with whom Justice Brennan and Justice Marshall join, dissenting.
Last Term, in Caldwell v. Mississippi, 472 U. S. 320 (1985) (a case not even cited by the Court in its controlling opinion, ante, p. 376), we recognized institutional limits on an appellate court’s ability to determine whether a defendant should be sentenced to death:
“Whatever intangibles a jury might consider in its sentencing determination, few can be gleaned from an appellate record. This inability to confront and examine the individuality of the defendant would be particularly devastating to any argument for consideration of what this Court has termed ‘[those] compassionate or mitigating factors stemming from the diverse frailties of humankind.’ When we held that a defendant has a constitutional right to the consideration of such factors, we clearly envisioned that that consideration would occur among sentencers who were present to hear the evidence and arguments and see the witnesses.” 472 U. S., at 330-331 (citations omitted; interpolation in original).
That statement in Caldwell is not an abstract disquisition on appellate courts generally. It concerns, in particular, the institutional limits of the Supreme Court of Mississippi in capital cases. Today, the Court ignores those recently stated limits and holds that the Mississippi Supreme Court may be competent to make, on a paper record, the findings required by Enmund v. Florida, 458 U. S. 782 (1982) — that Crawford Bullock, Jr., killed, attempted to kill, or intended to kill Mark Dickson, and thus deserves to die. The Court reaches that result by paying lipservice to the constitutional significance of Enmund while relegating Enmund findings to a position of judicial afterthought. The nature of the Enmund findings, however, dictates who must make them and at what point in the sentencing process they must be *395made. The Eighth Amendment requires that Enmund findings be made at the trial court level before the sentencer condemns a defendant to death. The Court’s misreading of Enmund threatens a retreat from the constitutional safeguards on the capital sentencing process that the Court has acknowledged in the decade since Gregg v. Georgia, 428 U. S. 153 (1976).
I
Bullock testified both at his trial and at his sentencing proceeding. He explicitly denied that he killed, attempted to kill, or intended to kill Dickson. See, e. </., Tr. 956, 983, 996, 1190. The jury’s verdict and sentence are entirely consistent under Mississippi law with Bullock’s testimony. As the Court recognizes, that law and the trial court’s instructions permitted the jury to convict him and to sentence him to death without finding any particular degree of personal participation in the killing. Ante, at 383-384.
The Court also recognizes that the Mississippi Supreme Court failed to make the required Enmund findings. That court affirmed Bullock’s conviction and death sentence based on its view of Bullock’s culpability under Mississippi’s law of aiding and abetting, which establishes a threshold far below Enmund’s constitutional minimum. Ante, at 389-390. The Mississippi Supreme Court explicitly based its account of the crime on Bullock’s written confession, see Bullock v. State, 391 So. 2d 601, 605, cert. denied, 452 U. S. 931 (1981), in which Bullock stated only that Tucker killed Dickson, and that he, Bullock, had no intention of robbing Dickson. Tr. 387-390. That confession provides no evidence that Bullock killed, attempted to kill, or intended to kill Dickson. Thus, the Court properly concludes that none of the required Enmund findings has been made.
■I — I I — I
The central message of Enmund is that the death penalty cannot constitutionally be imposed without an intensely indi*396vidual appraisal of the “personal responsibility and moral guilt” of the defendant. 458 U. S., at 801.
“The focus must be on his culpability, ... for we insist on ‘individualized consideration as a constitutional requirement in imposing the death sentence,’ Lockett v. Ohio, 438 U. S. 586, 605 (1978) (footnote omitted), which means that we must focus on ‘relevant facets of the character and record of the individual offender.’ Woodson v. North Carolina, 428 U. S. 280, 304 (1976).” Id., at 798 (emphasis in original).
See also Eddings v. Oklahoma, 455 U. S. 104, 110-112 (1982); Lockett v. Ohio, 438 U. S. 586, 603-604 (1978) (plurality opinion); Gregg v. Georgia, 428 U. S., at 199 (joint opinion).
Put simply, Enmund establishes a constitutionally required factual predicate for the valid imposition of the death penalty. Cf. ante, at 390. Like the statutory aggravating circumstances discussed in Zant v. Stephens, 462 U. S. 862 (1983), the Enmund findings “circumscribe the class of persons eligible for the death penalty.” 462 U. S., at 878. Just as, absent the finding of a statutory aggravating circumstance, “ ‘[a] case may not pass . . . into that area in which the death penalty is authorized’” under Georgia law, id., at 872, quoting Zant v. Stephens, 250 Ga. 97, 100, 297 S. E. 2d 1, 4 (1982), so too, absent a finding of one of the Enmund factors, a case may not pass into that area in which the death penalty is authorized by the Eighth Amendment.
