It would be amusing, if it were not so very serious, to imagine for a moment the majority’s visit to a haberdashery. The visit would be a fun one, because my friends in the majority would try on every hat in the shop, except, of course, the one that might conceivably fit. Here, the majority dons the head-wear of the jury, the state trial court, the state appellate court, the state post-conviction relief (PCR) court, and the federal district court, but then inexplicably leaves the premises without a passing glance at the cap befitting federal appellate judges reviewing under AEDPA the considered judgment of a state court that a defendant’s counsel was not ineffective and that there was no prejudice arising from that counsel’s allegedly deficient performance.
The majority spends a considerable amount of time defending its conclusion that Edward Lee Elmore is entitled to habeas relief on his ineffective assistance of counsel claim. But as Mark Twain is reputed to have said, “The more you explain it, the more I don’t understand it.” SEC v. Chenery Corp., 332 U.S. 194, 214, 67 S.Ct. 1760, 91 L.Ed. 1995 (1947) (Jackson, J., dissenting). Simply put, the majority’s rejection of the South Carolina PCR court’s determination that defense counsel’s alleged deficiencies did not result in prejudice cannot be squared with the deferential standards required under AEDPA, the facts of this case, or Supreme Court precedent. And in the course of its decision, the majority unjustly impugns the criminal justice system of South Carolina, slanders a deceased man who simply had the misfortune of discovering his neighbor’s mutilated body, and grants habeas relief to a prisoner whom overwhelming evidence suggests brutally raped and murdered an elderly woman in her home. For these reasons, I respectfully dissent.
I.
A.
Under the two-prong test set forth in Strickland v. Washington, 466 U.S. 668, *874104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), Elmore must show not only that his “counsel’s representation fell below an objective standard of reasonableness,” id. at 688, 104 S.Ct. 2052, but also that there was “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694, 104 S.Ct. 2052.
According to Strickland, courts should resolve ineffective assistance claims under the prejudice prong whenever possible. See id. at 697, 104 S.Ct. 2052 (“If it is easier to dispose of an ineffectiveness claim on the ground of lack of sufficient prejudice, which we expect will often be so, that course should be followed.”); see also Fields v. Attorney Gen., 956 F.2d 1290, 1297 (4th Cir.1992) (“If the defendant cannot demonstrate the requisite prejudice, a reviewing court need not consider the performance prong.”). It is clear beyond peradventure that whatever defense counsel’s alleged failings, the outcome of Elmore’s 1984 trial would almost certainly have been the same.1
B.
Only if one is prepared to throw Strickland deference to the winds can prejudice to Elmore be discerned. The majority’s approach here notwithstanding, we are neither the first court, nor the best positioned one, to consider Elmore’s claim. In collateral review, state courts should not be subordinate courts, and the procedural history in fact places a heavy duty of deference upon us. The standard of review here is among the best known in all of law. As amended by AEDPA, 28 U.S.C. § 2254 provides that “[a]n application for a writ of habeas corpus on behalf of a person in custody pursuant to the judgment of a State court shall not be granted” unless the state court’s decision either “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court,” 28 U.S.C. § 2254(d)(1), or was “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” Id. § 2254(d)(2). Unless it falls under one of these narrow exceptions, we cannot set aside a state court’s judgment on collateral attack.
It is not as though the state court missed the point. As the majority acknowledges, the state PCR court recognized that the clearly established law of Strickland governed Elmore’s claims. Ante at 856. The majority nevertheless concludes that the PCR court both “acted contrary to” and “unreasonably applied” Strickland in its prejudice analysis. Id. at 871. Specifically, it believes the PCR court contravened Strickland’s mandate to “consider the totality of the evidence,” id. at 868 (quoting Strickland, 466 U.S. at 695, 104 S.Ct. 2052), and that, “[p]roperly applied, the totality-of-the-evidence standard results in only one reasonable conclusion,” *875namely, that defense counsel’s alleged errors resulted in prejudice. Id. at 869.
The Supreme Court and Congress notwithstanding, the majority’s view is de novo in disguise. The PCR court’s application of Strickland’s prejudice prong cannot fall under the “contrary to” clause of § 2254(d)(1). To qualify under this provision, a state court must either “arrive[ ] at a conclusion opposite to that reached by [the Supreme] Court on a question of law” or “deeide[] a case differently than [the Supreme] Court has on a set of materially indistinguishable facts.” Williams v. Taylor, 529 U.S. 362, 412-13, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000). In dismissing Elmore’s prejudice claim, the PCR court committed neither of these two errors. As for the first, according to Strickland itself, the prejudice inquiry is a “mixed question[ ] of law and fact,” 466 U.S. at 698, 104 S.Ct. 2052, and the majority’s rejection of the PCR court’s analysis is heavily focused on factual considerations. See ante at 869-71. As for the second, the majority fails to provide a materially indistinguishable precedent. At best, it offers a short string-cite of cases, a number of which involved de novo rather than deferential review. See id. at 867-68. At the end of the day, the majority’s conflation of these two clauses only muddles the matter before us.
At bottom, the majority’s two attacks on the PCR court are really a single claim that the state court unreasonably applied Strickland by refusing to find prejudice in this case. This conclusion, however, cannot be squared with the deference required under § 2254. When applying the “unreasonable application” clause, it is important to remember that “an unreasonable application of federal law is different from an incorrect application of federal law.” Williams, 529 U.S. at 410, 120 S.Ct. 1495 (emphasis in original). As the Supreme Court has stated time and again, “[E]ven a strong case for relief does not mean the state court’s contrary conclusion was unreasonable.” See, e.g., Harrington v. Richter, — U.S. —, 131 S.Ct. 770, 786, 178 L.Ed.2d 624 (2011). We simply cannot overturn on collateral attack a state court’s conclusion that a claim lacks merit “so long as fairminded jurists could disagree on the correctness of the state court’s decision.” Id. (internal quotation marks omitted). This statute creates a “highly deferential standard for evaluating state-court rulings” that “demands that state-court decisions be given the benefit of the doubt.” Woodford v. Visciotti, 537 U.S. 19, 24, 123 S.Ct. 357, 154 L.Ed.2d 279 (2002) (per curiam) (internal quotation marks omitted).
