Willie James Pye v. Warden, Georgia Diagnostic Prison

50 F.4th 1025

• Court: Court of Appeals for the Eleventh Circuit
• Decided: October 4, 2022
• Docket: 18-12147
• Status: Published

Opinion

USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 1 of 135

[PUBLISH]

In the

United States Court of Appeals For the Eleventh Circuit

____________________

No. 18-12147 ____________________

WILLIE JAMES PYE, Petitioner-Appellant, versus WARDEN, GEORGIA DIAGNOSTIC PRISON,

Respondent-Appellee.

Appeal from the United States District Court for the Northern District of Georgia D.C. Docket No. 3:13-cv-00119-TCB ____________________ USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 2 of 135

2 Opinion of the Court 18-12147

Before WILLIAM PRYOR, Chief Judge, WILSON, JORDAN, ROSENBAUM, JILL PRYOR, NEWSOM, BRANCH, LUCK, LAGOA, and BRASHER, Circuit Judges. * NEWSOM, Circuit Judge, delivered the opinion of the Court, in which WILLIAM PRYOR, Chief Judge, and BRANCH, LUCK, LAGOA, and BRASHER, Circuit Judges, joined. JORDAN, Circuit Judge, filed an opinion concurring in the judg- ment, in which ROSENBAUM, Circuit Judge, joined. JILL PRYOR, Circuit Judge, filed a dissenting opinion, in which WILSON, Circuit Judge, joined. NEWSOM, Circuit Judge: More than a quarter century ago, Willie James Pye was con- victed by a Georgia jury of having kidnapped, robbed, gang-raped, and viciously murdered Alicia Yarbrough. The jury recommended that Pye be sentenced to death for his crimes, and the trial judge so sentenced him. Having exhausted his state post-conviction reme- dies, Pye filed a federal habeas corpus petition, arguing, as relevant here, that his trial counsel rendered him constitutionally ineffective assistance in connection with the sentencing phase of his trial. The district court denied relief, but a panel of this Court reversed and vacated Pye’s death sentence, holding that the state court’s rejec- tion of his ineffective-assistance-of-counsel claim was based on an

* Judge Grant is recused. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 3 of 135

18-12147 Opinion of the Court 3

unreasonable determination of the facts and involved an unreason- able application of clearly established federal law. See Pye v. War- den, Ga. Diagnostic Prison, 853 F. App’x 548, 570–71 (11th Cir.), reh’g en banc granted, 9 F.4th 1372 (11th Cir. 2021); 28 U.S.C. § 2254(d). We granted rehearing en banc to decide whether the state court’s decision that Pye is not entitled to relief on his ineffective- assistance claim warrants deference under the Antiterrorism and Effective Death Penalty Act (AEDPA). Because the state court rea- sonably concluded that Pye was not prejudiced by any of his coun- sel’s alleged deficiencies in connection with his sentencing proceed- ing, we affirm the district court’s denial of Pye’s petition and re- mand to the panel for further proceedings. I A The Georgia Supreme Court’s decision on direct appeal re- counts the grisly facts of Pye’s crimes: Pye had been in a sporadic romantic relation- ship with the victim, Alicia Lynn Yarbrough, but, at the time of her murder, Ms. Yarbrough was living with another man, Charles Puckett. Pye and two companions, Chester Adams and Anthony Freeman, planned to rob Puckett because Pye had heard that Puckett had just collected money from the settlement of a lawsuit. Pye was also angry because Puckett had USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 4 of 135

4 Opinion of the Court 18-12147

signed the birth certificate of a child whom Pye claimed as his own.

The three men drove to Griffin[, Georgia] in Adams’ car and, in a street transaction, Pye bought a large, distinctive .22 pistol. They then went to a party where a witness observed Pye in possession of the large .22. Just before midnight, the three left the party and drove toward Puckett’s house. As they were leaving, a witness heard Pye say, “it’s time, let’s do it.” All of the men put on the ski masks which Pye had brought with him, and Pye and Adams also put on gloves.

They approached Puckett’s house on foot and observed that only Ms. Yarbrough and her baby were home. Pye tried to open a window and Ms. Yar- brough saw him and screamed. Pye ran around to the front door, kicked it in, and held Ms. Yarbrough at gunpoint. After determining that there was no money in the house, they took a ring and a necklace from Ms. Yarbrough and abducted her, leaving the in- fant in the house. The men drove to a nearby motel where Pye rented a room using an alias. In the motel room, the three men took turns raping Ms. Yar- brough at gunpoint. Pye was angry with Ms. Yar- brough and said, “You let Puckett sign my baby’s birth certificate.”

After attempting to eliminate their fingerprints from the motel room, the three men and Ms. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 5 of 135

18-12147 Opinion of the Court 5

Yarbrough left in Adams’ car. Pye whispered in Ad- ams’ ear and Adams turned off onto a dirt road. Pye then ordered Ms. Yarbrough out of the car, made her lie face down, and shot her three times, killing her. As they were driving away, Pye tossed the gloves, masks, and the large .22 from the car. The police later recov- ered these items and found the victim’s body only a few hours after she was killed. A hair found on one of the masks was consistent with the victim’s hair, and a ballistics expert determined that there was a 90% probability that a bullet found in the victim’s body had been fired by the .22. Semen was found in the victim’s body and DNA taken from the semen matched Pye’s DNA. When Pye talked to the police later that day, he stated that he had not seen the vic- tim in at least two weeks. However, Freeman con- fessed and later testified for the State.

Pye v. State, 505 S.E.2d 4, 9–10 (Ga. 1998). Based on the evidence presented, a Georgia jury found Pye guilty of malice murder, kid- napping with bodily injury, armed robbery, burglary, and rape. Attorney Johnny Mostiler represented Pye at both the guilt and penalty phases of his trial. At sentencing, Mostiler—with help from his investigator Dewey Yarbrough, who had no relation to the victim—called eight witnesses to testify on Pye’s behalf: Pye’s sister Pam Bland, sister Sandy Starks, brother Ricky Pye, father Ern- est Pye, niece Chanika Pye, nephew Dantarius Usher, sister-in-law Bridgett Pye, and family friend Lillian Buckner. While Mostiler elicited some testimony about Pye’s impoverished upbringing— USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 6 of 135

6 Opinion of the Court 18-12147

for instance, that his childhood home lacked running water and heat—Pye’s witnesses mainly testified to his good moral character and asked the jury to show mercy by declining to impose a death sentence. The State, meanwhile, presented evidence of Pye’s rep- utation for violence in the community, earlier crimes and alterca- tions with Alicia Yarbrough, and the aggravating circumstances of the murder. The State also argued Pye would pose a danger to prison staff were he to remain incarcerated. The jury recom- mended a death sentence, which the trial court imposed, and the Georgia Supreme Court affirmed. See id. at 14. B Pye filed a petition for post-conviction relief in the Butts County Superior Court. He raised numerous grounds, including, as relevant here, that Mostiler had provided constitutionally inef- fective assistance of counsel during the sentencing phase of his trial by failing to “conduct an adequate pretrial investigation into [Pye’s] life, background, physical and psychiatric health to uncover and present to the jury evidence in mitigation.” Doc. 13-31 at 13; see Strickland v. Washington, 466 U.S. 668, 687 (1984) (holding that an attorney’s performance is constitutionally ineffective when he (1) renders deficient performance (2) that prejudices the defend- ant). The state court conducted a three-day evidentiary hearing. In support of his petition, Pye presented affidavit testimony from 27 witnesses, 24 of whom testified about matters relevant to his ineffective-assistance-at-sentencing claim. Many of these affiants asserted (1) that Pye’s childhood was marked by significant USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 7 of 135

18-12147 Opinion of the Court 7

poverty, abuse, and neglect—mitigating circumstances that, Pye argues, his trial counsel failed to present at sentencing—and (2) that they would have been willing to testify to these facts had they been asked to do so at the time of sentencing but were never contacted by Pye’s trial team. Two corrections officers who had known Pye during an earlier period of incarceration provided affidavit testi- mony that Pye was not a dangerous inmate. Pye also offered testi- mony from mental-health experts that he suffered from frontal- lobe brain damage that impaired his ability to plan and control his impulses—damage, they said, that was potentially caused by fetal alcohol syndrome. The State’s response to Pye’s petition, as relevant here, in- cluded testimony from Dewey Yarbrough. Yarbrough testified that he and Mostiler investigated Pye’s background in preparation for trial but found Pye’s family generally unwilling to cooperate in his defense or to help pursue other leads. The State also called its own mental-health expert, who testified that the facts of Pye’s crime, which involved significant premeditation and planning, weren’t consistent with frontal-lobe impairment or fetal-alcohol syndrome—though he acknowledged that Pye had cognitive defi- cits that would have affected his ability to function in the commu- nity. The Butts County court denied relief on all counts. The court concluded that Mostiler’s performance at sentencing wasn’t constitutionally deficient and that, even if it was, it didn’t prejudice Pye. With respect to evidence of Pye’s childhood of poverty and USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 8 of 135

8 Opinion of the Court 18-12147

abuse, the court concluded that any failure to investigate and pre- sent this evidence wasn’t prejudicial. In so holding, the court em- phasized (1) credibility concerns regarding the affidavit testimony presented at the state post-conviction proceedings; (2) evidence of Pye’s family’s unwillingness to cooperate in his defense at the time of trial; (3) the minimal connection between Pye’s background and the crimes he committed; (4) Pye’s age at the time of his crimes; and (5) the extensive aggravating evidence presented by the State at sentencing. See Doc. 20-40 at 64–67. With respect to Pye’s men- tal-health evidence, the court credited the testimony of the State’s expert that Pye was not as impaired as his witnesses suggested. Id. at 63–64. And with respect to Pye’s evidence of his behavior in prison and lack of future dangerousness, the court concluded that disciplinary reports in Pye’s prison records indicated “a history of insubordination, aggressiveness and propensity for violence to- ward those in authority” that negated any reasonable probability that testimony like that offered by the corrections officers during state post-conviction proceedings would have affected the out- come of sentencing. Id. at 61–62. The Georgia Supreme Court summarily denied Pye a certificate of probable cause to appeal. C Pye filed a petition for a writ of habeas corpus pursuant to 28 U.S.C. § 2254 in the United States District Court for the North- ern District of Georgia. In his amended petition, Pye alleged 16 claims for relief, including a claim that Mostiler provided ineffec- tive assistance during the penalty phase of the trial by failing to USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 9 of 135

18-12147 Opinion of the Court 9

conduct an adequate mitigation investigation. The district court rejected Pye’s ineffective-assistance claim, reviewing the state court’s decision deferentially under AEDPA. As to Strickland’s de- ficient-performance prong, the district court emphasized that Mostiler and Yarbrough had visited Pye’s home more than once and obviously knew about his childhood living conditions. The court held that Pye had failed to “rebut the theory that counsel could have reasonably determined that a strategy of humanizing [Pye], highlighting the fact that [he] did not have a violent reputa- tion, and begging for mercy would be preferred to attempting to provide excuses for [his] crimes because he had led a difficult life.” On prejudice, the district court concluded that while Pye had demonstrated an impoverished upbringing and that there was some evidence of “fighting by and among [his] family members,” he hadn’t “presented evidence that he was subjected to regular and brutal beatings, sexual abuse, or conditions so severe that the state had to step in and remove [him] and his siblings from the home or that his parents were charged with neglect,” as might overcome the aggravating evidence and thus undermine confidence in his sentence. Pye timely appealed. A three-judge panel of this Court reversed the district court and vacated Pye’s death sentence in an unpublished opinion. Pye, 853 F. App’x 548. The panel held that the district court erred in rejecting Pye’s sentencing-phase Strickland claim because the state court’s conclusions as to both deficient performance and prejudice were based on unreasonable factual determinations and involved USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 10 of 135

10 Opinion of the Court 18-12147

unreasonable applications of Strickland—and therefore weren’t en- titled to AEDPA deference. Id. at 563, 567. Engaging in de novo review, the panel held that Mostiler’s performance at sentencing was deficient because he failed (1) to conduct a sufficient investiga- tion into the potentially mitigating circumstances of Pye’s back- ground—specifically, his childhood history of extreme poverty and abuse; (2) to obtain a mental-health evaluation of Pye or otherwise uncover his mental deficiencies; and (3) to attempt to rebut the State’s argument about Pye’s future dangerousness. Id. at 563–65. The panel concluded that these deficiencies were prejudicial not- withstanding the aggravating evidence that the State presented at sentencing, and thus concluded that Pye was entitled to habeas re- lief. Id. at 570–71. The State filed a petition for rehearing en banc, which pre- sented a single issue: whether the panel’s review “of the state court’s determination that the petitioner failed to establish preju- dice at the sentencing phase” conflicts with Eleventh Circuit and Supreme Court precedent. App. Doc. 59 at 1. After this Court voted to rehear the case en banc, we issued a briefing notice that framed the issue somewhat more generally: whether the state court’s decision that Pye’s trial counsel “did not render constitu- tionally ineffective assistance during the penalty phase of trial” was contrary to or an unreasonable application of clearly established federal law or based on an unreasonable determination of the facts. App. Doc. 62 at 1. Pye’s opening en banc brief focused almost ex- clusively on the prejudice prong of Strickland’s two-part test, USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 11 of 135

18-12147 Opinion of the Court 11

noting that because “the [state] did not seek rehearing of” the panel’s deficient-performance holding, it had abandoned that issue. En Banc Br. of Appellant at 70 n.20. The State’s brief in response addressed only prejudice but said, in a footnote, that it had focused on prejudice because “it [was] easier to do so in this case” and that it was “not conceding that trial counsel’s performance was defi- cient.” En Banc Br. of Appellee at 39 n.8. Given the State’s failure to seek rehearing on or brief the merits of the deficient-performance issue, we won’t consider whether the district court erred in holding that the state court’s conclusion as to deficient performance at sentencing was reasona- ble and entitled to deference. Instead, we will assume that Mostiler’s performance was deficient and evaluate only the state court’s conclusion that Pye was not prejudiced by these alleged de- ficiencies. II A We review de novo a district court’s denial of habeas relief on an ineffective-assistance-of-counsel claim, which presents a mixed question of law and fact. See Connor v. Sec’y, Fla. Dep’t of Corr., 713 F.3d 609, 620 (11th Cir. 2013). When a state court has adjudicated a habeas petitioner’s claim on the merits, we review its decision under AEDPA’s “highly deferential” standards. Davis v. Ayala, 576 U.S. 257, 269 (2015). Under those standards, we may not grant the writ unless the state USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 12 of 135

12 Opinion of the Court 18-12147

court’s “adjudication of the claim . . . (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). To meet the “unreasonable application” standard, “a prisoner must show far more than that the state court’s decision was merely wrong or even clear error.” Shinn v. Kayer, 141 S. Ct. 517, 523 (2020) (per curiam) (quotation marks omitted). The decision must be “so obviously wrong that its error lies beyond any possibility for fairminded disagreement.” Id. (quotation marks omitted). When it comes to factual determinations, “[s]tate court fact-findings are entitled to a presumption of correctness unless the petitioner re- buts that presumption by clear and convincing evidence.” Conner v. GDCP Warden, 784 F.3d 752, 761 (11th Cir. 2015); see 28 U.S.C. § 2254(e)(1). Overall, “[a] state court’s determination that a claim lacks merit precludes federal habeas relief so long as fairminded ju- rists could disagree on the correctness of the state court’s decision.” Harrington v. Richter, 562 U.S. 86, 101 (2011) (quotation marks omitted). When the final state court decision on the merits doesn’t come with reasons—as here, where the Georgia Supreme Court summarily denied Pye a certificate of probable cause to appeal the denial of his habeas petition—the federal court must “‘look through’ the unexplained decision to the last related state-court USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 13 of 135

18-12147 Opinion of the Court 13

decision that does provide a relevant rationale” and “presume that the unexplained decision adopted the same reasoning.” Wilson v. Sellers, 138 S. Ct. 1188, 1192 (2018). B Before diving into the merits, we pause to clarify three points about AEDPA’s standard of review. 1 1 First, despite some lingering confusion—including among the parties here—it’s not (any longer) the law that a federal court should decline to defer to a state court’s factual determinations if it concludes that those findings “lacked . . . fair support in the record.” Rose v. McNeil, 634 F.3d 1224, 1241 (11th Cir. 2011) (quotation omitted). Unlike this case, Rose involved a habeas petition filed “before the effective date of [AEDPA],” so it applied “pre-AEDPA law.” Id. at 1240. The pre-AEDPA version of the federal habeas statute “provided that factual findings of a state court were pre- sumed to be correct unless ‘the Federal court on consideration of the record as a whole concludes that such factual determination is not fairly supported by the record.’” Fugate v. Head, 261 F.3d 1206,

1 It seems clear enough that our dissenting colleagues don’t much like AEDPA,

whose “abstruse language,” they say, leaves much to “imagination and rumi- nation.” Dissenting Op. at 12. Not to put too fine a point on it, but the statute is what it is and says what it says. Congress passed it, and President Clinton signed it. See U.S. Const. art. I, § 7. It is now our job to apply it according to its terms. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 14 of 135

14 Opinion of the Court 18-12147

1215 n.11 (11th Cir. 2001) (emphasis added) (quoting 28 U.S.C. § 2254(d)(8) (1994)). In AEDPA, Congress eliminated and replaced the fair-support-in-the-record standard. Under the amended stat- ute, a state court’s factual determinations are “presumed to be cor- rect,” and the petitioner has the burden of proving otherwise “by clear and convincing evidence.” 28 U.S.C. § 2254(e)(1). 2 Second, even if a petitioner successfully carries his burden under § 2254(e)(1)—showing by clear and convincing evidence that a particular state-court factual determination was wrong—he does not necessarily meet his burden under § 2254(d)(2): Even if the state court made a clearly erroneous factual determination, that doesn’t necessarily mean the state court’s “decision” was “based on” an “unreasonable determination of the facts in light of the evi- dence presented in the State court proceeding.” Id. § 2254(d)(2). Depending on the importance of the factual error to the state court’s ultimate “decision,” that decision might still be reasonable “even if some of the state court’s individual factual findings were erroneous—so long as the decision, taken as a whole, doesn’t con- stitute an ‘unreasonable determination of the facts’ and isn’t ‘based on’ any such determination.” Hayes v. Sec’y, Fla. Dep’t of Corr., 10 F.4th 1203, 1224–25 (11th Cir. 2021) (Newsom, J., concurring) (quoting 28 U.S.C. § 2254(d)(2)); see also Miller-El v. Cockrell, 537 USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 15 of 135

18-12147 Opinion of the Court 15

U.S. 322, 341 (2003) (noting that subsections (e)(1) and (d)(2) are “independent requirements”). 2 3 Third, although the Supreme Court’s decision in Wilson in- structs us to “review[] the specific reasons given by the state court and defer[] to those reasons if they are reasonable,” 138 S. Ct. at 1192, we are not required, in assessing the reasonableness of a state

2 The dissent vehemently objects to our explanation of the “interplay” be- tween §§ 2254(d)(2) and (e)(1)—not, to be clear, to the merits of our explana- tion, but to the fact that we have offered it at all. Calling our decision to ad- dress those sections’ relationship “irregular[],” “wrong,” and “odd at best,” the dissent equates our discussion to a “‘takeover of the appeal.’” Dissenting Op. at 33 (quoting United States v. Sineneng-Smith, 140 S. Ct. 1575, 1579, 1581 (2020)); see also Concurring Op. at 1 (citing Sineneng-Smith, 140 S. Ct. at 1579–80). We respectfully disagree. This case doesn’t remotely present the situation that the Supreme Court confronted in Sineneng-Smith or that this Court recently debated in United States v. Campbell, 26 F.4th 860 (11th Cir. 2022) (en banc). Those cases considered the questions whether and under what circumstances it is appropriate for an appellate court to consider sua sponte a discrete legal issue, claim, or defense that the parties haven’t squarely presented. Here, by contrast, in elaborating on the relationship between §§ 2254(d)(2) and (e)(1), we are simply explaining how two adjacent statutory provisions interact—in short, how the law works. And of course, once “an issue or claim is properly before the court”—as Pye’s entitlement to relief un- der 28 U.S.C. § 2254 plainly is—“the court is not limited to the particular legal theories advanced by the parties, but rather retains the independent power to identify and apply the proper construction of governing law.” Kamen v. Kem- per Fin. Servs., 500 U.S. 90, 99 (1991); cf. Babb v. Sec’y, Dep’t of Veterans Affs., 992 F.3d 1193, 1208 & n.9 (11th Cir. 2021) (doubting that “the correct stand- ard” is an issue subject to ordinary forfeiture rules). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 16 of 135

16 Opinion of the Court 18-12147

court’s reasons for its decision, to strictly limit our review to the particular justifications that the state court provided. Rather, in or- der to “give appropriate deference to [the state court’s] decision,” id., having determined the reasons for the state court’s decision, we may consider any potential justification for those reasons. If, as here, the “specific reason[]” for a state court’s decision to deny ha- beas relief was that the petitioner wasn’t prejudiced by his counsel’s deficient performance, we can, in evaluating whether that “reason [was] reasonable,” id., consider additional rationales that support the state court’s prejudice determination. We have so held repeat- edly, both before and since Wilson. See, e.g., Whatley v. Warden, Ga. Diagnostic & Classification Ctr., 927 F.3d 1150, 1178 (11th Cir. 2019) (“[O]ur review is not limited to the reasons the [state court] gave in its analysis.”), reh’g en banc denied, 955 F.3d 924 (11th Cir. 2020); id. at 1182 (“[W]e are not limited to the reasons the [state court] gave and instead focus on its ‘ultimate conclusion.’” (citation omitted)); Meders v. Warden, Ga. Diagnostic Prison, 911 F.3d 1335, 1350 (11th Cir. 2019) (“We have explicitly rejected the prop- osition that a state court decision involves an unreasonable appli- cation of federal law and is not entitled to deference unless that court’s opinion on its face ‘shows its work’ by explicitly mentioning ‘all relevant circumstances’ that the defendant argues in support of relief.” (citation omitted)); Lee v. Comm’r, Ala. Dep’t of Corr., 726 F.3d 1172, 1223 (11th Cir. 2013) (“Under Supreme Court and our Circuit precedent, a state court’s written opinion is not required to mention every relevant fact or argument in order for AEDPA def- erence to apply. . . . [W]e still examine what other ‘implicit USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 17 of 135

18-12147 Opinion of the Court 17

findings’ the state court could have made in its denial of a federal claim.”); cf. Butts v. GDCP Warden, 850 F.3d 1201, 1232 (11th Cir. 2017) (deferring to a state court’s “ultimate conclusion” under AEDPA despite the court’s “unreasonable finding” regarding what happened at the sentencing hearing). Both Pye and our dissenting colleagues assert that Wilson prohibits us from considering justifications that support the reasons underlying the state court’s decision but that, for whatever reason, the state court didn’t explicitly memorialize in its written opinion. For reasons we will explain, we disagree. Given the vigor with which the dissent presses its Wilson- based argument—and the fact that it is, for all practical purposes, the lone basis on which the dissent sidesteps AEDPA deference— it’s worth explaining our position in some detail. The dissent stren- uously—and stridently—insists that because the Butts County Su- perior Court issued a written opinion, we are obliged by Wilson to limit our review not just to the “reasons” for that court’s decision— as relevant here, that Pye wasn’t prejudiced by Mostiler’s allegedly deficient performance—but also, at an even more granular level, to the particular justifications that the court provided to support those reasons. Indeed, the dissent goes so far as to assert that our contrary view is a “[n]onsense” “gambit” that “nullifi[es]” Wilson. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 18 of 135

18 Opinion of the Court 18-12147

Dissenting Op. at 16, 18. Respectfully, we don’t think so. Here’s why. 3 With a promise to return to Wilson in short order, we begin with first principles. Put simply, our approach is the one that AEDPA’s plain language requires. With respect to “any claim that was adjudicated on the merits in State court”—which Pye’s ineffec- tive-assistance claim indisputably was—the statute focuses exclu- sively on the reasonableness of the state court’s “decision.” In par- ticular, it states that a petitioner “shall not be granted” relief unless the state court’s “adjudication of the claim . . . resulted in a deci- sion” that was either contrary to or involved an unreasonable ap- plication of federal law or was based on an unreasonable determi- nation of the facts. 28 U.S.C. § 2254(d)(1)–(2) (emphasis added). AEDPA’s text couldn’t be any clearer: A federal habeas court is

3 A brief preface:The dissent nitpicks our terminology, deploying dictionaries as if we were engaged in statutory or contractual interpretation. See Dissent- ing Op. at 17. But, of course, everyone recognizes the difference between macro-level reasons and their constituent rationales—what we’ve called justi- fications. And to be clear, that distinction is hardly “nonsense.” Id. at 18. The law is shot through with similar gradations, and the fact that lines can be tough to draw doesn’t eliminate our responsibility to draw them. See, e.g., Yee v. City of Escondido, 503 U.S. 519, 534 (1992) (distinguishing between “claim[s]” and their constituent “argument[s]”); Kamen v. Kemper Fin. Servs., Inc., 500 U.S. 90, 99 (1991) (distinguishing between “issue[s]” and constituent “theo- ries”). What the dissent never does is convincingly explain why a federal court should be limited on habeas review to the precise explanation offered by the state court—in every jot and tittle, right down to the last syllable. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 19 of 135