The Court agrees that it would be wrong for Mississippi to execute Bullock without first determining that he killed, attempted to kill, or intended to kill Dickson. See, e. g., ante, at 378, 385, 386. But if that is so, then it was also wrong for the Mississippi jury to discharge “the truly awesome responsibility of decreeing death for a fellow human,” McGautha v. California, 402 U. S. 183, 208 (1971), without first considering the fundamental issue of his personal culpability. By condemning Bullock to die, the jury announced *397that he was not fit to live. This expression of the community’s ultimate outrage, unaccompanied as it was by any finding that Bullock possessed the degree of culpability required by Enmund, involved the kind of deprivation of human dignity which the Eighth Amendment forbids. Cf., e. g., Trop v. Dulles, 356 U. S. 86, 100-102 (1958) (plurality opinion); Weems v. United States, 217 U. S. 349, 366 (1910).
A
The question of how to cure this constitutional violation remains. The Court holds that an adequate remedy for the absence of Enmund findings can be supplied by “any court that has the power to find the facts and vacate the sentence.” Ante, at 386. I believe that, in this case, only a new sentencing proceeding before a jury can guarantee the reliability which the Constitution demands. But the Court’s decision today goes beyond a simple determination of how to cure an error that has already occurred. It tells the States, in effect, that it is no error for a jury or a trial judge to say that a defendant should die without first considering his personal responsibility and moral guilt, as Enmund requires. By turning the jury or trial court’s determination into what can be viewed only as a preliminary stage in the capital-sentencing process, the Court’s holding poses the threat of diffusing the sentencer’s sense of responsibility in the manner condemned in Caldwell. The Court thus ignores both the proper institutional roles of trial and appellate courts and the pragmatic and constitutional concerns with reliability that underlie those roles. In short, the Court’s holding rests on an improper equation of the wholly dissimilar functions of finding facts and of vacating a sentence because no facts have been found. Enmund established a clear constitutional imperative that a death sentence not be imposed by a sentencer who fails to make one of the Enmund findings. The Court confuses this imperative with the guarantee it purports to *398make today that a death sentence will not be carried out before someone makes an Enmund finding.
That this ignores a distinction with a constitutional difference is made clear by the Court’s decisions in Cole v. Arkansas, 333 U. S. 196 (1948), and Presnell v. Georgia, 439 U. S. 14 (1978). In Cole, the Court reversed a state appellate decision that had affirmed the defendants’ sentences by finding they had violated a different statutory provision from the one with which they had been charged. It recognized that the Due Process Clause requires that defendants “have the validity of their convictions appraised on consideration of the case as it was tried and as the issues were determined in the trial court.” 333 U. S., at 202. In Presnell, the Court acknowledged that the “fundamental principles of procedural fairness” announced in Cole “apply with no less force at the penalty phase of a trial in a capital case than they do in the guilt-determining phase of any criminal trial.” 439 U. S., at 16. It thus reversed a death sentence which the Georgia Supreme Court had affirmed on the basis of its own finding that evidence in the record would support a statutory aggravating circumstance that had not been found by the jury. Notably, in neither Cole nor Presnell did this Court consider whether the State Supreme Courts’ evidentiary findings were correct; whether their findings were right was entirely irrelevant to the question whether the Due Process Clause gave them the power to make such findings. The Court’s decision today gives a state appellate court carte blanche to engage in factfinding concerning issues that no one at trial thought to be relevant. Here, as the Court recognizes, “ ‘the entire case was essentially tried on the theory. . . that it was not necessary, either for the felony murder conviction or for the sentence to death, to find that Bullock had either the intent to kill or any personal participation in the killing.’” Ante, at 384, quoting Bullock v. Lucas, 743 F. 2d 244, 248 *399(CA5 1984) (concurring opinion); see also, e. g., Tr. 1155. The critical issue was never determined in the trial court.1
Far more than “[cjonsiderations of federalism and comity,” ante, at 391, should prevent this Court, and other federal habeas courts, from examining trial transcripts and making Enmund findings themselves. Considerations of reliability provide a compelling reason for requiring state trial courts to address this issue in the first instance. And, with respect to the question of reliability, the Mississippi Supreme Court is in no better position than is this Court to determine Bullock’s credibility.
