Ronnie Long v. Erik Hooks, No. 18-6980 (4th Cir. 2020)Annotate this Case
This opinion or order relates to an opinion or order originally issued on January 8, 2020.
ON REHEARING EN BANC PUBLISHED UNITED STATES COURT OF APPEALS FOR THE FOURTH CIRCUIT No. 18-6980 RONNIE WALLACE LONG, Petitioner – Appellant, v. ERIK A. HOOKS, Secretary, NC Dep't of Public Safety, Respondent – Appellee. ------------------------------ THOMAS ALBRIGHT, Professor and Director, Vision Center Laboratory, Conrad T. Prebys Chair in Vision Research, Salk Institute for Biological Studies; VALENA ELIZABETH BEETY, Professor of Law, West Virginia University School of Law; BARBARA E. BIERER, MD, Professor of Medicine, Harvard Medical School; DR. C. MICHAEL BOWERS, Clinical Associate Professor, Herman Ostrow School of Dentistry, University of Southern California; ARTURO CASADEVALL, MD, PhD, Chair, Molecular Microbiology & Immunology, Alfred & Jill Sommer Professor and Chair, Bloomberg Distinguished Professor, Johns Hopkins University; JESSICA GABEL CINO, Associate Professor of Law, Associate Dean for Academic Affairs, Georgia State University School of Law; SIMON A. COLE, Professor, Department of Criminology, Law & Society, University of California, Irvine; M. BONNER DENTON, Galileo Professor of Chemistry, Professor of Geosciences, The University of Arizona; SHARI SEIDMAN DIAMOND, Howard J. Trienens Professor of Law, Northwestern University Pritzker School of Law; DR. RACHEL DIOSO-VILLA, Senior Lecturer, School of Criminology & Criminal Justice, Griffith University; JULES EPSTEIN, Professor of Law, Director of Advocacy Programs, Temple University Beasley School of Law; DAVID L. FAIGMAN, Chancellor and Dean, John F.Digardi Distinguished Professor of Law, University of California Hastings College of the Law; LISA S. FAIGMAN, Professor, University of California Hastings College of the Law; NITA A. FARAHANY, Professor of Law & Philosophy, Duke University School of Law; BRANDON L. GARRETT, L. Neil Williams Professor of Law, Duke University School of Law; BRUCE GREEN, Louis Stein Chair of Law, Director, Stein Center for Law and Ethics, Fordham University School of Law; LISA KERN GRIFFIN, Carroll-Simon Professor of Law, Duke University School of Law; EDWARD J. IMWINKELRIED, Edward L. Barrett, Jr. Professor of Law Emeritus, UC Davis School of Law; INNOCENCE PROJECT, INC.; EISHA JAIN, Assistant Professor of Law, University of North Carolina School of Law; DR. DAVID KORN, Professor of Pathology, Harvard Medical School; JASON KREAG, Associate Professor of Law, University of Arizona James E. Rogers College of Law; DANIEL S. MEDWED, University Distinguished Professor of Law and Criminal Justice, Northeastern University; JENNIFER L. MNOOKIN, Dean, David G. Price & Dallas P. Price Professor of Law, UCLA School of Law; JOHN MONAHAN, PhD, Shannon Distinguished Professor of Law, Professor of Psychology, Professor of Psychiatry and Neurobehavioral Sciences, University of Virginia School of Law; ALAN MORRISON, Lerner Family Associate Dean for Public Interest & Public Service, George Washington University Law School; ROBERT P. MOSTELLER, J. Dickson Phillips Distinguished Professor of Law, University of North Carolina School of Law; ERIN MURPHY, Professor of Law, NYU School of Law; D. MICHAEL RISINGER, John J. Gibbons Professor of Law Emeritus, Seton Hall University School of Law; MICHAEL SAKS, Regents' Professor, Sandra Day O'Connor College of Law, Department of Psychology, Arizona State University; NICHOLAS SCURICH, PhD, Associate Professor, Department of Psychological Science, Department of Criminology, Law & Society, University of California, Irvine; GEORGE SENSABAUGH, Professor of the Graduate School, Professor Emeritus of Forensic and Biomedical Sciences, School of Public Health, University of California, Berkeley; DAN SIMON, Richard L. and Maria B. Crutcher Professor of Law & Psychology, Gould School of Law, Department of Psychology, University of Southern California; J.H. PATE SKENE, JD, PhD, Associate Research Professor of Neurobiology, Duke University Medical Center; CLIFFORD SPIEGELMAN, Distinguished Professor of Statistics and University Distinguished Professor, Department of Statistics, Texas A&M University; COLIN STARGER, Associate Professor, University of Baltimore School of Law; HAL STERN, Chancellor's Professor, Department of Statistics, Donald Bren School of Information and Computer Sciences, University of California, Irvine; WILLIAM A. TOBIN, Principal, Forensic Engineering International; JAMES L. WAYMAN, PhD, FIEEE, FIET, Office of Research, San Jose State University; ELLEN YAROSHEFSKY, Howard Lichtenstein Professor of Legal Ethics, Director, Monroe Freedman Institute for the Study of Legal Ethics, Maurice A. Deane School of Law, Hofstra University; SANDY ZABELL, Director 2 of Undergraduate Studies, Professor of Statistics and Mathematics, Northwestern University, Amici Supporting Appellant. Appeal from the United States District Court for the Middle District of North Carolina, at Greensboro. Catherine C. Eagles, District Judge. (1:16-cv-00539-CCE-LPA) Argued: May 7, 2020 Decided: August 24, 2020 Amended: August 26, 2020 Before GREGORY, Chief Judge, and WILKINSON, NIEMEYER, MOTZ, KING, AGEE, KEENAN, WYNN, DIAZ, FLOYD, THACKER, HARRIS, RICHARDSON, QUATTLEBAUM, and RUSHING, Circuit Judges. Vacated and remanded with instructions by published opinion. Judge Thacker wrote the opinion, in which Chief Judge Gregory and Judges Motz, King, Keenan, Wynn, Diaz, Floyd, and Harris joined. Judge Wynn wrote a concurring opinion, in which Judges Thacker and Harris joined. Judge Richardson wrote a dissenting opinion, in which Judges Wilkinson, Niemeyer, Agee, Quattlebaum, and Rushing joined. ARGUED: Jamie Theodore Lau, DUKE UNIVERSITY SCHOOL OF LAW, Durham, North Carolina, for Petitioner. Phillip Anthony Rubin, NORTH CAROLINA DEPARTMENT OF JUSTICE, Raleigh, North Carolina, for Respondent. ON BRIEF: Theresa A. Newman, DUKE UNIVERSITY SCHOOL OF LAW, Durham, North Carolina; G. Christopher Olson, Raleigh, North Carolina, for Appellant. Joshua H. Stein, Attorney General, Clarence Joe DelForge, III, Special Deputy Attorney General, NORTH CAROLINA DEPARTMENT OF JUSTICE, Raleigh, North Carolina, for Appellee. Karen A. Newirth, THE INNOCENCE PROJECT, INC., New York, New York; Breon S. Peace, Matthew Aglialoro, Willam Segal, CLEARY GOTTLIEB STEEN & HAMILTON LLP, New York, New York, for Amicus The Innocence Project, Inc. Brandon L. Garrett, L. Neil Williams Professor of Law, DUKE UNIVERSITY SCHOOL OF LAW, Durham, 3 North Carolina; Mark D. Harris, Adam W. Deitch, PROSKAUER ROSE LLP, New York, New York, for Amici Curiae. 4 THACKER, Circuit Judge: For 44 years Ronnie Wallace Long (“Petitioner”) has been in prison for a rape and burglary that he has consistently maintained he did not commit. At his 1976 trial, the State of North Carolina (the “State”) asked the jury to rely on the “perfect honesty” of the police officers in charge of investigating the crime. J.A. 576. 1 However, from the time of Petitioner’s conviction until now, a trickle of posttrial disclosures has unearthed a troubling and striking pattern of deliberate police suppression of material evidence, in violation of Petitioner’s due process rights pursuant to Brady v. Maryland, 373 U.S. 83 (1963). This suppressed, material evidence includes lab test results demonstrating that Petitioner was not linked to the crime scene in any way; a medical authorization demonstrating sperm taken from the victim was turned over to the police and never seen again; and most recently, 43 latent fingerprints lifted from the scene, none of which matched Petitioner. The Superior Court in Cabarrus County, North Carolina, denied Petitioner’s Motion for Appropriate Relief (“MAR”) (hereinafter the “MAR Court”). Specifically, the MAR Court concluded that the cumulative effect of the withheld Brady evidence would have had no impact on Petitioner’s trial. After exhausting state remedies, Petitioner then filed a federal habeas petition pursuant to 28 U.S.C. § 2254 (the “Petition”). The district court dismissed the Petition, concluding that the MAR Court’s decision did not involve “an unreasonable application of” clearly established federal law. J.A. 1657 (quoting 28 U.S.C. § 2254(d)(1)). 1 Citations to the “J.A.” refer to the Joint Appendix filed by the parties in this appeal. 5 We disagree. The MAR Court’s analysis subjected Petitioner to an enhanced burden, unreasonably applied Supreme Court law, and was objectively unreasonable. For these reasons, and as explained more fully below, we vacate the district court’s dismissal of the Petition. The district court left unaddressed the issue of whether the Petition can survive the threshold requirements set forth in the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”). Thus, we remand to the district court to permit consideration of this question in the first instance and to permit further discovery if requested by Petitioner. I. A. The Burglary and Rape Sarah Judson Bost, a 54 year old widow, was attacked in her Concord, North Carolina home on the evening of April 25, 1976, between 9:30 and 9:45 p.m. The perpetrator grabbed Ms. Bost from behind as she stepped from her kitchen into her den on the first floor of her home. With a knife to Ms. Bost’s throat, the perpetrator first demanded money, and when Ms. Bost found none in her purse, he proceeded to drag her to the foot of the stairs, strip off her clothing, and rape her. The perpetrator pushed Ms. Bost’s head down and to the side and told her, “[D]on’t look at my face.” J.A. 204. Ms. Bost clawed at the man’s leather jacket, pressing so hard that her fingernails were “nearly bent backwards.” Id. at 295. After 10 to 12 minutes, Ms. Bost’s telephone rang, causing the perpetrator to run out the front door. 6 Scared the assailant might come back, Ms. Bost ran naked to her neighbor’s home, and the neighbor called the police and an ambulance. Ms. Bost was transported to the local hospital, where she was examined by Dr. Lance Monroe, a gynecologist, who collected vaginal fluid samples. Also at the hospital, Ms. Bost talked to two Concord Police Department (“CPD”) officers, Detective David Taylor and Lieutenant George Vogler. The officers showed Ms. Bost a photo array of 13 black men “age group twenty to thirty,” 2 J.A. 307, but she could not identify any of them as her attacker. Petitioner was not included in this photo array. According to the initial case report, which was based on information Ms. Bost gave to officers directly following the rape, Ms. Bost gave the following description of the perpetrator: [A] black male, height, five foot, five to five foot nine, slender build, slim hips. Subject was plain spoken, used correct [E]nglish and at times spoke very softly. No speech defect, accent or noticeable brogue evident. Subject was wearing a dark waist length leather jacket, blue jeans with a dark toboggan pulled over his head. Could possibly have been wearing gloves. Id. at 369–70. Despite this record, Ms. Bost later insisted that her attacker was definitely wearing gloves. In addition, she said nothing about the perpetrator having facial hair to the officer preparing the initial case report. 3 2 There is no evidence Ms. Bost told Detective Taylor or Lieutenant Vogler how old her attacker was. 3 The record indicates that CPD Sergeant Jack Parnell provided the information for the initial case report, which was based on Ms. Bost’s initial description of the suspect. 7 Officers investigated the crime scene and found an open window on the second floor of Ms. Bost’s home. They also found several discarded matches on the floor of the upstairs bedroom, “burned almost completely out.” J.A. 336. The window led to a ledge onto a porch roof, and below that, a banister painted white. Officers noticed a latent shoe print on the banister and took a lift of the print. In an alleged attempt to find the perpetrator, Detective Taylor and Lieutenant Vogler asked Ms. Bost to come to the Cabarrus County Courthouse on May 10, 1976, two weeks after the rape. The officers told Ms. Bost they had “reason to believe that there might be somebody in the courthouse . . . somebody there that [she] could recognize.” J.A. 171–72. They also told her she may have to come back several times over several days. They asked her to sit in the courtroom and observe while cases were called. Understandably, Ms. Bost was nervous at the prospect of seeing her attacker again, so she disguised herself in a red wig and glasses, and sat with her neighbor in the second row of the courtroom. The officers sat in the courtroom as well, away from Ms. Bost. There were approximately 60 people in the courtroom, and around a dozen were African American men. According to Ms. Bost, except for Petitioner, none of these African American men “even closely resembled” the perpetrator -- one of them was “very light and tall, and all stooped over,” and “several . . . had afros.” J.A. 172. After 60 to 90 minutes of Ms. Bost observing the entire courtroom with Petitioner present, Petitioner’s name was called. He stood at the back of the courtroom and proceeded up the aisle to the front of the courtroom. Petitioner was wearing a “leather like” coat. J.A. 276. Ms. Bost signaled that Petitioner was the perpetrator before she heard him speak. See 8 id. at 277 (neighbor agreeing that “[a]t the time [Petitioner] was talking to the Judge, Ms. Bost had already said . . . he’s the one”). Nonetheless, Ms. Bost later told officers she knew the assailant by his voice, which she described as “a black voice,” id. at 170, or a “natural black man’s voice,” id. at 181. Petitioner was in the courtroom on a trespassing charge. According to the Petition, the CPD had arrested Petitioner on or around April 30, 1976 (five days after the rape) in a park located “behind and adjacent to” his parents’ house, where he lived at the time, and charged him with trespassing. J.A. 27 n.8. The trespass charge was dismissed on May 10, 1976, the day Petitioner appeared in court and Ms. Bost identified him as her assailant. After Petitioner left the courtroom, Ms. Bost met the officers in the hallway outside the courtroom, and they drove her to the police station. At the station, around 20 minutes after the courtroom identification, the officers showed Ms. Bost six to eight photographs of suspects, one of which was Petitioner. Of note, Petitioner was the only person in the photos wearing a leather coat, which was the type of clothing Ms. Bost initially identified her assailant as wearing. She chose Petitioner’s photo as the photo of her assailant and explained she had “[n]o doubt whatsoever” that he was the perpetrator. J.A. 227. Ms. Bost said the officers “could have” asked her to pick out Petitioner by name, but she did not know for sure. Id. at 179. That evening, May 10, 1976, Detective Taylor and Lieutenant Vogler showed up at Petitioner’s home. The officers asked him to come to the police station. Believing the police wanted to discuss his dismissed trespass charge, Petitioner followed them in his car 9 to the station. While he was at the station, he signed a waiver of Miranda 4 rights and submitted to questioning without an attorney. The officers then told Petitioner he was a suspect in Ms. Bost’s rape. The officers, upon Petitioner’s consent, went to the parking lot and searched Petitioner’s car. 5 They reported finding matchbooks; black gloves over the sun visor, which Petitioner admitted were his; and a toboggan, which Petitioner said he had never seen. He was promptly arrested for the rape of Ms. Bost and burglary of her home. Since that day 44 years ago, Petitioner has been in prison, all the while maintaining his innocence. B. Trial 1. The State’s Evidence The State called Ms. Bost, who testified that she got a “good, clear look” at the perpetrator’s face, J.A. 203, and had “[n]o doubt whatsoever” that Petitioner was the perpetrator, id. at 227. She also testified that the perpetrator “kept holding [her] face . . . and mashing [her] face to the side.” Id. at 245–46. He told her many times “don’t look at my face,” to which Ms. Bost replied, “I don’t want to see your face.” Id. at 197. Ms. Bost 4 Miranda v. Arizona, 384 U.S. 436 (1966). 5 At trial, on appeal, and in a state habeas petition, Petitioner challenged the constitutionality of the search of his car, contending that, despite the officers’ testimony, he did not give his consent to search the vehicle. Because the trial court found Petitioner did give consent based on the testimony of Detective Taylor and Lieutenant Vogler, Petitioner’s argument was rejected at every turn. 10 also testified that the assailant was wearing a toboggan hat that “came down . . . over his ears, and down his neck.” Id. at 201. She said she saw “the whites of his eyes,” and he was “looking up” and “toward the television.” Id. at 138. She also testified that she looked into his face when he was on top of her. Although she did not mention facial hair in her initial description, at trial, which was also after she had seen a photo of Petitioner with facial hair, Ms. Bost testified the perpetrator had “a tiny little mustache and a sort of spotted, unshaven beard,” just as Petitioner looked in the photo. Id. at 158. She also testified her assailant was a “light skinned” or “yellow” black man, “not a real black man.” J.A. 167–68, 201. The State also called Ms. Bost’s neighbor, who corroborated the events occurring after Ms. Bost fled her home after the rape, and in the courtroom identification. The State then called a series of CPD officers. First, the State called Sergeant Jack Parnell, who testified that he surveyed Ms. Bost’s house on the night of the attack and “when [he] got upstairs, . . . [t]he window was open.” J.A. 281. Sergeant Parnell also went to the neighbor’s house to talk to Ms. Bost, who, per the initial case report, gave a description of the perpetrator but said nothing about facial hair. The State then called Dr. Monroe, who testified that when he examined Ms. Bost, she was “in a very highly emotional state.” J.A. 295. She had “numerous scratches, bruises and cuts on her face. . . . Some of her fingernails looked like they had been traumatized, or nearly bent backwards.” Id. On her abdomen, “she had a bruise on the lower left side extending to the inner vaginal region”; “her legs were skinned and bruised”; and she had “a lot of . . . bruises” around her vagina and vulva, which could have been caused by “a 11 hand, knee, fist, teeth, or anything like that.” Id. at 296, 299. Dr. Monroe took fluid from Ms. Bost’s vagina, put it on a slide, and went to the hospital laboratory to examine it under the microscope. There, he found the fluid was “swarming with an extremely high count of live, very active, human spermatozoan.” Id. at 296. Dr. Monroe did not testify what happened to the vaginal fluid after he examined it. A series of other CPD officers also testified, including Detective Taylor. He made written notes of the information Ms. Bost gave him after her courtroom identification, which he read from the stand. The notes reflect that Ms. Bost told Detective Taylor that when she saw Petitioner during the prior courtroom identification, she “knew [Petitioner] was the man” because of his “profile, the coloring of his skin,” and she “will never forget the way he talked to [her]” and “knew his voice.” Id. at 314–15. Detective Taylor also testified that he went to Petitioner’s house to ask him to come to the police station. Even though he knew Petitioner was a suspect in Ms. Bost’s rape, he did not tell Petitioner why he was being asked to come to the station. Once there, Detective Taylor “advised [Petitioner] of his rights” and then told him he was a suspect in Ms. Bost’s rape. J.A. 330. Detective Taylor and Lieutenant Vogler also testified about the search of Petitioner’s car, which was in the visitor’s parking space of the police station. There, Detective Taylor testified he found black gloves, a toboggan, and some matchbooks. When questioned out of the presence of the jury on a suppression motion why Detective Taylor took matches from the car, he explained, “We obtained some matches from the scene of the crime. Some matches that had been struck and lit, and we took matches of [a] similar nature from [Petitioner’s] vehicle.” Id. at 335. Then the following exchange occurred: 12 Q: What kind of matches did you find? A: They were just paper matches. Q: Did they match? A: I didn’t match them, no, sir. Q: Then they were not matched? A: To my knowledge, they were not matched. Q: In other words, the matches you got out of the car do not match with those found at the scene of the crime? . . . [A:] I can’t testify to that. [COURT:] Do you have any information that they didn’t match? [A:] No, sir, they didn’t match. J.A. 334–35 (emphasis supplied). Detective Taylor, once in the presence of the jury, repeated that the matches he retrieved from Petitioner’s family car were “of [a] similar nature” to burned matches found in Ms. Bost’s home, which were believed to have been left by the suspect. J.A. 378. He also explained that after finding the toboggan and black gloves in Petitioner’s car, he asked Petitioner to remove his leather jacket. Detective Taylor gave the toboggan, gloves, matches, and Petitioner’s leather jacket to CPD Detective Van Isenhour. Detective Isenhour also testified at trial. He explained that he spent around two and a half hours at the crime scene looking for latent evidence, “such as fingerprints, or impressions of this type.” J.A. 383. On a painted wooden banister, Detective Isenhour noticed a latent shoeprint at the base of a column, which extended up to the roof and to the 13 open upstairs window. Using a fingerprint dusting brush and lifting powder, Detective Isenhour lifted the print to aid in finding the perpetrator. The State district attorney asked Detective Isenhour if he “ever relinquished possession and control of [the latent lift] since the time that [he] made it,” to which Detective Isenhour replied, “No, sir.” J.A. 408. Detective Isenhour then testified he took Petitioner’s shoes and made impressions of them on May 10, 1976, just after Petitioner was arrested. Then, the following exchange occurred: Q: What did you do on . . . the 11th of May, sir? A: I went to Raleigh, North Carolina, to the State Bureau of Investigation Lab. . . . Q: Officer Isenhour, did you take any items with you when you went? A: I did. Q: What did you take? A: I took the pair of shoes which I received from Mr. Long . . . . I took two inked impressions, one of the left shoe and one of the right shoe which I had made on May the tenth, and I took the latent lift, which I had lifted from the top of the bannister column at [Ms. Bost’s residence]. J.A. 411. Detective Isenhour was then asked, “[H]as that pair of shoes ever left your possession and control since that time?” and Detective Isenhour responded, “It has not.” Id. at 412; see also id. at 413 (testifying the inked impressions also never left his possession and control). Detective Isenhour then testified that he received a “black leather type coat,” a “green cloth toboggan,” a “pair of black gloves,” and “a quantity of match books” from 14 Detective Taylor. J.A. 415. Detective Isenhour also testified that the leather jacket had been “in [his] custody and control” “since [he] received it.” Id. SBI latent evidence expert Dennis Mooney testified that the latent prints on Ms. Bost’s banister “could have been made” by Petitioner’s shoes. J.A. 423. But he did not “have an opinion satisfactory to [him]self. . . as to whether or not” they were, and it was not a “positive identification.” Id. at 424, 425. Agent Mooney further explained that in his six years of examining latent prints, there had been occasions when he could make a positive identification, but this was not one of them. 2. Petitioner’s Case Again, the rape occurred between 9:30 and 9:45 p.m. on Sunday, April 25, 1976. During trial, several witnesses testified that on that day, Petitioner had attended a class reunion planning meeting and socialized with friends until around 8:00 p.m., at which point a friend took him home. He made plans with those same friends to go to a party in Charlotte later that night. Petitioner’s mother testified that Petitioner was at home with her from around 8:30 p.m. until “about twenty-five minutes after” 10:00 p.m. J.A. 477. During this time, Petitioner was in his room on the second floor of their home. Petitioner participated in a group phone conversation with his mother; the mother of his son, Janice Spears; and his two year old son (who was with Spears). Petitioner’s mother talked to her grandson for five or ten minutes on a downstairs phone and then hung up. Petitioner also talked to his son for five or ten minutes. Spears talked to Petitioner for a period of time, and Spears 15 testified the entire phone call lasted approximately “forty-five minutes,” from around 9:00 p.m. to 9:45 p.m. J.A. 468. Petitioner’s mother further testified that during the time period between 8:30 p.m. and approximately 10:25 p.m., when she was not on the phone, she was in the house, in a place where she could easily see if anyone were to leave the house. She testified that the windows leading out of Petitioner’s room were rollout windows, so that there was no way someone could climb out of them. After the phone call with Spears, Petitioner’s mother heard Petitioner playing tapes in his room. When Petitioner’s father arrived home around 10:25 p.m. with the family car, Petitioner came downstairs and took the car from his father to head to Charlotte to meet his friends at the party. Petitioner’s father corroborated that he arrived home “sometime after ten.” J.A. 494. Other witnesses testified to seeing Petitioner from 10:30 p.m. on that night. Ruthene Stokes testified that she saw Petitioner around 10:30 p.m., when he drove by Barber Scotia College and offered her a ride to the party in Charlotte. She remembered the time because her boyfriend was late picking her up. Doris Nesbitt, Petitioner’s girlfriend at the time, testified that Petitioner arrived at her house in Charlotte around 11:00, they went to the party together, and they left the party around 2:30 a.m. Terrence Stocks testified that he saw Petitioner between 11:00 p.m. and midnight at the party in Charlotte and did not notice cuts, scratches, bruises, or anything unusual about him. Kenneth Byers testified he saw Petitioner at the class reunion meeting earlier that night, and also afterward at the Charlotte party. He said Petitioner was wearing the same clothes in Charlotte that he was wearing at the class reunion meeting, and he did not have visible scratches, bruises, or marks on his 16 jacket or his face. Many witnesses also testified to the common trend of African American men wearing leather jackets at that time. Two witnesses also recalled that Petitioner was wearing light khaki pants, not blue jeans, that night, because Byers made fun of Petitioner for wearing khaki pants. None of the witnesses ever saw Petitioner wear a green toboggan, but rather, the witnesses testified he usually wore a leather hat. 3. Closing Arguments and Verdict During the State’s closing argument, it relied on three main points: (1) Ms. Bost’s identification was accurate; (2) it was corroborated by and consistent with the physical evidence in the case; and (3) the police acted honestly. The prosecutor argued, “[Ms. Bost’s] testimony is clothed with the clothing of