Kenneth T. RICHEY, Petitioner-Appellant, v. Betty MITCHELL, Warden, Respondent-Appellee, 395 F.3d 660

COLE, J., delivered the opinion of the court, in which DAUGHTREY, J., joined. SILER, J., (pp. 688-92), delivered a separate dissenting opinion. OPINION COLE, Circuit Judge. Kenneth T. Richey was arrested, convicted, and sentenced to be executed for intentionally starting an apartment fire that killed two-year-old Cynthia Collins. He continues to maintain his innocence. Following a series of unsuccessful appeals in the Ohio state courts, Richey petitioned the district court for a writ of habeas corpus, and the court denied his request. Richey now appeals this decision.

Because constitutional errors have undermined our confidence in the reliability of Richey’s conviction and sentence, we REVERSE the decision below. I. BACKGROUND On June 30, 1986, at approximately 4:15 a.m., a fire started in the second-floor apartment of Hope Collins, who was elsewhere at the time. Her two-year-old daughter, Cynthia, was alone in the apartment, and she died in the fire. The Fire Chief initially blamed the fire on an electric fan, but then asked Assistant State Fire Marshal Robert Cryer to investigate further. Cryer arrived at the apartment at 6:30 a.m. and spent most of the day investigating.

The next day, Cryer told the prosecutor’s office that he believed that the fire had resulted from arson. Cryer based his conclusion solely on his belief that some of the burn patterns he found at the apartment demonstrated the presence of accelerants. He found no empty containers of flammable liquids. From the apartment, Cryer took nothing that the Ohio Arson Crime Laboratory (“State Arson Lab”) could test for the presence of accelerants, nor did he order the scene secured at the end of his June 30 investigation. Instead, Cryer authorized the building’s owners to clean the apartment.

The owners discarded the damaged living room carpet, which ended up at a local garbage dump. The police investigation quickly focused on Richey. On the morning of June 30, within hours of the fire, he was interviewed by the police chief. On July 1, Richey was arrested for arson and gave a tape-recorded interview to the police, who were joined by Cryer and a prosecutor. Although acknowledging that he was intoxicated and therefore did not remember much of what happened early in the day on June 30, Richey denied starting the fire.

Because gasoline and paint thinner were stored in an unlocked greenhouse across the street from the apartment building, the State theorized that Richey had obtained these materials from that greenhouse. However, the owner of the greenhouse was unable to determine whether any gasoline or paint thinner was missing, and Cryer took none of the accelerants from the greenhouse to compare with the materials at the fire scene.

On July 10, 1986, a grand jury charged Richey with: (1) aggravated felony murder; (2) aggravated arson; (3) breaking and entering; (4) involuntary manslaughter during the commission of a felony; and (5) child endangerment. The court appointed two attorneys to defend Richey against the charges, to which he pled not guilty, not guilty by reason of insanity, and not competent to stand trial by reason of insanity. The trial court ordered Richey to undergo a psychiatric examination, and he eventually withdrew his insanity plea and was adjudged competent to stand trial.

In late 1986, Richey waived his right to a trial by jury, and agreed to be tried before a three-judge panel, under Ohio Revised Code

Section 2945.06. The heart of the indictment against Richey was the charge of aggravated felony murder, which, if proven, made him eligible for the death penalty. According to the statute in effect at the time, an aggravated felony murderer must have “specifically intended to cause the death of the person killed[ — ]the prosecution must prove the specific intent of the person to have caused the death by proof beyond a reasonable doubt.” Ohio Rev. Code Ann. § 2903.01(D). The State concedes that it presented no evidence that Richey specifically intended to kill Cynthia Collins.

Rather, the State hypothesized that Richey set fire to the Collins apartment so that he could kill his ex-lover, Candy Barchet, and her new boyfriend, Mike Nichols, who were spending the night together in the apartment below. The testimony at trial established that Barchet moved into the building on June 15, and that within a few days she and Richey progressed to a sexual relationship. Apparently, Richey frequently told Bar-chet that he loved her and would kill any other men she dated.

John Butler testified that on June 24, he had sex with Barchet, and that when Richey learned of this encounter, he confronted Butler while carrying a knife. Right after the confrontation, Richey broke his hand by punching a door. On June 29, Richey attended a party hosted by Peggy Villearreal, who lived next door to Hope Collins.

Barchet brought Nichols to the party, kissed him in plain view of the other guests, told Richey that she wanted to date Nichols, and later took Nichols home with her. ■ Several witnesses testified that Richey — who was intoxicated that night — became upset upon learning of Bardlet’s new boyfriend. One witness testified that Richey proclaimed that “[i]f I can’t have [Barchet], nobody else can.” Three other witnesses testified that Richey told them that Building A, the subsection in which Barchet resided, “was going to burn” that night.

Hope Collins testified that around 2:00 a.m., as the party began to wane, Richey asked her if he could sleep on her sofa that night, but she refused. Collins testified that Richey offered to steal some flowers for her from the greenhouse located across the street, but’she declined his gesture. Shortly after 3:00 a.m., a friend of Collins drove up to the building and asked Collins to go out with him that night. Collins told him that she did not have a babysitter. According to Collins, Richey volunteered to “keep an eye” on Cynthia, as long as he could sleep on Collins’s couch.

Collins testified that at 3:30 a.m.,- with Cynthia in Richey’s care, she went out with her friend. The fire started at about 4:15 a.m. Five eyewitnesses testified that after Richey emerged onto the scene: (1) he repeatedly hollered ' that “[tjhere’s a baby in the house”; (2) he repeatedly attempted to enter the burning apartment building to save Cynthia’s life; (3) he proceeded so far into the building that “he came back out coughing and spitting up”; and (4) and police eventually had to restrain him from entering the building.

The Assistant Fire Chief stated that Richey’s efforts to save Cynthia “constituted that of a person who completely was disregarding his own safety.” Conversely, another tenant testified that upon viewing the fire, Richey boasted that “[i]t looks like I did a helluva good job, don’t it.” As part of its investigation, the State eventually retrieved six samples of debris remaining from the fire. Several of those samples came from the carpet that had first found its way into the garbage dump. On the afternoon of. July 1, nearly two days after the fire broke out, the Deputy Sheriff retrieved the carpet from the dump.

One piece of carpet was recovered from atop the garbage pile, and another was partially covered by trash. Once removed, the carpet was placed in the sheriffs parking lot. The carpet stayed in the parking lot — located no more than forty feet away from gasoline pumps — for three weeks, before it was finally taken to the State Arson Lab for testing. Similarly, a wood chip sample was not removed from Collins’s apartment for testing until July 17, nearly three weeks after the fire.

These samples were analyzed by the State Arson Lab using gas chromato-grams, which one of the State’s forensic chemists, Dan Gelfius, described at trial as “scientific instrumentation that allows the differential migration of the components of hydrocarbons to separate and to give ... a pattern similar to the identification of fingerprints.” Basing his conclusions on a method of analyzing the chromatograms that has neither been published nor peer-reviewed, Gelfius testified that both a sample of carpet from Collins’s living room and a sample of wood from her balcony contained paint thinner, and that another sample of the living room carpet contained gasoline.

The State detected no traces of accelerants on the other three samples from the apartment, the clothing and boots that Richey wore on June 29, or the bandage that covered his broken hand. Gelfius was unaware of the carpet’s removal to the local garbage dump and then to the sheriffs parking lot until he arrived to testify at trial.

He has since acknowledged that, had he known about these detours, he would have been “very concerned” and “probably work wouldn’t have been done on this case.” The missing links in the carpet’s chain of custody did not disturb Assistant State Fire Marshal Cryer, because he believed that the same paint thinner was found on both the living room carpet and the balcony.

However, this turned out to be false: the Chief of the State Arson Lab, Mohammed Gohar, and Gelfius have since acknowledged that the State Arson Lab did not — and would not have even been able to — “match” the paint thinner purportedly found on the deck to the paint thinner purportedly found on the carpet. Yet the State’s scientific evidence went unchallenged at trial. Richey’s forensic expert for the trial was L. Gregory DuBois of the firm CTL Engineering.

Richey’s trial counsel acknowledged that DuBois “was not recommended by any civil defense attorney or any other attorney.” Instead, trial counsel learned about DuBois’s firm through an advertisement that he received in the mail. Trial counsel stated that he “called CTL because they advertised [that] they had experts in arson investigation, and [DuBois] was the one [that CTL] sent up.” CTL sent DuBois, whose formal education was limited to a bachelor’s degree in metallurgical engineering and a partially completed stint in business school.

