, a "new rule" within the meaning of Teague. Therefore, unless his claim meets one of the narrow exceptions to the Teague nonretroactivity principle, we are barred from considering it.

"Teague provides that a new constitutional rule can apply retroactively on federal collateral review only if the new rule (1) puts certain kinds of primary, private conduct beyond the power of the criminal law-making authority to proscribe or (2) is a rule of procedure that is implicit in the concept of ordered liberty." Fisher, 169 F.3d at 306 (internal quotation marks and citation omitted). In addition, our court has recently adopted a third narrow exception.

a.

Burdine seeks, inter alia, shelter within the second exception -- "a rule of procedure that is implicit in the concept of ordered liberty". This exception is "reserved for watershed rules of criminal procedure that implicate the fundamental fairness and accuracy of the proceeding". Id. Burdine asserts that both elements of that exception are violated when a capital defendant is denied assistance of counsel during a substantial portion of trial.

Under the circumstances of this case, including the claim'snot being presented until nearly 11 years after trial, Burdine's withholding evidence, and the impossibility of identifying the portions of trial during which Cannon slept, it is not necessary to create a new rule of presumed-prejudice in order to promote fundamental fairness and ensure an accurate determination of guilt or innocence or punishment. Those goals can be achieved satisfactorily -- and with far greater assurance of accuracy -- under the Strickland actual prejudice analysis. See Tippins, 77 F.3d at 686 ("Ordinarily, episodes of inattention or slumber are perfectly amenable to analysis under the Strickland prejudice test."). That is especially true here, where: Burdine, who sat beside Cannon throughout trial, has neither stated in an affidavit nor testified that he observed Cannon sleeping (indeed, he even withheld evidence on this point); and the witnesses' testimony at the evidentiary hearing, regarding the amount of sleeping and when it occurred, cannot be corroborated by reviewing the trial transcript.

b.

Burdine also claims an exception to Teague for constitutional rights susceptible of vindication only on habeas review, asserting that his presumed-prejudice claim could not have been raised on direct appeal because he was represented by the same counsel who slept during his trial (even though Burdine requested that Cannon represent him on appeal) and that, in any event, the claim required development of facts outside the trial record. Subsequent to oral argument before the panel, our court, in Jackson v. Johnson, 217 F.3d 360, 364 (5th Cir. 2000), adopted a somewhat similar, quite narrow, third Teague exception.

The Texas intermediate appellate court affirmed Jackson's conviction for aggravated assault. Jackson neither filed a timely motion for rehearing with that court nor sought discretionary review by the Texas Court of Criminal Appeals. Id. at 363. On habeas review, Jackson claimed his attorney rendered ineffective assistance by failing to timely file a motion for rehearing with the intermediate appellate court. Id. at 361, 363.

Our court concluded that holding an "opportunity to file a motion for rehearing should be considered the last step in [Jackson's] first appeal of right ... would surely create a new rule" under Teague. Id. at 363-64. But, it held Jackson's claim satisfied "a third narrow exception to Teague, heretoforeunrecognized by the courts". Id. at 364. "When an alleged constitutional right is susceptible of vindication only on habeas review, application of Teague to bar full consideration of the claim would effectively foreclose any opportunity for the right ever to be recognized". Id.32

Arguably, the right asserted by Jackson was one that could never be raised on direct appeal. Id. at 364. In any event,Jackson must, at the very least, be limited to its facts, so that it does not swallow the rule announced in Teague. The holding inJackson has obvious, wide-ranging implications concerning the limits mandated by Teague for habeas review, including whether the new exception itself is Teague-barred. See note 32, supra. TheJackson exception has not been applied since Jackson was decided in July 2000, although it was cited in Clark v. Johnson, 227 F.3d 273, 283 n.4 (5th Cir. 2000), cert. denied, 121 S. Ct. 1129 (2001). Cf.Soffar v. Johnson, 237 F.3d 411, 450, 452 (5th Cir. 2000) (citingJackson, but referring to "either of the two exceptions to nonretroactive applicability"), vacated, 253 F.3d 227 (5th Cir. 2001).