It is not as though the Supreme Court has uttered these sentiments only once. It has repeated them time and time again, in part because actions of lower federal courts have made that repetition necessary. Standards of review are not advisory. They are not catechisms whose repetition is designed to make us numb. AEDPA commands federal courts to show respect for state court factual findings in order “to further the principles of comity, finality, and federalism.” Williams, 529 U.S. at 436, 120 S.Ct. 1479. These findings “shall be presumed to be correct,” and can only be overcome “by clear and convincing evidence.” 28 U.S.C. § 2254(e)(1). The South Carolina courts did not kiss this case off. Despite having at its disposal a 183-page state PCR court opinion rejecting sixteen separate allegations of error, the majority takes it upon itself to adjudicate Elmore’s Strickland claims anew and overturn a litany of state court factual findings. Like overreaching appellate courts in the past, the majority simply has given “ § 2254(d) no operation or function in its reasoning.” Richter, 131 S.Ct. at 787.
*876If the requirements of § 2254 were not enough to send a strong cautionary signal, Strickland itself imposes an additional level of deference for reviewing ineffective assistance claims. “The standards created by Strickland and § 2254(d) are both highly deferential, and when the two apply in tandem, review is doubly so.” Id. at 788 (citations omitted). On the performance prong, courts are not permitted to “second-guess counsel’s assistance” or to engage in an “intrusive post-trial inquiry” due to the fact that an appellate court’s hindsight often fails to take account of the actual circumstances on the ground. See Strickland, 466 U.S. at 689-90, 104 S.Ct. 2052. And as for the prejudice prong, there must be a “substantial, not just conceivable,” likelihood that counsel’s alleged errors led to a different result. Cullen v. Pinholster, — U.S. —, 131 S.Ct. 1388, 1403, 179 L.Ed.2d 557 (2011) (internal quotation marks omitted).
These two prongs — and the deferential approach required by each — necessarily overlap. See Correll v. Ryan, 539 F.3d 938, 951 (9th Cir.2008) (“[Djeficient performance and prejudice questions may be closely related.”). It is therefore unsurprising that courts regularly apply the “doubly deferential” standard of Strickland and AEDPA to both the performance and prejudice prongs. See, e.g., Cullen, 131 S.Ct. at 1410-11; Foust v. Houk, 655 F.3d 524, 534 (6th Cir.2011). This makes good sense. If appellate courts must be wary of deeming counsel’s conduct to be deficient, they should also be cautious before deciding that that conduct actually resulted in prejudice.
The majority pays lip service to these principles, but their ultimate role in its decision is limited to perfunctory opening remarks. See ante at 849-51. Had the majority begun its analysis by seriously applying the doubly deferential standard required by Strickland and AEDPA to the PCR court’s adjudication of Elmore’s prejudice claims, it would have quickly concluded that it could not grant relief. Instead, it decided to engage in what amounts to yet another trial of Elmore, only this time without a jury, live evidence, or the other accoutrements of adversarial process that give a trial its rightful name. It is not unusual that sound principles of legal structure fall into disrepair in a habeas case. See, e.g., Richter, 131 S.Ct. at 792 (reversing a federal appellate court’s refusal to show deference to a state court’s adjudication of a Strickland claim on collateral attack). But I am sorry to see an example so extreme.
II.
Three basic facts demonstrate that the majority’s rejection of the PCR court’s prejudice analysis ignores the principles of deference set down by Congress and the Supreme Court. First, my friends in the majority stand in a lonely, lonely place. No court — either state or federal — has ever before found that any of defense counsel’s alleged errors prejudiced Elmore. Second, unlike the juries that convicted Elmore and the PCR court that upheld that conviction, the majority has not heard a shred of witness testimony. Finally, the majority disregards the actual evidence and spins a fanciful conspiracy theory that cannot be squared with the facts of this case. These are all symptoms of a court doing its own thing rather than one mindful of its role when assessing Strickland claims on collateral attack.
A.
To begin, the majority’s decision runs up against the striking fact that before today, no court had ever found that the conduct of Elmore’s counsel resulted in constitutional prejudice. Instead, after examining *877this issue in extensive detail, the state PCR court, the federal magistrate judge, and the federal district court each concluded that the Elmore could not make this showing. This fact is significant because we can only reverse the PCR court’s decision here if there is no possibility that “fairminded jurists could disagree.” Richter, 131 S.Ct. at 786. While the previous federal opinions impose on us no duty of deference as state opinions do, see Conner v. Polk, 407 F.3d 198, 204 (4th Cir.2005), they demonstrate that the majority has reached the remarkable conclusion that every single judge to have previously considered this issue has been unreasonable. The only reasonable conclusion, says the majority, is its own.
The majority focuses on four evidentiary issues to support its claim that the supposed failings of Elmore’s counsel resulted in prejudice. First, it contends that had defense counsel more fully investigated Dr. Conradi’s expert opinion regarding Mrs. Edwards’s time of death, the jury would have discovered that it was more likely her murder occurred on Sunday afternoon rather than Saturday night. Ante at 870. But the state PCR court rejected this precise argument. It began by noting that even Elmore’s own expert “conceded that there was a possibility that the time of death could have been on the Saturday night.” It then reiterated the strong circumstantial evidence supporting a Saturday night time of death. Mrs. Edwards had planned to leave on a trip Sunday morning, a trip she never had the opportunity to undertake. Instead, on Sunday, her car remained parked at her home. When her body was discovered Monday afternoon, the television set was on with a TV Guide open to Saturday evening. The Sunday and Monday newspapers were lying in her driveway. In light of this evidence, it is little wonder that the PCR court determined that Elmore “has failed to show a reasonable probability that had a pathologist be[en] retained in 1984, the result of the proceeding would have been not guilty.”
Neither the magistrate judge nor the district court found this conclusion to be unreasonable. The magistrate judge, for instance, concluded that Elmore “has failed to show a reasonable probability that the result would have been different.” And the district court pointed out that unlike Elmore’s expert, Dr. Conradi — the forensic pathologist who performed Mrs. Edwards’s autopsy and testified that the time of death was likely Saturday night— “had first-hand information from the autopsy which credibly supported her analysis,” such as the limited extent of rigor mortis present when she examined the body. Moreover, the district court noted that Dr. Conradi “took into account factors that [Elmore’s expert] assumed she had missed, such as body temperature, the extent of decomposition, and the fact that the body had been transported.” In this battle of the experts, it is hard to see how the majority could dismiss the reasoned conclusions of every previous court and declare that Elmore’s witness was the clear winner.