Although accredited in Ohio as an accident reconstructionist, he primarily performed “vehicle accident reconstructions.” He had no accreditations in arson or fire investigations.

Moreover, although DuBois was a member of the American Society for Testing Materials, the membership was solely through CTL; DuBois had no individual membership in the group, and he had attended no “seminars or other educational venues sponsored by the ASTM.” Similarly, although DuBois was a member of the International Association of Arson Investigators (“IAAI”), his involvement was limited to receiving quarterly publications in the mail, and at the time he was hired by Richey’s trial counsel, he had attended no IAAI seminars on fire investigation.

Finally, although Du-Bois had participated in numerous arson investigations with CTL, he did so in a supporting role: “[t]ypically, [he] would review the results [of the investigations] with the chemist who performed the analy-ses.” DuBois’s resume, which trial counsel had received, indicated that he worked as a metallurgical engineer, and that his arson-related training consisted of only two two-day courses, neither of which involved the subject of burn patterns. Both courses were taught by personnel from the State Arson Lab, whose inculpatory conclusions DuBois was hired to review.

Moreover, DuBois admitted that he admired Mohamed Gohar — Chief of the State Arson Lab, who oversaw the testing in Richey’s case- — and he believed that Gohar stayed at the forefront of technology and was “quite authoritative in his field.” At the time he was hired, DuBois believed that the state fire marshal’s office did a better job of fire sample analysis than his own employer did. Although Richey’s trial counsel had been appointed on July 14, 1986, and received the State Arson Lab reports on August 5, 1986, he did not meet with DuBois (or any other potential expert) until September 11, 1986.

At that meeting, according to Du-Bois, trial counsel told him that “he was interested in keeping costs to a minimum, .... and he told [DuBois] that he wanted about ten hours in the initial investigation.” Trial counsel did not ask DuBois what type of work was necessary to investigate the fire; according to DuBois, “the ten hours was given to me by [trial counsel] as an implied budget, that this was what he could afford in this particular case to have the time spent, and he wanted to know what could I do for ten hours worth of time.” It was not until November 18, 1986 that trial counsel contacted DuBois again, and in the interim DuBois did no further work on the case.

By this point, despite having had no idea what conclusions DuBois would draw, trial counsel had already put his name on the witness list. When his investigation finally began, DuBois contacted Gohar. The ' two “work[ed] together, went through each chromatogram, and [DuBois] had [Gohar] show [him] which standard he used and what result he got, and which sample matched which chromatogram.” DuBois conducted no independent testing of the samples, nor did he interview Cryer or otherwise attempt to analyze the growth and spread of the fire.

Trial counsel, however, “wasn’t aware that [DuBois] had not done any [independent testing] until well after the trial.” Although trial counsel never asked DuBois to perform any independent testing, he “assumed [DuBois] would do more than just look at [the] reports of Cryer.” Furthermore, when DuBois presented his findings to trial counsel, counsel did not ask him what his basis was for those findings.

DuBois, who submitted no written report to trial counsel, confirmed that counsel “was surprisingly nonargumentative with me [and] didn’t challenge me on' what I thought or why I thought what I did or anything.” Nor did DuBois recall counsel ever asking him about any problems with the' State’s evidence. Rather, admitted trial counsel, “once [DuBois] confirmed that everything Cryer had done was accurate and appeared to be in order, we decided at that point not to use him.” Reading between the lines, the State then subpoenaed DuBois to testify against Richey.

DuBois contacted trial counsel and asked for his help in resisting the subpoena, but he received none. . According to DuBois, he was told by Richey’s counsel that “I’ll [DuBois] just have to do what I have to do.” Consequently, Rich-ey’s sole forensic expert testified at Rich-ey’s trial against him, acknowledging that he found nothing wrong with the State’s conclusions. Trial counsel neither cross-examined DuBois nor presented any other scientific evidence.

When informed, following trial,' about the full extent to which the carpet had been diverted prior to arriving at the State Arson Lab, DuBois responded that “I don’t think [that counsel] told me the entire sequence of events involving the carpeting after the fire and before trial.” When asked whether this information would have affected his acquiescence to the State’s scientific conclusions, DuBois remarked that: I don’t know if I knew that at the time I did this investigation. And it raises a problem with spoilation of this evidence.

I’m having trouble right now telling you whether or not that changes my opinion or affects it. I’m looking — I looked at this fire and I looked at a lot of things together in reaching my conclusion about this fire. And if you take out a piece of this puzzle, then I would have to re-evaluate what’s left about this fire. The panel found Richey guilty on all counts (except involuntary manslaughter, which was a lesser alternative to aggravated felony murder), including aggravated felony-murder. At the death penalty sentencing hearing, Richey presented an array of evidence about his background and upbringing.

He had received his first mental evaluation at age thirteen, and had been repeatedly treated at mental institutions. A psychologist testified that Richey suffers from a borderline personality disorder and antisocial personality disorder, and that he functions at the emotional level of a ten- or eleven-year-old. A social worker testified that Richey suffers from “histrionic behavior disorder.” The panel unanimously sentenced Rich-ey to death. It concluded that “Richey purposely and with specific intent caused the death of Cynthia Collins” and disagreed that mitigation was warranted by Richey’s attempt to save her.

Instead, the panel opined that “[t]his final contention of the defendant was immediately disposed of when placed in juxtaposition to the unre-futed evidence that the defendant pulled the fire alarm from the hall prior to torching the decedent’s apartment and effectively eliminated the chance that Cynthia Collins would awaken and escape her own demise.” The “unrefuted” evidence supporting the panel’s conclusion that Richey had pulled the fire alarm — which even the State did not argue — was the testimony of Cryer, who noted that he saw the smoke detector hanging from the ceiling by its wires, eight inches below the ceiling, when he arrived at the apartment the morning after the fire. (At trial, Hope Collins testified to her belief that the smoke detector had been connected and functioning in the hours prior to the fire.) On direct appeal, Richey was represented by counsel from the Ohio Public Defender Commission.

After he reviewed the trial record, appellate counsel discussed the appealable issues with his supervisor, and expressed his view that trial counsel’s “representation of Mr. Richey had been markedly deficient and that the deficiencies had severely prejudiced Mr. Richey’s defense.” According to appellate counsel: [My supervisor], who is a personal friend of [trial counsel], listened to my views regarding the prejudicial deficiencies of [trial counsel’s] representation of Mr.

Richey, and responded by saying that I should do the best job I could without raising the ineffective assistance of trial counsel issues. [My supervisor] did not tell me that the ineffectiveness arguments I wanted to make lacked merit. He simply told me not to make them .... [he] left me with no doubt that if I were to push on the ineffective assistance of counsel issues, he would not be pleased and my job rating would suffer. This was not the first time [my supervisor] had directed me and others in our office not to raise ineffectiveness issues in a brief.

On an earlier occasion I had to sneak into the office after hours to put an ineffective assistance of counsel argument back in a direct appellate brief after [my supervisor] had taken it out. Yet appellate counsel averred that “[t]he errors made by [trial counsel] were too egregious to be completely ignored. Therefore, I raised ineffective assistance of trial counsel as an appellate issue ... but did so in a relatively weak fashion so that I would not have too severe a confrontation with [my supervisor].

Furthermore, I did not raise many of the specific instances of trial counsel’s ineffectiveness that I had noted in reviewing the record, and did not research what I regarded as other potentially meritorious ineffective assistance of counsel arguments.” The Ohio Court of Appeals affirmed Richey’s conviction and sentence. State v. Richey, 1989 WL 156561 (Ohio Ct.App. Dec.28, 1989); State v. Richey, 1989 WL 156562 (Ohio Ct.App. Dec.28, 1989). By a vote of four to three, the Ohio Supreme Court affirmed. State v.

Richey, 64 Ohio St.3d 353, 595 N.E.2d 915 (1992), rehearing denied 65 Ohio St.3d 1421, 598 N.E.2d 1172 (table). The U.S. Supreme Court denied certiorari. Richey v. Ohio, 507 U.S. 989, 113 S.Ct. 1592, 123 L.Ed.2d 157 (1993), rehearing denied, 508 U.S. 934, 113 S.Ct. 2401, 124 L.Ed.2d 303. At this point Richey retained new counsel, who uncovered new evidence and retained new scientific experts. Counsel submitted an affidavit from Villearreal, who recanted her trial testimony, claiming that “I never heard Kenny Richey say anything about burning ‘A Building’ or about using Marine Corps tactics to burn the building.