The Court’s conclusion that we should allow the States to adopt capital punishment schemes that depend on appellate factfinding because “it is by no means apparent that appellate factfinding will always be inadequate,” ante, at 388, n. 5 (emphasis added), turns on its head the heightened concern with reliability that has informed our review of the death penalty over the past decade.2 See, e. g., Caldwell v. Mississippi, *400472 U. S., at 328-329; California v. Ramos, 463 U. S. 992, 998-999 (1983); Beck v. Alabama, 447 U. S. 625, 637-638 (1980); Lockett v. Ohio, 438 U. S., at 604 (plurality opinion); Gardner v. Florida, 430 U. S. 349, 358-359 (1977) (opinion announcing judgment); Woodson v. North Carolina, 428 U. S. 280, 305 (1976) (plurality opinion). I believe that the Eighth Amendment not only requires that the sentencer make Enmund findings before it decides that a defendant must die, but also requires that the Enmund factfinder be present at the trial, to see and hear the witnesses.
The Court long has recognized the special competence of trial courts which formed the basis for Caldwell’s discussion of the “institutional limits on what an appellate court can do.” 472 U. S., at 330. In a variety of contexts, the Court has relied upon the New York Court of Appeals’ explanation *401in Boyd v. Boyd, 252 N. Y. 422, 429, 169 N. E. 632, 634 (1930):
“Face to face with living witnesses the original trier of the facts holds a position of advantage from which appellate judges are excluded. In doubtful cases the exercise of his power of observation often proves the most accurate method of ascertaining the truth. . . . How can we say the judge is wrong? We never saw the witnesses. ... To the sophistication and sagacity of the trial judge the law confides the duty of appraisal. . . . His was the opportunity, the responsibility and the power to decide.”
See, e. g., Wainwright v. Witt, 469 U. S. 412, 434 (1985) (quoting Boyd); Marshall v. Lonberger, 459 U. S. 422, 434 (1983) (same).
Our precedents are not to the contrary. Although we held in Spaziano v. Florida, 468 U. S. 447 (1984), that neither the Sixth nor the Eighth Amendment required jury sentencing in capital cases, we made that determination in the face of a Florida statute which “plac[ed] responsibility on the trial judge to impose the sentence in a capital case.” Id., at 465 (emphasis added). In the relevant respects, a trial judge in a capital case is more like a jury than he is like an appellate court. Like the jury, he has seen the witnesses, and is well positioned to make those “determinations of demeanor and credibility that are peculiarly within a trial judge’s province.” Wainwright v. Witt, 469 U. S., at 428.3
*402B
The Court’s discussion of “the nature of our ruling in Enmund,” ante, at 384, reveals a reliance on three premises: first, Enmund “does not impose any particular form of procedure upon the States,” ante, at 386 (emphasis omitted); second, Enmund “‘does not affect the state’s definition of any substantive offense, even a capital offense,”’ ante, at 385, quoting Reddix v. Thigpen, 728 F. 2d 705, 709 (CA5), cert. denied, 469 U. S. 990 (1984); and, third, Enmund is a “substantive limitation on sentencing” amenable to traditional proportionality review, ante, at 386. None of these propositions justifies the Court’s holding today.
That we have refused “‘to say that there is any one right way for a State to set up its capital sentencing scheme,”’ ante, at 387, quoting Spaziano, 468 U. S., at 464, does not mean that there are no wrong ways. As has been shown, a capital-sentencing scheme that permits an appellate court to *403make Enmund findings sacrifices reliability needlessly to no discernible end, and cannot satisfy the Eighth Amendment.
That Enmund does not restrict the State’s power to define offenses is equally beside the point. A State’s decision to define a crime as “capital” cannot “automatically . . . dictate what should be the proper penalty,” Lockett v. Ohio, 438 U. S., at 602 (plurality opinion), and does not empower the State to execute a defendant who neither killed, nor attempted to kill, nor intended to kill. In Coker v. Georgia, 433 U. S. 584 (1977), for example, Georgia’s definition of rape as a capital offense did not dispose of the Eighth Amendment issue. Both Justice O’Connor’s dissent in Enmund and the Court of Appeals’ opinion in Reddix — the authorities upon which the Court relies —recognize the distinction, which seems to elude the Court, between defining an offense and being entitled to execute a defendant. See Enmund, 458 U. S., at 810, and n. 19 (O’Connor, J., dissenting) (Enmund did not contest his conviction for felony murder; his “sole challenge is to the penalty imposed”); Reddix, 728 F. 2d, at 709 (the State may convict a defendant of a capital crime without requiring an instruction on intent; “Enmund, however, will ‘bar a death penalty’ ” absent such an instruction, quoting Skillern v. Estelle, 720 F. 2d 839, 847 (CA5 1983) (emphasis in Skillern), cert. denied, 469 U. S. 873 (1984)). A State remains free to define felony murder as it wishes; but it can execute a felony murderer who has been sentenced to death only by a sentencer who has determined that he possesses the degree of culpability discussed in Enmund.