18-12147 Opinion of the Court 19

tasked with reviewing only the state court’s “result[ing] deci- sion”—not the constituent justifications for that decision. Consistent with the statutory text, the Supreme Court unan- imously held in Harrington v. Richter, 562 U.S. 86, that a state court’s decision rejecting a petitioner’s post-conviction claim is en- titled to AEDPA deference even “when state-court relief is denied without an accompanying statement of reasons”—e.g., in a “one- sentence summary order.” Id. at 92, 96. 4 Notably, in so holding, the Court emphasized § 2254(d)’s “terms”: “There is no text in the statute requiring a statement of reasons. The statute refers only to a ‘decision,’ which resulted from an ‘adjudication.’” Id. at 98. And “determining whether a state court’s decision resulted from an un- reasonable legal or factual conclusion does not require that there be an opinion from the state court explaining the state court’s rea- soning.” Id. Even “[w]here a state court’s decision is unaccompa- nied by an[y] explanation” at all, the Court explained, “the habeas petitioner’s burden still must be met by showing there was no rea- sonable basis for the state court to deny relief.” Id. “AEDPA de- mands,” the Court concluded, that the federal court determine what “arguments or theories” either “supported or . . . could have

4Justice Ginsburg concurred in the judgment. See 562 U.S. at 113–14 (Gins- burg, J., concurring in the judgment). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 20 of 135

20 Opinion of the Court 18-12147

supported . . . the state court’s decision.” Id. at 102 (emphasis added). 5 Consistent with both AEDPA’s plain language and the logic of the Supreme Court’s decision in Richter, this Court has long (and consistently) held that where, as here, a state court rejects a petitioner’s claim in a written opinion accompanied by an explana- tion, the federal habeas court reviews only the state court’s “deci- sion” and is not limited to the particular justifications that the state court supplied. See, e.g., Parker v. Sec’y for Dep’t of Corr., 331 F.3d 764, 785 (11th Cir. 2003) (“[W]e review the state court’s ‘deci- sion’ and not necessarily its rationale.”); Gill v. Mecusker, 633 F.3d 1272, 1288–93 (11th Cir. 2011) (“[T]he statutory language focuses on the result, not on the reasoning that led to the result.”); What- ley, 927 F.3d at 1178 (“[O]ur review is not limited to the reasons the Court gave in its analysis.”). Our decisions in that respect are part and parcel of our recognition that “overemphasis on the lan- guage of a state court’s rationale would lead to a grading papers approach that is outmoded in the post-AEDPA era.” Jones v. Sec’y, Fla. Dep’t of Corr., 834 F.3d 1299, 1311 (11th Cir. 2016) (quotation

5 This language from the Supreme Court’s unanimous decision in Richter— from which the Court didn’t recede in Wilson—constitutes a full answer to the dissent’s charge that we have somehow “miss[ed] half of the text and con- text” of § 2254(d). Dissenting Op. at 20. It is telling—and more than a little ironic—that the dissent simultaneously criticizes us both for engaging § 2254(d)’s full text—including its “based on” clause—and for ignoring it. Com- pare id. at 20 with id. at 32–33. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 21 of 135

18-12147 Opinion of the Court 21

marks omitted); see also Meders, 911 F.3d at 1350 (noting that “a line-by-line critique of the state court’s reasoning” is “not the proper approach”). And to be clear, ours is hardly an outlier view; rather, it rep- resents the overwhelming consensus position. Surveying courts across the country, the Fifth Circuit recently summarized that “most of the courts of appeals” have held that even where a state court rejects a petitioner’s claim in a reasoned decision, the federal “habeas court must defer to a state court’s ultimate ruling rather than to its specific reasoning.” Sheppard v. Davis, 967 F.3d 458, 467 n.5 (5th Cir. 2020) (collecting cases so holding from the First, Sec- ond, Sixth, Seventh, Eighth, Tenth, and Eleventh Circuits); accord, e.g., Holland v. Rivard, 800 F.3d 224, 236–37 (6th Cir. 2015) (“[I]t is the decision of the state court, not its reasoning, to which AEDPA deference applies.”). And indeed, given Richter, that’s the only rule that makes any sense at all. On the dissent’s view, if a state court offers no explanation whatsoever for its decision to deny a petitioner relief, its decision is, per Richter, entitled to full AEDPA deference, and the federal court should indulge any “argument[] or theor[y]” that “could have supported” that decision. 562 U.S. at 102. But, the dissent insists, as soon as the state court gives it the old college try and writes an opinion, the federal court is stuck, so to speak, with the specific justifications articulated therein. As this Court recently explained, such a rule “would be irrational.” Meders, 911 F.3d at 1351 (observing that “[i]t would be irrational to afford deference to USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 22 of 135

22 Opinion of the Court 18-12147

a decision with no stated explanation”—as Richter clearly re- quires—“but not afford deference to one that states reasons, albeit not as thoroughly as it could have”). The “irrational[ity]” of the dissent’s position exists on at least two levels. First, and most ob- viously—and most perversely—it would incentivize state courts to issue unreasoned, summary decisions as a means of guaranteeing maximum AEDPA deference. Second, as a sister circuit has ex- plained, any state court’s written opinion is necessarily “partial”; it will never perfectly and exhaustively capture every justification un- derlying the court’s decision: [E]ven if we assume that deference to the state court’s decision is warranted only when there is some possi- bility that the court specifically contemplated “rea- sonable” grounds for denying relief, the issuance of a written opinion with deficient reasoning does not eliminate such a possibility. Just as there is more than one way to skin a cat, there often is more than one way to resolve an appeal, and not every possible ap- proach makes it into an opinion.

Williams v. Roper, 695 F.3d 825, 837 (8th Cir. 2012). The lone question, then, is whether the Supreme Court’s de- cision in Wilson instituted an entirely new and different AEDPA regime, whereby a federal court reviewing a state court’s (neces- sarily partially) reasoned decision is strictly limited to the particular justifications that the state court memorialized in its written opin- ion. The dissent insists that it did; Wilson, the dissent says, is USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 23 of 135

18-12147 Opinion of the Court 23

“materially indistinct” from, and thus “controls,” this case. Dis- senting Op. at 13–15. But the dissent overreads (and thus misreads) Wilson. As the Wilson Court itself acknowledged—and as the dis- sent here does not dispute—it confronted only a very narrow ques- tion: “[T]he issue before [it],” the Court said, was solely “whether to ‘look through’ [a] silent state higher court opinion to the rea- soned opinion of a lower court” in applying AEDPA deference. 138 S. Ct. at 1195. To be sure, the Court answered that specific ques- tion “yes.” But just as surely, there is no indication that, in so do- ing, the Court silently upended the existing AEDPA standard as it applies to reasoned state-court decisions, thwarted § 2254(d)’s plain language, and abrogated a nationwide circuit consensus. 6

6 Perhaps in an effort to camouflage the breadth of its position, the dissent repeatedly emphasizes the similarities between the procedural postures of this case and Wilson—in both, the dissent reminds us, a reasoned lower-court opinion was summarily affirmed on appeal. See Dissenting Op. at 14, 19 n.16, 22–23 n.19. And to be clear, contrary to the dissent’s suggestion, see id. at 14– 16, we have followed Wilson’s holding to a T, “look[ing] through” the Georgia Supreme Court’s summary order to the Butts County Superior Court’s deci- sion. See supra at 12–13. But the dissent insists—wrongly—that Wilson does so much more. Disregarding both the Wilson Court’s own specification of the narrow issue before it, see 138 S. Ct. at 1195, and our subsequent reaffirmation that Wilson dealt only with the question of “which state court decision we are to look at if the lower state court gives reasons and the higher state court does not,” Meders, 911 F.3d at 1350, the dissent insists that Wilson changed how AEDPA applies to all reasoned decisions, regardless of procedural posture. That’s the only way to make sense of the dissent’s criticism of our post-Wilson decision in Whatley, which the dissent admits “arose in a different procedural posture” USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 24 of 135

24 Opinion of the Court 18-12147

To support its position, the dissent points to a passage in Wilson in which it says the Supreme Court “h[eld] that AEDPA ‘requires’ a federal habeas court to look to the last reasoned state court decision and then ‘train its attention on the particular rea- sons—both legal and factual—why state courts rejected a state pris- oner’s federal claims.’” Dissenting Op. at 14–15 (quoting Wilson, 138 S. Ct. at 1191–92, 1195–96 (quoting, in turn, Hittson v. Chat- man, 135 S. Ct. 2126, 2126 (2015) (Ginsburg, J., concurring in denial of certiorari))). But the dissent’s quotation stops short. Notably, the Supreme Court went on, in language that the dissent omits, to clarify that, having divined the “reasons” for the state courts’ deci- sion, the federal court should “give appropriate deference to that decision.” Wilson, 138 S. Ct. at 1192. Notably as well, for the latter proposition the Court cited—as binding and with approval—its earlier decision in Richter. See id. The dissent’s confident contention to the contrary notwith- standing, there is simply nothing in Wilson that clearly confines a

but which it nonetheless says “conflicted with Wilson.” Dissenting Op. at 22– 23 n.19. And it’s the only way to make sense of the dissent’s criticism of our citation to “pre-Wilson” cases that didn’t involve summary affirmances, see id. at 26 & n.21, or its own reliance on post-Wilson cases that didn’t involve summary affirmances, see id. at 26–30. Bottom line: To the dissent, Wilson isn’t remotely limited to the “look through” issue that the Supreme Court said it was tackling; rather, Wilson sub silentio revolutionized AEDPA’s application to all state-court decisions. We simply—but vehemently—disagree. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 25 of 135

18-12147 Opinion of the Court 25

federal habeas court to the precise justifications that a state court provides in its written opinion. 7 To the extent there is any doubt about that, our own post-Wilson precedent resolves it. For start- ers, we recently rejected the very misreading of Wilson that today’s dissent advocates: Wilson was about which state court decision we are to look at if the lower state court gives reasons and the higher state court does not. It was not about the specificity or thoroughness with which state courts must spell out their reasoning to be entitled to AEDPA deference or the level of scrutiny that we are to apply to the reasons that they give.

Meders, 911 F.3d at 1350. 8 And even more recently, we held—and then declined to reconsider, over the exact same Wilson-based

7The dissent also overreads Wilson’s one-paragraph discussion of Premo v. Moore, 562 U.S. 115 (2011), to mean that, having “look[ed] through” a state supreme court’s summary order and presumed that it adopted the lower state court’s reasoning, “we must focus exclusively on the reasons actually given.” Dissenting Op. at 31; see also id. at 14. But nothing in that paragraph is incon- sistent with our approach here—or says that we must do anything. State-court reasoning is usually reasonable—and considering only a state court’s reasons before determining its decision to be reasonable proves little. Considering only those reasons before determining the decision to be unreasonable— which is what the dissent advocates here—would be a different thing alto- gether. 8The dissent suggests that Meders held that when a state court’s justifications are unreasonable, we must refuse AEDPA deference. See Dissenting Op. at 22–23 n.19. Put simply, we just don’t see in that decision what our dissenting colleagues do. Moreover, and in any event, in Meders we deferred under USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 26 of 135

26 Opinion of the Court 18-12147

objection that today’s dissent recycles—that a federal habeas court’s “review is not limited to the reasons the [state court] gave in its analysis.” Whatley, 927 F.3d at 1178 (“[O]ur review is not limited to the reasons the [state court] gave in its analysis.”), reh’g en banc denied, 955 F.3d 924. And for what it’s worth, others share our skepticism about the dissent’s ambitious understanding of Wil- son’s reach. See Sheppard, 967 F.3d at 467 n.5 (“[I]t is far from cer- tain that Wilson overruled sub silentio the position—held by most of the courts of appeals—that a habeas court must defer to a state court’s ultimate ruling rather than to its specific reasoning.”); Thompson v. Skipper, 981 F.3d 476, 483–84 (6th Cir. 2020) (Nal- bandian, J., concurring) (“Federal courts have never been required to confine their habeas analysis to the exact reasoning that the state court wrote, and [nothing in] Wilson v. Sellers . . . compels us to change our analysis.” (citation omitted)); id. at 484 (“[N]othing in Wilson suggests that federal courts cannot look to any other reason for supporting the state court[’s] decision and applying AEDPA def- erence.”). 9

AEDPA, see 911 F.3d at 1355, so any suggestion that our decision there created binding “law” precluding deference is a non-starter. 9 The dissent insists that several of our sister circuits have “refined their ap- proach” in the wake of Wilson and that, in fact, the “great weight of authority” is on its side. Dissenting Op. at 26–30. With respect, the dissent has overread those courts’ decisions in the same way that it has overread Wilson itself. Be- hind the cited cases’ rote quotations of Wilson’s language, one finds important nuance that the dissent overlooks. In three of the dissent’s cases, the courts expressly did not confine themselves to the particular justifications proffered USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 27 of 135

18-12147 Opinion of the Court 27

by the state courts whose decisions they were reviewing, but rather consid- ered others—the very thing the dissent insists Wilson forbids. In Coleman v. Bradshaw, 974 F.3d 710 (6th Cir. 2020), for instance, the Sixth Circuit did not, as Judge Nalbandian has explained, “constrain its analysis to the exact reasons that the state court discussed.” Thompson v. Skipper, 981 F.3d 476, 485 (6th Cir. 2020) (Nalbandian, J., concurring). So too, in Porter v. Coyne-Fague, the First Circuit, after noting that the state court had failed to cite or discuss a key fact, did not—as the dissent here claims Wilson requires—proceed straight to de novo review, but rather first considered whether there was another “possi- ble explanation of the state court’s decision.” 35 F.4th 68, 79 (1st Cir. 2022). And in Scrimo v. Lee, the Second Circuit, after determining that “it was error to exclude [certain w]itnesses’ testimony for [the state court’s] reason,” went on to ask “whether the [w]itnesses’ testimony could have been excluded on other grounds.” 935 F.3d 103, 116 (2d Cir. 2019). The dissent’s Fourth and Seventh Circuit citations are similarly una- vailing. In holding that a state court’s decision was reasonable, the Fourth Circuit in Richardson v. Kornegay adopted pretty much exactly the approach that we’ve outlined here: It defined the state court’s “particular reason” at a relatively high level of generality—namely, that “the trial court did not abuse its discretion”—and then distinguished that reason from its underlying “ra- tionale,” which it said “support[ed] finding no abuse of discretion.” 3 F.4th 687, 697–98 (4th Cir. 2021) (quotation omitted). And in Winfield v. Dorethy, the Seventh Circuit declined to decide exactly how § 2254(d) applied, instead refusing relief on § 2254(a) grounds. See 956 F.3d 442, 455 (7th Cir. 2020). So, after careful review of the dissent’s two-page string cite, it turns out that just one circuit—the Ninth—has employed its sweeping rule. See Kipp v. Davis, 971 F.3d 939, 948–60 (9th Cir. 2020). But the Ninth Circuit had limited federal habeas courts’ review to the state courts’ specific justifications long before Wilson was decided and has done so on a different theory than the dissent proffers here—namely, on the ground that certain factual determina- tions rendered the “fact-finding process itself . . . defective” rather than that “the resulting finding[s]” were themselves substantively unreasonable. See id. at 953–55 (applying its earlier decision in Taylor v. Maddox, 366 F.3d 992, 1008 USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 28 of 135

28 Opinion of the Court 18-12147

Absent a clearer indication that Wilson “meant to strike the widespread method of applying AEDPA without even[] mention- ing the overhaul that would result,” id., we decline to read the Su- preme Court’s decision as aggressively as the dissent does. C Back to this case: AEDPA’s deferential standard of review governs the state court’s application of Strickland, which itself places a demanding burden on a convicted defendant to show that he was prejudiced by his counsel’s deficient performance. “In the capital sentencing context, the prejudice inquiry asks whether there is a reasonable probability that, absent the errors, the sen- tencer . . . would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.” Shinn, 141 S. Ct. at 522–23 (quotation marks omitted). “A reasonable proba- bility means a substantial, not just conceivable, likelihood of a dif- ferent result.” 10 Id. at 523 (quotation marks omitted). And this

(9th Cir. 2004)). Even before Wilson, we had noted the circuit split over whether “state court’s fact-finding procedures” can render its decision unrea- sonable under AEDPA, and we declined to adopt the Ninth Circuit’s approach in Taylor. See Landers v. Warden, 776 F.3d 1288, 1298 (11th Cir. 2015). No circuit, so far as we can tell, shares the dissent’s view that Wilson somehow changed the way AEDPA applies to reasoned state-court decisions. 10Here, because Georgia law requires jury unanimity as a prerequisite to the imposition of capital punishment, prejudice requires a substantial likelihood that at least one juror would have voted against the death penalty. See O.C.G.A. § 17-10-31(c). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 29 of 135

18-12147 Opinion of the Court 29

probability must be sufficient for the reviewing court to determine that counsel’s errors were “so serious as to deprive the defendant of a fair trial . . . whose result is reliable.” Strickland, 466 U.S. at 487. Applying AEDPA to Strickland’s prejudice standard, we must decide whether the state court’s conclusion that Mostiler’s performance at the sentencing phase of Pye’s trial didn’t prejudice him—that there was no “substantial likelihood” of a different re- sult—was “so obviously wrong that its error lies beyond any possi- bility for fairminded disagreement.” Shinn, 141 S. Ct. at 523–24 (quotation marks omitted). So without respect to what we might (or might not) have concluded about prejudice were we free to re- view that issue de novo, we lack the power to grant relief so long as the state court’s conclusion wasn’t that wrong.11 III The state court’s conclusion that Pye wasn’t prejudiced by any of Mostiler’s alleged deficiencies was not “contrary to” and did not “involve[] an unreasonable application of, clearly established Federal law,” nor was it “based on an unreasonable determination of the facts.” 28 U.S.C. § 2254(d). While the state court might have made some debatable calls as to the weight that it ascribed to dif- ferent pieces of evidence—and made at least one dubious factual

11 Moreover, “because the Strickland standard is a general standard, a state court has even more latitude to reasonably determine that a defendant has not satisfied that standard.” Knowles v. Mirzayance, 556 U.S. 111, 123 (2009). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 30 of 135

30 Opinion of the Court 18-12147

statement—its ultimate decision to deny relief was not “so obvi- ously wrong that its error lies beyond any possibility for fairminded disagreement.” Shinn, 141 S. Ct. at 523 (quotation marks omitted). To assess whether an allegedly deficient aspect of a lawyer’s performance was prejudicial, courts must “consider the totality of the available mitigation evidence—both that adduced at trial, and the evidence adduced in the habeas proceeding—and reweigh it against the evidence in aggravation.” Porter v. McCollum, 558 U.S. 30, 41 (2009) (alteration adopted) (quotation marks omitted). Here, Pye alleges multiple grounds on which Mostiler’s performance was deficient: His failure to (a) reasonably investigate mitigation evi- dence concerning Pye’s background, (b) discover evidence of Pye’s mental-health issues, and (c) attempt to rebut the State’s future- dangerousness argument by presenting evidence of Pye’s nonvio- lent behavior in prison. Before us, Pye also argues (d) that he was prejudiced by Mostiler’s deficient failure to introduce at sentencing residual-doubt evidence supporting his version of the events sur- rounding his crimes.12 Accordingly, the state court’s conclusion

12 The district court did not discuss the residual-doubt issue as it pertained to the sentencing phase of the trial—presumably because Pye framed his argu- ment that Mostiler should have presented more evidence supporting his ver- sion of events primarily in terms of Mostiler’s deficient performance during the guilt phase. See Doc. 1 at 68–72; Corrected Initial Br. of Appellant at 161– 72. Still, we will assume for argument’s sake that Mostiler’s performance was deficient on this ground—that he should have introduced additional residual- doubt evidence at either the guilt or sentencing phase of trial—and explain why it still wasn’t unreasonable for the state court to conclude there is “simply USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 31 of 135

18-12147 Opinion of the Court 31

that there was no sentencing-phase prejudice is reasonable and en- titled to deference if its prejudice determinations with respect to each alleged deficiency, and with respect to the deficiencies cumu- latively, were reasonable. They were. A It was reasonable for the state court to conclude that Mostiler’s failure to further investigate Pye’s difficult childhood and present this mitigating evidence at sentencing wasn’t prejudi- cial. The court’s prejudice determination with respect to this defi- ciency was based on (1) its decision to discount the affidavit evi- dence presented at the state post-conviction proceedings due to concerns about their credibility; (2) evidence of Pye’s family’s un- willingness to cooperate in his defense at the time of trial; (3) the minimal connection between Pye’s background and the crimes he committed; (4) Pye’s age at the time of those crimes; and (5) the extensive aggravating evidence presented by the State at sentenc- ing. Neither the court’s weighing of these factors nor its ultimate prejudice determination was contrary to or based on an unreason- able application of federal law, or based on an unreasonable deter- mination of the facts.

no reasonable probability that the result of the penalty phase . . . would have been different if the new evidence had been submitted at trial.” Doc. 20-40 at 67. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 32 of 135

32 Opinion of the Court 18-12147

1 First, the discounting of the affidavits. The state court ex- pressed both (a) concerns about the credibility of the affidavits gen- erally and (b) specific concerns based on inconsistencies that it iden- tified in the affidavits of Curtis Pye, Ricky Pye, Lolla Mae Pye, and Arthur Lawson. While the court’s determination that these affida- vits contained inconsistencies might have been debatable, it wasn’t “clear[ly] and convincing[ly]” erroneous. 28 U.S.C. § 2254(e)(1). Nor—after giving effect to the presumptive correctness of the state court’s assessment of the affidavits’ inconsistencies, see id.—was it unreasonable for the state court to view the “affidavit evidence al- leging abuse and deprivation,” taken as a whole, “with caution.” Doc. 20-40 at 66. The state court read the affidavits of Curtis, Ricky, and Lolla Mae Pye as stating that Mostiler didn’t speak with them at all be- fore trial—whereas contemporaneous billing records show that Mostiler in fact met with all three. Pye argues that these affidavits are best read to assert only that Mostiler didn’t talk to the affiants about Pye’s childhood. So, the argument goes, the affidavits weren’t inconsistent with the record, the presumption is rebutted, and it was thus unreasonable for the state court to discredit them. We disagree. However debatable, the state court’s interpretation of these affidavits was not clearly and convincingly erroneous. Starting with Pye’s brother Curtis, he testified in his affidavit about his family’s impoverished circumstances and his parents’ al- coholism and fighting, and concluded his affidavit as follows: USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 33 of 135

18-12147 Opinion of the Court 33

No one talked to me about any of this before Willie James’s trial. Johnny Mostiler and his assistant Dewey know me. Mr[.] Mostiler represented me be- fore. He didn’t get in touch with me or ask me any questions about the house Willie James was raised in or what he was like as a child. If he had, I would have said all the things I’ve said in this statement, and I would have testified to all these things if he had asked me to.

Doc. 16-24 at 83. The state court’s quotation of this passage used ellipses in a way that made it seem like Curtis definitively stated that Mostiler didn’t speak to him at all before trial: “No one talked to me . . . before Petitioner’s trial. Johnny Mostiler and his assistant Dewey know me . . . He didn’t get in touch with me.” Doc. 20-40 at 65 (omissions in original). But without regard to whether the state court’s interpretation of this affidavit was the most natural, it wasn’t clearly and convincingly erroneous. That is particularly so given the fact that Curtis stated, disjunctively, that Mostiler didn’t either “get in touch with me” or “ask me any questions about the house.” Moreover, if Curtis meant to convey only that Mostiler didn’t talk to him about Willie’s childhood, it’s unclear why he would have said that “Johnny Mostiler and his assistant Dewey know me”—a statement that would make most sense if Mostiler had failed to contact Curtis at all. Given this entirely plausible in- terpretation of Curtis’s affidavit—with which, tellingly, the dissent doesn’t squarely contend, see Dissenting Op. at 39–40—the state USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 34 of 135

34 Opinion of the Court 18-12147

court’s factual finding that Curtis’s affidavit contradicted the record evidence was not clearly and convincingly erroneous. Turning, then, to Ricky Pye’s affidavit. It stated, in pertinent part, that— The investigator, a man named Dewey, came by the house and talked to my dad about the charges against Willie. He didn’t ask about Willie James and how he came up, or how we all were raised. Dewey never spoke to me about those things. . . . I took the stand to testify later on in the trial. No one talked to me about my testimony before I went. I never spoke to Mr. Mostiler about what to say, and he didn’t meet with me or ask me any questions before my turn for testimony. I knew who he was because he repre- sented me before that.”