corroboration at every point. Everything Ms. Bost told you. Every word she uttered is fully and entirely corroborated by the evidence as was seen by the officers in her home.” J.A. 526 (emphases supplied). The prosecutor reiterated, “We have shown that Ms. Bost’s testimony is not only accurate, but totally consistent with every piece of physical evidence existent. Everything she says happened that is capable of being corroborated by physical evidence . . . is so corroborated.” Id. at 536 (emphases supplied). Defense counsel’s closing argument highlighted the weakness of identification evidence in general, and cross-racial identifications in particular. Defense counsel also argued, “There is no evidence that Ms. Bost, who testified that she scratched and fought for her life, . . . . No evidence [of] any skin, hair, or anything the State can connect with 17 Ronnie Wallace Long. None of that was brought here, and if it exist[s], we don’t know about it, and maybe it wouldn’t fit this man, if it exist[s].” J.A. 568. In response, on rebuttal, the State argued, [I]f the suggestion is that these police officers have wronged this man, what would have taken my good police officers, in the still of the night, to go down there and take this jacket and rub it up and down on that white post, then take it to the S.B.I. lab and say: Hey man, is that white paint on that coat? But you don’t see any on here. . . . All perfect honesty . . . . J.A. 576 (emphasis supplied). Then, the State referred to the fact that SBI Agent Mooney did not say Petitioner’s shoes were a conclusive match for the latent print on the banister, arguing, “The officer says: No, I cannot take this and prove to you conclusively that this shoe made [the print]. He said a similar shoe did. Complete and total honesty. . . . [I]t’s another example of the honesty of these police officers.” Id. at 579 (emphasis supplied). The jury returned a guilty verdict for first degree rape and first degree burglary, and Petitioner was sentenced to life imprisonment on both counts. C. State Court Post-Conviction Petitioner filed a direct appeal, contending that the “pretrial identification procedures were so impermissibly suggestive that admission of [Ms. Bost’s] in-court identification violated due process of law.” State v. Long, 237 S.E.2d 728, 730 (N.C. 1977). The North Carolina Supreme Court held that the identification procedures were not “impermissibly suggestive” and therefore, Ms. Bost’s out of court identifications were “properly admitted.” Id. at 731 (internal quotation marks omitted). Petitioner also argued 18 that the search of his automobile at the police station was illegal because it was conducted without his consent. The North Carolina Supreme Court also rejected this argument, relying on the trial court’s finding that Petitioner “gave his permission to Officers Vogler and Taylor to search his automobile.” Id. at 732 (internal quotation marks omitted). Finally, Petitioner argued the latent shoe print evidence should not have been admitted because it was nonprobative. Id. at 734. Again, the state supreme court rejected this argument, explaining, “Officer Mooney’s testimony on direct examination that the shoe print corresponded with shoes taken from [Petitioner] at the time of his arrest was clearly competent as tending to connect the accused with the crime.” Id. at 734. Nearly ten years later, on August 1, 1986, Petitioner filed a motion for appropriate relief contending: (1) the police illegally searched his car at the police station without his consent, resulting in the discovery of the toboggan, gloves, and matches; (2) his jury was improperly composed, demonstrating a “systematic exclusion of blacks,” J.A. 636 -- of the 49 jurors called for selection, only two were African American -- and in support, providing testimony from the Chairman of the Cabarrus County Jury Commission, who stated that the local sheriff single-handedly removed prospective jurors whom he believed “are supposed to be disqualified” from the jury roll lists, id. at 640–41; and (3) his trial and appellate counsel were ineffective. After a hearing, the state court denied Petitioner’s motion on July 8, 1988. 19 D. The First Federal Habeas Petition On April 24, 1989, Petitioner filed a pro se petition for a writ of habeas corpus in the district court pursuant to 28 U.S.C. § 2254, alleging: (1) the trial court improperly admitted evidence seized during an illegal search of Petitioner’s vehicle; (2) the trial court denied Petitioner an adequate challenge to the selection of the jury; and (3) Petitioner received ineffective assistance of trial counsel generally and in regard to the challenge to the jury pool selection process. The district court dismissed the petition on May 3, 1990. See Long v. Dixon, No. 4:89-cv-278 (M.D.N.C. May 3, 1990). E. The 2005–2009 State MAR Proceedings On April 20, 2005, the North Carolina Center for Actual Innocence (“NCCAI”) filed a motion for location and preservation of evidence, asking the MAR Court to (1) “requir[e] the State to locate and preserve evidence obtained” in Petitioner’s case; (2) require the CPD, SBI, and district attorney’s office “to produce for inspection the evidence and records of their evidence custodians relating to any and all physical evidence collected” in Petitioner’s case; and (3) require the State “to provide defense counsel with any and all test results or reports conducted by the [SBI] or other law enforcement agencies on the physical evidence collected in this matter.” J.A. 766. The MAR Court granted the motion and, on June 7, 2005, on its own initiative, the MAR Court ordered the NorthEast Medical Center to locate and preserve all biological evidence in its possession as well. 20 At a hearing on June 16, 2005, the SBI reported “the only evidence found by the SBI” was the latent shoe print. J.A. 776. At the same hearing, the CPD opined that the only item it had in its possession was the master case file, which the district attorney had reviewed and found “nothing . . . of evidentiary value.” Id. Ultimately, on July 12, 2005, Petitioner received medical records released by NorthEast Medical Center, after they were viewed in camera and partially redacted by the MAR Court. These medical records revealed the following: • Ms. Bost’s “wrists [we]re markedly sore and swollen . . . [This] occurred when she was trying to beat her assailant over the head with her hands and wrists,” and her “fingernails [we]re all sore and some of them have been bent backward which the patient thinks occurred when she was trying to scratch her assailant and fighting back”; • Dr. Monroe combed Ms. Bost’s pubic hair and put it into a plastic bag; • Dr. Monroe took Ms. Bost’s vaginal fluid and placed it on two swabs, which were then placed into a test tube and stoppered; and • An authorization, dated April 26, 1976, indicated that the “combed pubic hair” and “1 test tube with vaginal swabs and secretions” were “released to Marshall J. Lee of the Concord Police Department.” J.A. 793, 777 (internal quotation marks omitted). In addition, the CPD master case file revealed that “certain physical evidence” was delivered by Detective Isenhour to the SBI lab on May 11, 1976, for testing. J.A. 777. In the master case file were two reports from Detective Isenhour (collectively, the “Isenhour reports”). One, which is undated, indicated that Detective Isenhour took only the latent 21 shoe print evidence to the SBI lab for testing. See id. at 1484–85 (the “Undated Report”). The other, dated May 12, 1976, indicated that Detective Isenhour actually took 13 other pieces of evidence from the scene, Ms. Bost, and Petitioner to the SBI lab as well. See id. at 1480–83 (the “May 12 Report”). Of note, the Undated Report states that clothing belonging to Petitioner was “collected by [Detective Isenhour] and held for investigative uses,” contradicting the May 12 Report, which demonstrates the clothing was actually taken to the SBI lab. Id. at 1484. And six months later, in January 2006, reports documenting the SBI test results of that evidence were finally turned over to Petitioner (the “SBI reports”). Together, the Isenhour reports and SBI reports demonstrated the following: • In addition to the latent shoeprint evidence, Detective Isenhour took the following items to the SBI for testing: (1) a green toboggan, (2) black gloves, (3) a black leather jacket, (4) one plastic bag containing a head hair from Petitioner, (5) one plastic bag containing a pubic hair from Petitioner, (6) one glass test tube containing carpet fibers taken from Ms. Bost’s home, (7) another glass test tube containing carpet fibers from Ms. Bost’s home, (8) one glass test tube containing paint from the scene, (9) one plastic bag containing the suspect’s hair from the scene, (10) one plastic bag containing pubic hair from Ms. Bost, (11) one plastic bag containing match books, (12) one plastic bag containing three partially burned matches, and (13) one plastic bag containing Ms. Bost’s clothing, see J.A. 1454; • Detective Isenhour left the items in SBI custody for eight days, see J.A. 1464, 1468, contrary to his trial testimony that the items remained in his “custody and control” the entire time, id. at 792, 415; • The suspect hair found at the scene was “different from [Petitioner’s] hair,” J.A. 1472; 22 • The “suspect” hair found at the scene had “a reddish sheen,” was “more reddish” than Petitioner’s hair, and “could be negroid or . . . Mongolian,” J.A. 1471–72; • “No hair or hair fragments similar to [Petitioner’s] were found in [Ms. Bost’s] clothing,” J.A. 1474; • Examination of the toboggan, black gloves, and leather jacket “failed to reveal the presence of any fibers or paint similar to those [submitted],” J.A 1454; • Examination of the matches found in Ms. Bost’s bedroom “failed to reveal sufficient identifying characteristics to allow the examiner to give an opinion with regard to their origin relative to the matchbooks [found in Petitioner’s car],” J.A. 1454; • Analysis of the shoe print revealed “[t]here were an insufficient number of distinct characteristics noted by which to effect any identification,” but Petitioner’s shoes “could have made” the prints, J.A. 1464. Once this evidence had come to light, on March 15, 2007, Petitioner filed another motion to locate and preserve evidence from University of North Carolina (“UNC”) Hospitals “to search for any test reports relating to any biological evidence in this case,” including “serological testing.” J.A. 777–78. On April 26, 2007, UNC Hospitals informed Petitioner that “after a diligent search, [it] ha[d] no medical records in the name of [Ms. Bost] or [Petitioner].” Id. at 778. On August 27, 2008, Petitioner filed the MAR that is at the center of this appeal (the “2008 MAR”). The 2008 MAR asked the state court for a new trial based on due process violations pursuant to Brady v. Maryland and the North Carolina Constitution. 23 It contended the medical records, Isenhour reports, and SBI reports were suppressed, favorable, and material to Petitioner. The MAR Court held a hearing on the 2008 MAR on November 20, 2008. Trial counsel from both the prosecution and defense testified, as did Agent Rick Cone, who worked in the SBI Trace Evidence Division from 1971 to 1982 and tested the matches; Jeffrey Hollifield, owner and operator of a private forensic lab; and Jennifer Remy, a hair analyst with the SBI. Notably, at the MAR hearing, the assistant district attorney who prosecuted Petitioner was asked, “If you had had the SBI reports would you have provided them to defense counsel?” He answered, “Oh, yeah. Oh, yes. . . . [A]ny kind of scientific tests . . . was just automatic.” J.A. 1145. On February 25, 2009, the MAR Court denied Petitioner’s 2008 MAR. The MAR Court concluded that, pursuant to Brady, Petitioner “has not shown by a preponderance of the evidence that the claimed evidence was withheld by the State, that it was exculpatory, or that the result likely would have been different with the claimed evidence.” J.A. 1359 (emphasis supplied). It also concluded that the “cumulative [e]ffect of any items with any value is so minimal that it would have had no impact on the outcome of the trial,” and Petitioner “failed to prove by a preponderance of the evidence that the cumulative [e]ffect of these items are material or would have changed the result at trial.” Id. (emphases supplied). Petitioner appealed to the North Carolina Supreme Court, which split three to three, with one Justice abstaining, on the merits of his claims, resulting in a procedural affirmance of the MAR Court’s decision. See State v. Long, 705 S.E.2d 735 (N.C. 2011). 24 F. Instant Federal Habeas Proceedings On May 26, 2016, Petitioner filed the Petition pursuant to 28 U.S.C. § 2254, after this court granted his motion to file a second or successive petition pursuant to 28 U.S.C. § 2244, see Order, In re Long, No. 16–295 (4th Cir. filed May 24, 2016), ECF No. 6. Attached to the petition was yet another group of withheld evidence -- results from the CPD (disclosed on June 4, 2015) demonstrating that 43 latent fingerprints had been lifted from the crime scene and tested. 6 An independent expert analyzed the prints, and excluded Petitioner as the source of the prints. In the Petition, Petitioner alleges the State violated his constitutional rights pursuant to Brady v. Maryland “by failing to disclose SBI reports and notes, the victim’s medical records, and Detective Isenhour’s reports,” all of which resulted in a “verdict unworthy of confidence.” J.A. 62, 64 (formatting omitted). Petitioner recognizes this is a successive petition, and his claims are filed outside of the one-year statute of limitations set forth in AEDPA. See 28 U.S.C. §§ 2244(b), (d)(1). Nonetheless, he makes a claim of actual innocence to serve as a gateway to his substantive claims. The State filed a motion for summary judgment, arguing, “A correct application of Brady shows the result reached by the MAR [C]ourt . . . is objectively reasonable,” and in 6 The fingerprint evidence was unearthed thanks to the efforts of the North Carolina Innocence Inquiry Commission (the “NCIIC”). At Petitioner’s counsel’s request, the NCIIC agreed to review his case under the Postconviction DNA Testing Assistance Program. Upon inquiry by the NCIIC, the CPD located three envelopes containing 43 latent fingerprint lifts the CPD had taken from the scene. 25 any event, “Petitioner has failed to show a reasonable probability of a different result, as required by Brady.” Br. Supp. Mot. at 38, Long v. Perry, No. 1:16-cv-539 (M.D.N.C. filed June 17, 2017), ECF No. 7. The magistrate judge issued an Opinion, Order, and Recommendation (the “Recommendation”), advising that the district court should grant the State’s summary judgment motion and dismiss the Petition. 7 Specifically, the magistrate judge recommended the following, which the district court adopted in full. 8 1. Favorability Brady v. Maryland instructs that “the suppression by the prosecution of evidence favorable to an accused” violates due process where the evidence is “material either to guilt or to punishment, irrespective of the good faith or bad faith of the prosecution.” 373 U.S. 83, 87 (1963). As to favorability, the district court reasoned that the MAR Court’s conclusion that the SBI reports on paint, carpet fibers, and hair were not exculpatory “runs contrary to” the Supreme Court’s decision in Kyles v. Whitley, 514 U.S. 419 (1995). J.A. 1659. Rather, 7 Before the instant dismissal, the district court dismissed the Petition without prejudice to give Petitioner a chance to exhaust his unexhausted fingerprint-based Brady claim. However, Petitioner then filed a request for a certificate of appealability and an appeal with this court, claiming that he “unequivocally disclaimed . . . any independent claim based upon newly discovered latent fingerprint evidence”; therefore, this court remanded, and held Petitioner did not “present the district court with a mixed petition that required dismissal.” Long v. Perry, 699 F. App’x 260, 261 (4th Cir. 2017). 8 For ease of reference, we refer to the magistrate judge’s Recommendation, which the district court adopted, as the “district court.” The district court added no substantive analysis in its order adopting the Recommendation. 26 the evidence “would have had some weight and its tendency would have been favorable to Petitioner.” Id. at 1660 (alteration omitted). And on the analysis of the matches, the district court concluded the MAR Court “also contradicted Kyles.” Id. In addition, the district court pointed out that, as a factual matter, the MAR Court unreasonably determined that the report on the matches “actually favored the state more than Petitioner” because the testimony it relied on (Detective Taylor stating the matches did not match the book in Petitioner’s car) was actually made outside the presence of the jury. Id. at 1661. And with regard to the medical records, the district court pointed out that the MAR Court generally “interpreted the concept of favorable evidence too narrowly” because it conflated the terms “exculpatory” and “favorable.” Id. at 1663. Nonetheless, the district court found the MAR Court “did not misapply” the law with regard to the medical records because “the alleged acts and omissions of Dr. Monroe and/or staff working under his direction cannot be imputed to the state” and thus, does not “impeach the quality of the state’s investigation.” Id. at 1664. However, the authorization reflecting that Sergeant Lee took possession of Ms. Bost’s biological evidence, and “suggest[ing] that the CPD did not disclose the existence of the rape kit to the prosecution and did not maintain adequate safeguards for that evidence,” “does possess a degree of favorability to Petitioner, in that it tends to impeach the quality of the state’s investigation.” Id. at 1665–66 (footnote omitted). The district court also recognized the impeaching nature of the withheld evidence, concluding the MAR Court “unreasonably applied Brady and Kyles in failing to deem favorable the evidence of conflicts between Detective Isenhour’s testimony and the SBI reports, as well as the evidence of the two differing versions of Detective Isenhour’s 27 summary reports.” J.A. 1669. Specifically, the district court noted the MAR Court “failed to account for the impeaching value of the two differing versions of the summary reports” -- that is, the Undated Report, which stated that Detective Isenhour only took shoeprint evidence to the SBI as he testified, and the May 12 Report, which “listed the numerous additional items of evidence Detective Isenhour delivered to the SBI about which he did not testify.” Id. at 1669–70. The State does not challenge these favorability conclusions on appeal. 2. Materiality Regardless of the MAR Court’s errors on favorability, the district court hinged its dismissal of the Petition on materiality. It first recognized that the MAR Court’s “phrasing of the [Brady] burden” does not match the standard adopted in Kyles. J.A. 1679. Nonetheless, “in another conclusion of law, the  MAR Court indicated that the withheld evidence would have had no impact on the outcome of the case.” Id. at 1680 (emphasis in original). Thus, “[b]ecause the  MAR [C]ourt found that no chance of a different outcome existed, by logical extension, the  MAR [C]ourt concluded that Petitioner had not shown a ‘reasonable probability’ of a different outcome.” Id. (emphasis in original). The district court also reasoned that the suppressed evidence does not “undermine the victim’s identification in any material way.” Id. at 1683. Thus, the MAR Court “reasonably applied the United States Supreme Court’s Brady jurisprudence in concluding that the evidence in question, considered cumulatively, did not qualify as material.” Id. at 1693–94. 28 The district court granted summary judgment for the State and dismissed the Petition, but issued a certificate of appealability (“COA”) on the following question: “Did the state court unreasonably apply Supreme Court precedent in concluding that the suppressed exculpatory evidence in question, considered cumulatively, did not qualify as material?” J.A. 1725. Petitioner timely appealed. II. We review the district court’s dismissal of a petitioner’s § 2254 petition de novo. See Grueninger v. Dir., Va. Dep’t of Corr., 813 F.3d 517, 523 (4th Cir. 2016). A petitioner is entitled to relief on his § 2254 petition if the adjudication on the merits of his claim in state court: (1) resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or (2) resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding. 28 U.S.C. § 2254(d). III. Unreasonable Application of Clearly Established Federal Law This appeal centers on the MAR Court’s conclusion that, as to materiality, the “cumulative effect of any items with any value is so minimal that it would have had no impact on the outcome of the trial.” J.A. 1359 (hereinafter, “the No Impact Conclusion”). The district court and the dissent have recognized that the MAR Court unreasonably applied Supreme Court law in imposing an erroneously high burden -- the MAR Court 29 required Petitioner to demonstrate the withheld evidence would have changed the result at trial “by a preponderance of the evidence,” rather than asking him to demonstrate a “reasonable probability of a different result.” J.A. 1679–80; Dissenting Op. at 109 (“[The] state court’s conclusion described a preponderance burden, [which] directly contradicts Supreme Court precedent.”); see Kyles, 514 U.S. at 434 (“[A] showing of materiality does not require demonstration by a preponderance that disclosure of the suppressed evidence would have resulted ultimately in the defendant’s acquittal.”). However, because the MAR Court found the cumulative effect to have “no impact,” the district court and the dissent believe it did not unreasonably apply federal law. In other words, because the MAR Court found “no chance of a different outcome,” it necessarily concluded that there would be no “‘reasonable probability’ of a different outcome.” Id. at 1680 (emphasis in original). This conclusion is wrong for three reasons. First, the No Impact Conclusion was informed by the incorrect burden imposed by the MAR Court. Second, the MAR Court’s erroneous favorability conclusion necessarily rendered its No Impact Conclusion unreasonable. Third, in any event, the No Impact Conclusion itself is objectively unreasonable. We address each in turn. A. The No Impact Conclusion was Informed by the Incorrect Burden 1. First and foremost, the No Impact Conclusion refers to, and necessarily depends on, the MAR Court’s erroneously imposed burden. The MAR Court reasoned: 30 As to the cumulative [e]ffect of the items of evidence the defense alleges they did not receive, this court finds, based on the findings of fact and conclusions of law stated herein, that the contents of several of the items the defense alleges they did not receive were fully addressed in front of the jury; that other materials contained in the reports were more favorable to the State’s case than the defendant’s; and that any remaining matters that were not presented to the jury were of little or no value to the case as a whole; and that the cumulative [e]ffect of any items with any value is so minimal that it would have had no impact on the outcome of the trial. J.A. 1358–59 (emphases supplied). The bolded language is key to the analysis here because the “findings of fact and conclusions of law” referenced in the MAR Court’s decision are actually “based on” the preponderance burden that is contrary to Brady and Kyles. Id. at 1358. Thus, in the paragraph quoted above, when the MAR Court concludes there was “no impact” on the outcome of the trial, it necessarily refers to its pervasive error that, as to each individual piece of evidence, Petitioner did not demonstrate “by a preponderance of the evidence” that the “result of the proceeding would have been different if the evidence had been disclosed.” See J.A. 1356 ¶ 2 (general), ¶ 7 (SBI reports), ¶ 8 (shoeprint analysis), ¶ 9 (hair analysis), ¶ 10 (paint, fiber, and matches analysis), ¶ 11 (Isenhour reports). The No Impact Conclusion, therefore, is merely a cumulative view of the impact of each erroneously analyzed piece of Brady evidence, not a separate ground for relief. That the No Impact Conclusion is colored by the erroneous Brady burden comes into even clearer focus in the paragraph directly following the No Impact Conclusion, which demonstrates that the MAR Court viewed the cumulative effect of the evidence through the improper Brady lens: 31 When balancing the strength of the State’s case with the cumulative [e]ffect of the items of evidence the defense alleges they did not receive, [Petitioner] has failed to prove by a preponderance of the evidence that the cumulative [e]ffect of these items are material or would have changed the result at trial. J.A. 1359 (emphasis supplied). Finally, the No Impact Conclusion explicitly states that it is “based on the findings of fact and conclusions of law stated herein,” and those findings and conclusions are infected by the erroneous Brady burden. J.A. 1358–59 (emphasis supplied). The MAR Court then sums up, explaining, “In summary, [Petitioner] has failed to prove by a preponderance of the evidence that there exists newly discovered evidence . . . of a nature as to show that upon another trial a different result would probably be reached.” Id. at 1359 (emphasis supplied). The phrase “[i]n summary” necessarily means that the MAR court viewed the No Impact Conclusion through the erroneous Brady lens, and could not have relied on it as a ground divorced from that incorrect standard. 2. The dissent relies on an outlier Supreme Court decision that the State itself discounted at oral argument before the en banc court, Wetzel v. Lambert, 565 U.S. 520 (2012). See Oral Arg. at 57:45–58:00, Long v. Hooks, No. 18–6980 (4th Cir. May 7, 2020), https://www.ca4.uscourts.gov/oral-argument/listen-to-oral-arguments (State attorney: “I don’t think this is a Wetzel case.”). In the face of a glaring and pervasive constitutional error -- which the dissent acknowledges -- appearing several times in the MAR Court’s decision, the dissent, citing Wetzel, nonetheless bends over backward to pluck out a single 32 phrase of a single sentence of the MAR Court’s Conclusions of Law as separately “sufficient” to reject Petitioner’s § 2254 claim. Dissenting Op. at 115. We disagree with this contorted view. The “alternative ground” doctrine set forth in Wetzel does not contemplate a case such as this. In Wetzel, the state court addressed a post-conviction Brady claim made by defendant James Lambert, who was convicted and sentenced to death for the murder of two patrons during a robbery of a bar. See 565 U.S. at 521. Lambert argued that the Commonwealth of Pennsylvania withheld a police activity sheet, which noted that one of the Commonwealth’s primary witnesses, Bernard Jackson, who was involved with the murders and identified Lambert as one of his accomplices, also noted that a man named Woodlock was “a co-defendant.” Id. The police activity sheet, however, did not indicate the crime to which Jackson was referring (and Jackson had been involved in other robberies). See id. Nonetheless, Lambert claimed the police activity sheet was exculpatory because “it suggested that someone other than or in addition to him” was involved with the murders. Id. The Pennsylvania Supreme Court rejected the Brady claim, holding that the police activity sheet was not material for two reasons: first, it was ambiguous as to the particular crime in which Woodlock was involved, and the idea that “someone else was involved in the . . . robbery [was] purely speculative at best”; and second, the activity sheet “would not have materially furthered the impeachment of Jackson at trial as he was already extensively impeached by [other witnesses].” Wetzel, 565 U.S. at 523 (internal quotation marks omitted). On federal habeas review, the district court denied the writ, but the Third Circuit reversed, relying only on the impeachment grounds and concluding that it was “patently 33 unreasonable for the Pennsylvania Supreme Court to presume that whenever a witness is impeached in one manner, any other impeachment evidence would be immaterial.” Id. (internal quotation marks omitted). The United States Supreme Court reversed, holding that the Third Circuit erred because it “overlooked the determination of the state courts that the [police activity sheet was] entirely ambiguous,” as an alternative to the impeachment determination. Wetzel, 565 U.S. at 524 (internal quotation marks omitted). Significantly, the Court noted that the state court used the word “[m]oreover,” “confirming that [the impeachment determination] was an alternative basis for its decision” Id. at 524–25 n.