I don’t recall exactly what I said at [Richey]’s trial, but if I said anything about [Richey] threatening to burn A Building or about using Marine Corps tactics, I was wrong.” Villearreal further averred that during and after the party, Richey “was at ease, having fun and enjoying himself. He was not angry about anything[, and] expressed no anger toward Candy Barchet, Mike Nichols or anybody else.” Villearreal also contended that Cynthia “was always into everything, including matches and lighters.

Sometimes when Hope and I would be drinking coffee and smoking cigarettes, Cynthia would grab our lighters and run into her bedroom.” Villearreal detailed a number of incidents in which Cynthia’s action failed to comport with basic fire safety, including one in which Cynthia placed a lit cigarette between sofa cushions, and also averred that according to Hope, Cynthia had twice set fire to their beds, using, respectively, a curling iron and a cigarette lighter.

Vil-learreal also declared that she ate dinner at Collins’s apartment the night of the fire, and that the smoke alarm was “hanging loose.” Indeed; noted Villearreal, “Hope’s smoke detector was almost always unhooked. It was very sensitive and would go off when we were just smoking cigarettes or even when just the oven was on. So Hope used to keep it unbooked.” According to Villearreal, her knowledge of these circumstances was no mystery to either the prosecutor or Richey’s trial counsel.

Specifically, Villearreal averred that “before Richey’s trial for arson and murder, [she] spoke to [his] lawyer ... about Cynthia Collins’ playing with fire.” Villearreal expressed surprise that during Richey’s trial, “no one asked me any questions about Cynthia’s playing with fire.” As for the prosecutor, Villearreal stated that she “also spoke with the .prosecuting attorney ... before [Riehey]’s trial.

I probably also told him about Cynthia’s playing with fire but I don’t specifically recall.” Richey’s new forensic experts both castigated the State’s scientific conclusions and surmised that DuBois’ accession to them amounted to professional incompetence.

Richard Custer, a specialist in fire reconstruction, testified that — even in 1986, when the State tested the samples— Cryer’s conclusions were based on “unsound scientific principles.” First, Custer opined that the burn pattern could also have resulted from a fire that occurred naturally, and that Cryer’s theory of the accelerant’s pour pattern and location would have required the use of ten gallons of fuel. Second, Custer testified that the irregular burn patterns on the deck— which Cryer believed indicated that accel-erants were used — were caused simply by intense radiation.

Indeed, elaborated Custer, whether or not an accelerant had been used, the fire would have spread downward, and any evidence of accelerant in the gaps would have been destroyed by the fire itself. Third, Custer discounted Cryer’s belief that accelerant use was evinced by the fire’s speed, noting that rapidly spreading fires are common in residences with modern furnishings and room sizes. Finally, Custer dismissed Cryer’s explanation for the state of the smoke detector- — that it had been dislodged prior to the apartment’s ignition- — as speculative.

Custer noted that the high temperature around the ceiling could easily have dislodged even a smoke detector that was mounted. Custer concluded that the fire scene reflected an accidentally set fire that experienced “flashover” — a possible collateral consequence of the fire’s radiated heat causing additional objects to combust. Custer produced three computer models detailing how the fire would have progressed had it spread accelerant-free, and determined that flashover would have occurred in the two most likely scenarios.

Moreover, based on his knowledge of Hope Collins’s smoking habits at the time, photographs that showed the presence of alcoholic beverages and smoking paraphernalia on the evening of the fire, and the fact that Collins and some of her friends had been using marijuana that evening, Custer concluded that the fire was likely caused by a carelessly discarded cigarette that had smoldered, for several hours, between the cushions of the living room couch. Andrew Armstrong, Ph.D., a chemist and forensic scientist, echoed Custer’s conclusions.

Reviewing the State’s raw data — and also applying what he believed to be contemporaneous scientific standards — Armstrong concluded that “there is no evidence of an identifiable ignitable liquid in any of the samples from the fire scene.” The sample of wood obtained from the balcony revealed no paint thinner; the chromatogram from the sample did not “even look close” to the pattern associated with paint thinner, and instead revealed only turpentine, which occurs naturally in the type of wood used to construct patios.

The pattern purportedly revealing paint thinner in the carpet, opined Armstrong, did not conclusively indicate paint thinner, as it was also consistent with common household products such as furniture polish or insecticide. Moreover, although the State concluded that both of the aforementioned samples contained paint thinner, Armstrong testified that the two samples did not even reveal the same substance. Armstrong also testified that the carpet sample which the State concluded contained gasoline lacked the “five peak component” pattern indicative of gasoline.

Although acknowledging that the ASTM had yet to adopt the “five peak” standard in 1986, Armstrong testified that he and other experts employing gas chromatography knew of and relied upon the standard at that time.

Noted Armstrong, “Many carpet samples that have been exposed to fire conditions contain [the compounds identified by the State Arson Lab] in some concentration.” Armstrong stressed that only the “five peak” standard can eliminate false positives, and that “the elimination of false positives is of utmost importance.” To the extent that the carpet did reveal gasoline, elaborated Armstrong, it would have been gasoline that had evaporated, and it would have been impossible to determine when the gasoline had reached the carpet.

Armstrong agreed that the carpet could have absorbed gasoline from the parking lot — which, recall, was located near gasoline pumps — on which the carpet had rested for three weeks. Finally, Armstrong touched on a broader problem with the State’s methodology that undermined its conclusions: no control samples had been compared with the tainted ones. In April 1994, Richey moved to reopen his appeal, arguing that he lacked the effective assistance of appellate counsel. His motion was denied by the court of appeals, which was affirmed by the Ohio Supreme Court. See State v.

Richey, 73 Ohio St.3d 523, 653 N.E.2d 344 (1995). Richey’s petition for postconviction relief was denied by the trial and appellate courts, State v. Richey, 1997 WL 722782 (Ohio Ct.App. Nov. 18, 1997), and denied review by the Ohio Supreme Court, State v. Richey, 81 Ohio St.3d 1467, 690 N.E.2d 1287 (Ohio Mar. 11, 1998) (table). Richey’s second state postconviction petition was similarly denied by the trial and appellate courts, State v. Richey, 2000 WL 681642 (Ohio Ct.App. May 26, 2000), and further review was again denied by the Ohio Supreme Court, State v.

Richey, 90 Ohio St.3d 1427, 736 N.E.2d 25 (Ohio Oct. 4, 2000) (table). Advancing to federal district court, Richey filed a petition for a writ of habeas corpus, alleging that both his conviction and sentence were tainted by constitutional errors. The district court denied Rich-ey’s petition. Although it refused to conduct an evidentiary hearing, the district court cited, in defense of the State’s forensic conclusions, an extra-record

article written by forensic scientist Anthony Cafe.

Cafe has subsequently averred that the district court “miscited and misunderstood my published articles” and that “most of the world’s leading forensic scientists in this field would be horrified if they saw the chromatograms used to convict Kenny Richey.” Cafe warned that “[i]f Kenny Richey were executed on the basis of this scientific evidence, then these chromato-grams will become historical documents, examined by scientists all over the world and used to show just how wrong forensic evidence can be.” The district court certified for appellate review eight of Richey’s grounds for relief.

Richey timely appealed, and we added three claims to the certificate of appeala-bility. Included in the certificate of ap-pealability is the question of whether there was insufficient evidence to convict Richey of the crimes with which he was charged or tried, as well as whether he suffered ineffective assistance of counsel. II. ANALYSIS We apply the familiar habeas rules. Because Richey’s petition was filed after April 24, 1996, it falls under the restrictions imposed by the Antiterrorism and Effective Death Penalty Act (“AEDPA”). Lindh v. Murphy, 521 U.S. 320, 326, 117 S.Ct. 2059, 138 L.Ed.2d 481 (1997).

AED-PA requires that we respect any determination made by the state court unless it: (1) “was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court,” or (2) “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceedings.” 28 U.S.C. §§ 2254(d)(l)-(2); see also Williams v. Taylor, 529 U.S. 862, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000).

Within these constraints, we review the district court’s legal conclusions de novo, but we may upset its factual determinations only if they were clearly erroneous. Lott v. Coyle, 261 F.3d 594, 606 (6th Cir.2001). A. Sufficiency of the Evidence The heart of the prosecution against Richey — and the basis for his death sentence — was that he committed aggravated felony murder, which is prohibited by Ohio Revised Code § 2903.01.