Doc. 16-24 at 99–100. Whether or not the best reading, the state court’s interpretation of Ricky’s affidavit as suggesting that Mostiler didn’t speak to him at all before his testimony wasn’t clearly and convincingly erroneous. Ricky said that Dewey talked to “[his] dad”—not him—and his statement that he knew who Mostiler was because Mostiler had represented him before suggests that he didn’t meet Mostiler again in the context of preparing for Willie’s trial. Moreover, Ricky’s statement that “[n]o one talked to me about my testimony before I went” suggests—or at the very least could reasonably be read to suggest—that he didn’t speak to Mostiler at all before his testimony. (Again, the dissent offers no response to this plausible interpretation of Ricky’s affidavit. See USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 35 of 135

18-12147 Opinion of the Court 35

Dissenting Op. at 40.) So, given the fact that Mostiler did speak to Ricky for an hour about a month before trial, the state court’s find- ing that Ricky’s affidavit testimony contradicted the record evi- dence was not clearly and convincingly erroneous. Next, Pye’s mother, Lolla Mae. As relevant here, she testi- fied by affidavit as follows: “No one took the time to talk to me about all anything before Willie’s trial. Nobody ask me all about how I grew up, how I came to be married to Ernest, and how I raised Willie and my other children.” Doc. 16-24 at 97. The state court read this statement as suggesting that no one from Willie’s trial team spoke to her at all—contradicting Dewey Yarbrough’s testimony and Mostiler’s billing records that reflect their meeting with her. While the phrase “all anything” is unusual and could re- flect a typographical error, the most natural correction of such an error would simply be to remove the word “all.” That would leave the statement, “No one took the time to talk to me about anything before Willie’s trial.” And if “all anything” wasn’t a typo but just a nonstandard turn of phrase, then the statement could plausibly be interpreted to suggest that Mostiler and Yarbrough didn’t speak to Lolla Mae at all. So again, the state court’s interpretation of Lolla Mae’s statement and its resulting finding of an inconsistency was not clearly and convincingly erroneous. Finally, the affidavit of social worker Arthur Lawson. The state court questioned Lawson’s credibility because, while he ini- tially testified that he had observed Lolla Mae intoxicated when she was pregnant, he later submitted another affidavit to “clarify [this] USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 36 of 135

36 Opinion of the Court 18-12147

inaccurate statement” with the explanation that he had “no direct knowledge” that Pye’s mother drank during pregnancy. Doc. 20- 40 at 65. He had simply presumed that she did so based on general indications that she’d been drinking. See Doc. 20-6 at 17. If we were evaluating Lawson’s testimony de novo, we might not view this clarification as a particularly negative reflection on his affida- vit’s credibility. But the state court didn’t say that it was discount- ing Lawson’s affidavit completely, and its finding that the correc- tion diminished Lawson’s credibility was not clearly and convinc- ingly erroneous, nor was it in any way unreasonable for the court to give some weight in its prejudice determination to the fact that Lawson needed to correct an earlier inaccurate statement. Despite these reasonable credibility concerns with some of the affidavits, Pye argues that it was unreasonable to discount all of the affidavits based on perceived credibility issues with just a few of them. It’s true that for many of the affidavits that speak to Pye’s childhood neglect and abuse, neither the state court nor the State have offered specific reasons to doubt their truth besides the gen- eral concern with “artfully drafted” affidavit testimony collected many years after trial. And in Porter v. McCollum, the Supreme Court held, with respect to evidence adduced from deposition tes- timony taken during habeas proceedings, that it was “unreasonable to discount to irrelevance the evidence of [the petitioner’s] abusive childhood, especially when that kind of history may have particular salience for a jury evaluating [the petitioner’s] behavior in his rela- tionship with [the victim].” 558 U.S. at 43. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 37 of 135

18-12147 Opinion of the Court 37

But here, there’s no indication (and the dissent has pointed to none) that the state court discounted the contents of the affida- vits “to irrelevance”—the court merely stated that it “reviewed the Petitioner’s affidavit evidence with caution.” Doc. 20-40 at 66 (em- phasis added). And “when the mitigating weight given to the post- conviction evidence is unclear, ‘we must presume that state courts know and follow the law.’” Evans v. Sec’y, Dep’t of Corr., 703 F.3d 1316, 1329–30 (11th Cir. 2013) (quoting Woodford v. Visciotti, 537 U.S. 19, 24 (2002)). That presumption isn’t defeated here because the Supreme Court hasn’t defined a standard that courts must fol- low in weighing the credibility of affidavit evidence produced in habeas proceedings. So, the state court’s decision to view the affi- davit evidence “with caution” was neither contrary to nor an un- reasonable application of clearly established federal law. Nor was the state court’s finding that the affidavits had been artfully drafted clearly and convincingly erroneous. For one thing, the court identified material inconsistencies in the affidavits pro- vided by Curtis, Ricky, and Lolla Mae Pye and Arthur Lawson. For another, there was substantial uniformity across the affidavits in terms of the language used to describe both Mostiler’s alleged fail- ure to discuss Pye’s background with the affiants and the affiants’ willingness to testify at sentencing had they been asked—a uni- formity that could plausibly suggest artful drafting of the sort that might create reasonable credibility concerns. See Waters v. Thomas, 46 F.3d 1506, 1513–14 (11th Cir. 1995) (“[T]he existence USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 38 of 135

38 Opinion of the Court 18-12147

of [habeas] affidavits, artfully drafted though they may be, usually proves little of significance.”). 13 Ultimately, even if—presuming the correctness of the state court’s factfinding—we might have given the affidavit testimony more weight in a prejudice analysis, it wasn’t unreasonable for the state court to discount the affidavits, to some degree, based on the inconsistencies it found in several of them and general concerns about after-the-fact artful drafting applicable to all of them. 14 And

13 Although the dissent acknowledges that the state court could determine that

the affidavits indicated artful drafting, it seems to suggest that there was no “evidence” here to support such a determination. Dissenting Op. at 43. But even aside from the fact that “the affidavits themselves,” id. at 43 n.33, con- tained sufficient evidence—for instance, the conspicuously parallel phrasing— the dissent offers no support for its insinuation that extrinsic corroborating evidence is required, nor are we aware of any. See, e.g., Nejad v. Attorney Gen., State of Ga., 830 F.3d 1280, 1284–92 (11th Cir. 2016) (holding that the absence of documentary proof to corroborate a witness’s testimony didn’t constitute the sort of clear and convincing evidence necessary to reject a state court’s credibility determination). To the contrary, federal habeas courts do require robust evidence—which is lacking here—before disturbing a state court’s credibility determinations. 28 U.S.C. § 2254(e)(1). 14 Another factor that supports the state court’s conclusions about the affida- vits’ credibility is the lack of corroborating evidence in the contemporaneous records—particularly regarding whether Pye was subject to regular physical abuse. The State argues, and Pye doesn’t really dispute, that there is no men- tion of him being abused in two decades’ worth of records from the Georgia Department of Human Resources and Division of Family & Children Services or in his school records. This fact, on balance, supports the state court’s deci- sion to view the affidavit evidence with caution, regardless of whether the court said as much in its written opinion. See supra Part II.B.3. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 39 of 135

18-12147 Opinion of the Court 39

the state court’s reasonable concerns about the affidavits’ credibil- ity properly played a role in the court’s overall prejudice analysis: Doubt about the affidavits’ credibility equates to reasonable uncer- tainty about (1) the truth of their depiction of Pye’s childhood as abusive and destitute, (2) whether the affiants actually would have testified to such mitigating factors at sentencing years earlier, and therefore (3) whether there was a substantial likelihood that Mostiler’s failure to fully investigate Pye’s background affected the outcome of sentencing. 2 Second, Pye’s family’s unwillingness to cooperate at trial. In order for Mostiler’s alleged failure to adequately investigate Pye’s background to have resulted in prejudice, Pye’s family would have had to have been willing (1) to cooperate during Mostiler’s pretrial investigation, (2) to take the stand during sentencing, and (3) to tes- tify frankly about the extreme neglect and abuse that Pye allegedly suffered in their home. If, contrary to what the affidavits said years later, Pye’s family was less than fully cooperative at the time of sen- tencing, it would undermine Pye’s contention that he was preju- diced. It was not unreasonable for the state court to consider the “evidence suggesting [Pye’s] family’s unwillingness to cooperate” as weighing against a finding of prejudice. Doc. 20-40 at 67. That evidence included (1) a contemporaneous memo from Mostiler’s files noting that Pye’s brothers didn’t respond to his phone calls, see Doc. 19-11 at 93; and (2) Dewey Yarbrough’s state post-convic- tion testimony that Pye’s family “w[as] not willing to work with USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 40 of 135

40 Opinion of the Court 18-12147

[them],” “didn’t put any effort forth on any of the contacts” he made with them, Doc. 19-11 at 24, and wasn’t willing to help him pursue leads, Doc. 14-41 at 85–86. Of course, some of Pye’s family did testify at sentencing. But that doesn’t mean that it was unrea- sonable for the state court to consider the family’s general uncoop- erativeness as undermining any argument about prejudice. 3 Third, the lack of “nexus” between Pye’s background and his crimes. 15 It wasn’t clearly and convincingly erroneous for the state court to find, nor was it unreasonable for it to weigh in its prejudice analysis the fact, “that there is little, if any, connection between [Pye’s] impoverished background and the premeditated and horrendous crimes in his case.” Doc. 20-40 at 66. Citing Ten- nard v. Dretke, 542 U.S. 274 (2004), and Williams v. Taylor, 529 U.S. 362, 367–68 (2000), Pye argues that the state court’s decision in that respect was unreasonable because the Supreme Court “has rejected a causal nexus requirement in order for penalty phase evi- dence to mitigate a capital crime” and has “given full weight to strikingly similar mitigation despite its lack of bearing upon the crime.” En Banc Br. of Appellant at 45–47. But Tennard held that mental-capacity evidence, in particular—not just any background evidence—can be mitigating regardless of nexus. 542 U.S. at 287. And while Pye is correct that the Supreme Court has decided in

15 The dissent offers no response to our assessment of the state court’s decision

in this respect. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 41 of 135

18-12147 Opinion of the Court 41

individual cases, like Williams, that the failure to present back- ground evidence was prejudicial despite a lack of nexus to the de- fendant’s crimes, those cases do not establish a per se rule that the degree of connection between background and crime can never play any role in a court’s prejudice analysis. And indeed, such a rigid rule would contradict the com- monsense prejudice standard, which assesses the likelihood that counsel’s failures changed the outcome of sentencing: Background circumstances that are closely linked to the defendant’s crime are naturally more likely to influence jurors than those that aren’t. Here, the state court reasonably concluded that Pye’s childhood poverty and neglect aren’t strongly connected to his crimes of gang-rape and murder 16—and this factor could properly have played a role in the court’s overall prejudice evaluation. Moreover,

16 Pye also argues that his childhood abuse—“the extreme domestic violence” that he experienced—would have had “particular salience for a jury evaluat- ing” his relationship with Alicia. En Banc Br. of Appellant at 47. Even if Pye is right, that wouldn’t mean the state court’s finding with respect to Pye’s childhood poverty was clearly and convincingly erroneous (or that the use of that finding in its prejudice determination was unreasonable): The state court mentioned only Pye’s “impoverished background,” not the history of domes- tic violence that Pye allegedly suffered—so it’s unclear whether and to what extent the connection between Pye’s alleged history of abuse and his crimes played in the court’s prejudice analysis. As already explained, the court may well have reasonably discounted the credibility of the affidavits alleging that Pye was abused or the likelihood that any witnesses would have testified to that abuse at sentencing—and for that reason didn’t proceed to consider the link between domestic abuse and Pye’s murder of his ex-girlfriend. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 42 of 135

42 Opinion of the Court 18-12147

given that it’s unclear from the court’s opinion the extent to which the court relied on this “nexus” factor in its prejudice analysis, “we must presume that [the] state court[] kn[e]w and follow[ed] the law.” Evans, 703 F.3d at 1329–30 (quotation marks omitted). 4 Fourth, Pye’s age at the time of the crime. It wasn’t unrea- sonable for the state court to give less mitigating weight to evi- dence about Pye’s childhood because he was 28 years old when he committed his crimes. It’s true that in Porter, the Supreme Court held that it was “unreasonable to discount to irrelevance the evi- dence of [the petitioner’s] abusive childhood” even though he was 54 years old at the time of the trial. 558 U.S. at 37, 43. But Pye overreads Porter when he claims that it makes the state court’s treatment of Pye’s age “patently unreasonable.” En Banc Br. of Appellant at 52 (quoting Pye, 853 F. App’x at 566). Neither Pye nor the dissent points to anything in Porter that explicitly forbids courts from considering age as one factor among many in their prejudice analyses—just as the state court did here. Cf. Evans, 703 F.3d at 1329–30 (noting that we must presume that state courts “know and follow the law” when determining what mitigating weight to give to post-conviction evidence). Regardless of whether we would read Porter de novo as signaling that habeas courts generally shouldn’t weigh age heavily in their prejudice analyses, it wasn’t contrary to or an unreasonable application of “clearly established Federal law,” 28 U.S.C. § 2254(d)(1), for the state court to consider Pye’s age as a factor weighing against prejudice. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 43 of 135

18-12147 Opinion of the Court 43

Moreover, as just explained with respect to “nexus,” a per se prohibition on the consideration of a defendant’s age in the preju- dice analysis would make little sense given that standard’s require- ment that we determine the likely impact of the unpresented evi- dence on the jury: Childhood neglect and abuse are certainly more likely to influence the jury if the defendant was barely an adult at the time of the crime than if he was significantly older. We do not interpret Porter as abrogating our precedents treating a defendant’s age at the time of his crime as an appropriate factor for a court to consider (among others) when conducting a Strickland prejudice analysis. See, e.g., Tompkins v. Moore, 193 F.3d 1327, 1337 (11th Cir. 1999) (noting that “where there are significant aggravating cir- cumstances and the petitioner was not young at the time of the capital offense, evidence of a deprived and abusive childhood is en- titled to little, if any, mitigating weight” (quotation marks omit- ted)); Bolender v. Singletary, 16 F.3d 1547, 1561 (11th Cir. 1994); cf. Francis v. Dugger, 908 F.2d 696, 703 (11th Cir. 1990) (per curiam) (noting that “the fact that [the defendant] was thirty-one-years old” when he committed the crime weighed in favor of finding that trial USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 44 of 135

44 Opinion of the Court 18-12147

counsel made a reasonable “decision not to investigate family childhood background”). 17

17 In discussing Pye’s age at the time of his crimes, the state court made two points. First, it offered the strange—and likely clearly erroneous—summary that “trial counsel could have reasonably decided, given the heinousness of this crime and the overwhelming evidence of [Pye’s] guilt, that remorse was likely to play better than excuses.” Doc. 20-40 at 66 (emphasis added). All here agree that Mostiler’s strategy at sentencing had nothing to do with “remorse”; it was focused instead on asking the jury for mercy. Second, and more broadly, the state court emphasized that “‘evidence of a deprived and abusive childhood is entitled to little, if any, mitigating weight’ when the defendant is ‘not young’ at the time of the offense.” Id. at 67 (citing Tompkins v. Moore, 193 F.3d 1327, 1337 (11th Cir. 1999)). The state court’s “remorse”-based statement—nestled in a sub-justifi- cation of a larger justification—doesn’t undermine the reasonableness of the state court’s overall rejection of Pye’s ineffective-assistance claim, or even ren- der the court’s constituent no-prejudice determination unreasonable. Per- spective is critical. The state court offered at least five justifications for its de- termination that Mostiler’s failure to introduce evidence of Pye’s childhood wasn’t prejudicial: because of (i) that evidence’s unreliability; (ii) the seeming unwillingness of the family to testify to it; (iii) its lack of nexus to Pye’s crime; (iv) Pye’s age; and (v) the aggravated nature of the rape-murder at issue. Within the state court’s discussion of one of those five justifications—pertain- ing to Pye’s age—one of its two sub-justifications was mistaken. With respect, the “remorse” issue is a sideshow—the proverbial flea on the hair of the tail of the dog. There is no indication—none—that the state court’s single misstate- ment regarding remorse “resulted in” a “decision” that was “based on” unrea- sonable determination of the facts. 28 U.S.C. § 2254(d)(2); see supra Part II.B.2. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 45 of 135

18-12147 Opinion of the Court 45

5 Fifth, and finally, the aggravating factors. 18 The state court found especially relevant the “extensive evidence presented in ag- gravation by the State during sentencing.” Doc. 20-40 at 67. That conclusion was far from unreasonable. The mitigating evidence not presented as a result of counsel’s deficient performance must be weighed “against the evidence in aggravation.” Porter, 558 U.S. at 41. We’ve repeatedly held that even extensive mitigating evi- dence wouldn’t have been reasonably likely to change the outcome of sentencing in light of a particularly heinous crime and significant aggravating factors. See, e.g., Windom v. Sec’y, Dep’t of Corr., 578 F.3d 1227, 1251 (11th Cir. 2009) (per curiam) (noting that given the strength of the State’s case “and the nature of the crimes them- selves,” the state court didn’t “unreasonably apply Strickland when it found that the available mitigating evidence, taken as a whole, did not outweigh the aggravating nature of [the defendant’s] crimes” (citing Payne v. Allen, 539 F.3d 1297, 1318 (11th Cir. 2008))); Suggs v. McNeil, 609 F.3d 1218, 1232 (11th Cir. 2010) (ex- plaining that significant aggravating facts are “difficult to over- come” and holding that a state supreme court’s prejudice decision wasn’t unreasonable). Here are the aggravating factors that the jury heard about Pye at sentencing: He had previously struck Alicia in the back with

18 The dissent offers no response to our assessment of the state court’s decision

in this respect. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 46 of 135

46 Opinion of the Court 18-12147

a gun, had been arrested for burglary, and had a “very bad” repu- tation for violence in the community. He enlisted two accomplices to kidnap Alicia—leaving an infant he thought was his alone at her home—and drive her to a motel room where the three men each raped her at gunpoint. Pye, 505 S.E.2d at 8–10. Then, they took her out onto a dirt road, where Pye ordered her to lie face-down on the ground, before he shot her in the back twice, after which she begged him not to shoot her in the head. Despite the oppor- tunity to show mercy, Pye shot her in the head anyway. Alicia took between 10 and 30 minutes to die, during which time she would have been conscious almost until the end, “crawl[ing] . . . in the dark” and “alone.” Doc. 13-11 at 88–89. It wasn’t unreasonable for the state court to weigh these aggravating factors heavily in its eval- uation of whether the presentation of additional mitigating evi- dence about Pye’s background would have changed a juror’s vote for the death sentence. 19 * * * The state court’s task in conducting its Strickland prejudice analysis was to assess probabilities—to determine, by weighing the aggravating and mitigating evidence, whether there was a “sub- stantial” likelihood that the outcome of sentencing would have been different had Mostiler conducted a more complete investiga- tion into Pye’s background. Shinn, 141 S. Ct. at 523. In doing so,

19Our concurring colleagues have likewise (and quite sensibly) emphasized the extremely aggravated nature of Pye’s crime. See Concurring Op. at 2–3. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 47 of 135

18-12147 Opinion of the Court 47

the court discounted, to some extent, the affidavit testimony that it received, and factored in the competing evidence that Pye’s fam- ily was generally uncooperative at the time of the trial, the tenuous connection between the mitigating evidence and Pye’s crimes, and Pye’s age when he committed those crimes. None of these choices individually resulted in a decision that was contrary to or involved an unreasonable application of clearly established federal law, or was based on an unreasonable determination of the facts. And it wasn’t “so obviously wrong [as to be] beyond any possibility for fairminded disagreement,” id. (quotation marks omitted), for the state court to conclude that on balance, given the significant aggra- vating evidence, there wasn’t a substantial likelihood that the jury would have voted for anything less than death even had Mostiler conducted a constitutionally adequate investigation into Pye’s background. B We next consider whether the state court’s conclusion that Pye wasn’t prejudiced by Mostiler’s failure to obtain a mental- health evaluation of Pye or present mental-health evidence at sen- tencing was either based on an unreasonable determination of the facts or contrary to or an unreasonable application of clearly estab- lished federal law. See 28 U.S.C. § 2254(d). It was not. To begin, in determining the facts, it was not clearly and convincingly erroneous (or unreasonable more generally) for the state court to view the evidence of Pye’s alleged brain damage as conflicting and to question the severity of the condition it reflected. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 48 of 135

48 Opinion of the Court 18-12147

One of Pye’s experts at the state habeas proceeding, Dr. Eisenstein, found that Pye had frontal-lobe impairment and brain damage— which suggested to him that Pye had an impaired ability to plan and control his impulses. But the State’s expert, Dr. King, testified that the facts of the crime, which involved significant premedita- tion and planning, were inconsistent with frontal-lobe impairment. The reason, he said, is because individuals with frontal-lobe dam- age have significant “disinhibition of responses and impulses in all areas” and “wouldn’t choose out a particular victim at a particular time and then engage in premeditation, goal directedness, trying to cover [their] tracks.” Doc. 14-44 at 69. Dr. King testified that the tests conducted by Dr. Eisenstein weren’t sophisticated enough “to identify that particular kind of specific brain damage,” id. at 68, and he expressed skepticism of the suggestion—made by another of Pye’s experts, Dr. Pettis—that Pye might have had a “failure to thrive” or “fetal alcohol syndrome,” id. at 72–73. Still, Dr. King agreed that even though Pye didn’t meet the threshold for mental retardation, he had cognitive “deficits in a number of areas” that would have “affect[ed] his ability . . . to function in the commu- nity.” Id. at 80. While the state court didn’t explicitly make a fac- tual finding about Pye’s alleged brain damage, it would have been reasonable for it to find that, given the testimony presented, he had cognitive deficits but not frontal-lobe impairment or fetal-alcohol syndrome. It was reasonable for the state court to conclude based on these facts that there wasn’t a substantial probability that the USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 49 of 135

18-12147 Opinion of the Court 49

presentation of mental-health evidence would have changed the outcome of Pye’s sentencing. While Pye may be correct in arguing that the only reasonable factual conclusion based on the evidence presented at the state habeas proceeding is that he has cognitive deficits, that doesn’t mean that it was unreasonable for the state court to find that no prejudice resulted from the failure to present this mental-health evidence at sentencing. Given the fact that Pye had sufficient mental faculties to “plan a robbery,” “le[a]d two fel- low co-defendants in the kidnapping, rape, and murder of his for- mer girlfriend,” “attempt[] to avoid detection by authorities through disposal of the murder weapon and accessories,” and “fab- ricate[] an alternative sequence of events,” Doc. 20-40 at 62, and in light of the aggravating factors already described, the jury could well have been unmoved even if Mostiler had obtained a mental- health evaluation and presented an expert’s testimony about Pye’s cognitive defects. It wasn’t unreasonable for the state court to find that there wasn’t a substantial likelihood of a different sentencing outcome.20

20 Pye also points to statements in records from his first prison stint that he seemed “unstable,” reported that he heard voices calling his name, exhibited a “flat affect” and a “rather fragile composure,” and displayed “elements of psychotic withdrawal” and “depression . . . severe enough to suggest consid- eration of chemotherapy.” En Banc Br. of Appellant at 49–50 (quoting Doc. 15-19 at 12–16). Pye faults the state court for not considering this evidence in its prejudice analysis. See id. at 50 (citing Pye, 853 F. App’x at 567). But under AEDPA, we do not assess “whether the state court considered and discussed every angle of the evidence”: “There is no text in [§ 2254(d)] requiring a USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 50 of 135

50 Opinion of the Court 18-12147

The state court’s prejudice determination regarding the mental-health evidence also didn’t contradict or unreasonably ap- ply clearly established federal law. There is no per se rule that the failure to present evidence of a defendant’s cognitive defects at sen- tencing is prejudicial for purposes of the Strickland ineffective-as- sistance analysis. While Pye cites Porter, that case noted that “it was not reasonable to discount entirely the effect” that the defend- ant’s mental-health expert’s testimony might have had on the jury. 558 U.S. at 43 (emphasis added). And here, “[n]othing in the opin- ion” of the state court “suggests that the mitigating effect of [Pye’s] mental health problems was ‘discount[ed] entirely.’” Evans, 703 F.3d at 1330 (quoting Porter, 558 U.S. at 43). Porter didn’t create a per se rule that the failure to present evidence of brain damage or cognitive defects is always prejudicial; rather, it held only that in

statement of reasons.” Lee, 726 F.3d at 1211 (quotation marks omitted); see supra Part II.B.3. Here, the state court’s overall prejudice determination with respect to Pye’s mental-health evidence was reasonable notwithstanding the unaddressed mental-health information in his prison records. As the district court correctly noted, the mitigation value of Pye’s psychological state when he went to prison for the first time—years before he killed Alicia—was low because “it is not at all surprising that someone who had just arrived at a state prison to begin serving a ten-year sentence would be depressed and confused.” Doc. 68 at 66. Moreover, the same intake form also confirmed that Pye had “no history of mental health treatment and did not show overt signs of severe depression, anxiety, or perceptual disturbance.” Id. (citing Doc. 15-19 at 11, Doc. 19-11 at 94). So, a fairminded habeas jurist could conclude that, even if these records had been presented to the trial jury, they wouldn’t have been substantially likely to make a difference. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 51 of 135

18-12147 Opinion of the Court 51

that case, given that particular petitioner’s brain damage, the fail- ure to present mental-health evidence was prejudicial. 558 U.S. at 43–44; see also Richter, 562 U.S. at 101 (explaining that in evaluat- ing whether a state court’s application of federal law was unreason- able, “[t]he more general the rule, the more leeway courts have in reaching outcomes in case-by-case determinations”); Knowles, 556 U.S. at 123 (noting that Strickland is a “general standard”). And the petitioner in Porter, in contrast to Pye, presented largely unrebut- ted evidence that he had PTSD from his military service that “could manifest in impulsive, violent behavior,” “suffered from an ex- treme mental or emotional disturbance,” and “was substantially USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 52 of 135

52 Opinion of the Court 18-12147

impaired in his ability to conform his conduct to the law.” 558 F.3d at 36. 21 Moreover, in addition to there being no per se rule of preju- dice based on unpresented mental-health evidence, “we have held that ‘the indication of brain damage . . . can often hurt the defense as much or more than it can help.’” Evans, 703 F.3d at 1329 (quot- ing Haliburton v. Sec’y, Dep’t of Corr., 342 F.3d 1233, 1244 (11th Cir. 2003)); see Haliburton, 342 F.3d at 1244 n.30 (noting defense attorney’s testimony that presenting evidence of a defendant’s

