concurring in the judgment.
Today’s judgment reverses the decision below on the grounds that, in deciding not to apply the “manifest injustice” exception to the law of the case, the Court of Appeals wrongfully failed to consider a newly discovered transcript from petitioner’s trial. The judgment is correct, but the judgment is also not worth making, serving no purpose but to extend the scandalous delay in the execution of a death sentence lawfully pronounced more than 18 years ago.
As a general matter, I agree with Justice Stevens’ frequently recited view that “[i]t is not appropriate for this Court to expend its scarce resources crafting opinions that correct technical errors in cases of only local importance where the correction in no way promotes the development of the law.” Anderson v. Harless, 459 U. S. 4, 12 (1982) (Stevens, J., dissenting). “‘To remain effective, the Supreme Court must continue to decide only those cases which present questions whose resolution will have immediate importance far beyond the particular facts and parties involved.’” Board of Ed. of Rogers v. McCluskey, 458 U. S. 966, 971 (1982) (Stevens, J., dissenting) (quoting address of Chief Justice Vinson before the American Bar Association (Sept. 7,1949)). I am willing to make an exception from that rule in capital cases — but only where there is a realistic likelihood that the “technical error” affected the conviction or the sentence. Here that is fanciful.
To begin with, the rediscovered transcript covers only the closing statements in the case, and petitioner’s claim goes *361only to the fact that his trial counsel made one rather than two arguments in closing.. It would be a rare case in which the omission of one particular argument from an attorney’s closing statement would be dispositive of an ineffective-assistance claim. It is simply not true, moreover, that the transcript “flatly contradices],” ante, at 358, the testimony of petitioner’s trial counsel in the initial habeas hearing — or at least not true in the sense that it shows trial counsel was lying. The transcript confirms that, as trial counsel recalled, he had argued that the death penalty was inappropriate in any circumstance. In fact, he read to the jury large portions of Justice Brennan’s opinion in Furman v. Georgia, 408 U. S. 238 (1972), which is certainly an eloquent argument that the death penalty is improper in any case. The transcript does reveal that counsel had not argued in mitigation that the killing was impulsive — which, petitioner now claims, shows that counsel’s prior testimony was “false,” Pet. for Cert. 19. That is not so. Although counsel stated at one point that he was “sure” he had argued the impulsive-killing point, a few lines earlier in the transcript he had said that “I would assume that I argued [it]” (emphasis added), and had made clear that “a lot of this is really not from actual recollection.” Tr. 70-71 (Nov. 10, 1982). Petitioner’s ha-beas counsel understood the import of this, describing (in his posthearing brief) trial counsel’s testimony to have been that “he probably argued that the killing was impulsive and not planned.” Petitioner’s Post-Hearing Brief in No. C80-247R (ND Ga., Dec. 2, 1982), p. 25, n. 6 (emphasis added); see also Objections to Magistrate’s Report and Recommendation in No. C80-247R (ND Ga., Sept. 12, 1983), p. 8, n. 1.
There is absolutely zero likelihood that counsel’s misrecol-lection (or misreconstruction) that he had made an “impulsiveness” argument to the jury made the difference in the 1986 finding that his assistance was not ineffective. Petitioner’s ineffectiveness contention had not been directed to the inadequacy of counsel’s closing argument, but rather to his *362failure to introduce mitigating evidence — character witnesses and the like — during the sentencing phase. See Petition for Writ of Habeas Corpus (Dec. 3, 1980), ¶ 49, p. 12, Petitioner’s Post-Hearing Brief (Dec. 2, 1982), pp. 17-30, Magistrate’s Report and Recommendation (Aug. 26, 1983), pp. 8-18, and Objections to the Magistrate’s Report and Recommendation (Sept. 12, 1983), pp. 2-12, in No. C80-247R (ND Ga.); Dobbs v. Kemp, 790 F. 2d 1499, 1513 (CA11 1986), modified in part, 809 F. 2d 750 (CA11 1987). In addressing that contention, neither the District Court nor the Court of Appeals relied on the content of the closing argument, for the obvious reason that it does not constitute evidence. The District Court described the closing argument in setting forth the events of the sentencing phase, but did not mention it in its legal analysis, ruling instead that counsel had made a “reasonably substantial investigation” in searching for witnesses to testify on petitioner’s behalf and a “reasonable tactical decision not to present evidence based on the information before him.” No. C80-247R (ND Ga., Jan. 13, 1984), p. 25. In affirming the District Court, the Court of Appeals elaborated on what it thought was the crucial “tactical decision” of the trial counsel (which was not a decision to omit evidence of mitigation in favor of a stunning closing argu-j ment): “Knowing of [petitioner’s] poor reputation in the community, [counsel] testified in the district court that he did not want to put on any ‘positive’ character testimony for fear that it would not be persuasive and would prompt damaging counter evidence from the prosecution.” Dobbs v. Kemp, 790 F. 2d, at 1513-1514. The closing argument was mentioned only in a footnote, which stated: “Although [counsel] did not present any mitigating evidence, his testimony in the district court reveals that he did make a closing argument in mitigation.” Id., at 1514, n. 15.
I think it unimaginable that, if this transcript had been available in 1986 — showing that only Justice Brennan’s moving Furman argument, and not also an “impulsiveness” ar*363gument had been made — the Court of Appeals would have found that there was ineffective assistance of counsel. (Indeed, if it were our practice to correct technical errors I would vote to reverse a finding that assistance otherwise effective was rendered ineffective by that omission.) But I think it soars beyond the unimaginable, into the wildly delirious, to believe that the Court of Appeals will find that the newly discovered transcript demonstrates such “manifest injustice” that it warrants making an exception to the law-of-the-case doctrine, so that the already-decided ineffectiveness question should be reopened. There was, in short, no reason to grant this petition and correct this (admittedly clear) technical error, except to place another obstacle in the way of a death penalty that has been suspended within this Court’s “death is different” time warp since 1974.