Richey argues that to convict a person of aggravated felony murder under the statute, the State is required to prove that the defendant intended to kill the person who actually died (in this case, Cynthia Collins), but that here, the State introduced no evidence that Richey actually intended to kill Cynthia. The State disputes this argument on the merits, and also urges that the claim has been procedurally defaulted. Because a discussion of the procedural default in this case requires an understanding of the merits, we will discuss the merits of the claim first. 1.

Merits The district court rejected Richey’s statutory argument, ruling that “this claim is not cognizable in a federal habeas court as it is one touching purely on state law.” In so holding, the district court misunderstood Richey’s claim. According to Fiore v. White, 531 U.S. 225, 121 S.Ct. 712, 148 L.Ed.2d 629 (2001), the federal constitution’s Due Process Clause “forbids a State to convict a person of a crime without proving the elements of that crime beyond a reasonable doubt.” Id. at 228-29, 121 S.Ct. 712.

Therefore, the state law question regarding the elements of the crime predicates the enforcement of Richey’s federal constitutional right. Thus, we must first look to state law to determine whether the State was required to prove, as an element of the crime, that Richey intended to kill the victim. a. Elements of Aggravated Felony Murder At the time of the fire, the statute at issue provided: (

B) No person shall purposely cause the death of another while committing or attempting to commit, or while fleeing immediately after committing or attempting to commit ... aggravated arson or arson .... * * * (

D) No person shall be convicted of aggravated murder unless he is specifically found to have intended to cause the death of another.... If a jury in an aggravated murder case is instructed that a person who commits or attempts to commit any offense listed in division (

B) of this

section may be inferred ... to have intended to cause the death of any person who is killed during the commission of, attempt to commit, or flight from the commission of or attempt to commit the offense, the jury also shall be instructed that the inference is non-conclusive, that the inference may be considered in determining intent, that it is to consider all evidence introduced by the prosecution to indicate the person’s intent and by the person to indicate his lack of intent in determining whether the person specifically intended to cause the death of the person killed, and that the prosecution must prove the specific intent of the person to have caused the death by proof beyond a reasonable doubt.

O.R.C. § 2903.01(law effective prior to July 1996) (emphasis added). Although subsection B, the felony murder

section of the statute, mandates no nexus between the specific intent of the arsonist and the identity of the victim, subsection D, which informs the intent required under subsection B, mandates that the jury be instructed to consider whether the defendant “specifically intended to cause the death of the person killed, and that the prosecution must prove the specific intent of the person to have caused the death by proof beyond a reasonable doubt.” O.R.C. § 2903.01(D) (emphasis added).

The plain meaning is unmistakable: according to the statutory text, the defendant must have killed the person that he actually intended to kill. Indeed, Ohio courts have interpreted subsection D of the statute in this exact way. In State v.

Maurer, 15 Ohio St.3d 239, 473 N.E.2d 768 (1984), the court upheld an aggravated felony murder conviction under subsection D because the judge instructed the jury that it had to find beyond a reasonable doubt that not only did the defendant Mil the victim in the course of a felony (a Mdnapping), but also that “there was present in the mind of the defendant a specific intention to Mil Dawn Marie Hendershot,” the person whom the defendant was charged with killing. Id. at 779; see State v. Brewer, 1984 WL 6695 at *3 (Ohio Ct.App.

July 18, 1984) (affirming conviction where the jury was instructed to determine “whether defendant specifically intended to cause Tewksbury’s death,” where Tewksbury was the person killed); State v. Grant, 67 Ohio St.3d 465, 620 N.E.2d 50, 62 (1993) (upholding an aggravated felony murder conviction where the defendant “burn[ed] down an occupied home, known to contain children, in order to Mil the children") (emphasis added); see also Clark v.

Jago, 676 F.2d 1099, 1102 (6th Cir.1982) (“Ohio has left the company of those jurisdictions which have traditional ‘felony murder’ offenses - [ujnder Ohio law ... purpose to Mil is an essential element of the crime of aggravated murder.”). The State argues, however, that subsection D has no bearing on the elements of aggravated felony murder, and “does not require the State to do anything — it concerns jury instructions.” This argument is curious, to say the least. First, it is clear from the statute and case law that the State is required to do something: it is required to prove the element of specific intent.

Second, there is no meaningful difference between requiring the State to do something and requiring that the jury be informed that the State is required to do something. While the second formulation requires more words to articulate, its practical effect is identical. Nor is it fathomable, as the State suggests, that Ohio’s General Assembly intended to require specific intent when the defendant was tried by a jury, but not when the defendant was tried by a panel of judges. Such a reading would mean that the elements of the crime can change depending upon who the fact-finder is.

Because such a dichotomy would be unconstitutionally irrational, we are unable to embrace the State’s

interpretation. See I.N.S. v. St. Cyr, 533 U.S. 289, 299-300, 121 S.Ct. 2271, 150 L.Ed.2d 347 (2001) (noting that statutes should be interpreted to avoid unconstitutional results). Consequently, in order to convict a defendant of the crime of aggravated felony murder in Ohio, the State must prove that the defendant specifically intended to kill the victim. Not only does subsection D provide for the specific intent requirement, it provides for the one way that the State can more easily satisfy this requirement.

The statute creates the possibility of a permissive inference by a jury, whereby the jury can infer from the circumstances of the felony, that the defendant intended to cause the death of the person actually killed. Ohio Rev. Code Ann. § 2903.01(D). The statute makes clear, however, that this inference is not conclusive, and that the jury must consider all the evidence on both sides, including any evidence presented by the defendant “to indicate his lack of intent in determining whether the [defendant] specifically intended to cause the death of the person killed...” Id.

Although the statute details a less onerous means by which the State can prove the specific intent of a defendant to kill the actual victim, even in this circumstance, the statute does not change the substance of what the State must prove: that the defendant intended to kill the very person who actually died. See State v. Brewer, 1984 WL 6695, at *3 (Ohio Ct.App.). b. Transferred Intent In Richey’s case, the State presented no evidence that he specifically intended to kill Cynthia Collins, and in fact conceded this point during oral argument before the state court.

Instead, the State presented evidence that Richey had a motive to kill Candy Barchet and/or Mike Nichols, and thus, the State assumed that the doctrine of transferred intent could be used to find that Richey specifically intended to kill Cynthia. Because nothing in the language of O.R.C. § 2903.01 supports the idea that the Ohio General Assembly affirmatively intended for transferred intent to apply to aggravated felony murder, the State’s sole argument would be that the doctrine of transferred intent, derived from the common law, applies to all intent crimes. However, the plain text of O.R.C. § 2903.01(

D) indicates that the statute not only fails to affirmatively permit transferred intent, it actually precludes it. First, as previously stated, subsection D already specifies the single way in which the State is permitted to utilize a more flexible approach to proving specific intent: the permissible inference. Thus it would be overreaching for this Court to read into the statute another flexible approach, such as transferred intent. Second, as a general rule of statutory construction, every word has a meaning, and the courts should not render statutory language superfluous. Hibbs v.

Winn, — U.S. —, 124 S.Ct. 2276, 2286, 159 L.Ed.2d 172 (2004); D.A.B.E., Inc. v. Toledo-Lucas County Bd. of Health, 96 Ohio St.3d 250, 773 N.E.2d 536, 543 (2002). In this case, the State urges us to construe the phrase “intended to cause the death of the person killed” to mean “intended to cause the death of another.” The difference in meaning is profound, and it is worth noting that the Ohio General Assembly has actually used the phrase “death of another” in another statute: its traditional felony murder statute. See Ohio Rev.

Code Ann. § 2903.02(B) (stating that “[n]o person shall cause the death of another as a proximate result of the offender’s committing or attempting to commit an offence of violence that is a felony”). The fact that similar language was not used in this statute appears purposeful, suggesting that Ohio chose to use a heightened intent requirement here, to make aggravated felony murder deserving of the death penalty. Fortunately, we need not go so far as to interpret Ohio’s statutory language to determine whether the law permitted transferred intent in aggravated felony murder cases.

The Ohio courts themselves offered insight on this issue. At the outset, the assumption that the doctrine of transferred intent broadly overlays all intent crimes is incorrect. “Ohio courts have noted that the state legislature is aware of the doctrine of transferred intent at times applying the doctrine and other times refusing to apply the doctrine to a statute.” In re A.C.T., 158 Ohio App.3d 473, 816 N.E.2d 1098, 1100 (2004).