21 Nor do the other cases that Pye cites, see En Banc Br. of Appellant at 59–60, establish a rule that the failure to present evidence of any sort of brain damage or cognitive deficiency is necessarily prejudicial. While Jefferson v. GDCP Warden noted that “evidence of brain damage . . . profoundly change[s] the character of the penalty phase of the proceedings by fundamentally transform- ing [the defendant’s] sentencing profile,” it noted only that this sort of evi- dence “may establish prejudice.” 941 F.3d 452, 483 (11th Cir. 2019) (emphasis added). Moreover, Jefferson is distinguishable because it applied a stricter pre- AEDPA standard of review to the state court’s decision. Id. at 455. Nor is Sears v. Upton, 561 U.S. 945 (2010) (per curiam), directly analogous to Pye’s case. There was “clear and compelling evidence” in that case that the peti- tioner had “pronounced frontal lobe pathology” and was “among the most impaired individuals in the population in terms of ability to suppress compet- ing impulses.” Id. at 949–50 (quotation marks omitted). Finally, unlike Pye, the defendant in Rompilla v. Beard, suffered from “organic brain damage, an extreme mental disturbance significantly impairing several of his cognitive functions,” which was likely caused by fetal-alcohol syndrome and substan- tially impaired his capacity to “appreciate the criminality of his conduct or to conform his conduct to the law.” 545 U.S. 374, 392 (2005) (quotation marks omitted). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 53 of 135

18-12147 Opinion of the Court 53

abusive background and brain damage can counterproductively “paint a picture of Frankenstein” for the jury); cf. Windom, 578 F.3d at 1249 (holding that it “was not objectively unreasonable” for the state post-conviction court to find no prejudice where there was overwhelming evidence of premeditation, despite counsel’s “failure to investigate and present a mental health mitigation de- fense”). Given the conflicting evidence about the extent of Pye’s mental-health issues and the lack of clearly established federal law requiring a finding of prejudice based on the failure to present evi- dence of cognitive deficits, the state court’s conclusion on this issue was not “so obviously wrong [as to be] beyond any possibility for fairminded disagreement.” Shinn, 141 S. Ct. at 523 (quotation marks omitted). C We must next determine whether it was unreasonable for the state court to conclude that Pye suffered no prejudice as a result of Mostiler’s failure to rebut the State’s argument about his future dangerousness in prison. It was not. Pye’s argument rests largely on the state post-conviction tes- timony of two corrections officers—Ellenberg and Pittman—who supervised Pye during his prior incarceration as part of a youthful- offender program. See En Banc Br. of Appellant at 50–51 (citing Doc. 16-24 at 49, 70–71). But even assuming the truth of these of- ficers’ testimony—that they didn’t consider Pye a security concern USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 54 of 135

54 Opinion of the Court 18-12147

and that he was less dangerous than most inmates they encoun- tered—it was reasonable for the state court to conclude that this sort of evidence wouldn’t have been substantially likely to change the outcome of sentencing for three reasons: (1) prison records show evidence of Pye’s insubordination and aggressiveness; (2) Pye became increasingly violent after his first incarceration; and (3) fur- ther evidence that Pye wasn’t a violent person would have been cumulative. 22 First, the prison records. It was not clearly and convincingly erroneous for the state court to conclude that these records indi- cated that Pye had a “history of insubordination, aggressiveness and propensity for violence toward those in authority.” Doc. 20- 40 at 61. Had Mostiler presented testimony from corrections offic- ers about Pye’s behavior during his initial period of incarceration, the State likely would have presented later prison records, which contained at least 15 disciplinary reports, including those pertaining to fights with other inmates and instances of insubordination cate- gorized as “High”- and “Greatest”-level offenses. For instance, on October 12, 1989, Pye “became hostile and aggressive” toward cor- rections officers after being removed from his dorm. Doc. 15-20 at

22 The state court didn’t expressly discuss Factors (2) and (3), but in assessing whether a state court’s reasons for its decision were reasonable—here, whether it was reasonable for the state court to conclude that Pye wasn’t prej- udiced by Mostiler’s failure to present evidence about his behavior in prison— we can consider additional rationales that support the state court’s conclu- sions. See supra Part II.B.3. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 55 of 135

18-12147 Opinion of the Court 55

18. After he was instructed to assume the shakedown position, “Pye came off the wall in an aggressive manner” while shouting, “[m]other fucker get your hands off me”—leading to officers wres- tling him to the floor as he struggled with them. Id. at 16, 19. Pye disputes the state court’s characterization of his prison behavior, arguing that there’s no record indicating that he was ever violent toward prison personnel. See En Banc Br. of Appellant at 51. He describes one incident in which he fought with another in- mate as being mere horseplay that didn’t result in any injuries and asserts that the prison found him not guilty of assault in connection with another incident in which he fought an inmate. Id.; see Docs. 15-19 at 51; 15-20 at 8. But Pye doesn’t dispute that his prison rec- ords contain many instances of insubordination. Overall, Pye hasn’t rebutted by clear and convincing evi- dence that his prison records “indicate a history of insubordination, aggressiveness and propensity for violence toward those in author- ity.” Doc. 20-40 at 61. It may be debatable whether one should infer from Pye’s October 12, 1989 incident with prison staff and his altercations with other inmates that Pye had a history of “aggres- siveness and propensity for violence toward those in authority.” But Pye has not rebutted the presumption of correctness that AEDPA affords to state-court determinations of fact. See 28 U.S.C. § 2254(e)(1). And it wasn’t unreasonable for the state court to rely on this characterization of Pye’s prison records in assessing whether Mostiler’s failure to offer testimony rebutting the State’s future dangerousness argument was prejudicial. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 56 of 135

56 Opinion of the Court 18-12147

Second, the mitigating value of the officers’ testimony. Tes- timony from witnesses like Ellenberg and Pittman about Pye’s be- havior in prison likely would have had minimal value in swaying the sentencing jury. Ellenberg and Pittman knew Pye only during his incarceration in the youthful-offender program at Lee Arren- dale Correctional Facility for several years in the late 1980s. But Pye admits that when he aged out of that program and was trans- ferred to Frank Scott Correctional Institute 23 in 1988, “[h]is behav- ior became agitated and he incurred disciplinary reports for insub- ordination.” Doc. 43 at 63. So, even if Officers Ellenberg and Pittman had testified at Pye’s sentencing, they wouldn’t have been able to speak to Pye’s behavior at Frank Scott, and the State could have painted a picture of Pye as a man who became increasingly troubled and violent as he got older. And, of course, at the time of sentencing, the jury had just concluded—contrary to the officers’ testimony that Pye was generally nonviolent—that Pye had vio- lently raped and murdered Alicia Yarbrough. Thus, there is little chance that the officers’ testimony would have swayed any mem- ber of the jury: Even if Pye was generally nonviolent when he was incarcerated as part of a youthful-offender program years earlier, that says little about how dangerous he would be during a future

23While Pye’s brief in support of his habeas petition referred to this facility as “Robert Scott State Prison,” the state court and the contemporaneous discipli- nary reports in the record refer to the institution as “Frank Scott Correctional Institute.” Compare Doc. 43 at 63, with Docs. 20-40 at 61; 15-20 at 17. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 57 of 135

18-12147 Opinion of the Court 57

period of incarceration after he had become progressively more troubled and been convicted of rape and murder. Third, cumulativeness. During sentencing, Pye’s sister and father both testified that he was not violent, and several other wit- nesses testified about his kindness. This testimony would have served as a counterweight to the State’s argument about Pye’s fu- ture dangerousness. Further evidence from corrections officers as to Pye’s nonviolent nature would have been at least partially cu- mulative. See Cullen v. Pinholster, 563 U.S. 170, 200 (2011). The fact that the jury heard some testimony that Pye was generally non- violent further supports the reasonableness of the state court’s con- clusion that Pye wasn’t prejudiced by Mostiler’s failure to discover and present testimony like that offered by Ellenberg and Pittman at the state post-conviction proceedings. Together, these factors make it unlikely that the corrections officers’ testimony would have changed the outcome of Pye’s sen- tencing. At the very least, it wasn’t unreasonable for the state court to conclude that there wasn’t a “substantial likelihood” that the presentation of such testimony would have resulted in a different sentence. Shinn, 141 S. Ct. at 524. D Finally, Pye contends that he was prejudiced by Mostiler’s failure to present evidence of Alicia Yarbrough’s cocaine habit— including evidence that she had cocaine in her system the night she died—and testimony from Linda Lyons that Alicia called Pye on USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 58 of 135

58 Opinion of the Court 18-12147

the night of her murder. See En Banc Br. of Appellant at 62–68. Pye frames this as residual-doubt evidence that he says would have supported his story that Alicia voluntarily met him at the motel to trade sex for drugs—negating the aggravating circumstances of the rape and kidnapping committed alongside Alicia’s murder—and that it could have persuaded the jury not to impose the death sen- tence. But even if Mostiler should have presented this additional evidence supporting Pye’s version of events during the guilt or pen- alty phase of trial, it was reasonable for the state court to conclude that his failure to do so wasn’t prejudicial. To begin, it wasn’t clearly and convincingly erroneous for the state court to find that Lyons’s post-conviction affidavit testi- mony was unreliable. In this affidavit, Lyons—Alicia’s friend and neighbor—said that Alicia called a local motel from Lyons’s house and that Lyons heard her ask for Pye’s room and arrange for some- one to pick her up—presumably to get drugs. See Doc. 16-24 at 66. The state court pointed out the inconsistency between this state- ment and what Lyons told a police investigator about 12 hours af- ter seeing Alicia for the last time: Lyons heard Alicia call “some- one” at a local motel and “ask for room #27,” and Alicia told the “unknown party” on the other end of the line that she “was going to call the police on them for selling drugs out of the motel.” Doc. 12-9 at 3. The lack of positive identification of the person Alicia was calling and Alicia’s threat to call the police on that person—a relevant fact not reported in Lyons’s post-conviction affidavit, see Doc. 16-24 at 66—are significant differences between Lyons’s initial USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 59 of 135

18-12147 Opinion of the Court 59

story and the affidavit she prepared years later for the state habeas proceedings. Given these discrepancies, it wasn’t clearly and con- vincingly erroneous for the state court to find that Lyons’s habeas affidavit testimony was unreliable. Even if Lyons’s affidavit testimony is weighed alongside the additional evidence of Alicia’s cocaine use that Pye says Mostiler should have presented, there still isn’t a substantial likelihood that this evidence would have changed the outcome of sentencing. That’s because Pye’s version of events, in addition to being only weakly supported by Lyons’s unreliable affidavit, is implausible in light of the evidence produced at trial. As the state court noted, within about 24 hours of the crime, Georgia investigators exam- ined the residence where Alicia had been living with Charles Puck- ett. They found that the front door had been forced open, with the door, door jamb, and locking mechanisms “broken and shattered from a violent force initiated from the exterior” as though the door had been “kicked open.” Doc. 12-2 at 108–09. That finding was consistent with the testimony of Pye’s co-defendants Anthony Freeman and Chester Adams that Pye kicked in Alicia’s door when he forced himself into the home, but inconsistent with the story that Pye now says he could have told to raise residual doubt—that Alicia willingly went to Pye’s motel to get drugs. Pye’s explanation that Alicia kicked in her own door is implausible. And Pye would have had little reason to murder Alicia if she had gone willingly to his motel room to exchange sex for drugs, but every reason to kill her if he’d kidnapped, robbed, and raped her. At the very least, it USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 60 of 135

60 Opinion of the Court 18-12147

wasn’t unreasonable—“so obviously wrong [as to be] beyond any possibility for fairminded disagreement,” Shinn, 141 S. Ct. at 523 (quotation marks omitted)—for the state court to conclude that in light of this competing evidence, testimony from Lyons and addi- tional proof of Alicia’s cocaine use wouldn’t have created residual doubt substantially likely to change the outcome of sentencing. E Even if the state court’s prejudice determination as to each ground of allegedly deficient performance was reasonable, we must still decide whether its conclusion as to the cumulative prej- udice, constituted an unreasonable application of Strickland. See Strickland, 466 U.S. at 694–96; United States v. Blakey, 14 F.3d 1557, 1561 (11th Cir. 1994) (discussing cumulative effect of counsel’s er- rors). This question asks whether it was reasonable for the state court to conclude that there was no substantial likelihood that at least one juror would have voted against imposing the death pen- alty had Mostiler not committed all the errors that Pye alleges (and we assume) that he committed—i.e., if Mostiler had conducted a more thorough investigation of Pye’s background and presented additional evidence of his neglected and (possibly) abusive child- hood, discovered and presented evidence of Pye’s cognitive defi- ciencies, offered testimony about Pye’s generally nonviolent be- havior when he was previously incarcerated, and introduced addi- tional residual-doubt evidence. But even considering Mostiler’s al- leged deficiencies cumulatively, it wasn’t unreasonable for the state court to conclude that Pye has failed to establish prejudice: The USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 61 of 135

18-12147 Opinion of the Court 61

extensive aggravating circumstances of Pye’s crimes weighed heav- ily in favor of the jury imposing a death sentence, and the difficul- ties already described, which prevent Pye from establishing preju- dice with respect to any individual deficiency—including § 2254(e)(1)’s presumption of correctness, credibility concerns with the habeas affidavits, conflicting mental-health evidence, and min- imally relevant and conflicting evidence regarding Pye’s behavior in prison—could also, to a fairminded jurist, preclude him from es- tablishing cumulative prejudice. No precedent applying AEDPA to state-court prejudice de- terminations compels a different result. While Pye argues that the background evidence that Mostiler should have presented parallels the evidence in Williams v. Taylor, 529 U.S. 362 (2000), and Romp- illa v. Beard, 545 U.S. 374 (2005), those cases “offer no guidance with respect to whether a state court has unreasonably determined that prejudice is lacking” because the Supreme Court “did not ap- ply AEDPA deference to the question of prejudice in those cases.” Pinholster, 563 U.S. at 202. And even if those precedents were in- structive, the balance of aggravating and mitigating factors is sig- nificantly different in Pye’s case. Pye’s crimes against Alicia could reasonably be considered more aggravated than the robbery and murder that the petitioner in Williams committed. 529 U.S. at 367– 68. In addition, Pye’s argument comparing his background to the petitioner’s in Williams also assumes the truth of the affidavits pre- sented at his state post-conviction proceeding. But see supra Part III.A.1. In Rompilla, there were significant aggravating factors— USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 62 of 135

62 Opinion of the Court 18-12147

the murder was committed by torture during a felony and the de- fendant had a significant history of violent felony convictions—and the petitioner’s background was characterized by abuse and ne- glect similar to what Pye alleges. 545 U.S. at 378, 391–92. But in that case, unlike here, there was credible contemporaneous evi- dence in the petitioner’s file (which his attorneys hadn’t examined) that suggested that he was schizophrenic and had a third-grade level of cognition, and later testing showed “an extreme mental dis- turbance . . . likely caused by fetal alcohol syndrome.” Id. at 391– 92. Thus, the mitigating evidence in Rompilla was significantly stronger than the evidence presented here. Lastly, in Porter, the mitigating evidence that defense counsel failed to present was also significantly stronger than what Pye has presented: Had counsel performed competently, the jury would have heard about the pe- titioner’s “heroic military service in two of the most critical—and horrific—battles of the Korean War” and his “struggles to regain normality upon his return from war,” including PTSD that could “manifest in impulsive, violent behavior.” 558 U.S. at 36, 41. Given the reasonableness of the state court’s weighing of the evidence and the lack of contrary precedent, AEDPA requires us to defer to that court’s cumulative-prejudice conclusion because it wasn’t contrary to or an unreasonable application of the Supreme Court’s precedents, based on an unreasonable determination of the facts, or “so obviously wrong that its error lies beyond any possi- bility for fairminded disagreement.” Shinn, 141 S. Ct. at 523 (quo- tation marks omitted); see 28 U.S.C. § 2254(d). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 63 of 135

18-12147 Opinion of the Court 63

IV In conclusion, a brief word about today’s dissent—which like so (so, so, so) many before it, is framed around an extended allusion to Lewis Carroll’s Alice-based novels. See Parker B. Pot- ter, Jr., Wondering About Alice: Judicial References to Alice in Wonderland and Through the Looking Glass, 28 Whittier L. Rev. 175 (2006) (noting that, as of almost 20 years ago, some 1000 judi- cial opinions had referenced Carroll’s works). What the dissent lacks in originality, it more than makes up for in spice. It accuses us of all manner of things—peddling “[n]onsense,” Dissenting Op. at 18, “bury[ing]” unreasonable legal conclusions and factual find- ings, id. at 3, 35, “nullif[ying]” Supreme Court precedent, id. at 16, and “invent[ing]” reasons to “prop up” the state court opinion, id. at 24. Respectfully, none of those things are true. Our dissenting colleagues’ objections notwithstanding, the fact is that the standard embodied by 28 U.S.C. § 2254, as amended by AEDPA, is “difficult to meet . . . because it was meant to be.” Richter, 562 U.S. at 102. While AEDPA “stops short of imposing a complete bar on federal-court relitigation of claims already rejected in state proceedings,” we have authority to grant relief only “where there is no possibility fairminded jurists could disagree that the state court’s decision conflicts with [the Supreme Court’s] prece- dents.” Id. Section 2254(d) “reflects the view that habeas corpus is a guard against extreme malfunctions in the state criminal justice systems, not a substitute for ordinary error correction through ap- peal.” Id. at 102–03 (quotation marks omitted). The rationale for USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 64 of 135

64 Opinion of the Court 18-12147

this principle is well established: “Under AEDPA, state courts play the leading role in assessing challenges to state sentences based on federal law.” Shinn, 141 S. Ct. at 526. “Federal habeas review of state convictions frustrates both the States’ sovereign power to punish offenders and their good-faith attempts to honor constitu- tional rights.” Richter, 562 U.S. at 103 (quoting Calderon v. Thompson, 523 U.S. 538, 555–56 (1998)). It “disturbs the State’s significant interest in repose for concluded litigation, denies society the right to punish some admitted offenders, and intrudes on state sovereignty to a degree matched by few exercises of federal judicial authority.” Id. (quoting Harris v. Reed, 489 U.S. 255, 282 (1989) (Kennedy, J., dissenting)). Put simply, we have the power to overturn a state court’s decision on the merits of a petitioner’s habeas claim only in rare circumstances. Pye has not shown that this is one of them. The district court’s denial of habeas relief with respect to Pye’s ineffective-assistance-of-counsel-at-sentencing claim is AFFIRMED and the case is REMANDED to the panel for proceed- ings consistent with this opinion. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 65 of 135

18-12147 JORDAN, J., Concurring 1

JORDAN, Circuit Judge, joined by ROSENBAUM, Circuit Judge, con- curring in the judgment: I join Parts I and II of Judge Jill Pryor’s dissent (with the ex- ception of the last paragraph on page 30). But despite reservations with the majority opinion, I concur in the judgment denying Mr. Pye habeas relief, and write to explain why. In deciding this appeal, the majority resolves an important issue of first impression in our circuit—the relationship between 28 U.S.C. §§ 2254(d)(2) and 2254(e)(1)—in a single paragraph. This issue is of significant complexity, as evidenced by the literature dis- cussing the caselaw and the different interpretive approaches that exist. See, e.g., Randy Hertz & James S. Liebman, 1 Fed. Habeas Corpus Prac. & Proc. § 20.2[c] (7th ed. & 2020 update); Brian R. Means, Postconviction Remedies § 28.3 (June 2021 update); Justin F. Marceau, Deference and Doubt: The Interaction of AEDPA § 2254(d)(2) and (e)(1), 82 Tul. L. Rev. 385, 396–440 (2007). We did not ask the parties to address this issue, and they did not brief it. In the absence of adversarial presentation, I would not decide it here. See United States v. Sineneng-Smith, 140 S. Ct. 1575, 1579–81 (2020). Like the panel, and as set forth in Judge Jill Pryor’s dissent, I think the state court made a number of significant and unreasona- ble factual determinations. See Pye v. Warden, 853 F. App’x 548, 562–63, 566–67 (11th Cir. 2021); Jill Pryor Dissent at 35–49. I would conduct plenary review as to the prejudice prong of Strickland v. Washington, 466 U.S. 668, 687 (1984), and deny relief because Mr. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 66 of 135

2 JORDAN, J., Concurring 18-12147

Pye has not made the requisite showing. See Berghuis v. Thomp- kins, 560 U.S. 370, 390 (2010) (“Courts can . . . deny writs of habeas corpus under § 2254 by engaging in a de novo standard when it is unclear whether AEDPA deference applies because a habeas peti- tioner will not be entitled to a writ of habeas corpus if his or her claim is rejected on de novo review[.]”). To show prejudice under Strickland, Mr. Pye must demon- strate “a reasonable probability that, but for his counsel’s ineffec- tiveness, the jury would have made a different judgment,” and be- cause Georgia law requires a unanimous jury recommendation of death the focus is on whether one juror would have come to a dif- ferent conclusion. See Andrus v. Texas, 140 S. Ct. 1875, 1886 (2020). Although the reasonable probability standard does not re- quire Mr. Pye to show that his counsel’s performance more likely than not affected the outcome, the likelihood of a “different result must be substantial, not just conceivable.” Harrington v. Richter, 562 U.S. 86, 111–12 (2011). Each capital case, and every capital defendant, is different. Generalizations, at least when it comes to the prejudice determina- tion, are therefore difficult to make. For me, this is one of those cases in which the totality of the new mitigating evidence—taking into account some of its limitations and its partly contested na- ture—does not satisfy the reasonable probability standard. When juxtaposed against the brutality and cruelty of the premeditated kidnapping, gang rape, and murder of Ms. Yarbrough—whose child Mr. Pye claimed was his—after her plea for mercy, I do not USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 67 of 135

18-12147 JORDAN, J., Concurring 3

believe there is a substantial likelihood that one juror would have made a different recommendation as to punishment. In other words, there is not a reasonable probability that one juror would have “concluded that the balance of aggravating and mitigating cir- cumstances did not warrant death.” Strickland, 466 U.S. at 695. Cf. Krawczuk v. Secretary, 873 F.3d 1273, 1297–98 (11th Cir. 2017) (“[U]nder de novo review, we readily conclude that Krawczuk failed to establish a reasonable probability that, had he presented the above mitigating evidence [abandonment, isolation, lack of su- pervision, neuropsychological damage, mental disorders, emo- tional and physical abuse, depression symptoms, and sexual abuse by strangers on one occasion] the outcome of the proceedings would have been different. . . . In reaching this conclusion, we weigh the totality of the mitigating evidence against the aggravat- ing factors, considering the substantial weight due to aggravation in light of the brutal nature of [the] murder. . . . Krawczuk’s cruelty and premeditation make it unlikely that he would have received a different sentence.”). The Strickland prejudice analysis is, of course, a predictive human endeavor based on a hypothetical construct. See Evans v. Secretary, 703 F.3d 1316, 1334 (11th Cir. 2013) (en banc) (Jordan, J., concurring). But it is the framework the Supreme Court has given us, and the one we must apply. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 68 of 135 USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 69 of 135

18-12147 JILL PRYOR, J., Dissenting 1

JILL PRYOR, Circuit Judge, joined by WILSON, Circuit Judge, dis- senting: When she stepped through the looking glass, Alice found a world of opposites, nonsense, and “impossible things.” 1 Walking toward a thing is best accomplished by walking away from it.2 Time runs backwards. Alice can read “Jabberwocky” only by view- ing it through a mirror, and, even so, finds it “rather hard to under- stand.”3 As the author of the panel opinion in this case, which ap- plied the familiar legal standards set forth in the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), I feel as though I too have stepped through the looking glass. But what happened during Alice’s time through the looking glass was a dream. This, case, un- fortunately, is not. Willie James Pye was convicted of an aggravated crime. He brutally raped and murdered his former girlfriend. Despite over- whelming evidence to the contrary, Mr. Pye maintained his inno- cence and insisted on a defense strategy focused on proving it. When Mr. Pye’s family members understandably were uncooper- ative in helping him try to prove the unprovable in the guilt phase of his trial, Mr. Pye’s lawyer, Johnny Mostiler, and investigator, Dewey Yarbrough, largely gave up on attempting to rally the

1 Lewis Carroll, Through the Looking Glass 47 (2022). 2 Id. at 15. 3 Id. at 10. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 70 of 135

2 JILL PRYOR, J., Dissenting 18-12147

family members for the penalty phase, to save their client’s life. The majority opinion does not defend trial counsel’s performance, so I will not go on about his shortcomings in this case. Suffice it to say that after presenting a meager case in opposition of the death penalty, Mr. Mostiler gave a canned closing argument that his op- posing counsel anticipated and rebutted, to disastrous result. Rather than defending trial counsel’s performance, the ma- jority opinion concludes that Mr. Pye has failed to show preju- dice—or, more precisely, that the state habeas court’s determina- tion that he hadn’t shown prejudice was not, in AEDPA’s terms, “contrary to” and did not “involve[] an unreasonable application of, clearly established Federal law”; nor was it “based on an unrea- sonable determination of the facts.” 28 U.S.C. § 2254(d). In federal habeas, that is checkmate. But in reaching its conclusion, the ma- jority opinion makes two moves that do not belong on any chess board this side of the mirror. So I dissent. The majority opinion’s first move is to declare that federal courts may find that a reasoned state court decision withstands AEDPA deference by turning to justifications the state court never even hinted at. This is the opposite of what the Supreme Court has instructed, and the majority’s attempt to wiggle out from under Supreme Court precedent is unconvincing. The majority opinion supports its declaration with a half-baked textual analysis. And it relies on cases holding—uncontroversially—that a state court’s de- cision is not unreasonable just because it did not address and reject each one of a petitioner’s arguments and pieces of evidence USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 71 of 135