Second, the majority claims that had defense counsel pursued the theory that the police lied about discovering forty-five of Elmore’s pubic hairs in Mrs. Edwards’s bed, there was a reasonable probability that his client would have been acquitted. See ante at 870-71. Once again, every court to have previously considered the evidence has come to the opposite conclusion. After hearing testimony from the SLED agents who discovered the pubic hairs, the state PCR court determined that they “credibly testified about the existence and discovery of the hair.” It then found that there was no break in the chain of custody of that evidence, as the hairs were *878secured in a zip lock bag at the crime scene and delivered to Lieutenant Earl Wells who kept them until trial. The court concluded that Elmore’s “bare assertions” that there was a break in the chain of custody raising the possibility of tampering were “without merit” and dismissed this Strickland claim.
The magistrate judge similarly refused to find prejudice and adopted the state court’s finding that Elmore “has not shown a break in the chain of evidence with regard to the hair.” The district court reached the same conclusion. Noting that “the PCR court made very specific factual findings regarding the credibility of the state’s witnesses” and “the collection of hairs from the victim’s bed” and had considered Elmore’s arguments, it refused to overturn the state court’s determination that Elmore had not been prejudiced. Observing that Elmore “has failed to establish that there was any break in the chain of custody,” it concluded that the PCR court’s determination here was simply “not a decision that was contrary to, or involved an unreasonable application of clearly established federal law.” The majority should have followed that course today.
Third, the majority claims that defense counsel’s decision not to investigate three unidentified fingerprints on Mrs. Edwards’s toilet and exterior door frame was prejudicial. See ante at 870-71. During the PCR proceedings, Elmore’s experts argued that these prints formerly considered unidentifiable were in fact identifiable, and not a match for Elmore. But as the PCR court concluded, this did not result in any constitutional prejudice. As it observed, the jury was well aware that Elmore’s fingerprints were found only on Mrs. Edwards’s outside doorframe and not within her house. Given that these new arguments would have not told the jury anything new, the PCR court concluded that Elmore’s prejudice argument was so unconvincing as to “strain[ ] his own credibility.”
For similar reasons, the magistrate judge found that Elmore was not prejudiced, noting that the experts at the PCR hearings disagreed “about the ability to identify one of those prints.” The district court agreed and dismissed Elmore’s “assertion that fingerprints of an unknown third party were found at the crime scene” as “incorrect.” In particular, it observed that Elmore’s own expert could not compare the toilet print with the victim’s fingerprints and consequently was unable to rule out the distinct possibility that the print inside Mrs. Edwards’s home was in fact Mrs. Edwards’s own.
Fourth, the majority contends that defense counsel’s lack of investigation into the presence of a Caucasian hair in “Item T” — the material found on Mrs. Edwards’s chest that was originally (and mistakenly) identified as blue fibers — was prejudicial. See ante at 870-71. This evidence came to light late, after most state post-conviction proceedings had come to a close, and it ultimately extended litigation in state court by several years and necessitated a separate opinion.
That state court opinion, however, made clear that this new evidence did not amount to much. To start, the PCR court found that the initial failure to disclose the hair by SLED Agent Wells was not intentional. As it observed, Elmore’s “position that Agent Wells knowingly withheld this evidence is refuted by ... the fact that it was he, not another source, who discovered the missing evidence and reported the discovery” to the South Carolina Attorney General’s Office. In addition, Wells repeatedly “offered himself for cross examination” in these new proceedings, but Elmore refused to take advantage of the opportunity. As the PCR court concluded, *879“I do not find that SLED Agent Wells knowingly failed to disclose the hair evidence and accept his sworn statement that his previous testimony was incorrect but was based upon his then recollection.”
Moreover, the state court observed that there was only one strand of Caucasian hair in the set that did not belong to the victim. Of course, a body lying on or flailing about the floor is bound to pick up a stray hair or two. As the PCR court concluded, “One hair from the victim’s body from the bedroom floor that could have come from various sources does not mandate a new trial.” The magistrate judge reached a similar conclusion and the district court did so as well, noting that a “single third-party Caucasian hair found on the victim’s body is greatly outweighed by the substantial evidence against [Elmore].” As I will discuss in greater detail, that substantial evidence includes the pubic hairs belonging to Elmore found on the victim’s bed, the blood matching Mrs. Edwards’s blood type found on Elmore’s clothes, his lack of a consistent alibi for his whereabouts on Saturday night, his lie to the police that he did not know Mrs. Edwards, and his confession to his cellmate. Against this mountain of evidence, Elmore is able to construct no more than a small hill.
On each one of these four claims of prejudice, the state PCR court, the magistrate judge, and the district court reached the opposite conclusion from the one the majority puts forth today. The majority, however, runs roughshod over these findings, dismissing, for example, the circumstantial evidence in support of Dr. Conradi’s analysis as “staged,” ante at 870, despite the fact that no court has ever found anything that would support this extreme allegation. And even though the majority admits that the “value of the toilet print to Elmore is somewhat circumscribed,” it still contends that defense counsel’s decision not to investigate this print fully resulted in prejudice because it constituted “evidence of police ineptitude and deceit.” Id. at 871. I have no idea where the majority is coming up with these things. Disparaging rhetoric cannot substitute for evidence, the foundation on which the PCR court’s factual findings rest. Nor can it displace the credibility determinations that are uniquely a trial court’s to make. Such speculative reasoning does not exhibit deference of any kind, let alone the doubly deferential standard of review mandated by Strickland and AEDPA.
B.
The majority not only disregards the considered judgment of three separate courts, but does so without hearing a single second of witness testimony. But with only the written record in front of us, we as appellate judges are in the worst possible position to evaluate witness credibility. As the Supreme Court has observed: “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.” Marshall v. Lonberger, 459 U.S. 422, 434, 103 S.Ct. 843, 74 L.Ed.2d 646 (1983) (citation omitted). We must therefore always approach a lower court’s factual determinations with an eye toward our inherent limitations. This is especially true on collateral attack, where principles of comity and federalism command that we grant the factual findings of state courts great respect. See Miller-El v. Cockrell, 537 U.S. 322, 339-40, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003).