Moreover, to accept Burdine's contention that the Jacksonexception applies because he could not raise his claim on direct appeal (because Cannon was also his appellate counsel) could lead to great procedural abuse. A defendant will know his lawyer was asleep during trial, but may hope nevertheless to receive a favorable verdict, and therefore does not want a new trial. Accordingly, he decides to take no action at trial about the sleeping. If the verdict is unfavorable, the defendant may remain silent about the sleeping and, as in this case, have that lawyer represent him on appeal (who better to do so?), the defendant still hoping to prevail and, therefore, post-trial, not bringing up the sleeping for fear it might result in an unwanted new trial. But, if he loses on appeal, the defendant on habeas can finally raise -- and under the majority's new rule perhaps receive a new trial on -- the matter about which he was aware -- and indeed permitted -- when it occurred years before at trial -- his lawyer's sleeping, about which he then said not a word! Obviously, this totally undermines the goal of finality.

Some might view this as an extreme, perhaps absurd, scenario. Perhaps so. More likely not. See Tippins, 77 F.3d at 688 (noting "[l]awyers may sometimes affect a drowsy or bored look to downplay an adversary's presentation of evidence" and "a per se rule would 'give ... unscrupulous attorneys a delayed-trigger weapon to be sprung at some later strategic phase of the proceeding if events developed very badly for a defendant'" (quoting People v. Winkler, 71 N.Y.2d 592, 598, 528 N.Y.S.2d 360, 363, 523 N.E.2d 485, 488 (1988))); Prada-Cordero v. United States, 95 F. Supp. 2d 76, 81-82 (D.P.R. 2000) ("court should be cognizant that attorneys may use the appearance of sleep as a strategic tool to downplay the importance of an adversary's presentation"; "[m]oreover, a rule that required a finding of prejudice whenever an attorney slept during a trial would provide unscrupulous practitioners with a safety valve to annul trials that they feel they are at risk of losing").

This is what will be permitted by presumed-prejudice based on this record. Courts already have concerns about some of the tactics utilized by criminal defendants and their counsel, especially in capital cases. Why would this court add to the avenues for abuse of court processes unless there is a clear-cut, compelling, constitutionally-required reason to do so? On this record, that reason is not before us.

Burdine requested that Cannon be appointed to represent him on appeal. On 26 March 1984, approximately two months after he was sentenced, he wrote to the trial judge asking, unsuccessfully, that Cannon be replaced; nevertheless, he then stated: "My family and I still feel Mr. Cannon done [sic] the best he could during the trial". This was consistent with his earlier praise to Cannon. In a letter dated 30 January 1984, the day the jury rendered its verdict in the penalty phase, Burdine wrote to Cannon: "with what little defense we had to work with, I don't feel in my heart that you could have done a better job defending me". A week later, he again wrote to Cannon: "again let me emphasize my satisfaction in your representation of me during the trial.... [W]ith what little we had to work from, I expected a sentence of this nature, although a life sent[ence] would have been more relaxing".

On this record, especially in the light of the recent admission, Burdine's request that Cannon represent him on appealmay well constitute a forfeiture or waiver of Burdine's presumed-prejudice claim. In the now vacated panel majority opinion, we concluded that this request did not constitute a waiver. 231 F.3d at 957. But, that was before Burdine's habeas counsel's subsequent admission to our en banc court that he withheld evidence: that, during trial, Burdine nudged Cannon. This amply demonstrates Burdine knew -- as would anyone -- that Cannon should not be sleeping, that he should be participating. Nevertheless, despite this knowledge, Burdine voluntarily asked for Cannon to represent him on appeal. This is a classic case of forfeiture or waiver of this, and possibly other, ineffective assistance claims tied to Cannon's sleeping. See United States v. Olano, 507 U.S. 725, 733 (1993) ("Waiver is different from forfeiture. Whereas forfeiture is the failure to make the timely assertion of a right, waiver is the intentional relinquishment or abandonment of a known right." (internal quotation marks and citations omitted)). Again, this recent admission has totally changed the landscape for this presumed-prejudice claim.

Although the majority apparently is not concerned about this withheld-evidence, much less its effect, I am troubled greatly, to say the least, by wide-ranging abuses that can result where, as here, a criminal defendant sits next to counsel during trial; makesno mention then of counsel's sleeping; requests that the same counsel represent him on direct appeal; and then, over ten years after trial, claims presumed-prejudice because counsel slept during trial, despite defendant's never, by affidavit or testimony, stating under oath counsel engaged in such conduct. Obviously, the recent extra-judicial admission is even more troubling.

