WILLIAMS v. STATE (2026)

ON REMAND FROM THE SUPREME COURT OF THE UNITED STATES

In Thornell v. Jones, the petitioner asserted that his lawyer provided ineffective assistance at sentencing in his capital murder trial by not introducing mitigating evidence of his mental illness and childhood trauma. 602 U.S. 154, 161–62, 166 (2024). He argued that the attorney's mistakes warranted resentencing because a reasonable sentencer might have considered the omitted evidence relevant to his moral culpability. Id. at 165. The Ninth Circuit agreed. Id. at 162–63. But the Supreme Court explained that an Arizona sentencing judge almost certainly would have weighed the serious aggravating circumstances of the petitioner's crime more heavily than all his mitigating evidence, including new and corroborating evidence of psychological disorders, cognitive impairment, childhood physical and sexual abuse, and substance abuse. Id. at 164–71. So it rejected his ineffective-assistance-of-counsel-at-sentencing claim, concluding that he had not shown that he was prejudiced by his counsel's alleged incompetence. Id. at 165.

The claim here is similar, and after issuing Thornell, the Supreme Court vacated our first opinion in this case and remanded for reconsideration. We now conclude that Williams likewise has not shown that he was prejudiced by his trial attorneys' substandard mitigation investigation. There is no reasonable probability that the sentencer, an Alabama trial court, would not have sentenced Williams to death even had it considered his new mitigation evidence. Alabama sentencing courts generally weigh the aggravating circumstances of his crimes—burglary, rape, and murder—more heavily than the kinds of mitigating evidence he offered, including episodic, unreported childhood sexual abuse. We therefore reverse the district court's judgment and remand for reinstatement of Williams's death sentence.

I.

A few weeks before Thanksgiving, almost 30 years ago, Williams raped and murdered his neighbor, Melanie Rowell. After crawling through her window at about 1:00 a.m., he took a knife from the kitchen counter and walked up the stairs toward her bedroom, taking his pants off along the way. He stepped over a baby gate and “peeked in” at Rowell's two toddlers, asleep in their bedroom across the hall from their mother. Williams then crossed into Rowell's room, where he climbed on top of her sleeping body. He began tearing off her clothes while holding the knife to her throat and covering her mouth with his hand to muffle her screaming. He held her down, strangling her until she stopped resisting and quit breathing. Only then did he rape her for 15 to 20 minutes.

Williams left Rowell's half-naked body sprawled on the floor of her bedroom with a smear of his semen on her abdomen. Her children were alone in the apartment until the next afternoon, when her mother came to the house, worried that Rowell had not dropped off the children that morning as usual. She found Rowell blue and cold, with her teddy-bear print nightshirt pushed up over her belly button. Rowell's 15-month-old daughter was sitting on the bed in her room; her two-year-old son did not answer when his grandmother called, and she found him standing on his own bed across the hall “just staring” at her.

Eighteen days later, after a night of drinking and smoking marijuana, plus a minor argument with a friend, Williams again went to the home of a female neighbor in the middle of the night “with sex in mind.” This time it was an older woman he had known since he was a child. At about 2:00 a.m., he took his pants off outside her home and crawled in through her window as she slept. She awoke to find him on his hands and knees by her bed with an erect penis, naked except for a shirt and gloves. He jumped on top of her, sexually assaulted her, and put his hands around her throat to choke her. She struggled and pleaded with him until dawn, when he panicked and fled.

Williams was soon arrested for that burglary and sexual assault. He eventually pleaded guilty to a first-degree burglary charge, and confessed while in custody to raping and murdering Rowell.

At his trial for Rowell's murder, the State introduced overwhelming evidence of his guilt, including his oral confession and two written ones that followed. A forensic scientist also testified that DNA analysis matched blood from Rowell's sheets and semen found on her body to Williams. The jury found him guilty of capital murder.

The same jury heard a sentencing proceeding, during which the State relied on the evidence it had introduced during the guilt phase. For his part, Williams called two witnesses: his mother, Charlene Williams, and his great-aunt, Eloise Williams.

His mother testified that she was 16 years old and unmarried when he was born; she said Williams sometimes lived with other family members because she was so young. She added that he never had a good male role model because his father was not around. Still, he attended church regularly as a child, and played football and basketball until a devastating knee injury during his senior year. After that, his mother told the jury, he acted as though he had “lost all hope.” It did not help that just before high school graduation, he learned that he was one half-credit short and could not graduate.

Though he hung out with a rough crowd after dropping out of school, Williams's mother said that after about two years he decided to “straighten up” and joined Job Corps. At first, he seemed to be doing well, but about a month before the program ended he was kicked out for fighting. After that, just two weeks before the murder, Williams “started hanging out a lot,” and only wanted to “sleep all day and stay up all night.”

When Williams's great-aunt Eloise took the stand, she reiterated that his childhood home life “was not very good.” His mother “did not know a lot about motherhood and responsibilities” and his father was not involved, which meant Williams was moved around a lot. Williams was sad and withdrawn as a child, she said, because he wanted to be with his mother. And though he was close with his great-grandfather and a great-uncle, they both died when he was a teenager, leaving him without any male role models.

After high school, Great-Aunt Eloise noticed a change in Williams—his dreams of “being a great ballplayer” were over and he could not find steady work. He also thought his mother did not care about him. Not long before the murder, she said, he started drinking and maybe even doing drugs. She added that after his arrest Williams told her that he was sorry for what he had done—he said that he had repented and asked the Lord to forgive him.

