dissenting:
Respectfully, and with great regret, I dissent. Today a majority of the court impermissibly shrinks the Great Writ, that most “fundamental instrument for safeguarding individual freedom from arbitrary and lawless state action,” Harris v. Nelson, 394 U.S. 286, 290-91, 89 S.Ct. 1082, 1086, 22 L.Ed.2d 281 (1969), and in doing so diminishes us all. The majority applies the wrong legal analysis — whether there was sufficient evidence of guilt without the erroneously admitted taped confession — to arrive at the wrong result — a denial of the writ of habeas corpus to Kamathene Adonia Cooper. The court reaches its own “inescapable conclusion” that the evidence presented against Cooper was “overwhelming” without ever coming to grips with the proper legal issue — what effect the erroneously admitted confession had on the jury at the actual trial at which Cooper was convicted. Application of the correct analysis does not require that we free Cooper, but it surely does require that we afford him a new trial.1
I.
Although the majority states the correct standard of harmless-error review, whether the erroneous admission into evidence and playing to the jury of Cooper’s detailed and devastating taped confession had a “‘substantial and injurious effect in determining *376the jury’s verdict,’ ” Op. at 370, the majority fails to follow the standard in the most basic manner: the majority never once discusses the effect the error had on the jury at the trial that Cooper actually received. Instead, the majority excises Cooper’s taped confession — the evidence upon which Cooper’s trial judge concluded the State’s case “[hjinged”— and considers only the properly admitted evidence. By analyzing only this portion of the evidence the majority reaches its own conclusion that “Cooper murdered Stewart.” Op. at 371. The error in this approach can be summed up in a single, common-sense question: Can an appellate court fairly decide whether an error was harmless without once considering the effect of that error on the verdict of the jury which originally considered the trial evidence? The obvious answer is no. Without considering the effect of the error the most an appellate court can do is retry the defendant on appeal with the remaining non-erroneous evidence. That is’ what the majority does today, and in doing so disregards fifty years of precedent establishing the approach for harmless-error review and adopts a harmless-error approach heretofore unknown at law in any context.
When assessing whether an error is harmless, it simply is not enough for an appellate court to ignore the effect of the erroneously admitted evidence, conduct an independent assessment of the remaining evidence and decide that it is sufficient to establish guilt, or that guilt would be found at another trial. Instead, as every criminal harmless-error case since Kotteakos v. United States, 328 U.S. 750, 66 S.Ct. 1239, 90 L.Ed. 1557 (1946), has held, the inquiry must focus on the trial that actually occurred, and whether “the error had substantial and injurious effect or influence in determining the jury’s verdict” in that trial. Brecht v. Abrahamson, 507 U.S. 619, 637, 113 S.Ct. 1710, 1722, 123 L.Ed.2d 353 (1993) (quoting Kotteakos, 328 U.S. at 776, 66 S.Ct. at 1253). The majority states the proper standard, but its analysis is totally at odds with that standard, because it never considers the effect that the error in this case had on the verdict of the actual jury that convicted Kamathene Adonia Cooper. The majority’s approach thus hypothesizes a jury verdict that never was.
Kotteakos held that under the federal harmless-error statute a court must measure whether the error had a “substantial and injurious effect” upon the jury’s decision. Kotteakos, 328 U.S. at 776, 66 S.Ct. at 1253. In establishing this standard, Kotteakos made the relevant inquiry crystal clear: “it is not the appellate court’s function to determine guilt or innocence. Nor is it to speculate upon probable reconviction and decide according to how the speculation comes out---- Those judgments are exclusively for the jury____” Kotteakos, 328 U.S. at 763, 66 S.Ct. at 1247 (citations omitted). The question is not whether the jurors were “right in their judgment, regardless of the error or its effect upon the verdict. It is rather what effect the error had or reasonably may be taken to have had upon the jury’s decision. The crucial thing is the impact of the thing done wrong on the minds of other men, not on one’s own, in the total setting.” Id. at 764, 66 S.Ct. at 1247-48 (citations omitted) (emphasis added). Thus, if “the error did not influence the jury, or had but very slight effect” the error was harmless. Id. at 764, 66 S.Ct. at 1248. But, if a court cannot conclude “that the judgment was not substantially swayed by the error,” the error is harmful. Id. at 765, 66 S.Ct. at 1248.