Burdine contends, alternatively, that, on direct appeal, he could not have asserted the claim now at issue because it required development of facts outside the trial record. But, Burdine sat next to Cannon throughout trial. Based on his recent admission, Burdine was certainly aware at trial of Cannon's sleeping and, other than nudging Cannon, took no action then to try to correct it. In the light of his concern, he likewise could have brought Cannon's sleeping to the attention of the trial court during trial, which may, at the very least, have allowed development of the facts at that time. As noted, the trial judge testified at the state habeas hearing he would have stopped the trial if he had noticed Cannon sleeping. Or, Burdine could have otherwise advised the trial judge or a family member. Along this line, perhaps Cannon's sleeping could have been the basis for a new trial.

Therefore, especially in the light of the recent withheld-evidence admission, the quite narrow Teague-exception adopted inJackson is not applicable: Burdine's claim is not one that could never be raised on direct appeal under any circumstances. Restated, the alleged constitutional right he claims is not "susceptible of vindication only on habeas review". Jackson, 217 F.3d at 364 (emphasis added).

The majority's short-lived "new" rule does not fall within any exception to Teague. But, even assuming the rule is not new, Burdine still is not entitled to presumed-prejudice under the facts and circumstances of this case.

The majority holds a constructive denial of counsel occurs when defense counsel is "repeatedly unconscious through not insubstantial portions" of the guilt-innocence phase of a capital murder trial, warranting presumed-prejudice. Again, there is no state-finding that Cannon was "repeatedly unconscious" for either "substantial" or "not unsubstantial" portions; the Court of Criminal Appeals expressly rejected the trial court's recommended conclusion, which used the word "substantial" in describing Burdine's "allegation"; and the only presumptively correct, and therefore binding, somewhat specific state-finding is that Cannon "dozed and actually fell asleep during portions of [Burdine's] trial on the merits, in particular during the guilt-innocence phase when the state's solo prosecutor[] was questioning witnesses and presenting evidence". Ex parte Burdine, No. 379,444-B, at 13.

As discussed supra, the Supreme Court has articulated three reasons for presuming, rather than requiring proof of, prejudice for actual or constructive denial of counsel: (1) "[p]rejudice in these circumstances is so likely that case-by-case inquiry into prejudice is not worth the cost", Strickland, 466 U.S. at 692; (2) "such circumstances involve impairments of the Sixth Amendment right that are easy to identify", id.; and (3) where "the prosecution is directly responsible", such circumstances are "easy for [it] to prevent", id.; see also Robbins, 528 U.S. at 287 & n.15; Cronic, 466 U.S. at 658.

In the light of the circumstances of this case, none of those justifications supports presumed-prejudice. First, as the state court found, neither the prosecutor nor the trial judge was aware of Cannon's sleeping. And, although Burdine's counsel now maintains Burdine nudged Cannon during trial, Burdine at trial (and counsel on habeas) withheld that evidence. Accordingly, the State could not have easily prevented the sleeping. Thus, the basis on which Tippins was distinguished in Morales v. United States, 143 F.3d 94, 97 (2d Cir. 1998) ("[a] lawyer asleep in open court has abandoned his client in front of the judge and the prosecutor; his ineffectiveness and helplessness are therefore both 'easy to identify' and 'easy for the government to prevent'" (citation omitted)), is not applicable in the case at hand.

Second, the claimed prejudice is not easy to identify, just the opposite. The claim was not raised until over ten years after trial, after it was first raised by another death row inmate. Therefore, a determination of precisely when counsel slept has been rendered impossible due to the passage of time and the lack of any indication in the trial transcript, discussed infra, as to when the conduct occurred. Nor can it be determined from the witnesses' testimony at the state evidentiary hearing. Accordingly, it is impossible to determine whether, for example, Cannon slept during the presentation of crucial, inculpatory evidence, or during the introduction of unobjectionable, uncontested evidence.

Finally, for circumstances where, as here, counsel sleeps for unidentified portions of a trial, prejudice is not so likely that case-by-case inquiry into prejudice is not worth the cost. Again, the majority states that its rule "is limited to the egregious facts found by the state habeas court in this case". Maj. Op. at 28. But, as noted, the state habeas court made no finding that Cannon's dozing and sleeping rose to the level of unconsciousness, and, in any event, no finding quantifying the amount of sleeping or what evidence was then being presented. Had the state habeas court actually found that Cannon was repeatedly unconscious during a substantial portion of trial, presumed-prejudice might well be warranted.