Alabama law at the time required the jury to weigh the aggravating and mitigating evidence and return an “advisory verdict” recommending either life in prison without parole or death. See Ala. Code § 13A-5-46(e) (1994). The trial court would then decide on a sentence after weighing the same evidence as the jury plus the advisory verdict, a presentence investigation report, and any additional argument or evidence presented by the parties. Id. § 13A-5-47. The relative weight accorded to the aggravating and mitigating circumstances was “strictly within the discretion of the sentencing authority”; the court was required to consider any evidence proffered in mitigation, but it was not obliged to find that the evidence was mitigating or to assign it any significant weight. Smith v. State, 908 So. 2d 273, 298–99 (Ala. Crim. App. 2000); Ex parte Slaton, 680 So. 2d 909, 924 (Ala. 1996).

For Williams, the jury recommended the death penalty by a vote of 11 to 1. But before imposing its sentence, the state trial court took more evidence. The State put Rowell's mother on the stand, where she explained the impact of the murder on the family, especially Rowell's children. Then Williams testified, apologizing to Rowell's family, his own family, and the court. He also submitted six character letters from family members and friends, each of whom described him as a good person, with most attributing his crimes to drug and alcohol use. One relative said that Williams “lived a terrible childhood life” before moving in with his great-aunt, describing his feeling of being abandoned by his mother and emphasizing that he did not get “the tender loving care that every child needs.”

The trial court found that the jury's guilty verdict plus the other evidence established one aggravating factor: murder committed during the course of a rape, robbery, burglary, or kidnapping. See Ala. Code § 13A-5-49(4) (1994). But that single aggravating circumstance applied “in two ways”—murder during a rape and murder during a burglary. The court found one statutory mitigating circumstance, lack of significant prior criminal activity. It also found several nonstatutory mitigating circumstances related to his childhood, including “family problems, problems in early childhood dealing with his living arrangements, his problems resulting from the end of a promising athletic career, [and] his attainment of his GED after failing to graduate from high school.” See Ala. Code §§ 13A-5-51, 13A-5-52. Considering all of these, the court decided that the aggravating circumstance outweighed the mitigating circumstances. It sentenced Williams to death.

On direct appeal, both the Alabama Court of Criminal Appeals and the Alabama Supreme Court affirmed the conviction and sentence. Williams v. State, 795 So. 2d 753 (Ala. Crim. App. 1999), aff'd sub nom. Ex parte Williams, 795 So. 2d 785 (Ala. 2001). The United States Supreme Court denied his petition for certiorari. Williams v. Alabama, 534 U.S. 900 (2001).

Less than a year later, Williams filed a petition for postconviction relief under Rule 32 of the Alabama Rules of Criminal Procedure. He argued that his trial attorneys were ineffective during the penalty phase of his trial for failing to discover and present details of his “history of abuse and neglect,” including the fact that he had been “sexually abused by an older male” when he was a child.

The state circuit court denied those claims on the merits. It explained that most of what Williams proffered in his Rule 32 petition—including evidence of neglect and abandonment by his parents, his history of living with different family members, the high school knee injury that ended his hopes for a sports career, and the deaths of his great-grandfather and great-uncle—was cumulative of testimony presented through his mother and great-aunt at trial. The court also noted that Williams failed to identify a single instance of abuse by a family member in his petition, and that his claims of childhood physical and sexual abuse were contradicted by a pretrial mental evaluation report stating that Williams had denied any such abuse. The Alabama Court of Criminal Appeals affirmed without reviewing the merits of the ineffective-assistance claims because Williams had previously raised ineffective-assistance claims on direct appeal. See Ala. R. Crim. P. 32.2(a)(4). The Alabama Supreme Court denied his petition for a writ of certiorari.

Turning to federal court one week later, Williams filed for habeas corpus relief under 28 U.S.C. § 2254, claiming (among other things) that his attorneys had provided ineffective assistance by failing to investigate his background and present an adequate mitigation case during the penalty phase of his trial. The district court held an evidentiary hearing at which Williams testified and called ten additional witnesses, including family members, a childhood friend, and two psychologists.

Williams and his family members testified about his childhood, explaining that he lived with his mother until he was six years old, when he moved in with his great-aunt, Eloise, and his great-uncle, Robert. Eloise was consistent and reliable, but had strict rules and did not tolerate misbehavior.

When he was 12 or 13 years old, relatives said, Williams got tired of the rules at his great-aunt's house and moved back in with his mother. But that arrangement only lasted about a year, and Williams spent the rest of his childhood moving from home to home with various relatives.

Williams was troubled and unhappy, the witnesses said, because he felt that his mother abandoned him. He did visit her frequently, and sometimes stayed with her for the summer, but that was not always for the best, including during the summer when he was 12. His mother's boyfriend was also living in the house, and Williams saw him abuse his mother both physically and emotionally. That included a black eye and a “busted lip,” and Williams once tried to stab the boyfriend after he hit her. Another year, he lived with his mother and sisters in an old house next door to his great-grandparents, but the house had no air conditioning, no hot water, an outhouse rather than a bathroom, and a wood-burning stove for heat—electricity came from extension cords rigged to his great-grandparents' trailer.

The witnesses added that several of Williams's childhood caretakers were alcoholics, including his mother, great-grandmother, and great-uncle Robert, who drank to excess almost daily when he was not working. As for Williams, he began drinking in his early teens and increased his alcohol use as he got older.

He was also exposed to adult sexuality as a child when he shared a bed with his mother and would sometimes wake to find one of her boyfriends sleeping in the bed (though he did not witness any sexual activity). And when he was ten, an older male cousin had Williams watch him having sex “as a way of showing [him] how to do it with a woman.” That same year, Williams began having sex with a girl his age.