18-12147 JILL PRYOR, J., Dissenting 3

supporting his claims. To turn this unremarkable principle into support for its holding, the majority must refract the light shed by these cases beyond what the laws of nature allow. Second, the majority opinion holds—on an issue of first im- pression in this Court that was never briefed or argued by the par- ties—that a state court’s findings of fact may be clearly erroneous but not sufficiently important to meet the “unreasonable” AEDPA standard. Even if we assume for argument’s sake that this holding is correct, when combined with the majority opinion’s disregard of Supreme Court precedent requiring us to review exclusively the reasons the state habeas court actually gave, the holding creates a practically impossible path to relief for habeas petitioners. If federal courts can bury unreasonable findings under an avalanche of new reasons the state court never gave, then unreasonable findings will virtually never be important enough to satisfy the majority’s test. In Part I, I describe what happened in this case. Although the majority opinion mostly gets the facts right, I will highlight some nuances that, I think, the majority opinion has missed. In Part II, I describe habeas review under AEDPA and explain the majority opinion’s major errors in describing AEDPA deference. In Part III, using the proper AEDPA analysis, I examine the state habeas court’s decision and conclude that de novo review of the prejudice to Mr. Pye’s defense is warranted. In Part IV, I demonstrate why, on a de novo review, Mr. Pye has shown prejudice. In Part V, I conclude by summarizing the majority’s errors and the impact they will have unless the Supreme Court sets us right again. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 72 of 135

4 JILL PRYOR, J., Dissenting 18-12147

I. BACKGROUND As the majority opinion recounts, the facts of Mr. Pye’s crime are indeed aggravated. Mr. Pye had dated the victim, Alicia Lynn Yarbrough, on and off for a time.4 When the crime was com- mitted, however, Ms. Yarbrough was living with another man, Charles Puckett, and their infant. Mr. Pye and two associates, Ches- ter Adams and Anthony Freeman, drove to the home of Ms. Yar- brough and Mr. Puckett, apparently intending to rob Mr. Puckett. Mr. Pye was angry that Mr. Puckett had signed the birth certificate of Ms. Yarbrough’s child, whom Mr. Pye believed was his child. When they arrived, Mr. Puckett was not at home; Mr. Pye forcibly took Ms. Yarbrough from the home, leaving the infant behind. The three men drove to a hotel and rented a room, where each man repeatedly raped Ms. Yarbrough. The men eventually took Ms. Yarbrough from the hotel room, put her into Mr. Adams’s car, and drove away from the hotel. At Mr. Pye’s direction, Mr. Adams pulled the car onto a dirt road. Mr. Pye ordered Ms. Yarbrough out of the car, made her lie face down, and shot her three times as she begged for her life. She died of the wounds. Mr. Freeman confessed and implicated the other two men. The trial court appointed Mr. Mostiler, the county public de- fender, to represent Mr. Pye. Mr. Mostiler was assisted by Mr. Yar- brough. Mr. Pye maintained his innocence, despite the

4 Ms. Yarbrough was not related to investigator Dewey Yarbrough. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 73 of 135

18-12147 JILL PRYOR, J., Dissenting 5

overwhelming evidence of his guilt, and testified in his own de- fense. The jury found him guilty. At the trial’s penalty phase, the State presented testimony from three witnesses who collectively spoke about Mr. Pye’s pre- vious conviction and incarceration for burglary, reputation for vi- olence about a decade before the murder, and a previous violent altercation involving Ms. Yarbrough. Mr. Mostiler presented testi- mony from eight lay witnesses: Mr. Pye’s sister Pam Bland, sister Sandy Starks, brother Ricky Pye, father Ernest Pye, 15-year-old niece Cheneeka Pye,5 nephew Dontarious Usher, 6 sister-in-law Bridgett Pye, and family friend Lillian Buckner. These witnesses testified that Mr. Pye was of good moral character and asked the jury for mercy on his behalf. Ernest Pye testified that Mr. Pye had the makings of a normal childhood: he “liked to play basketball, liked to play with kids,” was “always smil[ing,]” and was never known to fight. Doc. 13-11 at 48. 7 Some of the witnesses said Mr. Pye and Ms. Yarbrough seemed to have a good relationship.

5 I use the spelling “Cheneeka” here because it is used in Ms. Pye’s postcon- viction affidavit. At trial, the court reporter spelled her name “Chanika” with- out confirming on the record the correct spelling. See Doc. 13-11 at 51. 6 Here, too, I use the spelling reflected in Mr. Usher’s postconviction affidavit. At trial, the court reporter spelled his name “Dantarius” without confirming on the record the correct spelling. See Doc. 13-11 at 59. 7 “Doc.” numbers refer to the district court’s docket entries. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 74 of 135

6 JILL PRYOR, J., Dissenting 18-12147

Mr. Mostiler asked a couple of the witnesses about Mr. Pye’s early life. Ms. Starks testified that she and her siblings were raised in a house with no “running water in the bathroom” and only a “wooden heater.” Id. at 67. But Ms. Bland testified that the family “had a four-bedroom” home. Id. at 30. And Ms. Starks told the jury that, above all, the family “had love.” Id. at 67. In closing, the prosecutor—who had tried several capital cases against Mr. Mostiler—anticipated what Mr. Mostiler would say to the jury in defense of Mr. Pye’s life, down to the letter. He told the jury Mr. Mostiler would quote from William Shake- speare’s The Merchant of Venice, “the quality of mercy is not strained,” and from the Bible’s Beatitudes, “blessed are the merciful for they shall obtain mercy.” Id. at 83. Mr. Mostiler did exactly that, revealing that he had not bothered to tailor his argument to Mr. Pye’s case. The prosecutor also relied on some old tricks. As he had in previous cases he had tried against Mr. Mostiler, he told the jury that, if left in prison for life, Mr. Pye would “for sure kill a guard to get out.” Id. at 86–87. Even though Mr. Mostiler well knew that the prosecutor had used this argument in previous cases, Mr. Mostiler had nothing prepared to refute the prosecutor’s assertion. He simply told the jury that Mr. Pye wouldn’t kill a prison guard. A classic “just trust me” with nothing to back it up. The jury did not just trust Mr. Mostiler. After finding four statutory aggravating factors, the jury unanimously recommended a sentence of death, which the trial court imposed. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 75 of 135

18-12147 JILL PRYOR, J., Dissenting 7

After Mr. Pye’s convictions and sentence were upheld on di- rect appeal, he sought postconviction relief in state court. This en banc proceeding is concerned with only one of Mr. Pye’s postcon- viction claims: that his trial counsel rendered ineffective assistance in failing to investigate and present mitigating evidence at the pen- alty phase of his trial, including evidence of his family background, mental health challenges, and cognitive impairment, as well as ev- idence to “counter the State’s evidence of aggravated culpability.” Rompilla v. Beard, 545 U.S. 374, 380–81 (2005). 8 To prove his claim in state court, Mr. Pye was required to demonstrate that trial counsel’s performance fell “below an objec- tive standard of reasonableness,” taking into account prevailing professional norms, and that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland v. Washington, 466 U.S. 668, 688, 694 (1984). Mr. Pye presented evidence of deficient per- formance and resulting prejudice, but because the majority opinion assumes Mr. Mostiler performed deficiently, I do not recount the evidence relating solely to that element and instead focus on Mr. Pye’s evidence that went to the state habeas court’s prejudice- prong reasoning.

8 Although the majority opinion addresses Mr. Pye’s argument that trial coun- sel also should have leaned more heavily on residual doubt, see Maj. Op. at 30 n.12, I do not address it here because in my view Mr. Pye is entitled to relief even without it. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 76 of 135

8 JILL PRYOR, J., Dissenting 18-12147

At the postconviction evidentiary hearing, Mr. Yarbrough testified about the defense team’s work on Mr. Pye’s case. He tes- tified that the primary focus of the defense was to prove Mr. Pye’s innocence, a strategy Mr. Pye insisted upon. Mr. Yarbrough testi- fied that Mr. Pye told him “to go out and contact his family mem- bers,” and so he contacted four or five of them, including Mr. Pye’s mother, father, and two to three siblings. Doc. 19-11 at 23. Mr. Yar- brough testified that the family “didn’t put any effort forth on any of the contacts I made with them.” Id. at 24. He recalled one family member saying “that [Mr. Pye] got himself into this, [and] he can get himself out of it.” Id. When asked to explain “[h]ow. . . the lack of family involvement affect[ed the] investigation,” Mr. Yarbrough clarified that the family was unhelpful “as far as helping prove [Mr. Pye’s] innocence.” Id. at 24–25 (emphasis added); see id. at 25 (“[T]hey were not willing to . . . put the effort forward to prove what I was trying to prove that I was told to try to prove by Wil- lie.”). Mr. Yarbrough recounted: “I can remember thinking, and I want to say this was during, right before the sentencing phase, you know, I just don’t care about going back over there and trying to get them here.” Id. at 25 (emphasis added). At the hearing, habeas counsel offered undisputed evidence that Mr. Pye is of low intellectual functioning, bordering on intel- lectual disability. 9 They also offered undisputed evidence that

9 As the state habeas court explained, “[i]t is undisputed among the mental health professionals who have evaluated [Mr. Pye] that [his] intellectual USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 77 of 135

18-12147 JILL PRYOR, J., Dissenting 9

Mr. Pye has suffered from depression for nearly all his life. And counsel offered undisputed evidence through affidavits that Mr. Pye experienced a traumatic childhood and adolescence, dur- ing which physical and emotional abuse, extreme parental neglect and endangerment, and abject poverty pervaded his daily life, as well as a resulting troubled adulthood. Finally, habeas counsel in- troduced evidence that Mr. Pye previously had adapted well to car- ceral life and had been trusted by prison staff. The state habeas court denied Mr. Pye’s habeas petition, concluding, as relevant here, that he failed to show that any defi- cient performance by Mr. Mostiler prejudiced him. The court found that evidence of low intellectual functioning would not have swayed the jury. The court noted the affidavit testimony rebutting the State’s contention of future dangerousness but emphasized that Mr. Pye’s corrections records showed several instances of insubor- dination and aggression. The court thus found no reasonable prob- ability that Mr. Pye’s resulting sentence would have been different had the jury heard positive testimony about his adaptation to prison.

functions are in the low to borderline range.” Doc. 20-40 at 18. The majority opinion also discusses dueling witness testimony about the presence and se- verity of brain damage. See Maj. Op. at 7, 47–53. I do not discuss that testi- mony here because I believe Mr. Pye is entitled to relief even without the con- tested evidence of brain damage. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 78 of 135

10 JILL PRYOR, J., Dissenting 18-12147

The state habeas court also concluded that trial counsel’s failure to investigate and present evidence of Mr. Pye’s family back- ground did not cause prejudice. The majority opinion characterizes the state habeas court’s conclusion as based on five reasons: (1) the state habeas court’s “decision to discount the affidavit evidence pre- sented at the state post-conviction proceedings due to concerns about their credibility”—specifically, supposed “artful drafting”; (2) “evidence of Pye’s family’s unwillingness to cooperate in his de- fense at the time of trial”; (3) “the minimal connection between Pye’s background and the crimes he committed”; (4) “Pye’s age[, 28,] at the time of those crimes”; and (5) “the extensive aggravating evidence presented by the State at sentencing.” Maj. Op. at 31, 37. The majority opinion’s identification of the fourth reason is correct but incomplete—the state habeas court found because Mr. Pye “was 28 years old at the time of these crimes, trial counsel could have reasonably decided, given the heinousness of this crime and the overwhelming evidence of Petitioner’s guilt, that remorse was likely to play better than excuses.” Doc. 20-40 at 66; see Maj. Op. at 44 n.17 (acknowledging the court’s “remorse” finding). Add the court’s reasoning that Mr. Pye’s evidence of (6) low intellectual functioning and (7) lack of future dangerousness would not have swayed the jury, and we can see that the state habeas court supplied seven total reasons for its no-prejudice determination.10 And so,

10 Do I think we need to take a tally of a state court’s reasons? No. I do so here only to compare the state habeas court’s reasoning with the majority opinion, which adds reasons that the state habeas court did not give. Because the USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 79 of 135

18-12147 JILL PRYOR, J., Dissenting 11

the state habeas court denied Mr. Pye relief from his death sen- tence. After the Supreme Court of Georgia denied Mr. Pye a certif- icate of probable cause to appeal the state habeas court’s order, Mr. Pye filed a federal habeas petition. Focusing primarily on prejudice, the district court rejected the petition but granted Mr. Pye a certif- icate of appealability on the claim we address today. After briefing, and with the benefit of oral argument, a panel of this Court—of which I was a member—held that the state habeas court’s rejection of Mr. Pye’s ineffective-assistance-of-counsel claim was contrary to and involved an unreasonable application of clearly established fed- eral law and was based on unreasonable factual determinations in light of the state court record. See 28 U.S.C. § 2254(d). On de novo review, we concluded that Mr. Pye had shown deficient perfor- mance and prejudice and, therefore, was entitled to habeas relief.11

universe AEDPA tasks us with reviewing is more limited than what the ma- jority opinion describes, I use numbers to define precisely what we should be reviewing. 11 Our opinion, which was not listed for publication, set out no new law. Ra- ther, it simply applied precedent to the facts of the case. See 11th Cir. R. 35-3 (“[E]rror asserted in the panel’s misapplication of correct precedent to the facts of the case[] are matters for rehearing before the panel but not for en banc consideration.”). As I will explain in the next section, the majority opinion has used this case not, as it professes, to “clarify” AEDPA’s application, Maj. Op. at 13, but to narrow even further the nearly impossible path to relief available to a state prisoner in federal habeas. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 80 of 135

12 JILL PRYOR, J., Dissenting 18-12147

II. AEDPA AEDPA was enacted in 1996 in part to streamline the federal review of state prisoners’ habeas petitions, but in practice the stat- ute has done anything but. 12 Its abstruse language also has left much to the imagination and rumination of jurists and litigants alike throughout the more than quarter of a century it has been in place. The basics, though, are simple enough. AEDPA bars federal courts from granting habeas relief to a petitioner on a claim that was adjudicated on the merits in state court unless the state court’s adjudication of the claim “(1) resulted in a decision that was con- trary to, or involved an unreasonable application of, clearly estab- lished 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). These standards are, we all agree, “highly deferential.” Burt v.

12 See Joseph L. Hoffmann & Nancy J. King, Rethinking the Federal Role in State Criminal Justice, 84 N.Y.U. L. REV. 791 (2009) (criticizing the costliness and inefficiency of federal habeas review of state criminal cases); Nancy J. King, Fred L. Cheesman II & Brian J. Ostrom, Final Technical Report: Habeas Litigation in U.S. District Courts, NAT’L CTR. STATE CTS. 59 (2007), https://www.ojp.gov/pdffiles1/nij/grants/219559.pdf (analyzing the statis- tics of all federal habeas cases and explaining that the “[o]verall disposition time per case has increased on average since AEDPA” (emphasis in original)). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 81 of 135

18-12147 JILL PRYOR, J., Dissenting 13

Titlow, 571 U.S. 12, 18 (2013). 13 The majority and I disagree, how- ever, on how that “highly deferential” review should be conducted. Several years ago, our Court fractured over how to apply AEDPA’s highly deferential standard in a case materially indistinct from this one: a case in which a Georgia state habeas court issued a reasoned decision and the Supreme Court of Georgia declined to issue a certificate of probable cause to appeal. A majority of our en banc Court borrowed a standard from a Supreme Court case that applied AEDPA’s deferential review in the absence of any reasoned state court decision, Harrington v. Richter. Applying Richter’s standard, the majority held that in conducting AEDPA review we were not limited to the reasons the state habeas court supplied; in- stead, the Court could imagine what theories supported or “could have supported” the state court’s denial of habeas relief and exam- ine whether “it is possible fairminded jurists could disagree that those arguments or theories are inconsistent with” clearly estab- lished federal law. Wilson v. Warden, 834 F.3d 1227, 1235 (11th Cir. 2016) (en banc) (majority opinion) (quoting Harrington v. Richter, 562 U.S. 86, 102 (2011)). But see id. at 1247–49 (Jill Pryor, J., dissent- ing) (explaining why Richter’s standard did not apply and arguing we should follow Premo v. Moore, 562 U.S. 115 (2011), an AEDPA case in which the Supreme Court had looked through an

13 We review de novo a district court’s rejection of habeas relief on an inef- fective-assistance-of-counsel claim. Pope v. Sec’y, Fla. Dep’t of Corr., 752 F.3d 1254, 1261 (11th Cir. 2014). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 82 of 135

14 JILL PRYOR, J., Dissenting 18-12147

unreasoned decision to a reasoned one and examined the “particu- lar reasons why the state court rejected the claim on the merits” (internal quotation marks omitted)). The Supreme Court overruled our en banc decision, holding that AEDPA “requires” a federal habeas court to look to the last reasoned state court decision and then “train its attention on the particular reasons—both legal and factual—why state courts re- jected a state prisoner’s federal claims,” and then “defer[] to those reasons if they are reasonable.” Wilson v. Sellers, 138 S. Ct. 1188, 1191–92 (2018) (emphasis added) (internal quotation marks omit- ted); see id. at 1195–96 (“[W]e focus[] exclusively on the actual rea- sons given by the lower state court, and we defer[] to those reasons under AEDPA.” (emphasis added) (citing Premo, 562 U.S. at 132)). The Court, noting that “Richter did not directly concern the issue before [it]—whether to ‘look through’ the silent state higher court opinion to the reasoned opinion of a lower court in order to deter- mine the reasons for the higher court’s decision,” held that “Richter does not control here.” Id. at 1195. Without a doubt, then, the Court rejected Richter’s approach in cases with reasoned decisions. Wilson controls this case. Here, as there, a reasoned Georgia state court decision was followed by an unreasoned denial from the Supreme Court of Georgia of a certificate of probable cause to ap- peal. Here, as there, we must “train [our] attention on the particu- lar reasons—both legal and factual”—why the state habeas court rejected Mr. Pye’s ineffective-assistance-of-counsel claim and “de- fer[] to those reasons” only, I repeat, only “if they are reasonable.” USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 83 of 135

18-12147 JILL PRYOR, J., Dissenting 15

Id. at 1191–92. Wilson commands us to review a limited uni- verse—the state habeas court’s seven reasons, which I described above. So our task in this appeal consists of three steps: (1) look at these seven reasons, (2) defer to them if they are reasonable, and (3) if they are not, conduct a de novo review. Reading the majority opinion, you would at first think the majority and I agree about Wilson’s direct application to this case. See Maj. Op. at 12–13. Turn the page, though, and the majority opinion veers into another world entirely, one where “things go the other way.”14 Despite Wilson’s clear dictate that we examine the particular reasons the state habeas court actually provided “and defer[] to those reasons if they are reasonable,” Wilson, 138 S. Ct. at 1192 (emphasis added), and the majority’s apparent acceptance of this rule, see Maj. Op. at 15, the majority opines that we can “consider any potential justification” for the state court’s decision. Maj. Op. at 16 (emphasis added). In other words, according to the majority we can examine what “could have supported” the state court’s decision. Richter, 562 U.S. at 102. This violates step 1, above. And then, the majority says, AEDPA deference applies “only [to] the state court’s resulting decision—not [to] the constit- uent justifications for that decision.” Maj. Op. at 18–19 (emphasis in original) (alteration adopted) (internal quotation marks omit- ted). This violates steps 2 and 3, above. All of it violates Wilson.

14 Carroll, supra note 1, at 4. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 84 of 135

16 JILL PRYOR, J., Dissenting 18-12147

This feels like déjà vu. Compare Wilson, 834 F.3d at 1247–49 (Jill Pryor, J., dissenting), with id. at 1235–36 (majority opinion), overruled sub nom. by Wilson, 138 S. Ct. at 1195–96. 15 How does the majority opinion attempt to justify its nullification of Wilson? In a two-part gambit. First, the majority opinion sidesteps Wilson’s dictate that we focus exclusively on the reasons a state court supplied by imagining two categories of support for a state-court decision: reasons and justifications. Reasons, the majority says, are high-level determina- tions like “the petitioner wasn’t prejudiced by his counsel’s defi- cient performance.” Maj. Op. at 16. Justifications, the majority says, are something more granular—like why the petitioner was not prejudiced. So, the majority surmises, it can marshal new justifica- tions in support of a state habeas court’s disposition because

15 Or, even, déjà vu of déjà vu, because Wilson was not the most recent occa- sion when I argued, unsuccessfully, to my colleagues that the Supreme Court had already resolved an issue for us, only to have the Supreme Court overturn and once again remind us what it has said. Compare Ovalles v. United States, 905 F.3d 1231, 1283 (11th Cir. 2018) (en banc) (Jill Pryor, J., dissenting) (ex- plaining that the Supreme Court had already said that the “language” used in 18 U.S.C. § 924(c)’s residual clause “require[d]” a categorical approach (quot- ing Leocal v. Ashcroft, 543 U.S. 1, 7 (2004))), with id. at 1242, 1244 (majority opinion) (stating that the Supreme Court hadn’t “provide[d] a detailed expla- nation” when it said in Leocal that the § 924(c)’s language “requires” a cate- gorical approach, and holding that a conduct-based approach applied instead), abrogated by United States v. Davis, 139 S. Ct. 2319, 2328 (2019) (reiterating that “It’s not even close; the statutory text [of § 924(c)’s residual clause] com- mands the categorical approach” (citing Leocal, 543 U.S. at 7)). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 85 of 135

18-12147 JILL PRYOR, J., Dissenting 17

justifications are different from reasons, and Wilson said only that we must examine the state court’s reasons. This distinction between reasons and justifications is nonex- istent in the caselaw, and that should come as no surprise. Justifica- tions are not different from reasons, they are reasons. Black’s Law Dictionary defines “justification” as “[a] lawful or sufficient reason for one’s acts or omissions.” Justification, Black’s Law Dictionary (11th ed. 2019) (emphasis added). Merriam-Webster defines the term to mean “the showing in court of a sufficient lawful reason why a party charged or accused did or failed to do that for which he is called to answer,” or “something that constitutes such a rea- son.” Justification, Merriam-Webster Unabridged (emphasis added). The Oxford-English Dictionary defines “justification” as “The action of or result of showing something to be just, right, or reasonable; vindication. Also: the grounds on which this is done; a justifying circumstance; a good reason.” Justification, Oxford-Eng- lish Dictionary (emphasis added). So if a federal court is tasked with reviewing only the state court’s reasons, so too is it tasked with reviewing only its justifica- tions. They are one and the same. Second, the majority opinion casts aside, or diminishes to meaninglessness, its admission that “Wilson instructs us to ‘re- view[] the specific reasons given by the state court and defer[] to those reasons if they are reasonable.’” Maj. Op. at 15 (quoting Wil- son, 138 S. Ct. at 1192). Now, the majority opines, “only” the deci- sion, id. at 18–19 (emphasis in original)—the “you win or you lose” USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 86 of 135

18 JILL PRYOR, J., Dissenting 18-12147

on the claim—gets AEDPA deference. Maybe the majority is shift- ing gears altogether, arguing—inconsistently with its reasons ver- sus justifications nonsense—that we do not defer to reasons at all, only to decisions. I think, though, that the majority is saying some- thing more—that even if we review a state court’s reasons, and even if those reasons represent an “unreasonable application of fed- eral law” or an “unreasonable determination of the facts,” 28 U.S.C. § 2254(d)(1), (2), the state court’s decision nonetheless is worthy of deference so long as we can imagine theories that “‘could have sup- ported’” the decision. Maj. Op. at 19–20 (emphasis in original) (quoting Richter, 562 U.S. at 102). Or, put another way, the major- ity opines that although we can focus “‘exclusively on the actual reasons given by the lower state court, and . . . defer[] to those rea- sons under AEDPA” if they are reasonable, if the actual reasons given are unreasonable, then we should revert to Richter, imagine reasons that would support the ultimate decision, and hold fast to AEDPA deference. Wilson, 138 S. Ct. at 1195–96; see Maj. Op. at 25 n.7. Either way, I’m confounded. If the majority opinion is correct, then Wilson’s look- through rule does no work. Whether the majority is saying that we defer only to the ultimate decision of the lower state court, or that we defer to the ultimate decision despite any wrong-beyond-fair- minded-disagreement reasoning, examining a state court’s reason- ing would be a meaningless, make-work exercise. That is because we could always skip that step and start making up reasons to sup- port the state court’s decision. This Court’s en banc majority in USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 87 of 135

18-12147 JILL PRYOR, J., Dissenting 19

Wilson would have been correct because federal courts would have no need to train their attention on a state court’s reasons when they could just imagine their own, perhaps better, reasons why a claim would fail. In the same vein, the Supreme Court would have had no occasion to take the case, and Wilson would not exist. That seems to be the world the majority is living in, as it clings to Richter16 and expressly relies on the Court’s reasoning in that case, see Maj. Op. at 19–20 (reviving the “could have sup- ported” standard)—a case that simply “does not control” here. Wil- son, 138 S. Ct. at 1195. 17 And there is more. The majority opinion

16 Recently, in a concurring opinion in a federal habeas case, Judge Newsom opined that “there may also be some tension between Richter and the Court’s . . . more recent decision in” Wilson, “which—albeit in a different con- text—seemed to privilege a state court’s ‘reasons’ over its ‘decision.’” Hayes v. Sec’y, Fla. Dep’t of Corr., 10 F.4th 1203, 1215 n.1 (11th Cir. 2021) (Newsom, J., concurring). Although at the time he wrote his concurrence in Hayes Judge Newsom explained that he would “leave for another day and for those higher on the food chain” how to resolve the tension he perceived between Supreme Court decisions, id., it appears that now he has decided we as an inferior Court can overlook the Supreme Court’s discussion and rejection of Richter not in a “different context,” but in the very same context we have here. Id. 17 Contrary to the majority opinion’s characterization of my position, see Maj. Op. at 23–24, I do not suggest Wilson held that a state court’s decision is not entitled to AEDPA deference and that somehow only the reasons in support of that decision are. That view of Wilson cannot be squared with Richter. Ra- ther, Wilson says that when a state court provides reasons, we give AEDPA deference to its decision by examining whether the reasons supporting the de- cision were reasonable. When a state court provides no reasons, we have no reasons to review, and Richter’s rule controls. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 88 of 135

20 JILL PRYOR, J., Dissenting 18-12147

gives short shrift to the entirety of § 2254(d). The majority empha- sizes that AEDPA asks whether a state court’s adjudication of a claim “resulted in a decision.” Maj. Op. at 18. From there, the ma- jority extrapolates that only—or regardless of flaws in a decision’s reasoning—a court’s “resulting decision” should receive deference. The majority’s reading misses half of the text and context. Here is the rest of the story. Under subsection (d), we cannot grant a writ of habeas corpus on a claim a state court decided on the merits “un- less the adjudication of the claim . . . resulted in a decision that was contrary to, or involved an unreasonable application of, clearly es- tablished Federal law,” or “resulted in a decision that was based on an unreasonable determination of the facts.” 28 U.S.C. § 2254(d) (emphasis added). In tasking federal courts with determining whether a decision involved, or was based on, certain egregious errors, the statute directs us to examine how, or why—that is to say, the reasons, if any, for the decision. And those reasons should rule the day “unless” they were “unreasonable,” id.—that is, unless they are unworthy of the deference the statute confers upon them. If the how and the why were unreasonable, the “unless” of the stat- ute is satisfied, no further deference is authorized or warranted.