In the absence of such findings, however, the only way to determine whether the facts actually found by the state habeas court are "egregious" is to conduct a detailed, painstaking examination of the record. In circumstances such as these, where the sleeping is not noticed by the trial judge or prosecutor, and where there is no way to easily and quickly determine from either the trial or post-conviction record when the sleeping occurred, and where there are no presumptively correct state-findings regarding the soundness or depth of the sleep or when it occurred, the majority's rule cannot be applied without closely examining the record. Otherwise, how can a reviewing court determine either that the sleep was deep enough to reach a state of "unconsciousness", or that it occurred for "not insubstantial portions" of the trial, or both?

Again, such examination precludes presumed-prejudice, because "once it is necessary to examine the trial record in order to evaluate counsel's particular errors, resort to a per sepresumption is no longer justified by the wish to avoid the cost of case-by-case litigation". Scarpa v. DuBois, 38 F.3d 1, 14 (1st Cir. 1994) (refusing presumed-prejudice where defense counsel's argument effectively conceded only disputed elements of charged crimes), cert. denied, 513 U.S. 1129 (1995).

[T]here are real dangers in presuming prejudice merely from a lack of alertness. Prolonged inattention during stretches of a long trial (by sleep, preoccupation or otherwise), particularly during periods concerned with ... uncontested issues, or matters peripheral to a particular defendant, may be quantitatively substantial but without consequence. At such times, even alert and resourceful counsel cannot affect the proceedings to a client's advantage.

Tippins, 77 F.3d at 686. Tippins observed, however, that prejudice becomes "'inherent' at some point, 'because unconscious or sleeping counsel is equivalent to no counsel at all'". Id. (quoting Javor, 724 F.2d at 834). The court concluded "that Tippins suffered prejudice, by presumption or otherwise, if his counsel was repeatedly unconscious at trial for periods of time in which defendant's interests were at stake". Id. at 687. But, again, there are no state-findings in this case that Cannon was "repeatedly unconscious" for periods of the trial "in which [Burdine's] interests were at stake".

For many cases allowing presumed-prejudice, the circumstances justifying that presumption are quickly, easily, and clearly discernible. See, e.g., Geders, 425 U.S. at 91 (court prevented defendant from consulting with counsel during overnight recess between defendant's direct and cross-examination); Davis v. Alaska, 415 U.S. 308, 318 (1974) (presumed-prejudice where defendant denied right of effective cross-examination); Hamilton, 368 U.S. at 55 (defendant denied counsel at arraignment); Hughes v. Booker, 220 F.3d 346, 352 (5th Cir. 2000) (attorney withdrew from representation of defendant on appeal without filing sufficient brief); Russell, 205 F.3d at 770-72 (testimony implicating defendant in conspiracy presented during counsel's two-day absence due to illness); Blankenship v. Johnson, 118 F.3d 312, 317 (5th Cir. 1997) (appointed counsel did "nothing whatsoever" on state-requested discretionary appeal).33

Moreover, and as should be quite obvious by now, the presumed-prejudice exception is an extremely narrow one. See Childress, 103 F.3d at 1229 ("constructive denial of counsel as described inCronic affords only a narrow exception to the requirement that prejudice must be proved"); Craker, 805 F.2d at 542 ("A constructive denial of counsel occurs ... in only a very narrow spectrum of cases where the circumstances leading to counsel's ineffectiveness are so egregious that the defendant was in effect denied any meaningful assistance at all." (internal quotation marks and citation omitted)); Hollenback v. United States, 987 F.2d 1272, 1275 (7th Cir. 1993) ("cases in which an inherently prejudicial constructive absence of counsel has been found involve particularly egregious conduct that is the functional equivalent of actual absence of counsel").

Prejudice has not been presumed for claims of denial of effective-assistance due to counsel's alleged impairment because of alcohol, drug use, or a mental condition. See, e.g., Burnett v. Collins, 982 F.2d 922, 928-30 (5th Cir. 1993) (alcohol abuse);Berry v. King, 765 F.2d 451, 454 (5th Cir. 1985) (addiction to illegal drugs), cert. denied, 476 U.S. 1164 (1986); Buckelew v. United States, 575 F.2d 515, 521 (5th Cir. 1978) (poor health);Dows v. Wood, 211 F.3d 480, 485-86 (9th Cir.) (Alzheimer's disease), cert. denied, 121 S. Ct. 254 (2000); Smith v. Ylst, 826 F.2d 872, 875-76 (9th Cir. 1987) (mental illness), cert. denied, 488 U.S. 829 (1988); Hernandez v. Wainwright, 634 F. Supp. 241, 245 (S.D. Fla. 1986) (intoxication during trial), aff'd, 813 F.2d 409 (11th Cir. 1987).