Williams himself testified that between the ages of four and six, he was sexually assaulted three or four times by a boy who was several years older. The boy touched Williams's “groin area,” he said, and had Williams touch him too. He also “penetrate[d] me from behind,” Williams said. At the time, Williams thought the sexual contact was a game, but testified that as he got older, he felt ashamed, depressed, and even suicidal over it. Still, he said, he had kept the story to himself, even denying any sexual abuse to his habeas attorney—at least at first. Years after he was convicted, he told the court, he shared the information with his attorney because he had become more comfortable with him.

Williams also called two psychologists to testify about the impact of the childhood sexual abuse and other negative events. Neither said he had PTSD at the time of the murder. One diagnosed him with “complex posttraumatic stress syndrome,” and said it arose from childhood instability and inability to develop an attachment with a primary caretaker—not to mention the childhood sexual assaults and exposure to “general disinhibited behavior” from the adults in his life. The other cited the negative impact of those same three aspects of Williams's childhood, though he added that Williams likely did not perceive the sexual molestation as wrong or sexual until years later. He said his examination yielded some posttraumatic symptoms, but short of what would be required for PTSD. Both experts agreed that the “primary negative, painful, traumatizing experiences” in Williams's life were not the childhood sexual assaults but the lack of a stable, consistent home life and the sense of abandonment and betrayal he felt when his mother left him with other relatives. Indeed, Williams himself had told one expert that his high school knee injury was the “most bothersome, upsetting” traumatic event.

The experts also testified that Williams displayed hypersexual and hyperaggressive behavior to compensate for his shame and insecurity from the sexual abuse. That behavior included a stunning 150–200 sexual partners between the ages of 10 and 21, the time of his arrest. And his “anger and the tendencies towards violence” manifested themselves in frequent fighting because he reacted violently to the slightest insult. According to one of the experts, Williams's hypersexualization, tendency toward violence, and alcohol use were the “core factors” in understanding why he committed murder. He explained that in his opinion, if Williams had not been sexually abused as a child, or if he had not been drinking on the night of his crimes, he would not have raped and murdered Melanie Rowell.

The State called a psychologist in rebuttal. He questioned the complex trauma syndrome diagnosed by Williams's expert, explaining that it did not exist in standard psychology diagnostic manuals. And Williams could not be diagnosed with posttraumatic stress disorder, the psychologist explained, because none of his childhood experiences (including the sexual abuse) caused clinically defined trauma—that is, a reaction of horror and fear for one's own life or the life of another. Williams's troubling behaviors, he said, including the rape and murder, were because of his substance abuse.

The district court granted Williams's federal habeas petition. It held that his trial attorneys provided ineffective assistance of counsel by failing to adequately investigate and present mitigating evidence during the penalty phase of his trial. Williams v. Alabama, No. 1:07-CV-1276-KOB, 2021 WL 4325693, at *16–17, *52 (N.D. Ala. Sept. 23, 2021). This Court affirmed in a split decision. Williams v. Alabama, 73 F.4th 900, 902 (11th Cir. 2023) cert. granted, judgment vacated, 144 S. Ct. 2627 (2024); id. at 912–15 (Grant, J., dissenting). The case now returns to us after the United States Supreme Court granted Alabama's petition for certiorari, vacated the panel's earlier judgment, and remanded for reconsideration in light of Thornell v. Jones, 602 U.S. 154 (2024). Alabama v. Williams, 144 S. Ct. 2627 (2024).

II.

On appeal from a district court's judgment on a § 2254 petition, we review legal questions and mixed questions of law and fact de novo, and we review findings of fact for clear error. Sullivan v. Sec'y, Florida Dep't of Corr., 837 F.3d 1195, 1204 (11th Cir. 2016). Both components of an ineffective-assistance-of-counsel claim—deficient performance and prejudice—are mixed questions of law and fact that we review de novo. Strickland v. Washington, 466 U.S. 668, 698 (1984); Jefferson v. GDCP Warden, 941 F.3d 452, 473–74 (11th Cir. 2019).

III.

To satisfy the Strickland standard for evaluating ineffective-assistance-of-counsel claims, a petitioner must show both that his counsel's performance was deficient and that the deficient performance prejudiced his defense. Strickland, 466 U.S. at 687. If the petitioner does not meet his burden on either component, the claim fails. Id. at 687, 697.

To show deficient performance, a petitioner must establish that his attorneys “made errors so serious that counsel was not functioning as the ‘counsel’ guaranteed the defendant by the Sixth Amendment.” Id. at 687. And to show prejudice, he must demonstrate “a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different.” Id. at 694. For a penalty-phase claim in a capital case, “the question is whether there is a reasonable probability that, absent the errors, the sentencer—including an appellate court, to the extent it independently reweighs the evidence—would have concluded that the balance of aggravating and mitigating circumstances did not warrant death.” Id. at 695. “A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id. at 694.

Thornell clarified that the prejudice analysis requires habeas courts to weigh the aggravating and mitigating circumstances in the same way that the state sentencing judge and appellate courts would have done. 602 U.S. at 164–65. Federal courts must give due consideration to aggravating circumstances that are weighed heavily by state courts. Id. at 164, 170–71. And they must be mindful that although state courts are required to consider any mitigating evidence proffered by the defendant, they are not required to find that evidence persuasive. See id. at 164–65, 166–70. With these points in mind, we consider Williams's claim anew.

IV.

To start, we have no doubt about our earlier holding that trial counsel's cursory mitigation investigation was deficient, so we need not revisit it. See Williams, 73 F.4th at 909–10; see also id. at 912 (Grant, J., dissenting). Even so, unless Williams can show that deficient performance prejudiced him, he cannot receive relief. Strickland, 466 U.S. at 687, 697.