Nor, to be clear, do I suggest that we can never consider reasons supporting a state habeas court’s decision that the state court did not provide. If we conduct a de novo review of the record, either by assuming that standard applies or after concluding that the state court’s decision does not withstand AEDPA def- erence, then we may marshal our own reasons why a petitioner is not entitled to relief. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 89 of 135

18-12147 JILL PRYOR, J., Dissenting 21

This is what Wilson says in instructing federal courts to “de- fer[] to [a state court’s] reasons if they are reasonable.” Wilson, 138 S. Ct. at 1191–92 (emphasis added). If they are not, the Su- preme Court’s body of AEDPA cases shows, we are “free to pro- vide habeas relief.” 18 See, e.g., Wiggins v. Smith, 539 U.S. 510, 534 (“In deferring to counsel’s decision not to pursue a mitigation case despite their unreasonable investigation, the Maryland Court of Appeals unreasonably applied Strickland. Furthermore, the court partially relies on an erroneous factual assumption. The require- ments for habeas relief established by 28 U.S.C. § 2254(d) are thus satisfied.”). Building on the shaky foundation of its feeble reasons-ver- sus-justifications distinction, outdated view of Richter’s scope, and partial textual analysis, the majority seeks to buttress its defer- ence-at-every-turn holding with our previous decisions. But those decisions stand only for the wholly unremarkable principle that a state court’s decision that otherwise is reasonable is not unreason- able simply because it fails to discuss every fact or argument a pe- titioner advances. Maj. Op. at 16–17 (citing Lee v. Comm’r, Ala.

18 Wiggins v. Smith, 539 U.S. 510, 542 (2003) (Scalia, J., dissenting) (character- izing the Court’s earlier decision in Williams v. Taylor, 529 U.S. 362 (2000): “Williams had surmounted § 2254(d)’s bar to habeas relief because we held that the Virginia Supreme Court’s analysis with respect to . . . preju- dice . . . was both ‘contrary to,’ and ‘an unreasonable application of,’ our clearly established precedents,” so the Court was “free to provide habeas re- lief . . . .” (emphasis omitted)). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 90 of 135

22 JILL PRYOR, J., Dissenting 18-12147

Dep’t of Corr., 726 F.3d 1172 (11th Cir. 2013)); Meders v. Warden, Ga. Diagnostic Prison, 911 F.3d 1335 (11th Cir. 2019)). It does not follow from this principle, however, that we can rely on and give deference to reasons never mentioned by the state habeas court to conclude that the decision withstands AEDPA deference. The for- mer merely reflects that we do not “engage[] in a line-by-line cri- tique of the state court’s reasoning.” Meders, 911 F.3d at 1350. The latter violates Wilson. See id. at 1349 (“[Wilson’s holding] does not mean we are to flyspeck the state court order or grade it. What it means is we are to focus not merely on the bottom line ruling of the decision but on the reasons, if any, given for it.” (emphasis added)). 19

19 No matter how many times the majority opinion says so, Meders does not support its reading of Wilson. The language from Meders the majority opinion cites, see Maj. Op. at 16, 21–22, 25, all concerned this same simple idea: that we do not flyspeck a state court’s opinion. And, of course, how we did things before the Supreme Court decided Wilson in 2018, see Maj. Op. at 16–17, 20– 21 (citing cases that pre-date Wilson), does not answer how we do things in light of Wilson. The majority opinion also relies heavily on this Court’s post-Wilson decision in Whatley v. Warden, 927 F.3d 1150 (11th Cir. 2019). There, in a case that arose in a different procedural posture than this case and Wilson, a panel stated that “our review is not limited to the reasons the [state] [c]ourt gave in its analysis,” id. at 1178, and that we “instead focus on [the state court’s] ultimate conclusion,” id. at 1182 (internal quotation marks omitted). The Whatley panel never mentioned Wilson. Nor did it mention this Court’s earlier post- Wilson decision in Meders, 911 F.3d at 1349, where we correctly applied Wil- son. My former colleague Beverly Martin asked this Court to rehear Whatley en banc because it conflicted with Wilson and Meders. See Whatley v. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 91 of 135

18-12147 JILL PRYOR, J., Dissenting 23

The majority opinion quotes our decision in Lee v. Commis- sioner for the proposition that we may “examine what other ‘im- plicit findings’ the state court could have made in its denial of a federal claim.” Lee, 726 F.3d at 1223; see Maj. Op. at 16–17. Di- vorced from the context of the lengthy Lee opinion, this language may seem alluring. But in Lee, a case we decided before the Su- preme Court decided Wilson, the petitioner argued that the state postconviction court’s decision involved an unreasonable applica- tion of clearly established law “because that opinion did not men- tion or discuss every relevant fact or argument he offered in sup- port of his . . . claim.” Lee, 726 F.3d at 1210–11. We held, as we have numerous times before and since, that a state court is not re- quired to show its work. Id. at 1211–12. We further explained that there may be some “implicit findings” of a state court “which, though unstated, are necessary to that ruling.” Id. at 1217 (empha- sis added) (internal quotation marks omitted). Those necessary im- plicit findings, we said, “are entitled to deference under § 2254(d).” Id. (internal quotation marks omitted). We gave as an example a Batson challenge: if we know from the record that defense counsel failed to rebut the prosecutor’s race-neutral explanation for a strike

Warden, 955 F.3d 924, 924–27 (11th Cir. 2020) (Martin, J., dissenting from de- nial of reh’g en banc). She was right. In any event, because Whatley conflicted with Meders, under our prior panel precedent rule, only Meders is good law on this point. See United States v. Levy, 379 F.3d 1241, 1245 (11th Cir. 2004) (“[W]here there is conflicting prior panel precedent, we follow the first in time.”). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 92 of 135

24 JILL PRYOR, J., Dissenting 18-12147

and that the trial court ultimately ruled the strike proper, we can infer “that the trial court implicitly found the prosecutor’s race- neutral explanations to be credible, thereby completing step three of the Batson inquiry.” Id. at 1220 (internal quotation marks omit- ted). That necessary finding, though implicit, is entitled to AEDPA deference. Id. The majority opinion extrapolates from Lee that it can con- sider unlimited reasons unstated by state habeas courts. But even setting aside the illogic of this extrapolation—that because we should not flyspeck state court opinions, we can violate Wilson’s express dictates—the reasons the majority opinion invents to prop up the state habeas court’s decision were in no way “necessary” to the state habeas court’s ruling. See, e.g., Maj. Op. at 38 n.14 (citing the supposed lack of evidence corroborating affidavit testimony, a reason the state habeas court did not provide and that is not neces- sary to its rejection of the affidavits), 53–54 & n.22 (holding that it was reasonable for the state habeas court to conclude that evidence to rebut future dangerousness would not have been substantially likely to change the outcome of Mr. Pye’s sentencing “for three reasons,” two of which the state habeas court did not mention, and were not necessary to the state court’s conclusion). Thus, they are not “implicit findings” of the type Lee contemplated. To read Lee any more broadly—that is, to read Lee to permit what the majority opinion undertakes—is to violate Wilson. The majority opinion laments that Wilson’s rule as I see it “would incentivize state courts to issue unreasoned, summary USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 93 of 135

18-12147 JILL PRYOR, J., Dissenting 25

decisions as a means of guaranteeing maximum AEDPA defer- ence.” Maj. Op. at 22. Ironically, the en banc majority in Wilson resisted the look-through rule on the opposite ground, that it “does nothing less than impose an opinion-writing standard.” Wilson, 834 F.3d at 1238–39. Setting that aside, “the matter is empirical,” and the majority opinion has no data to back up its concern. Wil- son, 138 S. Ct. at 1197. But to the extent such an incentive exists, it has existed since the Supreme Court decided Richter in 2011. See Richter, 562 U.S. at 99 (“The issuance of summary dispositions in many collateral attack cases can enable a state judiciary to concen- trate its resources on the cases where opinions are most needed.”). Further, state courts stand to be reversed by more than one court, and on direct review it seems a state court’s incentive would be to show its work. Even as it pertains to federal review, I trust that state court judges will value thoroughness for the parties’ and the pub- lic’s sake. And setting aside unsupported empirical matters, defer- ring to the reasons a state court supplied “is more likely to respect what the state court actually did” than deferring to reasons the court never supplied and perhaps never even considered. Wilson, 138 S. Ct. at 1197. 20

20 The majority opinion also notes that “any state court’s written opinion is necessarily ‘partial.’” Maj. Op. at 22. Undoubtedly. But this is just another way to say that a court is not required to discuss every fact or argument. It does not answer the question here, which is how we apply AEDPA deference to “what the state court actually did.” Wilson, 138 S. Ct. at 1197. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 94 of 135

26 JILL PRYOR, J., Dissenting 18-12147

Lastly, the majority says my position conflicts with “the overwhelming consensus position” of our sister circuits. Maj. Op. at 21. In support, the majority opinion cites a recent Fifth Circuit case collecting (the majority fails to mention) pre-Wilson decisions of the First, Second, Sixth, Seventh, Eighth, and Tenth Circuits, as well as one from our Court. Sheppard v. Davis, 967 F.3d 458, 467 n.5 (5th Cir. 2020). 21 Of course, we are obliged to follow Wilson, as intervening Supreme Court precedent, not decisions that pre- dated it. Just as importantly, the majority opinion fails to acknowledge that several of these circuits, and others whose deci- sions Sheppard did not cite, have refined their approach in the wake of Wilson. See, e.g., Porter v. Coyne-Fague, 35 F.4th 68, 74–75 (1st Cir. 2022) (citing Wilson’s requirement that federal courts defer to the “specific reasons” given by the state court, examining those rea- sons, concluding that they did not withstand AEDPA deference, and applying de novo review) 22; Scrimo v. Lee, 935 F.3d 103, 111–

21 To be sure, the Fifth Circuit, pre-Wilson, applied the rule the majority ap- plies today. In Sheppard, the Fifth Circuit assumed without deciding that Wil- son overruled its prior precedent. 967 F.3d at 467–68. The majority opinion does not mention this nuance, either. 22 The majority opinion correctly notes that in Porter the First Circuit “con- sidered whether there was another ‘possible explanation of the state court’s decision.’” Maj. Op. at 26–27 n.9 (quoting Porter, 35 F.4th at 79). But that was only because, as the court noted at the outset, the “relevant passages of the state court’s opinion are terse to the point of obscuring the precise mechanics of its reasoning” and because of their ambiguity could be read in two different USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 95 of 135

18-12147 JILL PRYOR, J., Dissenting 27

12 (2d Cir. 2019) (“consider[ing] the rulings and explanations of the trial judge,” citing Wilson)23; Coleman v. Bradshaw, 974 F.3d 710, 719 (6th Cir. 2020) (citing Wilson and stating that “AEDPA requires this court to review the actual grounds on which the state court

ways. Porter, 35 F.4th at 77. The court nevertheless concluded that “the state court decision—depending on how it is read—either unreasonably applies Bat- son’s second step or is premised on an unreasonable determination of the facts,” so “there is no need to identify which of these roads the state court traveled because both of them lead to the same destination.” Id. “Either way, the state supreme court’s decision is not entitled to deference under AEDPA.” Id. The Porter court evaluated the state court’s reasoning under both alterna- tive readings of that reasoning before proceeding to de novo review. That is not at all akin to the approach the majority opinion advances here. 23 The majority opinion’s rebuttal to my citation of Scrimo misses the mark as well. The majority opinion suggests that in its effort to determine whether the state court’s order withstood AEDPA deference, the Second Circuit in Scrimo looked for other reasons why the state court may have excluded chal- lenged testimony. See Maj. Op. at 26–27 n.9. Not so. The Scrimo court asked “whether the [w]itnesses’ testimony could have been excluded on other grounds” because a negative answer to that question was essential to meet the petitioner’s burden on federal habeas to prove that the trial court’s exclusion of the testimony had a substantial and injurious effect or influence on the jury’s verdict. Scrimo, 935 F.3d at 115–16 (citing Brecht v. Abrahamson, 507 U.S. 619, 638 (1993)). The state trial court ruling that the Second Circuit was reviewing had not addressed whether the error was harmless, so no AEDPA deference applied to the harmlessness determination. See Brown v. Davenport, 142 S. Ct. 1510, 1517 (2022) (requiring a federal habeas petitioner to satisfy Brecht and AEDPA only “[a]fter a state court determines that an er- ror at trial did not prejudice a criminal defendant”). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 96 of 135

28 JILL PRYOR, J., Dissenting 18-12147

relied”) 24; Winfield v. Dorethy, 956 F.3d 442, 454 (7th Cir. 2020) (“Having found the state court’s ‘specific reasons’ for denying

24 The majority opinion notes that one Sixth Circuit judge in a later opinion characterized the Coleman decision as not “constrain[ing] its analysis to the exact reasons the state court discussed.” Maj. Op. at 26–27 n.9 (quoting Thompson v. Skipper, 981 F.3d 476, 485 (6th Cir. 2020) (Nalbandian, J., con- curring)). Judge Nalbandian’s reading of Coleman, for which he did not secure a majority, does not convince me that Sixth Circuit does not share my view of Wilson. In Coleman, the petitioner pursued a Brady claim based on the prosecution’s alleged failure to disclose exculpatory evidence that another person, Sapp, had confessed to the murder for which Coleman had been convicted. Cole- man, 974 F.3d at 716. In support, Coleman pointed to a letter from Sapp in which he admitted to a killing and an affidavit Coleman’s postconviction counsel prepared for Sapp in which Sapp admitted to killing the victim. Id. at 717. The state court in Coleman rejected the petitioner’s Brady claim be- cause the affidavit lacked credibility and the letter, which did not name the victim, was too indefinite to be material. Id. at 718. Reviewing the state court’s decision, the Coleman panel first concluded that the state habeas court “reasonably determined that the Sapp letter was not material.” Id. at 719. This was precisely the reason the state court had employed as to the let- ter. Second, addressing the affidavit, the Coleman panel noted, as a preliminary matter, that “Coleman does not contend that the state should have disclosed the Sapp affidavit.” Id. The affidavit was not Brady material because it was prepared by Coleman’s counsel. Rather, the affidavit was “evidence of al- leged Brady evidence (i.e., evidence of Sapp’s [] confession).” Id. The panel then addressed the state court’s determination “that the Sapp [a]ffidavit lacked any credibility” and “the necessary implication . . . that Coleman failed to establish that Sapp’s [] confession ever occurred.” Id. The Coleman panel’s preliminary observation that the affidavit was not itself Brady evidence was merely a point of clarification (that the evidence USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 97 of 135

18-12147 JILL PRYOR, J., Dissenting 29

relief, the next question is whether that explanation was reasonable thereby requiring our deference.” (citing Wilson, 138 S. Ct. at 1188)) 25; see also Richardson v. Kornegay, 3 F.4th 687, 697–98 (4th Cir. 2021) (“Sitting as a federal habeas court, we must identify ‘the particular reasons—both legal and factual—why state courts re- jected a state prisoner’s federal claims.’ And the particular reason for rejecting this claim was that the trial court did not abuse its dis- cretion in excluding [the expert witness’s] testimony.” (quoting Wilson, 138 S. Ct. at 1191–92)) 26; Kipp v. Davis, 971 F.3d 939,

allegedly was evidence of Brady material, not Brady material itself); it was not a reason why the state court reasonably rejected Coleman’s Brady claim. And the panel’s observation that if the affiant was not credible, then what he professed was not true, is the same kind of necessary “implicit finding[]” our Court has long swept within the purview of AEDPA deference. See Lee, 726 F.3d at 1217. In sum, Coleman does not suggest that a federal court may marshal any rea- sons that “could have supported” a state court’s decision in deferring to that decision. 25 The Seventh Circuit in Winfield did not, as the majority opinion says, “de- cline[] to decide exactly how § 2254(d) applied.” Maj. Op. at 26–27 n.9. The court said exactly how § 2254(d) applied, in the language I have quoted above. Rather, the court accepted for the sake of argument that the district court cor- rectly concluded that AEDPA deference did not apply, holding that the district court erred on a different ground. Winfield, 956 F.3d at 455. 26 Contrary to the majority’s claim that Richardson supports going beyond the reasons a state court articulates, see Maj. Op. at 26–27 n.9, in Richardson the Fourth Circuit hewed not only to the state court’s “particular reason”— no abuse of discretion in excluding the expert’s testimony—but also to the state court’s “rationale supporting finding no abuse of discretion,” which the USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 98 of 135

30 JILL PRYOR, J., Dissenting 18-12147

948–60 (9th Cir. 2020) (reviewing whether the stated reasons of a state habeas court were reasonable, concluding they were not, ap- plying de novo review, and granting petitioner relief) 27. Respect- fully, then, the great weight of authority is not at all in the majority opinion’s favor. 28 Quite to the contrary.

Richardson court enumerated and discussed, without seeking reasons outside the scope of the state court’s order. Richardson, 3 F.4th at 697–98. 27 The majority’s suggestion that although the Ninth Circuit “has employed [my] sweeping rule,” it has done so for an entirely different reason, Maj. Op. at 26–28 n.9, overlooks the most basic of points: Wilson has subsumed pre- Wilson reasoning. The Ninth Circuit has recognized as much. See Kipp, 971 F.3d at 953 n.10 (“The Warden argues that there were several additional [reasons why the state court’s decision was correct] . . . but we may look only to the reasoning of the California Supreme Court.” (citing Wilson, not a pre- vious line of Ninth Circuit cases)). 28 The Third Circuit has not cited Wilson in a precedential opinion. But in unpublished opinions, it has applied Wilson the way I do here. See, e.g., Gibbs v. Admin’r N.J. State Prison, 814 F. App’x 686, 689–91 & n.6 (3d Cir. 2020) (unpublished) (citing Wilson and deferring only to the two reasons why the state habeas court concluded counsel had not performed deficiently and the one reason why it concluded there was no prejudice: (1) that either counsel was unaware of allegedly biased jurors until after trial and so did not defi- ciently fail to object during voir dire; (2) or, alternatively, if counsel knew of the biased jurors, those jurors could have been an advantage to the defense and so any failure to object during voir dire was strategic; and (3) the allega- tions of bias were too vague to amount to sufficient evidence of bias). The same can be said of the Tenth Circuit. See Straub v. Goodrich, 842 F. App’x 263, 267–69 (10th Cir. 2021) (unpublished). The Eighth Circuit has not cited Wilson in any reported case. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 99 of 135

18-12147 JILL PRYOR, J., Dissenting 31

This side of the mirror, Wilson holds true: as a federal court constrained by AEDPA, we must focus exclusively on the reasons actually given by the state habeas court and defer to those reasons, and those reasons alone, under AEDPA. If those reasons are “that wrong,” Maj. Op. at 29 (emphasis in original), then the decision is unworthy of AEDPA deference. 29

29 And no, I do not seek to “camouflage the breadth” of my position on def- erence to the reasons a state court supplied. See Maj. Op. at 23–24 n.6. Nor do I “insist[] that Wilson changed how AEDPA applies to all reasoned decisions, regardless of procedural posture,” or that “Wilson sub silentio revolutionized AEDPA’s application to all state-court decisions.” Id. Wilson expressly (not sub silentio) reminded us (rather than revolutionized) how AEDPA applies to reasoned decisions, characterizing the inquiry as “straightforward”: Deciding whether a state court’s decision “involved” an unrea- sonable application of federal law or “was based on” an unrea- sonable determination of fact requires the federal habeas court to train its attention on the particular reasons—both legal and factual—why state courts rejected a state prisoner’s federal claims, and to give appropriate deference to that decision. This is a straightforward inquiry when the last state court to decide a prisoner’s federal claim explains its decision on the merits in a reasoned opinion. In that case, a federal habeas court simply reviews the specific reasons given by the state court and defers to those reasons if they are reasonable. We have affirmed this approach time and time again. Wilson, 138 S. Ct. at 1191–92 (emphasis added) (internal quotation marks and citations omitted). Applying that logic to a reasoned decision layered beneath an unreasoned one, Wilson confirmed that the approach it had affirmed “time and time again” applied in the “look through” context. Id. at 1192. It is the Richter “could have supported” approach that is the exception to the rule, USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 100 of 135

32 JILL PRYOR, J., Dissenting 18-12147

Before leaving my discussion of AEDPA, I echo Judge Jor- dan’s concerns about the majority opinion’s “clarif[ication],” Maj. Op. at 13, of the interplay between § 2254(d)(2), which permits a federal court to grant a state prisoner a writ of habeas corpus if the relevant state court decision was “based on an unreasonable deter- mination of the facts in light of the evidence presented in the State court proceeding,” and § 2254(e)(1), which provides, “In a proceed- ing instituted by an application for a writ of habeas corpus by a per- son in custody pursuant to the judgment of a State court, a deter- mination of a factual issue made by a State court shall be presumed to be correct. The applicant shall have the burden of rebutting the presumption of correctness by clear and convincing evidence.” See Jordan Concurring Op. at 1. Despite the Supreme Court’s having repeatedly dodged the question of the precise interplay between the two statutes, a split among the circuits in how to interpret them, and no briefing or argument from either party on the ques- tion, the majority opinion—citing to a concurrence written by its author—declares that habeas relief is warranted only if a petitioner proves both (1) by clear and convincing evidence that at least one individual state court fact-finding was erroneous and (2) that the

because it applies only when there is no reasoned decision to look to. It is hard to imagine how the Supreme Court could have been any clearer. Would we expect it to have thought to specify that we are not free to add “justifications” of our own devising to the “particular” and “specific reasons” to which we defer if reasonable? I think not. The majority opinion’s attempt to overcom- plicate the inquiry smacks of obfuscation. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 101 of 135

18-12147 JILL PRYOR, J., Dissenting 33

error or errors were important enough that the state court’s deci- sion was based on the finding or findings and was unreasonable as a result. See Maj. Op. at 14–15 (citing Hayes v. Sec’y, Fla. Dep’t of Corr., 10 F.4th 1203, 1224–25 (11th Cir. 2021) (Newsom, J., concur- ring)). Given the irregularity of deciding an issue of such im- portance without any notice to or briefing by the parties, this should strike anyone paying attention as odd at best. See United States v. Sineneng-Smith, 140 S. Ct. 1575, 1579, 1581 (2020) (“[C]ourts normally decide only questions presented by the par- ties. . . . No extraordinary circumstances justified the panel’s take- over of the appeal.” (alteration adopted) (internal quotation marks omitted)). Worse, the majority opinion uses its newly crafted rule to deny Mr. Pye relief, even though he had no chance to argue against it or that he should prevail under it. This is wrong. Nevertheless, I believe Mr. Pye should prevail even if the majority opinion’s reading of the interplay between § 2254(d)(2) and (e)(1) is correct. So I will assume only for purposes of this dis- sent that the majority opinion is correct on this point.30

30 The panel opinion in this case, following the lead of previous decisions of our Court that did not define precisely the interplay between the two subsec- tions, explained that habeas relief is unwarranted unless it is, in relevant part, “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding[s]” and that state court factual deter- minations “are entitled to a presumption of correctness unless the petitioner rebuts that presumption by clear and convincing evidence.” Pye v. Warden USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 102 of 135