* (emphasis supplied); see also id. at 524 (describing the impeachment determination as an “alternative ground”). If the ambiguity determination was reasonable (which the Court did not decide), the impeachment decision was “beside the point.” Id. Thus, the Court reasoned that it would not impose the burden of retrial “unless each ground supporting the state court decision is examined and found to be unreasonable under AEDPA.” Id. at 525 (emphasis in original). Here, unlike the ambiguity and impeachment grounds considered in Wetzel, the No Impact Conclusion and the improper Brady analysis cannot be isolated from one another so as to be considered “alternative.” See Alternative, Oxford English Dictionary, 2d ed. (2004) (defining “alternative” as “[s]tating or offering the one or the other of two things of which either may be taken,” and “Of two things: Such that one or the other may be chosen, the choice of either involving the rejection of the other”); see also Alternative, Oxford English Dictionary Online, https://www.oed.com/view/Entry/5803?redirected From=alternative#eid (Nov. 22, 2019) (defining “alternative” as “characterized by . . . 34 disjunction”) (filed as ECF attachment). To the contrary, as explained above, the No Impact Conclusion is inextricably intertwined with Petitioner’s Brady claim because the No Impact Conclusion refers to, and necessarily depends on, the MAR court’s Brady conclusion. 3. For these reasons, the No Impact Conclusion cannot serve as an independent conclusion that there is no “reasonable probability” of a different result at trial. There is no dispute that the MAR Court held Petitioner to an improperly high burden, and thus it unreasonably applied clearly established federal law. B. Erroneous Favorability Conclusions Render the No Impact Conclusion Unreasonable The district court concluded that, as to favorability, the MAR Court misapplied Kyles at just about every turn. It concluded -- and the State does not challenge -- that as to the paint/carpet fiber report, the matches report, the hair report, the authorization for release of medical records to the CPD, and the false Isenhour testimony, the MAR Court unreasonably concluded this evidence was not favorable to Petitioner. It is clear the MAR Court had an improper view of favorable evidence, conflating favorable evidence with exculpatory evidence. For example, the MAR Court stated that Petitioner did not show “the claimed evidence was withheld by the State, that it was exculpatory, or that the result likely would have been different with the claimed evidence.” J.A. 1359 (emphasis supplied). But Kyles expressly rejects the notion that evidence must be “impeachment or exculpatory evidence” in order to be “favorable.” 514 U.S. at 450– 35 51. Rather, evidence is favorable under Brady if it would have “some weight” and a “tendency [to be] favorable” to Petitioner. Id. at 451. Indeed, the district court explained that the MAR Court’s conclusion that the withheld SBI Reports concerning the carpet fibers, paint, and hair found at the crime scene “did not qualify as exculpatory” was “contrary to Kyles,” and that such evidence “would have had some weight and its tendency would have been favorable to [Petitioner].” J.A. 1659–60 (quoting Kyles, 514 U.S. at 451). The district court also reasoned that, as a factual matter, the MAR Court “unreasonably determined that the SBI Matches report actually favored the state more than [Petitioner],” because Agent Cone did not testify at trial, and Detective Taylor’s testimony that the matches at the scene “didn’t match” those found in Petitioner’s car was made outside the presence of the jury. Id. at 1661–62. The district court also observed that, with regard to the rape kit evidence, the MAR Court “interpreted the concept of favorable evidence too narrowly” because it equated favorability with exculpation and failed to recognize the impeaching nature of that evidence. J.A. 1663. The district court noted that the failure to disclose the existence of the rape kit “possess[es] a degree of favorability to Petitioner, in that it tends to impeach the quality of the state’s investigation,” and the MAR Court “ran afoul of and/or unreasonably applied Kyles by denying relief on the grounds that the victim’s medical records did not qualify as exculpatory.” Id. at 1665–66. Finally, the district court explained that, in failing to account for the “impeaching value of the two differing versions of the” Isenhour reports and the conflicts between the Isenhour reports and the SBI reports, the MAR court unreasonably applied Brady and Kyles 36 in failing to deem such withheld evidence favorable. J.A. 1669 (emphasis in original). We agree. The Supreme Court has held, “When the reliability of a given witness may well be determinative of guilt or innocence, nondisclosure of evidence affecting credibility” violates the Constitution. Giglio v. United States, 405 U.S. 150, 154 (1972) (internal quotation marks omitted). The MAR Court, due to its misguided notions of favorability, failed to even consider this point in its materiality analysis of the cumulative impact of the withheld evidence. But remarkably, the district court then treats these MAR Court favorability errors and misapplications as having no bearing on the No Impact Conclusion. Yet per Supreme Court precedent, it must. In Kyles, the Supreme Court explained that “favorable evidence is material, and constitutional error results from its suppression by the government, ‘if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different.’” 514 U.S. at 433–34 (quoting United States v. Bagley, 473 U.S. 667, 682 (1985)) (emphasis supplied). Moreover, “the constitutional duty [to disclose under Brady] is triggered by the potential impact of favorable but undisclosed evidence.” Id. at 434 (emphasis supplied). The district court and the dissent fail to recognize that the MAR Court directly relied on its erroneous favorability findings to support its No Impact Conclusion. The MAR Court specifically referenced “the cumulative [e]ffect of any items with any value” as having “no impact,” but clearly the MAR Court did not properly analyze the “value” of the withheld evidence. J.A. 1359. Accordingly, because the MAR Court’s No Impact 37 Conclusion was infected by its incorrect and unreasonable favorability analysis, it involves an unreasonable application of clearly established Supreme Court law. C. The No Impact Conclusion is Objectively Unreasonable Finally, the No Impact Conclusion is an unreasonable materiality determination pursuant to § 2254(d). The MAR Court’s overall determination that the cumulative effect of the evidence at issue would have no impact on Petitioner’s trial is “so lacking in justification that [it] was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement[.]” Richardson v. Branker, 668 F.3d 128, 149 (4th Cir. 2012) (alteration in original) (quoting Harrington v. Richter, 562 U.S. 86, 103 (2011)). 1. Significance of Evidence In reaching the No Impact Conclusion, the MAR court unreasonably minimized (and thereby, improperly weighed) the significance of the withheld evidence, erroneously stating that “several of the items [of withheld evidence] were fully addressed in front of the jury” and other evidence had “minimal” value. J.A. 1359. This adds to the objective unreasonableness of the No Impact Conclusion. Tellingly, in its briefing, the State does little to assert a substantive defense of the officers involved. Instead, the State attempts to minimize the significance of the evidence. See, e.g., Resp’t’s Br. 22 (“The impeachment value of [Detective Isenhour’s report demonstrating he did not retain custody of the latent shoeprint] is marginal and could have 38 been explained by Isenhour as a mistake, misunderstanding, or the report itself could be incorrect.”); id. (“Regardless of the reason for th[e] inconsistency [in whether Detective Isenhour relinquished custody of the other evidence], it does not change the results of any of the SBI reports.”). While these may have been useful arguments for the State to make to the jury at trial, the rule is not that only unassailable evidence must be disclosed to the defense. Rather, it is clearly established federal law that any favorable and material evidence must be disclosed. On this point, three of the MAR Court’s conclusions and/or the State’s arguments in particular are worth addressing: (1) the suppressed evidence is cumulative of counsel’s arguments at trial; (2) defense counsel could have tested certain items or questioned witnesses at trial whether certain items had been tested, but made a strategic decision not to do so; and (3) the suppressed evidence does not undermine Ms. Bost’s identification of Petitioner as her rapist. Each of these claims is meritless. a. Cumulative Evidence The State argues the evidence it withheld was merely cumulative of defense counsel’s arguments at trial. But surely the State is aware (or at least should be) that it is elemental that counsel’s arguments are not evidence in a case. It is literally black letter law. See In re D.L., 603 S.E.2d 376, 382 (N.C. Ct. App. 2004) (“Statements by an attorney are not considered evidence.”); see also N.C. Pattern Jury Inst. Crim. 101.37 (“The final arguments of the lawyers are not evidence, but are given to assist you in evaluating the evidence.”). 39 Even if this were not the case, the State is wrong for two additional reasons. First, the SBI reports carry much more weight than simply an unsupported argument from counsel. If the SBI reports had not been suppressed, Petitioner’s trial counsel would not have had to ask the jury in closing argument to take his word for it that the forensic evidence could not connect Petitioner to the crime scene; rather, the jury could take the SBI’s word for it. Indeed, we are quite confident that if the SBI lab had found affirmative forensic evidence that tied Petitioner to the crimes charged, the State not only would have disclosed it, but would have been shouting it from the rooftops at trial. Further, if the SBI reports had not been suppressed, the prosecutor would not have been able to vouch for the officers in his closing argument that Ms. Bost’s identification was “totally consistent with every piece of physical evidence existent.” J.A. 536. Nor would the prosecutor have been able to opine as to the officers’ “perfect honesty.” Id. at 576. Certainly not. Second, even more material than one exculpatory test result is the cumulative effect of the legion of exculpatory test results in this case. As Petitioner’s trial counsel testified during the MAR Court’s 2008 evidentiary hearing: [T]he tests . . . have both an individual and a cumulative effect. . . . I got one test here that does not implicate you. Okay. I’ve got a second test that does not implicate you. And now the jury is paying attention. And now I’ve got a third test and a fourth test, and pretty soon it creates a snowball effect that you’re not the [suspect]. And that’s why I believe every one of those tests was critical. J.A. 1099–1100; see also id. at 1286 (operator of private forensic laboratory testifying that in a violent crime such as this one, it would be “very unlikely to not find some sort of trace 40 evidence” in any of the items submitted for analysis). Similarly, as noted in the amicus brief filed in this case by forensic science scholars: [J]urors use a coherence-based reasoning method, in which they integrate the whole of the evidence that they receive. That is, a piece of strong inculpatory evidence can make the entire evidence set appear inculpating. By the same token, including an exculpating item can push the evidence towards a conclusion of innocence. Critically, evidence is not independent: it is related, and thus the exclusion of evidence of innocence can make an entire case against a defendant seem far more compelling than it is. Amici Curiae Br. of Professors & Scholars 8 (citations and internal quotation marks omitted) (emphasis in original). In short, results of forensic analyses, whether inculpating or exculpating, are a critically important type of evidence. Indeed, the significant impact of this evidence makes it all the more important that forensic analyses be disclosed to defense counsel and the court in all cases. These tests are probative and can be powerful evidence of innocence; conversely, “[w]ithout a narrative to contrast the prosecution’s story, a jury will have little reason to give any weight to the defense.” Amici Curiae Br. of Professors & Scholars 8. b. Defense Strategy The State further argues that the suppressed evidence is not material because Petitioner’s trial counsel could have learned of the SBI lab results by (1) testing the evidence themselves; or (2) questioning witnesses at trial as to whether certain items had been tested. But, according to the State, Petitioner made a “strategic decision” not to do so. Resp’t’s Br. 29. Similarly, the district court faulted defense counsel for failing to ask 41 Dr. Monroe “whether or not he had performed any further testing on the spermatozoa,” such as connecting it to Petitioner. J.A. 1691 n.16. Such an approach turns the burden of proof completely on its head. The burden of proof in criminal cases rests with the State, and remains with the State throughout the course of the trial. It is unquestionably not a defense counsel’s responsibility to elicit testimony about potentially harmful forensic evidence against his/her own client by blindly questioning witnesses in front of the jury. It is axiomatic that it is not the defendant’s job to prove himself innocent. Rather, it is the State’s job to build the case against the defendant. And when the State tests evidence in an effort to build that case, it is the State’s responsibility to turn over the results to the defendant -- whether those results are inculpatory or exculpatory -- rather than hide the fact that the tests ever occurred in the first place. c. Victim’s Identification Finally, the State places undue weight on the “strength” of Ms. Bost’s identification of Petitioner as her assailant, repeatedly asserting that the suppressed evidence is immaterial because it does not “diminish” this identification. Resp’t’s Br. 30. But the suppressed evidence does not need to explicitly diminish Ms. Bost’s identification for it to be material. Viewed cumulatively, the suppressed evidence casts the alleged accuracy of Ms. Bost’s cross-racial identification in a questionable light, and completely undermines the crux of the State’s closing argument: that Ms. Bost’s identification was “totally consistent with every piece of physical evidence existent.” J.A. 536. 42 Here, the officers brought Ms. Bost into a courtroom where Petitioner was singled out and already cast in a negative light. “[T]he courtroom setting itself was enough to prejudice the victim’s identification of her assailant[,] as it suggested to the victim that the parties present were already in legal trouble.” Innocence Project Amicus Br. 24–25. And in the subsequent photo array provided by the police only 20 minutes later, Petitioner was the only one wearing a leather coat, which of course was the item of clothing Ms. Bost recalled her assailant wearing, as well as the item of clothing he was wearing in the courtroom. Indeed, the State’s conduct in suppressing the forensic test results likely caused Ms. Bost’s identification to carry more weight with the jury than it deserved, since there was zero forensic evidence to contradict it. In reality, every bit of forensic evidence in this case at worst directly contradicted Ms. Bost’s purported identification and, at best, failed to support it. The false bolstering of Ms. Bost’s identification in this case is highlighted by the prosecutor’s statements to the jury -- now known to be demonstrably untrue -- that “[e]very word [Ms. Bost] uttered is fully and entirely corroborated by the evidence as was seen by the officers in her home . . . [a]nd the latent evidence found by the officers,” J.A. 526–27, and that: Ms. Bost’s testimony is not only accurate, but totally consistent with every piece of physical evidence existent. Everything she says happened that is capable of being corroborated by physical evidence. . . is so corroborated . . . . Every piece of physical evidence points unerringly to the fact that [Ms. Bost] told you exactly what happened that night unerringly. J.A. 536. 43 Not true. Had the Isenhour reports and the SBI reports excluding Petitioner been disclosed, the jurors could have more credibly questioned and considered the reliability of Ms. Bost’s identification in the absence of other evidence. Accordingly, the fact that Ms. Bost testified that she had “[n]o doubt whatsoever” that Petitioner was her assailant only served to make the suppressed evidence, which did not corroborate her identification, all the more crucial for the jury to hear. J.A. 227. Withholding this evidence distorted the strength of Ms. Bost’s identification, and thereby, distorted the strength of the State’s case in total. 2. Cumulative Effect of Evidence Viewing the withheld evidence cumulatively as we must, there is no “possibility for fairminded disagreement” that the cumulative effect of the suppressed evidence -favorable to the defendant and suppressed by the State in order to make its own case appear stronger -- would have impacted Petitioner’s trial. Richardson, 668 F.3d at 149 (quoting Harrington, 562 U.S. at 103). a. Following is a list of evidence the jury would have had the opportunity to consider, but for the officers’ Brady violations. • Impeachment evidence against Detective Isenhour. The Isenhour reports and SBI reports would have completely undermined Detective Isenhour’s testimony and the State’s case, as he was one of the key witnesses. See Monroe v. Angelone, 323 F.3d 286, 315 (4th Cir. 2003) (“[T]he jury, had it had been shown that a major prosecution witness was testifying falsely, is likely to have been more sympathetic to 44 [the defendant’s] entire defense.”). The SBI reports demonstrate that he lied to the jury 9 about how many items he took to the SBI lab for testing, and he lied to the jury about whether he kept the items in his custody. 10 In fact, the only test results Detective Isenhour turned over were ones marginally favorable to the State. These reports, together with the missing rape kit, medical records, and 43 latent fingerprints, 11 demonstrate a pattern of deceitfulness and suppression that not only signifies that state actors conducted themselves in a corrupt manner, but also that they believed the withheld evidence was material enough to hide. • Impeachment evidence against Detective Taylor. The undisclosed SBI reports can also be used to impeach and undermine Detective Taylor’s testimony. He told the court out of the presence of the jury that the matches found at the scene “didn’t match” the matchbooks found in Petitioner’s car, even though they were of a “similar nature.” J.A. 335. But Detective Taylor’s testimony that the matches he retrieved from Petitioner’s family car were “of [a] similar nature” was also directly contrary to the withheld SBI Report, wherein SBI Agent Cone concluded that four of the five matchbooks collected in Petitioner’s family car were eliminated as possible origins for the burned matches in Ms. Bost’s home. Of note, although Agent Cone could not exclude the fifth matchbook on that basis, he indicated that the burned matches from the scene 9 We take issue with the dissent downplaying Detective Isenhour’s testimony as merely painting an “incomplete picture” of the testing he requested. Dissenting Op. at 103. No. The simple fact is that he lied on the stand at trial. 10 Of particular note, Detective Isenhour is now a convicted felon. On May 22, 2020, the State filed a letter with this court, explaining that in 1987, Detective Isenhour was convicted of possessing a stolen United States check and failure to appear, in violation of 18 U.S.C. § 1708 and § 3146. He was sentenced to a total of four years of imprisonment. See Letter, Long v. Hooks, No. 18–6980 (4th Cir. filed May 22, 2020), ECF No. 74 (citing J.A. 1599–1602). 11 The dissent downplays the fingerprint evidence, stating that Ms. Bost testified “immediately after the attack and at trial” that her attacker wore gloves. See Dissenting Op. at 104. But the dissent fails to acknowledge that Ms. Bost at first told officers the attacker “[c]ould possibly have been wearing gloves.” J.A. 370 (emphasis supplied). 45 “probably did not originate from” this matchbook. Id. at 1463. And if Detective Taylor is impeached, that would undermine his trial testimony that Petitioner gave consent for him to search his vehicle, which led to the discovery of the toboggan, the gloves, and the matches. It would also call into question his orchestration of Ms. Bost’s courtroom identification procedure. • The missing rape kit. Ms. Bost’s rape kit, which included pubic hair and a test tube with the suspect’s semen, 12 was provided to the CPD following Ms. Bost’s hospitalization. Specifically, the authorization for release form reflects, “Sergeant [sic] Marshall J. Lee with the CPD picked up the victim’s biological specimens at the hospital at 12:35 a.m. on April 26, 1976.” J.A. 1665. The complete absence of the specimen -- perhaps the most crucial evidence of all in a rape case -- with no paper trail whatsoever, calls into question other officers’ handling of evidence. • The legion of SBI test results that did not implicate Petitioner. The results of the SBI testing would also put the State’s argument in a new and damaging light. The State’s closing argument relied heavily on the idea that the victim’s identification was corroborated by every piece of physical evidence available. Not so. And at the MAR hearing, an operator of a private forensic laboratory testified that in a violent crime such as this one, it would be unlikely not to find 12 The dissent opines that in 1976, semen analysis “could only identify a man’s blood type if he secreted blood group antigens,” and “[t]oday, juries may expect precise forensic analyses of biological evidence, yet we must take care not to transplant those expectations where ‘modern forensic techniques . . . of course would not have been available.’” Dissenting Op. at 119–20 n.5 (quoting Babick v. Berghuis, 620 F.3d 571, 577 (6th Cir. 2010)). But this misses the point of the analysis. Kyles teaches that, in assessing the significance of withheld evidence, “however the evidence would have been used,” we ask if it “would have had some weight and its tendency would have been favorable” to the defendant. 514 U.S. at 451. Surely semen evidence -- the most crucial evidence in a rape case in 1976 and today -- would have “some weight” and a tendency to be favorable to Petitioner, even if the semen was only tested using an antigen test. Indeed, the dissent’s argument begs the question of why the semen evidence was collected in the first instance if it would be of no value, and why the State faults Petitioner for failing to test this evidence himself or cross-examine Dr. Monroe about it. The State cannot have it both ways. 46 some sort of trace evidence in any of the items submitted for analysis. b. Victim Identification In light of this new evidence the jury could consider, in our cumulative assessment, we must also look closely at the victim identification. It is important to note that Ms. Bost’s description of the perpetrator (1) did not match Petitioner; and (2) changed over time. Further, the courtroom identification procedure and follow-on photo array 20 minutes later were so suggestive that, coupled with the favorable SBI test results, they could cause a reasonable juror to pause. First, in the hours immediately following the rape, Ms. Bost described the perpetrator as “light skinned,” J.A. 168; a “yellow black man,” id. at 248, “not your normal black person,” id.; and “not a real black man,” id. at 201. But Petitioner is a black man with dark skin. See id. at 26; Exhibit, Long v. Perry, No. 1:16-cv-539 (M.D.N.C. filed May 26, 2016), ECF No. 1 Ex. 4. She also said she knew Petitioner by his voice, but Ms. Bost signaled to her neighbor that Petitioner was the attacker before she heard him speak. See J.A. 277 (neighbor agreeing that “[a]t the time [Petitioner] was talking to the Judge, Ms. Bost had already said . . . he’s the one”). She described his voice as “a black voice,” id. at 170, and a “natural black man’s voice,” id. at 181. During the rape, Ms. Bost said she had a good opportunity to see the perpetrator’s face, but she also testified that there was a struggle; that a toboggan was pulled down over his ears; that the perpetrator kept pushing her head to the side and saying, “don’t look at 47 my face,” J.A. 197; and she told Detective Taylor that “everytime [sic] she would turn her head to look at [the assailant], he would slam her head back down on the floor,” id. at 304. As set forth in the initial case report, in her initial description, Ms. Bost did not tell Officer Parnell the perpetrator had facial hair. In fact, she did not describe any facial features at all. But after having seen a picture of Petitioner with facial hair shown to her by the officers, she testified at trial the perpetrator had “[a] tiny little mustache and a sort of spotted, unshaven beard,” exactly as Petitioner looked in that picture. Id. at 158. She also first said he “[c]ould possibly” have been wearing gloves, and then later insisted he was wearing gloves. Id. at 370. Only 20 minutes after the courtroom identification, the officers took Ms. Bost to the police station and showed her six to eight photographs of suspects, one of which was Petitioner. Of note, Petitioner was the only person in the photos wearing a leather coat, which was the type of clothing Ms. Bost initially identified her assailant as wearing. She chose Petitioner’s photo as the photo of her assailant. Of further note, when the trial court asked Ms. Bost, “Is it possible [the officers] could have asked you to pick out [Petitioner by name]?” Ms. Bost replied, “They could have, but I don’t know.” J.A. 179. c. Considering both the exculpatory and impeachment effects of the suppressed evidence, together with the shortfalls in Ms. Bost’s identification and consistent testimony from alibi witnesses, the withheld evidence “could reasonably be taken to put the whole case in such a different light as to undermine confidence in the verdict.” Kyles, 514 U.S. at 435. It follows, then, that the MAR Court’s conclusion that the suppressed evidence 48 would have “no impact” on the outcome of trial is not just wrong, but patently unreasonable. IV. AEDPA Threshold Requirements In order to have his habeas petition entertained by a federal court, Petitioner must also satisfy the stringent requirements of AEDPA. The district court declined to address these requirements, opting to decide the matter on the merits. See J.A. 1654 (“Even assuming, arguendo, that no affirmative defense barred Petitioner’s Brady claim, that claim fails on the merits.”). Because his claims arise in this second or successive habeas application, Petitioner is required to demonstrate: “the factual predicate for the claim could not have been discovered previously through the exercise of due diligence”; and “the facts underlying the claim, if proven and viewed in light of the evidence as a whole, would be sufficient to establish by clear and convincing evidence that, but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” 28 U.S.C. § 2244(b)(2)(B)(i), (ii). Even though this court granted Petitioner’s motion for authorization to file a second or successive petition, “[c]learing this threshold is only [his] first step.” In re Stevens, 956 F.3d 229, 233 n.2 (4th Cir. 2020). Before a court can review the merits of a successive application, “it must determine that the applicant actually ‘satisfies the requirements of’ § 2244(b).” Id. (quoting 28 U.S.C. § 2244(b)(4)). In addition, Petitioner 49 must also demonstrate his § 2254 petition was filed within the AEDPA statute of limitations. See 28 U.S.C. § 2244(d). 