Although there have not been many judicial pronouncements on the narrow issue of applying transferred intent to aggravated felony murder, all the cases that directly touch upon it suggest that the doctrine is inapplicable here. In State v. Sowell, 39 Ohio St.3d 322, 530 N.E.2d 1294 (1988), the defendant relied on Ohio Rev. Code Ann. § 2903.1(

D) to argue that transferred intent could not support his conviction for aggravated (non-felony) murder under § 2903.1(A). In rejecting his argument, the Ohio Supreme Court drew a distinction between the two subsections’ treatment of transferred intent; such that subsection D could not be read to vitiate the incorporation of transferred intent in subsection A. See id. at 1305. In so doing, the court explicitly limited its application of transferred intent to cases of aggravated murder, not. aggravated felony murder. See id.

This suggests that the doctrine of transferred intent did not apply to subsection D, and at the very least, makes clear that the doctrine of transferred intent is not so over-arching as to be automatically applicable in proving all intent crimes. One Ohio court has gone further to state that: The doctrine [of transferred intent] has been applied for many years in Ohio but has apparently been removed by the legislature from application in aggravated murder cases. In revising R.C. 2903.01(D), the legislature [enacted the text at issue]...

The legislature did not remove the doctrine of transferred intent from application in determining the absence or presence of purpose to kill in murder, as opposed to aggravated murder, convictions. The limitation of the legislative reference to aggravated murder implies to a point that the legislature intended for the doctrine of transferred intent to have applicability in situations involving lesser crimes such as murder. State v.

Mullins, 76 Ohio App.3d 633, 602 N.E.2d 769, 771 (1992) (upholding a murder conviction based on transferred intent where the defendant, though was initially indicted on a charge of aggravated murder, was actually found guilty by the jury on a charge of non-aggravated murder). The preceding state case law — and it is the only Ohio case law at the time that squarely addressed this issuq-Hbndicates that transferred intent does not apply to subsection D of the aggravated felony murder statute. Accordingly, in determining what the required elements of aggravated felony murder were under Ohio Rev.

Code Ann. § 2903.1, we are compelled to adhere to Ohio’s specific pronouncements at the time of Richey’s conviction and state appeals. Therefore, because the State chose to prosecute Richey for aggravated felony murder (subsection B and

D) of the statute, rather than straight aggravated murder (subsection A), the State had to prove, as an element of the crime, that Richey intended to kill Cynthia Collins. The State does not address the aforementioned case law head-on. Instead, the State claims that the case law was addressed and disposed of by the Ohio Supreme Court. Thus, the State turns a blind eye to the status of the law at the time of Richey’s conviction and the filing of his appeal; it instead relies on the Ohio Supreme Court’s pronouncement in Rich-ey’s case to support its

interpretation of what elements it needed to prove. First, it is unclear if the issue of transferred intent as applied to subsection D of the statute was truly decided by the Ohio Supreme Court. The State points to a portion of the court’s opinion entitled “Sufficiency of the Evidence,” where the Ohio Supreme Court explains why there was sufficient proof that Richey was the arsonist. State v. Richey, 64 Ohio St.3d 353, 595 N.E.2d 915, 924 (1992). Although this

section addressed many issues, the court stated that, “The fact that the intended victim escaped harm, and that an innocent child, Cynthia Collins, was killed instead, does not alter Richey’s legal and moral responsibility.” Id. at 925. The court then proclaimed that “transferred intent is firmly rooted in Ohio law.” Critically, it cited only eases addressing subsection A of the statute-cases dealing with aggravated murder and not aggravated felony murder, for which Richey was convicted.

However, even assuming that the Ohio Supreme Court’s brief statement in Rich-ey’s case was meant to interpret the aggravated felony murder statute, that court, although it has the ultimate authority in interpreting state law, would have been offering a novel, or at least unforeseen,

interpretation of the statute. Prior to the holding in Richey’s case, both the text of the statute and its treatment by the Ohio courts strongly indicated that transferred intent could not be applied to aggravated felony murder. We have not found a single Ohio case where transferred intent was applied and upheld where the defendant was charged with aggravated felony murder, whereas the language in Sowell and Mullins suggests the contrary. Thus, under the State’s

interpretation of the Ohio Supreme Court’s opinion, the court would have been expanding what at the time was the known scope of the statute, raising what the United States Supreme Court has described as a most serious due process concern: “that a statute precise on its face has been unforeseeably and retroactively expanded by judicial construction.” Bouie v. City of Columbia, 378 U.S. 347, 351, 84 S.Ct. 1697, 12 L.Ed.2d 894 (1964). It is well settled that the government cannot establish the sufficiency of the evidence by resorting to an unforeseeable judicial construction of a criminal statute.

The Supreme Court in United States v. Lanier, 520 U.S. 259, 117 S.Ct. 1219, 137 L.Ed.2d 432 (1997), stated what it deemed to be the three manifestations of the fair warning requirement: First, the vagueness doctrine bars enforcement of a statute which either forbids or requires the doing of

an act in terms so vague that men of common intelligence must necessarily guess at its meaning and differ as to its application. Second ... the canon of strict construction of criminal statutes, or rule of lenity, ensures fair warning by so resolving ambiguity in a criminal statute as to apply it only to conduct clearly covered.

Third, although clarity at the requisite level may be supplied by judicial gloss on an otherwise uncertain statute, due process bars courts from applying a novel construction of a criminal statute to conduct that neither the statute nor any prior judicial decision has fairly disclosed to be within its scope. Id. at 266,117 S.Ct. 1219 (internal citations and quotations omitted). See Gall v.

Parker, 231 F.3d 265, 305-06 (6th Cir.2001) (granting writ of habeas corpus upon holding that elimination of an element of a crime, or altering the necessary proof by the state supreme court violated Due Process Clause when both the plain text of the statute and prior state court decisions instructed otherwise); see also Rabe v. Washington, 405 U.S. 313, 316, 92 S.Ct. 993, 31 L.Ed.2d 258 (1972) (reversing as violative of the Due Process Clause a state law conviction that rested on unforeseeable judicial

interpretation of a criminal statute). Accordingly, if the Ohio Supreme Court deliberated on, and decided the issue presented before us now, it would have been a novel

interpretation of the statute, because the plain meaning and the court pronouncements in Sowell and Mullins indicated the opposite of what the State now urges. Prior to Richey’s appeal before the Ohio Supreme Court, an Ohio appeals court (.Mullins) actually cited to subsection D of the aggravated felony murder statute as an example of an instance where the Ohio General Assembly read out of a statute the transferred intent doctrine. Furthermore, to the extent that the statute is ambiguous, the rule of lenity ensures that the requirement of fair warning is met by resolving doubts in favor of the defendant.

United States v. Bass, 404 U.S. 336, 348, 92 S.Ct. 515, 30 L.Ed.2d 488 (1971). Although the Ohio Supreme Court is the ultimate arbiter of the State’s laws, the fact that the law prior to Richey’s final state appeal points to an

interpretation of the statute without transferred intent raises serious due process concerns of retroac-tivity and notice. It must be said that this Court cannot decide what Ohio law should be, and that the Ohio Supreme Court could certainly decide that transferred intent was meant to apply to subsection D of the statute, were its decision to be made in a vacuum.

However, because there was not a single judicial pronouncement suggesting that transferred intent could be used under subsection D, and because the prior case law affirmatively indicates to the contrary, the Ohio Supreme Court would have raised a federal constitutional concern of impermissible retroactivity, had it authoritatively read transferred intent into subsection D. Therefore, if we accept the State’s

interpretation of the Ohio Supreme Court’s opinion as having done this, it would have been a retroactive or novel application of a statute which would have violated Richey’s clearly established rights under the Due Process Clause. The State could point to no Ohio law, prior to Richey’s own case, that stands for the proposition that transferred intent applies in aggravated felony murder cases. Thus, in compliance with Ohio law, we must find that the State could not apply transferred intent in order to satisfy the specific intent requirement in an aggravated felony murder case.

Whether or not the Ohio Supreme Court reached the issue of transferred intent as applied to subsection D, the State did not meet its burden of proving all the required elements of aggravated felony murder. Based on the state of the law at the time of his actions, the only way that Richey could have been constitutionally convicted of aggravated felony murder would have been upon a showing that Richey intended to kill the person that actually died. Because it is undisputed that there was no evidence to this effect, Richey’s conviction necessarily lacked the support of sufficient evidence. 2.

