Williams v. Trammell

782 F.3d 1184, 2015 U.S. App. LEXIS 5826, 2015 WL 1600424

• Court: Court of Appeals for the Tenth Circuit
• Decided: April 10, 2015
• Docket: 12-5190
• Status: Published

Opinions

PHILLIPS, Circuit Judge.

In this habeas case, Jeremy Alan Williams challenges his Oklahoma conviction for first-degree murder and his accompanying sentence of death. The district court denied relief but issued a certificate of appealability, giving Williams the ability to appeal his claims of ineffective assistance of counsel. In addition, this court also agreed to hear Williams’s sufficiency-of-the-evidence and cumulative-prejudice claims. Exercising jurisdiction under 28 U.S.C. §§ 1291 and 2253(a), we agree with the district court and conclude that Williams is not entitled to relief.

I. BACKGROUND

The following facts come from the direct-appeal decision of the Oklahoma Court of Criminal Appeals (OCCA) unless otherwise noted. See Williams v. State, 188 P.3d 208, 214-218 (Okla.Crim.App. 2008). We presume that the OCCA’s factual findings are correct. See 28 U.S.C. § 2254(e)(1).

On the morning of June 22, 2004, two gunmen (one wearing a black-hooded sweatshirt and the other wearing a white-hooded sweatshirt) robbed the First Fidelity Bank in Tulsa, Oklahoma. Williams, 188 P.3d at 214. Both men wore ski masks. Id. During the robbery, the gunmen shot three people — -bank customer Howard Smith, bank president Mark Poole, and bank teller Amber Rogers. Id. When the gunmen entered the bank, the one wearing white ordered Poole to open the safe. He complied, but the safe would not open because it was on a fifteen-minute time delay. Id. Not long after, Smith entered the bank. He saw the gunman wearing white, but not the one wearing black. As Smith raised his arms, the gunman in black shot him twice from behind. Id.; (Trial. Tr. vol. Ill at 745-46.) That gunman then went behind the teller area, where the gunman wearing white was arguing with Poole. The gunman in black shot Poole in his side, with the bullet traveling through his right arm before entering his chest. (Trial Tr. vol. III at 666, 679-80). Then the one in white stood above Poole and also shot him, hitting Poole in the leg. Williams, 188 P.3d at 214; (Trial Tr. vol. Ill at 671). As the two gunmen left, the one wearing white turned around and fired a shot that killed Rogers as she lowered her head and crouched on the floor. Id.; (Trial. Tr. vol. III at 721-22). Smith and Poole survived their gunshot wounds.

A witness’s description of the getaway car led police to Jeremy Williams and, soon after, to Alvin Jordan. The state charged both men with first-degree murder (under alternate theories of malice murder and felony murder), armed bank robbery, and shooting with intent to kill. Williams alone went to trial.

The evidence connecting Williams to the robbery was compelling. One of Jordan’s girlfriends testified that, sometime before the June 22 bank robbery, she overheard Williams tell Jordan about having previously robbed a bank located on the second floor of a building. First Fidelity was on the second story of a multi-use office building. In fact, a single gunman had robbed that same bank on May 11. After arresting Williams for the second bank robbery, police matched his fingerprints to those lifted from the bank after this first robbery.

Before the June 22 bank robbery, the same girlfriend went to Williams’s apartment with Jordan. While there, she saw a revolver resembling the one that the masked gunman dressed in black used on June 22. She also heard Williams tell Jordan that he would kill if he had to.

Another one of Jordan’s girlfriends placed Williams, Jordan, and the alleged getaway driver together at 4:00 a.m. the morning of the crime. In addition, Jordan’s aunt placed the three men together soon after the robbery and testified that Williams had boasted that he had shot some people and that he had divided the money with Jordan and the driver. According to her testimony, Williams said that he and Jordan each came away with $1100, leaving $700 for the driver. The bank reported just under $3000 stolen during the June 22 bank robbery.

Williams’s girlfriend testified that he arrived at their apartment later that morning with the same wad of stolen cash. That evening, the girlfriend saw Williams retrieve a ski mask and guns from the yard of an abandoned house and wipe the guns clean. Williams owned those guns, and their caliber and appearance matched the firearms used in the robbery. Police later determined that Williams’s DNA matched that found on the ski mask and that a footprint left at the bank matched the shoes he was wearing when police arrested him.'

On top of all this, Williams testified that he had robbed First Fidelity in May. He said he had jumped off the second-floor balcony when fleeing, just as one of the June 22 robbers had done. Nevertheless, Williams maintained that he did not rob First Fidelity on June 22.

Both gunmen shot people during the robbery — although it was not entirely clear who shot Amber Rogers. The state’s theory was that Williams was the gunman in black and that Jordan was the gunman in white. Eyewitnesses said that the gunman in black shot Smith from behind while the gunman in white commanded Rogers to unload the till. The gunman in black then went behind the teller area, where he and the other gunman both shot Poole; the gunman in black first shot Poole because Poole could not immediately open the time-delayed safes. As the gunmen fled the bank, one turned around and delivered the fatal shot to Rogers as she crouched on the floor. A bank employee identified the gunman in white as the killer. Yet in the wall behind Rogers’s teller station, investigators found a slug of the same caliber as the revolver used by the robber in black. Still images from the bank’s security cameras showed both gunmen in various positions, but they did not clearly depict how and when Rogers had been shot.

The state argued that it did not matter whether Williams was the actual trigger-man. The felony-murder charge certainly did not depend on it, and Williams could be guilty of malice-murder too, so long as he aided and abetted Jordan. The trial court instructed the jury to this end. Ultimately, using separate verdict forms, the jury found Williams guilty of both felony murder and malice murder.

At the penalty phase of trial, the state argued that Williams deserved the death penalty because of three aggravating circumstances: (1) the murder involved a great risk of death to more than one person; (2) the murder was committed to avoid arrest or prosecution; ' and (3) Williams posed a continuing threat to society. The state presented evidence of the life-threatening nature of Smith’s and Poole’s injuries and impact statements from Amber Rogers’s family. Otherwise, the state relied on the evidence it presented at trial.

The defense conceded the presence of the first aggravating circumstance but argued the state had failed to prove the other two. The defense also presented evidence of several mitigating circumstances, arguing that Williams: (1) did not have a prior criminal record; (2) was likely to be rehabilitated; (3) was just 21 at the time of the murder; (4) was under the influence of an emotional disturbance or intoxicants or both; and (5) had a difficult upbringing and home life. Social historian, Dr. Wanda Draper, and Williams’s mother recounted Williams’s turbulent family history for the jury.

In the end, the jury found that the murder involved a great risk of death to more than one person and that Williams was a continuing threat to society. The jury further found that the mitigating factors did not outweigh the aggravating factors and voted to impose the death penalty.

The OCCA affirmed the convictions and sentence on direct appeal, see Williams, 188 P.3d 208, and later denied post-conviction relief, see Williams v. State (First Application for Post-Conviction Relief), No. PCD-2006-1012, slip op. (Okla.Crim. App. Jan. 13, 2009) (unpublished). In response to Williams’s habeas petition, the federal district court denied his claims without an evidentiary hearing. But the district court did issue a certificate of appealability for two claims: (1) ineffective assistance of counsel (mostly during the guilt phase of trial) and (2) ineffective assistance of counsel at sentencing. This court then expanded the certificate of appealability to include two more claims: (1) sufficiency of the evidence to support Williams’s malice-murder conviction and (2) cumulative error. These four claims are now before us on appeal.

II. DISCUSSION

A. Standard of Review

“In general, if a convicted state criminal defendant can show a federal habeas court that his conviction rests upon a violation of the Federal Constitution, he may well obtain a writ of habeas corpus that requires a new trial, a new sentence, or release.” Trevino v. Thaler, — U.S. -, 133 S.Ct. 1911, 1917, 185 L.Ed.2d 1044 (2013). 28 U.S.C. § 2254 governs our review of habeas petition and focuses on how the state court resolved the claim. Byrd v. Workman, 645 F.3d 1159, 1165 (10th Cir.2011).

For claims that the state court adjudicated on the merits, we will only grant habeas relief if a petitioner establishes that the state court decision was “contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States,” or “was based on an unreasonable determination of facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(l)-(2).

We review de novo claims that the state court did not adjudicate on the merits. Hooks v. Workman, 689 F.3d 1148, 1163-64 (10th Cir.2012). A habeas petitioner must first exhaust his claims in state court before a federal court may review them. § 2254(b)(1)(A).

B. Sufficiency of the Evidence

We begin with Williams’s claim that the evidence was insufficient to support his conviction for first-degree malice murder. Even if Williams were to prevail, he would still be guilty of first-degree felony murder based on the jury’s separate verdicts. Why not then disregard this claim altogether and let Williams’s first-degree murder conviction stand on felony-murder grounds? There are at least two reasons why addressing the sufficiency claim is the better course.

First, the OCCA construed Williams’s verdict as one of malice murder. Williams, 188 P.3d at 225. This is Oklahoma’s practice in cases involving separate convictions of malice and felony murder because it avoids the need to vacate the underlying felony conviction (otherwise a source of double-jeopardy concerns). See Alverson v. State, 983 P.2d 498, 521 (Okla. Crim.App.1999). Thus, upholding Williams’s first-degree-murder conviction based solely on felony murder would disturb the OCCA’s preferred construction. It would also require dismissal of the underlying robbery conviction. See Harris v. Oklahoma, 433 U.S. 682, 682, 97 S.Ct. 2912, 53 L.Ed.2d 1054 (1977) (per curiam) (‘When, as here, conviction of a greater crime ... cannot be had without conviction of the lesser crime ... the Double Jeopardy Clause bars prosecution for the lesser crime, after conviction of the greater one.”).

Second, if Williams’s first-degree-murder conviction rested on felony murder alone, we would need to address a separate question: whether Williams is even eligible for the death penalty. See Ring v. Arizona, 536 U.S. 584, 609,122 S.Ct. 2428,153 L.Ed.2d 556 (2002) (holding that a jury, not a judge, must find facts necessary for imposition of the death penalty); Tison v. Arizona, 481 U.S. 137, 158, 107 S.Ct. 1676, 95 L.Ed.2d 127 (1987) (clarifying that the death penalty may be imposed on a felony murder defendant who was not the actual killer and who had no specific intent to kill, if evidence shows “major participation in the felony committed, combined with reckless indifference to human life.”) Although Williams would like us to reach this question, there is no need to do so if the evidence supporting his malice-murder conviction was sufficient. For the reasons discussed below, we conclude that it was.

In Jackson v. Virginia, the Supreme Court held that a conviction based on insufficient evidence violates the Due Process Clause of the Fourteenth Amendment. 443 U.S. 307, 316, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979). Weighing this constitutional guarantee against the jury’s exclusive role as fact-finder, Jackson extended the familiar sufficiency-of-the-evidence standard to the habeas realm: “the relevant question is whether, after viewing the evidence in the light most favorable to the prosecution, any rational trier of fact could have found the essential elements of the crime beyond a reasonable doubt.” Id. at 319, 99 S.Ct. 2781 (emphasis original).

In applying Jackson, we look to state law to determine the essential elements of the crime at issue. Id. at 324 n. 16, 99 S.Ct. 2781. Here, Oklahoma’s first-degree murder statute provides that:

A person commits murder in the first degree when that person unlawfully and with malice aforethought causes the death of another human being. Malice is that deliberate intention unlawfully to take away the life of a human being, which is manifested by external circumstances capable of proof.