34 JILL PRYOR, J., Dissenting 18-12147

Even so, the majority opinion’s construction of the two sub- sections of § 2254, when paired with its backwards reading of Wil- son, produces disastrous results. That is because, instead of asking whether any of the stated reasons supporting the state habeas court’s decision represented a clearly erroneous finding of fact and then asking whether those reasons were sufficiently important to make an order “based on” them “unreasonable,” we as federal

Ga Diagnostic Prison, 853 F. App’x 548, 558 (11th Cir. 2021) (internal quota- tion marks omitted). We elaborated that relief is not warranted unless a peti- tioner proves that “the state court’s findings lacked even fair support in the record.” Id. at 559 (quoting Rose v. McNeil, 634 F.3d 1224, 1241 (11th Cir. 2011)); see also, e.g., Jenkins v. Comm’r, Ala. Dep’t of Corr., 963 F.3d 1248, 1263–64 (11th Cir. 2020) (citing Rose in an AEDPA case); Smith v. Comm’r, Ala. Dep’t of Corr., 924 F.3d 1330, 1336–37 (11th Cir. 2019) (same); Terrell v. GDCP Warden, 744 F.3d 1255, 1268 (11th Cir. 2014) (same). The majority opinion, pointing out that Rose applied pre-AEDPA law, rules that in passing AEDPA Congress “eliminated and replaced the fair-support-in- the-record standard” Maj. Op. at 14. Now, the majority opinion says, the standard is § 2254(e)(1)’s “presumed to be correct” absent “clear and convinc- ing evidence” standard. Id. (internal quotation marks omitted). Although cor- rect that Rose was pre-AEDPA, the majority opinion fails to explain how Rose’s standard, when articulated alongside the “clear and convincing” burden of § 2254(e)(1), is materially different. If a petitioner proves that a factual de- termination lacks any fair support in the record, how has he not also proved by clear and convincing evidence that the factual determination was incorrect? Put another way, if the ultimate question is whether a petitioner has “show[n] that the state court’s decision is so obviously wrong that its error lies beyond any possibility for fairminded disagreement,” Shinn v. Kayer, 141 S. Ct. 517, 523 (2021) (emphasis added) (internal quotation marks omitted), then how is it different to inquire whether fair support in the record is lacking? I fail to see the daylight between the two. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 103 of 135

18-12147 JILL PRYOR, J., Dissenting 35

courts can bury reasons that flunk AEDPA deference under a mountain of unstated and un-relied-upon reasons that would with- stand AEDPA deference. And then, even if one or a handful of the state court’s reasons were based on clearly erroneous fact-findings, amongst any number of made-up reasons that could have sup- ported the state court’s decision, those findings will never be prom- inent enough to meet the majority opinion’s test. 31 I have journeyed far only to return to where I began: AEDPA requires that we must be highly deferential of state court decisions. In cases such as this one, where the state habeas court provided a reasoned decision, we review exclusively the reasons the state habeas court gave for denying relief, deferring to those reasons under § 2254(d). If those reasons fail § 2254(d)’s test, then we are “unconstrained by § 2254’s deference and must undertake a de novo review of the record.” Daniel v. Comm’r, Ala. Dep’t of Corr., 822 F.3d 1248, 1260 (11th Cir. 2016) (internal quotation marks omitted); see also Cooper v. Sec’y, Dep’t of Corr., 646 F.3d 1328, 1353 (11th Cir. 2011) (“When a state court unreasonably

31 Of course, where Richter’s “could have supported” standard does ap- ply—that is, when there is no reasoned state court decision to review—federal courts may do precisely the kind of hypothetical inquiry the majority opinion undertakes in this case. There are reasons for this difference: the “principles of federalism and comity that underlie federal collateral review.” Wilson, 834 F.3d at 1248 (Jill Pryor, J., dissenting). When a state court gives reasons for denying a state prisoner postconviction relief, a federal court should not give the back of its hand to the state court’s given reasons. When a state court does not give reasons, a federal court has no reasons to respect, only the judgment. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 104 of 135

36 JILL PRYOR, J., Dissenting 18-12147

determines the facts relevant to a claim, we do not owe the state court’s findings deference under AEDPA, and we apply the pre–AEDPA de novo standard of review to the habeas claim.” (in- ternal quotation marks omitted)). In the next section, I explain why the state habeas court’s decision was unreasonable under § 2254(d), and then in the follow- ing section, I conduct a de novo review of the merits of Mr. Pye’s penalty phase ineffective-assistance-of-counsel claim. III. ANALYSIS UNDER AEDPA Mr. Pye claims that his trial counsel was ineffective in failing to investigate and present evidence about his traumatic childhood and adolescence, mental health problems, and low intellectual functioning. He also claims that his counsel failed to investigate and present evidence to rebut the State’s claim of future dangerousness. And, he argues, there is a reasonable probability that, had the jury heard this evidence, at least one juror would have voted against a death sentence. See O.C.G.A. § 17-10-31(c). Mr. Pye was required to show that his trial counsel rendered deficient performance and that the deficient performance preju- diced his defense. Strickland, 466 U.S. at 686, 688, 694. Counsel’s deficient performance causes prejudice when “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Id. at 694. “A rea- sonable probability is a probability sufficient to undermine confi- dence in the outcome.” Id. In determining whether there is a USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 105 of 135

18-12147 JILL PRYOR, J., Dissenting 37

reasonable probability of a different result, “we consider ‘the total- ity of the available mitigation evidence—both that adduced at trial, and the evidence adduced in the habeas proceeding’—and ‘re- weig[h] it against the evidence in aggravation.’” Porter v. McCollum, 558 U.S. 30, 41 (2009) (alteration in original) (quoting Williams v. Taylor, 529 U.S. 362, 397–98 (2000)). Because AEDPA applies to Mr. Pye’s claim, the specific question we must ask is whether the state habeas court’s determination that he failed to demonstrate prejudice “involved” or was “based on,” § 2254(d)(1), (2), an “error [that] lies beyond any possibility for fairminded disa- greement.” Shinn v. Kayer, 141 S. Ct. 517, 520 (2020) (internal quo- tation marks omitted). Training our attention on the reasons the state habeas court actually supplied in its no-prejudice determination, we must re- view these seven reasons: (1) there is no reasonable probability that evidence of low intellectual functioning would have swayed one juror; (2) there is no reasonable probability that evidence rebutting the State’s case of future dangerousness would have swayed one juror because prison records showed instances of insubordination and aggression; and, as to evidence of Mr. Pye’s family background, the evidence would not have swayed one juror because (3) the af- fidavit testimony was of little value due to artful drafting; (4) Mr. Pye’s family was unwilling to cooperate at the time of trial; (5) there was no nexus between Mr. Pye’s background and the crime he committed; (6) the extensive aggravating evidence; and (7) that because Mr. Pye was 28 at the time of the crimes, the USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 106 of 135

38 JILL PRYOR, J., Dissenting 18-12147

aggravating evidence was extensive, and the evidence of his guilt was overwhelming, remorse was likely to play better than excuses. Several of these reasons were an unreasonable application of clearly established law or represented an unreasonable determina- tion of facts in light of the state court record. See Pye v. Warden Ga. Diagnostic Prison, 853 F. App’x 548 (11th Cir. 2021). Without reiterating everything in the panel opinion, and because the unrea- sonableness of three reasons in particular renders the state court’s decision unworthy of AEDPA deference, I will limit my discussion in this section to reasons 3, 4, and 7. A. Value of affidavit testimony on mitigation I begin with reason number 3. The state habeas court gave two reasons for discounting the mitigating evidence in the affidavit testimony Mr. Pye’s state postconviction counsel amassed. First, the court said, postconviction affidavits usually are unpersuasive, in part because they are artfully drafted long after the fact. Second, the court opined that the affidavits in this case reflected artful draft- ing because a few of them contained perceived inconsistencies. I will take these in reverse order. The state habeas court’s finding that the affidavits in this case reflected artful drafting due to inconsistencies it perceived was unreasonable in light of the record—or, as the majority sometimes says, “clearly and convincingly erroneous.” See, e.g., Maj. Op. at USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 107 of 135

18-12147 JILL PRYOR, J., Dissenting 39

37. 32 The state habeas court focused on four mitigation affidavits. The first affidavit was by Mr. Pye’s brother, Curtis Pye. The court recounted that “Curtis Pye testified . . . ‘No one talked to me . . . before [Mr. Pye’s] trial. Johnny Mostiler and his assistant Dewey [Yarbrough] know me . . . He didn’t get in touch with me.’” Doc. 20-40 at 65 (purporting to quote Curtis Pye’s affidavit). “How- ever, Mr. Mostiler’s billing records in Petitioner’s case reflect that Mr. Mostiler interviewed Curtis Pye for one hour approximately one month prior to trial.” Id. The ellipses are the state court’s, not mine. In concluding that Curtis had apparently lied in his affidavit about his contact with Mr. Mostiler, the state habeas court omitted with ellipses a key portion of Curtis’s testimony. Curtis did not tes- tify that no one talked to him before Mr. Pye’s trial—the contradic- tion the state habeas court purported to identify given Mr. Mostiler’s billing records. Rather, after describing the Pye fam- ily and Mr. Pye’s upbringing, he testified that “[n]o one talked to me about any of this before Willie James’s trial. . . . [Mr. Mostiler] didn’t get in touch with me or ask me any questions about the house Willie James was raised in or what he was like as a child.” Doc. 16-24 at 83 (emphasis added). After testifying about the

32 I mostly stick with the reasonable/unreasonable terminology, although I believe that Mr. Pye has also rebutted a presumption of the facts’ correctness by clear and convincing evidence. See 28 U.S.C. § 2254(e)(1). Perhaps one day the Supreme Court will clear up the confusion about the interplay between 28 U.S.C. §§ 2254(d) and 2254(e)(1). USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 108 of 135

40 JILL PRYOR, J., Dissenting 18-12147

mitigating circumstances in Mr. Pye’s childhood and adolescence, Curtis testified that no one talked to him “about any of this”—“this” being the circumstances to which he had just testi- fied—before trial. He again clarified that no one asked him ques- tions about Mr. Pye’s childhood and upbringing. Reading his iso- lated statements about contact with the defense team in context with his entire affidavit makes clear that Curtis never denied meet- ing with Mr. Mostiler altogether. When read “in light of the evi- dence presented in the State court proceeding,” 28 U.S.C. § 2254(d)(2), as a whole—that is to say, in context with the entire affidavit and entire factual record, see Dunn v. Reeves, 141 S. Ct. 2405, 2412 (2021)—it is to me beyond fairminded disagreement that the state court’s finding of an inherent conflict was in error. The second affidavit the state habeas court addressed was Mr. Pye’s brother Ricky Pye’s. According to the state habeas court, Ricky “testified . . . ‘I never spoke to Mostiler about what to say [at trial], and he didn’t meet with me or ask me any questions before my turn for testimony.’” Doc. 20-40 at 66 (alteration in original) (quoting Ricky Pye’s affidavit). But, the court continued, “[t]he af- fidavit makes no mention of Mr. Mostiler’s one hour interview with him, also approximately one month prior to trial.” Id. The state habeas court’s statements are literally true, but they lend no evidentiary support to the court’s finding that the affidavit was mis- leading. Ricky testified that no one talked to him “about what to say”—what he would testify to—not that no one talked to him at all. Doc. 16-24 at 99; see also id. at 100 (Ricky Pye testifying: “We USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 109 of 135

18-12147 JILL PRYOR, J., Dissenting 41

had it real tough growing up and Mr. Mostiler and Dewey never asked about that.”). Third was the affidavit of Mr. Pye’s mother, Lolla Mae Pye. The state habeas court found misleading her testimony that “[n]o one took the time to talk to me about all anything before Willie’s trial,” given that Mr. Yarbrough’s testimony and Mr. Mostiler’s bill- ing records showed otherwise. Doc. 16-24 at 97; see Doc. 20-40 at 66. Once again, the state habeas court tore a snippet of affidavit testimony from its context. Before the passage the state habeas court quoted, Lolla Mae testified extensively about the mitigating circumstances in Mr. Pye’s background. The passage in Lolla Mae’s affidavit that immediately follows the one the state habeas court quoted, together with her lengthy mitigation testimony, leaves no room for doubt that Lolla Mae meant she was never asked about mitigating circumstances: Nobody ask me all about how I grew up, how I came to be married to Ernest, and how I raised Willie and my other children. I would have been willing to talk about my life with Willie James’s lawyer or investiga- tor, or with any doctor or psychologist working on his case. I would have told about all the things I de- scribed here, and testified to the jury about them if they wanted me to. Doc. 16-24 at 97 (emphasis added). In context, there is no mislead- ing statement. Lolla Mae stated straightforwardly that she was not asked about the mitigating circumstances to which she had just USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 110 of 135

42 JILL PRYOR, J., Dissenting 18-12147

testified. The state habeas court’s finding is unreasonable in light of the evidence before it. The state habeas court found that the fourth affidavit, that of social worker and truancy officer Arthur Lawson, reflected artful drafting. Mr. Lawson initially testified: “I showed up at the home to find [Lolla Mae] intoxicated on many visits. This was equally true when she was pregnant.” Doc. 16-24 at 61. He later submitted a second affidavit in which he clarified how he knew she was intox- icated: “when I visited the home there were indications that she had been drinking by the way she spoke and her general behavior. This was equally true when she was pregnant.” Doc. 20-6 at 17. Mr. Lawson’s clarification of the basis for his opinion that Lolla Mae often was intoxicated, if indicative of anything, suggests the very opposite of artful drafting. In tweaking his testimony, Mr. Lawson ensured that the evidence he provided did not over- reach the limits of his personal knowledge. In my mind it is beyond fairminded disagreement that Mr. Lawson’s slight clarification re- flected a desire for accurate conveyance of personal knowledge, not artful drafting. Based on these unreasonable findings of fact about four of the 24 affidavits containing mitigation evidence, the state habeas court decided to discount all of the affidavit evidence about Mr. Pye’s family background. Doc. 20-40 at 66. Here the majority opinion acknowledges that “for many of the affidavits that speak to [Mr.] Pye’s childhood neglect and abuse, neither the state court nor the State have offered specific reasons to doubt their truth besides USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 111 of 135

18-12147 JILL PRYOR, J., Dissenting 43

the general concern with ‘artfully drafted’ affidavit testimony col- lected many years after trial.” Maj. Op. at 36. And this is the point: without any evidence in the record to demonstrate artful drafting, the state habeas court could not reasonably have made a finding that the affidavits were artfully drafted. 33 And the state habeas court could not then have a solid foundation upon which to base its decision to discount the evidence contained in these affidavits. The majority opinion maintains that there was evidence of artful drafting because “there was substantial uniformity across the affidavits” in describing Mr. Mostiler’s failure to discuss Mr. Pye’s background with them and their willingness to testify at sentencing had they been asked to do so. Maj. Op. at 37. I agree that substantial uniformity of affidavits may be evidence of artful drafting, though the affidavits must be viewed as a whole and in context. But there are three problems with the majority opinion’s conclusion here. First, the state habeas court never cited the similarity between the affidavits as a reason for finding they were artfully drafted. So, un- der Wilson, we do not consider that reason when examining whether the state court’s errors were beyond the realm of fair- minded disagreement. Wilson, 138 S. Ct. at 1191–92, 1195–96.

33 I do not, as the majority accuses, suggest that “extrinsic corroborating evi- dence” of artful drafting “is required” to make this determination. Maj. Op. at 38 n.13. The state court could have used as evidence the affidavits themselves. But, as the majority opinion acknowledges, the state court did not point to any such evidence. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 112 of 135

44 JILL PRYOR, J., Dissenting 18-12147

Second, “substantial uniformity” is a stretch. These were not boilerplate affidavits that all recited the same statements. Compare, e.g., Affidavit of Curtis Pye, Doc. 16-24 at 83 (“No one talked to me about any of this before Willie James’s trial. Johnny Mostiler and his assistant Dewey know me. Mr. Mostiler represented me before. He didn’t get in touch with me or ask me any questions about the house Willie James was raised in or what he was like as a child. If he had, I would have said all the things I’ve said in this statement, and I would have testified to all these things if he had asked me to.”), with Affidavit of Ricky Pye, Doc. 16-24 at 99–101 (“[Mr. Yar- brough] didn’t ask about Willie James and how he came up, or how we all were raised. Dewey never spoke to me about those things. . . . I took the stand to testify later on in [the penalty phase of] the trial. No one talked to me about my testimony before I went. I never spoke to Mr. Mostiler about what to say, and he didn’t meet with me or ask me any questions before my turn for testimony. . . . We had it real tough growing up and Mr. Mostiler and Dewey never asked about that. . . . If Mr. Mostiler had asked me about these things on the stand, I would have told the jury the same things I’ve said here.”), and Affidavit of Lolla Mae Pye, Doc. 16-24 at 97 (“No one took the time to talk to me about all anything before Willie’s trial. Nobody ask me all about how I grew up, how I came to be married to Ernest, and how I raised Willie and my other children. I would have been willing to talk about my life with Willie James’s lawyer or investigator, or with any doctor or psy- chologist working on his case. I would have told about all the USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 113 of 135

18-12147 JILL PRYOR, J., Dissenting 45

things I described here, and testified to the jury about them if they wanted me to.”). Third, and even more to the point, we require a petitioner seeking to substantiate an ineffective-assistance claim for failing to investigate and present evidence of mitigation to show that trial counsel did not contact postconviction witnesses and that the wit- nesses would have been available to testify at the time of sentenc- ing. See Maj. Op. at 39. If the witnesses had not included this testi- mony, Mr. Pye’s claim necessarily would have failed. It cannot be that simply because these two facts—lack of contact by counsel about mitigation and availability to testify at sentencing—are pre- sent in every postconviction witness’s affidavit, the record supports a finding that the affidavits are artfully drafted. Otherwise, this is a “heads I win, tails you lose” scenario that the law surely does not countenance. The majority opinion makes one last effort to prop up the state habeas court’s devoid-of-context reading of the four affidavits: it says that the affidavits “lack . . . corroborating evidence in the contemporaneous records—particularly regarding whether Pye was subject to regular physical abuse.” Maj. Op. at 38 n.14. As the majority opinion acknowledges, however, the state habeas court made no mention of any lack of corroborating evidence. 34 Because

34 The state habeas court’s failure to mention the lack of corroborating evi- dence makes sense: there has never been a requirement that testimonial evi- dence be corroborated by contemporaneous documentary evidence. What is USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 114 of 135

46 JILL PRYOR, J., Dissenting 18-12147

the state habeas court did not supply this reason for its decision, we cannot consider it in our § 2254(d) analysis. See Wilson, 138 S. Ct. at 1191–92, 1195–96. B. Willingness of Mr. Pye’s family to cooperate with trial coun- sel As reason 4, the state habeas court found that “the family was not cooperative with the defense team during the pre-trial in- vestigation.” Doc. 20-40 at 64. From that finding, the court sur- mised that counsel did what he could with what little he had. Thus,

more, numerous details, including Mr. Pye’s severe poverty, neglect, low in- tellectual functioning, and mental health challenges, were in fact corroborated by contemporaneous documentary evidence. See, e.g., Doc. 15-12 at 10–11 (third grade record stating that Mr. Pye’s “major causes of absences” were “ill- ness, no shoes, missed the bus”); Doc. 15-14 at 49 (Department of Family and Children Services record from 1972, when Mr. Pye was about seven, noting that Mrs. Pye had requested food and clothing for the family). There also were at least suggestions that Mr. Pye was suffering from something more serious than neglect. For example, in third grade, his teacher observed that he was “fearful and unhappy.” Doc. 15-12 at 11. Another teacher noted that Mr. Pye’s sister, Pam, had “a very difficult home situation.” Doc. 16-13 at 60. Yet another teacher noted that Mr. Pye’s brother Andrew was “often troubled and upset about conditions at home.” Doc. 16-14 at 7. That there were not more con- temporaneous documents is no surprise to me. Mr. Pye grew up poor and Black in rural Georgia in the 1970s. Common sense tells us that the relative scarcity of law enforcement and governmental records was due to the family’s circumstances in the time and place in which Mr. Pye was raised. And, finally, in this case the testimonial evidence cross-corroborated: nearly every story in the affiants’ testimony was corroborated by another affiant’s testimony. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 115 of 135

18-12147 JILL PRYOR, J., Dissenting 47

the court concluded, Mr. Pye had not shown prejudice because trial counsel “did learn, to some extent, of the family’s impoverished circumstances” and presented those facts through Mr. Pye’s sisters. Id. It is of course true that trial counsel had some limited awareness of the family’s poverty and hinted at it in the penalty phase. But the state habeas court’s factual premise that Mr. Pye’s family members were uncooperative in the mitigation investigation finds no sup- port in the record. Seven of Mr. Pye’s family members testified at the penalty phase, so it cannot be that the entire family was unco- operative when it came to sentencing. Further, every family-mem- ber affiant testified under penalty of perjury that he or she would have been willing to speak to the defense team about mitigation before trial, and nothing in Mr. Yarbrough’s testimony about the family’s uncooperativeness in proving Mr. Pye’s innocence called that testimony into question. In concluding otherwise, the majority opinion cites record evidence that, it says, renders the state court’s finding reasonable. First, the majority opinion points to an undated memo from Mr. Mostiler’s file noting that “Willie’s brothers did not respond to my phone calls.” See Maj. Op. at 39 (citing Doc. 19-11 at 93). We know from the record, however, that Mr. Pye’s brother Ricky Pye cooperated with the defense because he testified at the penalty phase. Plus, the very next sentence in Mr. Mostiler’s memo was: “Willie’s sister Pam Bland is a good witness.” Doc. 19-11 at 93. And we know Ms. Bland cooperated with the defense because she, too, testified at the penalty phase. In context, then, Mr. Mostiler’s USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 116 of 135

48 JILL PRYOR, J., Dissenting 18-12147

undated note—which contains no information about the duration or timing of the noted condition and contradicts other evidence in the record when read as the state habeas court read it—offers no support for the state habeas court’s sweeping finding that Mr. Pye’s family was unwilling to cooperate in mounting a case in mitigation. Second, the majority opinion says, Mr. Yarbrough testified that the family members were uncooperative. See Maj. Op. at 39– 40. Again, in context, Mr. Yarbrough’s testimony does not support the state court’s finding as it relates to a case in mitigation because Mr. Yarbrough specified that the family was uncooperative in “helping prove [Mr. Pye’s] innocence,” Doc. 19-11 at 24–25, 35 not in the investigation of mitigating circumstances. By Mr. Yar- brough’s own testimony, he simply didn’t “care” about trying to get Mr. Pye’s family members to testify during the penalty phase. Id. at 25. The record unmistakably demonstrates that any failure to marshal family support in the penalty-phase investigation and presentation was due not to the family’s unwillingness to cooper- ate but rather to Mr. Yarbrough’s lack of care. To sum up on reason 4, the affidavit testimony Mr. Pye in- troduced in postconviction proceedings about the family’s willing- ness to cooperate if only they had been contacted and adequately prepared about a mitigation case based on Mr. Pye’s family

35 Of course the family could not help the defense team prove Mr. Pye’s in- nocence—he committed the crime. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 117 of 135

18-12147 JILL PRYOR, J., Dissenting 49

background directly contradicted the supposed evidence that the family was uncooperative. Thus, Mr. Pye has shown by clear and convincing evidence that the state habeas court’s finding—a find- ing that the court found to be “especially” important, Doc. 20-40 at 67—was unreasonable. 36 C. Remorse as the best strategy given Mr. Pye’s age As its seventh reason, the state habeas court found that be- cause Mr. Pye “was 28 years old at the time of the[] crime[], trial counsel could have reasonably decided, given the heinousness of this crime and the overwhelming evidence of [his] guilt, that re- morse was likely to play better than excuses.” Doc. 20-40 at 66. But there is no evidence in the record—none—that Mr. Mostiler at- tempted to or did offer Mr. Pye’s remorse to the jury as a reason not to sentence Mr. Pye to death. Quite to the contrary. Remorse would have been utterly inconsistent with the defense strategy be- cause Mr. Pye testified in his own defense at trial and denied that he had been present for the rape and murder. At the penalty phase, most of the defense witnesses stated their belief that Mr. Pye was innocent. Mr. Mostiler never mentioned remorse in his closing ar- gument. Thus, the state habeas court’s conclusion that Mr. Mostiler employed a strategy of remorse which likely played

36 Arguably, considering the prominence the state habeas court gave this un- reasonable finding, the state habeas court’s decision was “based on” it, 28 U.S.C. § 2254(d)(2), and so de novo review is appropriate without even ex- amining the court’s other findings. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 118 of 135

50 JILL PRYOR, J., Dissenting 18-12147

better to the jury than excuses was unreasonable in light of the rec- ord. Indeed, even the majority admits this finding was “likely clearly erroneous.” Maj. Op. at 44 n.17. Acknowledging the state habeas court’s error, the majority opinion downplays the import of the clearly erroneous finding by characterizing it as a “sideshow” to the more important age-related determination—a “statement” (that remorse was likely to play bet- ter than excuses) “nestled in a sub-justification” (that given Mr. Pye’s age, evidence of his guilt, and the heinousness of the crime, remorse was likely to play better than excuses) “of a larger justifi- cation” (that Mr. Mostiler’s failure to introduce evidence of Mr. Pye’s childhood was not prejudicial). Id. The state court’s “state- ment,” however, cannot be teased out from its “sub-justification.” It’s all one sentence, one thought: when a person is not young when he commits a crime, mitigating evidence from his childhood and adolescence is entitled to little weight; thus, remorse was the better strategy. The remorse “statement” is the conclusion of this thought. It is not a sideshow; rather, it is part and parcel of the main event. And the main event—what the majority terms a “sub-justi- fication” for the no-prejudice “larger justification,” was unreasona- ble. In Porter v. State, 788 So. 2d 917 (Fla. 2001), the Florida Su- preme Court addressed an ineffective-assistance-of-counsel claim by a prisoner on the state’s death row. In rejecting the claim, the court explained that because the defendant was not young at the time of his crime, “evidence of a deprived and abusive childhood is entitled to little, if any mitigating weight when compared to the USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 119 of 135

18-12147 JILL PRYOR, J., Dissenting 51

aggravating factors.” Id. at 924 (emphasis added) (quoting Bolender v. Singletary, 16 F.3d 1547, 1561 (11th Cir. 1994)). Thus, “[a]ny presentation of this factor would therefore have been insignifi- cant.” Id. After this Court upheld that determination as reasonable, see Porter v. Att’y Gen., 552 F.3d 1260, 1270 (11th Cir. 2008) (citing Bolender, 16 F.3d at 1561), the Supreme Court reversed, conclud- ing that the Florida Supreme Court’s analysis was unreasonable, see Porter, 558 U.S. at 41. In Mr. Pye’s case the state habeas court cited cases, all pre- dating the Supreme Court’s decision in Porter, for the proposition that “‘evidence of a deprived and abusive childhood is entitled to little, if any, mitigating weight’ when the defendant is ‘not young’ at the time of the offense.” Doc. 20-40 at 67 (quoting Tompkins v. Moore, 193 F.3d 1327, 1337 (11th Cir. 1999), and Housel v. Head, 238 F.3d 1289, 1295 (11th Cir. 2001), which cited Tompkins); see id. at 67 (also citing Francis v. Dugger, 908 F.2d 696, 703 (11th Cir. 1990), Mills v. Singletary, 63 F.3d 999, 1025 (11th Cir. 1995), and Bolender v. Secretary, 16 F.3d at 1561). These cases all stood for the same proposition: evidence of a deprived and abusive childhood is entitled to “little, if any” (as stated in Tompkins and Dugger), or “insignificant” (as stated in Bolender, cited in Mills) weight when a defendant commits a crime a decade or so after reaching adult- hood. These are the same cases the majority opinion now cites to conclude that the state habeas court’s treatment of Mr. Pye’s age was reasonable. See Maj. Op. at 43–44. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 120 of 135

52 JILL PRYOR, J., Dissenting 18-12147

But these are the very same reasons—indeed, based on the very same cases—that the Supreme Court held to be unreasonable in Porter. Thus, the state habeas court’s finding was unreasonable under clearly established Supreme Court precedent. Porter, 558 U.S. at 41. 37 * * * The state habeas court’s numerous consequential unreason- able determinations reflect an “extreme malfunction[] in the state criminal justice system.” Reeves, 141 S. Ct. at 2411 (alteration adopted) (internal quotation marks omitted). Even under the ma- jority opinion’s reading of the interplay between 28 U.S.C. § 2254(d)(2) and I(1)—and without the majority opinion’s nullifica- tion of Wilson—it is clear to me that the state court’s errors were of sufficient importance that we can say without difficulty that its ultimate decision was “based on” them. Thus, I would hold that the state court’s decision does not withstand AEDPA deference, and that we should apply de novo review.

