(dissenting).
It is not the function of this court to try petitioner’s habeas corpus case de novo; our function is to review the decision of the district court denying the petition, and we are required to accept the factual findings of that court unless clearly erroneous.1 When that standard of review is applied, I cannot agree with the majority that the record in the case calls for reversal, and for that reason I dissent.
At the outset, I accept the historical facts of the case as set out in the majority opinion. I also accept the proposition that the failure of petitioner’s trial counsel to interview Brown, and perhaps other persons involved in the ease, cannot be justified by counsel’s general policy of never interviewing prospective witnesses for the prosecution. Such a policy is certainly impermissible, regardless of what else it may be called, and I agree with the district court that by whatever standard counsel’s conduct may be measured, he failed in the abstract to provide petitioner with the adequate representation of counsel that is called for by the sixth amendment to the Constitution of the United States as carried forward into the fourteenth amendment.
As the majority recognizes, however, the inquiry does not end at that point. It is also necessary to determine whether counsel’s breach of duty was so prejudicial to the petitioner as to require reversal of his conviction on federal constitutional grounds.
With respect to prejudice the majority says that it was not necessary for the petitioner to demonstrate by a preponderance of the evidence that adequate representation would have produced an acquittal, and that his burden was limited to “showing that the alleged error itself sufficiently undercut the reliability of the trial process to have prejudiced the petitioner’s right to a fair trial.” At 413-414. And the majority also says, “ . . .we must consider whether the effects of that breach of duty [to interview] raise a serious question as to the reliability of the trial process in this case.” At 414. For present purposes at least, I will accept the majority’s definition of the burden of proof as far as prejudice is concerned.
While the majority is critical of the failure of counsel to interview the young men who were in the company of the defendant and Brown before and after the killing of Dougherty and of the failure of counsel to interview the people whom the petitioner claimed to have seen at the home of Brown’s mother-in-law while he was allegedly looking for Brown, which interviews might or might not have been productive, and while the majority suggests that counsel would have done well to have tried to locate certain cab drivers who, if they had been found, might or might not have supplied information helpful to the defense, it seems to me that the majority’s decision on the issue of prejudice is based ultimately on the failure of counsel to interview Brown himself while Brown was confined in jail in St. Louis.
*418The majority is of the view that had Brown been interviewed, he perhaps would have recanted that part of his confession which implicated petitioner in the murder, and that such a recantation would have been of substantial help to petitioner at his trial regardless of whether or not Brown was called as a witness for either side.
Although the majority concedes that it cannot be said with assurance that Brown would have made such a recantation had he been interviewed by counsel (at 416, first sentence, and n.ll appearing on that page), nevertheless, the majority is of the view that the possibility of such a partial recantation was strong enough to cause the failure of counsel to interview Brown to result in a serious undercutting “of the reliability of the trial process which prejudiced petitioner’s right to a fair trial.” It is at this point that I disagree with the majority.
In my opinion, to say that there was a strong possibility, to say nothing of a probability, that Brown would have exculpated petitioner had Brown been interviewed by counsel, is to engage in pure speculation. And I do not think that a conviction, the validity of which has twice been considered by the Supreme Court of Missouri, in post-conviction proceedings and which has also been considered carefully by the district court should be overturned on such a basis.
Recantations of confessions and changes of story by convicts months or years after their convictions are by no means unusual. There are many motives for such recantations and changes, and they are highly untrustworthy.
I do not think that the Missouri courts were required or that the federal district court was required to accept or give any great weight to Brown’s belated affidavit and post-conviction testimony, including his statement that if he had been interviewed while in jail and had been called as a witness, he would have absolved petitioner of any involvement in the robbery and killing of the cab driver.
It must be remembered that Brown kept silence for some two years after petitioner was convicted. He explained his silence by saying that he expected help of an unspecified kind from two alleged accomplices, Anderson and Williams, the latter of whom is now dead, and that he decided to tell the truth after he became convinced that they were not going to afford him any help whether legitimate or illegitimate.
It cannot be ascertained at what point in time, if ever, Brown became convinced that he was not going to receive any help from Anderson and Williams, but the fact that he kept silence for two years justifies the inference that he was harboring his hope of assistance for a substantial period of time after petitioner’s trial and conviction, and there is simply no real reason to believe that he would have changed his original story had petitioner’s attorney talked to him before petitioner’s trial.
Let it be assumed, however, that Brown would have recanted had he been interviewed by counsel. While such recantation may have been of some value to defense counsel in cross-examining prosecution witnesses, it is clear that the most obvious help that Brown could have given petitioner would have been to take the stand and tell the jury that petitioner had left the cab before the killing and had nothing to do with it.
The trouble with that from petitioner’s standpoint is that apart from Brown the State had no witness that could from his own knowledge place petitioner in the cab when the driver was killed, and in fact the State was not able to place petitioner in the cab at the time of the killing by means of direct evidence. Had Brown been called as a witness, and had he exculpated petitioner, the State would have impeached him immediately by reference to his confession. Had the defense called Brown, it would have run the risk that the jury would not accept the recantation, and the jury would have gone to its deliberations with knowledge that Brown had implicated petitioner originally. The State did not call Brown, and when it failed to do so, it took a serious risk that the jury would not be convinced of the petitioner’s guilt on the strength of testimony of *419the witnesses who were called. That risk might well have been eliminated from a practical standpoint had Brown been called as a defense witness; however, had the defense called Brown, it would have taken a serious risk that Brown would recant his recantation and would testify directly that petitioner participated in the robbery that resulted in the killing.