In Alabama courts, the question of aggravating and mitigating circumstances in a capital case is about weighing—not counting. “The determination of whether the aggravating circumstances outweigh the mitigating circumstances is not a numerical one, but instead involves the gravity of the aggravation as compared to the mitigation.” Ex parte Clisby, 456 So. 2d 105, 108–09 (Ala. 1984); see Ala. Code. § 13A-5-48. And the “weight to be attached to the aggravating and the mitigating evidence is strictly within the discretion of the sentencing authority.” Gobble v. State, 104 So. 3d 920, 989 (Ala. Crim. App. 2010) (citation omitted). As the Supreme Court has directed, we follow Alabama's approach in considering whether an Alabama sentencing court would have concluded that the postconviction evidence tipped the balance in favor of life rather than death. See Thornell, 602 U.S. at 164–65, 170.

A.

We start with “the totality of the available mitigation evidence” from both the trial and the later habeas proceeding. Williams v. Taylor, 529 U.S. 362, 397–98 (2000). The sentencing court heard about Williams's unhappy childhood as he was shuffled from relative to relative: his mother's abandonment; his father's absence; the death of his great-grandfather and great-uncle; and the knee injury that ended his plans for an athletic career. It heard about his failure to graduate with his high school class; his participation in Job Corps, where he obtained his GED; his difficulty finding work after being sent home from Job Corps for fighting; and his drug and alcohol use leading up to and on the night of the murder. It also heard Williams's confessions to the police and his expression of remorse, and it considered character letters from friends and family members.

Those family members testified to many of the same facts during the federal habeas proceeding—in more detail, no argument. They also provided more information on certain topics: his mother's poverty; his exposure to domestic violence, alcohol use, and adult sexuality; his use of drugs and alcohol as a teenager; and his extended family's troubling sexual history. Williams testified for the first time about his experience of childhood sexual abuse, and he introduced expert testimony linking the abuse to his hypersexual and increasingly aggressive behavior.

Alabama trial courts, though, frequently accord “little weight” or “very little weight” to similar evidence. That includes cases with instability, poverty, and absent parents, see James v. State, 61 So. 3d 357, 377 (Ala. Crim. App. 2010); parental abandonment, instability, and a deprived childhood, McNabb v. State, 991 So. 2d 313, 331 (Ala. Crim. App. 2007); a dysfunctional family, few positive role models, exposure to criminals and violence, PTSD, and substance abuse, Hosch v. State, 155 So. 3d 1048, 1066, 1121 (Ala. Crim. App. 2013); childhood rape and abuse, extreme poverty, parental abandonment, an alcoholic caretaker, domestic violence, and no positive male role models, McMillan v. State, 139 So. 3d 184, 268 (Ala. Crim. App. 2010); childhood rape and abuse, an alcoholic caretaker, domestic violence, and substance abuse, Sneed v. State, 1 So. 3d 104, 129–30 (Ala. Crim. App. 2007); instability, parental death, and drug abuse, Flowers v. State, 922 So. 2d 938, 961 (Ala. Crim. App. 2005); and poverty, domestic conflict, an alcoholic caretaker, and substance abuse, Jones v. State, 43 So. 3d 1258, 1286 (Ala. Crim. App. 2007). A difficult and unsettled childhood featuring impoverished living conditions and exposure to substance abuse and violence is, tragically, so common that it provides little mitigating value, at least as Alabama sentencing courts see it. After all, “many people who experience similar or even worse childhoods do not grow up and commit capital murder.” James, 61 So. 3d at 377.

What's more, the new testimony also showed that the serious disadvantages Williams suffered were at least to some degree offset by positive experiences and relationships. One of his sisters, for instance, described their mother as a “great mother,” even before she quit drinking. Another sister described Great-Aunt Eloise in glowing terms. Williams's great-grandfather was also said to be a good, caring, and gentle man who helped raise Williams and had a close and loving relationship with him. His great-uncle too was a role model and a “great provider.” Plus, Williams attended church and participated in sports throughout his childhood, and one of his own psychological experts cited his intelligence, athletic abilities, and supportive extended family as “protective factors” during his childhood.

We have no doubt that Williams's most significant new evidence was his testimony that he was sexually abused as a child. But episodic or isolated instances of abuse like those Williams described have not been seen by Alabama trial courts as outweighing serious aggravating circumstances. And that's especially so if the petitioner did not tell anyone about the abuse until many years later.¹ In Sneed v. State—in which the defendant was convicted as an accomplice to murder during a robbery—the sentencing court gave “little” mitigating weight to evidence that the defendant “was likely raped at a young age by a virtual stranger and told no one of this until he was out of high school.” 1 So. 3d at 129–30. And in McMillan v. State—also a murder-during-robbery case—the trial court assigned “very little” mitigating weight to evidence that the defendant was raped as a child, noting that the defendant first disclosed the childhood sexual abuse to a defense expert after his arrest for capital murder. 139 So. 3d at 213–15.

Nor would the psychologists' testimony linking the sexual assaults Williams suffered as a child to Rowell's rape and murder have been entirely mitigating. In fact, that kind of testimony may have even backfired. See Waldrop v. State, 987 So. 2d 1186, 1200 (Ala. Crim. App. 2007). Alabama courts, after all, have described such evidence as a “double-edged sword” because it also supports a finding of future dangerousness. State v. Mitchell, 377 So. 3d 94, 121 (Ala. Crim. App. 2022); Harris v. State, 365 So. 3d 1075, 1128 (Ala. Crim. App. 2021); Davis v. State, 9 So. 3d 539, 566 (Ala. Crim. App. 2008). Expert testimony about his hyperaggressive behavior and compulsive sexuality might also have opened the door to evidence of his second burglary and sexual assault, as further proof to the jury of his future dangerousness. Though it is not weighed as a separate aggravating factor, future dangerousness is a proper consideration when evaluating the weight of the statutory aggravating circumstances. Floyd v. State, 289 So. 3d 337, 431 (Ala. Crim. App. 2017). And the question is “of inestimable concern at the penalty phase” of a capital trial in Alabama. Id.