37 The majority opinion points out that there isn’t “anything in Porter that explicitly forbids courts from considering age as one factor among many in their prejudice analyses.” Maj. Op. at 42. That is true enough, but in this case the state habeas court did not consider age as a factor. Instead, it repeated pre- cisely the mistake that the Supreme Court corrected in Porter. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 121 of 135

18-12147 JILL PRYOR, J., Dissenting 53

IV. DE NOVO REVIEW Typically, once AEDPA deference is pierced, I would begin my de novo review 38 by explaining why Mr. Mostiler rendered de- ficient performance during the penalty phase. But because the ma- jority opinion assumes deficient performance, I refer to our panel opinion for our analysis. See Pye, 853 F. App’x at 560–65. That leaves prejudice. Here we ask, “whether the entire postconviction record, viewed as a whole and cumulative of miti- gation evidence presented originally, raised a reasonable probabil- ity that the result of the sentencing proceeding would have been different if competent counsel had presented and explained the sig- nificance of all the available evidence.” Debruce v. Comm’r, Ala. Dep’t of Corr., 758 F.3d 1263, 1275 (11th Cir. 2014) (internal quo- tation marks omitted). Postconviction counsel produced evidence that Mr. Pye suffered severe physical and emotional abuse, neglect, endangerment, and privation as a child. Counsel produced evi- dence that Mr. Pye began displaying symptoms of depression early in his childhood, depression that followed him into adulthood. Counsel produced evidence that Mr. Pye’s intellectual capacity is low, bordering on intellectually disabled. This “consistent, unwa- vering, compelling, and wholly unrebutted” evidence, Ferrell v.

38 Although we disagree on whether Mr. Pye is entitled to relief, I note Judge Jordan’s and Judge Rosenbaum’s agreement that the state habeas court’s deci- sion is not entitled to AEDPA deference and so our review should be de novo. See Jordan Concurring Op. at 1. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 122 of 135

54 JILL PRYOR, J., Dissenting 18-12147

Hall, 640 F.3d 1199, 1234 (11th Cir. 2011), “paints a vastly different picture” of Mr. Pye leading up to the crime than the evidence Mr. Mostiler presented to the jury, Debruce, 758 F.3d at 1276. Even in the face of the aggravated crime Mr. Pye committed, and the ag- gravating evidence presented in the penalty phase, I would con- clude, as did the panel, that there is a reasonable probability that at least one juror would have voted for a sentence less than death had the jury heard what we now know about Mr. Pye. See Wiggins, 539 U.S. at 536. First, the new mitigating evidence. Had Mr. Mostiler ade- quately investigated and presented a case in mitigation of the death penalty, the jury would have heard that Mr. Pye was raised in ab- ject poverty by parents who managed to feed and clothe their 10 children by the slimmest of margins. The family lived in a kind of poverty rarely witnessed in the United States, occupying a small four-room house with makeshift walls to separate the sleeping ar- eas and no indoor plumbing or central heating. The jury would have heard that Mr. Pye suffered from ex- treme neglect. At the time of Willie’s birth, his mother Lolla Mae struggled as the sole provider for her six children. Her husband Ernest, whom people called “Buck,” was incarcerated and working on a chain gang. Lolla Mae took whatever work she could get, working all the way up until Willie’s birth and then resuming working immediately afterward. Whether to go to one of her jobs or out drinking (which she did even while pregnant), Lolla Mae typically left Willie alone with his siblings all day. This left the older USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 123 of 135

18-12147 JILL PRYOR, J., Dissenting 55

children—the oldest only 10 years old—to care for the younger ones. The youngest children, Willie included, would spend the day outside in the dirt, often crying all day because the older children lacked the skills necessary to care for infants and toddlers. Willie had little to eat, consuming watered-down milk as an infant and primarily bread and gravy through childhood. With no money for it, he received virtually no medical care. The jury would have heard how the Pye home reflected this neglect. According to a police officer, “[t]he conditions were filthy and the rooms in total disarray every time we entered.” Doc. 16-24 at 22. Mr. Lawson, the school’s social worker, observed that the condition of the house was “deplorable.” Id. at 62. “The house was never clean; piles of filth, scraps and garbage were strewn every- where.” Id. at 62. On one visit, finding the home “so unsanitary” that it created a risk to “the health of the children”—specifically, “the small children had not been bathed and there was spoiled food sitting around”—Mr. Lawson reported the Pye home to the De- partment of Family and Children Services (DFACS). Id. at 61–62. DFACS did not intervene. The jury would have heard that Mr. Pye was raised in a home rife with alcohol abuse and domestic violence. Lolla Mae and Buck drank excessively. Buck, in fact, was notorious around town for his drinking and violent behavior, and Willie and his siblings USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 124 of 135

56 JILL PRYOR, J., Dissenting 18-12147

were ostracized from the community because of the family’s noto- riety. 39 Buck was extremely physically violent; “calls [to law en- forcement] about violence in the Pye home were constant.” Doc. 16-24 at 20. Buck would hit Lolla Mae and throw things at her in front of the children. On at least one occasion, he attacked Lolla Mae with a knife; on another occasion, he hit her over the head with a bottle. Lolla Mae also was violent toward Buck, sometimes threatening him with a knife. The jury would have heard that Mr. Pye experienced fre- quent and often severe physical and emotional abuse at the hands of his father and mother. “Beatings and tirades were [Buck’s] only interaction with his children.” Id. at 60. His verbal assaults were “downright cruel”: he called the children “worthless” and used “every foul expletive he could manage.” Id. at 21–22. “Willie defi- nitely got the worst of his father’s nasty comments.” Id. at 26. Buck “would tell Willie that he was so stupid that he just couldn’t be his kid.” Id. Buck would say “that Willie was born because [Lolla Mae] was messing around while he was in prison, and that he was sick of looking at a kid that belonged to some other guy.” Doc. 16-25 at 2.

39 Buck and Lolla Mae bought alcohol with their individual meager earnings and from the government assistance check the family received because of their son Ernest Pye, Jr.’s disability. “It was general knowledge in the [community] that Junior . . . limped because his father hit him with a tire iron while he was still recovering from a broken hip and the hip never healed properly.” Doc. 16-24 at 73. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 125 of 135

18-12147 JILL PRYOR, J., Dissenting 57

Buck “would tell the rest of [the Pye] kids that there was stuff wrong with Willie [], and that [they] shouldn’t pay attention to him, all with Willie standing right in front of him.” Id. Buck also “beat the devil out of [the] children,” and “Willie definitely got the worst of [those] violent outbursts.” Doc. 16-24 at 26, 60. Lolla Mae beat the children too, and although the school’s social worker Mr. Lawson counseled her about the abuse, she did not stop. The jury would have learned that as the older Pye boys reached their teenage years, they too began to drink excessively and engage in physical violence, beating their father when he was drunk and abusive. When police responded to calls at the Pye house, what they found “was absolute chaos,” id. at 20, with brawl- ing between Willie’s parents and older siblings. For their part, “[t]he younger kids would head for the hills when the fighting started,” often hiding in a clearing in the woods near the home. Id. at 22. Willie was one of the children who ran and hid. As he got older, he tried to play the role of peacemaker, “pull[ing] [his] par- ents apart” when they fought. Id. at 36. In response, Buck would “blast Willie right across the head and he’d go flying.” Doc. 16-25 at 2. The jury would have heard that as a child Mr. Pye “took the comments about not belonging to [his] father hard.” Id. “He was quieter and took things to heart. The most important thing to Wil- lie was to be like everyone else, and [Buck] was constantly telling him that he wasn’t.” Id. When he got upset, Willie “would find any place he could . . . be alone—the bed, the woods, under the porch. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 126 of 135

58 JILL PRYOR, J., Dissenting 18-12147

Then he’d lie down and curl up and just stare at nothing.” Id. If a sibling tried to talk to him, he would act like no one was there. Id. The jury would have learned that Mr. Pye struggled in school because of the home life he experienced and because of his borderline intellectual functioning. Willie often was absent from school because he lacked basic necessities at home: shoes and a place warm enough to dress. When Willie attended school, he per- formed poorly—in some instances in the lowest one percentile— and attended classes for slow learners. He tried hard but could not succeed, and he left school before the end of junior high. Willie was teased by his peers at school, both because he was behind academ- ically and because he lacked adequate clothing—what little he wore often was shared with his many siblings, was seasonally inap- propriate, and was dirty. The jury would have been informed that Mr. Pye’s low in- tellectual functioning was documented into adulthood. After being convicted of burglary and sentenced to prison, notes from the prison psychologist indicated that “[i]ntellectually, [Mr. Pye] is probably in the low average range but his test scores are signifi- cantly lower”—for example, he was reading and writing at a fourth-grade level. Doc. 15-19 at 12–13. The psychologist opined that Mr. Pye “may need special ed help, probably in the learning disabled area.” Id. at 11. Mr. Pye asked to be given job training for barbering, but he failed the aptitude test for it. The records stated that Mr. Pye “[a]ppears to need educational upgrading and adjust- ment prior to retesting.” Id. at 14. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 127 of 135

18-12147 JILL PRYOR, J., Dissenting 59

The jury would have heard that, unsurprisingly, Mr. Pye was depressed. Again, as a child Willie would run away from his family and disassociate. Even as he reached adulthood, he contin- ued to experience long depressive episodes. While Mr. Pye was serving time for his burglary conviction, the prison psychologist in- dicated that he was “very depressed,” “severe enough” to warrant medication and “more counse[l]ing than the average.” Doc. 15-19 at 11, 13, 16. When Mr. Pye left prison in 1990, his depressive epi- sodes continued, and, it seems, worsened. Also of his previous incarceration, the jury would have heard that the prison in which Mr. Pye was first housed, Lee Arren- dale Correctional Facility, was dangerous. “New inmates could ex- pect to be terrorized upon arrival by the guys that were already there. Most were either physically or sexually assaulted, or both.” Doc. 16-24 at 50. Mr. Pye, “a smaller guy” who came across as “very weak,” “confused,” and “vulnerable,” was considered by prison staff “to be at risk for victimization.” Doc. 15-19 at 9, 11, 70. Indeed, a former cellmate of Mr. Pye’s recalled being told that Mr. Pye was raped when he first arrived at the prison. The jury would have heard that despite the environment and the trauma it brought, officials at Lee Arrendale considered Mr. Pye to be generally trustworthy and not an escape or safety risk. The psychologist opined that Mr. Pye was “[n]ot likely to be violent or potentially dangerous” and was “very unlikely to be- come a predator”; she found “[n]o evidence of escape.” Doc. 15-19 at 9, 11. A psychological report indicated that Mr. Pye “should be USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 128 of 135

60 JILL PRYOR, J., Dissenting 18-12147

able to adapt to average security arrangements.” Id. at 15. The psy- chologist’s prediction was correct: Mr. Pye made a notably positive impression on some of the staff he encountered. Guards who pro- vided postconviction testimony regarded Mr. Pye as “completely respectful . . . in a way that most of the inmates were not.” Doc. 16-24 at 49. “He was never menacing, never made any threatening remarks, never did anything but joke around and take care of his assigned work.” Id. These guards had “no reservations about [Mr. Pye] working throughout the dorm area, even during times when he was not closely supervised.” Id. at 71. He even helped the guards “keep the rest of the unit safe” by disclosing knowledge of other prisoners’ weapons or plans for disruption—likely at enormous personal risk to himself. Id. at 49. Had trial counsel adequately in- vestigated Mr. Pye’s previous conviction, he would have un- earthed this evidence, and the jury would not have heard, unrebut- ted, the prosecution’s argument that Mr. Pye would kill a prison guard to escape. Finally, had trial counsel adequately investigated and pre- sented a case in mitigation of the death penalty, the jury would not have heard, without the correct context, that Mr. Pye was raised in a “four bedroom house” or that the family “had love” to offset the lack of modern conveniences. The jury would not have heard, de- void of the context of the abuse he meted out, Mr. Pye’s father’s brief testimony about Willie’s supposedly unremarkable child- hood. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 129 of 135

18-12147 JILL PRYOR, J., Dissenting 61

The wealth of mitigating evidence the jury would have heard had Mr. Mostiler not rendered deficient performance is pre- cisely the kind of mitigating evidence the Supreme Court and this Court have held can demonstrate prejudice. See Rompilla, 545 U.S. at 390, 393 (finding prejudice based on mitigating evidence that Rompilla had low intellectual functioning, “was reared in [a] slum environment”; his parents “were both severe alcoholics who drank constantly” his “mother drank during her pregnancy” his “father, who had a vicious temper, frequently beat [his] mother”; his “par- ents fought violently, and on at least one occasion his mother stabbed his father”; he “was abused by his father who beat him when he was young”; “he was subjected to yelling and verbal abuse”; the family “had no indoor plumbing in the house, he slept in the attic with no heat, and the children were not given clothes and attended school in rags”); see also Porter, 558 U.S. at 33–34 (“Porter routinely witnessed his father beat his mother,” his “father was violent every weekend, and by his siblings’ account, Porter was their father’s favorite target, particularly when Porter tried to protect his mother”; Porter “attended classes for slower learners and left school when he was 12 or 13”); Wiggins, 539 U.S. at 516–17 (“[P]etitioner’s mother, a chronic alcoholic, frequently left Wiggins and his siblings home alone for days.”); Williams, 529 U.S. at 396 (“Williams was borderline [intellectually disabled] and did not ad- vance beyond sixth grade in school”; “prison officials . . . described Williams as among the inmates least likely to act in a violent, dan- gerous or provocative way” (internal quotation marks omitted)); USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 130 of 135

62 JILL PRYOR, J., Dissenting 18-12147

Ferrell, 640 F.3d at 1234 (petitioner was “especially” targeted for abuse by his father). 40

40 The majority opines that the cases in which the Supreme Court has granted habeas relief presented mitigating circumstances that were “significantly stronger” than those present in Mr. Pye’s case. Maj. Op. at 62. Respectfully, the majority is hair-splitting. In Rompilla, for example, the Supreme Court noted that a constitutionally ad- equate investigation into Mr. Rompilla’s prior conviction would have led counsel to discover: Rompilla’s parents were both severe alcoholics who drank constantly. His mother drank during her pregnancy with Rompilla, and he and his brothers eventually developed seri- ous drinking problems. His father, who had a vicious temper, frequently beat Rompilla’s mother, leaving her bruised and black-eyed, and bragged about his cheating on her. His parents fought violently, and on at least one occasion his mother stabbed his father. He was abused by his father who beat him when he was young with his hands, fists, leather straps, belts and sticks. All of the children lived in terror. There were no expressions of parental love, affection or approval. Instead, he was subjected to yelling and verbal abuse. His father locked Rompilla and his brother Richard in a small wire mesh dog pen that was filthy and excrement filled. He had an isolated back- ground, and was not allowed to visit other children or to speak to anyone on the phone. They had no indoor plumbing in the house, he slept in the attic with no heat, and the children were not given clothes and attended school in rags. Rompilla, 545 U.S. at 391–92. Nearly sentence for sentence, this paragraph could have been written about Mr. Pye. Nevertheless, the majority opinion homes in on three details it says Mr. Romp- illa had in his background that Mr. Pye lacked: evidence of schizophrenia, a USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 131 of 135

18-12147 JILL PRYOR, J., Dissenting 63

What is more, this is not a case where the type of mitigating evidence adduced during the state habeas proceedings would have undermined counsel’s strategy at sentencing. Mr. Mostiler focused his penalty-phase presentation on mercy; mitigating evidence of the type postconviction counsel uncovered “would have easily and directly supported the approach counsel offered at sentencing.” Id. at 1235. If the prosecution had asked a more informed jury, “If Wil- lie James Pye does not deserve the death penalty, who are you sav- ing it for?,” Doc. 13-11 at 90, there is a reasonable probability that at least one juror would not have seen Mr. Pye as someone so un- worthy of grace. Second, the aggravating evidence. This of course includes evidence the State would have introduced to rebut the defense’s new mitigating evidence. Wiggins, 539 U.S. at 534. Mr. Pye’s was,

third-grade level of cognition, and likely fetal alcohol syndrome. Maj. Op. at 62. Mr. Pye also had a documented, serious mental health condition: he suf- fered from depression. Mr. Pye also had documented, significant cognitive im- pairments. He performed in the lowest one percentile of his classmates and attended classes for slow learners—and, like Mr. Rompilla, left school near the end of junior high. See Rompilla, 545 U.S. at 391. When he reached adulthood, he was reading and writing at a fourth-grade level—similar to Mr. Rompilla. See id. Finally, Mr. Pye’s evidence showed that his mother drank while preg- nant, and one of his experts opined that Mr. Pye suffered from a fetal alcohol spectrum disorder (one of which is fetal alcohol syndrome). I see very little daylight between the wealth of mitigating evidence counsel failed to uncover in Rompilla and the wealth of mitigating evidence counsel failed to uncover here. It cannot fairly be said that the mitigating evidence in Rompilla was “significantly stronger” than the evidence here. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 132 of 135

64 JILL PRYOR, J., Dissenting 18-12147

without a doubt, an aggravated crime with aggravating circum- stances, including Mr. Pye’s history with Ms. Yarbrough. Mr. Pye and two others kidnapped and raped Ms. Yarbrough at gunpoint, and then Mr. Pye shot her multiple times as she was lying on a roadside and left her to die. The State presented compelling evi- dence that Mr. Pye had been violent with Ms. Yarbrough before and that on this night she remained alive for up to 30 minutes after he shot her, was conscious for most that time, and attempted to stand or crawl to safety. Mr. Pye’s conduct resulted in the trial court’s imposition of four statutory aggravating circumstances. But the Supreme Court and this Court have found prejudice in highly aggravated cases. See, e.g., id. at 514–15, 535 (finding prej- udice even though defendant robbed and drowned an elderly woman); Ferrell, 640 F.3d at 1204–05, 1234–36 (finding prejudice even though defendant robbed and murdered, execution-style, his elderly grandmother and young cousin); Cooper, 646 F.3d at 1331, 1353–56 (finding prejudice even though the state had proven that the triple execution-style murders—apparently committed in the presence of an eight-year-old child—satisfied five aggravating fac- tors). Moreover, the extreme domestic violence Mr. Pye experi- enced—in part because his father, imprisoned around the time of his conception and birth, questioned his parentage—would have contextualized some of the circumstances of the undeniably hor- rific crime Mr. Pye committed—a crime that involved extreme do- mestic violence apparently fueled by questions about Ms. Yar- brough’s child’s parentage. See Ferrell, 640 F.3d at 1235; see also USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 133 of 135

18-12147 JILL PRYOR, J., Dissenting 65

O.C.G.A. § 17-10-30(b) (requiring the judge to instruct the jury that it can consider “any mitigating circumstances”). I have little doubt that had Mr. Mostiler introduced evidence that Mr. Pye posed no serious threat while incarcerated and had trusting, congenial relationships with guards, the State would have introduced evidence that while serving his time for burglary Mr. Pye was sometimes insubordinate.41 Even so, these instances would have added very little in the way of support for the prosecu- tor’s assertion that Mr. Pye would murder a prison guard to escape prison. Similarly, although there is some evidence in the record that Mr. Pye occasionally moved or spoke in an aggressive manner, the records reveal no real violence toward guards 42 and no propen- sity for an escape attempt. Reweighing the evidence in mitigation against the evidence in aggravation, I am convinced that the “mitigating evidence, taken as a whole, might well have influenced the jury’s appraisal of

41 The majority opinion states that many of the instances of insubordination were “categorized as ‘High’ and ‘Greatest’-level offenses.” Maj. Op. at 54. Sort of. Those designations are listed under “Warden’s Disposition Recommenda- tion,” see, e.g., Doc. 15-20 at 1, indicating that they reflect not the severity of the infraction itself, but the Warden’s view of how severe the disciplinary re- sponse should be in proportion to the infraction committed. 42 The one instance of aggression toward a guard the majority opinion high- lights—when Mr. Pye refused to take a shakedown posture—resulted only in failure to follow instructions/insubordination charges. Mr. Pye was not charged with assault or any violent offense. USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 134 of 135

66 JILL PRYOR, J., Dissenting 18-12147

[Mr. Pye’s] moral culpability.” Wiggins, 539 U.S. at 538 (internal quotation marks omitted). Although surely it “is possible that a jury could have heard it all and still have decided on the death penalty, that is not the test.” Rompilla, 545 U.S. at 393. I would conclude, upon a de novo review, that Mr. Pye has shown “a reasonable prob- ability that at least one juror would have struck a different balance” between life and death. Wiggins, 539 U.S. at 537. Thus, I would conclude that he has shown prejudice under Strickland and is enti- tled either to a new penalty phase or to be resentenced without the penalty of death. V. CONCLUSION Deciding that she has had enough of the characters she en- countered through the looking glass, Alice, immersed in a giant chessboard, captures the Red Queen, puts the Red King into check- mate, and awakens from the dream. She emerges in her home, sur- rounded by her belongings and her precious pet kittens. All is right again. The majority opinion, by ignoring the Supreme Court’s opinion in Wilson, traps our Court behind the looking glass. At this point, only the Supreme Court can set things right again. This side of the looking glass, the reality for Mr. Pye is that he experienced the unthinkable as an infant, child, and adolescent. He is chronically depressed and has borderline intellectual func- tioning. When weighing his background against the undeniably horrendous crime he committed, the state habeas court USCA11 Case: 18-12147 Date Filed: 10/04/2022 Page: 135 of 135

18-12147 JILL PRYOR, J., Dissenting 67

egregiously missed the mark. But the majority opinion—even while acknowledging some of the problems in the state court’s de- cision—buries those problems under a mountain of reasons the state habeas court never employed, in violation of Wilson. For Mr. Pye and others who come after his case, though deserving of a second chance to convince a jury to spare their lives under AEDPA as framed by Wilson, they will never get that chance. The writ of habeas corpus is illusory—impossible—even, to obtain. I dissent.