Further, in Chapman v. California, which established harmless-error review for constitutional error in criminal cases, the Supreme Court recognized that the harmless-error inquiry must focus on whether “the evidence complained of might have contributed to the conviction.” Chapman v. California, 386 U.S. 18, 23, 87 S.Ct. 824, 827, 17 L.Ed.2d 705 (1967). Thus, in determining that the error in Chapman was not harmless, the Court relied upon the fact that the prosecution “continuously and repeatedly” mentioned the error. Id. at 25, 87 S.Ct. at 829.
This understanding of harmless-error review remains black letter law. As Justice Scalia recently stated for a unanimous Court:
Harmless-error review looks, we have said, to the basis on which “the jury actually rested its verdict.” Yates v. Evatt, 500 U.S. 391, 404 [111 S.Ct. 1884, 1893, 114 *377L.Ed.2d 432] (1991) (emphasis added). The inquiry, in other words, is not whether, in a trial that occurred without the error, a guilty verdict would surely have been rendered, but whether the guilty verdiet actually rendered in this trial was surely unattributable to the error. That must be so, because to hypothesize a guilty verdict that was never in fact rendered— no matter how inescapable the findings to support that verdict might be — would violate the jury-trial guarantee. See Rose v. Clark, 478 U.S. 570, 578 [106 S.Ct. 3101, 3106, 92 L.Ed.2d 460] (1986); id. at 593 [106 S.Ct. at 3114] (Blackmun, J., dissenting); Pope v. Illinois, 481 U.S. 497, 509-510 [107 S.Ct. 1918, 1925-26, 95 L.Ed.2d 439] (1987) (Stevens, J., dissenting).
Sullivan v. Louisiana, 508 U.S. 275, 279-80, 113 S.Ct. 2078, 2081-82, 124 L.Ed.2d 182 (1993) (parallel citations omitted).
It is true that the Supreme Court directed in Brecht that when assessing whether á constitutional trial error is harmless on habeas review, a court applies the somewhat less stringent evidentiary standard of Kotteakos instead of the Chapman direct review standard. Compare Brecht, 507 U.S. at 638, 113 S.Ct. at 1722 (error harmless if it had no “substantial and injurious effect” on the jury’s verdict), with Chapman, 386 U.S. at 24, 87 S.Ct. at 828 (error harmless if clear “beyond a reasonable doubt” it' had no effect on the jury’s verdict). But, in Brecht the Supreme Court certainly did not hold that federal courts should suddenly abandon the clear teaching of Kotteakos, Chapman, and every other type of harmless-error review, and focus solely on whether .the properly admitted evidence provided sufficient proof of guilt.
Rather, in Brecht the court expressly held that “the Kotteakos harmless-error standard applies” to collateral habeas relief. Brecht, 507 U.S. at 638, 113 S.Ct. at 1722. Moreover, the Brecht Court applied the Kotteakos standard in its entirety. Id. Both the majority and Justice Stevens’ concurrence focus on the effect of the error on the judgment within the trial as a whole.
The Brecht Court began its analysis by noting'that “[t]he State’s references to [the error] were infrequent, comprising less than two pages of the 900-page trial transcript in this case.” Id. at 639, 113 S.Ct. at 1722. The Court contrasted the small role the error played to the multiple references to similar permissible evidence, as well as the rest of the evidence presented in the ease. The Court then “eonclude[d] that the [error] which occurred at petitioner’s trial did not ‘substantially influence’ the jury’s verdict.” Id. The Brecht Court’s analysis accords with Kotteakos: tiie key question was the error’s influence upon the jury, measured by the prosecution’s minimal use of the erroneously admitted evidence, in comparison to the evidence presented at trial as a whole.