13 A. Due Diligence Because of the extreme and continuous police misconduct in this case, the withheld items could not have been discovered previously, even with the exercise of due diligence on the part of Petitioner. The prosecutor in this case maintained an open file policy, and under these circumstances, “defense counsel may reasonably rely on that file to contain all materials the State is constitutionally obligated to disclose under Brady.” Strickler v. Greene, 527 U.S. 263, 283 n.23 (1999). In the context of a Brady claim, “a defendant cannot conduct the reasonable and diligent investigation . . . to preclude a finding of procedural default when the evidence is in the hands of the State.” Id. at 287–88 (internal quotation marks omitted). In the context of the materially indistinguishable diligence provision of § 2254(e)(2)(A)(ii), “[d]iligence . . . depends upon whether the prisoner made a reasonable attempt, in light of the information available at the time, to investigate and pursue claims in state court.” Williams v. Taylor, 529 U.S. 420, 435 (2000); see id. at 432 (in discussing 13 If Petitioner proves actual innocence as a gateway pursuant to § 2244(b)(2)(B), he necessarily survives the lower hurdle for statute of limitations purposes pursuant to § 2244(d)(1). See McQuiggin v. Perkins, 133 S. Ct. 1924, 1935 (2013) (concluding that cases in which new evidence shows “it is more likely than not that no reasonable juror would have convicted [the petitioner]” can bypass the one-year limitations period in § 2244(d)(1) (internal quotation marks omitted)). 50 due diligence requirement, explaining, “a person is not at fault when his diligent efforts to perform an act are thwarted, for example, by the conduct of another[;] [f]ault lies, in those circumstances, . . . with the person who interfered with the accomplishment of the act”). 14 Indeed, “[w]hen police or prosecutors conceal significant exculpatory or impeaching material in the State’s possession, it is ordinarily incumbent on the State to set the record straight.” Banks v. Dretke, 540 U.S. 668, 675–76 (2004); see also Douglas v. Workman, 560 F.3d 1156, 1181 (10th Cir. 2009) (per curiam) (petitioner satisfied due diligence requirement where “agents of the State prevented the facts underlying the claim from becoming known when they failed to turn over . . . undisclosed evidence”). And there is nothing that would have put Petitioner on notice that there were undisclosed lab reports or a document demonstrating that rape kit evidence was turned over to the CPD. Indeed, as noted, at the MAR hearing, the assistant district attorney who prosecuted Petitioner was asked, “If you had had the SBI reports would you have provided them to defense counsel?” He readily answered “yes.” J.A. 1145. Yet the CPD apparently withheld this favorable evidence even from the prosecutor. This case is unlike others where the undisclosed evidence was nonetheless accessible and known to defense counsel. See Blackman v. Davis, 909 F.3d 772, 779 (5th Cir. 2018) (petitioner did not satisfy due diligence requirement where state failed to disclose 911 call, but defense counsel was on 14 We recognize that the due diligence inquiries under § 2254(e)(2)(A)(ii) and § 2244(b)(2)(B)(i) involve two different points of reference. Namely, § 2254(e)(2)(A)(ii) is tethered to the petitioner’s prior state court proceedings, whereas § 2244(b)(2)(B)(i) is tethered to the petitioner’s prior federal habeas proceedings. Here, that distinction makes no difference, as the State continued to withhold evidence after both proceedings had concluded. 51 notice that a recording or transcript of the call existed because it was discussed at trial by the eyewitness and the paramedic who responded to the scene); Spitznas v. Boone, 464 F.3d 1213, 1226 (10th Cir. 2006) (where factual predicate was a letter that had been “contained in the state district court file since the time [the presentence report] was prepared,” petitioner could not satisfy due diligence requirement). Perhaps most striking about this case, however, is that even when Petitioner obtained habeas counsel in 2005 and filed the motion for location and preservation of evidence that led to the unearthing of the evidence presented in this appeal, it took years for the evidence to trickle out. The SBI initially reported that it had no evidence related to the case, and the CPD denied having any physical evidence. Only after CPD located its master case file did the slow process of obtaining the SBI reports begin. And then, after Petitioner thought he had received all of the withheld evidence in 2006, nearly a decade later emerged 43 latent fingerprints taken from the crime scene that had not been turned over. For these reasons, Petitioner satisfies the due diligence requirement of § 2244(b)(2)(B)(i). B. Actual Innocence Under AEDPA, Petitioner’s second or successive § 2254 petition can only survive if he can demonstrate “the facts underlying the claim, if proven and viewed in light of the evidence as a whole, would be sufficient to establish by clear and convincing evidence that, but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” § 2244(b)(2)(B)(ii). This is also referred to as the “actual 52 innocence” standard and can serve as a “gateway” through which otherwise procedurally defaulted or out-of-time claims can be addressed. See McQuiggin v. Perkins, 133 S. Ct. 1924, 1928 (2013) (statute of limitations); Schlup v. Delo, 513 U.S. 298, 327 (1995) (successive petitions). In viewing the “evidence as a whole,” a court makes its § 2244(b)(2)(B)(ii) decision “unbounded ‘by the rules of admissibility that would govern at trial’ -- based on ‘all the evidence, including that alleged to have been illegally admitted [and that] tenably claimed to have been wrongly excluded or to have become available only after the trial.’” United States v. MacDonald, 641 F.3d 596, 612 (4th Cir. 2011) (quoting Schlup, 513 U.S. at 327– 28). “Or, to say it another way, the ‘court must consider all the evidence, old and new, incriminating and exculpatory, without regard to whether it would necessarily be admitted under [evidentiary rules].’” Id. (quoting House v. Bell, 547 U.S. 518, 538 (2006)). The question, then, is whether the newly discovered evidence, and all the evidence old and new, clearly and convincingly demonstrates that “but for constitutional error, no reasonable factfinder” would have found Petitioner guilty of these crimes. § 2244(b)(2)(B)(ii). The district court did not reach this question in view of its decision granting the State’s summary judgment motion. Therefore, we express no opinion on the “evidence as a whole,” and we remand this issue to the district court for consideration of the actual innocence question in the first instance. And given the extensive suppression of evidence that occurred in this case, Petitioner, at his request, shall be afforded on remand the opportunity to conduct further discovery before the district court makes its final determination on the actual innocence question. 53 V. The dissent freely admits it is “beyond dispute” that the prosecution withheld evidence. Dissenting Op. at 94. And the dissent freely admits the MAR Court’s decision “directly contradicts Supreme Court precedent,” id. at 109, but chalks such contradiction up to “sloppy writing” and calls it a “peripheral problem,” id. at 95. Further, the dissent freely admits that because of the fallibility of eyewitness testimony, “particularly within the racial historical context,” an “injustice may have occurred.” Id. at 122. Yet, by hiding behind an ill-constructed façade of blind deference to the MAR Court, the dissent ignores the injustice of these admitted constitutional errors: A man has been incarcerated for 44 years because, quite simply, the judicial system has failed him. Rather than overstepping our judicial role, as the dissent contends, today we remain faithful to our oath by “administer[ing] justice.” 28 U.S.C. § 453. For the foregoing reasons, we hold that the MAR Court’s adjudication of Petitioner’s Brady claims resulted in a decision that is an unreasonable application of Supreme Court precedent and objectively unreasonable. Therefore, we vacate the district court’s dismissal of the Petition, and we remand for consideration of the actual innocence question in the first instance, with the Petitioner being afforded the opportunity for discovery before the district court makes a final determination on that actual innocence question. We urge the district court to act with dispatch, considering the age of Petitioner and number of years he has already been incarcerated. VACATED AND REMANDED WITH INSTRUCTIONS 54 WYNN, Circuit Judge, with whom Judges THACKER and HARRIS join, concurring: Without a doubt, no reasonable jury could find Mr. Long guilty based on the undeniable facts before us today: suppressed physical evidence failing to link Mr. Long to the crime scene, the perjured testimony of investigating officers, missing key biological evidence, and an eyewitness identification obtained through means now illegal in North Carolina. So, while I am in full accord with the majority that Mr. Long would be entitled to additional discovery if it were necessary, I believe there to be no need for it. Rather, justice demands that we immediately grant Mr. Long the relief he has pursued for fortyfour years. This case exemplifies how our current habeas precedent incentivizes and rewards bad faith on the part of police and prosecutors. Our habeas precedent rewards state actors guilty of Brady violations for committing additional constitutional violations in order to subject the Brady claim to a higher standard of review. But keeping an innocent man in prison should not be considered a “reward.” Such a result is of no benefit to the principles of law and order: for every innocent man behind bars due to mistaken identity, a guilty one remains free. Not just free from prison—free to commit additional heinous crimes. 1 Meanwhile, public confidence in the judicial system falters, making it more difficult for law enforcement to do their jobs—and weakening our democracy. 1 E.g., DNA Exonerations in the United States, Innocence Project, https://www.innocenceproject.org/dna-exonerations-in-the-united-states/ (last visited August 17, 2020) (noting dozens of murders and rapes committed by the actual perpetrators of initial crimes after innocent defendants falsely confessed to those initial crimes). 55 The violent racial history of this country necessarily informs the background of this case: a Black man accused of raping a white woman is tried in 1976 by an all-white jury in a county with strong ties to the woman’s family, because defense counsel feared that any attempt to relocate the case would land them instead in a county with significant controlling influence by the Klan. Those historical facts lend gripping context to the egregious constitutional violations at the heart of this case. At the end of the day, this record shows compellingly that we should give justice to Mr. Long now. I. A. From the outset, though I, like the majority, employ the shorthand “actual innocence” to refer to the standard presented in § 2244(b)(2)(B), it is important to emphasize that habeas petitioners need not come forward with irrefutable evidence pointing to a different perpetrator or establishing an unassailable alibi. Rather, the question is whether, “but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” 28 U.S.C. § 2244(b)(2)(B)(ii); see also McQuiggin v. Perkins, 569 U.S. 383, 395 (2013) (noting a similar standard for petitioners seeking an exception to the filing deadline of § 2244(d)(1), namely, they must present “new evidence show[ing] it is more likely than not that no reasonable juror would have convicted” them (internal quotation marks omitted)). For a reasonable factfinder to find the applicant guilty, of course, the factfinder would have to “f[i]nd him guilty beyond a reasonable doubt.” McLeod v. Peguese, 337 F. App’x 316, 324 (4th Cir. 2009) 56 (unpublished but orally argued); see also, e.g., In re Bolin, 811 F.3d 403, 408 (11th Cir. 2016) (referring to the reasonable-doubt standard). Thus, an actual-innocence analysis must take into consideration “the evidence as a whole.” § 2244(b)(2)(B)(ii). This means the Court looks at all evidence, “old and new, incriminating and exculpatory, without regard to whether it would necessarily be admitted under [trial evidentiary rules].” United States v. MacDonald, 641 F.3d 596, 612 (4th Cir. 2011) (quoting House v. Bell, 547 U.S. 518, 538 (2006)). 2 So, the Court must consider “all the evidence, including that alleged to have been illegally admitted [and that] tenably claimed to have been wrongly excluded or to have become available only after the trial.” Id. (alteration in original) (quoting Schlup v. Delo, 513 U.S. 298, 328 (1995)). 3 2 New evidence can, of course, include advances in scientific knowledge that cast old evidence in a new light. Cf. Gimenez v. Ochoa, 821 F.3d 1136, 1143 (9th Cir. 2016). A defendant convicted before the existence of DNA testing, for example, could certainly be exonerated by a later DNA analysis. So, too, may we consider advances in our understanding of the psychology of memory formation, eyewitness identification, and implicit bias. See Brian R. Means, Federal Habeas Manual § 11:30 (2020 ed.) (collecting cases emphasizing the breadth of the “evidence as a whole,” including this Court’s decision in MacDonald). 3 Notably absent from this list is evidence that the prosecution was unable to introduce at trial. Rather, the evidence to be considered is (1) all trial evidence, even if arguably illegally admitted; (2) all evidence the defense alleges to have been wrongly excluded at trial; and (3) all evidence that became available after trial. Even if evidence properly excluded at trial can be considered in this Circuit, cf. Hyman v. Brown, 927 F.3d 639, 659 (2d Cir. 2019), its reliability must be questioned in light of the principles of evidence, and the Court must analyze only what a properly instructed, reasonable jury would make of it. Thus, my dissenting colleague’s insinuative references to an unsolved rape case in Washington, D.C., are unavailing. See Dissenting Op. at 97. While the prosecution was aware of that case, they did not bring it up at trial. And for good reason: neither the victim nor a witness identified Mr. Long from a photo lineup, and no charges were ever filed. Thus, that case is entitled to no weight in our analysis here. 57 B. As discussed by the majority, the trial evidence against Mr. Long came down to: (1) Ms. Bost’s eyewitness identification; (2) the alleged corroboration of that identification by the physical evidence, including the footprint, the “matches of [a] similar nature,” and the discovery of the toboggan and gloves in Mr. Long’s car; and (3) the honesty of the police. J.A. 378; see Majority Op. at 17. Lacking countervailing affirmative evidence, the defense team was left to mount challenges to each of these points by: (1) pointing to the weaknesses of eyewitness identification, especially cross-racial identification by a white woman who admitted to having little interaction with African Americans; (2) noting the discrepancies between Ms. Bost’s initial description of her attacker and Mr. Long; (3) asking the jury to look at the leather jacket for evidence (or lack thereof) of paint transfer; (4) pointing the jury to a light hair in the toboggan (and noting that Mr. Long’s hair was black), but without any supporting forensic tests regarding the origin of the hair; and (5) implying that the police planted the toboggan. As affirmative evidence, the defense could present only Mr. Long’s alibi. With the weight of evidence slanting firmly toward the prosecution, it is no surprise the jury returned a verdict of guilty. But we now know better. The mountain of evidence once concealed by the State has slowly been revealed. This new evidence directly contradicts the State’s second and third arguments: significant evidence did not implicate Mr. Long, and the police did not act honestly. Moreover, today we know much more about the pitfalls of eyewitness identification, especially across racial lines, and especially where the identification procedures lacked protections against influence (conscious or otherwise) by investigating 58 officers. This new evidence, in turn, necessarily casts the trial evidence in a new light. Faced with these realities, no reasonable juror could conclude that the State met its burden to prove beyond a reasonable doubt that Mr. Long committed this crime. C. 1. A Black man broke into a white woman’s North Carolina home. Putting a knife to her throat, he raped her. Though terrified, she tried to remain clear-headed: she memorized his face. She told herself that if she survived, she would make sure he would pay for his terrible crime. She believed she would never forget his face. The same day, she reported the crime and described her attacker to the police. Shortly thereafter, she identified him in person. Physical evidence also appeared to tie him to the crime. He was arrested, tried, and sentenced to life in prison by an all-white jury. She was elated, knowing the person who harmed her could never hurt anyone else. While the foregoing could easily be a summary of the attack on Ms. Bost and her identification of Mr. Long, it isn’t. Instead it describes the identification of Ronald Cotton by Jennifer Thompson in another unfortunate case with similar facts. 4 She picked him from 4 See Jennifer Thompson, I Was Certain, But I Was Wrong, N.Y. Times (June 18, 2000), https://www.nytimes.com/2000/06/18/opinion/i-was-certain-but-i-waswrong.html; Jennifer Thompson-Cannino, Ronald Cotton & Erin Torneo, Picking Cotton: Our Memoir of Injustice and Redemption 124 (2009); see also Innocence Project Amicus Br. at 9–10. For another similar case, see Nat’l Registry of Exonerations, Perry Mitchell, U. Mich. L. Sch. (June 2012), https://www.law.umich.edu/special/exoneration/Pages/casedetail.aspx?caseid=3479 (discussing the case of Perry Mitchell, a Black man convicted of raping a seventeen-yearold white girl at knife-point after she confidently identified him from the third photo array she was shown, who was later exonerated by DNA evidence). 59 a photo lineup, and then again in a physical lineup. She was absolutely positive. And after another man, Bobby Poole, bragged about committing the crime, she testified that she had never seen Poole before. Moreover, Ms. Thompson’s confidence was bolstered by the physical and other purported evidence against Mr. Cotton. “There was [a] piece of rubber [the police] found at [Ms. Thompson’s] apartment, which was found to be consistent with the shoes [the police] took from Ronald Cotton’s apartment[.]” Jennifer Thompson-Cannino, Ronald Cotton & Erin Torneo, Picking Cotton: Our Memoir of Injustice and Redemption 67 (2009) [hereinafter Picking Cotton]. Mr. Cotton’s boss tipped off the police after seeing Ms. Thompson’s composite sketch of her attacker, which Mr. Cotton’s boss thought resembled him. Id. at 44. A witness had seen Mr. Cotton riding his bicycle in Ms. Thompson’s neighborhood early in the morning, wearing a shirt and gloves matching Ms. Thompson’s description. Id. at 18, 44, 67. He had previous convictions for breaking and entering— including having pleaded guilty at age sixteen to breaking and entering with intent to commit rape against a teenage girl. Id. at 44, 82–84. He gave conflicting alibi accounts, and the alibi he ultimately provided was merely testimony from his family saying he was at home asleep on the couch. Id. at 66, 86. He had a utility knife on him when he came to the police station, and he owned a flashlight matching one stolen from a second rape victim’s home. Id. at 77, 82. And a cellmate told police he had heard Mr. Cotton talking in his sleep about the rape. Id. at 88–89. Ms. Thompson was sure this could not all be coincidence. Id. at 67–68. 60 But in 1995, after Mr. Cotton had served eleven years in prison, DNA evidence exonerated him—and implicated Mr. Poole. Ms. Thompson had been certain that Mr. Cotton was her attacker, and that she had never seen Mr. Poole. She had it exactly backwards. 5 Mr. Cotton’s case emphasizes three details that reappear in this case. First, that Ms. Bost sincerely believed her identification to be correct. Second, that the brutal, terrifying crime occurred exactly as she said it did, with the exception of one important detail: the identity of her attacker. And third, that it is not farfetched that someone who endures a violent crime, does everything in her power to memorize her attacker’s face, immediately reports the attack, and then identifies the culprit, does so erroneously. In short, it has happened before, and it can happen again. To say so suggests no ill will on the part of the victim; nor does it cast any doubt on the fact of the attack itself. Rather, it simply points to a deficit in human cognition: difficulty forming memories of faces we do not know, especially in a traumatic situation, and especially across racial lines. 5 Years later, Ms. Thompson and Mr. Cotton wrote a book together which described these events, explained the failings of eyewitness identifications, and emphasized their eventual friendship, founded on grace and forgiveness. See generally Picking Cotton. Their book makes apparent how lucky Mr. Cotton was to have ever been exonerated. He happened to come across Mr. Poole early on in his time in prison—after Mr. Poole was convicted of additional attacks on women—and thought Mr. Poole resembled Ms. Thompson’s composite sketch. Id. at 99–100. He persisted in seeking help from counsel, and ultimately was blessed with pro bono counsel who pursued the state version of habeas relief. Id. at 163–66. Ultimately, they were able to obtain DNA testing—giving Mr. Cotton another stroke of luck that the samples still existed and were in good shape. Id. at 212. Mr. Poole only confessed when confronted with the DNA evidence. Id. at 213. And it was in the nick of time—Mr. Poole died of cancer a few years after Mr. Cotton’s exoneration. Id. at 252, 265. Mr. Cotton’s narrow path to exoneration raises the question of how many other innocent people have not been so lucky and remain behind bars. 61 Recently, the Third Circuit confronted this issue by convening a task force consisting of “a diverse group of judges, lawyers, professors, and law enforcement agents . . . to study the issue of eyewitness identification.” Third Circuit Task Force, 2019 Report of the United States Court of Appeals for the Third Circuit Task Force on Eyewitness Identifications, 92 Temp. L. Rev. 1, 6 (2019) [hereinafter Task Force Report]. The Task Force noted that “[e]yewitness misidentification is the single greatest source of wrongful convictions in the United States.” Id. at 10 (internal quotation marks omitted). And it cataloged modern understandings of the psychology surrounding eyewitness identification, noting that over the past several decades, “a burgeoning body of research has shed light on how human perception and memory function. We now understand that the brain does not work like a video camera.” Id. at 12; see also Dennis v. Sec’y, Pa. Dep’t of Corr., 834 F.3d 263, 314 (3d Cir. 2016) (en banc) (McKee, C.J., concurring) (“In the last thirty years, over 2,000 studies have examined human memory and cognition and their relationship to the reliability of eyewitness identifications. This impressive body of scholarship and research has revealed that eyewitness accounts can be entirely untrustworthy. As the International Association of Chiefs of Police has concluded, ‘[o]f all investigative procedures employed by police in criminal cases, probably none is less reliable than the eyewitness identification.’” (alteration in original) (footnote omitted)). The Task Force Report noted that Ms. Thompson’s identification of Mr. Cotton “was the result of a suggestive eyewitness process that unintentionally resulted in a wrongful conviction.” Task Force Report at 10. Thus, even though “[l]aw enforcement employed identification procedures considered standard, including a composite sketch, a 62 photo array with a subsequent lineup, and finally an in-court identification,” those guardrails were not enough to prevent the police from “suggestively reinforcing the accuracy of her identification of Cotton”—even if they did not intend that result. Id. at 9– 10. Mr. Cotton’s case was not an anomaly: erroneous eyewitness identification played a role in ninety-three percent of wrongful rape convictions later overturned by DNA evidence. See Innocence Project Amicus Br. at 10. To prevent the recurrence of tragedies like Mr. Cotton’s, the Task Force Report provides several suggested best practices, including: (1) double-blind administration of eyewitness identification procedures; 6 (2) selecting fillers (i.e., lineup members who are not suspected of committing the crime) using “all the features included in the witness’s description of the perpetrator,” since “[i]t is obviously crucial that the suspect not stand out”; (3) avoiding the use of show-ups, 7 which are “inherently suggestive,” but if necessary to the investigation, conducting any show-ups “no later than two hours after the eyewitness observed the perpetrator”; and (4) giving a witness “only one opportunity to make an identification of the same suspect.” Task Force Report at 13–14, 16, 42–43 (emphasis added). 