Okla. Stat. tit. 21, § 701.7(A). Additionally, Oklahoma law punishes as a principal any person who aids and abets the commission of a crime. Okla. Stat. Ann. tit. 21, § 172 (“All persons concerned in the commission of crime, whether it be felony or misdemeanor, and whether they directly commit the act constituting the offense, or aid and abet in its commission, though not present, are principals.”). On this point, the state court instructed the jury that a principal “is one who directly and actively commits the act(s) constituting the offense or knowingly and with criminal intent aids and abets in the commission of the offense or whether present or not, advises and encourages the commission of the offense.” (Pleadings vol. VI at 1056). The state trial court also instructed the jury that:

One who does .not actively commit the offense, but who aids, promotes, or encourages the commission of a crime by another person, either by act or counsel or both, is deemed to be a principal to the crime if he knowingly did what he did either with criminal intent or with knowledge of the other person’s intent. To aid or abet another in the commission of a crime implies a consciousness of guilt in instigating, encouraging, promoting, or aiding in the commission of that criminal offense.

(Pleadings vol. VI at 1057).

Williams argues, that even if the evidence was sufficient to prove his involvement in the robbery and his intentional shooting of Smith and Poole, no evidence proved that he caused the death of Amber Rogers. The state does not argue otherwise. Nor did the OCCA believe that the state’s evidence pointed to Williams as the actual killer. Williams, 188 P.3d at 226. Accordingly, like the OCCA, we address whether the evidence was sufficient to support Williams’s conviction under an aiding and abetting theory.

A conviction for aiding and abetting can rest on a wide range of underlying conduct, including “acts, words or gestures encouraging the commission of the offense, either before or at the time of the offense.” Wingfield v. Massie, 122 F.3d 1329, 1332 (10th Cir.1997) (quoting VanWoundenberg v. State, 720 P.2d 328, 333 (Okla.Crim.App. 1986)) (internal quotation marks omitted). Some mental state beyond “mere assent” or “acquiescence” is also required, Wingfield, 122 F.3d at 1332, but in the malice-murder context, the OCCA has required even more. To convict an aider and abettor as a principal in a first-degree-malice-murder prosecution, the state must prove: “(1) that the defendant [that is, the aider and abettor] personally intended the death of the victim; and (2) that the defendant aided and abetted with full knowledge of the perpetrator’s intent.” Id. (citing Johnson v. State, 928 P.2d 309, 315 (Okla.Crim. App.1996)).

At least, this was the law. On Williams’s direct appeal, the OCCA suggested in a footnote that Johnson’s two-pronged intent requirement might be outdated:

According to Appellant’s brief, we must determine whether the evidence was sufficient to show that either Williams shot and intended to kill Amber Rogers, or Williams aided and abetted the Rogers’ killer with a personal intent to kill or he aided and abetted with full knowledge of the intent of the killer. See Johnson v. State, 1996 OK CR 36, ¶ 20 928 P.2d 309, 315. We overrule the language in Johnson which indicates this is the proper test and we continue to abide by the general aiding and abetting language. See Banks v. State, 2002 OK CR 9, ¶ 13, 43 P.3d 390, 397 (“Aiding and abetting in a crime requires the State to show that the accused procured the crime to be done, or aided, assisted, abetted, advised or encouraged the commission of the crime.”) We note that Appellant would even lose this proposition under the Johnson test, because his involvement was such that he personally had the intent to kill or knew that his codefendant had the intent to kill, when Amber Rogers was shot.

Williams, 188 P.3d at 225 n. 18.

We are unsure what to make of footnote 18. On one hand, the OCCA appears to have rejected the two-pronged intent requirement from Johnson. On the other hand, as we discuss below, the OCCA still seems to consider Williams’s case under Johnson in evaluating the sufficiency of the evidence.

Further muddling matters is the OCCA’s treatment of Banks. The OCCA cited Banks to state its adherence to “the general aiding and abetting language” as the “proper test” for malice murder under an aiding and abetting theory. Williams, 188 P.3d at 225 n. 18 (citing Banks, 43 P.3d at 397 (“Aiding and abetting in a crime requires the State to show that the accused procured the crime to be done, or aided, assisted, abetted, advised or encouraged the commission of the crime.”)). Yet, the OCCA in Banks also stated nearly identical language to that from Johnson:

To convict Banks of malice aforethought murder, the jury had to find that he caused the unlawful death of a human with malice aforethought, or aided and abetted another in the commission of the murder with the personal intent to kill, and with knowledge of the perpetrator’s intent to kill.

See Banks, 43 P.3d at 397 (emphasis added).

We are uncertain why the OCCA overruled Johnson without any mention of this similar language from Banks. We also note that, as far as we can tell, the only meaningful difference between the Johnson and Banks standards is that Johnson states that the aider and abettor must intend the death “of the victim,” 928 P.2d at 315, and Banks does not, 43 P.3d at 397.

Further adding to our confusion is the OCCA’s statement that it •will abide by its “general aiding and abetting language.” Williams, 188 P.3d at 225 n. 18. This language — found in the very next sentence of Banks — features no mens rea requirement at all, but simply provides that the aider and abettor must advise, encourage, assist — or, father unhelpfully, aid and abet. Banks, 43 P.3d at’ 397. This suggests to us that an Oklahoma conviction for aiding and abetting malice murder may no longer require intent of any kind.

That would cause serious problems. We generally disfavor offenses that require no mens rea. See Staples v. United States, 511 U.S. 600, 606, 114 S.Ct. 1793, 128 L.Ed.2d 608 (1994). One of the “basics” about aiding and abetting is the intent requirement — “a person aids and abets a crime when (in addition to taking the requisite act) he intends to facilitate that offense’s commission.” Rosemond v. United States, — U.S. -, 134 S.Ct. 1240, 1248, 188 L.Ed.2d 248 (2014). Oklahoma’s provision for aiding and abetting, which apparently requires nothing more than “advising or encouraging,” seems to miss the mark.

Williams asserts that, whatever the OCCA did, it did not give him the benefit of Johnson. According to him, footnote 18 shows that the OCCA failed to consider whether the evidence was sufficient to prove the crime. He maintains that no rational jury could have found him guilty of the essential elements of aiding and abetting malice murder as those elements are set forth in Johnson — namely, that he “intended” Rogers’s death. See Johnson, 928 P.2d at 315. Additionally, he argues that the OCCA’s overruling of Johnson violated due-process limitations on the retroactive application of new rules of law. See Rogers v. Tennessee, 532 U.S. 451, 459, 121 S.Ct. 1693, 149 L.Ed.2d 697 (2001).

The state responds that Williams cannot raise the Rogers argument because he failed to exhaust it in state court. Of course, Williams had no way of knowing that the OCCA would purportedly overrule Johnson on direct appeal. He did seek rehearing on that basis, which the OCCA denied. Still, according to the state, Williams did not fairly present his claim because a petition for rehearing is discretionary. See Castille v. Peoples, 489 U.S. 346, 349, 109 S.Ct. 1056, 103 L.Ed.2d 380 (1989) (holding that presentation of claims to a State’s highest court on discretionary review does not satisfy exhaustion requirements of 28 U.S.C. § 2254.)

We assume (without deciding) that the state is right and that Williams should have raised his “ex post facto” argument by way of post-conviction application. Because footnote 18 intertwines with Williams’s properly exhausted Jackson claim, however, we will consider Williams’s arguments on the merits. See 28 U.S.C. § 2254(b)(2) (“An application for a writ of habeas corpus may be denied on the merits, notwithstanding the failure of the applicant to exhaust the remedies available in the courts of the State.”) In the end, despite the OCCA’s confusion in, and our concern about, footnote 18, we do not believe Williams is entitled to habeas relief because of it.

While the OCCA may have overruled Johnson, it also indicated that Williams’s sufficiency claim failed under that very standard. See Williams, 188 P.3d at 225 n. 18 (“Appellant would even lose this proposition under the Johnson test.”) True, as Williams points out, the OCCA then proceeded to misstate Johnson in the very next breath — claiming it requires evidence of intent to kill or knowledge of the perpetrator’s intent, instead of evidence that “the aider and abetter personally intended the death of the victim and aided and abetted with full knowledge of the intent of the perpetrator.” Johnson, 928 P.2d at 315 (emphasis added). Even so, in the body of its opinion, the OCCA was faithful to Johnson. It considered whether the evidence was sufficient to prove Williams’s intent to kill, and it discussed Williams’s knowledge of Jordan’s intent. Williams, 188 P.3d at 226. Although the OCCA overruled Johnson, it still evaluated Williams’s arguments under what Johnson previously required. See id. at 226. Regardless, and no matter what footnote 18 says or means, we believe the OCCA weighed the evidence against the essential elements of the crime.

In one respect, however, the OCCA did not adopt Williams’s view of the elements. As Johnson stated, a person aids and abets malice murder if he intends the death “of the victim.” 928 P.2d at 315. Given this, Williams argues the evidence in his case needed to show — but did not show — that he intended the death of Amber Rogers, not just anyone. The OCCA disagreed. In its view, it was enough that there was sufficient evidence of Williams’s general intent to Mil and his knowledge of Jordan’s similar general intent. Williams, 188 P.3d at 226.

Addressing Williams’s Jackson claim, we decline the invitation to consider whether the OCCA should have required proof of intent to kill Rogers because, in our view, this is fundamentally a matter of state law. See Anderson-Bey v. Zavaras, 641 F.3d 445, 448-52 (10th Cir.2011) (rejecting Jackson claim that amounted to a challenge to the state court’s interpretation of an “uncertain” statutory term.) This is not to say that courts may simply ignore elements previously (and unequivocally) deemed essential in resolving a sufficiency-of-the-evidence claim. However, before Williams’s case, the OCCA had not yet addressed the requirements of aiding and abetting malice murder under truly analogous circumstances — that is, where two gunmen shot different people during the same criminal enterprise, but only one of the shooting victims died. Other cases addressing aiding and abetting in the malice-murder context involved the death of the targeted victim. See, e.g. Young v. State, 12 P.3d 20, 29-30, 40 (Okla.Crim. App.2000); Torres v. State, 962 P.2d 3, 8, 16-17 (Okla.Crim.App.1998).

It would make sense in those cases that the question would be phrased as whether the aider and abettor intended the death “of the victim” — ie. the targeted person who died. Here, however, the OCCA determined that, where there are multiple targets and only one death, the gunman whose target survives may be convicted of first-degree malice murder if he knows his cohort intended to kill, and the cohort’s gunshot turns out to be fatal. Although it would have been reasonable to reach a different conclusion, the OCCA did not— and its interpretation is authoritative. See Estelle v. McGuire, 502 U.S. 62, 67-68, 112 S.Ct. 475, 116 L.Ed.2d 385 (1991) (“[I]t is not the province of a federal habeas court to reexamine state-court determinations on state-law questions.”).

Nor are we persuaded that the OCCA’s resolution of Williams’s Jackson claim offended the due process guarantee of fair warning. The Supreme Court has “repeatedly held that a state court’s interpretation of state law, including one announced on direct appeal of the challenged conviction, binds a federal court sitting in habeas corpus.” Bradshaw v. Richey, 546 U.S. 74, 76, 126 S.Ct. 602, 163 L.Ed.2d 407 (2005) (per curiam) (emphasis added). This makes sense in light of the discussion above; it is not unusual that courts need to clarify and interpret prior opinions as new circumstances and fact patterns come up against the law — particularly against common law doctrines (such as intent). See Rogers, 532 U.S. at 461, 121 S.Ct. 1693. So long as any interpretation (or alteration) is not “unexpected and indefensible by reference to the law which had been expressed prior to the conduct in issue,” there are no due process concerns. Id. at 462, 121 S.Ct. 1693 (citation omitted). Whatever measure of evolution the OCCA took in extending its aiding and abetting law to Williams’s malice-murder conviction, we can hardly say it was unexpected or indefensible.