Procedural Default Aside from the merits, the State argues that Richey has defaulted this claim because he failed to press this argument during his direct appeal. “In all cases in which a state prisoner has defaulted his federal claims in state court pursuant to an independent and adequate state procedural rule, federal habeas review of the claims is barred unless the prisoner can demonstrate cause for the default and actual prejudice as a result of the alleged violation of federal law, or demonstrate that failure to consider the claims will result in a fundamental miscarriage of justice.” Coleman v.

Thompson, 501 U.S. 722, 750, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991). Richey concedes that he did not raise the claim at trial or on direct appeal, as he was required to by Ohio rules. The first problem with the State’s argument that the claim was defaulted is the question whether the state courts actually relied on an independent ground to deny Richey relief on this claim. The ground is independent “only if the state court rendering judgment in the case clearly and expressly stated that its judgment rested on a procedural bar.” Simpson v. Sparkman, 94 F.3d 199, 202 (6th Cir.1996).

In other words, a lapsed claim survives if the state court overlooked the default and decided the claim anyway. See Warden v. Hayden, 387 U.S. 294, 297 n. 3, 87 S.Ct. 1642, 18 L.Ed.2d 782 (1967). Thus, if, as the State contends, the Ohio Supreme Court did decide Richey’s claim regarding transferred intent, this Court is free to review the claim without concern regarding the procedural bar.

While we express doubt regarding whether the Ohio Supreme Court truly considered transferred intent as applied to subsection D of the statute, it is clear that the Ohio Supreme Court considered the issue of transferred intent generally with regards to Richey’s case. The court stated that “the culpability of a scheme designed to implement the calculated decision to kill is not altered by the fact that the scheme is directed at someone other than the actual victim.” Richey, 595 N.E.2d at 925 (internal quotations omitted).

Moreover, at no point did the Ohio Supreme Court hold, or even intimate, that Richey had forfeited the chance to contest the sufficiency of the evidence supporting his conviction. Because the United States Supreme Court requires a “ ‘plain statement’ that a [state court] decision rests upon an adequate an independent state ground[ ],” Michigan v. Long, 463 U.S. 1032, 1042, 103 S.Ct. 3469, 77 L.Ed.2d 1201 (1983), we must interpret this ambiguity to mean that the Ohio Supreme Court reached the issue of transferred intent. Therefore, there was no independent state ground barring our review.

Additionally, Richey argues that the ineffective assistance of his trial and appellate counsel supplies the cause needed to excuse his procedural default. Indeed, “[i]neffeetive assistance of counsel can supply the cause that, together with prejudice, would excuse a procedural default.” McFarland v. Yukins, 356 F.3d 688, 699 (6th Cir.2004). To demonstrate cause for the procedural default, Richey must show that both his trial and appellate counsel were ineffective.

He must demonstrate that trial counsel was ineffective for failing to challenge the sufficiency of the evidence, and he must establish that appellate counsel was ineffective for failing to argue that trial counsel was ineffective for failing to challenge the sufficiency of the evidence.

Whether appellate counsel was ineffective for failing to raise the ineffectiveness of trial counsel (predicated on trial counsel’s failure to challenge the sufficiency of the evidence) turns, of course, on whether appellate counsel would have been ineffective, absent the default below, for failing to challenge the sufficiency of the evidence. To convince us that his counsel was ineffective, Richey must demonstrate that: (1) counsel’s performance was deficient; and (2) that the deficiency produced prejudice. Strickland v. Washington, 466 U.S. 668, 687, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984).

Richey encounters a threshold hurdle in asserting the ineffectiveness of his appellate counsel: he may not rely on ineffective assistance as cause to excuse the procedural default of his transferred intent/sufficiency of the evidence claim, if the ineffectiveness assistance claim is itself procedurally defaulted. Ohio Rule of Appellate Procedure 26(

B) provides that “[a] defendant in a criminal case may apply for reopening of the appeal from the judgment of conviction and sentence, based on a claim of ineffective assistance of appellate counsel. An application for reopening shall be filed ... within ninety days from journalization of the appellate judgment unless the applicant shows good cause for filing at a later time.” Richey’s motion to reopen his appeal in the Ohio courts — in which he asserted that his appellate counsel was deficient — was untimely, and the Ohio Supreme Court denied the motion as untimely. See Richey, 653 N.E.2d at 345.

The question remains, however, whether the rule was an “adequate” basis to preclude federal relief. “[A] rule is adequate if it is regularly or consistently applied by the state court[.]” White v. Schotten, 201 F.3d 743, 751 (6th Cir.2000). Richey argues that the rule does not fall within this criterion because it provides no definition of what constitutes “good cause” for an untimely filing and because the Ohio courts had neither articulated meaningful standards nor consistently applied any type of criteria.

The district court below held that Richey had not defaulted his ineffective assistance claim because “certainly as of the date [that] Richey filed his Rule 26(

B) application, the ‘good cause’ rule had not yet been applied with the regularity necessary to constitute an adequate and independent state procedural rule.” See also Anderson v. Attorney General, 342 F.3d 1140 (10th Cir.2003) (“[SJtate procedural rules that bar ineffective assistance of counsel claims are further reviewed with a healthy degree of skepticism.”) (internal quotations omitted); Page v. Frank, 343 F.3d 901, 909 (7th Cir.2003) (same). We agree with the district court below. We measure whether Rule 26(

B) was “firmly established and regularly followed by the time as of which it [was] to be applied.” Ford v. Georgia, 498 U.S. 411, 423-24, 111 S.Ct. 850, 112 L.Ed.2d 935 (1991) (quotations omitted). Although we did not decide this question directly in White, we did observe that “the state courts have not achieved consensus on what constitutes ‘good cause’ to excuse non-compliance with Rule 26(B).” Id. at 751. This was certainly true at, and even after, the time when Richey was required to file his motion to reopen under 26(B). Cf. Mitchell v.

Mason, 325 F.3d 732, 739 (6th Cir.2003) (“[I]t is the Warden’s burden to demonstrate that the rule was already firmly established and regularly followed as of the time the state court invoked the rule.”). In some instances, the Ohio courts failed to excuse the most minor of violations caused by circumstances beyond the control of both the litigant and his lawyer. See, e.g., State v.

Winstead, 74 Ohio St.3d 277, 658 N.E.2d 722 (1996) (finding no good cause for a filing that was one day late, and that had resulted “because the overnight courier that appellant’s counsel had used failed to deliver the application before the deadline for filing expired”). In other cases, the courts found good cause for significant violations even yriien no explanation whatsoever was provided by the litigant, solely because the litigant was proceeding pro se. See, e.g., State v. Nitenson, 1994 WL 69894 at *2 (Ohio Ct.App.

Feb. 24, 1994) (finding good cause “in the interests of justice” even though “application was filed more than one year after we journalized our affirmation of his conviction and he gives no sworn statements whatsoever in his application”). Nor were these inconsistent cases outliers in an otherwise consistent application of 26(B). Compare, e.g., State v. McCarter, 1993 WL 107800 (Ohio Ct.App. Aug. 12, 1993); State v. Klein, No. 49260 (Ohio Ct.App. March 15, 1994); with State v. Wright, 1994 WL 398805 (Ohio Ct.App. July 29, 1994); State v. Fields, No. 95 CA-08-048 (Ohio Ct.App. April 21, 1994).

Thus, we find that Richey’s challenge to the effectiveness of his appellate counsel is preserved. We now turn to the merits of Richey’s ineffective assistance of counsel claim as an excuse for his procedural default. The deficiency of Richey’s trial and appellate counsel stems from the same shortcoming: that they failed to grasp that the State did not prove (and, indeed, had not even attempted to prove) that Richey specifically intended to cause the death of Cynthia Collins, as was required by the aggravated felony murder statute.

The failure to understand the elements of the charged offenses and hold the State to proving those elements is among the most basic responsibilities of competent counsel. “At the least, defense counsel in a criminal case should understand the elements of the offenses with which his client is charged and should display some appreciation of the recognized defenses thereto.” Scarpa v. DuBois, 38 F.3d 1, 10 (1st Cir.1994).

The circumstances of Richey’s case magnified counsel’s obligation to understand the elements, because his conviction of aggravated felony murder subjected him not just to imprisonment, but to execution. It would be one thing if counsel had failed to grasp a subtle question of statutory

interpretation or had failed to raise an issue that would have been uncovered by only the most astute of jurists. However, as we explained above, at the time of Rich-ey’s trial and appeal — not the time of the Ohio Supreme Court’s ruling — both the plain text of the Ohio Code and the judicial decisions interpreting it indicated that aggravated felony murder required specific intent, and that the doctrine of transferred intent was unavailable to the State. See Prou v.