In this case, however, these warnings have fallen upon deaf ears. Despite the *880fact that every jury and judge to have heard witness testimony has not been convinced of Elmore’s innocence, the majority suggests that he was an innocent man framed by the police and prejudiced by his attorney’s supposed failings. But unlike the two juries that convicted Elmore or the state PCR court that rejected his Strickland claims, the majority has observed not one of the following witnesses testify:
• Jimmy Holloway, the neighbor of Mrs. Edwards who discovered her mutilated body and the man the majority now suggests is responsible for her murder. See ante at 855, 870.
• Dr. Sandra Conradi, the forensic pathologist who performed Mrs. Edwards’s autopsy and estimated that her time of death was Saturday night.
• Dr. Jonathan Arden, a New York forensic pathologist who neither performed Mrs. Edwards’s autopsy nor spoke with Dr. Conradi, but who concluded that the likelihood of a Saturday night time of death was highly unlikely, though not “physically impossible.”
• Lieutenant Frank DeFreese, the SLED fingerprint analyst who saw the pubic hairs in Mrs. Edwards’s bed as well as lifted Elmore’s fingerprint from the frame of the back door to Mrs. Edwards’s home and an officer the majority now suggests committed perjury. See id. at 870-71.
• Four officers of the Greenwood Police Department who confirmed De-Freese’s impression that Elmore’s fingerprint on the door frame was “recent” or “fresh.” See id. at 794.
• Ira Parnell, the SLED agent who assisted DeFreese at the crime scene by collecting forty-five of Elmore’s pubic hairs from Mrs. Edwards’s bed and who, according to the PCR court, “credibly testified about the existence and discovery of the hair.”
• Lieutenant Thomas Henderson, the SLED agent who questioned Elmore about the blood on his shoes and the man the majority now suggests “plant[ed]” that evidence. See id. at 834.
• John C. Barron, the SLED serologist who determined that some of the blood on Elmore’s clothing matched Mrs. Edwards’s blood type and testified “without equivocation” that he maintained control of the stained clothing until trial.
• Rodger Morrison, an Alabama serologist who criticized Barron’s analysis and hypothesized that further testing could have ruled out the possibility that Mrs. Edwards’s blood was on Elmore’s clothes. But see infra note 2 (observing that DNA testing has since confirmed that the blood on Elmore’s clothes matched Mrs. Edwards’s DNA).
• Hayward Starling, a self-employed forensic scientist who criticized SLED’s investigative procedures and claimed that the print lifted from the blood-smeared toilet was identifiable, but who “could not eliminate” the possibility that the print belonged to Mrs. Edwards.
• Lieutenant Earl Wells, the SLED chemist who examined the hair evidence and an agent the majority now accuses of “negligently or knowingly misreport[ing]” evidence despite the PCR court’s findings to the contrary. See ante at 870.
• Skip Palenik, a self-employed microscopist who argued that the number of pubic hairs found on Mrs. Edwards’s bed was atypical and unusual.
*881• James Gilliam, the inmate who testified that Elmore confessed to him that “he went there to rob the lady, and she started screaming, so he had to kill her,” and whose later recantation the PCR court found lacked credibility-
• Arlie Capps, the jail administrator Elmore accuses of inducing Gilliam to testify falsely but whose testimony the PCR court found to be “consistently credible.”
• Sergeant Alvin Johnson, the police officer who testified that neither he nor any other officer had ever discussed Elmore with Gilliam prior to receiving Gilliam’s letter detailing Elmore’s confession.
• Mary Alice Harris (nee Dunlap), the former girlfriend of Elmore who threw his shirt in the trash when he appeared at her mother’s home with a swollen lip after midnight on Sunday, January 17.
• Donnie, Susan, and Frances Mosley, the members of Dunlap’s family who each confirmed Elmore’s time of arrival that evening.
• Major James Coursey, the police officer who testified that at the time of his arrest, Elmore repeatedly denied knowing Mrs. Edwards until confronted with one of her checks made out to him.
• Dr. Jonathan Venn, the psychologist who contended that Elmore’s statement that he did not know Mrs. Edwards was insignificant in light of his memory problems.
• Geddes Anderson, John Beasley, and Billy Garrett, the attorneys who represented Elmore in his three trials and who testified extensively about their defense strategies.
• Edward Lee Elmore, the defendant himself.
Even the above sources do not exhaust the list. Despite not hearing a second of testimony from this roster of witnesses— some of whom testified at trial, some of whom testified at the PCR proceedings, and some of whom testified at both — the majority treats Elmore’s experts who never participated in the original investigation as unbiased oracles of truth, see ante at 803-20, while suggesting that some of the prosecution’s key witnesses “were outright dishonest.” Id. at 870. This analysis ignores the Supreme Court’s admonition that we are ill-equipped to engage in such sensitive credibility determinations. See Marshall, 459 U.S. at 434, 103 S.Ct. 843 (“[Fjederal habeas courts [have] no license to redetermine credibility of witnesses whose demeanor has been observed by the state trial court, but not by them.”).
Had the members of the majority actually been in court and then rendered its string of pejorative assessments, I would be prepared to credit them. But unless we are prepared to hold that trials and hearings are mere useless exercises, I would respect the views of those who were there. Without ever hearing a shred of witness testimony, the majority has effectively decided to take on the role of a jury in the sky and acquit Elmore. It does so despite the fact that no court before us has been in a worse position to evaluate Elmore’s conviction and no court before us has doubted it. To invalidate it on largely factual grounds, as the majority does today, contravenes AEDPA and all that it stands for.
C.
Finally, the majority disregards the actual evidence supporting Elmore’s conviction. Given the strength of the case against Elmore, it is hard to believe that even a flawless defense attorney would *882have been able to obtain an acquittal here. And it is even harder to believe that any jury would have accepted the conspiracy theory concocted by Elmore and embraced by the majority today. Elmore’s conviction has withstood nearly three decades of intense scrutiny from multiple juries and judges for the best of reasons: there is so much evidence that supports it.
1.