6 “Double-blind administration . . . occurs when the officer administering the procedure does not know which person in the display is the suspect.” Task Force Report at 13. This is crucial “so as to avoid ‘conscious or unconscious verbal or behavioral cues that could influence the eyewitness’[s] identification.’” Innocence Project Amicus Br. at 25 (quoting Identifying the Culprit: Assessing Eyewitness Identification, Nat’l Res. Council Nat’l Acads. 91–92 (2014)). 7 “A show-up or field identification is a procedure wherein only one person—the suspect—is presented to a witness for the purpose of identification. Show-ups typically occur live and relatively soon after the crime has been reported in the same vicinity.” Task Force Report at 41 (footnote omitted). 63 These best practices are so well-established today that the State of North Carolina itself has recognized their importance by requiring, by statute, that officers employ nearly all of them. North Carolina’s Eyewitness Identification Reform Act requires that officers conducting live lineups or photo arrays use double-blind identification procedures or approved alternatives “that achieve neutral administration.” N.C. Gen. Stat. § 15A284.52(c)(3); see id. § 15A-284.52(a)(3), (b)(1). Moreover, officers must use fillers who “generally resemble the eyewitness’s description of the perpetrator, while ensuring that the suspect does not unduly stand out from the fillers.” Id. § 15A-284.52(b)(5). And show-ups are only permissible “when a suspect matching the description of the perpetrator is located in close proximity in time and place to the crime, or there is reasonable belief that the perpetrator has changed his or her appearance in close time to the crime, and only if there are circumstances that require the immediate display of a suspect to an eyewitness.” Id. § 15A-284.52(c1)(1). No such best practices were observed here. If, as noted by the Task Force Report, the procedures used in Ms. Thompson’s identification of Mr. Cotton were subpar and almost certainly contributed to her errant identification, those employed in Ms. Bost’s identification of Mr. Long were absolutely dismal. The police settled on Mr. Long as a possible suspect by May 5, 1976. Yet, despite having a photograph of him, they did not perform a photo lineup. Instead, on May 5, Detectives David Taylor and Van Isenhour visited Ms. Bost at her home and asked her to accompany officers to court on May 10, as Mr. Long was to present that day on an unrelated minor charge. Detectives Taylor and Isenhour suggested to Ms. Bost that her attacker might be present in the courtroom and 64 instructed her to wear a disguise “so that if the suspect was there he would[ ]not recognize her.” J.A. 1432. Ms. Bost testified that she thought this procedure “was strange” and that she “was so afraid,” “was nervous,” and was “scared to death.” J.A. 149, 153. While the police did not tell her that the suspect would definitely be in the courtroom, common sense suggests she would have assumed there was a reason the police were putting her through this ordeal. 8 Yet she also testified that there were no other men who “even closely resembled” Mr. Long in the courtroom on May 10. J.A. 172. Thus, instead of using a photo lineup, the police opted to put the victim through an extremely stressful, in-court identification procedure that was not double-blind, 9 lacked even a single appropriate filler, was inherently suggestive, and was followed immediately by a second identification opportunity—a photo lineup in which the fillers again did not match Ms. Bost’s description of the suspect. 10 In other words, the police did not undertake the procedures “considered standard” at the time of Ms. Thompson’s identification of Mr. Cotton less than a decade later (“a composite sketch, a photo array with a subsequent lineup, and finally an in-court identification”). Task Force Report at 9. And they certainly 8 Ms. Thompson has explained that, while the police told her that “[t]he suspect may or may not be included in” the photo array, she “assumed they must have had a suspect. Why would they want me to drive all this way if they didn’t? All I had to do was pick him out. And if I failed to do that, would he go free? Would he find me?” Picking Cotton at 32– 33. 9 Ms. Bost and Detective Taylor could see each other from their positions in the courtroom. 10 According to Ms. Bost, one of the fillers “looked like it might have been a woman,” and only Mr. Long’s photograph depicted a person wearing a leather jacket. J.A. 160. Additionally, the fillers sported a variety of types of facial hair (or had no facial hair at all). 65 did not abide by the best practices we know today—and which North Carolina now requires its officers to use. 11 Intentionally or not, there can be no doubt that the officers’ suggestive tactics influenced the identification, such as by inflating Ms. Bost’s confidence in her identification of Mr. Long. Id. at 53; see also Picking Cotton at 271 (Ms. Thompson noting that after the photo and physical lineups, “by the time I went into court, everything added up for me: I was defiantly confident that Ronald Cotton was the one”). The fact that eyewitness identification is deeply flawed is something we all, lay people and experts alike, understand much better today than in the 1970s. Indeed, the concept was novel to the public even in the 1990s, but has received significant attention since then, through news reports and the efforts of organizations like the Innocence Project (founded in 1992). In their book, for example, Mr. Cotton and Ms. Thompson describe participating in an interview with Larry King in which the prosecutor and defense counsel from their case “talked a lot about DNA, how DNA was even better than eyewitness identification because people can make mistakes.” Picking Cotton at 224. Of course, everyone today understands that DNA is significantly more trustworthy than eyewitness identification; it goes without saying. The pair also participated in a PBS Frontline story in 1996 “about how eyewitnesses can make mistakes.” Id. at 236. What was newsworthy in the 1990s is now common sense. 11 Incredibly, the State continues to argue that this eyewitness identification procedure was appropriate and even ahead of its time. Oral Arg. at 41:50–42:10, available at https://www.ca4.uscourts.gov/OAarchive/mp3/18-6980-20200507.mp3; see Response Br. at 16 (“[T]his was a very strong eye-witness identification.”). 66 In sum, “[t]he voluminous studies conducted on the subjects of memory and eyewitness identifications make it painfully clear that many of the identification procedures used in this case were inconsistent with the fundamental concept of neutral inquiry. As a result, [Ms. Bost’s] identification lack[s] many of the basic indicia of reliability. Yet, the jury that convicted [Mr. Long] was completely unaware of these problems.” Dennis, 834 F.3d at 315–16 (McKee, C.J., concurring). A reasonable juror today would know more about these matters and, therefore, necessarily question the strength of the identification. 2. Putting aside the specific faults of the police identification procedures in this case, today we know more about general shortcomings in eyewitness identification— shortcomings that cast doubt on the prosecution’s case. For example, although Ms. Bost confidently identified Mr. Long at trial, we now know that “[t]here is no scientific basis for correlating time-of-trial confidence with accuracy.” Task Force Report at 57 (citing studies from the past three years); see also United States v. Greene, 704 F.3d 298, 309 n.4 (4th Cir. 2013) (citing “considerable research showing that an eyewitness’s confidence and accuracy have little correlation”). Armed with modern science, we can see several factors at play in Ms. Bost’s identification of Mr. Long that tend to cripple eyewitness identifications. First, we know that stress, especially that induced by weapons, can inhibit accurate memory formation and retention, rendering identifications less reliable. See Task Force Report at 77–81. For example, a 2008 study “found that correct identification rates dropped to 18 percent [in high-stress scenarios] compared with 75 percent for identifications 67 following low-anxiety scenarios.” Innocence Project Amicus Br. at 18. Similarly, “[t]he visible presence of a weapon during a witnessed event”—such as the knife brandished in this case (and in Ms. Thompson’s)—“can influence eyewitness accuracy by directing attention towards the weapon and away from the perpetrator.” Task Force Report at 19. This Court has previously recognized that an identification may be unreliable where a victim’s “degree of attention to the [perpetrator] at the time of the offense was greatly diminished due to her reasonable fear and the distraction of having a weapon pointed at her.” Greene, 704 F.3d at 308. Moreover, today, the difficulties with cross-racial identification are clear. For instance, a 2001 meta-analysis demonstrated that study participants were more than fifty percent more likely “to falsely identify a novel other-race face” than an own-race face. Innocence Project Amicus Br. at 20 (quoting Christian A. Meissner & John C. Brigham, Thirty Years of Investigating the Own-Race Bias in Memory for Faces: A Meta-Analytic Review, 7 Psychol. Pub. Pol’y & L. 3, 15 (2001)). The impact appears to be generally greater where, as here, the eyewitness has not “had regular interactions and involvement with persons of the same ethnic group as the perpetrator.” Task Force Report at 19. Next, Ms. Bost’s identification of Mr. Long did not occur until more than two weeks after the attack. Delays in time can, unsurprisingly, frustrate a witness’s ability to identify a suspect. E.g., Kenneth A. Deffenbacher et al., Forgetting the Once-Seen Face: Estimating the Strength of an Eyewitness’s Memory Representation, 14 J. Experimental Psychol.: Applied 139, 147 (2008) (“[C]onservative advice to a trier of fact would be that the typical eyewitness viewing a perpetrator’s face that was not highly distinctive would 68 be expected to have no more than a 50% chance of being correct in his or her lineup identification (six-person lineup) at a 1-week delay.” (emphasis added)). Additionally, Ms. Bost testified that her attacker wore a toboggan that was pulled “down over his ears, and down his neck.” J.A. 201. Research has found “that wearing a disguise as simple as a hat reduce[s] identification accuracy.” Task Force Report at 89; see also Young v. Conway, 698 F.3d 69, 80 (2d Cir. 2012). Finally, today we understand that witnesses may incorporate details about an individual they have identified into their memory of the suspect. E.g., Nancy K. Steblay, Scientific Advances in Eyewitness Identification Evidence, 41 Wm. Mitchell L. Rev. 1090, 1108 (2015) (“[E]yewitness veracity is dually cursed by the likelihood that original memory of the crime event will be tainted by new external information and that the witness will be unable to effectively parse information into what she knows now, versus what she knew at the time of the crime . . . . An eyewitness may replace (or confuse) a perpetrator’s face with another image—of an innocent lineup member, a police composite, or a face seen in a mug-shot or other post-event context.” (footnotes omitted)). We see this phenomenon at work here. On the night of the attack, the police recorded Ms. Bost’s description of her rapist. The description does not mention facial hair and describes both the leather jacket and the toboggan merely as “dark,” without specifying color. J.A. 1428. Yet she later testified that her attacker did have facial hair—as did Mr. Long at the time—and that the jacket and toboggan introduced into evidence (which were black and green, respectively) were those worn by her attacker. She positively identified the clothing even though she could not even 69 describe the color of the items on the night of the attack—let alone be able to testify, months later, to further details, such as the number of pockets on the jacket. To summarize, even before considering the suppressed evidence in this case, a modern jury would have a much improved understanding of cognition and memory—and the lessons learned from other cases, like the 1995 exoneration of Mr. Cotton, which occurred many years after Mr. Long was sentenced—than the original jury did. They would also have much more guidance on the flaws in the identification procedures used and why those defects mattered. No reasonable juror could have found, on the basis of Ms. Bost’s identification alone, that Mr. Long was guilty beyond a reasonable doubt. Yet after considering the suppressed evidence—which overwhelmingly controverts the other trial evidence—the entirety of the State’s remaining case dangles on that single thread. D. Turning now to the rest of the State’s case, first, the evidence that the prosecution suppressed would have seriously undermined the physical evidence tying Mr. Long to the crime—and pointed to a different assailant. The police deliberately hid test after test that did not link Mr. Long to the crime scene. See Majority Op. at 22–23. This forensic evidence would necessarily give a jury pause—and it casts the remaining evidence in a far different light. See Professors Amicus Br. at 8 (“[J]urors use a coherence-based reasoning method, in which they integrate the whole of the evidence that they receive. . . . [A] piece of strong inculpatory evidence can make the entire evidence set appear inculpating. By the same token, including an exculpating item can push the evidence towards a conclusion of innocence.” (citations and internal quotation marks omitted)). 70 At trial, the only concrete evidence for the defense was Mr. Long’s alibi testimony. In contrast, the prosecution had Ms. Bost’s identification, the items of clothing that seemed to match those worn by her attacker, the shoeprint that could have been made by Mr. Long’s shoes, and the matches “of [a] similar nature.” J.A. 378. The prosecution could thus argue that Ms. Bost’s “testimony is not only accurate, but totally consistent with every piece of physical evidence existent.” J.A. 536. Given the balance of evidence, the jury understandably interpreted the forensic evidence as supporting Mr. Long’s guilt—even though each piece of forensic evidence was, at best, ambiguous. But the suppressed evidence entirely shifts this balance—meaning that a reasonable jury, in search of proof of guilt beyond a reasonable doubt, would recalibrate its analysis of the ambiguous trial evidence supposedly incriminating Mr. Long. We now know that lab tests could not conclude that the matches found in Mr. Long’s car were any more “similar” to the matches found at the crime scene than any other matches in the world. A hair located at the scene, which forensic testing revealed could have come from a Black individual—and therefore, given Ms. Bost’s admitted lack of general social interaction with African Americans, may well have come from her attacker—did not match Mr. Long. In fact, none of his hair was located at the scene or on her clothing. 12 And the forensic report on the hair that was retrieved from the scene noted that it was “reddish” in color—which aligns with Ms. Bost’s description of her attacker as a light-skinned Black man. J.A. 1471. By contrast, Mr. Long is a dark-skinned Black man with black hair that 12 In fairness, no suspect hair was located on her clothing. 71 took on “a brownish gray color” under the microscope. Further, no paint or fibers from the scene were discovered on Mr. Long’s jacket or gloves, or on the toboggan supposedly found in his car. No fingerprints from Mr. Long were found at Ms. Bost’s house. 13 And while Mr. Long’s shoes “could have made” a print found at the scene, the analyst could not “effect any identification.” J.A. 1464 (emphasis omitted). While the absence of evidence is not evidence of absence, Dissenting Op. at 116, in our criminal justice system, evidence is required to convict. That simply means that in the absence of evidence, the State has not met its burden to prove guilt beyond a reasonable doubt. E. 1. Next, there is the matter of the police officers’ purported “perfect honesty,” as the State put it during closing arguments. J.A. 576. Our criminal justice system indeed rests on the honesty of those who investigate crimes: absent their forthrightness, the system becomes one of corruption, cronyism, and conviction without evidence. In short, their honesty is necessary to our constitutional order. I do not take lightly, therefore, the conclusion that officers or other investigators, such as lab analysts, discharged their duties dishonestly. But nor can I ignore such a 13 While Ms. Bost testified at trial that her attacker was wearing gloves, the police reports of her description on the night of the attack were less clear. The first report mentions gloves, but the more complete description recorded at the hospital shortly thereafter states only that the suspect “could possibly [have] been wearing gloves.” J.A. 1428 (emphasis added). 72 conclusion when the evidence is clear; in fact, all who seek to uphold the rule of law must acknowledge when it has been flouted. Officers have a duty to investigate truthfully; as noted, our system of justice depends upon it. Unfortunately, we well know that some officers sometimes lie. E.g., United States v. Slager, 912 F.3d 224, 234 (4th Cir.) (noting that several of the defendant police officer’s statements concerning his killing of Walter Scott were directly contradicted by a witness’s video evidence—including the particularly significant lie that Scott grabbed the officer’s taser, supposedly justifying the use of deadly force), cert. denied, 139 S. Ct. 2679 (2019). Wishing otherwise does not make it so. The majority opinion has already recounted several ways in which the officers responsible for investigating this case acted dishonestly, but I will summarize those points again here. At trial, Detective Isenhour—a key witness for the prosecution, who was himself later convicted for a crime of dishonesty, that is, possession of a stolen U.S. Treasury check—perjured himself on the stand. That is, he lied. When asked during trial what items he brought with him to the lab at the State Bureau of Investigation (“SBI”) for forensic testing a mere four months earlier, Detective Isenhour purported to list them: a “pair of shoes” from Mr. Long, “two inked [shoe] impressions,” “and . . . the latent lift” of a shoeprint from the banister. J.A. 411 (emphasis added). There was no hedging, and his meaning is unmistakable: he had taken those items, and only those items, to the SBI. That was false. In fact, he had brought more than a dozen items to the lab—but the test reports 73 on the others had failed to provide results even remotely helpful to the prosecution’s case. So they were, conveniently, “forgotten.” 14 And not just forgotten on the stand at trial: the reports, which the SBI recorded were sent to Detective Isenhour, vanished from the police’s files. They were uncovered, decades later, only in the SBI’s records. The clear inference is that someone removed the reports from the police’s case file. Detective Isenhour also lied at trial about whether the items had remained in his possession and control since he first received them. He testified that they had so remained. Again, that was not true. In fact, he had left the items he discussed at trial—as well as those he decided not to mention—with the SBI for several days. Perhaps he feared that inquiries into the time the items spent at the lab would lead to discovery of the test reports whose existence he wanted to conceal. Moreover, Detective Isenhour covered his tracks in writing. On May 12, 1976, he prepared a summary report listing all the items of evidence he had brought to the SBI the previous day. He later prepared a second, undated report, which claimed to be “intended to show what type [of] work was done and on which dates evidence was collected and processed.” J.A. 1484. The second report describes receiving “oral and written results” that “the shoes worn by [Mr.] Long could have made the latent impression recovered” from the scene. J.A. 1484. 15 But the second report does not mention any other forensic test results. 14 The State concedes that Detective Isenhour’s testimony was “at least incorrect.” Oral Arg. at 46:30–46:42. 15 That suggests the report was created on May 19 or later, as that was the date of the written shoeprint report. 74 Additionally, the report states only that Mr. Long’s clothing had been “collected . . . and held for investigative uses”; there is no mention that the clothing had been brought to the lab. Id. Neither report was disclosed to the defense. The rational inference is that Detective Isenhour created the second report to support his trial testimony, whether or not he ultimately used it that way. Detective Isenhour was not the only officer who acted deceitfully. Detective Taylor testified, outside the presence of the jury, that “[t]o [his] knowledge, [the matches found in Mr. Long’s vehicle] were not matched [to those found at the crime scene].” J.A. 335. When pressed by the presiding judge, who asked whether he “ha[d] any information that they didn’t match,” Detective Taylor stated, “No, sir, they didn’t match.” Id. Such an affirmative statement suggests knowledge of the underlying SBI report analyzing this issue. Yet Detective Taylor did not disclose the existence of that report. And before the jury, he stated only that the two sets of matches were “of [a] similar nature.” J.A. 378. He failed to mention that a lab report did not support his testimony. 16 16 The majority notes the discrepancy between Detective Taylor’s testimony before the jury and his testimony outside the jury’s presence. See Majority Op. at 12–13. But this point bears repeating because the State argued to the MAR Court that Detective Taylor’s testimony given outside the presence of the jury was in fact heard by the jury—and argued that this meant “the jury was left” with an impression even more favorable to Mr. Long than the actual matchbook report. J.A. 1004. The MAR Court perpetuated this error, relying on this supposed fact twice in its opinion. See J.A. 1352 (stating that at trial, defense counsel “elicited testimony that the [burned] matches did not match [the matchbook found in Mr. Long’s car, which] . . . . was inaccurate, but to the benefit of [Mr. Long]”); J.A. 1358 (“[T]he jury was misled to believe that the matches did not match which was to [Mr. Long]’s benefit.”). 75 Disclosure of the reports would have undermined the State’s case regarding physical evidence; the State could no longer argue that Ms. Bost’s identification was “totally consistent with every piece of physical evidence existent.” J.A. 536. But it also would have damaged the credibility, and allowed for the impeachment, of Detectives Isenhour and Taylor—the two officers who approached Ms. Bost with their unorthodox (and highly suspect) plan for the in-court identification of Mr. Long. It would have raised the question of why, if this evidence was not meaningful, the officers chose to hide it. E.g., Carrillo v. County of Los Angeles, 798 F.3d 1210, 1227 (9th Cir. 2015) (noting that an officer’s suppression of exculpatory evidence, combined with that officer’s affirmative statements contradicting the suppressed evidence, “implicitly recognizes that [the suppressed] evidence . . . would have been very helpful to the defense”). And it would have cast the case in an entirely different light, where implications about planted evidence—such as the toboggan supposedly found in Mr. Long’s car—no longer seemed so farfetched. 17 Cf. Killian v. Poole, 282 F.3d 1204, 1209 (9th Cir. 2002) (“[T]he finders of fact were deprived of the fundamental inference that if [a witness] lied about X, Y and Z, it is quite likely that he lied about Q, R and S.”). I would be remiss if I did not also mention the missing semen sample—which police collected from the hospital, with the victim’s consent, a few hours after the rape. The semen 17 I note that Mr. Long’s claim of actual innocence does not rest on the inference that the police planted the toboggan, however fair an inference that may be. The police claimed to locate the toboggan under the seat of Mr. Long’s car. But the car was a family car in which friends also sometimes rode, including in the days before his arrest. It is not hard to imagine that someone else—days, weeks, even years earlier—could have forgotten their hat under the seat of the car. 76 sample was released along with the victim’s pubic hair combings. The former simply vanished into the ether, while the latter were tested (though those test results, too, disappeared for several decades). Yet Mr. Long’s trial counsel was “reasonably certain” he had asked the prosecution whether a rape kit was taken and was “not told that any such thing existed.” J.A. 1036. And it appears that the prosecutors believed that to be true. In fact, even the police no longer have a record of the semen sample; the only reason Mr. Long knows about it is because the hospital still had a record of its release to the police. Again, key evidence was somehow lost from the police file. The semen sample held significant exculpatory power. 18 But moreover, the officer who collected the sample from the hospital, Sergeant Marshall Lee, testified at trial on an unrelated matter. Yet defense counsel, knowing nothing of the semen sample’s existence, had no reason to cross-examine Sergeant Lee. 19 Finally, the officers’ motives must be questioned when one considers their chosen method of identification. Despite having a photograph of Mr. Long, they subjected Ms. Bost to a terrifying in-court ordeal, complete with a disguise in case her attacker was 18 Although DNA testing was not available in 1976, there were other forensic tests that could be performed on a semen sample. See Hilliard v. Spalding, 719 F.2d 1443, 1445 (9th Cir. 1983) (discussing the technology available in 1975, such as “test[ing] a sample of seminal fluid taken from the victim and compar[ing] it with samples of a defendant’s saliva and blood,” which in some circumstances could “conclusively exculpate an individual”). 19 The district court found that the examining doctor’s testimony that he had confirmed the presence of sperm provided Mr. Long with the opportunity “to argue that the state failed to make any meaningful use of the biological evidence, such as connecting it to [him].” J.A. 1691. But the court failed to note that Mr. Long did not know that the police had collected a test tube containing a sperm sample and then (at best) lost it—a fact tending to show either police incompetence or dishonesty, both of which are distinct points from the one the court found Mr. Long could have elicited based on the trial testimony. 77 present and recognized her. And, although they had purportedly identified a key suspect who they believed had committed a heinous, brutal crime, they chose to wait at least five days before having the victim identify him. Five days during which the alleged rapist could have attacked more women, fled the state, or returned to harm Ms. Bost. The officers in this case plainly did not act with “perfect honesty.” In fact, some of them acted with deliberate deceit—at the time of the investigation, on the stand at trial, and in the ensuing decades, when they never revealed the existence of the suppressed evidence. The impact on Mr. Long’s case was profound. The impact on his life was disastrous. 