United States, 199 F.3d 37, 48 (1st Cir.1999) (“Where, as here, an attorney fails to raise an important, obvious defense without any imaginable strategic or tactical reason for the omission, his performance falls below the standard of proficient representation that the Constitution demands.”). Indeed, we have held that counsel can even be ineffective for failing to raise a legal claim that turns on an unresolved question of law, so long as counsel is aware that an unresolved issue exists. See Combs v.

Coyle, 205 F.3d 269, 286 (6th Cir.2000) (resolving circuit split in defendant’s favor, and then holding that counsel’s failure to raise the objection was deficient because he “should have realized that the [practice at issue] was at least constitutionally suspect” — as a result, “[c]oun-sel’s failure to have objected at any point is inexplicable, and we can perceive no possible strategic reason for such a failure”); see also United States v.

Bass, 310 F.3d 321, 327-30 (5th Cir.2002) (finding that appellate counsel was ineffective for failing to challenge sufficiency on direct appeal when defendant’s conduct fell outside the statute). It follows with even greater force that when both text and precedent compelled the conclusion, and the conclusion would have gutted, as a matter of law, the primary charge and death sentence, counsel was deficient for failing to raise it. Having established that both trial and appellate counsel were deficient, Richey must also demonstrate prejudice, a showing which is straightforward in this ease.

Because we have already determined that specific intent was clearly a requirement and the State failed to meet this requirement, Richey has demonstrated the necessary prejudice from the claimed violation of federal law. By failing to challenge the sufficiency of the evidence, trial counsel allowed the State to convict Richey and sentence him to death for a crime in which it had failed to satisfy one of the elements.

Although the State may have been able to apply the doctrine of transferred intent for other crimes, such as felony manslaughter or perhaps even aggravated murder, the State chose to charge Richey with aggravated felony murder. At the time of the fire, the State could not convict someone of aggravated felony murder without proving that the defendant specifically intended to kill the person who was actually killed.

Had Rich-ey’s attorneys presented a proper challenge of the application of transferred intent to subsection D of the statute, Richey could not have been constitutionally convicted of aggravated felony murder, because the State did not prove Richey’s specific intent to kill the victim. Fiore v. White, 531 U.S. 225, 228-29, 121 S.Ct. 712, 148 L.Ed.2d 629 (2001). And by failing to raise this (and trial counsel’s ineffectiveness) on appeal, appellate counsel again deprived Richey of dispositive relief. Richey was clearly prejudiced by his attorney’s deficiencies.

Thus, the procedural default of Richey’s transferred intent/sufficiency claim is excused because of the ineffective assistance of both his trial and appellate counsel. Because the Ohio Supreme Court considered the general claim regarding transferred intent, the state procedural rule did not constitute an independent, unreviewable state law ground; however, even if it did, the trial and appellate defaults were excused by cause (the result of ineffective assistance of trial and appellate counsel) and prejudice (inherent in the fact that we find the claim to be meritorious).

In either case, Richey’s claim was preserved for review. B. Ineffective Assistance of Counsel In a challenge that implicates his convictions on all charges, Richey also contends that his trial counsel was constitutionally ineffective in the way he handled his forensic expert and his failure to challenge the scientific evidence of arson. The State argues that this claim has been procedurally defaulted and disputes the claim on the merits. 1. Default The State argues that this claim has been defaulted because Richey neglected to raise it on direct appeal.

As the Ohio Supreme Court has explained, however, “[a]ny allegations of ineffectiveness based on facts not appearing in the [trial] record should be reviewed through the post conviction remedies.” State v. Coleman, 85 Ohio St.3d 129, 707 N.E.2d 476, 483 (1999). See also Byrd v. Collins, 209 F.3d 486, 521 (6th Cir.2000).

As our analysis below will illustrate in detail, Richey’s claims regarding his counsel’s ineffective handling of the scientific evidence depend almost entirely on evidence outside the trial record: namely, (1) the testimony of DuBois, who explained trial counsel’s limited oversight, supervision, and engagement; (2) the testimony of trial counsel, which illustrated the process (or lack thereof) by which he decided to hire DuBois, as well as his oversight (or lack thereof) over the course of DuBois’s scientific investigation; and (3) the testimony of Custer and Armstrong, who explained the type and quality of scientific analysis that a reasonably competent expert would have performed at the time of Richey’s trial.

Thus, Richey was prohibited from raising these claims on direct appeal. Instead, Richey was required to raise the claims during his postconviction challenge. He did so. During the postconviction proceedings, Richey argued that he was “wrongfully convicted on the basis of false scientific testimony because trial counsel unreasonably relied upon DuBois, failed to conduct an adequate investigation for a competent expert and failed to call an expert to rebut the State’s false scientific testimony,” and supported these claims with the statements of trial counsel, DuBois, Custer, and Armstrong.

The state courts evaluated and decided the claim on the merits. See Richey, 1997 WL 722782 at *3. We therefore may do the same. 2. Merits As previously noted, Strickland requires that Richey demonstrate both deficient performance and resulting prejudice. a. Deficient Performance In order to show deficient performance, a defendant must show that counsel’s representation “fell below an objective standard of reasonableness.” Strickland, 466 U.S. at 688, 104 S.Ct. 2052.

A “court deciding an actual ineffectiveness claim must judge the reasonableness of counsel’s challenged conduct on the facts of the particular case, viewed as of the time of counsel’s conduct.” Id. at 690, 104 S.Ct. 2052. “In making that determination, the court should keep in mind that counsel’s function, as elaborated in prevailing professional norms, is to make the adversarial testing process work in the particular case.” Id.

At the outset, the State, and the dissent to this opinion, argue that Richey has no constitutional right to the “effective assistance of an expert.” The district court also noted that it was not unreasonable for trial counsel to “decline [to conduct] further expert-shopping once that expert rendered his opinion.” Both might be true if trial counsel had done everything he was supposed to have done, and— through no fault of trial counsel — the expert had simply come up short. See Skaggs v.

Parker, 235 F.3d 261, 268 (6th Cir.2000) (defense expert’s erratic and incoherent testimony during trial was not attributable to the failings of counsel given that the expert had been recommended by two colleagues and that counsel had used the expert before). But Richey argues, and the record reflects, that the failures of Richey’s expert were largely caused by the failures of Richey’s counsel. See Bloom v. Calderon, 132 F.3d 1267, 1277 (9th Cir.1997) (“[Counsel's failure to adequately prepare his expert and then present him as a trial witness [amounts to] constitutionally deficient performance.”).

As Richey’s counsel observes, “[i]ncompetence cannot excuse incompetence.” First, trial counsel hired DuBois, based solely on a promotional flier that he happened to come across. DuBois was unqualified for the job because he worked primarily in another field, lacked training to do the job, and came into the litigation believing that the State’s experts were more reliable than he was. Moreover, the minimal training he had received came from the very people whose conclusions he was being hired to review.

The dissent notes that “There is no suggestion in the record that counsel was put on notice that DuBois was either incompetent or unqualified to serve as Richey’s expert.” However, we have held that the deficiencies of an expert can be imputed to counsel when counsel has failed to adequately research and screen an expert witness. See Glenn v. Tate, 71 F.3d 1204, 1210 n. 5 (6th Cir.1995) (finding deficient performance because “we are not prepared to assume that Drs. Ramani and Siddal would have been the experts retained by the defense ... if counsel had done their homework”).

Thus, counsel owes more to his client than a passive duty to watch for red flags of incompetence. Moreover, trial counsel waited two months after he was retained, and a full month after he received the State’s scientific results, to contact and meet with Du-Bois. After their first meeting, he limited DuBois’s initial investigation to ten hours — -without any basis for doing so. See Skaggs, 235 F.3d at 270 (finding deficient handling of expert not excused “simply because counsel believed it would not be worth their time to request additional money from the court”); Loyd v.

Whitley, 977 F.2d 149, 158 (5th Cir.1992) (finding deficient performance when counsel “wrongly assumed that funds were unavailable and he abandoned what he knew to be an important pursuit”). After DuBois’s stunted investigation, counsel waited another two months to meet with DuBois, who did no work in the interim. Third, counsel failed to thoroughly inform DuBois about a critical problem with the State’s scientific evidence — that the carpet had been housed in a garbage dump and on a parking lot located near gasoline pumps.