The Court also would justify its holding by reference to the discussion of Eighth Amendment principles of proportionality in Solem v. Helm, 463 U. S. 277 (1983). The Court’s discussion mistakenly amalgamates review and essentially de novo factfinding. Certainly, the Court is correct that “the decision whether a sentence is so disproportionate as to violate the Eighth Amendment in any particular case . . . has long been viewed as one that a trial judge or an appellate *404court is fully competent to make.” Ante, at 386. But the Eighth Amendment demands more than that the reviewing court decide whether the sentencer has properly weighed the seriousness of the offense and the severity of the punishment. The Eighth Amendment binds the sentencer as well. The joint opinions in Gregg v. Georgia, 428 U. S. 153 (1976), Proffitt v. Florida, 428 U. S. 242 (1976), and Jurek v. Texas, 428 U. S. 262 (1976), all explicitly rested their approval of the capital-sentencing schemes before them on the combination of channeled factfinding by the sentencer and appellate review. In Gregg, an “important additional safeguard” was provided by the Georgia Supreme Court’s review of “whether the evidence supports the jury’s finding of a statutory aggravating circumstance,” as well as by the exercise of comparative proportionality review. 428 'U. S., at 198. In Proffitt, “meaningful appellate review” was provided because the appellate court had before it written findings justifying the imposition of the death penalty. 428 U. S., at 251. In Jurek, the jury had to make specific findings, which were then subject to appellate review. 428 U. S., at 269, 276. To permit States to collapse factfinding and review into one proceeding is to abandon one of the most critical protections afforded by every capital-sentencing scheme to which the Court previously has given its approval.
Enmund “insisted] on ‘individualized consideration as a constitutional requirement in imposing the death sentence,’” 458 U. S., at 798 (emphasis added), quoting Lockett v. Ohio, 438 U. S., at 605, and not merely in reviewing the sentence imposed. The sentencer is not relieved of the duty to consider whether the severity of the defendant’s crime justifies the death penalty by the availability of proportionality review. Enmund places a substantive limitation on a process that precedes proportionality review.
C
This case demonstrates graphically why a trial-court sentencer must make the Enmund determination. Under *405Mississippi law, “the jury is the sole player in the judicial process who may vote to send an accused to die.” Wiley v. State, 449 So. 2d 756, 762 (Miss. 1984); see also Williams v. State, 445 So. 2d 798, 811 (Miss. 1984), cert. denied, 469 U. S. 1117 (1985). To the extent that Enmund places a substantive limitation on sentencing, then, Bullock is entitled to insist that the sentencing jury heed its limits. Caldwell suggests that to postpone Bullock’s right to an Enmund determination is effectively to deprive him of that right because, in Mississippi, capital review is “conducted with a presumption of. . . correctness.” Wiley, 449 So. 2d, at 762; see Caldwell, 472 U. S., at 331; see also Miss. Code Ann. §99-19-105 (Supp. 1985). The Mississippi Supreme Court examines the record solely to see whether a reasonable jury could have concluded that Bullock killed, attempted to kill, or intended to kill, rather than whether Bullock in fact did any of those things. Saying that Bullock might have acted with the requisite culpability does not satisfy the constitutional requirement that Bullock actually have acted with that degree of blameworthiness.
Hicks v. Oklahoma, 447 U. S. 343 (1980), makes clear that the former inquiry is simply insufficient to satisfy due process. In Hicks, the Court vacated a sentence imposed, as Oklahoma law required, by a jury which had relied upon an invalid statutory provision despite the fact that the Court of Criminal Appeals had affirmed the sentence as within the permissible range. Hicks held that when a State vests the sentencing power in the trial jury, a defendant has “a substantial and legitimate expectation that he will be deprived of his liberty only to the extent determined by the jury in the exercise of its statutory discretion.” Id., at 346. A state appeals court cannot reform a defendant’s sentence, thus denying him the right actually to be sentenced by a jury “simply on the frail conjecture that a jury might have imposed a sentence equally as harsh as that [affirmed by the appellate court]. Such an arbitrary disregard of the petitioner’s right *406to liberty is a denial of due process of law.” Ibid, (emphasis in original).