The case against Elmore was anything but a flimsy one. To start with, the physical evidence points directly to his guilt. The state PCR court found that forty-nine hairs were recovered from Mrs. Edwards’s bed. According to Lieutenant Wells, two of those hairs were consistent with Mrs. Edwards’s head hairs, two of them were consistent with her pubic hairs, and forty-five of them were consistent with Elmore’s pubic hairs. Elmore has since conceded that those forty-five pubic hairs belong to him. Ante at 810 n. 16. His thumbprint was also found on the exterior frame of the back door to Mrs. Edwards’s home and multiple police officers testified that the print was relatively “fresh.” See id. at 794. And the jeans and shoes that Elmore admitted wearing on Saturday were spattered with blood that matched Mrs. Edwards’s blood type, but not his own. On cross-examination, he could give no explanation of where this blood came from.2
If that were not enough, the prosecution also provided an overwhelming amount of circumstantial evidence of Elmore’s guilt. First, Elmore had no consistent alibi for Saturday evening, the time in which Mrs. Edwards was murdered. On the day of his arrest, Elmore signed a written statement in front of Lieutenant Henderson stating that he was with his girlfriend Mary Alice Dunlap at a K-Mart “until nine-thirty p.m.,” when she was picked up by her brother. According to this statement, Elmore then “stop[ped] only for a minute to get a beer and then went straight on” to see his girlfriend and her family, with whom he then stayed until “sometime before midnight.” But at trial, Dunlap and her family members testified that Elmore left them a little before 10:00 p.m. and did not return until 12:30 in the morning, nursing a newly swollen lip. On the stand, Elmore changed his story as well, claiming that he “stayed awhile ... talking” at a convenience store before going to see Dunlap “around ten-thirty or eleven.” Evidently the jury — which again was far better positioned to evaluate Elmore’s credibility than the majority — -did not find these shifting alibis to be persuasive.
What is more, Elmore lied to the police about even knowing the victim. It is undisputed that Elmore knew Mrs. Edwards, *883having done household work for her on several occasions. In fact, on the day he was arrested he had a piece of paper in his wallet with her name and number on it. But when Elmore was arrested, he denied knowing who she was until the officers confronted him with checks she had written to him, at which point he changed his tune. He then told the arresting officer that “if in fact he did kill Mrs. Edwards, that he did not remember doing it.”
Finally, this already strong case was further bolstered by Elmore’s confession to his cellmate, James Gilliam. When Elmore was arrested, Gilliam was in jail for receiving stolen goods, and the two knew each other from having once lived in the same apartment complex. Gilliam testified that Elmore told him about how he had gone to Mrs. Edwards’s house to rob her but “the lady started screaming and she wouldn’t stop and so he had to kill her.” Elmore also queried Gilliam about whether his efforts at eliminating the forensic evidence of the crime had been successful. Specifically, Elmore described how he had wiped his fingerprints and asked Gilliam “if you had sex with somebody and you washed up, could you tell that you have had sex with that person.” Gilliam wrote a letter to the police about this conversation and testified about it consistently throughout Elmore’s three jury trials.
But in the state PCR proceedings, Gilliam changed course, testifying that Arlie Capps, a jail administrator, had come to him offering a deal if Gilliam could get a confession from Elmore and that Gilliam had invented the confession to reap those benefits. One of Gilliam’s two stories is false, of course. But which? The question turns on credibility and thus fundamentally belongs to the state PCR court. See, e.g., Wilson v. Ozmint, 352 F.3d 847, 858 (4th Cir.2003) (“Credibility determinations ... are presumed to be correct absent clear and convincing evidence to the contrary.”) (internal quotation marks and citation omitted). The PCR court heard extensive evidence on this issue, including testimony from jail administrators, other prisoners, and Gilliam himself. Its conclusion was emphatic: “Why Gilliam chose to recant his testimony is not evident, but that does not mean that his new version has credibility and I hold that it does not.” In particular, it found that “there has been no showing that there was any ‘deal.’ ” Probing the inconsistencies in Gilliam’s new story, the court called his recantation “illogically conflicting and incredible” and added that “[t]he gross speculation that Gilliam’s testimony was scripted is pure fiction.”
Thus, even if Elmore’s counsel had provided ineffective assistance with respect to a particular evidentiary issue, the likelihood that that error changed the outcome of the trial is hardly “substantial.” See Cullen, 131 S.Ct. at 1403. The state’s case against Elmore was built on numerous pieces of evidence which subsequent findings did not draw into question. The irony of this case is that the majority lambasts the state PCR court for “considering less than the totality of the evidence,” ante at 868, while ignoring or diminishing the state’s total arsenal of evidence against Elmore at the same time. See id. at 871 (dismissing the “bloodspots” on Elmore’s clothes, his “thumbprint on the back door frame,” the evidence of his “guilty conscience,” and “Gilliam’s account of Elmore’s spontaneous jailhouse confession” as “flimsy”). By attempting to divide and conquer discrete evidentiary issues while ignoring the rest of prosecution’s case, it is the majority — and not the PCR court— which has “unreasonably discounted evidence” by “minimizing its import and evaluating it piecemeal.” See id. at 868. The unfairness of calling the state court out for what is manifestly the majority’s own fail*884ing only underscores the injustice of today’s decision.
2.
In the face of all this evidence, the majority is forced to fall back on Elmore’s conspiracy theory that multiple members of the South Carolina law enforcement system framed Elmore. Once again, this accusation is not supported by the findings of any previous court. According to this account, Officers Coursey and Henderson removed the incriminating pubic hairs from Elmore while he was detained and then Agents Parnell and DeFreese lied about recovering this evidence from Mrs. Edwards’s bed. See id. at 810 n.16, 855-56. Officer Henderson then “planted” blood matching Mrs. Edwards’s blood type on Elmore’s “pants and shoe.” See id. at 834, 852. What is more, Agent DeFreese misreported the fingerprints on Mrs. Edwards’s toilet and exterior door frame as unidentifiable, yet another example, as the majority puts it, of “police ineptitude and deceit.” Id. at 871. According to Elmore, these prints belonged neither to him nor to Mrs. Edwards and would have been exculpatory if DeFreese had been honest. Id. at 803. Finally, SLED Agent Wells “negligently or knowingly misreported” the contents of Item T as being solely blue fibers. Id. at 870. Ultimately, the majority concludes, the “evidence leads one to distrust the competence or motivations of the police investigators.” Id. at 871.