See Glenn, 71 F.3d at 1210 (“[D]e-fense counsel should obviously have worked closely with anyone retained as a defense expert to insure that the expert was fully aware of all facts that might be helpful to the defendant.”); see also Affinito v.

Hendricks, 366 F.3d 252, 260 (3d Cir.2004) (“When the key issue in a criminal case is whether the defendant suffered from a diminished capacity, we can think of nothing more critical than ensuring that the defense’ psychiatric expert has as complete and accurate a description of the facts and circumstances surrounding the crime as possible.... [Failure to provide the expert with this information] was not a trial tactic, it was gross incompetence.”); Bloom, 132 F.3d at 1273 (finding deficient performance when counsel possessed information that defendant had suffered extensive abuse as child, but counsel “failed to provide the information to Dr.

Kling or to any other physician who examined [defendant]”). Although DuBois does not remember precisely what trial counsel told him, he did acknowledge that “I don’t think- [that counsel] told me the entire sequence of events involving the carpeting after the fire and before trial.” Fourth, trial counsel kept himself in the dark about all aspects of his expert’s analysis. counsel had no idea that DuBois, rather than performing his scientific tests or even conducting his own analysis of the test results, was essentially being spoon-fed the results by Gohar (who was the chief of the State Arson lab).

Only after the trial did counsel learn that DuBois performed no testing or independent analysis; counsel had never bothered to instruct or simply ask DuBois about the investigation and analysis he had hired him to perform. Moreover, counsel did not ask for a written report from DuBois explaining his conclusions, nor did counsel ask DuBois about the bases for his conclusions, or even if the State’s conclusions suffered from any flaws or gaps.

Despite DuBois’s endorsement of the State’s forensic findings — -which seriously damaged Richey’s case — counsel, in DuBois’s words, “was surprisingly nonargumentative with me [and] didn’t challenge me on what I thought or why I thought what I did or anything.” Even though trial counsel was not a scientist, this should not relieve him of his responsibility to understand the evidence being used to convict and execute his client. Furthermore, aside from the .presentation of testimony from his own expert, he would have to be sufficiently informed to cross-examine the Stated experts. See Dees v.

Caspiri, 904 F.2d 452, 455 (8th Cir.1990) (“[C]ounsel had a duty to garner the expertise necessary to cross examine [the State’s expert].”). This responsibility was heightened here, given that the State called to the stand not only its own experts but also DuBois. See Glenn, 71 F.3d at 1210 (“We can only assume that defense counsel, not having done their homework, were not prepared to interrogate [their court-appointed experts] about the basis for the very damaging conclusions they stated.”).

That counsel was a lawyer, rather than scientist, should not have left him unable to comprehend even the basics of the science at issue. This is particularly so when the chain of custody, which is a legal issue, was critical to the reliability of the scientific evidence. Fifth, trial counsel prematurely placed DuBois on his witness list and then failed to mitigate this mistake. Because trial counsel dragged his feet in hiring and debriefing DuBois, he had to put DuBois on his witness list before he was aware how DuBois would testify.

See Skaggs, 235 F.3d at 270 (rejecting time-pressure as excuse for counsel’s performance when the time pressure was itself caused by counsel’s “waitfing] until the eleventh hour to prepare for the penalty phase and to line up a psychiatric expert to testify on [defendant’s] behalf’). This became a problem, as once counsel decided against using DuBois, the State subpoenaed him to testify on its behalf. Nor, at this point, did counsel attempt to quash the subpoena or otherwise contest the testimony, or even cross-examine DuBois when he testified. Finally, counsel failed to offer any competing scientific evidence.

Of course, it is not always the case that “counsel must continue looking for experts just because the one he has consulted gave an unfavorable opinion.” Dees, 904 F.2d at 454. However, the testimony of Custer and Armstrong makes it clear that, even in 1986, a reasonably diligent attorney would have found witnesses to attack the State’s conclusions. And “[w]here there is substantial contradiction in a given area of expertise, it may be vital in affording effective representation to a defendant in a criminal case for counsel to elicit expert testimony rebutting the state’s expert testimony.” Knott v.

Mabry, 671 F.2d 1208, 1213 (8th Cir.1982). Nor was this a case in which counsel would have had to uncover an expert willing to spew junk science, because Armstrong and Custer both agreed that the scientific standards prevailing in 1986 would have seriously undermined the State’s results, and both testified that they would have been available to apply such standards. See Soffar v. Dretke, 368 F.3d 441, 477 (5th Cir.2004) (“As was made evident during the state habeas proceedings, ... defense counsel would not have had to look far to find a ballistics expert who could have provided testimony to aid his defense.”).

Instead, counsel adopted a defense that rendered Richey a sitting duck. Given the testimony of witnesses that Richey was upset at his ex-girlfriend and had made threatening comments, trial counsel’s theory — that the fire was intentionally started by someone else — was doomed to fail. See Foster v.

Lockhart, 9 F.3d 722, 726 (8th Cir.1993) (presentation of alibi defense did not excuse counsel’s failure to investigate possibility that defendant in rape case was impotent, given that “[defendant’s] alibi defense was vulnerable”- — “[t]he vulnerability of the alibi defense shows the unreasonableness of the attorney’s failure to investigate further and present the impotency defense”). And even if counsel had reasonably believed that this defense had merit, he could not have made an informed choice without reasonably investigating the alternatives.

See Soffar, 368 F.3d at 474 (“[A]n actual failure to investigate cannot be excused by a hypothetical decision not to use its unknown results.”). Trial counsel’s failure to screen, supervise, or engage DuBois left Richey with little more than “a warm body with a prefix attached to his name,” Skaggs, 235 F.3d at 273 n. 3 (quoting Ramdass v. Angelone, 187 F.3d 396, 411 n. 1 (4th Cir.1999) (Murnaghan, J., dissenting)). In a case anchored by scientific evidence, the failure to subject this evidence to meaningful adversarial testing was woefully deficient.

Moreover, we are unable to articulate any sound professional reason why counsel did not do so. The dissent argues that counsel was obligated to provide Du-Bois’s name as a potential witness, and that the prosecution’s use of the defense’s expert does not violate the defendant’s rights.

While we agree that Richey’s rights were not violated simply because the State used DuBois as its own witness, we believe that given the context here— where the defendant had a single expert witness who conducted no independent analysis and then became the witness for the other side without any cross-examination — there was a massive failure by trial counsel in his handling of the expert witness. Accordingly, we must conclude that counsel’s performance was “outside the wide range of professionally competent assistance.” Strickland, 466 U.S. at 690, 104 S.Ct. 2052.

The state court’s conclusion to the contrary was unreasonable. b. Prejudice We proceed to measure prejudice. In order to show prejudice, a defendant must demonstrate that “there is a reasonable probability that, but for counsel’s errors, the factfinder would have had a reasonable doubt about his guilt.” Combs, 205 F.3d at 290. “A reasonable probability is a probability sufficient to undermine confidence in the trial’s outcome.” Foster, 9 F.3d at 726. In determining whether there was prejudice, we “must consider the totality of the evidence before the judge or jury.” Strickland, 466 U.S. at 695, 104 S.Ct. 2052.

The prejudice inquiry is complicated slightly in this case because trial counsel’s errors were twofold: deficiencies plagued both his hiring and handling of DuBois. Under either scenario, however, the evidence of prejudice is overwhelming. Had counsel made the effort to find a qualified expert, rather than blindly hiring DuBois, the expert would have had the expertise and wherewithal to undermine the State’s evidence that the fire was caused by arson. Custer and Armstrong highlighted a litany of irregularities in the State’s scientific evidence.

First, Custer revealed alternative explanations for the circumstances that led Cryer to finger arson as the culprit, and surmised that the fire was more consistent with an accidental outbreak. Second, Armstrong opined that “there is no evidence of an identifiable ignitable liquid in any of the samples from the fire scenes.” Moreover, the blunders that Armstrong highlighted — the State’s failure to use accepted methodology, use control groups, and eliminate other explanations, to name a few — would likely have led the factfinder to adopt the defense’s understanding of the science.

Custer and Armstrong similarly explained, without refutation, that the scientific standards upon which they relied were widely accepted in 1986, leaving little doubt that counsel would have been able to find a competent expert had he bothered to look for one. See Horsley v. Alabama, 45 F.3d 1486, 1495 (11th Cir.1995) (“[T]o prove prejudice by failure to investigate and failure to produce a certain kind of expert witness, a habeas petitioner must demonstrate a reasonable

[…]