The majority distinguishes those cases on the basis that, unlike a drunk or drugged lawyer, who "exercises judgment, though perhaps impaired" (to say the least; such "impaired judgment" may well be worse than none at all), an unconscious lawyer (according to the majority, because he is sleeping) is not capable of exercising any judgment. Maj. Op. at 27-28. The majority maintains such an unconscious attorney is no different from one who is physically absent. The flaw in its analysis is that it assumes, without any state-finding or record evidence, that Cannon was always so deeply and soundly asleep that he was always "unconscious".34 As also discussed, the state-finding that Cannon "dozed and actually fell asleep" does not support the majority's characterizing each episode as "unconsciousness". In any event, as discussed infra, and in the context of Cannon's otherwise quite meaningful assistance to Burdine, Burdine was not denied counsel.

Javor and Tippins do not fit comfortably within the framework of cases in which prejudice has been presumed. Nevertheless, as discussed supra, both are distinguishable from the circumstances at hand. In Javor, there was evidence that counsel slept during a substantial portion of trial; counsel failed to participate when evidence against the defendant was being presented; counsel had stated to counsel for a co-defendant that he had missed some of the testimony; other counsel often "nudged" and "kicked" counsel to awaken him; and the judge was at times concerned about counsel's inattentiveness. Javor, 724 F.2d at 833-34. In Tippins, there was evidence that Tippins' counsel slept every day of the trial, including during the testimony of a critical prosecution witness and during damaging testimony by a co-defendant; and the sleeping was noticed by both the trial judge and prosecutor, as well as jurors and witnesses, some of whom even heard him snoring. Tippins, 77 F.3d at 687-89.

In contrast, Cannon's sleeping was not nearly so obvious. His sleeping, although noticed by the court clerk and several jurors, went unnoticed by the trial judge and prosecutor. As detailedinfra, the trial transcript reflects few long passages without some activity by Cannon. Again, despite Burdine's federal habeas counsel's admitting, at en banc oral argument, that he withheld evidence that Burdine at times nudged Cannon during trial to try to awaken him, Burdine, who sat next to Cannon throughout trial, has never stated by affidavit or testified that he observed Cannon sleeping or dozing during trial. In fact, as also noted, at the conclusion of trial Burdine requested that Cannon be appointed to represent him on direct appeal and, post-trial, was very complimentary of Cannon's representation of him at trial.

Some of the evidence presented against Burdine, such as photographs of the items taken from the murder/robbery victim, was not contested and, in the light of Burdine's admission that he participated in the robbery, but not the murder, was not harmful to his interests. At this stage, we simply do not know, and cannot ever know on this record as it now stands, whether the sleeping occurred during the presentation of that evidence, or whether it occurred when other, inculpatory, evidence was being presented. To presume prejudice under these circumstances requires not only the creation of a new rule, but also requires unbridled speculation, with no possibility of accuracy. Again, it is more than worth the cost to determine whether there was actual prejudice. This could be done on remand.

The following review of the record on appeal demonstrates that presumed-prejudice for Burdine is bottomed on speculation, rather than, as required, on extremely easy to identify instances where prejudice is extremely likely.35 Again, this case should be remanded for the district court to conduct a review for actual prejudice vel non, as well as to consider the numerous other issues raised by Burdine.

Voir dire took seven days. The balance of the trial lasted six: three for presentation of evidence at the guilt-innocence phase; one during which the jury was charged, closing arguments were made, and the jury reached its verdict for that phase; one for presentation of evidence in the penalty phase; and one during which the jury was charged, closing arguments were made, and the jury reached its verdict for that phase.

Strickland, the jury foreman, testified at the state habeas evidentiary hearing that: on several occasions at trial, Cannon appeared to "nod off" or "doze"; he noticed the dozing "more than two, but maybe not more than five times"; the dozing occurred during the guilt-innocence phase, typically in the afternoon, after the lunch recess, when witnesses were being questioned or other evidence was being presented; and the episodes were for 30 seconds or less.

Davis, another juror, testified at the hearing that she noticed Cannon repeatedly falling asleep during "quite a bit" of the guilt-innocence phase, especially during the afternoons of the second and third days of testimony.