Having addressed footnote 18 and Williams’s due-process arguments on that score, we turn now to the sufficiency of the evidence. As discussed, the OCCA weighed the evidence against the requirements set forth in Johnson, with the exception of the particular requirement that Williams needed to intend the death of Amber Rogers. We approve that weighing. The question now is whether the evidence met the constitutional threshold — or, more precisely, because we address the question on habeas — whether the OCCA’s determination that the evidence was sufficient to support the jury’s verdict was itself reasonable. Hooks, 689 F.3d at 1167.

Once the jury concluded that Williams robbed First Fidelity, the evidence was susceptible to limited interpretations. See Torres v. Mullin, 317 F.3d 1145, 1155 (10th Cir.2003) (deciding that a rational juror could conclude that the defendant had the requisite intent to kill, despite evidence that was susceptible to interpretation). As the state argued, the jury could have concluded that Williams and Jordan jointly planned to rob the bank and to kill whoever stood in their way. Alternatively, the jury might have concluded that the plan was just to rob the bank and that Williams never intended for anyone to die. But we cannot consider that option, because the jury found that Williams shot with intent to kill when it convicted him of two counts of that crime.

Another possibility is that Williams intended to kill but that Jordan did not. But the evidence belies this theory because Jordan shot Rogers at close range in the middle of her torso while she was crouching on the ground. This leaves just one possible interpretation that might lead to acquittal: that Williams intended to kill without knowing that Jordan intended to do the same. This is an unflattering defense to say the least. Apparently, Williams would like us to believe that he knew himself to be capable of murder but that he thought better of his friend.

The OCCA was skeptical of this notion on direct appeal and said as follows:

It is clear that [Williams] intended to kill at the bank. It is also clear that he knew that his codefendant was armed with a loaded weapon and both of them had spoken of killing, “if they had [to],” in preparation for this robbery. If he had intended to kill when he shot, how could he not know that his codefendant also shot with intent to kill? These two defendants acted with one accord and the evidence shows that they shot each person with intent to kill.

Williams, 188 P.3d at 226.

Some of the OCCA’s observations are more persuasive than others. Least persuasive in our view is the observation that Jordan spoke of killing if he had to. We find no record support for attributing this remark to Jordan. In fact, it was Williams who said that “he would kill if he had to” — as the OCCA correctly noted in its recitation of the record. (Trial Tr. vol. V at 1106.) The OCCA may have merely intended to observe that Jordan participated in conversations about killing. Regardless, this finding, one that Williams does not challenge here, was just one supporting the OCCA’s conclusion.

Somewhat more persuasive is the OCCA’s observation that Williams’s own clear intent to kill made it unlikely that Williams could be ignorant that Jordan shot with a different purpose. This might suggest that the OCCA believed that Williams became aware of Jordan’s intent to kill when Jordan first shot Poole. That could be. More likely, however, we think the OCCA simply meant to point out that it was unlikely that Williams was ignorant of Jordan’s intent to kill based on his own obvious intent.

This brings us to the OCCA’s final and most persuasive point-that Williams and Jordan acted together with a shared understanding. The following evidence supports this conclusion:

• According to one witness, Williams and Jordan hung out nearly every day in the six months leading up to the robbery. Another witness said that the two “was running together,” and neither one was really the leader of the other. (Trial Tr. vol. V at 1118, 1157).'

• During this time, Jordan was nearly always armed — and Williams supplied him with a gun on at least one occasion.

• Williams knew that Jordan used guns to get what he wanted. In fact, Williams testified that Jordan had pulled a gun on him more than once.

• Williams also testified that he knew that Jordan had shot some people during his robbery of a convenience store the week before the robbery.

• Williams testified that he “wasn’t surprised” that Jordan had shot people at the convenience store. (Trial Tr. vol. VII at 1547.) His friend had a reputation for being particularly “crazy” and “violent.” (Trial Tr. vol. V at 1148).

• Jordan did not back out of the robbery even though Williams said that he was prepared to kill if he had to.

• Finally, following the robbery, there was no evidence that Williams was surprised that Jordan shot people (including Amber Rogers) at close range when they were inside the bank.

In view of this evidence, we cannot say that the OCCA’s decision was contrary to or an unreasonable application of Jackson. Certainly, the OCCA’s reasoning was not altogether clear, accurate, or comprehensive. In supporting our conclusion, we rely on additional evidence that the OCCA did not. Still, we pay deference to the OCCA’s ultimate decision — “after all, what matters is that the evidence support the OCCA’s result.” Torres, 317 F.3d at 1156 (emphasis added). We are confident that the evidence supporting Williams’s malice-murder conviction was constitutionally sufficient. See Jackson, 443 U.S. at 313-14, 99 S.Ct. 2781 (stating that conviction can occur only when there is “evidence that is sufficient fairly to support a conclusion that every element of the crime has been established beyond a reasonable doubt.”).

C. Ineffective Assistance of Counsel (Guilt Phase)

1. The Applicable Standard

We turn now to Williams’s claim that his lawyer inadequately responded to the state’s evidence and witnesses — primarily during the guilt phase. Although Williams had two lawyers, he appears to focus this ineffective-assistance claim on lead counsel’s alleged failures. We imagine this is because Williams contends that lead counsel was under the influence of drugs and alcohol throughout the trial.

To prevail on a Sixth Amendment claim of ineffective assistance, a defendant must show both that (1) counsel “committed serious errors in light of prevailing professional norms such that his legal representation fell below an objective standard of reasonableness,” and (2) there is “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Grant v. Trammell, 727 F.3d 1006, 1017 (10th Cir.2013) (internal quotation marks omitted) (citing Strickland v. Washington, 466 U.S. 668, 694, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984)).

2. The OCCA’s Decision

In rejecting on direct appeal the ineffective assistance claim based on his counsel’s failure-to-object during both trial and sentencing, the OCCA relied in part on its resolution of Williams’s related claim that the trial court committed plain error by not disallowing certain testimony even absent objection. See Williams, 188 P.3d at 230 n. 20 (referring to the discussion of Williams’s proposition that the state introduced highly prejudicial evidence during trial). This claim regarding certain trial testimony is not before us on habeas. Nonetheless, in evaluating Williams’s ineffective assistance claims, the OCCA considered whether the trial court’s admission of that testimony rose to the level of “plain error affecting substantial rights.” Id. at 222; see also Okla. Stat. Ann. tit. 12, § 2104 (“Nothing in this section precludes taking notice of plain errors affecting substantial rights although they were not brought to the attention of the court.”). Under Oklahoma law, an error affects the defendant’s substantial rights if it “affect[s] the outcome of the proceeding.” Hogan v. State, 139 P.3d 907, 923 (Okla. Crim.App.2006). The OCCA concluded that no proposition of error rose to this level.

Turning then to Williams’s Strickland claim, the OCCA referred back to its plain-error determinations:

Williams points out that counsel failed to object to the introduction of several pieces of testimonial and real evidence, which he has complained about in several propositions in this appeal. In discussing these propositions of error, we found that either there was no error or that the error did not rise to the level of plain error.

Williams also argues that counsel was ineffective for failing to object to several instances of prosecutorial misconduct which are raised as error in proposition six. We noted that counsel did object to the most egregious instances of misconduct.

We further find that counsel’s failure to object to the introduction of certain items of evidence and the prosecutor’s alleged misconduct did not rise to the level of ineffective assistance of counsel under the Strickland standard.

Williams, 188 P.3d at 231.

Williams does not challenge the OCCA’s approach to resolving his Strickland claim. Initially, we had. our own doubts as to whether the OCCA’s approach was faithful to federal law. True enough, when a defendant fails to show that a trial court’s admission of evidence was improper for some reason, it likely follows that the lawyer did not perform deficiently by failing to object to its admission.¹ See Cannon v. Mullin, 383 F.3d 1152, 1162 (10th Cir.2004) (finding no deficient performance in defense counsel’s failure to object when there was “no meritorious state-law objection available” to counsel).

However, it does not necessarily follow that there can be no prejudice under Strickland when there is no plain error under Oklahoma’s plain error standard. This is because, as far as we can tell, Oklahoma’s substantial rights/plain error standard requires a defendant to show more than what is necessary to satisfy the prejudice standard under Strickland. That is, for a defendant to meet Oklahoma’s plain error standard, he must show that the error affected the outcome of the proceeding. Hogan, 139 P.3d at 923. But to satisfy Strickland, a defendant need only show that counsel’s error created a reasonable probability that the proceeding’s outcome would be different. Strickland, 466 U.S. at 694, 104 S.Ct. 2052 (emphasis added).²

Of course, if the OCCA made a positive finding that certain evidence did not affect the outcome of trial, then this would necessarily include a finding that the evidence did not influence the outcome to some lesser degree — or put another way, that the evidence did not create a reasonable probability of a different outcome. Here, however, the OCCA simply concluded that there was no plain error. Under that general conclusion, it is impossible to say whether Williams lost because he just could not make the requisite showing that the error was outcome-determinative or because the OCCA affirmatively concluded that the error did not determine the outcome. The distinction is a subtle one, but it matters if the OCCA treated its plain-error determinations as dispositive.

The OCCA did not. Instead, that court made the “further finding” that Williams failed to satisfy the requirements of Strickland — not just the requirements of plain error. Williams, 188 P.3d at 231. Because of this, we believe that the OCCA rejected Williams’s claims under the appropriate federal standard. Moreover, based on our review of the record, we cannot say that the OCCA’s decision was an unreasonable application of that standard. See Thornburg v. Mullin, 422 F.3d 1113, 1138-39 (10th Cir.2005) (reaching the same conclusion when the OCCA found no plain error but also separately concluded that there was no prejudice under Strickland ).

None of this is to say that the OCCA’s disposition under Strickland was a model of clarity. Again, the OCCA simply concluded that Williams’s claims “did not rise to the level of ineffective assistance of counsel under the Strickland standard.” Williams, 188 P.3d at 231. From this statement, it is unclear which of Strickland’s prongs the OCCA believed Williams failed to satisfy. See Hooks, 689 F.3d at 1187 (recognizing that a court can decide a Strickland claim under either prong). The OCCA’s reference to its plain-error determinations does not help. By finding no error, the OCCA might have concluded that counsel did not perform deficiently. Conversely, it might just as reasonably have assumed deficient performance and concluded that there was no resulting prejudice, particularly since Williams often did not explain why his lawyer should have objected. When the OCCA found error not requiring reversal, it might still have concluded that the lawyer’s representation was reasonable. Alternatively, it might have concluded that any error (actual or assumed) did not undermine confidence in the outcome. In short, we have no way of knowing what the OCCA was thinking.

This uncertainty does not change our deference. Even “[w]here a state court’s decision is unaccompanied by an explanation, the habeas petitioner’s burden still must be met by showing there was no reasonable basis for the state court to deny relief.” Harrington, 131 S.Ct. at 784; see Lafler v. Cooper, — U.S. -, 132 S.Ct. 1376, 1396, 182 L.Ed.2d 398 (2012) (“The state court’s analysis was admittedly not a model of clarity, but federal habeas corpus is a guard against extreme malfunctions in the state criminal justice systems, not a license to penalize a state court for its opinion-writing technique.”) (internal quotation marks omitted). This is true when, as here, the state court does not reveal which element of a claim it found insufficient. See Harrington, 131 S.Ct. at 784. Our task is still to evaluate the reasonableness of the OCCA’s application of Strickland, considering the reasonableness of the theories that “could have supported” the OCCA’s decision. Id. at 786. This panel must identify what those theories are.