2. Nevertheless, the State has continued to hinder Mr. Long’s attempts to obtain the evidence that should have been disclosed more than forty years ago. In June 2005, an officer testified that the only evidence he had found related to the investigation was the master case file, and the SBI reported it had no additional records. But after the court permitted Mr. Long’s counsel to review the master case file—at which point counsel discovered Detective Isenhour’s May 12 report—the SBI located the missing test reports. The SBI finally provided those reports in January 2006—thirty years after Mr. Long’s arrest, trial, and conviction. And then, in 2015, yet more evidence was uncovered—three envelopes containing 43 latent fingerprints from the crime scene, which did not match Mr. Long’s. The fact that the physical prints still existed was not disclosed after the court order in 2005, despite the State’s representations that there remained no additional evidence to reveal. 78 Indeed, Mr. Long only learned of the prints due to an investigation by the North Carolina Innocence Inquiry Commission (“Commission”). 20 The Commission apparently ran the prints against “several alternate suspects,” in addition to Mr. Long. J.A. 1584. But when Mr. Long sought more information from the Commission about those other suspects, the State again stood in the way, refusing to consent to disclosure of the Commission’s files. 3. The exculpatory evidence that has come to light thus far is voluminous and weighty. But there also exists a very real possibility of other evidentiary problems of which we may never learn. Unconstitutional behavior in investigations is not limited to investigating officers. In recent years, the practices of North Carolina’s forensic lab, the SBI, have come under scrutiny—particularly after the high-profile 2010 exoneration of Gregory Taylor, whose conviction was based in part on an errant SBI report. 21 After Mr. Taylor’s exoneration, the state Attorney General commissioned an independent review of the policies and procedures of certain subsections of the SBI. Chris Swecker & Michael Wolf, An Independent Review of the SBI Forensic Laboratory, N.C. 20 When no DNA evidence was located and the fingerprints provided no match, the Commission closed its investigation without coming to a decision as to guilt or innocence. 21 See Nat’l Registry of Exonerations, Gregory Taylor, U. Mich. L. Sch. (Oct. 16, 2014), https://www.law.umich.edu/special/exoneration/Pages/casedetail.aspx?caseid=3677. Specifically, an SBI report indicated that there was a “chemical indication for the presence of blood” on Mr. Taylor’s truck. Id. Absent from the final report were the results of a subsequent test showing that, in fact, no blood was present. 79 Dig. Collections 2 (2010). 22 The reviewers located hundreds of North Carolina cases “contain[ing] lab reports that mention positive presumptive test results but omit the results [of] other more sensitive tests.” Id. at 3. More troubling still, “[i]t was determined that during the relevant time periods”—1987 through 2003—“lab files were not routinely produced to an accused defendant.” Id. at 4. In fact, it was “the sanctioned practice of some NC SBI Laboratory Analysts at the time to omit the results of certain negative or inconclusive confirmatory tests in final lab reports under certain circumstances, and this practice later became written SBI policy in 1997.” Id. at 12 (second emphasis added); see also id. at 18–22. Bad faith is not a prerequisite to the production of such abysmal results. In fact, the SBI reviewers noted that they found no evidence “that laboratory files or reports were concealed or evidence deliberately suppressed.” Id. at 28. Rather, they attributed the issues they uncovered to flawed policies, training, and management, and to a mindset “that the lab’s customer was law enforcement and reported results should be tailored primarily for law enforcement’s consumption.” Id. The SBI review involved blood and DNA analyses between 1987 and 2003; neither the type of analysis, nor the time frame, is directly relevant to this case. But the report is worth considering because it demonstrates how even elements further removed from direct suppression of evidence or perjured testimony can greatly impact a case. North Carolina’s SBI has had a lengthy history of, for one reason or another, failing to disclose key 22 Available https://digital.ncdcr.gov/digital/collection/p249901coll22/id/498661/. 80 at information that we now know would have prevented convictions of the innocent. And we may well question whether the SBI’s history means that the results of the lab tests in Mr. Long’s case were designed to be as friendly to the prosecution as they could possibly be, notwithstanding the fact that they did not support the State’s case—or that other tests were performed whose existence remains suppressed, and which may never be uncovered. F. Finally, in light of all of the above, a jury would surely view Mr. Long’s alibi with a much more open mind today. Mr. Long’s alibi evidence showed a young man enjoying a normal weekend evening: attending a class reunion meeting; socializing with friends; then returning home where he went upstairs to his room, talked to his girlfriend and young son on the phone, and listened to music, all the while excitedly waiting for his father to return home with the car so he could borrow it to make the half-hour drive to Charlotte for a party. Both the attendees of the meeting (which occurred before the attack on Ms. Bost) and the partygoers who saw Mr. Long later in the evening (after the attack) reported he was wearing khaki pants—not blue jeans, as Ms. Bost’s attacker wore. And at the party, he wore a black leather hat—not a green toboggan. Moreover, those at the party testified that Mr. Long acted normally and did not display any cuts or scratches. Nor is there anything in the record to suggest that there were scratches on his leather jacket. Is it possible that between talking to his son and checking on whether his father had brought the car home yet, Mr. Long changed his pants, sneaked out of his upstairs bedroom, walked to Ms. Bost’s home, broke in, raped her in such a brutal fashion that her fingernails 81 were bent backward from her attempts to scratch and fight off her attacker (but without getting any scratches on himself or his jacket), returned home, slipped back into his bedroom unseen, changed back into the pants he was wearing earlier, and then went about his normal evening? Many things are possible. Is such a sequence of events probable, or even plausible, such as to support a conviction beyond a reasonable doubt? I do not believe any reasonable juror would find so—particularly in light of the many other items of exculpatory evidence of which we are now aware. G. Considering “all the evidence put before the court” today, MacDonald, 641 F.3d at 610, Mr. Long has established by “clear and convincing evidence” that, but for the Brady violations, no reasonable juror would have found him guilty of this crime beyond a reasonable doubt, § 2244(b)(2)(B)(ii). The various pieces of evidence cast each other in a new light. The new forensic evidence would undermine the only physical evidence presented at trial. The officers’ dishonesty would undermine the integrity of the investigation. Both the new forensic evidence and the evidence of the officers’ dishonesty would strengthen the alibi evidence. And all of that would leave only the eyewitness identification—an identification conducted using highly suggestive procedures, whose validity therefore must be doubted. Mr. Long has satisfied the § 2244(b)(2)(B) gateway, and since he has also succeeded on the merits in his § 2254 petition, he is entitled to habeas relief. 82 II. Before closing, I add my voice and vote in this en banc proceeding to overturn our decision in Evans v. Smith, which held that Brady claims may be subjected to the strictures of “second or successive” petitions. 220 F.3d 306, 309 (4th Cir. 2000). 23 In my view it is an extraordinary fact that Mr. Long is subjected to the § 2244(b)(2) gateway at all. 24 That statute requires a habeas petitioner filing “a second or successive habeas corpus application” to meet the exacting standards of this gateway. 28 U.S.C. § 2244(b)(2). Mr. Long has not squarely challenged the applicability of this rule to his petition in this appeal. 25 On en banc review, however, we may reconsider our circuit precedent, and we ought to do so here. After all, “[t]he [Supreme] Court has declined to interpret ‘second or successive’ as referring to all § 2254 applications filed second or successively in time.” 23 Notably, this Court’s reasoning in Evans has been undermined by the Supreme Court’s more recent holding in Panetti v. Quarterman, 551 U.S. 930 (2007). For example, Evans emphasized that unlike the earlier Supreme Court holding in Stewart v. MartinezVillareal, “Evans’ Brady claim was not part of th[e] first petition.” Evans, 220 F.3d at 325 (citing Stewart v. Martinez-Villareal, 523 U.S. 637 (1998)). Panetti, however, involved a claim that was not raised in the initial petition. 551 U.S. at 942. Whether Evans remains good law in this Circuit after Panetti is an open question. We have cited Evans only once since Panetti, in an unpublished decision released the day after Panetti, which cited Evans for an unrelated point. See United States v. Richardson, 239 F. App’x 792, 793 (4th Cir. 2007) (per curiam). 24 Mr. Long must satisfy some actual-innocence standard regardless of the successiveness of his petition, since it was filed out of time. Majority Op. at 25. As the majority opinion notes, however, the showing of actual innocence is less stringent for a tardy petition than for a successive one. See id. at 50 n.13. Compare 28 U.S.C. § 2244(b)(2)(B)(ii), with McQuiggin, 569 U.S. at 395. I believe Mr. Long satisfies both standards, but I also believe it worth considering whether he need do so. 25 At oral argument, however, Mr. Long’s counsel did note that AEDPA was not intended to prevent cases like Mr. Long’s from going forward. Oral Arg. at 12:50–13:30. 83 Panetti v. Quarterman, 551 U.S. 930, 944 (2007) (collecting cases). Rather, there are “exceptions.” Id. at 947. Panetti elaborated on one such exception (related to mental competency for execution), but left the door open to others. Notably, the Court held that § 2244 should not be interpreted in a manner “that would ‘produce troublesome results,’ ‘create procedural anomalies,’ and ‘close our doors to a class of habeas petitioners seeking review without any clear indication that such was Congress’[s] intent.’” Id. at 946 (quoting Castro v. United States, 540 U.S. 375, 380–81 (2003)). And, noting that AEDPA’s purposes were to promote “comity, finality, and federalism,” the Court held that “[t]hese purposes, and the practical effects of [the Court’s] holdings, should be considered when interpreting AEDPA,” “particularly . . . when petitioners run the risk under the proposed interpretation of forever losing their opportunity for any federal review of their unexhausted claims.” Id. at 945–46 (internal quotation marks omitted); cf. McQuiggin, 569 U.S. at 393 (collecting post-AEDPA decisions that “see[k] to balance the societal interests in finality, comity, and conservation of scarce judicial resources with the individual interest in justice that arises in the extraordinary case” (alteration in original) (emphasis added) (quoting Schlup, 513 U.S. at 324)). Brady claims, as a category, represent a good candidate for exclusion from the “second or successive” requirements. After all, such claims relate to suppression of material, favorable evidence by the state. In other words, to subject Brady claims to the heightened standard of § 2244(b)(2) is to reward investigators or prosecutors who engage in the unconstitutional suppression of evidence with a “win”—that is, the continued 84 incarceration of a person whose trial was fundamentally unfair (and unconstitutional). For that reason, some courts have hesitated to sanction a rule as broad as the one we adopted in Evans. E.g., Brown v. Muniz, 889 F.3d 661, 668 n.5 (9th Cir. 2018) (“[S]hould exculpatory evidence be discovered by the State after the first habeas petition is filed, and is thereafter suppressed by the State over the course of post-conviction proceedings, [the second-or-successive rules would not apply because] . . . . the new claim would not have been ripe at the time of the initial filing.”), cert. denied sub nom. Brown v. Hatton, 139 S. Ct. 841 (2019); Crawford v. Minnesota, 698 F.3d 1086, 1090 (8th Cir. 2012) (concluding that “at least nonmaterial Brady claims in second habeas petitions” are subject to the second-or-successive rules, but noting that the question of whether “all Brady claims in second habeas petitions are second or successive regardless of their materiality” was “not presented” in the case); Douglas v. Workman, 560 F.3d 1156, 1193 (10th Cir. 2009) (declining to apply the § 2244(b)(2) requirements where a prosecutor acted affirmatively to conceal the facts underlying the petitioner’s Brady claim until after he filed his first habeas petition, since to do so “would be to allow the government to profit from its own egregious conduct,” and “[c]ertainly that could not have been Congress’s intent when it enacted AEDPA”). Moreover, some of our colleagues in other circuits have expressed serious reservations about imposing § 2244(b)(2)(B)’s limitations on Brady claims at all. E.g., Gage v. Chappell, 793 F.3d 1159, 1165 (9th Cir. 2015) (noting that the petitioner’s “argument for exempting his Brady claim from the § 2244(b)(2) requirements has some merit,” though the Court was bound to follow circuit precedent in concluding otherwise). 85 Notably, an Eleventh Circuit panel recently followed its circuit precedent requiring a petitioner raising Brady claims to satisfy the § 2244(b)(2)(B) standard—which, as a threejudge panel, it was required to do—but nevertheless explicitly, and at length, noted the panel’s unanimous disagreement with that precedent and called for en banc review of the matter. Scott v. United States, 890 F.3d 1239, 1243–44 (11th Cir. 2018) (citing Tompkins v. Sec’y, Dep’t of Corr., 557 F.3d 1257 (11th Cir. 2009)). The panel explained that the circuit’s precedent “eliminates the sole fair opportunity for these petitioners to obtain relief” and that, in the panel’s view, “Supreme Court precedent, the nature of the right at stake here (the right to a fundamentally fair trial), and the Suspension Clause . . . do not allow this. Instead, they require the conclusion that a second-in-time collateral claim based on a newly revealed actionable Brady violation is not second-or-successive for purposes of AEDPA.” Id. at 1243. The panel reached this conclusion even absent any of the evidence of bad faith or conscious suppression on the part of law enforcement that exists in this case. See id. at 1246. The Eleventh Circuit’s opinion in Scott warrants careful consideration. After all, Congress’s intention in enacting AEDPA was “to curb the abuse of the statutory writ of habeas corpus.” H.R. Rep. No. 104-518, at 111 (1996) (Conf. Rep.); see also Panetti, 551 U.S. at 945 (noting the legislative purpose of promoting “finality”). In many cases, the rules governing second or successive petitions advance that goal: defendants are discouraged from filing frivolous habeas petitions because, if they do, any later petitions will be subject to strict limitations. 86 But Brady claims almost necessarily will not be included in an initial petition, no matter how meritorious the initial petition. That is because, in most cases, the petitioner will have no way of knowing he has a Brady claim to raise until it is too late; the initial petition, after all, must generally be filed within one year of the conclusion of direct review. § 2244(d)(1). So, unless the petitioner becomes aware of the suppressed evidence shortly after trial, his claim will be subjected to higher standards—through no fault of his own. Consider, for example, a Black defendant tried by an all-white jury who raises a plausible Batson claim in his initial habeas petition. 26 It would be reasonable for him to do so; otherwise, he will have forfeited the claim forever. Yet if, ten or twenty or thirty years later, it comes to light that the state failed to disclose material favorable evidence, the defendant will now have to establish not only the merits of a Brady claim, but also that he can satisfy the stringent rules applicable to second or successive petitions. (This is no mere hypothetical; Evans presented this Court with a Brady challenge subsequent to an initial Batson-based petition. 220 F.3d at 312, 322–23.) 27 Worse still, this situation creates incentives for any state actors withholding material evidence to violate the petitioner’s other constitutional rights, if subtly. After all, if they succeed in prompting the petitioner to file an initial habeas petition, they position 26 Batson claims are brought pursuant to the Supreme Court’s holding that “a State may not discriminate on the basis of race when exercising peremptory challenges against prospective jurors in a criminal trial.” Flowers v. Mississippi, 139 S. Ct. 2228, 2234 (2019) (citing Batson v. Kentucky, 476 U.S. 79 (1986)). 27 Indeed, one wonders whether, in light of such situations, § 2244 as interpreted in Evans does not violate the Equal Protection Clause, to the extent that petitioners of some races are more likely to have cause to raise Batson challenges in initial habeas petitions than petitioners of other races. 87 themselves to better defend against any later habeas claims in the event the suppressed evidence is ever uncovered. Such an incentive structure promotes neither the interests of justice nor finality. Similarly, the current situation disproportionately impacts petitioners in Mr. Long’s precise boat: that is, those who were convicted and raised habeas claims before AEDPA’s enactment in 1996—but who became aware of suppressed material evidence only after 1996. This is because the standards for successive petitions were much less strict prior to 1996. See 2 Randy Hertz & James S. Liebman, Federal Habeas Corpus Practice and Procedure § 28.3[a] (7th ed. 2019). Had the State disclosed the evidence it had withheld from Mr. Long for so long before 1996, he would have been in a much better position to raise his Brady challenge than he was when they actually disclosed it years later. And of course, yet another standard would have applied had the State disclosed the evidence before 1989, when Mr. Long filed his first federal habeas petition. Mr. Long had no control over when the State finally started to disclose the evidence. Surely Congress did not intend such a perverse result. Yet, under the precedent of this and other circuits, this is precisely how § 2244 currently works. Our Court, and others, should reconsider this precedent. And of course, Congress could choose to resolve the issue definitively by explicitly excluding Brady claims from the second-or-successive rules. III. I conclude by noting that it is imperative we not lose sight of the context in which this case was investigated and prosecuted. 88 Mr. Long, a Black man, was tried in “small town” 1970s North Carolina by an allwhite jury for the rape of the white widow of a prominent local business executive. J.A. 703. 28 Concord and the surrounding county “had a reputation of being run by” the company Ms. Bost’s husband had worked for. Id. And four members of the jury were either employees of that company or married to an employee. Moreover, the community, inside and outside the courtroom, was “racially polarized.” J.A. 677. Despite all these drawbacks, Mr. Long’s counsel decided against moving for a change in venue for fear of being transferred to an area with heavy Klan activity—which would have been “getting out of the frying pan into the fire.” J.A. 680. Mr. Long thus faced a steep uphill battle to persuade the jury of his innocence before he ever set foot in the courtroom. And that uphill climb became an impossibly steep— perhaps even vertical—one when one considers the particular crime in this case: an interracial rape. The history of policing sexual contact between Black men and white women— including using accusations of inappropriate or violent conduct, from wolf whistles to rapes, to imprison or lynch Black men—is a long and troubling one in this country, particularly in the South. These dynamics took on particular ferocity in the aftermath of the Civil War. See Randall Kennedy, Interracial Intimacies: Sex, Marriage, Identity, and Adoption 189 (2003). White supremacists argued that “the abolition of slavery and the awarding of civil rights to blacks had led to a sharp increase in sexual crimes against white 28 The population of the City of Concord at the time was around 18,000. 89 women.” Id. Ambassador Claude Bowers wrote that rape was “the foul daughter of Reconstruction.” Id. (quoting Claude G. Bowers, The Tragic Era: The Revolution After Lincoln 308 (1929)). And academics “concocted theories to legitimate the claim that black men were dangerous subhumans predisposed to rape.” Lynching in America: Confronting the Legacy of Racial Terror, Equal Just. Initiative 49 (3d ed. 2017), https://eji.org/wpcontent/uploads/2019/10/lynching-in-america-3d-ed-080219.pdf [hereinafter Lynching Report]; cf. Michelle Alexander, The New Jim Crow: Mass Incarceration in the Age of Colorblindness 21–22 (2010) (arguing that the mass incarceration of Black men today plays a similar role in enforcing racial hierarchy as slavery and Jim Crow laws did for previous generations). Not surprisingly, therefore, “[o]f the several defenses that apologists offered for lynching, one of the most popular was the claim that black men’s lust for white women was irrepressible or so intense that nothing short of mob violence could control it.” Kennedy, supra, at 192; see also Lynching Report at 15 (“Charges of rape, while common, were ‘routinely fabricated’ and often extrapolated from minor violations of the social code, such as ‘paying a compliment’ to a white woman, expressing romantic interest in a white woman, or cohabitating interracially.”). Indeed, in 1951 in North Carolina, a Black man was prosecuted for assault with intent to rape for allegedly looking at a seventeen-year-old white girl inappropriately (from a distance of seventy feet, no less). Kennedy, supra, at 196. And “[m]ore than half of the lynching victims [the Equal Justice Initiative has] documented were killed under accusation of committing murder or rape.” Lynching Report at 32. 90 Even as lynchings became less common, American society continued to frown upon interracial contact. We cannot forget, for example, that at the time of Mr. Long’s trial, interracial marriage—with its implications for fully consensual sexual contact across racial lines—had only been legal across the country for less than a decade. See Loving v. Virginia, 388 U.S. 1 (1967). Nor can we forget that despite that legal victory, the vast majority of white Americans in the 1970s continued to disapprove of such marriages. 29 Officers and jurors who disapproved of interracial marriage were sure to be highly skeptical of a man accused of interracial rape. There is no doubt that Ms. Bost was brutally raped. This was not a case of a fabricated claim or of retaliation for mere flirtation. But that does not mean we can ignore the history and context that no doubt implicitly biased both the officers during the investigation and the jury when faced with the decision of whether to convict Mr. Long, specifically, of the crime. Cf. Picking Cotton at 76, 84–85 (Mr. Cotton discussing how the cards were stacked against him as a Black man accused of raping a white woman in smalltown North Carolina in the 1980s); id. at 121 (“If it was so important to the community 29 According to Gallup polls, in 1958, only 4% of respondents approved of Blackwhite marriages, while 94% disapproved. Joseph Carroll, Most Americans Approve of Interracial Marriages, Gallup (Aug. 16, 2007), https://news.gallup.com/poll/28417/mostamericans-approve-interracial-marriages.aspx. In 1972 and 1978, the disapproval figures had fallen somewhat, but remained steep: 60% and 54%, respectively. Id. White respondents in the 1970s were particularly disapproving, with the percentage approving ranging between a mere 25% and 33%. Id. Such numbers may be hard for younger Americans to fathom today—in 2013, the percentage approving had risen to 87%, including 84% of white respondents—but they cannot be forgotten when contemplating the historical context in which this case was investigated and tried. Frank Newport, In U.S., 87% Approve of Black-White Marriage, vs. 4% in 1958, Gallup (July 25, 2013), https://news.gallup.com/poll/163697/approve-marriage-blacks-whites.aspx. 91 that a black man be punished for these crimes, I guess any one would do, so they pinned it on me.”). Nor can we ignore the devastating fact that the officers suppressed evidence they knew tended to show Mr. Long’s innocence—even though Mr. Long was charged with first-degree rape at a time when that crime carried a mandatory death sentence in North Carolina. Woodson v. North Carolina, 428 U.S. 280, 287 n.6 (1976) (plurality opinion) (citing N.C. Gen. Stat. § 14-21 (1975)). 30 That is, the officers hid evidence despite knowing that doing so could lead directly to Mr. Long’s death. Such an action is repugnant in any context. But it takes on a particularly sinister meaning here, given our country’s historical treatment of Black men accused of raping white women. IV. Mr. Long has spent forty-four years in prison. He has maintained his innocence all along. And the State has maintained his guilt—even while concealing evidence showing otherwise. 30 The U.S. Supreme Court found North Carolina’s statute unconstitutional between the time of Mr. Long’s arrest and his trial. See Woodson, 428 U.S. at 305. That decision converted the mandatory sentence to life in prison. State v. Mathis, 239 S.E.2d 245, 250 (N.C. 1977). Today, however, Mr. Long’s sentence for first-degree rape would have been far less than life in prison. See N.C. Gen. Stat. §§ 14-27.21, 15A-1340.14, 15A-1340.16– .17 (labeling first-degree forcible rape as a Class B1 felony, for which the penalty cannot be life imprisonment unless (1) the defendant has multiple prior felonies (or at least fourteen prior misdemeanors) and (2) the jury finds aggravating factors warranting an increased sentence). Under the State’s current sentencing structure, Mr. Long may well have already served far more than the maximum permissible sentence for rape. And while he was also convicted of first-degree burglary, life imprisonment is no longer permitted for that crime. See N.C. Gen. Stat. § 14-52. 