concurring dubitante.
This appeal presents two different questions relating to the reliability of the complainant’s identification of L.G.T. as the person who stabbed and robbed her. The trial judge having found that the photograph of the lineup from which the complainant selected L.G.T. was unduly suggestive,1 the first issue is whether the identification was sufficiently reliable, notwithstanding the suggestive lineup, to support its admission into evidence. Although the point is arguable — it is hard to be sure that the flawed lineup did not “give rise to a very substantial likelihood of irreparable misidentification,” Neil v. Biggers, 409 U.S. 188, 197, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972) — I do not believe that it was unreasonable for the judge to deny L.G.T.’s motion to suppress the identification evidence. I therefore agree that the admission of the pretrial identification was proper, and I will not address that issue further.
The second question presented is whether the identification of L.G.T. from the lineup, and the in-court identification that followed, were sufficiently reliable, notwithstanding the virtual lack of any corroboration, to support the adjudication of guilt. Admissibility and sufficiency are, of course, entirely different issues; an item *962of evidence may be reliable enough to warrant its submission to the trier of fact for his or her consideration, but may still be insufficient, without more, to support a finding of guilt beyond a reasonable doubt. See, e.g., United States v. Hunter, 692 A.2d 1370, 1376 (D.C.1997). For the reasons stated below, I find the sufficiency issue very difficult. Nevertheless, I reluctantly vote to affirm.
The complainant identified L.G.T. from the unduly suggestive lineup photograph seven months after the assault and robbery. At that time, she told the detective, and stated in writing, that No. 6 in the lineup (L.G.T.) “very clearly looks like the man who stabbed me.” (Emphasis added.) Although the complaining witness subsequently asserted that she was certain of her identification, her initial words aré important because many people resemble each other, because witnesses tend to become more and more “certain” as the case progresses, especially if the police rely on their identifications,2 and because the complainant made it quite clear during the course of her testimony that she recognized the critical difference between an assertion that a person “looks like” the assailant and a statement that he “is” the assailant. Significantly, the lineup photo did not include either of the young men who, the complainant had said five months earlier, “might be” the person who stabbed and robbed her. One wonders whether she would have identified L.G.T. as the robber if photographs of all three individuals had been presented to her simultaneously.
The identification on which the adjudication of guilt in this ease was based involved a complete stranger whom the complainant had never seen before. She testified that at the time of the offense, she had observed her assailant for approximately two seconds through the side-view mirror of her moving automobile and for an estimated seven' seconds while the robbery itself was in progress. The situation was obviously a stressful one. The complainant was initially unable to describe the attacker’s clothing, and if L.G.T. in fact committed the crime, then she mis-estimated his height by five inches.3
This court and other courts have repeatedly recognized the unreliability of identifications of strangers made on the basis of brief observation under stressful conditions. See, e.g., Webster v. United States, 623 A.2d 1198, 1204 n. 15 (D.C.1993); Crawley v. United States, 320 A.2d 309, 312 (D.C.1974); cf. United States v. Wade, 388 U.S. 218, 228, 87 S.Ct. 1926, 18 L.Ed.2d 1149 (1967); see also Nathan R. Sobel, Eyewitness Identifioation, Legal AND PRACTICAL PROBLEMS § 1.1, at 1.1-1.3 (2d ed.1998) (hereinafter Sobel)4 and authorities there cited. In the words of Judge Lumbard:
Centuries of experience in the administration of criminal justice have shown that convictions based solely on testimony that identifies a defendant previously unknown to the witness [are] highly suspect. Of all the various kinds of evidence it is the least reliable, especially where unsupported by corroborating evidence.
Jackson v. Fogg, 589 F.2d 108, 112 (2d Cir.1978). “[T]he very real danger of mistaken identification is a threat to justice.” United States v. Greer, 176 U.S.App.D.C. *96389, 94, 588 F.2d 437, 442 (1976) (citations and internal quotation marks omitted).
“Scientific evidence concerning the unreliability of eyewitness identification has continued to mount since the Wade trilogy.” Sobel, supra, § 1.1, at 1.2 & n. 7 (citing psychological and other studies).5 In this case, for example, the prosecutor argued, and the trial judge effectively found, that the complainant’s mental image of the robbery remained imprinted in her recollection. But “[c]ontrary to popular understanding, our eyes and memories do not operate like a camera on which events are accurately recorded subject to retrieval at any time, but in fact memory can be altered to a significant extent by information perceived after the fact.” Id. § 1.1, at 1.2-1.3 (citing State v. Contreras, 674 P.2d 792, 801 (Alaska Ct.App.1983)).6
The trial judge relied heavily on the complainant’s lack of doubt regarding her identification of L.G.T. The Supreme Court has held that the identifying witness’ level of certainty is a legitimate factor in the reliability calculus. See Manson v. Brathwaite, 432 U.S. 98, 114, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977). Indeed, the contrary would be counter-intuitive, for “he is the one who did it” surely strikes one as more probative than “he may be the one.” Nevertheless, the research indicates that there is little if any relationship between the expressed confidence of an eyewitness and the accuracy of the identification. Sobel, supra, § 6.8, at 6.37 n. 2 (citations omitted). “[I]t is well recognized that the most positive eyewitness is not necessarily the most reliable.” Crawley, supra, 320 A.2d at 312 (citations omitted). Indeed, “[sjome studies have shown a negative correlation between professed certainty and accuracy, and ‘positive’ has been defined as ‘mistaken at the top of one’s voice.” ’ Webster, supra, 623 A.2d at 1204 Moreover, the complainant’s initial reaction to L.G.T.’s photograph in the unduly suggestive lineup was that L.G.T. “very clearly look[ed] like” her assailant. The assertions of certainty that he was the attacker came later. n. 15 (citation omitted).