We believe that the OCCA reasonably could have resolved Williams’s challenges under the first prong of Strickland by concluding that Williams’s lawyer did not commit any “serious errors in light of prevailing professional norms such that his legal representation fell below an objective standard of reasonableness.” Wackerly v. Workman, 580 F.3d 1171, 1176 (10th Cir. 2009) (internal quotation marks omitted). First, we consider Williams’s argument that lead counsel was constructively absent because of counsel’s substance abuse. We then will discuss Williams’s challenges involving his trial counsel’s alleged errors. Next, we address the challenges involving Officer Felton’s and Dyra Malone’s testimony. In those two cases, we assume deficient performance but nonetheless uphold the OCCA’s decision for lack of prejudice under prong two of Strickland. Finally, we decline to consider Williams’s challenge to alleged prosecutorial misconduct because he inadequately briefed it.

3. Trial Counsel’s Alleged Substance Abuse

Williams recognizes that Strickland is the default standard for ineffective-assistance claims, but he argues that a different standard should apply in his case. Because of the alleged substance abuse, Williams contends that his lawyer “may have been” constructively absent from trial, thereby entitling him to relief under United States v. Cronic, 466 U.S. 648, 658-59, 104 S.Ct. 2039, 80 L.Ed.2d 657 (1984). Williams raised this argument in his second application for state post-conviction relief after he read a list-serv email that his lawyer authored. In that email, the lawyer acknowledged the toll of taking on death-penalty cases. He also wrote that he “pop[s] valium like candy just to face the day,” yet he “can only lay off the valium and alcohol during trial.” (R. vol. I at 617, 747-49.) Both the OCCA and the district court rejected the notion that the substance abuse mattered. They also denied Williams’s requests for an evidentiary hearing to explore the subject.

Just as the OCCA and the district court did, we apply Strickland here. In evaluating Williams’s claims, we turn to Cronic, where the Supreme Court identified three extreme situations “so likely to prejudice the accused that the cost of litigating their effect in a particular case [under Strickland ] is unjustified.” Bell v. Cone, 535 U.S. 685, 695, 122 S.Ct. 1843, 152 L.Ed.2d 914 (2002) (citing Cronic, 466 U.S. at 658-59, 104 S.Ct. 2039). See also Hooks v. Workman, 606 F.3d 715, 724 (10th Cir.2010). Williams’s bases his constructive-absence argument on Cronic’s second situation: “a presumption of prejudice is warranted if counsel entirely fails to subject the prosecution’s case to meaningful adversarial testing.” Hooks, 606 F.3d at 724 (internal quotation marks omitted) (emphasis added). This means that the lawyer’s failures must run throughout the entire proceeding, Hooks, 689 F.3d at 1186, and that the lawyer’s performance “be so inadequate that, in effect, no assistance of counsel is provided.” Cronic, 466 U.S. at 654 n. 11, 104 S.Ct. 2039.³

There is simply no way that the isolated failures of Williams’s lawyer (assumed below) rise to this level of complete failure. See Bell, 535 U.S. at 696-97,122 S.Ct. 1843 (‘When we spoke in Cronic of the possibility of presuming prejudice based on an attorney’s failure to test the prosecutor’s case, we indicated that the attorney’s failure must be complete.”). Both of Williams’s lawyers fought vigorously for acquittal, and later, against the death penalty. The lawyers filed numerous pre-trial motions, made compelling arguments during first- and second-stage proceedings, asked for sidebars and hearings outside the jury’s presence, raised numerous objections to the state’s evidence and arguments, and presented a cogent theory of innocence. See Hooks, 689 F.3d at 1186 (collecting cases and finding Cronic standard inapplicable based on similar findings).

Even if Williams could prove that one of his lawyers was really under the influence of drugs or alcohol at trial, this would not negate their professional efforts on his behalf. For this reason alone, there is no need for an evidentiary hearing to assess Cronic’s applicability. See Littlejohn v. Trammell, 704 F.3d 817, 858 (10th Cir. 2013) (recognizing that, even when we owe no deference to the state court’s decision, we can grant a habeas petitioner an evidentiary hearing only when his allegations, if true, “would entitle him to habeas relief.”) (internal quotation marks omitted). Nor is there a need for a hearing under Strickland, where our concern is the objective reasonableness of the lawyer’s conduct — not the lawyer’s subjective reasoning. See Harrington v. Richter, 562 U.S. 86, 131 S.Ct. 770, 790, 178 L.Ed.2d 624 (2011) (“Strickland ... calls for an inquiry into the objective reasonableness of counsel’s performance, not counsel’s subjective state of mind.”).

We turn now to Williams’s other arguments of ineffective assistance of counsel. First, he argues that-his lawyer failed to object to several pieces of evidence and to various statements by the prosecutor — we count seven total alleged deficiencies. Second, he argues that his lawyer failed to prepare to impeach one of the state’s witnesses. Williams raised both of these challenges on direct appeal, although the OCCA resolved only his failure-to-object arguments. Even so, we presume “that the state court adjudicated the claim on the merits in the absence of any indication or state-law procedural principles to the contrary.” Johnson v. Williams, — U.S. -, 133 S.Ct. 1088, 1094, 185 L.Ed.2d 105 (2013). Although this presumption is rebuttable, see id. at 1096-97, Williams offers no argument why the presumption should not apply. Thus, the deferential standards of § 2254(d) apply to Williams’s failure-to-object and failure-to-prepare claims. Hooks, 689 F.3d at 1163. Yet, we note, the outcome would remain the same even under de novo review.

4. Failure to Object to Evidence of the Stolen Watch

Williams first complains that his lawyer should have objected to evidence about a watch found in the apartment of Jordan’s girlfriend, Tarina Clark. Notably, this is the same apartment where Williams, Jordan, and the getaway driver met the night before the June 22 robbery. The state first presented evidence of Williams’s DNA on the watch. On cross-examination, the prosecutor then asked Williams if he knew the watch was stolen. Williams said he did not know and that Jordan gave him the watch. Williams seems to argue that his lawyer should have objected to both the DNA evidence and to the question about whether he knew that the watch was stolen. Except for relevance, he does not identify a basis for the objection, but he instead complains that the evidence was irrelevant and portrayed him as a thief.

Regardless of whether the question elicited relevant evidence, we believe that the DNA evidence was highly relevant because it tied Williams to Clark’s apartment and corroborated her testimony about his presence there soon before the bank robbery. On these grounds, we imagine the trial court would have overruled any objection from Williams. The DNA evidence itself had nothing to do with the watch’s being stolen and, as the OCCA pointed out, the presence of Williams’s watch at Clark’s apartment corroborated her testimony that Williams was there. See Williams, 188 P.3d at 220.

Addressing the question about the watch’s status as stolen, we defer to the OCCA’s view that it was improper. See Okla. Stat. Ann. tit. 12, § 2404 (“Evidence of other crimes, wrongs, or acts is not admissible to prove the character of a person in order to show action in conformity therewith.”) The question elicited testimony that Williams possessed stolen things. Its sole purpose was to suggest Williams’s propensity to steal — as he rightly claims.

The OCCA nonetheless could reasonably conclude that trial counsel was not deficient for failing to object. As the OCCA recognized, the stolen-watch evidence was hardly significant when compared to the evidence of Williams’s other crimes, much of which the defense introduced. Williams, 188 P.3d at 220. Williams himself relied on his general thievery to explain the wad of cash that police seized from him the day of the robbery (he claimed that the money came from his stealing and selling his girlfriend’s TV). In addition, to distance himself from the shoeprint found at First Fidelity, he claimed that he had stolen several identical pairs of shoes and sold them to others. Other evidence did not tend to show innocence. Along the same line, the defense elicited testimony from Dyra Malone that Williams stole most of the things he owned. Williams himself admitted that he sold stolen weapons and used drugs. He also admitted that he robbed First Fidelity in May.⁴

Nor can we say that Williams’s counsel was deficient in introducing this evidence. The wad of cash and matching shoeprints needed an explanation. Moreover, by “owning up” to conduct he could hardly deny, the defense had a chance to bolster credibility. If Williams freely admitted his other crimes, the jury might conclude that he was telling the truth when he denied his involvement in the June 22 robbery. See Bullock v. Carver, 297 F.3d 1036, 1051 (10th Cir.2002) (“As a general matter, we presume that an attorney performed in an objectively reasonable manner because his conduct might be considered part of a sound strategy.”) (emphasis original). Had Williams’s lawyer objected to the evidence of the stolen watch, he might have drawn unnecessary attention to it or otherwise suggested to the jury that Williams had something to hide. A reasonable defense lawyer could choose a strategy of not objecting under the circumstances, and “it would be well within the bounds of a reasonable judicial determination for the state court to conclude” that counsel’s performance was not deficient. Harrington, 131 S.Ct. at 789.

5. Failure to Object to Photographs of Tarina Clark’s Apartment

The state introduced photographs of the apartment that Jordan shared with his girlfriend, Tarina Clark. Except perhaps relevance, Williams simply does not identify a valid objection to admission of the photographs. He repeats his assertion from direct appeal that the photographs prejudiced him. In particular, he complains that the photographs invited “impermissible inferences about his lifestyle and the people with whom he associated.” (Appellant’s Br. at 69).

We agree with the OCCA that the photographs had probative value. Williams, 188 P.3d at 223. They helped corroborate Clark’s testimony. For example, one photograph showed towels pushed up against the dishwasher; Clark had testified that she was angry with Jordan the night before the robbery for allowing the dishwasher to.overflow. Another photograph showed a pair of Fubu tennis shoes like the ones Williams owned (with soles matching the footprint taken from the first robbery of First Fidelity). Clark testified that Jordan owned those shoes but rarely wore them because they were too small. Accordingly, we think the OCCA reasonably could have concluded that Williams’s lawyer’s failure to object to the photographs was not deficient performance.

6. Failure to Object to Post-Autopsy Photographs of Amber Rogers

The state admitted without objection two photographs taken at the medical examiner’s office showing Amber Rogers’s nude body. Williams argues that any reasonably competent lawyer would have objected to the photographs, “which served only to evoke the passions and sympathy of the jury.” Appellant’s Br. at 70. Williams also suggests that the photographs were not relevant because the manner of death was not disputed and because the body bore signs of medical intervention by that point in time. The prosecutor referred to the photographs in second-stage closing argument.

The OCCA found that the trial court did not err in admitting the photographs, which “show[ed] the handiwork of the defendant” and “more closely depict[ed] the nature and extent of the gunshot wound on the victim’s body than any other evidence available.” Williams, 188 P.3d at 223. We similarly conclude that photographs of a victim’s body, while gruesome, can be relevant when they depict the extent of injuries and are probative of intent to kill. See, e.g., Wilson v. Sirmons, 536 F.3d 1064, 1115 (10th Cir .2008). Nothing Williams says leads us to believe that the photographs were so obviously prejudicial that counsel’s failure to object was deficient. The OCCA could have reasonably rejected this claim on the same theory.

During penalty-phase closing argument, the prosecutor referred to the photographs as the “most powerful image.” (Trial Tr. vol. VIII at 1872). But even if a reasonably competent lawyer would have objected to this, Williams does not explain why the same lawyer would have anticipated the prosecutor’s comment as a reason to object to the photographs themselves. See Harrington, 131 S.Ct. at 789 (“Reliance on the harsh light of hindsight ... is precisely what Strickland and AEDPA seek to prevent.”) (internal quotation marks omitted). The photographs are the subject of this challenge — nothing else.⁵

7. Failure to Object to Testimony of Treating Physicians

Amber Rogers’s treating physician testified at length about the medical treatment he gave Rogers before her death. This testimony included details of surgery. The state also presented testimony during the penalty phase from Smith’s and Poole’s treating physicians. These doctors similarly testified about what they did to treat the gunshot-related injuries.

On direct appeal, Williams contended that the testimony of the treating physicians was irrelevant. The OCCA disagreed:

The State is obligated to show that the death was caused by the criminal actions of the defendant. In order to show that, in this case, the State had to show that Amber Rogers died despite the heroic efforts of the surgery team. There was no plain error here.