Another juror, Engelhardt, testified that: on five or ten occasions, covering both phases of trial, he noticed Cannon "nodding" or "dozing"; and, on one occasion, Cannon had his eyes closed and his head bowed for at least ten minutes. But, Engelhardt could not remember what was occurring at the time of the incidents or whether they were in the morning or afternoon.

Berry, a court clerk who assisted the trial judge, testified that: she witnessed Cannon sleeping "a lot" and "for long periods of time" during questioning of witnesses; the longest instance was at least ten minutes; and there were "lots of incidents" when Cannon dozed for shorter periods.

As discussed below, an obvious possible critical stage would be the ten-minute period, or periods, during which, according to two witnesses, Cannon slept; but, as noted, that period, or those periods, cannot be tied to a particular point during the trial. Examining the trial transcript and minutes in conjunction with the above-described testimony, it is impossible to determine when Cannon slept, much less whether he did so during the presentation of contested, inculpatory evidence. Burdine concedes this.

The presentation of evidence commenced at 10:50 a.m. on Monday, 23 January 1984. The State's first witness, a homicide detective, testified about his investigation of the murder and the discovery of the victim's body. Direct examination was completed when the court recessed at noon for lunch, and covers pages 27-80 of the transcript. The transcript and minutes reflect the following activity involving Cannon during that direct examination: he objected (page 43); the jury retired while the court reporter marked 34 exhibits (pages 45-46); Cannon requested time to examine photographs and, at 11:22 a.m., outside the presence of the jury, made objections to some of the exhibits (pages 50-53); the jury returned at 11:29 a.m. (page 54); Cannon objected (pages 69-70); and Cannon questioned the witness on voir dire (pages 73-79).

Trial resumed at 1:30 p.m., Cannon's cross-examination of the homicide detective covering pages 83-90 of the transcript. The State's redirect is at pages 90-98, with Cannon speaking on the record at 91, 92, 95, and 97. Cannon's recross is at 99-101; further redirect, at 101-02.

The State's next witness, the medical examiner, testified regarding the victim's wounds and cause of death. For the direct examination, which covers pages 103-19, Cannon objected at 109 and 118; his cross-examination covers pages 119-23. The State's redirect covers 123-29, interrupted at 126-27 by a bench conference requested by Cannon. Cannon conducted recross at page 130. A bench conference outside the hearing of the court reporter was conducted at page 131.

The State's final witness for the first day of testimony was a detective, who testified about tracing the victim's stolen gun. The direct examination covers pages 131-37; Cannon's cross, 137-38. After the jury was excused for the day (2:57 p.m.), the trial judge began the previously discussed suppression hearing on the admissibility of Burdine's confession. That hearing began at 3:37 p.m., and covers pages 141-86. The first witness, called by the State, was a detective who went to California and participated in obtaining Burdine's confession. Direct examination covers pages 141-62, interrupted by Cannon's objections at 143-44, and 149; Cannon's cross-examination, 162-66; and further questioning by the court, 166-67.

After the State announced it had no other witnesses for the suppression issue, Cannon advised the court, at 168, that he intended to call Burdine for the limited purpose of testimony regarding the confession. Burdine's direct examination covers pages 169-80; cross, 180-84. Cannon announced he had nothing further to present, at 184-85, and the hearing was recessed for the day.

The suppression hearing continued the following morning, Tuesday, 24 January, and covers pages 190-203. The State called another detective who participated in obtaining Burdine's confession. Direct examination covers pages 191-93; cross by Cannon, 193-94. Cannon recalled Burdine as a witness, at 194. Direct examination covers pages 195-96; there was no cross-examination. Cannon presented argument on the motion to suppress at 197-99; the State, 199-201. After the court announced its findings and conclusions, at 201-02, Cannon stated, at 203, he had nothing further to present in connection with the motion.

Following completion of the suppression hearing that morning, the State's presentation of evidence resumed (10:17 a.m.). The State's first witness was the manager of the pawn shop where, after the murder, Burdine pawned a ring. Direct examination covers pages 204-10; cross, 210-12.

The next witness on this second day of testimony, a bank security administrator, authenticated a tape showing Burdine, after the murder, withdrawing money from an automatic teller machine. Direct examination covers pages 212-19, and includes showing the tape to the jury. At page 216, Cannon stated he had no objection to the tape; at 220, no questions for that witness.

The State's next witness was an automatic teller machine coordinator for a credit union where, after the murder, Burdine withdrew money from the victim's account. The direct examination covers pages 220-23. At 223, Cannon stated he had no questions for the witness, and a bench conference was conducted outside the hearing of the court reporter. At pages 224 and 226, Cannon stated that he had no objections to the tape of that transaction.