92 That evidence has now trickled out, revealing the truth that Mr. Long has declared for decades: he should not have been found guilty. Today, the Court remands to give the State yet another opportunity to disclose the evidence it should have disclosed nearly half a century ago. Based on the record in this case over the last fifteen years, I would not be surprised if more evidence does turn up. But since the evidence is sufficient today to grant Mr. Long the relief he has so long pursued, I would not wait for further proceedings on remand. Forty-four years is an unconscionably long period to wait for justice. It is time. 93 RICHARDSON, Circuit Judge, with whom Judges WILKINSON, NIEMEYER, AGEE, QUATTLEBAUM, and RUSHING join, dissenting: Ronnie Wallace Long is serving two life sentences after a North Carolina jury convicted him of rape and burglary in 1976. The prosecution failed to disclose evidence— that is beyond dispute. But the North Carolina courts concluded that the undisclosed evidence would not have impacted the outcome of Long’s trial. Thus, the evidence was immaterial under Brady v. Maryland, 373 U.S. 83 (1963), and the state post-conviction court denied Long relief. Now, Long asks the federal judiciary to appraise the decision of the North Carolina courts under the Antiterrorism and Effective Death Penalty Act (“AEDPA”), Pub. L. No. 104-132, 110 Stat. 1214 (1996). Under AEDPA’s strict dictates, our review is severely limited. State courts are presumptively competent to review claims of error and right wrongful convictions, and AEDPA reserves federal relief for extreme circumstances. See Burt v. Titlow, 571 U.S. 12, 19–20 (2013). Under AEDPA, this appeal does not concern whether I agree with the state court’s analysis. My own opinion is irrelevant. Rather, I must ask whether the North Carolina disposition of Long’s post-conviction claims was so objectively off the rails that it surpasses any possibility for fairminded disagreement. See 28 U.S.C. § 2254(d); Harrington v. Richter, 562 U.S. 86, 103 (2011). The majority finds this standard satisfied. I respectfully disagree. My disagreement centers on two matters. To start, I take issue with how the majority characterizes the evidence. Consider the three items highlighted in its introduction. First, lab tests with “negative” results do not and cannot “demonstrate” that 94 the petitioner was “not linked” to the crime scene in “any way”: The absence of evidence is not proof of absence. Majority Op. 5. Second, the lack of biological testing similarly fails to bear on the petitioner’s guilt or innocence one way or another. And third, Long does not even raise on appeal the fingerprint claim (which is based on undisclosed latent prints from the crime scene even though the victim described her attacker as wearing gloves). Of course, Long relies on other evidence, as I discuss below. But when I put all the undisclosed evidence together, I must decide only whether fairminded judges could debate whether the state court’s conclusion was correct. See 28 U.S.C. § 2254(d). Because I find that they could, Long does not overcome AEDPA’s limitation on relief. Next, I respectfully disagree with the majority’s interpretation of the North Carolina post-conviction decision. Could the state court’s decision be read to subject Long’s claims to an enhanced burden? Possibly, as the majority opinion aptly shows. But again, that is not the right question. AEDPA requires us to give state courts “the benefit of the doubt” whenever possible. Burt, 571 U.S. at 15. So as long as I can reasonably read the state court’s decision to avoid an objective and indisputable error, I must do so. See Wetzel v. Lambert, 565 U.S. 520, 524–25 (2012); see also Richter, 562 U.S. at 102. And here, I have no trouble doing just that. The federal courts do not overturn state-court convictions based on sloppy writing or peripheral problems; our role is cabined to correcting extraordinary errors. Because I find the state court’s decision at least debatable, I think AEDPA precludes disturbing Long’s North Carolina convictions. Thus, I respectfully dissent. 95 I. Background A. The 1976 burglary, rape, and investigation On April 25, 1976, at around 9:30 p.m., a man entered the home of 54-year-old (now deceased) widow, Sarah Bost, in Concord, North Carolina. He put a knife to her throat and demanded money. When Mrs. Bost could not find money in her purse, the man became angry, cursed her, threw her to the ground, ripped her clothes off, beat her, and raped her. The man repeatedly ordered Mrs. Bost not to look at his face, but she defied him in hope that she could identify her attacker if she survived. During the assault, the phone rang and startled the man. He pulled up his pants and went out the front door. Mrs. Bost ran unclothed out the back door to her neighbor’s house. Once there, she told her neighbor that a Black man had just raped her. The neighbor brought Mrs. Bost inside and called the police. The Concord Police Department investigated the attack. Officers gathered evidence from the scene and interviewed Mrs. Bost, who told them that she was “attacked and raped by a black male wearing a leather coat, toboggan, and  gloves.” J.A. 1429; see also J.A. 206 (“[H]e had on black gloves . . . [and he did not] remove those gloves.”). She described her attacker as around five foot five inches to five foot nine inches tall, with a “slender build and slim hips” and a thin mustache. J.A. 1428; see also J.A. 200–01, 305. She also said that he wore blue jeans and used “correct [E]nglish and at times spoke very softly” with no noticeable accent. J.A. 1428. An ambulance then took Mrs. Bost to a local hospital. 96 At the hospital, Mrs. Bost was examined by Dr. Monroe, a physician specializing in gynecology. Dr. Monroe observed extensive scratches, bruising, and lacerations from Mrs. Bost’s face to her legs. He also noted her “fingernails looked like they had been traumatized, or nearly bent backwards,” likely from an attempt to fight off her attacker. J.A. 295. During a pelvic exam, Dr. Monroe assembled a microscope slide of vaginal fluid that revealed an “extremely high count of live, very active, human spermatozoa.” J.A. 296. Mrs. Bost remained at the hospital for five days for observation and treatment. The day after the rape, officers showed Mrs. Bost a photographic lineup of thirteen male suspects, hoping she might identify her attacker. She did not identify a suspect from these photographs, which did not include a picture of Long. Less than two weeks later, officers asked Mrs. Bost to go to the local courthouse to observe the proceedings. The Concord Police had learned about a similar burglary and rape in Washington, D.C. In that case, the victim found Long’s Social Security card in her apartment after the attack. Based on the card left behind, the Washington Metropolitan Police sought Long, a Concord resident, for questioning. Having asked Mrs. Bost to come to court, officers informed Mrs. Bost that the man who raped her may, or may not, be present in court and that she should discretely notify them if she identified him. The officers also told her that she may have to come to court on two or more occasions. She was also asked to bring a neighbor or friend and to wear a disguise. At first reluctant, Mrs. Bost agreed. Mrs. Bost arrived at the courthouse on the morning of May 10, 1976, disguised by a red wig and glasses. Accompanied by her neighbor, she sat in the second row, while two 97 officers sat away from them. According to the officers, there were “approximately 60 to 65 persons in the court room either in the audience or persons on trial,” with “12 adult black males in the general age group” described by Mrs. Bost. J.A. 1433. As Mrs. Bost observed the proceedings, several cases were called involving black men as defendants. After about a half hour, Ronnie Wallace Long’s case was called. 1 Long, “wearing a medium brown leather coat, a l[ei]sure shirt flowered, no hat, [and] dress pants,” (clothing that Mrs. Bost later testified she’d “never seen . . . before,” J.A. 158), exited the row of seats to the left side of the gallery and “walked around to the defense table where he was readily visible by Mrs. Bost.” J.A. 1433. Mrs. Bost notified the officers that she had identified her attacker, telling them that “there was no doubt in her mind that this person Ronnie W. Long was the person who entered her house.” J.A. 1433; see also J.A. 314–15 (“I observed several black males in the courtroom . . . and when I saw him (meaning Ronnie Long) I knew that was the man . . . based on his general appearance . . . I will never forget his profile, the coloring of his skin. It was light, not colored real dark. Another reason, his mannerism and the way he walked . . . I knew his voice . . . I will never forget the way he talked to me. The way it sounded. His voice sounded relatively soft, but not real loud.”). Officers then took her to the police station, where she then identified Long in a photographic lineup. During her interview at the station, she again “assure[d]” officers with “no doubt in [her] mind,” that Long was her attacker. J.A. 1435. 1 Long was in court that day on an unrelated trespassing charge. See J.A. 27 n.8. 98 After Mrs. Bost identified Long as her rapist, the officers asked him to come to the police station. Long drove himself to the station, where he waived his Miranda rights and permitted the officers to search his car. The search revealed a pair of black gloves over the visor, a green toboggan under the driver’s seat, and several matchbooks. The officers seized these items, as well as Long’s leather jacket. One officer, Officer Isenhour, took impressions of the bottoms of Long’s shoes. Long was arrested and charged with burglary and rape. B. The trial Long’s trial began in Cabarrus County Superior Court in September 1976. On the first day, Long’s trial counsel moved to suppress Mrs. Bost’s courthouse and photographic identifications, arguing that they were impermissibly tainted by the officers’ actions. After permitting the parties to question Mrs. Bost without the jury present, the court denied Long’s motions. In its case in chief, the State called Mrs. Bost, along with Dr. Monroe and several law-enforcement officers. During direct examination, Mrs. Bost pointed to Long in the courtroom, identifying him as her attacker. She explained that Long repeatedly told her not to look at his face, and he “kept holding [her] face . . . and mashing [her] face to the side” as he raped her. J.A. 245–46. And although she told Long that she “d[idn’t] want to see [his] face,” J.A. 197, she did so anyways. Mrs. Bost testified that she got a “good, clear look” at her attacker’s face. J.A. 203; see also J.A. 135 (“I looked at his face. I couldn’t help but see his face.”); J.A. 136 (“He was across my legs and looking at me and there was no way not to see him.”); J.A. 138 (“How could I help but look at his face?”); J.A. 143 99 (“Just . . . when a man’s on top of you, how close is his face to you? It’s just right at me.”); J.A. 144 (testifying that she “g[o]t a good, clear unobstructed view of his face”). Mrs. Bost’s in-court description of her attacker was consistent with her preidentification descriptions to the police. In the pre-identification description, she described her attacker as a light-colored black male. See J.A. 1434 (May 10, 1976 identification) (“I will never forget . . . [the] coloring of his skin, it was light colored[,] not real dark.”). And when cross-examined about her initial descriptions to police and testimony on direct, she again confirmed that she had identified her attacker as “light skinned, or [a] yellow . . . black man.” J.A. 248. When defense counsel pressed for clarification, adding, “you meant, I take it, by yellow that he was not a very dark, or that he was an unusually light skinned black male,” Mrs. Bost responded, “[n]ot unusually light . . . ” as in “not your normal black person.” J.A. 248 (emphasis added). Long’s defense had several components. First, he sought to impeach Mrs. Bost’s testimony on the grounds that cross-racial eyewitness identification is often suspect. Second, the defense pointed to the lack of any physical evidence tying him to the crime scene. Third, Long introduced testimony about his whereabouts on the evening of the crime to establish an alibi. See Majority Op. 15–17 (describing the alibi evidence); Concurring Op. 81–82 (same). The jury convicted Long of both burglary and rape. He was sentenced to two life terms in prison, and his conviction was affirmed on direct appeal by the Supreme Court of North Carolina. North Carolina v. Long, 237 S.E.2d 728 (N.C. 1977). 100 C. Disclosure of new evidence In the years that followed, Long filed unsuccessful post-conviction petitions in state and federal court, including a 28 U.S.C. § 2254 federal habeas application. North Carolina v. Long, 377 S.E.2d 228 (Mem.) (N.C. 1989); Long v. Dixon, Civ. No. 4:89-cv-278 (M.D.N.C. May 3, 1990). Then, in 2005, he moved in state court for location and preservation of evidence, seeking any biological evidence to use in DNA testing and the pieces of clothing recovered, such as his black leather jacket and green toboggan. The judge granted Long’s motion, ordering the prosecution and law enforcement to locate and preserve all evidence related to his case. The order led to the disclosure of dozens of documents falling into three groups: (1) State Bureau of Investigation forensic reports documenting the testing of physical evidence; (2) the master case file on the investigation; and (3) excerpts of Mrs. Bost’s medical records from her hospitalization. The parties agree that this information should have been disclosed to Long. 1. Forensic test reports The State disclosed reports and handwritten notes from forensic tests conducted on evidence delivered to the lab by Officer Isenhour. The reports revealed that analysts (1) compared the single hair found at the crime scene with Long’s head and pubic hair samples and concluded that they did not match, noting that no other hairs were found on Mrs. Bost’s clothing, J.A. 1466; (2) examined Long’s leather jacket, gloves, and toboggan and did not find any trace of paint or carpet fibers matching samples taken from Mrs. Bost’s home, J.A. 1454–55; (3) compared five matchbooks from Long’s car with three burned matches 101 recovered from the upstairs windowsill and found insufficient evidence to conclude that they were linked, J.A. 1463; and (4) compared a latent shoeprint recovered from the frontporch banister of Mrs. Bost’s home with the inked impressions of Long’s shoe bottoms, concluding that Long’s shoes could have made the shoeprint, but there was insufficient information for a definite match, J.A. 1464. The testing of the hair, clothing, and matchbooks was not disclosed to Long’s defense counsel before trial. 2. Master case file The Master Case File contained two “Request[s] For Examination of Physical Evidence,” each written by Officer Isenhour to document items of evidence he delivered for forensic testing the day after Long’s arrest. J.A. 1456, 1465. The first request listed the latent shoeprint taken from outside Mrs. Bost’s house and inked impressions of Long’s shoes. See J.A. 1465. It asked the forensic analysts to “[e]xamine for identification from latent lift to known shoe-bottom impressions.” J.A. 1465. The second request listed 13 more items of evidence provided for forensic testing, including Long’s leather jacket, green toboggan, and leather gloves; paint and carpet samples taken from the crime scene; samples of Long’s and Mrs. Bost’s head and pubic hair; a “suspect hair from the scene”; matchbooks from Long’s car; burned matches obtained from the upstairs windowsill; and Mrs. Bost’s clothing. J.A. 1454. It requested the examination of Long’s clothing “for the presence of paint and fibers” and the comparison of any found with the samples taken from Mrs. Bost’s home. Id. It also requested a forensic comparison of the hair found at the scene (and any hairs found on Mrs. 102 Bost’s clothing) with the hair samples taken from Long, as well as a comparison of the burned matches from the windowsill with the matchbooks recovered from Long’s car. Neither request was disclosed to Long’s counsel before trial. Moreover, at trial, Isenhour offered an incomplete picture of the testing he had requested. While he testified to delivering the shoeprints for testing, he also said that the black leather jacket, the green toboggan, and black leather gloves remained in his “custody and control” since he received them from another officer during the investigation. J.A. 415–16. 3. The victim’s medical records The county hospital produced 26 pages of Mrs. Bost’s medical records from her hospitalization and medical examination hours after the rape. After in camera review, the judge authorized the release of 11 pages to Long’s post-conviction counsel. The released records showed that Dr. Monroe collected biological evidence of the rape in accordance with the hospital’s rape protocol: he prepared slides of live spermatozoa, took two swabs of vaginal secretions that he placed in test tubes, and obtained pubic combings. After the examination, the records show that the hospital released pubic hair and one test tube to an officer after authorization from Mrs. Bost. J.A. 1475–79. These records were not disclosed to Long’s defense counsel. Efforts to locate any biological evidence in 2007 were unsuccessful. D. State post-conviction proceedings After receiving the new evidence, Long filed a Motion for Appropriate Relief (“MAR”) (North Carolina’s version of a habeas petition), raising (1) an allegation that the state failed to disclose exculpatory information material to the defense in violation of the 103 Due Process Clause, the North Carolina Constitution, and Brady, and (2) a newly discovered evidence claim under state law. After an evidentiary hearing, the state court denied Long’s motion, finding that both of Long’s claims failed. The Supreme Court of North Carolina affirmed. North Carolina v. Long, 705 S.E.2d 735 (N.C. 2011). After the state denied relief, Long filed another federal habeas application in 2012. But because he had failed to receive pre-filing authorization from this court as required by AEDPA, the district court dismissed Long’s application for lack of jurisdiction. See Long v. Lancaster, No. 1:12-CV-119, 2012 WL 3151179, at *1 (M.D.N.C. Aug. 2, 2012). Long neither appealed nor sought the necessary pre-filing authorization. Then, in 2015, Long participated in the North Carolina Innocence Inquiry Commission’s Postconviction DNA Testing Assistance Program. Those efforts revealed 43 latent fingerprints taken from the crime scene that the state had not disclosed. The crime scene investigators collected the prints even though Mrs. Bost would explain— immediately after the attack and at trial—that her attacker wore gloves. Testing would exclude Long as the source of those prints. Long returned to federal court and requested pre-filing authorization for a successive federal habeas application, which we granted. See Order, In re Long, No. 16– 295 (4th Cir. filed May 24, 2016). Long then filed the application at issue here. The district court at first dismissed Long’s application after finding that it presented, along with the Brady claims, an unexhausted claim involving the undisclosed latent fingerprints. See Long v. Perry, No. 1:16CV539, 2016 WL 7235779 (M.D.N.C. Dec. 14, 2016). We reversed and remanded to the district court, finding that Long had “unequivocally 104 disclaimed” his latent fingerprint claim. Long v. Perry, 699 F. App’x 260, 261 (4th Cir. 2017). Without the fingerprint claim, Long’s application relies solely on his argument that “the MAR Order was contrary to, or an unreasonable application of, the Supreme Court’s clearly established Brady jurisprudence with respect to each of the three fundamental Brady components.” J.A. 46. The state moved for summary judgment on procedural and merits grounds, and the matter was referred to a magistrate. In a 66-page report and recommendation, the magistrate found on the merits that the state court’s application of Brady to the newly discovered evidence was reasonable. The district court adopted the magistrate’s report and recommendation and dismissed Long’s petition. In so doing, the district court also issued a certificate of appealability on the question before us today in this appeal: “Did the state court unreasonably apply Supreme Court precedent in concluding that the suppressed exculpatory evidence in question, considered cumulatively, did not qualify as material?” J.A. 1725. Applying the deferential review that AEDPA demands, I find that it did not. II. Discussion We review the district court’s decision de novo. Muhammad v. Kelly, 575 F.3d 359, 367 (4th Cir. 2009). And we “examine [Long’s] argument through the dual lens of the AEDPA standard and the standard set forth by the Supreme Court in Brady.” Richardson v. Branker, 668 F.3d 128, 144 (4th Cir. 2012). “Because of AEDPA’s heavy emphasis on the standards governing the review of the merits of a habeas application,” identifying the relevant standard is a good place to 105 start. Woodford v. Garceau, 538 U.S. 202, 206 (2003) (emphasis omitted). AEDPA aims to “further the principles of comity, finality, and federalism,” Williams v. Taylor, 529 U.S. 420, 436 (2000), by ensuring that state proceedings remain “the ‘main event’ . . . rather than a ‘tryout on the road’ for . . . the determinative federal habeas hearing.” Wainwright v. Sykes, 433 U.S. 72, 90 (1977). Accordingly, AEDPA exudes deference to the state courts, recognizing their central role in righting wrongful convictions and their “presumptive competen[ce] to adjudicate claims arising under the laws of the United States.” Tafflin v. Levitt, 493 U.S. 455, 458 (1990). A key vehicle by which Congress aimed to accomplish this objective is § 2254(d). See Williams, 529 U.S. at 404 (“It cannot be disputed that Congress viewed § 2254(d)(1) as an important means by which its goals for habeas reform would be achieved”). When a prisoner’s claims have already been adjudicated by a state court, this subsection limits the power of the federal courts to grant habeas relief to two grounds. First, if the state decision turns on a factual determination, that determination must be “objectively unreasonable in light of the evidence presented in the state-court proceeding.” Miller-El v. Cockrell, 537 U.S. 322, 340 (2003); § 2254(d)(2). Otherwise, the state decision must have been contrary to or involve an unreasonable application of, clearly established federal law. § 2254(d)(1); see Virginia v. LeBlanc, 137 S. Ct. 1726, 1728–29 (2017). 2 2 While the standard for satisfying § 2254(d) is demanding, Long’s successive application faces an even greater hurdle on remand: 28 U.S.C. § 2244(b)(2) requires establishing “by clear and convincing evidence that, but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” 106 But an unreasonable determination differs from an incorrect determination. See Williams, 529 U.S. at 410. “[S]o long as ‘fairminded jurists could disagree’ on the correctness of the state court’s decision,” that decision was reasonable under § 2254(d). Richter, 562 U.S. at 101 (quoting Yarborough v. Alvarado, 541 U.S. 652, 664 (2004)). In other words, the state court’s decision must not be “merely wrong” but “objectively unreasonable” as “even clear error will not suffice.” LeBlanc, 137 S. Ct. at 1728 (internal quotation marks omitted). Thus, § 2254 reflects the view that habeas relief is a “guard against extreme malfunctions in the state criminal justice systems.” Jackson v. Virginia, 443 U.S. 307, 332, n.5 (1979) (Stevens, J., concurring in the judgment). Of course, to determine whether the state court has unreasonably applied federal law, we must identify the federal law that applies. Here, the applicable federal law consists of the rules for determining whether the state violated a defendant’s Fourteenth Amendment Due Process rights under Brady v. Maryland, 373 U.S. 83 (1963). The state violates Brady when the prosecution (1) fails to disclose (2) material evidence (3) favorable to a criminal defendant. See 373 U.S. at 87; see also Kyles v. Whitley, 514 U.S. 419, 432 (1995). Evidence is material if it creates a “‘reasonable probability’ of a different result,” Kyles, 514 U.S. at 434, thus “undermin[ing] confidence in the outcome of the trial.” United States v. Bagley, 473 U.S. 667, 678 (1985). It is not enough for the withheld evidence to create the possibility of a different verdict; a different result must be reasonably probable. See Strickler v. Greene, 527 U.S. 263, 291 (1999); United States v. Agurs, 427 U.S. 97, 109–10 (1976); see also Richter, 562 U.S. at 112 (“The likelihood of a different result must be substantial, not just conceivable.”). 107 In reviewing the state court’s decision, we must take care to apply the AEDPA and Brady standards “in tandem.” Richter, 562 U.S. at 105. The question then, is not simply whether the new evidence creates a “‘reasonable probability’ of a different result,” Kyles, 514 U.S. at 434, but whether any fairminded judge could agree with the state court’s Brady assessment. See Richardson, 668 F.3d at 144. A. The state court’s conclusions Based on the evidence before it, the state court made the following “conclusion of law”: As to the cumulative [e]ffect of the items of evidence the defense alleges they did not receive, this court finds, based on the findings of fact and conclusions of law stated herein, that the contents of several of the items the defense alleges they did not receive were fully addressed in front of the jury; that other materials contained in reports were more favorable to the State’s case than the defendant’s; and that any remaining matters that were not presented to the jury were of little or no value to the case as a whole; and that the cumulative [e]ffect of any items with any value is so minimal that it would have had no impact on the outcome of the trial. J.A. 1358–59 (emphasis added). The court’s determination compels a specific outcome under Brady: if evidence has no impact on the trial outcome, then it must leave no reasonable probability of a different result. On this basis, the state court’s decision that no Brady violation occurred adheres to controlling Supreme Court precedent. See Strickler v. Greene, 527 U.S. 263, 291 (1999). But Long instead asks us to scrutinize the state court’s summary conclusions. Petitioner’s Br. 26 (citing J.A. 1359 ¶17). After determining the evidence would have “no impact,” the state court “in summary” wrote: 108 The Defendant has failed to prove by a preponderance of the evidence that his due process rights have been violated under Brady, in that he has not shown by a preponderance of the evidence that the claimed evidence was withheld by the State, that it was exculpatory, or that the result likely would have been different with the claimed evidence. J.A. 1359 (citations omitted). As Long correctly argues, this summary conclusion incorrectly articulates Brady’s burden of proof. Brady requires that the defendant show a reasonable probability that a jury would find him innocent, given the new evidence. It “does not require demonstration by a preponderance that disclosure of the suppressed evidence would have resulted ultimately in the defendant’s acquittal.” Kyles, 514 U.S. at 434 (emphasis added). Because the state court’s conclusion described a preponderance burden, it directly contradicts Supreme Court precedent. 