Williams also complains about the second stage testimony of the surgeons that treated the other victims who did not die. Again, there was no objection to this testimony, thus we review for plain error only. 12 O.S.2001, § 2104. Here, one of the aggravating circumstances alleged was that Williams created a great risk of death to more than one person. Although, Williams claims that evidence that these two victims were shot was sufficient to show a great risk of death to more than one person, our cases reveal that testimony about the nature and extent of gunshot wounds are relevant for this aggravating circumstance.

Williams, 188 P.3d at 224.

In connection with his Strickland claim, Williams argues that certain unspecified details of the doctors’ testimony were “unfairly prejudicial and were introduced only to inflame the passions of the jury.” Appellant’s Br. at 71. With respect to the penalty-phase testimony, he goes even further, claiming that the doctors’ testimony was “entirely irrelevant.” Id. Given the OCCA’s observations to the contrary, we cannot agree. In our view, it was reasonable to conclude that the testimony was relevant because it supported the state’s theory that Williams’s crime showed “a great risk of death to more than one person.” Williams, 188 P.3d at 224. We think it follows that the OCCA could have reasonably concluded that counsel’s failure to object on relevance grounds was not deficient performance.

8. Failure to Object to Detective Felton’s Testimony

The investigating police officer, Detective Felton, testified that shortly after Williams’s arrest, he saw abrasions and lacerations on Williams’s shin. When asked if he found anything significant about the injuries, Felton testified: “Meeting with the other detectives it was determined that one of the suspects had fled the bank by jumping off the second floor balcony. These looked just like injuries that one might receive by, you know, jumping and falling.” (Trial Tr. vol. V at 1181).

Williams contends that this testimony was plainly inadmissible because it was not based on Felton’s perceptions as a lay witness. Presumably, he thinks that his counsel should have objected for this reason.

The OCCA concluded that Felton’s testimony was improper because it was based on specialized knowledge. Williams, 188 P.3d at 225. We accept this and assume that a reasonably competent lawyer would have objected. Still, we think the OCCA could have reasonably concluded that there was no reasonable probability of acquittal but for counsel’s unprofessional errors. See Strickland, 466 U.S. at 694, 104 S.Ct. 2052.

"While Felton’s testimony certainly linked Williams to the crime, and while testimony from a police officer can be particularly persuasive to juries, we believe that Felton’s opinion may have bordered on pure speculation. The OCCA concluded similarly. Williams, 188 P.3d "at 225 (concluding that the opinion, without specialized knowledge, would be “pure speculation”). The photographs Felton referenced showed generic injuries that any number of accidents could have caused. The jury might even have believed Williams that he did not know what caused the injuries. Further, Felton did not flatly declare that Williams had jumped off the balcony; he merely said that Williams’s injuries were consistent with those someone might sustain by “jumping and falling.” Given the isolated, and limited nature of the offending testimony and the overwhelming evidence that Williams robbed the bank, we cannot begin to say that, absent this testimony, Williams would have had a reasonable probability of acquittal. The OCCA’s decision under Strickland was objectively reasonable.

9. Failure to Object to Dyra Malone’s Testimony

Williams next complains about a portion of the redirect testimony of his girlfriend, Dyra Malone. Malone testified on direct examination about her interactions with Williams on the day of the robbery. She said that Williams came to her apartment around 11:00 a.m. with “wads” of cash and said that he had “jacked a white man.” (Trial Tr. vol. V at 1019, 1023). She did not say that Williams had confessed to anything more. On cross-examination, Malone admitted that she had spoken with prosecutors at least a dozen times, and first spoken with defense counsel on the day she testified. She further testified that she did not ask Williams any questions after he displayed wads of cash after supposedly having “jacked a white guy.” (Trial Tr. vol. V at 1063). Malone briefly described her own arrest and how police had questioned her at length, but she did not say what the police had asked her or what she had told them.

On redirect examination, however, the prosecutor began with this:

Prosecutor: Ma’am, you had been asked about the statements to police. What did you tell police when they talked to you?

Malone: Um—

Prosecutor: Let me rephrase the question .... Did you tell police that Jeremy had told you something about a bank, his cousin Tony, and a lady, and that he said his cousin started shooting? Did you tell police .that Jeremy had told you that?

Malone: Yes.

Prosecutor: And that was in the first statement to police; is that correct?

Malone: Yes.

Prosecutor: And at the preliminary hearing you said that you didn’t say that, is that correct, or that he didn’t tell you that?

Malone: He didn’t .tell me that.

(Trial Tr. vol. V at 1069). Malone further explained this earlier statement by saying that police had threatened her that if she “didn’t start talking they were going to put [her] in jail.” (Trial Tr. vol. V at 1069). She then said that she had based her early statement on what the police “told [her] about what happened.” (Trial Tr. vol. Y at 1069).

Williams contends that his lawyer should have objected to this questioning because “neither the prosecution nor defense [had] impeached Dyra Malone with any prior inconsistent statements.” Appellant’s Br. at 68. We are unsure what Williams means by this. Perhaps he is saying that the question was improper because the prosecutor had no inconsistent statement from Malone’s earlier testimony to impeach. Before the OCCA, Williams argued that the prosecutor’s question was inadmissible because it exceeded the scope of cross-examination and because there was no limiting instruction. See Williams, 188 P.3d at 222.

While we may not have divined the objection that Williams wanted his lawyer to make, we think we understand his ultimate concern: the only reason the state questioned Malone about her first statement to the police was to introduce Williams’s admitting to the crime.

Even assuming that an effective lawyer would have objected, we would still conclude that the OCCA’s decision was reasonable under Strickland. Williams simply cannot show a reasonable probability of acquittal had his lawyer objected and the evidence stayed out.⁶ In reality, Malone’s testimony on redirect likely weighed in Williams’s favor. After Malone testified about her initial statement to the police, she declared it spoon-fed and coerced. All the worse for the state, she then emphasized that Williams had never confessed to her having committed the bank robbery. Of course, none of this changes the fact that Williams did confess to another witness, Beverly Jordan. Moreover, Malone was consistent in maintaining that Williams told her he had “jacked a white man.” (Trial. Tr. vol. V at 1023). We are confident that the jury’s verdict would have been the same without Malone’s redirect testimony. The OCCA reasonably could have rejected this claim on a lack of prejudice.

10. Failure to Impeach Officer Kennedy

As stated above, the OCCA did not purport to resolve this failure-to-impeach claim. Even so, because Williams presented this claim to the OCCA, we presume that the OCCA adjudicated it on the merits, particularly in the absence of any argument from Williams to the contrary. Johnson, 133 S.Ct. at 1094. Thus, we apply § 2254(d)’s deferential standard of review to Williams’s failure-to-object and failure-to-prepare claims. Hooks, 689 F.3d at 1163. In doing so, we see that the district court found that this alleged failure simply “did not rise to the level of a constitutional violation.” Williams v. Workman, 2012 WL 5197674, at *18 n. 10 (N.D.Okla. Oct. 19, 2012) (unpublished). We agree. In our view, the OCCA could have reasonably concluded that the lawyer’s performance was not deficient. We would reach the same result even under de novo review.

About an hour after the bank robbery, Police Officer Kennedy stopped Williams for a moving violation. In connection with this stop, Kennedy did a pat-down search and discovered a wad of cash. At trial, he testified that he counted this money and that it amounted to approximately $1100 (the amount a previous witness identified as Williams’s claimed share of the robbery proceeds). At the suppression hearing, however, Kennedy had testified that he had not counted the money for an exact amount. Instead, he estimated that it was over $1000 based on his fingering through the cash.

Williams argues that his lawyer should have properly prepared to cross-examine Kennedy by having the suppression-hearing transcript on hand. Without the transcript, Williams claims that his lawyer could not effectively impeach Kennedy on cross-examination; he could only ask Kennedy if he remembered testifying that he had not counted the money. When Kennedy responded, “I don’t remember — I did count the money,” that was the end of it. (Trial Tr. vol. IV at 983).

When reviewing ineffective-assistance claims, we must make every effort “to eliminate the distorting effects of hindsight.” Strickland, 466 U.S. at 689, 104 S.Ct. 2052. Here, we think this means considering whether Williams’s lawyer had any reason to anticipate that Kennedy would change his testimony, making resort to the suppression-hearing transcript necessary. Williams gives us no reason to think this change in testimony was anything other than a “remote possibility].” Harrington, 131 S.Ct. at 779.

What’s more, we must consider whether a reasonable lawyer would even have perceived any inconsistency in Kennedy’s testimony. We cannot conclude that a reasonable lawyer would have immediately been able to perceive the inconsistency. True, Kennedy testified at the suppression hearing that he had not counted the money, but he also testified that he had estimated the amount to be over $1000 based on having fingered through the cash. At trial, Kennedy’s testimony was not much different: there he estimated the amount to be approximately $1100 based on his having counted the money.

Despite this minor inconsistency, Williams’s lawyer still managed to detect that Kennedy’s story had changed, acquire the suppression-hearing transcript, and finally persuade the trial court to admonish the jury “to disregard [Kennedy’s] statement that he counted money” and that “it was a thousand dollars.” (Trial Tr. vol. VII at 1531-32). We can hardly say that this conduct falls below the high standard of objectively unreasonable performance under Strickland.

11. Failure to Object to Prosecutorial Misconduct

Williams points to “five general categories of prosecutorial misconduct to which [his lawyer] failed to object.” (Appellant’s Br. at 72). These include: (i) the prosecutor’s expression of his personal opinion in first-stage closing argument that Williams was guilty; (ii) “numerous instances” of arguing facts not in evidence; (iii) unnecessary ridicule of Mr. Williams; (iv) “numerous instances” of “evoking improper sympathy for the victim;” and (v) expressions of the prosecutor’s personal opinion from the second-stage closing argument that Williams deserved the death penalty. (Id. at 72-73). Williams clarifies that his challenge extends only to “prosecutorial misconduct to which [his lawyer] made no objection.” (Id. at 73). He acknowledges that his lawyer did in fact object to at least some of the alleged misconduct. What Williams says is that the “numerous instances of prosecutorial misconduct in both phases of trial” were “constitutionally deficient [failures] in contravention of Strickland.” (Appellant’s Br. at 78).

Once again, Williams does not explain what objections his lawyer should have made. We have overlooked this problem in connection with his other challenges, but even if we did so here, we would face the added difficulty of trying to determine what alleged misconduct forms the basis of his complaint. We can hardly be sure what Williams means when he describes these five general “categories.” Apart from mentioning first- and second-stage closing arguments (both of which were lengthy), he does not direct us with any specificity to where we might locate the prosecutor’s allegedly improper statements in the transcript. Indeed, he does not include a single quote from the transcript or a single cite to the record in this section of his brief or habeas petition.

Williams supplies only one supporting citation. He points to the OCCA’s rejection of his stand-alone challenge to the alleged prosecutorial misconduct on direct appeal. (Appellant’s Br. at 73); Williams, 188 P.3d at 231. The OCCA dedicated twenty paragraphs of its opinion to this challenge. Williams, 188 P.3d at 228-230. Sometimes the OCCA specified whether trial counsel objected below; other times it did not. Sometimes it quoted the alleged misconduct, other times it paraphrased Williams’s challenge. Sometimes it addressed Williams’s arguments by referring to other sections of its opinion. In no case did the OCCA conclude that the trial court had committed plain error. We have no way of knowing what previously alleged misconduct Williams now intends to incorporate in his Strickland challenge.

For these reasons, we lack the information to address this challenge in any meaningful fashion. Even if we were inclined to sift through the OCCA’s decision and the trial transcript, we feel ill-prepared to guess which comments might form the basis for this Strickland challenge and why.⁷ For these reasons, we must decline to address Williams’s argument. In doing so, we note that “[e]ven a capital defendant can waive an argument by inadequately briefing an issue.” Grant, 727 F.3d at 1025; Bronson v. Swensen, 500 F.3d 1099, 1104 (10th Cir.2007) (“[W]e routinely have declined to consider arguments that are not raised, or are inadequately presented, in an appellant’s opening brief.”).