While this story may make for a good movie, it does not stand up as a piece of legal analysis or bear resemblance to reality. For one, the majority reverses multiple factual findings, including witness credibility determinations, made by the state PCR court. Once again, we can only overcome these factual findings “by clear and convincing evidence,” 28 U.S.C. § 2254(e)(1), and we can only dismiss these credibility determinations if the PCR court made a “stark and clear” error. Cagle v. Branker, 520 F.3d 320, 324 (4th Cir.2008). Such showings have not been made here.
With regard to the pubic hairs, Agents Parnell and DeFreese testified that Agent Parnell collected that evidence from Mrs. Edwards’s bed. The majority attacks this testimony by treating the quantity of hair found on Mrs. Edwards’s bed with incredulity and imputing ill motives for the agents’ failure to photograph the bed or put the sheets in the record. See ante at 870. But the PCR court that heard this testimony expressly found that Parnell and DeFreese “credibly testified about the existence and discovery of the hair.”
There was no “clear and convincing” evidence suggesting this determination was incorrect. 28 U.S.C. § 2254(e)(1). Agent Parnell consistently testified that he discovered “almost a rectangular area of hair” that was “approximately thirty inches across by approximately eighteen inches up and down” and “running diagonally across the bed,” and the PCR court found this testimony to be credible. Parnell also repeatedly admitted that his decision not to photograph the bed was an “oversight,” but not an act done for conspiratorial reasons. Parnell’s account was additionally corroborated by Agent De-Freese, who, as the PCR court found, “testified that he was present and assisted Agent Parnell when the hair was seized from the victim’s bed.” And far from plotting to lie about the pubic hairs to incriminate Elmore, Parnell testified that he was “mistakenly under the opinion they were Caucasian” until Agent Wells informed him otherwise.
As for the decision not to take the sheets into evidence, Parnell explained that he believed they had little evidentiary *885value given the lack of stains. This is hardly proof of a police conspiracy. The PCR court put it directly: “the speculative presentation of the other witnesses that they normally seize such items does not create a Sixth Amendment violation.” And as for the quantity of pubic hairs, Agent Wells explained during Elmore’s 1984 trial that only four of the hairs fell out naturally; the other forty-one were pulled out by force. One can only imagine the kind of battle between a rapist and his victim that would lead to such an injury.
Turning to the blood on Elmore’s clothes, the PCR court rejected the notion that Agent Henderson had planted the incriminating blood on the basis of Agent Barron’s testimony. The state court found that Barron testified “[cjlearly” and “without equivocation” he “would not have given the [clothes] stains to Tom Henderson,” but instead maintained control over the evidence until trial. The majority nevertheless seizes on a dispute in the PCR proceedings over whether the laboratory records suggest that Henderson removed these items from the lab prior to Barron’s analysis. See ante at 817, 834. But the PCR court had already resolved this dispute in favor of Barron. After hearing witness testimony on the issue, the PCR court found that Elmore’s clothes were brought to the SLED lab on January 20, 1982, whereupon “Barron received the evidence and cut out the specimens prior to the return of the coat and jeans to Tom Henderson” on February 3. The majority’s dismissal of the state court’s acceptance of Barron’s interpretation without a compelling reason for doing so is yet another example of its refusal to show a modicum of deference.
As for Agent DeFreese’s alleged misrepresentation of the fingerprint evidence, the state PCR court treated his earlier testimony that the prints were unidentifiable as an honest mistake at worst. While noting that it was “regrettable” that Agent DeFreese concluded the fingerprint on Mrs. Edwards’s door frame was unidentifiable in 1982, the PCR court concluded that fact “does not support [Elmore’s] rhetorical assault on his reporting and testimony.” As the state court pointed out, De-Freese simply used a higher threshold for identifying prints than one of Elmore’s experts did. It then noted that even Elmore’s own experts disagreed over how many prints were identifiable, but that this fact did not mean they somehow perjured themselves. After “viewing the credibility of the witnesses based upon the entire record,” the PCR court determined that it “must reject [Elmore’s] assertions” that DeFreese testified falsely.
As for Agent Wells’s claim that the contents of Item T were solely blue fibers, the PCR court found that he did not “knowingly fail[ ] to disclose” this evidence. Of course, it is inconvenient for the majority’s conspiracy theory that it was Agent Wells himself “who discovered the missing evidence and reported the discovery” to the South Carolina Attorney General’s Office, see ante at 843, and that Wells repeatedly “offered himself for cross examination.” But inconvenient facts are apparently momentary obstacles for the majority, which once again, without listening to any testimony, has decided peremptorily that law enforcement officers are bad actors despite findings to the contrary from the judicial officer who actually heard them.
Apart from ignoring its proper role in the context of collateral attack, the majority also fails to provide a plausible motive for this supposed conspiracy. At most, it claims that Henderson planted blood on Elmore’s clothes because his mother was “a longtime neighbor of Mrs. Edwards” as well as an acquaintance of Jimmy Holloway. See ante at 808-09, 834. But it *886never explains why these local connections would compel Henderson to single out Elmore of all people to take the blame. Nor does it explain why four other law enforcement officers — Coursey, Parnell, De-Freese, and Wells — would decide to go along with Henderson’s scheme. And it does not begin to explain why a jail administrator like Arlie Capps would induce an inmate to perjure himself in order to convict Elmore. The majority neither asks these questions nor provides any answers.
3.
If this rampant speculation were not enough, the majority then slanders a deceased and presumptively innocent man by essentially accusing him of the grisly murder of Mrs. Edwards. Following Elmore’s lead, the majority suggests that Jimmy Holloway murdered his longtime neighbor and friend and then “staged” the “circumstantial evidence of a Saturday night death.” Ante at 870. This charge is, to quote the PCR court, a “shocking assertion” and “gross speculation.”
There are three possible reasons for pinning this heinous crime on Holloway. Not one stands up to the slightest scrutiny. First, according to Elmore’s expert Vincent Scalise, Holloway’s conduct surrounding his discovery of his mutilated neighbor was “very, very strange” and ultimately “suspect.” But what exactly did Holloway do to incur such suspicion? On Monday, January 18, he stopped by Mrs. Edwards’s house to check on his elderly neighbor. He noticed the Sunday and Monday newspapers laying on the ground, despite the fact she had told him she was leaving for a trip on Sunday morning. Concerned that she might be ill, Holloway knocked on the back door, which suddenly opened. Entering the residence, he noticed that the house was at high temperature and that the TV was on at a high volume. Venturing further, he discovered that Mrs. Edward’s commode was stained with blood and that there was a knife and large amount of blood on the bedroom floor. At that point, Holloway went next door to the home of Mrs. Clark — another one of Mrs. Edwards’s neighbors — to “see if something had happened to her and [if] she was in the hospital.” Upon finding out Mrs. Edwards was not in the hospital, Holloway returned to the house with Mrs. Clark in tow. To avoid contaminating the scene, he carefully stepped around the pool of blood and donned a pair of woolen gloves before he opened the closet door. After he discovered his neighbor’s mutilated body stuffed in the closet, Holloway left the house with Mrs. Clark and they called the police.