The State's next witness was the owner of the California shop where, post-murder, Burdine sold the victim's gun. The direct examination covers pages 226-36, interrupted by Cannon's objections at 229 and 235. At 236, Cannon stated he had no questions for the witness. At 237, a bench conference was conducted outside the hearing of the court reporter.

The State recalled its first witness from the first day of testimony, the homicide detective. He testified about Burdine's confession in California. Direct examination was conducted until the court recessed for lunch at 11:53 a.m. (page 257), and continued when trial resumed at 1:48 p.m.; it covers pages 237-63. For the pre-noon-recess testimony, covering 237-57, the record reflects activity by Cannon at 255 and 256; following the recess, at 259-60 and 262. Cannon's cross-examination appears at 264-75. The State's redirect, at 276-81, was interrupted by an objection by Cannon (page 278). Cannon's recross covers 281-83.

The State's next witness, a receptionist at the security service where both the victim and Burdine worked, testified regarding Burdine's use of an alias during that employment. Direct examination covers pages 283-90; at 287, Cannon stated he had no objection to an exhibit. His cross-examination covers 291-92.

The next witness was the California detective who arrested Burdine. Direct examination covers pages 293-302. Cannon requested a bench conference, and the court was in recess from 2:46 until 3:24 p.m. Cannon cross-examined the witness on page 304, after which another bench conference was conducted.

The final witness for the second day of testimony was the victim's roommate, who testified about his relationships with the victim and Burdine, his discovery of the victim's body, and various items of property taken in the robbery. Direct examination covers 305-37, interrupted by one objection by Cannon at 316.

During that direct examination, and at the request of a juror, court was in recess from 4:05 until 4:16 p.m. Thereafter, Cannon cross-examined the witness at pages 337-49. The State's redirect covers pages 349-52; Cannon's recross, 353-54. After the State rested at 4:31 p.m., the jury was excused for the day.

On Wednesday, 25 January, the third day of testimony, Cannon moved for an instructed verdict (judgment of acquittal) outside the presence of the jury. At 10:17 a.m., when the jury returned, the State re-opened to offer exhibits, and again rested. Burdine testified on direct examination at pages 363-418. (As an example of inactivity not equating with sleep, the prosecutor was silent for 36 pages during Burdine's direct examination, except for a brief request for some testimony to be read back.) Burdine's cross-examination, at 419-90, was interrupted by a bench conference, at page 485, and the noon recess, which lasted from 12:38 until 2:20 p.m. Only the last five pages of the State's cross-examination were conducted following the recess. Cannon conducted redirect of Burdine at pages 491-95. Burdine rested at 2:33 p.m.

In rebuttal, the State called another detective, who testified regarding Burdine's confession in California. The State's direct examination covers pages 496-512, interrupted by objections by Cannon at 504, 507, and 508-09. Cannon's cross-examination covers 513-15. Following a very brief redirect examination by the State, Cannon requested a bench conference (pages 517-18), which lasted from 2:56 until 3:12 p.m., following which the State rested. Cannon then called one additional witness. After both sides rested at 3:15 p.m., the jury was excused for the day.

The next day of trial (after three days of testimony) began at 10:50 a.m. on Thursday, 26 January. After the charge was read to the jury, the State presented its initial closing argument from 11:02 until 11:37 a.m.; Cannon presented closing argument from 11:37 a.m. until 12:07 p.m.; and the State presented its final closing argument from 12:07 until 12:15 p.m., following which the jury retired to deliberate. The jury returned a guilty verdict at 2:10 p.m., and court recessed for the day.

Penalty phase proceedings began at 10:30 a.m. the next day, Friday, 27 January. The State presented the testimony of four witnesses between 10:30 and 11:05 a.m., covering pages 607-36, and then rested. The record reflects activity by Cannon at 608, 609, 616, 618, 621, 622-23, 624, 625-26, 626, 627, 629, 631, and 635-36. Following a recess, court reconvened at 11:28 a.m., at which time Burdine rested without presenting any evidence. Court was recessed until the following Monday.