What has been said to this point sufficiently states my views as far as the result of this particular case is concerned. I would affirm. However, in view of the recurrence of post-conviction contentions of inadequate representation by counsel at original trials, the advancing of which has become a standard operating procedure for convicts, and which are most troublesome to judges and embarrassing to the attorneys involved, I feel it desirable to go somewhat beyond this ease.
The focus of discussion here has been as to whether or not petitioner sustained prejudice as a result of his attorney’s dereliction of duty. However, it is clear from the unpublished opinion of District Judge Nangle that he had some trouble with the standard to be applied in determining whether or not there had been a breach of duty by petitioner’s counsel; and he called attention to a number of our decisions which, to use his term, make the applicable standard “a bit unclear.” The cases cited were: DeBerry v. Wolff, 513 F.2d 1336 (8th Cir. 1975); Wolfs v. Britton, 509 F.2d 304 (8th Cir. 1975); Johnson v. United States, 506 F.2d 640 (8th Cir. 1974), cert. denied, 420 U.S. 978, 95 S.Ct. 1404, 43 L.Ed.2d 659 (1975); Crenshaw v. Wolff, 504 F.2d 377 (8th Cir. 1974), cert. denied, 420 U.S. 966, 95 S.Ct. 1361, 43 L.Ed.2d 445 (1975); McQueen v. Swenson, 498 F.2d 207 (8th Cir. 1974), and Cardarella v. United States, 375 F.2d 222 (8th Cir.), cert. denied, 389 U.S. 882, 88 S.Ct. 129, 19 L.Ed.2d 176 (1967).
This writer also had trouble with the applicable standard in cases that he considered while on the district bench. In Frazier v. Roberts, 310 F.Supp. 504, 512 (E.D. Ark.1970), the writer cited with approval the then recent case of Scalf v. Bennett, 408 F.2d 325 (8th Cir.), cert. denied, 396 U.S. 887, 90 S.Ct. 175, 24 L.Ed.2d 161 (1969), wherein it was held, 408 F.2d at 327-28, that habeas corpus relief would not be granted on account of incompetency of counsel or denial of effective counsel unless the trial was a farce or mockery of justice, or was shocking to the conscience of the reviewing court, or unless the purported representation was only perfunctory, in bad faith, a sham, a pretense, or without adequate opportunity for conference and preparation.2
A few years later, the writer thought, largely on the strength of McQueen v. Swenson, supra, that the standard announced in Scalf v. Bennett, supra, had been replaced by a “reasonable competency” standard such as had been adopted in other circuits. Clark v. Lockhart, 379 F.Supp. 1320, 1329 (E.D.Ark.1974), aff’d, 512 F.2d 235 (8th Cir. 1975), cert. denied, 423 U.S. 872, 96 S.Ct. 139, 46 L.Ed.2d 103 (1976).3
Post-McQueen cases mentioned by the district court and in the majority opinion of this court indicate that we may have returned to the “farce and mockery of justice” standard of the older cases if, indeed, we ever departed from it, and that we are at least paying lip service to that standard. I fear, however, that what we may really be doing is deciding cases of this kind by hindsight, on a purely ad hoc basis, and without any real reference to any particular standard of the quality of representation constitutionally required or relating to the extent of prejudice that will require a reversal if representation has been inadequate constitutionally. Such an approach, if it is being taken, is in my view undesirable and unfair to district judges, to state courts, and to *420lawyers who represent defendants in criminal cases.
I recognize that it is extremely difficult to articulate general standards to be applied in this area, and I also recognize that whether a given representation has been constitutionally inadequate and, if so, whether the inadequacy has so prejudiced the defendant as to call for a reversal, must necessarily depend upon a number of factors which vary from case to case. But, I also think that a district judge called upon to decide a case of this kind should have some guidelines to follow, and some assurance as to how his decision will be reviewed on appeal.
In my opinion, the standard laid down in Cardarella v. United States, supra, as amplified to some extent in Scalf v. Bennett, supra, is an adequate and proper standard to be followed in appraising the constitutional adequacy of representation by counsel in a given case. And I think that this court should announce its adherence to that standard, or to some other standard if the court prefers, and that thereafter we should confine ourselves to exercising our proper appellate function in reviewing decisions of the district court in cases in which they have been called upon to apply that standard. On the other hand, it behooves the district courts to make it clear in their findings or opinions that they have considered the question before them in the light of the prescribed standard.
. Fed.R.Civ.P. 52(a); Brown v. Swenson, 487 F.2d 1236 (8th Cir. 1973), cert. denied, 416 U.S. 944, 94 S.ct. 1952, 40 L.Ed.2d 296 (1974); Taylor v. Swenson, 458 F.2d 593 (8th Cir. 1972).
. Frazier v. Roberts, supra, was reversed by this court, 441 F.2d 1224 (8th Cir. 1971). However, this court did not refer to the standard of representation mentioned by the district court.
. Again, this court did not discuss the standard employed by the district court.