12. Conclusion

In sum, Williams has failed to show that he is entitled to relief with respect to any of the challenges above. In only two instances have we assumed that Williams’s legal representation was anything less than objectively reasonable. Again, we only assumed deficient performance in trial counsel’s failure to object to certain testimony of Dyra Malone and Officer Felton. In those instances, we have already said that the OCCA reasonably could have concluded that Williams failed to make the requisite showing of prejudice. We now go a step further and say that those two modest errors are insufficient to warrant relief under Strickland, even considering their combined prejudicial effect. See Hooks, 689 F.3d at 1187-88 (recognizing that resolving each allegation of ineffective assistance on prejudice grounds is not “sufficient to dispose of the claim because a further analysis of ‘cumulative prejudice’ ” is necessary); Cargle v. Mullin, 317 F.3d 1196, 1212 (10th Cir.2003) (“[A] decision to grant relief on ineffective assistance grounds is a function of the prejudice flowing from all of counsel’s deficient performance....”) (emphasis added).

D. Ineffective Assistance of Counsel (Penalty Phase)

Williams also argues that his lawyers should have presented more and better mitigating evidence during the penalty phase of trial. During this phase, his lawyers called social historian and developmental specialist, Dr. Wanda Draper, to testify about Williams’s background and social/cognitive development. Williams’s mother, Joni Williams, also testified. She offered similar background testimony and asked the jury to spare her son’s life. Still, in Williams’s view, his lawyers “wholly failed to satisfy their constitutionally required duty to thoroughly investigate and present mitigating evidence.” (Appellant’s Br. at 41-43). He claims that his lawyers should have engaged the services of a dedicated “mitigation specialist” and that they should have called more fact witnesses (including family members, a childhood pastor, teachers, and friends). He also claims that his lawyers were “effectively absent” once the trial court appointed Dr. Draper. (Appellant’s Br. at 42). In Williams’s view, trial counsel should have better prepared and assisted Dr. Draper. At the very least, Williams believes he is entitled to develop these arguments at an evidentiary hearing, which the OCCA and the district court both denied.

Strickland governs penalty-phase claims of ineffective assistance. See Wiggins v. Smith, 539 U.S. 510, 521, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003).⁸ Again, to prevail, Williams must show both that (1) counsel “committed serious errors in light of prevailing professional norms such that his legal representation fell below an objective standard of reasonableness,” and (2) there is “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Grant, 727 F.3d at 1017 (citing Strickland, 466 U.S. at 694, 104 S.Ct. 2052) (internal quotation marks omitted). “A reasonable probability is a probability sufficient to undermine confidence in the outcome.” See Strickland, 466 U.S. at 694, 104 S.Ct. 2052.

The first issue we encounter in this claim is whether Williams properly exhausted it in state court. We cannot grant habeas relief “unless it appears that the applicant has exhausted the remedies available in the courts of the State.” Gray v. Netherlands 518 U.S. 152, 161, 116 S.Ct. 2074, 135 L.Ed.2d 457 (1996); 28 U.S.C. § 2254(b). Before the federal district court, the state argued that Williams’s habeas claim differed from the one he presented to the OCCA. (Appellee’s Br. at Idle.) For instance, in his state post-conviction application, Williams named nine witnesses his lawyers either failed to contact or call as witnesses. The state pointed out, however, that Williams named a dozen different witnesses in his federal habeas petition.⁹ Additionally, according to the state, Williams argued for the first time in federal court that his lawyers failed to devote adequate time to Dr. Draper. He also presented a new affidavit from Dr. Draper that he never presented to the OCCA. In the state’s view, these differences changed the substance of Williams’s claim. The district court agreed and found the claim partially unexhausted — at least “[i]nsofar as Williams’s [ineffective-assistance claim] complains of counsels’ dealing with Dr. Draper.” Williams, 2012 WL 5197674, at *11-12.

We agree with the district court’s conclusion. A party exhausts a claim in state court when it has been “fairly presented.” Picard v. Connor, 404 U.S. 270, 275, 92 S.Ct. 509, 30 L.Ed.2d 438 (1971). “Fair presentation,” in turn, requires that the petitioner raise in state court the “substance” of his federal claims. Id. at 278, 92 S.Ct. 509. This includes not only the constitutional guarantee at issue, but also the underlying facts that entitle a petitioner to relief. Gray, 518 U.S. at 163, 116 S.Ct. 2074; see Fairchild v. Workman, 579 F.3d 1134, 1149 (10th Cir.2009) (“A claim is more than a mere theory on which a court could grant relief; a claim must have a factual basis, and an adjudication of that claim requires an evaluation of that factual basis.”) (citation omitted). Here, Williams did not present to the OCCA the factual basis for his “Draper claim.” Williams did not allege in his state post-conviction application that his lawyers failed to prepare or assist Dr. Draper, or that her testimony might have improved with more consultation. Instead, he argued that his lawyers should have called more witnesses and hired a mitigation specialist — i.e., someone else to investigate Williams’s background and presumably testify during the penalty phase.¹⁰ Williams argued that “[h]ad trial counsel fully investigated [his] family and social history, they would have discovered numerous sources of information from individuals that have known [him] for his entire life.” (See First Application for State Post-Conviction Relief, at 19). To that end, Williams supplied the OCCA with affidavits from his relatives (mostly uncles) stating what they would have testified to had they been called as witnesses. Williams’s focus in his state post-conviction relief application was thus on his lawyers’ failure to discover and call additional witnesses, not on any failure to adequately support a witness who did testify.

In federal court, however, we see different allegations, most of which Williams supports by referring to Dr. Draper’s newly submitted affidavit. See R. vol. I at 116 (“Counsel never conducted, or facilitated the conduct of, any further follow-up or individual meetings with any of these persons.”); id. at 118 (“No preparation occurred for Dr. Draper’s testimony with the trial team.”); id. at 119 (“Dr. Draper received no communication from ... any member of the Williams trial team about meeting before her testimony or to prepare for the presentation of that evidence.”); id. at 120 (“[Williams’s lawyers] wholly failed to prepare Dr. Draper for her trial testimony.”); id. at 123 (“Had counsel ensured that an adequate investigation of [Developmental Disorganized Detachment Disorder] and adequate preparation for Dr. Draper’s testimony had taken place, Dr. Draper would have given much more extensive testimony on this condition” and “Because of counsel’s failure to facilitate additional investigation or prepare Dr. Draper for trial testimony, no member of Mr. Williams’s defense team investigated the potential mental-health issues which may have [underlay] Mr. Williams’s suicide attempt.”); id. at 124 (“[A]s a consequence of counsel’s failure to ... adequately prepare Dr. Draper to testify concerning the issue of Mr. Williams’s future dangerousness, the prosecution effectively cross examined Dr. Draper” and “As a consequence of counsel’s failure in assisting Dr. Draper ... th§ demonstrative aid typically used by Dr. Draper — the ‘life path’ — was wholly inadequate.”).

Williams runs into another problem with Dr. Draper’s new affidavit. Even if Williams had exhausted these claims in state court, § 2254 would restrict our (and the district court’s) discretion to consider that affidavit. “[Section] 2254(e)(2) still restricts the discretion of federal habeas courts to consider new evidence when deciding claims that were not adjudicated on the merits in state court.” Cullen v. Pinholster, — U.S. -, 131 S.Ct. 1388, 1401, 179 L.Ed.2d 557 (2011). This is because the state trial on the merits should be the “main event,” rather than a “tryout on the road” for what will later be the determinative federal habeas proceeding. Wainwright v. Sykes, 433 U.S. 72, 90, 97 S.Ct. 2497, 53 L.Ed.2d 594 (1977).

In sum, Williams alleges that his lawyers failed to assist Dr. Draper in a number of ways, and that her expert mitigation testimony therefore suffered. Even though these allegations would typically come within Strickland’s ambit, they raise an entirely new complaint of deficient performance and resulting prejudice because Williams raises them now for the first time. See Gray, 518 U.S. at 163, 116 S.Ct. 2074 (“[I]t is not enough to make a general appeal to a [broad] constitutional guarantee ... to present the ‘substance’ of such a claim to a state court.”); Hawkins v. Mullin, 291 F.3d 658, 669 (10th Cir.2002) (“The fact that [the petitioner] asserted some ineffective-assistance claims in state court ... will not suffice to exhaust this significantly different federal habeas claim.”).

Like the district court, our exhaustion determination is limited to the “Draper claim” alone. Even though Williams names many new witnesses in his habeas petition, he does not contend that these witnesses would have said anything more than the relatives he named in his first post-conviction application. Accordingly, the “ultimate question” before the OCCA with respect to this claim was the same as the one before us — is there a reasonable probability of a different outcome but for counsels’ alleged failure to call additional witnesses? See Picard, 404 U.S. at 277-78, 92 S.Ct. 509 (“there are instances in which ‘the ultimate question for disposition’ ... will be the same despite variations in the legal theory or factual allegations urged in its support.”) (internal citation omitted).

In seeking to persuade us that he exhausted his “Draper claim,” Williams argues that his presentation of new evidence (Dr. Draper’s affidavit) did not change the substance of his claim — it merely supplemented it. But if the Draper claim is one he already presented to the OCCA — one that the OCCA presumably resolved on the merits — then the claim in this court must be based on the record Williams presented to the OCCA. See Cullen, 131 S.Ct. at 1398 (stating that “review under § 2254(d)(1) is limited to the record that was before the state court that adjudicated the claim on the merits”). As a result, we cannot consider the Draper affidavit, and his Draper claim necessarily fails.

Next, we consider what to do with this partially unexhausted claim. The district court concluded that remand was not appropriate because “Oklahoma would bar consideration of this precise claim on an independent and adequate state law procedural ground if Williams presented it in a third post-conviction application.” Williams, 2012 WL 5197674, at *12. After all, were Williams to return to state court at this point, he would raise this failure-to-prepare claim in what would now be a third application for post-conviction relief. Oklahoma requires a post-conviction relief applicant to raise all grounds for relief, which he actually knows or should have known through the exercise of due diligence in his original application for relief. See Cummings v. Sirmons, 506 F.3d 1211, 1222-23 (10th Cir.2007) (summarizing Oklahoma’s rule for bringing claims in post-conviction relief applications); see also Okla. Stat. tit. 22 §§ 1086, 1089(D)(2), (8)-(9).

There would be little question that Oklahoma’s rule against successive petitions would bar us from considering Williams’s habeas claim if the OCCA had applied the rule itself. The doctrine of procedural default prevents a federal court from reviewing “the merits of a claim— including constitutional claims — that a state court declined to hear because the prisoner failed to abide by a state procedural rule.” Martinez v. Ryan, — U.S. -, 132 S.Ct. 1309, 1316, 182 L.Ed.2d 272 (2012). Habeas review is improper under those circumstances because the state procedural rule, so long as it is “firmly established and consistently followed,” is a “nonfederal ground adequate to support the judgment.” Id.