As for reliability, the Court buries in a footnote an acknowledgment that “the question whether the defendant killed, attempted to kill, or intended to kill might in a given case turn on credibility determinations that could not be accurately made by an appellate court on the basis of a paper record, cf. Anderson v. Bessemer City, 470 U. S. 564, 575 (1985); Wainwright v. Witt, 469 U. S. 412, 429 (1985).” Ante, at 388, n. 5. The Court fails to notice that this is that “given case”: Bullock took the stand, at both the guilt and penalty phases of his trial, to deny having killed, having attempted to kill, or having intended to kill Dickson. See Tr. 956, 983, 996, 1190. I have read the trial transcript. Although I think the evidence is consistent with Bullock’s claim that the killing of Mark Dickson resulted from a drunken brawl between Tucker and Dickson that tragically got out of hand, cf. Bullock v. Lucas, 743 F. 2d, at 248 (concurring opinion), I must concede that a jury or judge who saw Bullock testify might well think he lied. I fail, however, to see how an appellate court confidently could conclude, without any indication from anyone who actually saw him testify, that Bullock’s account was so unworthy of belief that he was properly condemned to death.
Moreover, nothing in the Court’s opinion suggests that this case is at all unusual in this réspect.4 To permit the States *407to construct capital-sentencing schemes that by their very nature will be inadequate in cases such as this strikes me as an abdication of our responsibility under the Eighth Amendment to ensure that the system of capital punishment, as well as the imposition of the penalty on individual defendants, meets the Constitution’s requirements.5
Here, Bullock had a legitimate expectation that the sentencing jury would consider his personal responsibility and moral guilt before deciding to send him to die. Under Enmund, the only way to guarantee that such consideration has been given is to require the sentencer to determine that the defendant either killed, or attempted to kill, or intended to kill. That a jury might or could have made such a determination hardly provides a guarantee that this jury did. Because I believe every defendant is entitled to that guarantee, I would vacate the death sentence and remand the case with instructions to provide Bullock with a sentencing hearing before a jury. Inasmuch as the majority refuses to take this essential step, I dissent.
The Court’s attempt to distinguish Presnell on the ground that Spaziano v. Florida, 468 U. S. 447 (1984), rejected the assumption that juries have equivalent constitutional roles in determining guilt or innocence and in determining capital sentences, see ante, at 387-388, n. 4, is misplaced. Cole and Presnell hold that the Due Process Clause requires that appellate courts review convictions and sentences “as [they] were determined in the trial court.” Cole, 333 U. S., at 202 (emphasis added). Spaziano simply held that the Constitution permits trial judges, rather than trial juries, to make sentencing determinations. See infra, at 401, and n. 3.
The Court’s reliance on Sumner v. Mata, 449 U. S. 539 (1981), is misplaced. There, the Court held that the presumption of correctness accorded state-court findings of fact under 28 U. S. C. § 2254(d) extends to appellate findings as well as trial-court findings. 449 U. S., at 545-547. But the presumption of correctness is defeated by a showing that “the factfinding procedure employed by the State court was not adequate to afford a full and fair hearing.” 28 U. S. C. § 2254(d)(2). The question whether state procedures are “adequate” involves two distinct inquiries. The first is whether the procedure employed in a particular case in fact afforded' the defendant a full and fair hearing. The second is whether the procedure itself comports with due process. Bullock raises both those *400questions: he claims that in his case the Mississippi Supreme Court failed to use adequate procedures for making Enmund findings, and that a procedure which places the responsibility for making Enmund findings on the Mississippi Supreme Court is inherently inadequate. Sumner v. Mata does nothing to answer the latter question, because it assumes that the appellate court is constitutionally a proper factfinder. In Mata, this Court explicitly acknowledged that the trial-court record on which the California Court of Appeal based its findings concerning the suggestiveness of a photographic lineup was “completely adequate” for that purpose. 449 U. S., at 543. Sumner v. Mata therefore says nothing about how state-court findings are to be treated when the record on which they are based, by its very nature, is inadequate to permit factfinding in the first instance.