Justice Stevens’ critical fifth vote concurrence also established that the entire Kotteakos standard applies. Justice Stevens quoted most of the language discussed above from Kotteakos and concluded that:
The habeas court cannot ask only whether it thinks the petitioner would have been convicted even if the constitutional error had not taken place. Kotteakos is full of warnings to avoid that result. It requires a reviewing court to decide that “the error did not influence the jury,” and that “the judgment was not substantially swayed by the error.”
Brecht, 507 U.S. at 642, 113 S.Ct. at 1724 (Stevens, J., concurring) (footnote and citations omitted).
Just two terms ago in O’Neal v. McAninch, the Supreme Court reaffirmed that “the Kotteakos standard applie[s] in its entirety.” 513 U.S. 432, -, 115 S.Ct. 992, 996, 130 L.Ed.2d 947 (1995) (emphasis in original). The O’Neal Court directed that “if a judge has grave doubt about whether an error affected a jury in [a substantial and injurious] way,” that error is not harmless. Id. at-, 115 S.Ct. at 995. O’Neal also quotes, with approval, a passage from Kotteakos that makes clear the majority’s error in the case at hand:
If, when all is said and done, the [court] is sure that the error did, not influence the jury, or had but very slight effect, the verdict and the judgment should stand---But if one cannot say, with fair assurance, *378after pondering all that happened without stripping the erroneous action from the whole, that the judgment was not substantially swayed by the error, it is impossible to conclude that substantial rights were not affected. The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error itself had substantial influence. If so, or if one is left in grave doubt, the conviction cannot stand.
Id. at-, 115 S.Ct. at 995 (quoting Kotteakos at 764-65, 66 S.Ct. at 1248) (emphasis added).
Further, in every other arena where harmless-error analysis is applied courts have always been admonished to consider the effect the error had on the jury, and not the sufficiency of the remaining evidence. For example, under Federal Rule of Criminal Procedure 52(a), harmless-error review focuses on the effect of the error itself, within the context of the trial as a whole. United States v. Lane, 474 U.S. 438, 449, 106 S.Ct. 725, 732, 88 L.Ed.2d 814 (1986) (“The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error. It is rather, even so, whether the error had substantial influence.”); id. at 460, 106 S.Ct. at 737 (Brennan, J., concurring in part and dissenting in part) (quoting Kotteakos, 328 U.S. at 760-65, 66 S.Ct. at 1245-48). The approach under civil harmless-error review is identical. See 11 Charles Alan Wright, Arthur R. Miller, Mary Kay Kane, Federal Practice and Procedure § 2883, at 447 (1995) (quoting Kotteakos, 328 U.S. at 761-62, 66 S.Ct. at 1246-47); Black’s Law Dictionary 718 (6th Ed.1990) (“An error is ‘harmless’ if reviewing court, after viewing entire record, determines that no substantial rights of defendant were affected and that error did not influence, or had only very slight influence on verdict.”).
Notably, the majority opinion only cites once to Kotteakos, and completely ignores O’Neal’s statement (which postdates Brecht) that “the Kotteakos standard applie[s] in its entirety.” O’Neal, 513 U.S. at -, 115 S.Ct. at 996 (emphasis in original). In fact, the majority relies principally on only two citations to Supreme Court law to justify its approach. The first citation is Brecht’s statement that to prevail under Kotteakos a habeas petitioner must “establish that [the trial error] resulted in ‘actual prejudice.’ ” Op. at 371 (emphasis omitted) (quoting Brecht, 507 U.S. at 637, 113 S.Ct. at 1722) (quoting United States v. Lane, 474 U.S. 438, 449, 106 S.Ct. 725, 732, 88 L.Ed.2d 814 (1986)). If the Brecht Court meant to materially alter its approach to Kotteakos harmless-error review this quotation is not where it said so. The “actual prejudice” language is a quote from United States v. Lane, which relies heavily on Kotteakos, and specifically rejects the majority’s approach on the very same page it requires actual prejudice: “The inquiry cannot be merely whether there was enough to support the result, apart from the phase affected by the error.” Lane, 474 U.S. at 449, 106 S.Ct. at 732 (quoting Kotteakos, 328 U.S. at 765, 66 S.Ct. at 1248). Furthermore, as the case at hand powerfully demonstrates, a defendant suffers “actual prejudice” and then some when a jury has based its guilty verdict upon erroneously admitted evidence. This actual prejudice does not dissipate because a court of appeals, as opposed to a jury, later reaches the “inescapable conclusion” that the properly admitted evidence reweighed on appeal is “overwhelming.” Op. at 371.