We have previously held that the OCCA’s ban on successive post-conviction applications is just such a firmly established and consistently followed rule. See Thacker v. Workman, 678 F.3d 820, 835-36 (10th Cir.2012). What’s more, Williams does not have a good excuse for not including the failure-to-prepare arguments in his first post-conviction application. Martinez, 132 S.Ct. at 1316 (recognizing that “[a] prisoner may-obtain federal review of a defaulted claim by showing cause for the default and prejudice from a violation of federal law.”). While Williams attempts to blame his lawyers, he only argues that his appellate lawyer was ineffective for failing to raise his unexhausted claim on direct appeal. To be sure, this might excuse Williams’s noncompliance with Oklahoma’s separate rule requiring presentment of claims on direct appeal. See Berget v. State, 907 P.2d 1078, 1081 (Okla.Crim.App. 1995) (stating that “issues which were not raised on direct appeal, but could have been, are waived,” but considering “claims which, for whatever reason, could not have been raised on direct appeal.”). But this rule is not at issue here; rather, we are only concerned with the rule against successive post-conviction applications. See Okla. Stat. tit. 22 § 1086. Williams’s argument does nothing to show why we should excuse his noncompliance with the successive applications rule.

Of course, the OCCA did not bar Williams’s unexhausted failure-to-prepare claim because Williams never presented it. The procedural default described above is thus distinguishable from the “anticipatory procedural default” at issue here — where a petitioner fails to exhaust a claim and we, as a federal court, nonetheless conclude that the claim would be procedurally defaulted on remand. See Thacker, 678 F.3d at 839-41. According to the Supreme Court, a habeas petition is procedurally defaulted if the petitioner “failed to exhaust state remedies and the court to which the petitioner would be required to present his claims in order to meet the exhaustion requirement would now find the claims procedurally barred.” Coleman v. Thompson, 501 U.S. 722, 735 n. 1, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991); Watson v. State of N.M., 45 F.3d 385, 386 n. 1 (10th Cir.1995). This question is arguably more complicated than merely deferring to the state’s actual invocation of a procedural bar because we must predict what the state court would do. This case asks us to consider how certain we must be in our prediction.

In Williams’s view, there can be no room for uncertainty. He believes uncertainty exists in his case because Oklahoma sometimes forgives noncompliance with the bar on successive post-conviction applications. He points us to Valdez v. State, where the Oklahoma Court of Criminal Appeals not only considered, but also granted, a second application for post-conviction relief. 46 P.3d 703, 711 (Okla.Crim.App.2002). The OCCA made clear that it retains the power to grant a successive post-conviction application “when an error complained of has resulted in a miscarriage of justice, or constitutes a substantial violation of a constitutional or statutory right.” Id. at 710. Williams cites several other cases since Valdez in which the OCCA has similarly declined to apply this procedural bar. Given this case law, Williams suggests that we cannot predict whether the OCCA would apply a procedural bar to his claim. But inconsistently, Williams makes a prediction of his own: the OCCA will not apply a procedural bar to his unexhausted claim because it is “particularly strong,” especially when “supplemented by the facts in Dr. Draper’s affidavit.” (Appellant’s Reply Br. at 13).

We cannot agree with Williams that his ineffective-assistance claim is at all like those claims the Oklahoma courts have considered in other noncomplying successive petitions. Williams raises a run-of-the-mill Strickland claim that is far different from the “special case” the OCCA recognized in Valdez. See 46 P.3d at 711 n. 25. Valdez was special because the lawyers there knew that their client was a citizen of Mexico and nonetheless failed to comply with the Vienna Convention when they failed to contact the Mexican Consulate, thereby depriving the Consulate the ability to intervene and present its discovery that the defendant suffered from organic brain damage. Id. at 706, 709-10.

The other Oklahoma cases Williams cites involve equally compelling circumstances not present here. See, e.g., Mali-coat v. State, 137 P.3d 1234, 1235 (Okla. Crim.App.2006) (declining to apply a procedural bar to consider whether Oklahoma’s lethal injection protocol violated the Eighth Amendment’s prohibition against cruel and unusual punishment). As we have said before, albeit in the context of the OCCA’s own invocation of the successive-petition ban, the Valdez exception only applies in cases involving an “exceptional circumstance,” Black v. Workman, 682 F.3d 880, 917 (10th Cir.2012), and it is “insufficient to overcome Oklahoma’s regular and consistent application of its procedural-bar rule in the vast majority of cases,” Thacker, 678 F.3d at 835-36 (internal quotation marks omitted).

Considering Williams’s claim, the chances that the OCCA might excuse his noncompliance with the ban on successive-petitions are slim to none. True, we are not the state court and we can never predict with 100% accuracy how another court will resolve an unexhausted claim under its own procedural rules. But we do not believe this level of certainty is what the Supreme Court requires. Instead, we think it is enough if, looking to the state’s treatment of its procedural bar, the likelihood of default in the petitioner’s case is beyond debate or dispute. See Cummings, 506 F.3d at 1223.

Indeed, this was enough in Cummings, where a habeas petitioner also failed to exhaust a Strickland claim involving his lawyer’s failure to seek DNA testing of “critical evidence.” Id. at 1222. Despite the petitioner’s claim that he should not be subject the same procedural bar because he was innocent (presumably, an argument he would have made on remand in state court), we “readily concluded” that his claim was subject to an anticipatory procedural bar. Id. at 1223. It was “beyond dispute” that, “were [the petitioner] to attempt to now present the claim to the Oklahoma state courts in a second application for post-conviction relief, it would be deemed procedurally barred.” Id. at 1223. Nothing Williams says leads us to believe that the likelihood of a procedural bar in his case is any less certain.

Williams raises two other arguments in resisting an anticipatory procedural bar for his failure-to-prepare claim. First, as proof of the OCCA’s unpredictability and the need for remand, he points to the fact that in this very case the OCCA excused a procedural bar to consider other ineffective-assistance claims that he did not raise on direct appeal. See Williams, No. PCD 2006-1012, at *3-8. Williams raised these claims in his state post-conviction relief application contrary to Oklahoma’s general rule requiring that defendants present all claims on direct appeal. See Okla. Stat. tit. 22, § 1089(C). But just because the OCCA declined to apply one procedural rule — the presentment of claims on direct appeal — in one context does not mean that it would decline to apply an entirely different procedural rule — requiring a petitioner to raise all known (or should have known) claims in a single, initial application for post-conviction relief — in another.¹¹

Second, Williams argues that the procedural rule at issue in his case is not an adequate and independent state ground to bar his failure-to-prepare claim. Specifically, Williams says that the general prohibition against successive post-conviction applications intertwines with federal law, which would not preclude habeas review. He seems to say that, under the Valdez exception, a state court is required to consider the merits of a constitutional claim— thereby raising questions of federal law and undermining the very reason we defer to state procedural dismissals. But we have already rejected this argument. See Banks v. Workman, 692 F.3d 1133, 1145 (10th Cir.2012) (concluding that Oklahoma’s procedural bar is independent of federal law, notwithstanding the OCCA’s power to excuse default in “extreme cases”). And to whatever extent the Valdez exception raises questions of federal law, we have considered those federal questions in determining for ourselves that Valdez does not extend to Williams’s ineffective-assistance claim.

This brings us to the merits. The OCCA rejected the exhausted portion of Williams’s ineffective-assistance claim under Strickland. In its view, the mitigation-specialist claim failed because Williams had not shown what additional evidence the mitigation specialist would have unearthed. Williams, No. PCD 2006-1012, at *5. The OCCA then resolved Williams’s claim that his lawyers should have called other witnesses. It concluded that the lawyers decided not to use these witnesses (aunt, uncle and grandfather) after a reasonable investigation into mitigation evidence. Id. at *8. The OCCA further concluded that this decision “amounted to reasonable trial strategy, thus counsel was not ineffective for failing to utilize these relatives as mitigation witnesses.” Id.

As for the claim that the lawyers should have hired a mitigation specialist, we agree with the OCCA that Williams has failed to identify what other testimony (if any) a mitigation expert would have discovered. We also fail to understand why Dr. Draper did not adequately fulfill that role. Accordingly, we cannot say — as we must to reverse — that the OCCA’s decision was contrary to or an unreasonable application of federal law, or that it resulted in an unreasonable determination of the facts presented in the state courts. See 28 U.S.C. § 2254(d).

As for the claim involving uncalled witnesses, like the district court, we choose to resolve this claim under the prejudice-prong of Strickland. See Rompilla v. Beard, 545 U.S. 374, 390, 125 S.Ct. 2456, 162 L.Ed.2d 360 (2005) (“Because the state courts found the representation adequate, they never reached the issue of prejudice ... and so we examine this element of the Strickland claim de novo.”) In doing so, we assume that Williams can satisfy Strickland’s first prong (deficient performance) because his trial lawyers did not interview Williams’s family and friends. See Cole v. Trammell, 755 F.3d 1142, 1161 (10th Cir.2014) (assuming the same “because family and social history is one of the crucial areas of investigation emphasized in the ABA Guidelines”) (internal quotation marks omitted). Still, our assumption does not bring Williams the relief he desires; in the end, he cannot show that his lawyers’ failures resulted in prejudice under Strickland.

To show prejudice, Williams must show that his lawyers’ failures mattered — “namely, that there is ‘a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.’ ” Grant, 727 F.3d at 1018 (citing Strickland, 466 U.S. at 694, 104 S.Ct. 2052). When a petitioner alleges ineffective assistance of counsel stemming from a failure to investigate mitigating evidence at a capital-sentencing proceeding, “we evaluate the totality of the evidence — both that adduced at trial, and the evidence adduced in habeas proceedings.” Smith v. Mullin, 379 F.3d 919, 942 (10th Cir.2004) (quoting Wiggins, 539 U.S. at 536, 123 S.Ct. 2527). This includes weighing “the evidence in aggravation against the totality of available mitigating evidence.” Hooks, 689 F.3d at 1202. “In a system like Oklahoma’s, where only a unanimous jury may impose the death penalty, the question is whether it’s ‘reasonably probabl[e] that at least one juror would have struck a different balance.’ ” Grant, 727 F.3d at 1018-19 (quoting Wiggins, 539 U.S. at 537, 123 S.Ct. 2527). Here, the OCCA did not reach this prejudice question, but we are sure that there was no substantial probability, let alone a conceivable one, that one juror (or more) would have voted against the death penalty had counsel not (assumed to have) failed to call other witnesses. See Harrington, 131 S.Ct. at 792 (defining reasonable probability as the likelihood of a different result being “substantial, not just conceivable”).¹²

We reach this conclusion based on what Williams tells us the uncalled witnesses would have said. In the prejudice section of his brief, Williams claims that the uncalled witnesses would have testified about his “criminogenic upbringing,” that he “lack[ed] ... any parent figure for the majority of his life,” and “that he could potentially thrive and rehabilitate himself in prison.” (Appellant’s Br. at 46). He gives no specifics about what each witness might have said or why this might have changed the outcome at sentencing. Admittedly, in his habeas petition, Williams says the witnesses would have testified to slightly more — that he was born premature, that he was on a heart monitor during infancy, that the Department of Human Services came to check on him as a baby, that he got lost in his grandmother’s home, that he didn’t have a room of his own, that he witnessed police encounters involving his uncles, that he was greatly affected by his grandmother’s death, that his uncles were involved in gangs, and that he was exposed to violence and criminal activity from a young age. If we consider the affidavits Williams presented in state court, the uncles also may have testified that they would sometimes beat up on Williams. Some would also have testified, based on personal experience, that prison was difficult and that a sentence of life in prison would not let Williams “off the hook.”