Moreover, the opinion in Mata does not concern itself with explaining when an appellate court is constitutionally incompetent to find facts. That an appellate court is not always a proper factfinder is clear beyond doubt. Surely, the Court would not read Sumner v. Mata to foreclose habeas relief in cases where an essential element of the offense was not found at trial. Cf. ante, at 384. In § 2254(d)(2)’s terms, a “factfinding procedure” that vested in appellate courts the responsibility for determining an element of the offense would not be constitutionally “adequate.” Similarly, I believe, the Enmund findings concern the kind of facts that can be found only by someone who has actually seen and heard the witnesses when they testified.
Every State with a death penalty statute has implicitly recognized this essential point, even though not all of them have explicitly held that Enmund findings must be made by the trial court. The seven States whose schemes involve judge sentencing all vest the power to impose sentence in a judge who actually has seen the presentation of evidence and confronted the defendant. See Ala. Code § 13A-5-46 (1982); Ariz. Rev. Stat. Ann. §13-703 (Supp. 1985); Fla. Stat. §921.141 (1985); Idaho Code §19-2515 (Supp. 1985); Ind. Code §35-50-2-9 (Supp. 1985); Mont. Code Ann. §46-18-301 (1985); Neb. Rev. Stat. §§29-2520 and 29-2521 *402(1979). No State has placed the sentencing power, as opposed to the power to review sentences, in an appellate court. Every State provides for an evidentiary sentencing hearing, to be conducted in front of the sentencing authority, be it judge or jury.
Enmund identified 17 States in which the then-existing death penalty statutes potentially countenanced the execution of defendants who neither killed, attempted to kill, or intended to kill: Arizona, California, Connecticut, Florida, Georgia, Idaho, Indiana, Mississippi, Montana, Nebraska, Nevada, North Carolina, Oklahoma, South Carolina, South Dakota, Tennessee, and Wyoming. See 458 U. S., at 789, n. 5; id., at 792, nn. 12 and 13. Since Enmund, seven of those States have addressed the issue and apparently have concluded that the sentencer must make Enmund findings before imposing sentence. See State v. McDaniel, 136 Ariz. 188, 199, 665 P. 2d 70, 81 (1983); People v. Garcia, 36 Cal. 3d 539, 556-557, 684 P. 2d 826, 835-837 (1984), cert. denied, 469 U. S. 1229 (1985); Allen v. State, 253 Ga. 390, 395, n. 3, 321 S. E. 2d 710, 715, n. 3 (1984), cert. denied, 470 U. S. 1059 (1985); Miss. Code Ann. § 99-19-101(7) (Supp. 1985); State v. Stokes, 308 N. C. 634, 651-652, 304 S. E. 2d 184, 195 (1983); Hatch v. Oklahoma, 662 P. 2d 1377, 1382-1383 (Okla. Crim. App. 1983); State v. Peterson, 287 S. C. 244, 248, 335 S. E. 2d 800, 802 (1985). Five others — Connecticut, Montana, Nebraska, Nevada, and South Dakota — have not yet considered cases raising an Enmund claim.
1 assume that many capital defendants who neither killed, attempted to kill, nor intended to kill take the stand, at least at the sentencing hearing, since they know that if they convince the sentencer of their diminished level of personal culpability their lives will be spared. The considerations of federalism and comity identified by the Court are hardly best served by allowing the State to construct capital-sentencing schemes that require federal habeas courts to examine in every case the nature of the evidence presented in order to determine whether the State’s regular capital-sentencing procedure is satisfactory. It is far better, it seems to me, to establish a bright-line rule requiring the findings to be made by the trial court, especially since the Court has failed to identify a single reason why a *407State legitimately could prefer to vest the factfinding function in an appellate court.
The Court’s refusal to “determine what factfinding procedures would be adequate in the particular case before us,” ante, at 388, n. 5, strikes me as somewhat perverse. Although most of the cases we hear concern broad legal questions the resolution of which will affect many persons other than the actual parties, this should not blind us to the fact that our authority to reach those questions rests on the presence of a concrete case. The question as to what procedures would be adequate in this ease is not, as the Court suggests, “hypothetical.” Ibid. The believability of Bullock’s testimony is the critical factor, and the credibility judgment can be made, in the first instance, only by someone who has seen him testify. If anything is “hypothetical,” it is the Court’s assumption that an appellate factfinding procedure that is clearly inadequate for the actual case before it will be adequate in hypothetical eases not before it.