As the PCR court found, this behavior revealed nothing more than the fact that Holloway was a “legitimately concerned neighbor who feared that something had happened” to an elderly woman and friend. Even Scalise conceded as much, noting that many of Holloway’s actions on Monday were those of a concerned neighbor and that it was “not unusual” for someone coming across a crime scene to get another person before proceeding any further. We should be the last to presume to dictate what the normal response of a worried neighbor stumbling upon evidence of foul play should be, particularly when the only court to hear testimony on the matter found that it was normal.
Second, as the majority points out, Agent Henderson “confirmed that Holloway had been an early suspect.” Id. at 807. But the only reason for that classification was that Holloway discovered the body. According to Scalise himself, as a matter of basic police procedure, “in most cases” the first suspect is the “person that finds the body.” There was nothing more to suggest Holloway’s guilt, and *887Henderson soon ruled him out as a suspect as the evidence pointed increasingly to Elmore.
In any event, it would be helpful if the majority could explain exactly why Holloway would ever brutally rape and murder his longtime neighbor and friend. But it offers no possible motive to support this astonishing charge, nor is there any indication of one in the record. In contrast, Elmore had a very clear reason for murdering Mrs. Edwards that Saturday night. As he confessed to Gilliam, “he went there to rob the lady, and she started screaming, so he had to kill her.”
The third and perhaps real reason for Elmore to pin the murder on Holloway is the simple fact that he is no longer alive. Unlike Elmore, who is still with us and represented by vigorous counsel, Holloway can no longer speak for himself nor has he anyone to argue on his behalf. But while it may be tempting to purchase Elmore’s freedom at the cost of an innocent man’s good name, it is profoundly unjust.
D.
At the end of the day, no less than three separate courts have reviewed and rejected Elmore’s claims of prejudice. And not one of them has ever made any finding that would suggest that any alleged errors by Elmore’s counsel were prejudicial. Now, for the first time since Elmore’s conviction over a quarter-century ago, two fine colleagues have come to the opposite conclusion without hearing a minute of witness testimony and in the face of all the evidence available. In spinning this tale of deceit, fabrication, perjury, and corruption, the majority has unfairly impugned the South Carolina criminal justice system— directly in the case of its law enforcement officers and indirectly in the case of the prosecutors and judges who turned a blind eye to malfeasance of this magnitude. Seldom have so many been besmirched by so few. If this framing of Elmore actually took place, it would be worth all the opprobrium the majority could heap upon it and more. But the legal system places its faith in evidence and those who see and hear it. While no system of factual findings is perfect, the regime of appellate highhandedness and speculation is far worse. This road is no path to justice, not in any whole and rounded sense. In light of all the evidence available, the notion that the state PCR court incorrectly applied Strickland is simply implausible. The idea that it committed an “error well understood and comprehended in existing law beyond any possibility for fairminded disagreement,” Richter, 131 S.Ct. at 787, is beyond the pale.
III.
This sad decision is only compounded by the fact that it could have been so easily averted had the majority paid attention to controlling precedent. Just last term, the Supreme Court in Harrington v. Richter, -U.S. -, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011), reversed a federal appellate court’s use of the writ of habeas corpus to set aside a conviction on the grounds of ineffective assistance of counsel. I suppose we should be grateful that this recent and relevant case merits a courtesy mention by the majority, for that is all that it gets.
Richter could not be more on point. Like Elmore, Richter was convicted in state court for murder despite his contention that another person — in his case, a drug dealer named Joshua Johnson — manipulated the physical evidence against him. See Richter v. Hickman, 578 F.3d 944, 976 (9th Cir.2009) (en banc). Like Elmore, Richter sought relief from the state court system following his conviction on the basis of Strickland, arguing that his *888counsel provided ineffective assistance “for failing to present expert testimony” on the forensic evidence. Richter, 131 S.Ct. at 783. The California Supreme Court denied relief. Id. Then, like Elmore, Richter filed a habeas petition in federal district court, which again denied him relief. And, like Elmore, he was finally able to find judges who agreed with his position on appeal. Id,.
The Supreme Court reversed. It was emphatic in its holding that “habeas corpus is a guard against extreme malfunctions in the state criminal justice systems, not a substitute for ordinary error correction through appeal.” Id. at 786 (internal quotation marks and citations omitted). Applying the doubly deferential approach mandated by Strickland and AEDPA, the Court concluded that the state court could have determined that Richter’s counsel was effective even though he had “not consulted forensic ... experts or introduced expert evidence.” Id. at 788. As the Court observed, “Reliance on ‘the harsh light of hindsight’ to cast doubt on a trial that took place now more than 15 years ago is precisely what Strickland and AEDPA seek to prevent.” Id. at 789 (quoting Bell v. Cone, 535 U.S. 685, 702, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002)).
More importantly, the Supreme Court then held that the state court could reasonably have found that Richter was unable to show prejudice. The Court noted that apart from the forensic evidence, there was enough circumstantial evidence available to allow a state court to eliminate the possibility that Richter was prejudiced. Noting Richter’s “shifting story” and “the lack of any obvious reason” for Johnson to tamper with the physical evidence, the Court concluded that there was “ample basis for the California Supreme Court to think any real possibility of Richter’s being acquitted was eclipsed by the remaining evidence pointing to guilt.” Id. at 792.
The majority references Richter, as it must, but then proceeds to distinguish the case on the ground that the forensic evidence here “was always and obviously vital to the State’s case.” Ante at 863. My colleagues miss the larger point. If the California Supreme Court’s application of Strickland in Richter was reasonable, then the South Carolina PCR court’s application of Strickland here must be so as well.