There is one key reason why this excluded testimony falls short of that necessary to show prejudice under Strickland — the sentencing jury was already “well acquainted” with evidence of Williams’s “background and potential humanizing features.” Wong v. Belmontes, 558 U.S. 15, 23, 130 S.Ct. 383, 175 L.Ed.2d 328 (2009); see Wackerly, 580 F.3d at 1182 (finding no prejudice when petitioner argued that his lawyer should have introduced evidence that was cumulative of evidence the jury did hear). As the OCCA outlined, both Dr. Draper and Joni Williams testified at length about Williams’s upbringing. They told the jury about the circumstances of Williams’s conception and premature birth. They told the jury how Williams had no contact with his biological father and how Joni left Williams with her mother to raise him. They told the jury that Williams got lost in his grandmother’s home and that Williams was even more lost following his grandmother’s death by heart attack (which Williams witnessed). They told the jury that Williams’s uncles were a negative influence and that they exposed him to violence and drug abuse from a young age. Dr. Draper even told the jury that the uncles would beat up on Williams. Williams gives us no reason to think that other family and friends would have added to this humanizing evidence in any significant way. Even if other witnesses might have offered additional details or unique perspectives, we fail to see how the exclusion of this largely cumulative evidence might undermine confidence in the outcome. “Many of our cases have also refused to find prejudice when the evidence the defendant says counsel should have presented would have been cumulative of the evidence the jury actually heard.” See Grant, 727 F.3d at 1022.

The only non-cumulative information the uncles might have presented is their firsthand accounts of prison and its effects. Had the lawyers presented this testimony, Williams contends there is a reasonable probability that at least one juror might have voted differently. This argument is not without merit; undoubtedly, the jurors considered whether life imprisonment was sufficient punishment for Williams and, under the “continuing threat” aggravating circumstance, whether he might remain a violent person even in prison. But even if Williams’s uncles had told the jury that prison was horrible and that it has the potential to reform, we cannot overlook the “double-edged nature” of this evidence. Wackerly, 580 F.3d at 1178. For example, if the uncles had testified, then the state would have pointed out to the jury that most of Williams’s closest family members are convicted felons. Also likely, the prosecution would have elicited testimony about the criminal activities that landed the uncles in jail — as well as any criminal activities since. Further, the jury would have seen that much of Williams’s support system comprised of individuals who (at least at one point in life) were tied to gangs and organized violence. This evidence might have made a difference, “but in the wrong direction” for Williams. Wong, 558 U.S. at 22,180 S.Ct. 383.

By pointing this out, we do not mean to say that convicted felons cannot give powerful mitigation testimony. For example, in Harlow v. Murphy, on which Williams relies, the United States District Court for the District of Wyoming found ineffective assistance and resulting prejudice based on counsel’s failure to present testimony from a defendant’s fellow inmates. R. vol. I at 406-26 (citing No. 05-ev-039-B, at *38-58 (D. Wyo. Feb. 15, 2008) (unpublished)). The petitioner in that case murdered a prison guard during an attempted escape. The other inmates whom trial counsel did not call to testify would have refuted specific evidence that the defendant was violent during his time in prison. Id.

Here, however, Williams only says that his uncles would have told the jury that prison has both the power to punish and to reform. These rather obvious points would not have provided much benefit, if any, to Williams, particularly when paired with the more damaging information the jury would have learned had the uncles testified. Thus, in discussing the uncles’ testimony, we simply mean to point out that we cannot just consider how the uncles would have responded to friendly questions from the defense. We must also consider what rebuttal evidence the uncles would have “put into play” if the uncles actually testified. Wong, 558 U.S. at 22, 130 S.Ct. 383.

In his habeas petition, Williams also argues that several of his relatives would also have asked the jury to spare his life. The jury heard Joni Williams’s plea, and we do not doubt that there are others who love the defendant and would similarly implore the jury not to impose the death penalty. The fact remains that Williams has failed to explain why pleas from additional witnesses would have made a difference in the jury’s consideration of the aggravating and mitigating circumstances in his case, or in its decision to impose the ultimate sentence of death.

In sum, as sad and difficult as a defendant’s circumstances may be, “evidence of childhood abuse or neglect isn’t always severe enough to earn a jury’s sympathies.” Grant, 727 F.3d at 1021. The mitigation evidence presented here was obviously not enough to earn the jury’s sympathies. Even if trial counsel had hired a mitigation expert or called additional witnesses to testify, Williams cannot point to any other evidence that his lawyers should have presented that might have moved even a single juror’s sympathies his way. See Grant, 727 F.3d at 1018-19.

Williams would nonetheless like us to grant him an evidentiary hearing on this claim to consider new evidence that was not before the OCCA. He points out that claims of this variety usually require a hearing. Williams dedicates much of his brief to arguing why he is entitled to a hearing and why § 2254(e)(2), which only allows hearings under compelling circumstances, does not apply.

Under the provisions of § 2254(e)(2), a habeas petitioner cannot receive an evidentiary hearing whenever an “applicant has failed to develop the factual basis of a claim in State court proceedings.” Under those circumstances, a petitioner can only receive a hearing when “the facts underlying the claim would be sufficient to establish by clear and convincing evidence that but for constitutional error, no reasonable factfinder would have found the applicant guilty of the underlying offense.” 28 U.S.C. § 2254(e)(2). Williams argues that we should not hold him to such a high burden here because he tried to develop the factual basis for his claim in state court but the OCCA unreasonably refused- to hold a hearing.

But even if we agree with Williams that § -2254(e)(2) does not apply and afford him the benefit of our less-rigorous pre-AEDPA standard, we see no reason for a hearing in this case. See Cannon v. Mullin, 383 F.3d 1152, 1175 (10th Cir.2004). When § 2254(e)(2) is inapplicable, a habeas petitioner still is not entitled to an evidentiary hearing unless “(1) the facts were not adequately developed in the state court, so long as that failure [was] not attributable to the petitioner, and (2) his allegations, if true and not contravened by the existing factual record, would entitle him to habeas relief.” Id. (internal citation and quotation marks omitted).

We. agree that, under the less-rigorous pre-AEDPA standard, a hearing is usually required when a defendant alleges his lawyer failed to investigate his background. Without a hearing, it is often difficult to “reliably determine whether counsel’s investigation was deficient without knowing what was investigated, and the scope of the investigation can rarely be discerned from the trial record.” Fairchild, 579 F.3d at 1142. Even if we agree with Williams that a hearing would have shed light on his lawyers’ performance in this case, we fail to see how a hearing would give us any insight into the question of prejudice — our basis for resolving this claim. Even affording Williams the benefit of assuming that counsel performed deficiently and of the pre-AEDPA standard for determining whether to grant an evidentiary hearing, we cannot conclude that a hearing is-necessary. While Williams has alleged facts that, if true, would show deficient performance, this is already something we have assumed. However, Williams has not alleged facts that, if true, would demonstrate prejudice under Strickland. See Cannon, 383 F.3d at 1175. An evidentiary hearing is for determining if allegations have factual support; it is not for uncovering new allegations of which the habeas petitioner is currently unaware. See Medina v. Barnes, 71 F.3d 363, 366 (10th Cir.1995) (determining whether petitioner was entitled to an evidentiary hearing “to resolve any disputed facts underlying his claims”).

E. Cumulative Error

This leaves us with Williams’s claim of cumulative error. He argues that while each identified error may be insufficient to warrant habeas relief in isolation, “when viewed in their entirety there can be no doubt that [he] has not received effective assistance as required by the Sixth Amendment.” (Appellant’s Br. at 90).

“In the federal habeas context, a cumulative-error analysis aggregates all constitutional errors found to be harmless and analyzes whether their cumulative effect on the outcome of the trial is such that collectively they can no longer be determined to be harmless.” Cole, 755 F.3d at 1177.

Here, Williams’s “perfunctory assertion falls well short of what’s needed to overturn a judgment, let alone one as long-settled and repeatedly reviewed as this one.” Grant, 727 F.3d at 1025. Even so, we have already concluded that the cumulative effect of all presumed errors of counsel was insufficient to create a reasonable probability that, but for those errors, the jury would not have convicted Williams of first-degree murder or imposed the death penalty. Similarly, Williams has not persuaded us that the combined errors of counsel led to Williams receiving a trial that was not “fundamentally fair.” See Cole, 755 F.3d at 1177 (framing cumulative-error question in this fashion).

III. CONCLUSION

We affirm the district court’s judgment. We deny Williams’s motion to expand the certificate of appealability.

1 . The opposite, however, is certainly not true. A lawyer’s failure to object to error (even plain error) does not amount to ineffective assistance per se. Gordon v. United States, 518 F.3d 1291, 1300 (11th Cir.2008).

2 . In this context, both inquiries focus on the harm the contested evidence causes when it is stacked up against the other (uncontested) evidence of guilt. It follows that when a substantial-rights standard requires no more than Strickland, as is true of the federal plain-error standard, the standards are "virtually identical.” Close v. United States, 679 F.3d 714, 720-21 (8th Cir.2012) cert. denied, - U.S. -, 133 S.Ct. 464, 184 L.Ed.2d 286 (2012).

3 . Williams also argues in passing that his case involves "circumstances where competent counsel very likely could not render effective assistance." Appellant’s Br. at 83. Williams thus attempts to invoke the third Cronic exception, which applies when "counsel is called upon to render assistance under circumstances where competent counsel very likely could not." Bell, 535 U.S. at 696, 122 S.Ct. 1843. The example the Supreme Court gave in Cronic is Powell v. Alabama, 287 U.S. 45, 58, 53 S.Ct. 55, 77 L.Ed. 158 (1932), where counsel was only “appointed” when two lawyers offered to "assist” on the morning of trial. Williams does not point to any analogous institutional problems here. And to the extent he argues that his lawyer’s alleged drug abuse should qualify, this novel position is not "clearly established law” within the meaning of § 2254(d)(1). See Fairchild v. Workman, 579 F.3d 1134, 1139 (10th Cir. 2009) (law is not "clearly established” when it requires extraction of "general legal principles developed in factually distinct contexts.”).

4 . In his habeas petition and brief before us, Williams generally complains of all other-crimes evidence admitted without objection, but he only identifies the watch. Like the district court, we will not speculate what other evidence (if any) Williams might mean to include in this Strickland challenge.

5 . In his brief before us, Williams argues his lawyer should have objected to the testimony of the doctor who referenced the photographs as well. But Williams did not raise this argument in his habeas petition, so we do not address it here. See United States v. Windrix, 405 F.3d 1146, 1156 (10th Cir.2005) (declining to address a claim petitioner did not raise before district court).

6 . For the first time on appeal, Williams argues that Malone’s testimony on redirect prejudiced him because it somehow enabled the prosecution to introduce evidence of the May robbery. Williams did not argue this to the district court, so we do not consider it here. See Windrix, 405 F.3d at 1156. Regardless, Malone's redirect testimony did not concern the May robbery of First Fidelity.

7 . The district court did not engage in this guesswork either. It simply concluded in general terms that Williams had not demon-stated constitutionally ineffective assistance. Williams, 2012 WL 5197674, at *16, *19.

8 . Williams does not argue that Cronic is the applicable standard here, nor does he raise lead counsel’s alleged substance abuse.

9 . By our count, Williams actually supplemented his habeas petition with thirteen new witnesses.

10 . In his state post-conviction relief application, Williams also argued that his lawyers received inadequate compensation and that trial counsel was ineffective for failing to rehabilitate jurors who expressed doubts about imposing the death penalty, but he does not renew those claims in his federal habeas petition.

11 . This is all the more true here given our prior announcement that we will only consider the OCCA’s rule requiring presentment of ineffective-assistance claims on direct appeal as an adequate ground for procedural default under certain circumstances. See English v. Cody, 146 F.3d 1257, 1259 (10th Cir.1998). What’s more, Williams argued that his appellate lawyer was ineffective in failing to raise this claim on direct appeal, thereby giving the OCCA another reason to reach the merits. Neither of these concerns would weigh in favor of excusing Williams's entirely unrelated failure to raise his Draper-centric arguments in his first (or second) application for post-conviction relief.

12 . In his brief, Williams alludes to a claim that his lawyers failed to prepare Joni Williams. The parties do not focus on this claim — likely because Williams failed to develop it properly in his briefing. To the extent Williams raises such a claim, however, it fails under Strickland's prejudice prong as well. Williams does not even hint at what more his mother might have said if trial counsel better prepared her.