The second citation relied upon by the majority is the Brecht Court’s statement that “[mjoreover, the state’s evidence of guilt was if not overwhelming, certainly weighty.” Id. at 371 (quoting Brecht, 507 U.S. at 639, 113 S.Ct. at 1722). The majority ignores a key distinction between its approach and the approach appliéd in Brecht. The Supreme Court in Brecht did not consider the state’s properly admitted “evidence of guilt” standing alone, as the majority does today. The Brecht Court did not even give this evidence primacy. Rather it first considered the erroneously admitted evidence, noting that “[t]he state’s references to petitioner’s post-Miranda silence were infrequent, comprising less than two pages of the 900 page transcript in this case.” Brecht, 507 U.S. at 639, 113 S.Ct. at 1722. Then, and only then, the Brecht Court compared the effect on the jury of the properly admitted evidence of guilt *379with the effect of the erroneously admitted evidence, by weighing the state’s “infrequent” references to the erroneous evidence against its “extensive and permissible references” to other evidence. Id.
In sum, the majority mischaracterizes Brecht, and ignores the most relevant portions of Kotteakos and O’Neal in conducting its analysis. Thus, although in the case at hand the majority correctly recognizes that “principles of federalism, comity, and finality” underlie our habeas jurisprudence, Op. at 370, it ignores the equally important directive that on habeas review a federal court must consider whether its holding is “consistent with the basic purposes underlying the writ of habeas corpus.” O’Neal, 513 U.S. at -, 115 S.Ct. at 997. The Supreme Court has explained, “we are dealing here with an error of constitutional dimension — the sort that risks an unreliable trial outcome and the consequent conviction of an innocent person.” Id. In following a misguided and unprecedented approach today, the majority ignores Justice Scalia’s warning that harmless-error review must not “hypothesize a guilty verdict that was never in fact rendered — no matter how inescapable the findings to support that verdict might be — [because to do so] would violate the jury-trial guarantee.” Sullivan, 508 U.S. at 279, 113 S.Ct. at 2081-82.
II.
Any possible question as to whether the majority applies the wrong approach vanishes upon examination of its analysis. The majority never even acknowledges the obvious importance of the improperly admitted, detailed, taped, confession to the jury’s verdict. This omission is not inadvertent; any examination of the effect of the taped confession inevitably leads to the conclusion that the error in admitting it had a “substantial and injurious effect” upon the jury’s verdict, and renders the majority’s conclusion to the contrary unimaginable.
In the majority’s four paragraphs of analysis concerning the evidence presented at trial, the effect of the error upon the judgment is not mentioned once. Instead, the majority conducts its own hypothetical trial to save the State of South Carolina and Kamathene Cooper the “substantial risk and cost” of a new trial. Op. at 371.2 Although the majority claims not to “conduct an independent assessment of the evidence,” it totally disregards the piece of evidence most relied upon by the State at Cooper’s trial. The majority is able to - deem the evidence presented against Cooper “overwhelming” only by eliminating from the calculus the erroneously admitted taped confession, and considering the remaining evidence alone. Of course, this approach does nothing to establish the basis on which the actual jury in this case based its verdict. In light of the majority’s complete disregard for the actual trial Cooper received, it is ironic that it criticizes me for ignoring any “actual effect on the outcome of the trial.” Id. at 371 (emphasis in original). Even a cursory review of both opinions reveals it is the majority that never comes to grips with the error’s “actual effect on the outcome of the trial.”