For one, the PCR court’s opinion was far better supported than the state court opinion in Richter. Whereas the California Supreme Court dismissed Richter’s claim for relief in a one-sentence order, the PCR court here issued a 183-page opinion responding to Elmore’s numerous claims. While a state court need not provide a justification for its decision for the purposes of § 2254(d), Richter, 131 S.Ct. at 784-85, when it “explain[s] its reasoning with some care, it should be particularly difficult to establish clear and convincing evidence of error.” Sharpe v. Bell, 593 F.3d 372, 378 (4th Cir.2010).
More importantly, on the question of prejudice, the circumstantial evidence in Elmore’s case provided more than an “ample basis” for the PCR court to conclude that any chance of Elmore’s acquittal “was eclipsed by the remaining evidence pointing to guilt.” See Richter, 131 S.Ct. at 792. In addition to Elmore’s “shifting story” regarding his whereabouts on Saturday night as well as “the lack of any obvious reason” for Holloway to murder Mrs. Edwards and the police to frame an innocent man, see id., the already-strong circumstantial evidence against Elmore was cemented, inter alia, by his confession to Gilliam, his pubic hair on Mrs. Edwards’s bed, and the blood on his clothes matching Mrs. Edwards’s blood type. The problem with Elmore’s case is the sheer *889strength of the evidence against him, not the failure of counsel to remove that boulder from the road. Perhaps the day will come when we all realize that habeas corpus confers no freewheeling warrant to retry a case on appeal, see id. at 786, and the Supreme Court can rest from reiterating this basic point in case after case. Unfortunately, today’s decision shows that we still have a long way to go.
IV.
Thus far, I have discussed facts and deference, for those are the most concrete failings of the majority’s decision today. But the real injustice of this case occurs on an individual level. Thanks to the majority’s efforts, Mr. Elmore may go free because the crime he committed took place almost three decades ago. Perhaps Elmore believes that because so much time has passed since the murder, he should simply be let go.
My distinguished colleagues in the majority respond to the dissent with rhetoric and a protestation that they are not doing what in fact they are doing — overturning factual findings and credibility determinations of the state system that painstakingly heard the evidence in this case. But at the end of the day, our system is indeed grounded on facts and evidence. If the state courts had defaulted in their job, that would be one thing, but it is hard to find a case that received a more thorough review under the well-settled Strickland standard than this one did. If respect is not accorded the careful conclusions and precise findings of the state court in the case at bar, then I am hard-pressed to see the circumstances in which the majority would extend respect. Contrary to my brothers’ rejoinder, I have not the slightest hesitation in embarrassing any person or institution so long as that embarrassment is grounded in the raw materials of judicial process, not dispensed from the speculative perch on which we sit. The majority says it doesn’t “credit Elmore’s evidence,” ante at 872, but it discredits the courts assigned to weigh that evidence and assess it. Visiting reputational harm is not something federal courts do lightly. Before insinuating that law enforcement officers are “inept and corrupt,” id. at 872, or that “the real perpetrator was Holloway,” id. at 873, the majority should be able to point to some finding or some ruling by those who actually heard testimony that that might have been the case. It is little enough to ask.
Before this court grants Elmore relief, we should pay passing thought as to what happened to Dorothy Edwards on the night she was raped and murdered. According to Dr. Sandra Conradi, the forensic pathologist who performed Mrs. Edwards’s autopsy, the body of the seventy-five-year-old victim “was covered with injuries of all sorts, including stab wounds, blunt traumatic injuries,” “abrasions,” and “bruises.” Her lower arms were extensively damaged, most likely from her attempts to shield herself from attack. Despite her resistance, she was stabbed repeatedly in the head and neck and suffered multiple lacerations to her ears as well as the inside of her mouth. Her left cheek was crushed in as well. She also sustained thirty-three lesions on her chest area, most likely inflicted with a pair of bottle tongs, and two-thirds of those injuries occurred prior to her death. Her ribs were fractured, the front part of her chest had caved in, and her right lung was torn. After estimating Mrs. Edwards sustained roughly “seventy-five separate injuries,” Dr. Conradi concluded that she could not recall, in her “fifteen years of experience, seeing as much trauma and as varied trauma as in this particular case.”
*890The more heinous the crime, the more necessary it becomes not to lash out and affix blame, but the long, thorough layers of review accorded Mr. Elmore’s case do not portray a system acting in vengeance or in haste. It is right and good to believe in mercy and forgiveness for those who commit even the most cruel and brutal acts. But that redemption is not rightfully ours to bestow, and the criminal justice system does not serve its own humane, protective purposes by slipping into amnesia on something like this. Law does not permit it, nor should it. Not on this. Not on this.
I would affirm the judgment.
. In view of the fact that this case can be soundly resolved on the prejudice prong of Strickland, I see no need to go into the performance aspect of the case. By no means, however, do I concede the point that the state PCR court's finding that Elmore’s "[cjounsel was [not] 'deficient' ... with respect to the second trial" represents an unreasonable application of Strickland. In fact, the majority itself points out the significant efforts defense counsel made in attacking the state’s case. See ante at 800-02, 852-55. It is manifest that the problem here was not Elmore’s counsel, but the fact that even the most sterling advocate cannot overcome an overwhelmingly adverse case. To scapegoat Elmore's counsel for not performing impossible feats seems to me a deeply flawed application of the Strickland decision.
. DNA testing now confirms that the blood on Elmore's jeans and shoes fully or partially matches Mrs. Edwards’s DNA. The majority dismisses this latest installment of inculpatory evidence as "consistent with Elmore's position that the SLED agents ... planted Mrs. Edwards’s blood on his clothing.” See ante at 853. It also assumes without deciding that Cullen v. Pinholster, — U.S. —, 131 S.Ct. 1388, 179 L.Ed.2d 557 (2011), precludes consideration of this evidence given that it was developed following the first and second PCR orders. See id. at 851-52. For the sake of argument, I will address the majority on its own terms and rely only on the evidence available at the time of these PCR proceedings. Still, it is worth noting the lengths to which the majority has gone to grant Elmore relief. In the face of evidence that would seem to put Elmore’s guilt beyond question, the majority embraces the only possible explanation that would preserve its claim he was innocent: the unfounded theory that he was framed by the police. At this point, I am skeptical that any evidence, no matter how probative, would ever be able to shake the majority’s unquestioning and ill-supported faith in Elmore's innocence.