Plus, experts on both sides testified that Williams's alcohol use played a causative role in Rowell's murder. But that testimony would not have been helpful—Alabama courts typically do not assign much, if any, mitigating weight to voluntary drug or alcohol use. See, e.g., Jones, 43 So. 3d at 1287; Thompson v. State, 503 So. 2d 871, 882 (Ala. Crim. App. 1986).

Alabama courts, in short, do not typically give much mitigating weight to evidence of the sort offered at Williams's habeas hearing.

B.

By contrast, Alabama courts give heavy weight to the aggravating circumstance at issue here—murder committed during the course of a rape, robbery, burglary, or kidnapping. See Ala. Code. § 13A-5-49(4). It is common for Alabama sentencing courts to impose the death penalty when this aggravating factor applies, and for Alabama appellate courts to uphold that penalty after independently reweighing the aggravating and mitigating circumstances on appeal. See Williams, 795 So. 2d at 785 (collecting cases involving rape or attempted rape); Reeves v. State, 807 So. 2d 18, 50 (Ala. Crim. App. 2000) (collecting cases involving robbery); Walker v. State, 932 So. 2d 140, 161–62 (Ala. Crim. App. 2004) (collecting cases involving burglary); Ingram v. State, 779 So. 2d 1225, 1283 (Ala. Crim. App. 1999) (collecting cases involving kidnapping).

This remains true when murder during rape, robbery, burglary, or kidnapping is the only aggravating circumstance found to apply—even when the murder involves only one of the enumerated felonies. See, e.g., Osgood v. State, 341 So. 3d 170, 236–37 (Ala. Crim. App. 2016) (rape); McWhorter v. State, 781 So. 2d 257, 329–30 (Ala. Crim. App. 1999) (robbery); George v. State, 717 So. 2d 849, 857 (Ala. Crim. App. 1997) (burglary). Alabama courts have weighed this single aggravating circumstance more heavily than multiple significant mitigating circumstances, including that the defendant “was raised in extreme poverty; that he was abandoned by his mother; that he was physically abused as a child; that he was raped as a child; that he was a witness to his mother's and sister's abuse; that he was raised in the home of an alcoholic/drug addict; that he did not get the treatment he needed; that he had no positive male role models; that he suffered from psychological and emotional difficulties; and that his intellectual functioning was in the borderline range.” McMillan, 139 So. 3d at 268–69; see Riley v. State, 166 So. 3d 705, 766 (Ala. Crim. App. 2013).

And when more than one of the listed felonies is involved, Alabama courts weigh the aggravating factor more than once, typically by counting each applicable felony as a separate aggravating circumstance. See Belisle v. State, 11 So. 3d 256, 321–22 (Ala. Crim. App. 2007). The Alabama Court of Criminal Appeals has explained that to “hold otherwise would be to say that a person could commit a murder during the course of committing any number of the felonies enumerated in § 13A–5–49(4) and face no additional consequences at the penalty phase and in fact be considered to have committed only one of the felonies contained in this section. This was not the intent of the legislature.” Stewart v. State, 730 So. 2d 1203, 1212 (Ala. Crim. App. 1996).

The Alabama case most like this one tells the tale. In Osgood, the state courts found that the aggravating circumstance of murder during rape outweighed mitigating evidence of childhood experiences that were similar to, or even worse than, those experienced by Williams. See Osgood, 341 So. 3d at 236–37. The mitigating circumstances there included an “unsettled” childhood “being passed around from home to home”; parental abandonment; sexual abuse by a man in a bar when he was between three and six years old; sexual encounters with other children beginning at age nine and sexual involvement with a 24-year-old woman at age 14; malnutrition as an infant, possibly resulting in delayed brain development; substance abuse; a suicide attempt; admission to a psychiatric hospital as a teenager; and diagnoses of antisocial personality disorder, sexual addiction, and attachment disorder. Id.at 209, 236–37.

The Alabama trial court identified ten mitigating circumstances and acknowledged the expert testimony linking the defendant's crimes to his childhood. Id. at 236–37. But it concluded that the single aggravating circumstance of murder during rape outweighed the mitigating factors and sentenced Osgood to death. Id. at 213, 229. On appeal, the state appellate court independently reweighed the aggravating and mitigating circumstances and agreed that death was the appropriate sentence. Id. at 237.

Williams, in contrast, has not identified any case in which an Alabama sentencing court or court of appeals has held that mitigating circumstances like these balanced or outweighed the aggravating circumstance of murder during burglary and rape.

C.

Considering these examples and the absence of comparable cases going the other way, we think it plain what Alabama courts are likely to have done here, even considering the new evidence Williams now offers. The Alabama trial court already gave the aggravating factor of murder during rape, robbery, burglary, or kidnapping extra weight because the crime involved both rape and burglary. And it concluded that the aggravating circumstance outweighed the mitigating evidence of Williams's sad and unstable childhood, abandonment by his mother, dashed hopes of an athletic career, unsuccessful stint at Job Corps, and substance abuse, and sentenced him to death. The state appellate court independently weighed the evidence and reached the same conclusion. The evidence introduced at habeas consisted primarily of more detailed testimony concerning his unsettled and sometimes impoverished childhood, plus his testimony that he was sexually assaulted three or four times by an older boy and expert testimony linking the childhood instability, sexual abuse, and drug and alcohol use to his hypersexual and increasingly aggressive behavior, including Rowell's rape and murder. Weighing all of Williams's mitigating evidence against the aggravating circumstance found by the trial court at sentencing, we see no reasonable probability that the new evidence he presented at habeas would have made a difference in his sentence. To conclude otherwise, we would have to dismiss the statutory aggravator of murder during burglary and rape without giving it the significant weight accorded to that factor by Alabama courts or overvalue Williams's mitigation evidence in a way the Alabama sentencer almost certainly would not do.