The majority’s approach, however, may be the only one available to it because, as the panel concluded, “it is hard to envision a case in which erroneously admitted evidence was more heavily relied -on by the prosecution.” Cooper v. Taylor, 70 F.3d 1454, 1467 (4th Cir.1995).
Even a simple comparison of the volume of testimony relating to the brief earlier confessions with that relating to the erroneously admitted taped confession establishes the insignificance of the former, and overwhelming importance of the latter, in the ease the jury actually considered. There are- only three extremely brief references to the two earlier confessions in the entire record: (1) two paragraphs of Agent Grimsley’s testimony, (2) a single paragraph of Officer Vause’s testimony, which takes up a third of a page of trial transcript, and (3) Officer McKenzie’s even more circumspect testimony, stating only that “I didn’t hear exactly everything. *380I can remember him saying that he killed Mr. Stewart. I think he said the old man. Yes.” In contrast, the taped confession consumes nineteen pages of trial transcript, and details virtually every aspect of the case, including many facts never alluded to in the earlier statements. Id. at 1467-68. Moreover, a written, signed copy of the taped confession was provided to the jury to read as the tape of the confession was played to them, and then both the tape and transcript (unlike the isolated testimony as to the earlier confessions) were provided to the jury during its deliberations. Id. at 1460.
Furthermore, in his closing argument the prosecutor made not just “several,” Op. at 369, but more than fifteen different references to the taped confession. Every aspect of the prosecutor’s description of the facts to the jury comes straight from Cooper’s taped confession. All of these facts are preceded with “he said” or “did he not say,” to hammer home Cooper’s own voice stating the facts. As the panel noted, the closing argument:
is literally saturated with references to the taped confession. We pick just three examples. First, the prosecution explained to the jury that it did not need to rely on implied malice because “out of Cooper’s mouth” jurors had heard evidence of “real, hard, actual real world malice”:
... you look at the facts of this ease and when you hear Mr. Cooper’s own voice telling you that he took that kitchen knife with him thinking he would use it against Mr. Stewart to rob him or kill him. When he tells you in his own voice and signs a statement a day later saying after he had given me the newspapers, I got him to turn around and reach for the basketball so his back would be to him so I could clobber him with the chair.
There you see, not only what the law calls implied malice, because when you use a deadly weapon like a knife against somebody, that’s a fact you can use to infer malice, evil intent. But when you hear out of the words of somebody’s mouth, this is the way I did it. I took it ahead of time. Well, I didn’t know if I was going to kill him, but as I sat there, I just decided to do it. I formulated a plan. I got him to turn around. I made sure it was done. That’s real, hard, actual real world malice.
(emphasis added). Second, the prosecution pointed out that the jury had, by “listening” to what Cooper had “said,” learned his motive for murder:
Now, the law does not require us to prove motive. But you can look at facts and you can listen to what he said and you see motive there. What is the first thing this man, Kamathene Cooper did after he had rendered this man senseless and sent him towards his inevitable death?
He went through his pockets. He went through his pockets. And you will recall on his statement from reading along and listening. I looked, but I didn’t find anything. And he said I went in other rooms, but I didn’t find anything. And, of course, you see the evidence that he went through Mr. Cooper’s (sic) pocket. You see the evidence as corroborated in his statement that he did, in fact, go in that room where the stereo was and pull to one side that curtain not realizing what was there,
(emphasis added). Third, the prosecution concluded its argument with numerous rhetorical questions based on what Cooper “said” or “told” the jury:
Did he not say in the statement that he had the gentleman to bend over in the corner to look for a basketball? Is this not the basketball? Did he not say that he hit the man with a chair? Do we not have the chair? Did he not say that he was admitted into the house without having a break-in and Mr. Gravely not tell you that there was no sign of forcible entry[?]
Did he not tell you that he stabbed the man with a knife? And did Dr. Conrad not tell you that the man had a stab wound to the head that penetrated his brain? Did he not tell you that he stole a checkbook? And do we not have the stolen checkbook? Did he not take the officers to where the checkbook was *381and are there not photographs of him being there? Did he not say I took about six checks out? And are there not but about five or six checks missing? Did he not say that he went to Thomlinson’s? And did Mr. King not verify that he had been at Thomlinson’s? Point after point after point after point after point.