Finally, the probative value of the complaining witness’ in-court identification should be assessed in light of the suggestive character of the lineup photo, for
[ijt is a matter of common experience that, once a witness has picked out the accused at the line-up, he is not likely to go back on his word later on, so that in practice the issue of identity may (in the absence of other relevant evidence) for all practical purposes be determined there and then, before the trial.
Wade, supra, 388 U.S. at 229, 87 S.Ct. 1926 (citation and internal quotation marks omitted). Indeed “the influence of improper suggestion upon identifying witnesses probably accounts for more miscarriages of justice than any other single factor — perhaps it is responsible for more such errors than all other factors combined.” Id. (citation and internal quotation marks omitted). The unduly suggestive lineup guided the complainant’s attention to L.G.T., and there is substantial danger that it brought about her “positive” identifications.
In light of the foregoing considerations, I perceive a logical disconnect between, on the one hand, an essentially uncorroborated identification seven months after the fact, based on a brief and stressful period of observation and a seriously flawed lineup, and on the other hand, the requirement that guilt be proved beyond a reasonable doubt, i.e., to a moral certainty. I recognize that I am not the trial judge, but appellate judges also have a significant *964substantive role to play. This court has explained that
while we are mindful that the record on review is “cold,” we think it proper — and indeed necessary — for us to draw upon our own experience, value judgments, and common sense in determining whether the verdict reached was in keeping with the facts.
Crawley, supra, 320 A.2d at 312 (citation omitted). Because, as the Supreme Court has noted, eyewitness identification of strangers has led to the conviction and incarceration of many innocent defendants, routinely deferential review on the part of appellate courts risks further injustices- in the future. It is important, in my view, for judges not only to be aware of the extensive case law explaining the problems with this type of testimony, but also to incorporate these authorities into their decision-making calculus.
Nevertheless, there are strong arguments for affirmance. First, the record shows that the trial judge gave careful and thoughtful consideration to the issues of fact and law presented, and that he decided them as fairly and conscientiously as he could. He heard the evidence, which included not only the testimony of the complainant, whom the judge quite reasonably found to be exceptionally believable, but also L.G.T., whom the judge expressly found not to be credible. Second, the complainant testified persuasively that she took her testimonial responsibility very seriously because, above all, she did not want to make a false accusation against an innocent person.7 Third, we are required to view the record in the light most favorable to the District, id., and we may reverse the judgment only if no reasonable trier of fact could find guilt beyond a reasonable doubt. See In re T.M., 517 A.2d 1149, 1151 (D.C.1990). Although I apprehend that the, judge may have accorded insufficient weight to the perils of uncorroborated eyewitness identification in cases such as this, he was obviously a reasonable and fair-minded arbiter. Fourth, the essential teaching of Manson and other Supreme Court decisions seems to be that, notwithstanding the problems with eyewitness identification of strangers following brief and stressful encounters, the ultimate decision should be left to the trier of fact. Fifth, the proposition on which my separate opinion is based— namely, that the admissibility of an identification does not establish its sufficiency to sustain a finding of guilt beyond a reasonable doubt — was not raised at all by counsel for L.G.T.8 Finally, notwithstanding my misgivings about this case, and in spite of my apprehension that a doctrine which sustains findings of guilt on the type of evidence in this record may well invite unintentional miscarriages of justice in future cases, I find it difficult to articulate a consistent rule of law which would support reversal here without doing violence, Crawley notwithstanding, to the proper allocation of authority as between trial and appellate courts.9 Accordingly, and with considerable reluctance, I concur in the judgment.
. It appears that the other persons in the lineup were police officers considerably older that L.G.T.
."Indeed, experience establishes that an uncertain identification becomes more and more positive at every stage of the proceeding through the trial itself.” In re Dwayne W., 109 Daily Wash. L. Rptr. 1901, 1906 n. 11 (Super.Ct.D.C.1981) (quoting Nathan R. Sobel, Eyewitness Identification, Legal and Practical Problems 9 (1972)); see also Fredric D. Woocher, Note, Did Your Eyes Deceive You? Expert Psychological Testimony on the Unreliability of Eyewitness Identification, 29 Stan. L. Rev. 969, 985 (1977).
. The complaining witness explained that the assailant was on a small bicycle and that this made it difficult to estimate his height.
. This is the most recent edition of the treatise by Judge Sobel to which I refer in footnote 2.
. I note, however, that reliance on scientific writings poses an obvious problem, for the psychologist or other scientist cited by the court will not have been subject to cross-examination, and other experts may have different views.
. The Contreras opinion surveys, and quotes at length from, the scientific literature.
.The witness stated:
I wanted to make every effort to make sure that I didn’t identify the wrong person. Somebody's life is at stake. I value that quite seriously.
The judge credited the complainant, and no reasonable person reading the transcript would doubt her honesty. The question is whether the fact that she honestly believed to be true was true in fact.
. Counsel’s entire insufficiency argument in this court is that the identification was too unreliable to warrant its admission and that it therefore did not support the adjudication of guilt.
. But a rule leaving it all to the trier of fact is not satisfactory either.