D.

In closing, we offer a few brief responses to the dissenting opinion. First, we have not, as that opinion suggests, altered the Strickland standard in any way. Dissenting Op. at 1–2. As the dissenting opinion itself recognizes, “Thornell's language about ‘the sentencer’ is not new language—it comes directly from Strickland.” Id. at 4. The point of Thornell is that a federal court performing a Strickland prejudice analysis must focus on how the petitioner's new mitigating information would have impacted the state court's sentencing decision. See Thornell, 602 U.S. at 163–64. And when trying to determine how a state court would have weighed evidence, it is a good idea to consult state court opinions weighing similar evidence. See id. at 164–65. This is not a change in the Strickland standard. It is an instruction on how to correctly apply it.

Nor is it true, as the dissenting opinion suggests, that “[n]either Strickland nor Thornell defines who the proper ‘sentencer’ is.” Dissenting Op. at 6. Both are clear that for our purposes, the “sentencer” is the state court responsible for weighing aggravating and mitigating circumstances to determine whether death is the appropriate sentence. See Thornell, 602 U.S. at 163–64 (quoting Strickland, 466 U.S. at 695).

Second, we agree that Thornell does not require federal habeas courts to “defer to state courts and avoid their duty to independently ‘consider all the evidence—the good and the bad.’ ” Dissenting Op. at 5 (quoting Wong v. Belmontes, 558 U.S. 15, 26 (2009)). But we disagree with any suggestion that we have not done so in this case. We have evaluated the entirety of Williams's evidence, doing our best to follow the “persuasive guideposts” in Alabama cases as the Supreme Court directed in Thornell. Id.

Finally, the dissenting opinion cites workability problems in our approach, but those critiques are grounded, as the opinion admits, in the Strickland standard itself. See id. at 6–7. Strickland requires us to assess the probability of a different result based on how the state sentencer would have weighed the evidence presented to it. Strickland, 466 U.S. at 695. If an Alabama prisoner presents new evidence that would not have mattered to an Alabama sentencing court, then the prisoner has not met the Strickland standard for prejudice even if a sentencer in California may have been more sympathetic.

No doubt, the prejudice inquiry can be a difficult and uncertain task. But relying on state court opinions rather than our own unbounded intuition allows federal courts to avoid overstepping our habeas authority when reviewing state decisions.

* * *

All things considered, it is highly unlikely that the assistance of competent counsel during the sentencing phase would have led the Alabama trial court to impose a sentence of life in prison rather than death. See Strickland, 466 U.S. at 695. We do not excuse the failure of the defense team to properly investigate and present a mitigation case. But viewed in its entirety and weighed properly, the evidence developed in habeas review creates only the slightest possibility of a more favorable result at sentencing. That is not enough. See Harrington v. Richter, 562 U.S. 86, 111–12 (2011). We therefore REVERSE the district court's judgment and REMAND with instructions for the district court to reinstate Williams's death sentence.

REVERSED and REMANDED with instructions.

As the majority explains, the Supreme Court granted certiorari as to Alabama's petition and then vacated and remanded the case to us to consider in light of Thornell v. Jones, 602 U.S. 154 (2024). Once back here, Alabama fervently pressed the argument that Thornell clarified the Strickland ¹ prejudice standard that applies in the penalty phase of a trial when the defendant argues ineffective assistance of counsel for failure to present certain mitigating evidence.

And the majority adopts Alabama's conclusion without reservation. Rather than conducting an independent reweighing of the mitigating and aggravating factors as courts have done consistently in the past ² —and as this court originally did when Alabama first appealed ³ —the majority explains that in light of Thornell, “the prejudice analysis requires habeas courts to weigh the aggravating and mitigating circumstances in the same way that the state sentencing judge and appellate courts would have done.” Maj. Op. at 14 (citing Thornell, 602 U.S. at 164–65). Using that reasoning the majority now reverses the district court's grant of habeas relief to Williams and reinstates his death sentence. This amounts to more than a clarification—it effectively and detrimentally changes the Strickland standard.

I disagree with this interpretation for two reasons. First, a close reading of Thornell suggests that the Court did not intend to alter the Strickland standard in this way. Second, the “clarified” standard embraced by the majority creates significant workability problems, which unfairly erodes a critical federal constitutional protection for criminal defendants. Thus, I respectfully dissent.

I.

In Thornell, the Supreme Court reversed the Ninth Circuit's grant of habeas relief to a defendant claiming ineffective assistance of counsel, applying what it called, “the proper understanding of Strickland.” 602 U.S. at 165. The defendant claimed his counsel failed to present certain mitigating evidence. See id. at 161. The Ninth Circuit agreed, concluding that there was a “reasonable probability” that the defendant “would not have received a death sentence” if that evidence had been presented at sentencing. Id. at 162. But the panel “made no mention of the aggravating factors” and “did not consider the State's rebuttal evidence.” Id.