Cooper, 70 F.3d at 1466-67 (emphasis in original).
Finally, totally unacknowledged by the majority is the assessment of the State trial judge, who is in a far better position to assess the evidence than this court on habeas. In denying Cooper’s motion for a directed verdict, the State court expressly recognized:
... of course, the case hinges and mil stand or fall upon the alleged confession. And I have previously denied your motion to suppress that evidence. And in furtherance of that motion, of course, there was at least a day that elapsed after the making of the statement on the tape and then the typing it up and then giving it back to him. He had an opportunity to read it and refuse to sign it and deny it if he had so desired and I think those are questions of fact for the juiy, sir, and I would deny your motion.
(emphasis added). Thus, the trial court, which heard all of the evidence in the ease, was absolutely clear that the taped confession alone — not in conjunction with the brief earlier statements — would be determinative of the jury’s verdict.3 To ignore the State trial court’s considered view of the effect of the taped confession, as the majority does, is at odds with both the Supreme Court’s observation in Brecht that “state courts often occupy a superior vantage point from which to evaluate the effect of trial error,” Brecht, 507 U.S. at 636, 113 S.Ct. at 1721 and the majority’s own “healthy respect” for state courts. Op. at 369.
In sum, the case that the majority deems “overwhelming,” is a ease cobbled together from evidence that was at best a sidelight at Cooper’s actual trial. It is certainly not the case the jury actually considered when it convicted Kamathene Cooper. There can be no question that under the proper standard of harmless-error review,, which examines “the basis on which the jury actually rested its verdict” or “whether the guilty verdict actually rendered in this trial was ... unattributable to the error,” Sullivan, 508 U.S. at 279, 113 S.Ct. at 2081 (emphasis in original) (quotation marks omitted), the majority opinion is flatly wrong.
III.
In addition to the majority’s fundamentally flawed approach to harmless-error review, it errs in several other respects.
For example, the majority asserts that Cooper “offered no evidence of his own ... to contend that his confessions were not voluntary.” Op. at 371. In fact, the record establishes that Cooper has consistently maintained that all three confessions were involuntary, and presented expert testimony that he did not understand his rights and did not effectively waive them. Cooper, 70 F.3d at 1469. (Indeed, again before us Cooper continues to assert that like the taped confession, the two earlier confessions were obtained in violation of his “Fifth Amendment rights to remain silent;” a claim the majority simply ignores.) Voluntariness is a jury issue, and the taped confession severely impeded Cooper’s assertion that the other confessions were involuntary, because it allowed the jurors to hear Cooper, in his own voice, acknowledging his Miranda rights, and describing the killings. Without the tape Cooper could have pressed the officer’s vague recollections of his other two confessions, and perhaps established involuntariness. Moreover, as the panel concluded after it considered all of the evidence relied upon by the majority (and in much greater detail); “without the taped confession not only would the most powerful evidence against Cooper have been eliminated, but much of the other evidence against him would not have been avail*382able or would have been irrelevant.” Id. at 1467-68.
Moreover, the majority utterly fails to address the Supreme Court’s analysis in Arizona v. Fulminante, 499 U.S. 279, 111 S.Ct. 1246, 113 L.Ed.2d 302 (1991). It does not even cite Fulminante, in which the Supreme Court explained:
the defendant’s own confession is probably the most probative and damaging evidence that can be admitted against him____ [A] full confession in which the defendant discloses the motive for and means of the crime may tempt the jury to rely upon that evidence alone in reaching its decision____ [Thus,] a reviewing court [must] exercise extreme caution before determining that the admission of the confession at trial was harmless.
Fulminante, 499 U.S. at 296, 111 S.Ct. at 1257-58. In words directly applicable here, the Fulminante Court went on to hold that the admission of the defendant’s first confession was not cumulative or harmless, even though the State had properly admitted a second, more detailed confession because “the jury’s assessment of the [admissible] confession ... could easily have depended in large part on the presence of the [inadmissible] confession____” Id. at 298, 111 S.Ct. at 1258.