3 But this inaccuracy alone does not make the state court’s decision unreasonable. If another ground provided by the state court can sustain its decision, this error is “beside the point.” Wetzel, 565 U.S. at 524; cf. Woodford v. Visciotti, 537 U.S. 19, 23–24 (2003) (inconsistent descriptions of the burden of proof did not make the state court’s decision unreasonable under AEDPA). “Congress,” the Supreme Court has explained, “intended 3 The state argues that the preponderance language refers to the evidentiary standard in a North Carolina MAR hearing: “[T]he moving party has the burden of proving by a preponderance of the evidence every fact essential to support the motion.” N.C.G.S. § 15A-1420(c)(5); Respondent’s Br. 18–19. This reading may explain the court’s reference to a preponderance in its factual determinations. See, e.g., J.A. 1359 (determining, under a preponderance standard, the defendant did not prove that the state failed to disclose forensic reports). But it is different to say—twice in the same sentence—that the defendant must prove, under Brady, the likelihood of a different outcome by a preponderance of the evidence. 109 federal judges to attend with the utmost care to state-court decisions, including all of the reasons supporting their decisions, before concluding that those proceedings were infected by constitutional error sufficiently serious to warrant the issuance of the writ.” Williams, 529 U.S. at 386. And so we may not disturb a state court’s judgment unless “each ground supporting the state court decision is examined and found to be unreasonable.” Wetzel, 565 U.S. at 525; see also Richter, 562 U.S. at 102. As described above, the state court’s “No Impact Conclusion” is “sufficient to reject [Long’s] claim, so it is irrelevant that the court also invoked a ground of questionable validity.” Parker v. Matthews, 567 U.S. 37, 42 (2012) (citing Wetzel, 565 U.S. at 523−24); see also Littlejohn v. Trammell, 704 F.3d 817, 831 (10th Cir. 2013). Nevertheless, the majority refuses to credit the “No Impact Conclusion,” asserting that it was “informed” by an incorrect burden and rendered unreasonable by the favorability analysis. Majority Op. 30–31. I respectfully disagree. To begin with, the majority takes the wrong approach to the problem: It is too quick to attribute error to the “No Impact Conclusion.” As the Supreme Court has explained, AEDPA requires that we give the state courts “the benefit of the doubt.” Burt, 571 U.S. at 15; see also Williams, 529 U.S. at 386. Thus, we must resist the urge to too quickly conclude that other parts of the state court’s analysis infected a proper reason for its result. “[S]o long as ‘fairminded jurists could disagree’ on the correctness of the state court’s decision,” it must be sustained. Richter, 562 U.S. at 101; see also id. (When the state court’s reasoning is absent or ambiguous, a federal court must determine “what arguments 110 or theories . . . could have supported[ ] the state court’s decision; and then it must ask whether it is possible fairminded jurists could disagree.”). Here, I think the best reading—much less one fair reading—of the state court’s decision is that the No Impact Conclusion is not infected by other statements. And so I think that a fairminded judge could find it proper. I consider the majority’s counterarguments in turn. 1. The No Impact Conclusion is not infected by the incorrect Brady burden First, consider the majority’s assertion that the No Impact Conclusion “refers to, and necessarily depends on” the erroneously stated Brady burden. Majority Op. 30. In the majority’s view, because the state court found that no single piece of evidence, “by a preponderance of the evidence,” would have undermined the proceedings, the “No Impact Conclusion” cannot be sustained. Majority Op. 32 (quoting J.A. 1356). In my view, this misconstrues the state court’s decision and ignores the nature of the cumulative-evidence inquiry. Whether evidence is material under Brady “turns on the cumulative effect of all  evidence suppressed by the government.” Kyles v. Whitley, 514 U.S. at 421. Of course, this requires assessing the value of each item and adding them up, see id. at 436 n.10, but what matters is the cumulative assessment—not whether any particular item alone would have changed the result. 4 While the state court evaluated whether particular items would 4 Should any particular piece of evidence—standing alone—suffice to undermine our confidence in the trial, that would be captured by the cumulative assessment. 111 have changed the result of the trial, only its cumulative assessment decided the issue. And in its cumulative analysis, the state court added up the evidence without applying any preponderance filter, concluding that the impact of “any items with any value . . . would have had no impact on the outcome.” J.A. 1359. Thus, whether the state court properly assessed whether any individual piece of evidence would have changed the outcome of the trial is irrelevant. The majority strains to read an interaction between the state court’s conclusions about each individual piece of evidence and its ultimate conclusion on the cumulative effect of all the evidence. But I would take the state court at its word. “Any” items, at least to me, means that the state court assessed all the items. As the Supreme Court has confirmed, “any” has “an expansive meaning” bringing all types of an item within its reach. Dep’t of Hous. and Urban Dev. v. Rucker, 535 U.S. 125, 131 (2002) (internal citations and quotation marks omitted); see also Republic of Iraq v. Beaty, 556 U.S. 848, 856 (2009); Ali v. Fed. Bureau of Prisons, 552 U.S. 214, 219–220 (2008). And “any” value, as I understand the term, similarly means that the court added up the value of all the evidence. In doing so, the state court found that this evidence would not have impacted the outcome of the trial. This is what Brady requires, and I see no error with this cumulative conclusion. Next the majority turns to the paragraph following the state court’s “No Impact Conclusion,” arguing that this subsequent analysis infects the previous conclusion by erroneously applying Brady. Again, I disagree. The unqualified conclusion that the new evidence has ‘no impact’ logically precedes the erroneous preponderance determination. Moreover, it structurally precedes the “No Impact Conclusion.” So even without needing 112 to give the state court the benefit of the doubt it is due, I would find that the downstream error does not poison the upstream well. Wetzel teaches that if the ‘no impact’ conclusion—which alone is sufficient to reject Petitioner’s argument—is reasonable, “whatever [the state] courts had to say” in a subsequent conclusion “is beside the point.” 565 U.S. at 524; see also Parker, 567 U.S. at 42 (holding, where it was “not clear” that the state court relied “sole[ly]” on a “questionable” ground, that another “ground was sufficient to reject [Petitioner’s] claim, so it is irrelevant that the court also invoked a ground of questionable validity”). So I reject this argument. For these reasons, I think fairminded judges could conclude that other references to an erroneous Brady standard would not undermine the state court’s No Impact Conclusion. And I am not persuaded by the majority’s arguments to the contrary. 2. The No Impact Conclusion is similarly distinct from the favorability assessment On similar grounds, I reject the majority’s claim that the state MAR court’s erroneous conclusions about the favorability of the evidence “necessarily rendered its No Impact Conclusion unreasonable.” Majority Op. 30. The majority observes that “remarkably, the district court  treats these MAR Court favorability errors and misapplications as having no bearing on the No Impact Conclusion.” Majority Op. 37. This is so because the majority again rests on an infection argument: in the majority’s view, the state court relied on the erroneous favorability finding to reach its No Impact Conclusion—thus rendering its No Impact Conclusion infirm. See Majority Op. 37–38. 113 But again, this reading ignores the plain text of the state MAR court’s order. See J.A. 1359 (“any items with any value . . . would have had no impact on the outcome.”). And it matters not that the state court’s assessment of favorability concerned the same items it later found immaterial. Either a lack of materiality or favorability suffices to reject a Brady claim. And, as discussed above, so long as one conclusion is “sufficient to reject [Long’s] claim,” we must defer. Parker, 567 U.S. at 42 (citing Wetzel, 565 U.S. at 523−24). For instance, in Wetzel, the petitioner—convicted of murder, robbery, and other offenses at trial—argued that the Commonwealth violated Brady when it failed to disclose a police activity sheet that supposedly identified another participant in the robbery. See Wetzel, 565 U.S. at 521. According to petitioner, the activity sheet was exculpatory and would have impeached one of the Commonwealth’s primary witnesses. See id. at 521–22. The state court disagreed, finding that the activity sheet was “‘not exculpatory or impeaching’ but instead ‘entirely ambiguous.’” Id. at 524. And the state court determined that the contents of the activity sheet would be cumulative to other impeachment evidence. See id. These rationales supporting the state court’s conclusion were neither isolated from one another nor entirely separate—on the contrary, both turned on assessing the activity sheet’s contents. See Commonwealth v. Lambert, 884 A.2d 848, 855–56 (Pa. 2005). Despite focusing on the same evidence, either rationale was sufficient, so the Supreme Court vacated the Third Circuit’s decision to issue the writ. So too here. The Brady materiality and favorability determinations may require an assessment of the same evidence, but either provided an independent reason that permitted 114 the state court to reject Long’s claim. As long as the state court could have gotten one debatably correct, the second is beside the point. Cf. Blackston v. Rapelje, 780 F.3d 340, 354 (6th Cir. 2015) (noting that the state court’s rationales “overlap” and continuing to analyze each pursuant to Wetzel). Indeed, the Supreme Court has characterized Wetzel as a sufficiency inquiry, not a complete separateness inquiry. See Parker, 567 U.S. at 42 (“That ground was sufficient to reject [Petitioner’s] claim, so it is irrelevant that the court also invoked a ground of questionable validity.”) (citing Wetzel, 565 U.S. at 524–25); see also Littlejohn, 704 F.3d at 831 (explaining that Wetzel requires “a sufficient substantive ground”). Thus, because the “No Impact Conclusion” is “sufficient” to reject Long’s claim on Brady’s materiality prong, “it is irrelevant” that its favorability conclusion was deficient. Parker, 567 U.S. at 42. B. The “No Impact Conclusion” is debatably correct Of course, Long remains free to challenge the reasonableness of the state court’s “No Impact Conclusion,” and he does. The state court assessed, separately and cumulatively, (1) the forensic testing reports, (2) the master case file, and (3) Mrs. Bost’s medical records. As to these items, Long asserts the court “erroneously construed [their] value” and “ignored the value of impeachment evidence.” Petitioner’s Br. 27. In other words, Long asks us to dissect the judgment calls made by the state court. The majority bites. In their view, although three courts—the magistrate judge, the district court, and the panel—reached the opposite conclusion, there can be no “fairminded disagreement” about this. See Majority Op. 38 (quoting Richardson v. Branker, 668 F.3d 115 128, 149 (4th Cir. 2012)). But I think the conclusion debatable. Before turning to the cumulative impact of the evidence, I review each item individually to assess its probative value. See Kyles, 514 U.S. at 436 n.10. 1. Forensic testing reports Long claims that in failing to disclose these reports, the State “painted a picture of a very limited forensic investigation, when in fact the forensic investigation was extensive.” Petitioner’s Br. 10. According to Long, the negative results reflected in the reports tend to prove his innocence. And he contends that the effect of the reports with negative findings “would have bolstered and supported the arguments his counsel made,” such that “a jury would have been less likely to convict with each successive negative result.” Id. at 37–38. The state court discounted this evidence, explaining that the absence of evidence is not evidence of absence. See J.A. 1357–58. This determination finds support in decisions by other courts and in commentaries. Case v. Hatch, 731 F.3d 1015, 1043 (10th Cir. 2013) (absence of a match is “neither incriminating nor exonerating”); see also Paul Erwin Kish & Herbert Leon MacDonell, Absence of Evidence is Not Evidence of Absence, 46 J. FORENSIC IDENTIFICATION 160–64 (1996). And, as Long and the state court acknowledge, the reports’ conclusions about the hair, paint or carpet fibers, matchbooks, and shoeprint testing mirror the testimony at trial. See J.A. 1357 (“[T]he agent’s testimony at trial was consistent with his report and the jury learned everything that was contained in the report yet found the defendant guilty of all 116 charges.”). Cumulative evidence “is generally not considered material for Brady purposes.” Johnson v. Folino, 705 F.3d 117, 129 (3d Cir. 2013). Whether or not I would have come to the exact same conclusion, I cannot say that it was unreasonable for the state court to discount the reports of the forensic testing conducted. 2. Master case file Long’s Master Case File contained two documents about the forensic testing requested. The first is Officer Isenhour’s request for analysts to conduct a forensic comparison of the latent shoeprint to Long’s shoeprints. This document is material only inasmuch as it confirms law enforcement testimony that the State Bureau of Investigation conducted the analysis disclosed to the jury at trial. Thus, the first request is cumulative to what was presented at trial and, in any event, has no probative value in showing Long’s guilt or innocence. The second document lists 13 items of physical evidence for testing. Long concedes that the document itself does not suggest his guilt or innocence. Instead, he argues it provides impeachment evidence because it shows that Isenhour’s testimony “was false and concealed the true facts surrounding the evidence brought to the SBI lab.” Petitioner’s Br. 32. Contrary to the prosecution’s representations at trial, see J.A. 415–16, this document suggests that the physical evidence did not remain in Isenhour’s custody. Rather, he delivered the items to the State Bureau of Investigation. In determining the materiality of impeachment evidence, we consider “the salience of the subject matter of the impeachment.” Spicer v. Roxbury Corr. Inst., 194 F.3d 547, 117 561 (4th Cir. 1999). When impeachment evidence directly “relat[es] to the central issue” in the case, the prosecution’s case suffers a “more serious blow.” Id. Thus, we are more inclined to find that the evidence places the case in an entirely different light. See id.; see also United States v. Avellino, 136 F.3d 249, 256–57 (2d Cir. 1998). For example, in Giglio v. United States, the Supreme Court considered whether the government’s failure to disclose its promise not to prosecute a key witness in exchange for his testimony was material under Brady. 405 U.S. 150, 151 (1972). In Giglio, the government’s case depended “almost entirely” on one witness’s testimony. Id. at 154. Because that witness’s credibility “was therefore an important issue in the case” and the impeachment evidence directly undermined that credibility, the Supreme Court remanded the case for a new trial. Id. at 154–55; see also Kyles, 514 U.S. at 441 (finding suppressed evidence impeaching eyewitnesses to be material because “the essence of the State’s case was the testimony of eyewitnesses, who identified Kyles as [the] killer”) (internal quotation marks omitted). In contrast, where suppressed impeachment evidence focuses on a peripheral issue, it generally fails to shed new light on the case as Brady requires. See Turner v. United States, 137 S. Ct. 1885, 1894 (2017) (evidence “too distant from the main evidentiary points” fails to create a reasonable probability of a different result as required by Brady); see also United States v. Robinson, 627 F.3d 941, 952 (4th Cir. 2010) (investigating officers dismissed for misconduct were not the only witnesses linking petitioner with the crimes); Gilday v. Callahan, 59 F.3d 257, 272 (1st Cir. 1995). In Strickler, the Supreme Court analyzed whether suppressed documents providing grounds for impeaching a government witness were material. 527 U.S. at 263, 289. Unlike in Giglio, the Petitioner’s guilt in 118 Strickler did not ultimately depend on the witness at issue, so the Court reasoned that undermining one witness did not create a reasonable probability of a different result at trial given the other persuasive evidence. See id. at 292–94. Here, the impeachment value of the suppressed evidence does little to render the state court’s conclusion patently unreasonable. A reasonable judge could well consider the impeachment evidence peripheral. Cf. Richter, 562 U.S. at 101 (state court decision is reasonable “so long as ‘fairminded jurists could disagree’ on [its] correctness”). Isenhour failed to tell the truth. But the impeachment of Isenhour would not change the substance of the reports at issue, which do not exculpate Long; would not undermine the state’s theory at trial; and would not point to another perpetrator. The state’s case—as Long must acknowledge—was built on Mrs. Bost’s testimony, not Officer Isenhour’s testimony. So impeaching Mrs. Bost might place the case in a different light. But is not unreasonable to conclude that impeaching Isenhour about his evidence handling and management of nonexculpatory reports would not. 3. The victim’s medical records Finally, I turn to the documents in Mrs. Bost’s medical records. At trial, Long did not know that an officer took custody of one test tube containing biological evidence of the rape. Nothing in the record suggests this evidence was tested using the methods available in 1976. 5 5 Some forensic analyses common today—like DNA testing—did not exist at the time of this crime. The analysis of semen then could only identify a man’s blood type if he secreted blood group antigens. See Petitioner’s Br. 39 n.10. Today, juries may expect precise forensic analyses of biological evidence, yet we must take care not to transplant 119 The state court determined that the information was immaterial: “The mere possibility that an item of undisclosed information might have helped the defense or might have affected the outcome of the trial, does not establish materiality in the constitutional sense.” J.A. 1358 (quoting State v. Alston, 298 S.E.2d 631, 642 (N.C. 1983) (quoting Agurs, 427 U.S. at 109–10)). And the state court found that “[t]here is no evidence that the materials collected from the victim’s person were ever examined by the [State].” J.A. 1354. So the biological evidence was only “potentially useful” and, at least without bad faith, not material. See Arizona v. Youngblood, 488 U.S. 51, 58 (1988); see also Lovitt v. True, 403 F.3d 171, 186–87 (4th Cir. 2005). Moreover, any impeachment value of the lack of testing was limited. In United States v. Cole, 293 F.3d 153, 163 (4th Cir. 2002), we explained that suppressed impeachment evidence is not material when ample information exists for an effective cross examination. Long knew, of course, that Dr. Monroe had collected semen during Mrs. Bost’s examination—Dr. Monroe said as much in trial testimony. He also knew the only testing done on that evidence was observation under a microscope. Long was free to point out the potential exculpatory value of additional testing. Again, the question here is only whether the state court decision was a reasonable one. And I believe it was. those expectations where “modern forensic techniques . . . of course would not have been available.” Babick v. Berghuis, 620 F.3d 571, 577 (6th Cir. 2010). 120 4. Cumulative impact assessment With the above considerations in mind, I conclude by asking whether the state court’s determination that “the cumulative effect . . . is so minimal that it would have had no impact on the outcome of the trial,” J.A. 1358–59, was “so lacking in justification that [it] was an error well understood and comprehended in existing law beyond any possibility for fairminded disagreement.” Richardson, 668 F.3d at 149 (quoting Richter, 562 U.S. at 103). In Kyles, the Supreme Court underscored the significance of considering the cumulative effect of all suppressed evidence in determining whether the state violated Brady. Undertaking its own analysis, the Court concluded that a different result would have been reasonably probable had the state disclosed the substantial amount of evidence. Kyles, 514 U.S. at 441. The evidence’s bearing on the jury would have been considerable, especially with respect to the weight the jury could have given to the defendant’s theory that he had been framed by a police informant, who was actually the murderer. See id. at 449 n.19. But the disclosure of the evidence would have mattered so significantly in Kyles because it would have discredited two of the four eyewitnesses who were “the essence of the State’s case,” id. at 441, in addition to leaving more for the jury to weigh about the overall credibility of the evidence and whether the principal witness was being truthful, see id. at 450. Those pieces of evidence, however, fit together like a puzzle—depicting a coherent image consonant with the defense’s theory of the case. But where evidence “sheds no new light,” Gilday, 59 F.3d at 272, the reasonable probability of a different result 121 is severely diminished, even in spite of the “legion” of new available evidence. Majority Op. 40. The record adequately supports the state court’s conclusion about the strength of the prosecution’s evidence. Consider Mrs. Bost’s identification of Long. As the sole eyewitness, Mrs. Bost gave detailed descriptions of the attacker and his clothing as soon as Concord officers responded to the scene and again when interviewed at the hospital and in her home, in the days following the attack. When the original photo array the police presented to her did not include a photo of Long, she did not identify a suspect. Tellingly, she confidently chose Long out of a crowd in a courtroom with other Black males present, telling officers, “there is no doubt in my mind that [Long] is the man that raped me.” J.A. 1355. And she confirmed this again in a photo array following the courtroom identification. Her description matched the one given at the scene. Mrs. Bost again pointed him out before the jury, and she identified his jacket and gloves as “identical” to the ones worn by her rapist. J.A. 217–18, 249–50, 1355. The majority and Judge Wynn’s concurrence seek to create pervasive doubt as to Mrs. Bost’s eyewitness identification. Majority Op. 42–44; Concurring Op. 59–70. And we all do well to consider the fallibility of eyewitnesses, particularly within the racial historical context, see Concurring Op. 56, 88–92. And together they make a considerable case that injustice may have occurred. 6 Yet eyewitness testimony often forms the central, 6 Even so, I cannot agree with their characterizations of the evidence. Take, for instance, the majority’s discontent with Mrs. Bost identifying Long as having “yellow skin.” Mrs. Bost’s initial on-scene description to the responding officer the night of the attack was that her attacker was a “black male.” J.A. 1428, 1429 (April 25, 1976 police 122 if not exclusive, evidence in sexual assault cases. Cf. Neil v. Biggers, 409 U.S. 188, 200– 01 (1972); State v. Newman, 302 S.E.2d 174, 179 (N.C. 1983). And our statutorily mandated role in reviewing Long’s conviction is a limited one. Cf. In re Stevens, 956 F.3d 229, 233–34 (4th Cir. 2020) (noting how AEDPA’s limits on federal courts serves to highlight the central role of the state executive to evaluate convictions). We are directed to focus on the withheld evidence, and it is simply not enough that the withheld evidence could have led the jurors to “more credibly question and consider the reliability of Ms. Bost’s identification in the absence of other evidence.” Majority Op. 44. Turning next to the weight of the undisclosed evidence, the state court assessed the forensic reports, Officer Isenhour’s reports, and the medical records. The state court found that this evidence would not have changed the result at trial. Whether they could have aided the defense in some way is beside the point—the reasonable probability that the result at trial would remain unchanged is what matters. See Strickler, 527 U.S. at 291; Agurs, 427 U.S. at 109–10; see also Richter, 562 U.S. at 112 (“The likelihood of a different result must be substantial, not just conceivable.”). And the state court’s determination that it would was not so obviously in error to be beyond any possibility for fairminded reports); see also J.A. 805 (“victim’s initial description to the police . . . [was that her attacker’s] skin color . . . [was] black.”). Only after identifying Long in the courtroom on May 10 did Mrs. Bost describe the coloring of his skin as “light colored, not real dark.” J.A. 1434. Only when pressed for clarification about the pigmentation of his skin at trial, with Long in the courtroom, did Mrs. Bost describe him as more of a “yellow black man,” which she ultimately explained meant that he was not “unusually light,” but rather, just not a very dark black man. J.A. 248; see also J.A. 137 (“[By] yellow looking . . . [I mean that his skin was] [j]ust not totally black. Not like, you know, a real blue black, black man, you know.”); J.A. 201 (“He was light [skinned] . . . I don’t know how to describe [his complexion] except that he was not a real black man.”). 123 disagreement. Richardson, 668 F.3d at 149 (citing Richter, 562 U.S. at 103); see also LeBlanc, 137 S. Ct. at 1728 (noting that “even clear error will not suffice”). I might well have balanced the evidence differently as a juror or as a state-court judge. But the state court’s balancing of the cumulative evidence was not unreasonable. And § 2254(d) thus requires us to respect the state court’s determination. C. Discovery on remand I also object to the majority’s directive that “Petitioner, at his request, shall be afforded on remand the opportunity to conduct further discovery before the district court makes its final determination on the actual innocence question.” Majority Op. 53. While the history of non-disclosure may well counsel in favor of some discovery on remand, that decision—and the details of the type, scope, and areas—should be left in the first instance to the district court. See Rules Governing Section 2254 Cases in the United States District Courts 6(b), 7(a). And in reaching out to decide, sua sponte, an issue not before us, the majority goes too far. * * * The desire to right an apparent wrong is a natural tendency. The majority, in my view, succumbs to this noble urge. And in doing so, I think it oversteps the juridical role. The legislature has prescribed a severely limited review of state-court convictions for the federal courts. And as judges, we must not grab hold of broader legislative or executive powers. See THE FEDERALIST No. 78, at 465–71 (Hamilton) (C. Rossiter ed., 1961). The judicial role requires that we respect AEDPA’s dictates, along with the weighty interests of comity and federalism that law embodies. And it similarly requires the humility 124 to recognize that others may be better positioned to consider contested convictions. See In re Stevens, 956 F.3d at 233–34. Since I think the state court’s decision debatable, I conclude that AEDPA precludes habeas relief. Therefore, I respectfully dissent. 125