The Supreme Court found the Ninth Circuit erred in “at least three ways” in interpreting and applying Strickland. Id. at 162–65. First, it failed to sufficiently consider “the weighty aggravating circumstances” in the case. Id. at 164. The Ninth Circuit's initial opinion “did not mention those circumstances at all,” and though its amended opinion “at least mentioned the aggravating circumstances, ․ it failed to give them the weight that they would almost certainly be accorded by an Arizona sentencing judge.” Id. Second, the Ninth Circuit applied an “unsound” circuit rule that prohibits a court from assessing the relative strength of expert testimony. Id. And third, the Ninth Circuit found the district court erred by diminishing the persuasive value of the mitigating evidence because it saw no link between the evidence and the defendant's conduct. Id.

To correct the Ninth Circuit's errors, the Supreme Court conducted its own independent reweighing process. It turned to each piece of new mitigating evidence presented by the defendant at the evidentiary hearing and discussed what effect that evidence would have had if presented at the penalty phase of the trial.⁴ Id. at 166–71. The Court concluded that “[m]ost of the mitigating evidence [the defendant] presented at the federal evidentiary hearing was not new, and what was new would not carry much weight in Arizona courts.” Id. The evidence “would barely have altered the sentencing profile presented to the sentencing judge, and it is insufficient to show prejudice.” Id. at 166 (internal quotation marks omitted).

The majority points to language in Thornell requiring us to consider how “the sentencer” would have considered the aggravating and mitigating factors, and relatedly, the case's repeated references to how “Arizona courts” would have weighed the factors. See id. at 163–71 (quotation marks omitted). Still, Thornell's language about “the sentencer” is not new language—it comes directly from Strickland. Id. at 163 (quoting Strickland, 466 U.S. at 695). And the Court did not state that it intended to change the Strickland standard—it said that it proceeded with “the proper understanding of Strickland in mind.” Thornell, 602 U.S. at 165.

The main gist of Thornell is a far cry from changing the Strickland standard. Thornell's main points are that the Ninth Circuit did not sufficiently consider the aggravating circumstances, and the new mitigating evidence that the defendant presented was duplicative, uncorroborated, or directly contradicted by other record evidence. Id. at 164–66. Thornell uses “Arizona courts” and Arizona case law not as mandatory authority but as persuasive guideposts in its analysis of whether the new evidence would have meaningfully changed the outcome of the case.⁵ See id. at 166–71. But Thornell does not suggest that the reviewing court should defer to state courts and avoid their duty to independently “consider all the evidence—the good and the bad.” Wong v. Belmontes, 558 U.S. 15, 26 (2009) (per curiam). And we should not give up that critical, independent duty without clear direction from the Supreme Court.⁶

II.

Next, the majority's interpretation of Thornell presents significant workability problems. Neither Strickland nor Thornell defines who the proper “sentencer” is, and Thornell's alternating use of “Arizona courts” and “Arizona sentencing judge” while citing to Arizona Supreme Court decisions offers little clarity. Thornell, 602 U.S. at 165–70. Hypothetically speaking, it would seem most probable that “the sentencer” would be the sentencing judge, since in Thornell, the trial judge initially imposed the sentence. Id. at 161. So, “the sentencer” could be the trial judge, but Court's references to “Arizona courts” and citations to Arizona Supreme Court cases, at the very least, create some confusion about whose shoes, exactly, we are supposed to put on. And if we do not know who the sentencer is, it is nearly impossible to step into their shoes properly.

Indeed, stepping into the shoes of a “sentencer,” even if we know the identity of that “sentencer,” is nearly an impossible task. Sentencing is, by its very nature, a fact-driven process, and Thornell gives little instruction about how we stand in the place of the sentencer, other than citing to prior Arizona Supreme Court decisions from various years. See id. at 166–71. If the actual sentencing judge is the proper “sentencer,” then Arizona Supreme Court cases seem to be necessary, but also insufficient, evidence of how the individual judge would have weighed the evidence. Similarly, if “Arizona courts” generally are “the sentencer,” then why only look to Arizona Supreme Court cases? Cases from lower Arizona courts might also be relevant.

And if the Arizona Supreme Court is the proper “sentencer,” then Arizona Supreme Court cases are indeed the only appropriate sources to look to, but what is the appropriate time frame? Is a case from 2020 appropriate evidence of how the 1996 Arizona Supreme Court ⁷ would have weighed new evidence of mood disorders with the rest of the aggravating and mitigating evidence in the record? See id. at 167 (citing State v. Poyson, 475 P.3d 293, 298 (Ariz. 2020)). Perhaps not.

The majority's interpretation effectively creates at least fifty standards—at least one per state, maybe more depending on changes in the state court's disposition over time—for when Strickland prejudice exists for failure to present mitigating evidence. While federal courts can and do consider state law in determining whether prejudice exists, see, e.g., Wiggins v. Smith, 539 U.S. 510, 536 (2003), requiring us to decide the prejudice prong based on state law leads to a standard where Strickland prejudice means one thing in one state and something else in another. For example, following the majority's interpretation, Williams' counsel was not constitutionally ineffective in Alabama, but had his trial been conducted in a different state with different views on aggravating and mitigating evidence—Berkeley, California,⁸ perhaps—his counsel may have been considered constitutionally ineffective there.

That cannot be so. The Sixth Amendment right to effective assistance of counsel is a universal right afforded to all citizens of the United States, no matter what state they are from or what state they are charged, tried, and sentenced in.

The majority's decision unfairly erodes the critical Sixth Amendment right to effective assistance of counsel. The majority's interpretation of Thornell takes away our ability, and the ability of the district court, to serve as independent safeguards for constitutional violations in habeas cases. Federal habeas review serves as “a bulwark against convictions [and sentences] that violate fundamental fairness.” Engle v. Isaac, 456 U.S. 107, 126 (1982) (internal quotation mark omitted). If in conducting that critical review for constitutional violations, we must defer to how state courts would have weighed evidence, our role as an independent reviewer is effectively eliminated.⁹ The majority's interpretation of Thornell, therefore, frustrates the purpose of habeas review and allows for potential constitutional violations to go unchecked.