Although Fulminante was decided under the less strict Chapman evidentiary standard, the prejudice the taped confession caused Kamathene Cooper far outweighs the prejudice at issue in Fulminante, and the State of South Carolina has never asserted that Fulminante would be decided differently under the Brecht evidentiary standard. See Cooper, 70 F.3d at 1465. As Justice Stevens aptly stated in his Brecht concurrence, “Justice Kennedy’s cogent analysis [in Fulminante ] demonstrated that the error could not reasonably have been viewed as harmless under a standard even more relaxed than the one we announce today____ In the end, the way we phrase the governing standard is far less important than the quality of judgment with which it is applied.” Brecht, 507 U.S. at 643, 113 S.Ct. at 1725 (Stevens, J., concurring). Of course, Justice Stevens could hardly have foreseen the decision of today’s majority, which both follows a misguided legal approach and applies an analysis that flies in the face of Fulminante as well as fifty years of harmless-error precedent.
IV.
Today, a majority of the court disregards the correct approach to harmless-error review, which requires an examination of the trial that actually occurred. Instead, the court conducts its own shadow-trial and concludes that without the improperly admitted taped confession there was sufficient evidence to convict Cooper. This approach “hypothesize[s] a guilty verdict that was never in fact rendered” — the precise result Justice Scalia warned against in Sullivan v. Louisiana, 508 U.S. at 279, 113 S.Ct. at 2081-82.
If it is appropriate to analogize a criminal trial at which a man has been convicted of murder and sentenced to life imprisonment to a baseball game, that analogy — as clearly as anything else — exposes the fundamental flaw in the majority’s entire approach. When the majority retallies the “score” against Kamathene Cooper without “the government's] grand-slam home run,” Op. at 370, it unequivocally demonstrates its lack of understanding of harmless-error review. It is not enough for an appellate court to ask how the “game” would have turned out without the erroneous evidence, instead a court is required to ask how the State won the “game’’. In this case the State did not run up the score before hitting a grand slam; the entire “game” turned on the erroneous evidence.
The majority may well be right that South Carolina could in another trial convict Cooper, without reliance on the erroneously admitted taped confession. The majority is unquestionably wrong, however, in concluding that the error in admitting the taped confession in the trial at which Cooper was actually convicted was harmless. The admission of, and heavy reliance on, the taped confession at Cooper’s trial could only have had “substantial and injurious effect or influence” on “the jury’s verdict.” Brecht, 507 U.S. at 637, 113 S.Ct. at 1722.
*383More than one hundred years ago, the Supreme Court recognized that “[t]he great writ of habeas corpus has been for centuries esteemed the best and only sufficient defence of personal freedom.” Ex Parte Yerger, 8 Wall. 85, 95, 19 L.Ed. 332 (1868). I respectfully dissent ft-om today’s decision, which ignores precedent and cripples this most critical safeguard to the personal freedom we all cherish.
MURNAGHAN, ERVIN, HAMILTON, and MICHAEL, Circuit Judges, join this dissenting opinion.
. The court must, as the majority does, reach the question of whether the Edwards error in admitting the taped confession was harmless.' Regardless of whether the State could have pursued á claim that its admission did not violate Edwards after failing to file any exceptions to the magistrate judge's findings, the State did not do this. Rather, at oral argument, both before the panel and the en banc court, the State expressly and unequivocally waived this claim.
. In view of its holding, it is difficult to understand why the majority believes that ordering a new trial would result in "substantial risk." Op. at 371. But, if without the taped confession there is indeed a "substantial risk” that Cooper would not be convicted of murder, then obviously admission of the taped confession cannot be held harmless.
. To suggest, as the concurrence does, that on this record the erroneously admitted taped confession was "cumulative” to the short and poorly recollected testimony concerning Cooper's earlier statements is equivalent to suggesting that a blizzard is cumulative to a snow flurry.