On Monday, 30 January, proceedings commenced at 10:40 a.m., when the penalty-phase charge was read to the jury. The State presented argument from 10:45 until 11:15 a.m.; Cannon, 11:15 until 11:44 a.m.; and the State, 11:44 until 11:53 a.m. The jury retired to deliberate and reach a verdict on the penalty. As reflected in this quite detailed examination of the trial transcript, there are very few long stretches of transcript in which no activity by Cannon is reflected. Portions of the transcript reflecting no such activity involve the presentation of contested, inculpatory evidence, as well as uncontested testimony and exhibits, where Cannon's attentive participation was irrelevant to the quality of Burdine's defense.

For example, during direct examination of the first witness for the State, the prosecution introduced 31 crime scene photographs and questioned the witness about each one individually. Although the record reflects no activity by Cannon at 55-65, he had already examined the photographs outside the presence of the jury; he had objections to only six of the photographs; and the court had already overruled those objections.

Following the lunch recess on the second day of trial, the prosecution introduced 46 photographs of property recovered after Burdine's arrest, and questioned the detective about whether each photograph accurately depicted the items found. Those photographs were admitted into evidence without objection. The prosecutor questioned Wise's roommate about much of that same photographic evidence.

Another period during which no activity is reflected for Cannon was the State's cross-examination of Burdine. As noted, for the 72 pages of cross, only five follow the lunch recess, when Cannon's sleeping might have been more likely. As also noted, Cannon's defense strategy was that the acts of violence were committed by McCreight; and that Wise had taken advantage of Burdine in the homosexual relationship, by depositing Burdine's pay checks into his (Wise's) account, spending Burdine's money, and attempting to persuade Burdine to prostitute himself.

Obviously, when a criminal defendant elects to testify, a great number of trial strategy considerations, including frequency of objections during cross-examination, come into play. In fact, these considerations parallel those championed by Burdine's counsel at en banc oral argument in seeking to justify his reason for withholding the crucial evidence of Burdine's nudging Cannon during trial.

That there were a great number of such competing considerations for the trial of this case is especially true, on the facts in this case, with Burdine's admitting being present at the murder. For example, as discussed supra, Cannon did not object when, on cross, Burdine was asked about a possible characteristic of a homosexual relationship; this decision arguably was consistent with his defense strategy. Moreover, to have objected at certain points might have caused the jury to believe Cannon was trying to prevent Burdine from presenting relevant information; on the other hand, Cannon's not objecting might have been understood by the jury as showing Burdine had nothing to hide. Deciding whether to object is, in many instances, simply a classic example of trial strategy. As noted, during Burdine's direct examination, the prosecutor remained silent throughout 36 pages, except for a single request for testimony to be read back.

As the foregoing review of the record demonstrates, it is possible that unobjectionable evidence (or evidence which Cannon was already anticipating) may have been introduced while Cannon slept, without having a substantial effect on the reliability or fairness of Burdine's trial. Notwithstanding Cannon's sleep episodes, he provided at least some -- indeed, as noted, a great deal of -- meaningful assistance to Burdine and more than subjected the prosecution's case to "meaningful adversarial testing". Cronic, 466 U.S. at 659.

Along this line, and as discussed, although he ultimately was unsuccessful, Cannon vigorously contested the admissibility of Burdine's confession. Once the confession was admitted, Burdine's guilt was, for all intents and purposes, firmly established. In the light of Burdine's admissions in that confession and in his subsequent trial testimony that he was present during the robbery, knew it would occur, and witnessed part of Wise's murder, much of the evidence introduced by the State, especially evidence of the robbery, was essentially duplicative. Moreover, Cannon still sought to establish, through Burdine's testimony, that the confession, although already admitted in evidence, was coerced and inaccurate. He continued to press that point on direct appeal, again unsuccessfully.

In sum, Cannon's sleeping during unidentified portions of Burdine's trial did not result in its losing its character as a confrontation between adversaries, nor did it render the trial fundamentally unfair. Because Cannon provided meaningful assistance to Burdine, prejudice vel non to Burdine's defense, resulting from Cannon's sleeping, should be established underStrickland's two-prong test: Burdine should be required to demonstrate a reasonable probability that the outcome of the trial would have been different if, during the periods in which the transcript reflects no activity by Cannon, he had taken some action.

Presumed-prejudice should be rejected out of hand because of the withholding-evidence tactic employed by Burdine's habeas counsel. (Concomitantly, our remand instructions should include for the district court to hold an evidentiary hearing on that tactic.) In the alternative, the claim fails: it is Teague-barred. But, even if the claim is not barred, presumed-prejudice, on this record, cannot be granted. I respectfully dissent.