III.

Given the high stakes of this decision, and since it is based on a debatable reading of Thornell, absent clear direction from the Supreme Court that it intended to change the Strickland prejudice standard in this context, I would again affirm the district court's grant of habeas relief. Therefore, I respectfully dissent.

FOOTNOTES

1.   We accept the district court's finding that the childhood sexual encounters with an older boy occurred as Williams described them, and we assume for purposes of our analysis that the Alabama sentencing court also would have found that the assaults occurred and would have given this evidence some weight in mitigation. But the fact remains that Williams's failure to report the abuse until after he was convicted of murder and sentenced to death—along with reports that he denied any such abuse in a pretrial psychological examination and when initially asked by his collateral counsel—would have lessened the mitigating weight accorded to the abuse by the state court. See Sneed, 1 So. 3d at 129–30; McMillan, 139 So. 3d at 213–15. It is worth noting that in denying Williams's penalty-phase ineffective-assistance claim, the state post-conviction court discounted his assertion of childhood abuse based on the pretrial psychological evaluation report indicating that Williams had denied “childhood sexual, emotional, or physical abuse.”

1.   Strickland v. Washington, 466 U.S. 668 (1984).

2.   See Porter v. McCollum, 558 U.S. 30, 41 (2009) (per curiam) (The reviewing court 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.’ ” (alteration adopted) (quoting Williams v. Taylor, 529 U.S. 362, 397–98 (2000))); see also Wiggins v. Smith, 539 U.S. 510, 534 (2003) (“In assessing prejudice, we reweigh the evidence in aggravation against the totality of available mitigating evidence.”).Indeed, when conducting habeas review in this context, we, along with several of our sister circuits, routinely and independently reweigh the evidence in mitigation against the evidence in aggravation. See, e.g., Mashburn v. Comm'r, 80 F.4th 1292 (11th Cir. 2023); Jenkins v. Comm'r, 963 F.3d 1248 (11th Cir. 2020); Sandoval Mendoza v. Lumpkin, 81 F.4th 461 (5th Cir. 2023); Goodwin v. Johnson, 632 F.3d 301 (6th Cir. 2011).

3.   Williams v. Alabama, 73 F.4th 900, 911 (11th Cir. 2023) (“Considering the record before us—and given the highly deferential standard of review for factual findings—we conclude that Williams has established Strickland prejudice.”), cert. granted, judgment vacated, 144 S. Ct. 2627 (2024).

4.   The defendant first presented new evidence about mental illness, but the Court concluded that “[b]ecause none of [the defendant's] experts provided a real link between [his] disorders and the murders, their testimony would have done him little good in the Arizona courts.” Thornell v. Jones, 602 U.S. 154, 167 (2024). Next, he presented new evidence of cognitive impairment, but the Court concluded that the evidence “at most corroborates testimony the Arizona courts already credited,” and it “would have offered an insignificant benefit, if any at all.” Id. at 168 (citation modified). He also presented new allegations of physical and sexual abuse, but the Court concluded that the new allegations of sexual abuse were “entirely uncorroborated,” and the physical abuse allegations, while corroborated, were directly contradicted by other evidence in the record and not “causally connected to the murders.” Id. at 168–69. Finally, he presented some minimal new evidence of substance abuse, but the Court concluded that it was merely duplicative of the other evidence in the record. Id. at 169–70.

5.   The Court has conducted similar analyses to decide whether there was a “reasonable probability that the advisory jury—and the sentencing judge—would have struck a different balance.” Porter, 558 U.S. at 42 (internal quotation marks omitted).

6.   Justices Sotomayor, Kagan, and Jackson all dissented in Thornell. 602 U.S. at 172–73 (Sotomayor, J., dissenting, joined by Kagan, J.); id. at 173–74 (Jackson, J., dissenting). Neither dissent contemplates that Thornell alters the Strickland prejudice standard in any way; in fact, the dissents focus on whether the Supreme Court should have involved itself in the reweighing process at all.

7.   The Arizona Supreme Court first took up this case in State v. Jones, 917 P.2d 200 (Ariz. 1996).

8.   See Oral Arg. at 3:00–4:00, Williams v. Ala. (11th Cir. 2025) (No. 21-13734), ca11.uscourts.gov/system/files_force/oral_argument_recordings/21-13734_01212025.mp3. During oral argument, counsel for the State of Alabama distinguished between the kinds of evidence that might have been available in Berkeley, California versus St. Claire County, Alabama at the time of Williams' trial. See id. But counsel's distinction helps make our point more than it does his. Tethering a critical federal constitutional right to idiosyncrasies of particular American geographic regions distorts the meaning and the purpose of the right itself.

9.   I recognize that the Antiterrorism and Effective Death Penalty Act (AEDPA) requires us to defer to the state court for federal claims that have been “adjudicated on the merits in State court proceedings.” 28 U.S.C. § 2254(d). But AEDPA does not apply here. In a prior iteration of this case, we found that the Alabama Court of Criminal Appeals invoked an improper procedural bar as to Williams' claim for ineffective assistance of counsel for failure to investigate. Williams v. Alabama, 791 F.3d 1267, 1273 (11th Cir. 2015). We also found that there was no adjudication on the merits of that claim, so we remanded the case to the district court for de novo review on that issue. Id. Because there was no state court adjudication on the merits, AEDPA deference does not apply here. See Wiggins, 539 U.S. at 534.

Grant, Circuit Judge: