Dugas v. Coplan

428 F.3d 317 | 1st Cir. | 2005

Lead Opinion

LIPEZ, Circuit Judge.

Peter Dugas was convicted of arson in New Hampshire state court. He petitioned for a federal writ of habeas corpus on the grounds, inter alia, that he had received constitutionally ineffective assistance of counsel under Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). The district court found that trial counsel’s performance was deficient, but found no prejudice, granted summary judgment against Dugas, and denied the petition. We agree with the district court that counsel’s performance was deficient, but' conclude that further proceedings are needed in the district court to determine whether the error resulted in prejudice within the meaning of Strickland.

I.

Peter Dugas was the manager and part owner of Dugas Superette, a grocery store in Nashua, New Hampshire, owned principally by Dugas’s father, Edgar Dugas.¹ On October 23, 1998, just before midnight, the Nashua Fire Department responded to a three-alarm fire at the store. The firefighters forced entry through locked doors and observed heavy smoke throughout the building. They opened a closed (but unlocked) door to the basement and discovered considerably thicker' smoke and extreme heat, which indicated that the source of the fire was downstairs. They ventilated the basement and extinguished the fire.

A. The Fire Marshal’s Investigation

In the early morning of October 24, Inspectors Brian Donaldson and Richard Strand of the Nashua Fire Marshal’s Office arrived to investigate the fire. The basement was still full of smoke, and the firefighters were ventilating the basement with a mechanical -fan in the- stairway. After examining the exterior of the building and allowing some time for the smoke to dissipate, the inspectors entered the building and quickly determined that the fire had originated in the basement. They examined certain obvious potential sources — -the furnace, an oil tank, .the hot water tank, some electrical equipment, and an electrical panel — and ruled them out because that area of the basement had not suffered any charring or flame damage. Further investigation narrowed the source of the fire to the southeast corner of the basement, in a pile of papers on the.floor in front of a, set of shelves. They separated the pile of papers into smaller piles in a center aisle in the basement so as to better examine the stack, and determined that the outer edges of the pile were burned, but the inside, where the pile had been tightly packed, was not burned. They also observed an electric clock that had stopped at 10:44 PM, apparently because a circuit breaker tripped after fire damaged the wires.

Using techniques of fire “cause and origin” investigation, the inspectors ruled out natural causes (such as lightning, magnified sunlight, bacterial spontaneous combustion, or static electricity) and accidental causes (such as mechanical devices, heating devices, electrical equipment or wires, or discarded cigarette butts). They noted that the basement -vvas. filled with thick smoke and unusually high carbon monoxide levels. Since there was little flame damage outside the immediate area of the, pile, the inspectors theorized that the fire had begun quickly but then became oxygen-starved and reduced to a smolder, which was how the firefighters found it.

When the inspectors concluded that the fire had been intentionally set, they decided to gather further evidence. They summoned a canine handler and fire investigator with the State Fire Marshal’s Office, who arrived with a dog trained to detect petroleum distillates, which can be used as fire accelerants. The dog “alerted” to parts of both the original pile of papers and the parts that had been moved aside.. The inspectors removed samples of those sections for laboratory analysis, and placed them in five airtight containers.

Inspector Strand brought the samples to Morris Boudreau, a forensic chemist at the State Police Forensic Laboratory. Bou-dreau detected that some, though not all, of the samples contained ignitablé' liquids: medium petroleum distillates in some, and normal alkanes in others.² Both classes of fluids can be used as fire accelerants.

B. The Insurance Investigation

The companies that insured the market and separately-owned video and pizza concessions within it hired James Eddy, a private fire investigator, to conduct an independent cause and origin investigation of the fire. Eddy arrived at the market on October 27 and, like the Nashua fire inspectors, quickly determined the approximate area of the basement where the fire originated. Like them, he eliminated natural and accidental causes early on, and concluded that the fire had been intentionally set at the pile of papers. Out of an abundance of caution he contacted Nathaniel Johnson, an electrical engineer specializing in electrical fire investigation, to examine the electrical system. Johnson determined, after exhaustive study of every inch of wire in the basement, that it was definitely not an electrical fire.

Eddy also took samples of the charred paper, wood, and other items from the basement for analysis at an independent laboratory in Massachusetts. However, he never had the chance to send the samples to that lab. The Nashua Police Department wrote to the store’s insurers, requesting all evidence that Eddy had collected. The insurers directed him to comply. He never received any information from the state regarding the analysis (if any) performed on his samples, but later testified that “it’s my understanding that all of my samples were negative.”

C. The Videotape

The store had a video surveillance system, albeit, as it turns out, a barely serviceable one. The system used a.multiplexer to produce images from several different cameras and record them on an obsolete, belt-driven video tape recorder that could no longer generate an accurate synchronization signal and had other flaws. Worse yet, the store’s practice was to continually rewind and re-record over the same tape, which not only degraded and wore out the tape (the image had become virtually undecipherable) but also meant that a recording from one day might be (and, in fact, was) followed with a leftover recording from an earlier day. Moreover, since the system did not time stamp the recordings, it was impossible to determine when a given image was recorded. In short, the system produced unreadable, undated, and difficult-to-interpret images.

The Nashua police hired an electronic surveillance specialist to attempt to decipher the tape, but it was beyond even his technical capabilities. He referred it to Tom Edwards, a forensic video analyst who specializes in image enhancement. Edwards found that the tape’s timing was off (i.e., one second on the tape did not correspond to one second in the real world) and that, because of the multiple rerecordings, the sequencing was unreliable as well (i.e., the fact that one image came after another on the tape did not mean that the second image actually happened after the first image; in fact, it could have been a leftover from a previous taping days or months before). Edwards attempted to extract a usable sequence of events from the tape, adjust the speed to match real time, and enhance the images to make them more legible. The enhanced videotape appeared to show that Dugas left the store for the night with another employee; returned three minutes' later; entered an office at the back of the store; reappeared sixty-two seconds later; briefly appeared in the camera’s field of view again; and then ran towards the back of the store. The tape then went blank, presumably because of a loss of power to the camera. Because it was undisputed that Dugas and the employee had left the store just minutes after 10:00 PM, the police interpreted the events shown on the tape as happening at that time.

D. The Police Investigation

Dugas was present outside the market while the firefighters were putting out the blaze. He spoke briefly with Inspector Strand at the scene and signed a form consenting to a search to allow the inspectors to further investigate the fire. Inspector Strand described Dugas as “visibly shaken by the fire,” but cooperative regarding Strand’s questions. Dugas was also formally interviewed by Eddy (the private fire investigator working on behalf of the insurers) and a special investigator directly employed by the insurers.

On October 24, the day after the fire, a Nashua police fingerprint specialist visited the scene. He took two bottles of lighter fluid for latent fingerprint analysis. One bottle did not reveal any prints, and the other revealed prints that were unusable due to their poor quality. Although a Nashua police detective visited Dugas at his house that same morning, the detective did not ask for consent to search the house, nor for the clothing Dugas had worn the previous night.

Detectives twice interviewed Dugas and tape-recorded the interviews. Each time, Dugas described how, on the night of the fire, he had locked up the store shortly after 10:00 PM and left the building with an employee. He described driving to the Nashua Mall to make his night deposit at about 10:20 or 10:25, then passing by" the store again on his way home about ten minutes later. He left home again at approximately 11:30 PM to pick up his daughter at the mall. ■ Just after- he picked her up, his wife called him on his cellular phone to tell him about the fire. He stated that he had not returned to the store after locking up.

In the second interview, which was conducted after the police had received the enhanced videotape, the police told Dugas that the video showed him exiting and then re-entering the store seven or eight minutes later, and confronted him with pictures taken from the video. Dugas insisted that he did not reenter the store after locking up for the night, and noted that the video might not have been from that night.

E. The Defense Investigation

Realizing that he was a suspect, Dugas hired attorney Ray Raimo to represent him. Raimo was an experienced criminal defense attorney and former prosecutor, though this was apparently his first arson case. On December 15, 1998 — less than two months after the fire — a state grand jury indicted Dugas on one count of arson.

Like most attorneys, Raimo had no training in fire investigation; his scientific background consists of “high school chemistry and physics.”³ As part of his pretrial investigation, Raimo spoke to Donaldson (the Nashua fire inspector), Boudreau (the state forensic chemist), Eddy (the private fire investigator), and Johnson (the electrical fire investigator), and found all of them.to be credible and formidable witnesses. He also walked through the scene, and, in his lay opinion, concluded that the physical evidence was consistent with the state’s expert testimony. He realized that the impression a layman might get from the scene was “it must be arson.”

Raimo did not, however, conclude that the “not arson” line of defense lacked merit. While he noted that “it seemed like the State had some fairly decent testimony,” he recognized that “[tjheir expert testimony ... conflicted a bit, and we thought we could use that.” He discussed the strengths of the state’s experts with fellow defense attorneys (in what he described as “casual discussions with friends”) and recalls discussing the idea that “we were going to need somebody ... who was at least as well qualified” as the state’s experts to testify for the defense. He “read some materials” on arson (he could not recall the exact sources) and admitted that he did not understand the basic terminology or techniques of arson investigation. Viewing the “not arson” defense as a challenge, Raimo also decided “that we weren’t going to just rest on the defense it wasn’t arson,” and chose to also pursue the defense that, if the fire was arson, another person caused it.

Raimo followed up leads and issues pertaining to the defense that another person was responsible for the fire. Raimo did not, however, conduct further investigation into the “not arson” defense. Despite his earlier consideration of the possibility, Rai-mo did not hire an expert to testify for the defense about the state’s evidence on arson. He did not consult an expert in preparing for his cross-examination of the state’s arson experts. Nor did he conduct the research required to understand the principles of arson investigation on his own. Instead, in the end, his investigation of the arson issue consisted of his interviews with the state’s experts, his reading of “some materials” about arson, his assessment of the experts’ testimony and the fire scene based on his own inadequate understanding of arson forensics, and the advice about cross-examination that he gleaned from his “casual discussions” with other defense attorneys.

F. The Trial

Early in the trial, just before the jury viewed the scene of the fire, Raimo told the jury that “what we’re going to be asking ourselves during this trial is how this fire started and why.... I want to just make it clear ... where the State is bringing in all of these witnesses ... we think they’re wrong.” However, the basis of Raimo’s challenge to the state’s evidence on arson was less than clear and remained unclear.

The state’s case against Dugas consisted of several expert witnesses who testified that the fire was an arson; a videotape that showed Dugas leaving the store and then returning a few minutes later; motive evidence regarding recent troubles in Du-gas’s life; and testimony of the wife of Paul “P.J” Kulas, the man who Dugas argued was the real perpetrator of the crime, to establish an alibi for Kulas.

The state’s strongest evidence was its expert testimony on arson; the balance of its evidence was relatively weak. The videotape linking Dugas to the scene of the fire was barely viewable and undated. The state’s motive evidence was problematic: Dugas’s marriage was strained; his wife had recently spent three weeks in an alcohol treatment center; he had some credit card debt; he was working long hours; and he had “butted heads” with his father because Dugas wanted to transform the store from a general grocery to a more specialized seafood store, which his father resisted. But the state never presented a theory of how burning the store could possibly benefit Dugas.

In establishing the defense’s case, Rai-mo confined his challenge of the state’s arson evidence to cross-examination of the state’s expert witnesses. Through cross-examination, he attempted to raise the possibility that the fire had started accidentally. Raimo pointed out some arguably questionable evidence handling procedures, and explored on cross-examination something to which Boudreau had testified on direct: that normal alkanes could be found in printing ink and medium petroleum distillates could be found in pesticides.⁴

However, the focus of Raimo’s cross-examination of the state’s experts was unclear, and many of the experts’ scientific conclusions went unchallenged.⁵ Raimo did not ask the kinds of questions that a trained fire investigator or forensic scientist would consider important. Instead, his questions amounted to an unfocused set of miscellaneous criticisms and evinced his lack of scientific knowledge. Despite his earlier statement to the jury that he believed that the state’s arson experts were wrong, Raimo presented no alternative theory of the fire.

Raimo did, however, challenge the state’s evidence regarding motive and Du-gas’s whereabouts prior to the fire. Rai-mo presented the testimony of Gordon Rehnborg, the attorney representing Du-gas’s parents in their insurance policy claim. Rehnborg testified that Dugas was not a beneficiary of the insurance policy and, other than some minimal salary protection, would not receive any money from insurance payments.

With regard to the undated and hazy videotape, Dugas himself testified and acknowledged that, contrary to what he had earlier told detectives, he had in fact returned to the store after his initial departure.⁶ Dugas explained that he had forgotten this brief re-entry when he first spoke to police, but that seeing the still images from the videotape had jogged his memory. Dugas maintained his innocence.

Raimo also presented evidence pursuant to the defense theory that another person may have caused the fire. Raimo called James Briggs, an expert in fingerprinting for the Nashua Police Department, to testify that the police had not dusted various doors to the building for fingerprints, presumably suggesting that the police did not adequately consider the possibility of another perpetrator. Raimo also identified other potential suspects, particularly focusing on P.J. Kulas, an ex-employee who had quit the store after Edgar accused him of theft. Raimo called Edgar to testify about his confrontation with Kulas. However, the trial court did not permit Raimo to cross-examine Kulas’s wife, or certain other witnesses, on the subject of Kulas’s own criminal trial, which Raimo argued would have shed light on why Kulas would have set fire to the store.⁷

The eight-day trial concluded on July 22, 1999. After three days of deliberations, the jury returned a guilty verdict. The court sentenced Dugas to five to ten years in prison.

II.

After the conviction was affirmed on appeal, State v. Dugas, 147 N.H. 62, 782 A.2d 888 (2001) (“Dugas I ”), Dugas moved for a new trial in the superior court before the same judge who had presided over the trial. See State v. Dugas, No. 98-S-1899 (N.H.Super.Ct. Aug. 12, 2002) {“Dugas II ”). Dugas argued, inter alia, that Raimo had provided ineffective assistance of counsel by inadequately pursuing the “not arson” defense, particularly due to the lack of expert consultation.⁸ Raimo testified at the hearing on the motion for a new trial and, in testimony that the judge found “remarkably candid,” admitted that he had no justifiable reason not to consult an expert since there was no risk in doing so and the money was available. Id. at 4. He confessed that he “didn’t give it enough consideration at the time,” was overly confident in the strength of his case (or, put differently, the weakness of the state’s) and admitted that he had placed too much reliance in his second theory of the case (that, if the fire was arson, someone other than Dugas was the perpetrator) and not enough emphasis on challenging whether it was even arson. See id. at 4-5.

In support of his motion, Dugas offered the report of Michael Higgins, a proposed forensic expert. Higgins identified several issues that' a forensic consultant could have flagged in pretrial investigation.⁹ Dugas also moved for discovery so that he could develop a more detailed expert analysis. A separate hearing was held on the motion for discovery, during which Higgins testified and addressed the issues raised in his report.

The state court analyzed the ineffective assistance of counsel claim under New Hampshire constitutional, precedent that mirrors the Sixth Amendment standards articulated in Strickland v. Washington, 466 U.S. 668, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984), and is “at least as protective” as Strickland. State v. Henderson, 141 N.H. 615, 689 A.2d 1336, 1339 (1997). Under both the federal and state standards, the petitioner must prove that counsel’s performance was deficient and that the deficient performance prejudiced the defense. The court found that Raimo had weighed his options and made “a strategic decision regarding one of two defenses, based on more than twenty years experience and his familiarity with the various players involved in this particular case.” Dugas II, slip op. at 6. In assessing Rai-mo’s decision to forego any consultation with an arson expert, the court focused on Raimo’s statement that because any potential expert would need the state’s permission to view the fire scene, the prosecution “might want to talk to [his] expert,” which could be damaging. The court concluded that Raimo had properly considered “the benefits and perils of hiring an expert” and therefore, his performance was not deficient. It did not reach the question of prejudice, and it denied both the motion for new trial and the motion for discovery. The state supreme court declined to grant review.

Dugas then filed this petition for a federal writ of habeas corpus directed to ap-pellee, the warden of the New Hampshire State Prison for Men. The warden moved for summary judgment.

The district court issued a thoughtful opinion holding that Raimo’s performance had in fact been deficient. Dugas v. Warden, No. 03-376-JD (D.N.H. May 21, 2004), slip op: at 9-12 (“Dugas III ”). The court concluded that Raimo’s decision to forego expert advice regarding the state’s forensic evidence and attack that evidence without the benefit of expert guidance was below the standard of competence expected of practicing criminal defense lawyers. The court noted that “Raimo admitted that he offered no explanation to support the defense that the fire was not arson” and that “he had no independent basis to investigate the fire and that he accepted the states’ experts’ opinions as to the cause.” Id. at 10-11. The court also observed that Raimo may have been concerned that a non-testifying defense expert might have to be made available to the state. The district court explained that, to the extent Raimo’s decision was based on this concern, it was an incorrect understanding of New Hampshire law. The court also found the state court’s reliance on that incorrect understanding to be unreasonable:

[B]oth Raimo and the state court noted a risk inherent in notifying the state that the defense had an expert witness, which would have been necessary to access the fire scene. Raimo apparently believed, based on his testimony, that he would have had to make his expert available to the state for questioning, whether or not he intended to call the expert to testify at trial, which would be detrimental if the expert agreed with the state’s experts. But see N.H.Super.Ct. R. 98; Mello v. DiPaulo, 295 F.3d 137, 146 & n. 8 (1st Cir.2002). The state court recognized that ordinarily the defense would not have to disclose an expert who was used only for consultation but concluded, without explanation, that the necessary disclosure in this case presented a “peril” to the defense. Given the lack of legal support or explanation, the state court’s conclusion was unreasonable that the defense would have faced a “peril” if Raimo consulted an expert.

Id. at 10-12.

The district court concluded that “the record in this case does not show that in both failing to consult an expert and maintaining the defense that the fire was not arson, Raimo conducted a thorough or reasonable investigation of the fire or exercised reasonable professional judgment.” Id. at 12. With respect to the heightened standard for habeas corpus under 28 U.S.C. § 2254, the district court further found that “[b]ased on the record, the state court’s conclusion that Raimo carefully investigated the case and made a tactical decision not to consult with or hire an expert is both an unreasonable determination of the facts and an unreasonable application of the federal standard.” Id. at 11. It further pointed out that the state court’s conclusion was “based at least in part on a misunderstanding of the law” and found it to be an “unreasonable application of the Strickland standard.” Id. at 12.

Since the state court had not reached the question of prejudice, the district court decided that issue de novo. It noted that “Higgins’s opinions ... might have been helpful to the defense.” Id. at 15. However, it found that, overall, they “do not undermine the court’s confidence in the outcome of the criminal trial.” Id. at 16. Thus, the court held that Dugas had suffered no prejudice from the deficient performance of his counsel, granted the warden’s motion for summary judgment, and denied the petition.

III.

An ineffective assistance of counsel claim requires the petitioner to demonstrate (1) that “counsel’s representation fell below an objective standard of reasonableness,” and (2) “a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different.” Strickland, 466 U.S. at 688, 694, 104 S.Ct. 2052. “A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Id. at 694, 104 S.Ct. 2052. “[B]oth the performance and prejudice components of the ineffectiveness inquiry are mixed questions of law and fact.” Id. at 698, 104 S.Ct. 2052.

Our standard of review is slightly atypical because the state court adjudicated Strickland’s performance prong but did not reach the prejudice prong. “Appellate review of the district court’s denial of habeas relief is de novo, but we accord deference" to the state court as to issues it actually decided.” Ellsworth v. Warden, 333 F.3d 1, 3-4 (1st Cir.2003) (en banc) (citations omitted). Since the state court concluded that Raimo’s performance was not deficient under Strickland, we defer to that conclusion unless it “involved an unreasonable application of[] clearly established Federal law.” 28 U.S.C. § 2254(d)(1). Since the state court never decided the question of prejudice, we review that issue de novo. See Ellsworth, 333 F.3d at 4; Fortini v. Murphy, 257 F.3d 39, 47 (1st Cir.2001) (“Here, the federal claim was never addressed by the state courts.... [The Antiterrorism and Effective Death Penalty Act] imposes a requirement of deference to state court decisions, but we can hardly defer to the state court on an issue that the state court did not address.”).

A. Performance

1. Strickland Analysis

We apply the Strickland standard to evaluate an attorney’s strategic choices in light of the investigation that led to those choices:

[Sjtrategic choices made after less than complete invéstigation are reasonable precisely to the extent that reasonable professional judgments support the limitations., on investigation. In other words, counsel has a duty to make reasonable investigations or to make a reasonable decision that makes particular investigations unnecessary. In any ineffectiveness case, a particular decision not to investigate must be directly assessed for reasonableness in all the circumstances, applying a heavy measure of deference to counsel’s judgments.

466 U.S. at 690-91, 104 S.Ct. 2052. Since “the state court identifie[d] the correct governing legal principle [i.e., the Strickland performance test] from [the Supreme Court’s] decisions,” its conclusion that Rai-mo’s performance was not deficient constitutes an “unreasonable application” of that law only if the court “unreasonably applied] that principle to the facts of the prisoner’s case.” Williams v. Taylor, 529 U.S. 362, 413, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000).

We have no doubt that Raimo is an experienced attorney. But our inquiry is not whether Raimo is generally competent.¹⁰ Rather, we must decide whether, given the particular facts of this case, he fell below the constitutional standard of competence by inadequately investigating the “not arson” defense that he decided to pursue as part of his overall defense. We emphasize that the question is not whether Raimo should have presented a “not arson” defense. Instead, we focus on whether the investigation supporting his pursuit of the defense was itself reasonable. Cf. Wiggins v. Smith, 539 U.S. 510, 523, 123 S.Ct. 2527, 156 L.Ed.2d 471 (2003) (“[O]ur principal concern ... is not whether counsel should have presented a mitigation case. Rather, we focus on whether the investigation supporting counsel’s decision not to introduce mitigating evidence of [the defendant’s] background was itself reasonable.”) (emphasis in the original).¹¹

As noted, the record demonstrates that Raimo’s investigation consisted of his own visual assessment of the fire scene, his conversations with the state’s experts, some limited reading, and his conversations with other defense attorneys after work. Based on this investigation, Raimo mounted a “not arson” defense. He did not consult an expert in arson investigation or learn how to effectively use the terminology and techniques of arson investigation from his own research. In essence, Raimo “abandoned [his] investigation ... after having acquired only rudimentary knowledge of [the issues] from a narrow set of sources.” Wiggins, 539 U.S. at 524, 123 S.Ct. 2527 (finding counsel’s failure to adequately investigate mitigation evidence deficient).

In assessing the constitutional significance of Raimo’s investigation of the “not arson” defense, we recognize that “reasonably diligent counsel may draw a line when they have good reason to think further investigation would be a waste.” Rompilla v. Beard, — U.S. —, —, 125 S.Ct. 2456, 2463, 162 L.Ed.2d 360 (2005) (finding attorneys’ failure to investigate material they knew that the prosecution would rely on was ineffective). We also recognize that reasonably diligent counsel are not always required to consult an expert as part of pretrial investigation in a case involving the use of expert witnesses by the state. “A defendant’s lawyer does not have a duty in every case to consult experts even if the government is proposing to put on expert witnesses. There may be no reason to question the validity of the government’s proposed evidence or the evidence may be so weak that it can be demolished on cross-examination.” Miller v. Anderson, 255 F.3d 455, 459 (7th Cir.2001) (Posner, J.) (citations omitted) (finding defense counsel’s failure to consult scientific experts constituted deficient performance where defense was that defendant was not present at scene of crime), remand order modified by stipulation, 268 F.3d 485 (7th Cir.2001).

However, for multiple reasons, Raimo’s failure to thoroughly investigate the “not arson” defense in this case was constitutionally deficient. First, challenging the state’s arson case was critical to Dugas’s defense. Other than creating reasonable doubt that the fire was not arson, Dugas’s only defense was that “someone else did it” — a defense that is often difficult to mount and fraught with evidentiary problems, as Raimo’s attempt to cross-examine witnesses about the main alternative suspect demonstrates. Much of Dugas’s defense, therefore, depended on Raimo’s ability to convince the jurors that the State’s experts might be wrong in concluding that the fire was arson. Indeed, he told the jurors as much at the outset of the trial. Given this representation, it is unfathomable that he' did not undertake' a more thorough investigation into such a crucial aspect of the defense.¹²

Second, the arson evidence was the cornerstone of the state’s case. The state had little evidence beyond it. Raimo understood this. In fact, he was “shocked” at the outcome of the trial because he perceived the state’s motive evidence to be so weak. Yet he failed to conduct a thorough investigation into the most crucial aspect of the state’s case.¹³

Third, Raimo acknowledged that he lacked any knowledge of arson investigation and had never tried an arson case. He understood that he needed expert assistance to understand and challenge the state’s case.¹⁴ Yet he decided to accept the characterization of the fire scene by the state’s experts rather than conduct an independent investigation.¹⁵

Fourth, Raimo knew and admitted that a layperson would be likely to view the scene as an arson. He understood that expert testimony or a well-informed cross-examination on the scientific conclusions of the state’s experts would be necessary to shake the jurors’ views that they were dealing with an arson scene. Yet he did not prepare himself for that task.¹⁶

Fifth, Raimo conceded that he had at least some reason to believe that there were problems with the state’s arson case. He noted inconsistencies in the testimony of the state’s arson experts and recalls talking to colleagues about the need to hire a well-qualified expert to challenge the state’s arson case. “In assessing the reasonableness of an attorney’s investigation ... a court must consider ... whether the known evidence would lead a reasonable attorney to investigate further.... Strickland does not establish that a cursory investigation automatically justifies a tactical decision-” Wiggins, 539 U.S. at 527, 123 S.Ct. 2527. Raimo’s failure to follow through with his investigation of the arson issues resulted in a feeble defense that was contrary to his promise to the jury at the outset of the trial that “what we’re going to be asking ourselves during this trial is how this fire started and why.... I want to just make it clear ... where the State is bringing in all of these witnesses ... we think they’re wrong.”¹⁷ Raimo never gave the jury any reason to think that the state’s experts were wrong.¹⁸

Taken together, these circumstances— the importance of challenging the state’s arson case to Dugas’s defense, the unfulfilled promise to the jury to do so, the crucial role of the arson evidence to the state’s case, Raimo’s lack of knowledge and experience in arson investigation and arson cases, and Raimo’s initial awareness of problems with the state’s case — demonstrate the inescapable need for expert consultation in this case. Yet for reasons he was unable to explain, Raimo did not consult an arson expert as part of his investigation.

The state argues that Raimo’s inexplicable choice to curb his investigation was based on an informed, tactical decision. We disagree. We cannot conclude that, as the state suggests, Raimo’s failure to thoroughly investigate the “not arson” defense was justified by a tactical decision to pursue the defense that another person caused the fire or his overconfidence in that alternative defense. “A tactical decision to pursue one defense does not excuse failure to present another defense that ‘would bolster rather than detract from [the primary defense].’ ” Foster, 9 F.3d at 726 (finding that counsel failed to adequately pursue impotency defense in rape case).¹⁹ The “not arson” defense did not conflict with the “other perpetrator” defense. They offered alternative grounds for reasonable doubt. Pursuing one defense did not require Raimo to avoid pursuing the other.

Nor can we find that Raimo’s failure to consult an expert or educate himself on the techniques of defending an arson case is excusable based on a “tactical” decision to pursue uninformed cross-examination of the state’s experts. Where defense counsel’s “extensive cross-examination of the government’s expert strongly suggests that they consulted an expert,” Ruiz v. United States, 221 F.Supp.2d 66, 82 (D.Mass.2002), aff'd, 339 F.3d 39 (1st Cir.2003), or where the state’s “evidence may be so weak that it can be demolished on cross-examination,” Miller, 255 F.3d at 459, defense counsel’s investigation and pursuit of a defense may be deemed sufficient. However, Raimo’s cross-examination demonstrated a clear lack of understanding of arson investigation and the principles invoked by the state’s many expert witnesses. Without having consulted an expert or researched the scientific principles more thoroughly, Raimo was hopelessly unprepared to challenge the state’s expert witnesses.²⁰

In short, Raimo’s failure to thoroughly investigate the “not arson” defense and seek expert assistance cannot be classified as a conscious, reasonably informed tactical decision. This is particularly so because Raimo recognized that there were possible problems with the testimony of the state’s experts, and he told the jurors he would demonstrate those problems. He just never prepared himself to do the job. Under these circumstances, we must conclude that Raimo’s investigation of the “not arson” defense was deficient under the Strickland standard.²¹

2. Section 2254 Analysis

The state court’s contrary conclusion largely rested on the facts that (1) Raimo is an experienced attorney; (2) he spoke to other experienced attorneys; and (3) if he had consulted an expert, the state would know about it (though not what the expert determined) because the expert would need permission to visit- the fire scene. See Dugas II, slip. op. at 5-6. As the district court aptly explained, see Du-gas III, slip op. at 11-12, the state court’s decision rested on an unreasonable determination of fact and an unreasonable application of Strickland to the facts.

The argument that Raimo’s investigation was reasonable because Raimo is an experienced attorney and spoke to other experienced attorneys takes us only so far in a Strickland analysis. The ability to challenge the state’s experts was a crucial part of the important “not arson” defense. Rai-mo’s own background and his “casual conversations with friends” provided him with no basis for critically assessing the scientific conclusions and methodology of the state’s experts. Raimo’s attempt to impeach the state’s arson case at trial was largely ineffective. His investigation into the arson case was neither careful nor thorough. The state court’s contrary conclusion was an unreasonable application of Strickland.

In support of its ruling, the state court found that Raimo had “considered the benefits and perils of hiring an expert.” Du-gas II, slip. op. at 5. According to Raimo’s testimony in state postconviction proceedings, no such balancing of “benefits and perils” ever took place.²² Even, however, assuming that such a benefit-peril balancing had taken place, the state court was only able to identify one such “peril”:

There would be no obligation to disclose an expert to the State unless the expert would be called to testify at trial and, therefore, there was no patent danger in consulting an expert. Consultation with an expert may have provided Raimo with more knowledge of fires and arson so that he could have more effectively challenged the State’s evidence of arson. Unfortunately, there was no way to get an expert into the fire scene to evaluate it or collect samples without notifying the State.

Dugas II, slip. op. at 5 (emphasis added). The district court, in evaluating this logic, noted that the state court “concluded, without explanation, that the necessary disclosure in this case presented a ‘peril’ to the defense. Given the lack of legal support or explanation, the state court’s conclusion was unreasonable that the defense would have faced a ‘peril’ if Raimo consulted an expert.” Dugas III, slip op. at 12.

We agree with the district court. The peril, if any, associated with the state’s learning that the defense had hired an expert for pretrial consultation was minimal, approaching trivial. Perhaps the state might have obtained a sixth expert witness in addition to Donaldson, Strand, Eddy, Johnson, and Boudreau. Perhaps it would have prepared its experts for more vigorous . cross-examination. But, in -the context of this case, these “perils” could not have outweighed the benefits of consulting an- expert under any rational calculus. Raimo’s feeble pursuit of an important defense through a cross-examination that was entirely uninformed by any scientific knowledge further demonstrates that any such balancing (again assuming that one took place) led to ah unreasonable result.²³ •

The state court’s contrary conclusion is thus incorrect by “clear and convincing evidence,” 28 U.S.C. § 2254(e)(1), and therefore reflects “an unreasonable determination of the facts in light of the evidence presented in the State court proceeding,” id. § 2254(d)(2). Moreover, “[t]his partial reliance on an erroneous factual finding ... highlights the unreasonableness of the state court’s decision.” Wiggins, 539 U.S. at 528, 123 S.Ct. 2527. The state court’s conclusion that counsel made a tactical decision to forego expert consultation is inconsistent with how counsel actually proceeded and appears to be “more a post-hoc rationalization of counsel’s conduct than an accurate description.” Id. at 526-27, 123 S.Ct. 2527.²⁴ In short, the state court based its conclusion on a supposed benefit-peril balancing which, assuming it had taken place, could not rationally have led to the conclusion that the state court drew. Accordingly, the state court’s conclusion was an unreasonable application of Strickland within the meaning of § 2254.

B. Prejudice

The knottier question is whether Dugas was prejudiced by Raimo’s failure. Prejudice occurs when “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different. A reasonable probability is a probability sufficient to undermine confidence in the outcome.” Strickland, 466 U.S. at 694, 104 S.Ct. 2052.

In submitted documents and hearing testimony, Higgins alleges several defects in the state’s case which, if he had been consulted before trial, he would have highlighted for further investigation.²⁵ One of these issues — concerning flaws in the state forensic chemist’s analysis — gives us considerable pause. At least two of the remaining major issues²⁶ — concerning evidence of ventilation and certain smoke shadows — demonstrate other unexplored lines of questioning, which would have ben-efitted from expert consultation. While the flaws in the chemist’s analysis alone probably provide enough evidence of prejudice to survive summary judgment, “Strickland clearly allows the court to consider the cumulative effect of counsel’s errors in determining whether a defendant was prejudiced.” Kubat v. Thieret, 867 F.2d 351, 370 (7th Cir.1989). We thus find that these three issues, taken together, with particular emphasis on the potential flaws in the state forensic chemist’s analysis, are sufficient to raise a genuine dispute of material fact concerning prejudice.

We acknowledge that, even if a proper expert analysis would completely undermine the state’s chemical analysis evidence, Dugas would not be home free. Dugas would still have to contend with the fire investigators’ conclusion (independent of any chemical analysis) that the fire was intentionally set, although his cross-examination of these experts would be considerably stronger given expert consultation. Dugas also would have to address the issue of the video and his conflicting statements. While the tape was largely incomprehensible, Dugas testified at trial that he did indeed reenter the store after closing. That admission would seem damaging in light of his earlier denials.²⁷

However, such weaknesses were not confined to Dugas’s defense alone. The state’s evidence of motive was a potpourri of generalities about the stresses of marriage and business, without any theory of how burning the store could possibly benefit Dugas, financially or otherwise. Despite the deficiencies in the defense, it took the jury three days of deliberations to reach a guilty verdict after only eight days of trial. The length of jury deliberations can be one factor in determining how close the jury viewed the case to be.²⁸ In a close case, the failure of defense counsel to present certain evidence or effectively challenge the state’s evidence on important issues can be particularly prejudicial.

This case lay on a knife edge, and it would not have taken much to sway at least some jurors towards acquittal. Accordingly, the threshold for prejudice is comparatively low because less would be needed to unsettle a rational jury. Strickland, 466 U.S. at 696, 104 S.Ct. 2052 (“[A] verdict or conclusion only weakly supported by the record is more likely to have been affected by errors than one with overwhelming record support.”).

Boudreau (the chemist) testified simply that some of the samples contained medium petroleum distillates and normal al-kanes. He conceded that normal alkanes could be present for benign reasons:

Q. Now, do normal alkanes appear as a matter of course on paper, or in the paper-making process?

A. I can’t say for all paper, but for some paper they may appear as being a solvent for — for some of the printing inks.

Q. What about — could there be a benign or an innocent reason for the presence of normal alkanes present on paper even before burning?

A. Yes.

On cross-examination, he admitted that medium petroleum distillates might be a component of pesticide sprays:

Q. What does J.P. Chemical use in their pesticides in — when they spray in a commercial building?

A. I really wouldn’t be able to tell you.

Q. Would you be surprised to find medium petroleum distillates as a dispersant in pesticides?

A. No, I wouldn’t. I have read that sometimes — that the carrier is used — is a medium petroleum distillate used for pesticides, and I have checked — I have tested some samples myself, or I found a medium petroleum distillate. It may also be other petroleum distillates such as a heavier petroleum distillate might be used.

Q. So you wouldn’t be at all surprised to have — to find that bug spray is an ignitable liquid?

A. Correct. That bug spray may contain an ignitable liquid.

Raimo was able to proceed thus far without the benefit of expert advice. But Higgins has identified a potentially more serious flaw in Boudreau’s analysis. The applicable national standard from the American Society for Testing and Materials (ASTM) requires an ignitable liquid to be not merely detected, but specifically identified,²⁹ As Higgins explains:

[The ASTM standard] requires the exact identification of flammable liquids (accel-erants). [The state’s experts’] report ... gives a conclusion that they detected medium petroleum distillates.... As you can see they said they detected but nowhere does it say they identified. In other words, their conclusion says that they might have a flammable liquid, but what they fail to say is they may not have one.

The [standard] requires the exact identification of the flammable liquid and that standard chromatograms of that acceler-. ant should be maintained in the case folder.

Thus, Dugas alleges, a competent cross-examination of Boudreau would have challenged his failure to identify the specific accelerant, in apparent violation of ASTM standards. Put differently, it is arguably improper even to use the term “acceler-ant” unless one has been specifically identified.

The district court acknowledged that “Higgins’ opinions about the state’s experts’ failure to follow standards for identifying accelerants ... might have been helpful to the defense.” Dugas III, slip op. at 15. However, it did not find this to constitute prejudice, since “Dugas has not shown that the state’s experts were required to follow the standards Higgins cites. It also does not appear that the irregularities he charges would necessarily have been material to the state’s experts’ opinions.” Id.

It appears that Higgins’s laconic explanation of the ASTM standards issue led to a misunderstanding on the part of the district court. We do not understand Higgins’s critique to be merely that the lab did not follow ASTM standards; indeed, Du-gas does not suggest that it was required to do so. Rather, we understand Higgins’s critique — that the state’s experts “say[] that they might have a flammable liquid, but what they fail to say is they may not have one” — to mean that the results themselves are questionable. Specifically, Bou-dreau’s analysis — which, due to Raimo’s failure to consult an expert, was the only analysis applied to the samples — was designed not to distinguish between innocuous and nefarious explanations for the presence of ignitable liquids,³⁰

As Boudreau testified, his analysis is designed to answer one question only: “if any ignitable liquids are present in the sample.” His laboratory technique consists of five basic steps. First, an activated carbon strip is inserted into a small hole in the container holding the sample. Second, the container is heated in an oven “so that if there’s any ignitable liquids present in the debris it [sic] will rise to the top of the can .... [and] then be chemically bonded to the carbon strip.” Third, the strip is removed and bathed in a solvent that removes any ignitable liquids. Fourth, a gas chromatograph is used to separate the components of the ignitable liquids. Finally, a mass spectrometer is used to analyze each component.

As the preceding discussion shows, Boudreau’s laboratory technique is designed to detect ignitable liquids, and only ignitable liquids. He testified, and we have no reason to doubt, that the method he described is a nationally recognized procedure “for showing that [an] ignitable liquid is present.” However, this method has two important limitations. First, it cannot distinguish between different chemicals in the same class, e.g., between different types of medium petroleum distillate. See People v. Sykes, 341 Ill.App.3d 950, 276 Ill.Dec. 57, 793 N.E.2d 816, 824 (2003) (“A petroleum distillate can be identified, but the specific product from which it came cannot be identified.”); Commonwealth v. Scott, No. 93-12100-1-3, 1995 WL 808883, at *6, 3 Mass. L. Rep. 309, 1995 Mass.Super. LEXIS 847, at *16 (Mass.Super.Ct. Feb. 15, 1995) (“The conclusion that a medium petroleum distillate was detected is a elas-sification not an identification.”), aff'd sub nom. Commonwealth v. McQuade, 46 Mass.App.Ct. 827, 710 N.E.2d 996, 1002 (1999).

Perhaps more importantly, Boudreau’s method does not and cannot analyze any other component of the sample besides the ignitable liquids. For example, pesticides may contain ignitable liquids as solvents:³¹

Q.... The — the bug spray is the stuff that kills the bugs, right?

A. Yes, the compound that’s toxic to— to the insects.

Q. Okay. And the — but there is something that they use to disperse the toxic agent to the bugs?

A. Yes, basically to get the insecticide to dissolve, sometimes they would use a hydrocarbon, a petroleum distillate. And it may be a medium petroleum distillate in some cases.

In other words, if someone sprays onto a stack of papers a pesticide using a medium petroleum distillate for a solvent, both the solvent and the toxic active ingredient might be present on the sample.³² But Boudreau’s method of analysis — designed only to detect ignitable liquids — would simply report the presence of a medium petroleum distillate. Similarly, if a printing ink contained normal alkanes as a solvent for various other ink ingredients, Boudreau would dutifully report that normal alkanes were present in the sample — and nothing more.

We do not disparage Boudreau’s science. The state is interested in the answer to a particular question — does the sample contain ignitable liquids? — and Boudreau answers that question. However, the defense is (or ought to be) interested in a different question: does the sample contain compounds associated with pesticides, alkane-based inks, or other substances that could suggest that the ignitable liquids were present for benign reasons? Put differently, Boudreau’s technique can only inculpate, never exculpate.

If Raimo had consulted an arson investigator before trial, that investigator would have undoubtedly urged Raimo to demand the debris samples that Boudreau received — not to mention those that Eddy collected, but which the state seized — and submit them for independent testing. Indeed, Higgins regularly performs such analysis as a defense expert. See, e.g., State v. Dowdle, 148 N.H. 345, 807 A.2d 1237, 1243 (2002) (noting that Higgins was retained by the defense and obtained Bou-dreau’s samples and charcoal testing strips for independent testing); Scott, 1995 WL 808883 at *2, 1995 Mass.Super. LEXIS 847, at *6 (where the state criminal laboratory accidentally destroyed the fire debris samples, Higgins testified, as a defense expert, as to how he could have independently tested the samples to determine whether certain petroleum products were accelerants or natural components of fire). Higgins states that, if he had been retained here, he would have done the same. See also William W. Turner, Investigating Particular Crimes, 2 Am.Jur. Trials 171 § II.8 (“[CJounsel for an accused arsonist ... should, where possible, use discovery methods to determine the evidence on which the prosecution hinges its case.... Findings based on laboratory analysis or expert opinion should be submitted to his own experts to determine their validity.”). Such testing could both verify Boudreau’s findings, and, more importantly, determine whether the sample also contained other compounds (e.g., an insecticide active ingredient, or components associated specifically with alkane-based inks) suggestive of an innocent explanation for the ignitable liquids. If the independent test had revealed such compounds, it might have severely undermined Boudreau’s analysis by demonstrating an innocent explanation for his findings. By failing to secure an independent forensic analysis, Raimo eliminated the possibility of obtaining exculpating chemical evidence.³³

Of course, even if Boudreau’s chemical analysis could be completely undermined by independent examination, Dugas would still have to contend with the opinions of three separate fire investigators, each of whom concluded, before the chemical analysis was conducted, that the fire was intentional.³⁴ However, Higgins points to two other issues that raise questions about the completeness of the experts’ fire investigation and their conclusions.

First, Higgins notes that smoke shadows on the floor indicate that boxes were present near the top of the staircase to the basement at the time of the fire. Those boxes were apparently moved by the state at some point before the jury viewed the scene. Higgins explains that the smoke shadows demonstrate that the stacked boxes permitted, at most, four inches of clearance to get through to the basement door. Arguably, the arrangement of those boxes would have obstructed passage, and it would have been difficult for a person to cross the first floor, descend to the basement, light the fire, reascend to the first floor, and arrange the boxes, in the time shown on the videotape. The district court opined that this theory “is interesting but is not corroborated by any other evidence at trial. Most tellingly, Dugas did not testify that the boxes would have blocked his access to the basement.” Du-gas III, slip op. at 15-16. However, if Raimo had consulted an expert, he would have known the importance of the smoke shadows and the placement of the boxes. Raimo could have elicited testimony from Dugas about the possible placement of the boxes by the basement door. More importantly, Raimo could have presented evidence of the smoke shadows to the jury and questioned the fire investigators about how the obstruction of the passage affected their conclusion that the fire was intentionally set during the time frame they suggested.

Second, Higgins asserts that evidence from the fire scene demonstrates that the state’s investigation failed to account for the presence of an air vent and a half-open split door. He argues that the state’s entire theory of an oxygen-starved fire is incorrect, and that the fire was actually oxygen-fueled and fast-burning due to the ventilation. He suggests that, if the fire was fast-burning and discovered around midnight, it must have started later than the fire investigators had surmised.

The district court understandably found that Higgins’ theory on this issue would be difficult to advance. See Dugas III, slip op. at 14. The firefighters at the scene credibly testified that the basement contained especially thick smoke and unusually high carbon monoxide levels, which suggested that the fire had been oxygen-starved. The clock apparently stopped at 10:44, but firefighters arrived close to midnight to discover the smoke-filled room, with little fire damage on the side of the basement through which the fire was supposedly ventilated or on the upper floors.

However, the value of Higgins’ evidence is not just that it supports an alternate, albeit questionable, theory of the fire’s progression. Higgins also argues that the state’s experts’ failure to consider the air vent and their incorrect conclusion that the door to the basement was closed were “an indication of a poor or incomplete scene investigation.” Regardless of the strength of the alternative theory, Raimo could have used the actual evidence that Higgins found (the presence of the air vent and the smoke stains and melted items indicating that the door was partially open during the fire) in his cross-examination of the fee investigators.

In short, if Raimo had been advised by an expert, his cross-examination of the fee investigators could have been far more pointed. All three investigators testified that they concluded that the fire was arson because they had eliminated other causes. Such testimony reflects a recognized technique of fire cause and origin investigation, and we do not minimize its usefulness. But it also has its limits, and Raimo— unaware of those limits — did not press the investigators effectively. Indeed, based on his inadequate investigation, he had concluded before the trial began that the state’s experts’ conclusions were basically sound.

This case is similar in that respect to United States v. Correia, No. 00-10246, 2002 WL 31052766, 2002 U.S. Dist. LEXIS 17218 (D.Mass. Sept. 13, 2002), aff'd, 77 Fed.Appx. 12 (1st Cir.2003).³⁵ In Correia, as here, “there were no eye witnesses nor any other direct evidence as to the cause of the fire .... [and] the fire investigator ] testified that he determined the cause of fire by process of elimination.” Id. at *2, 2002 U.S. Dist. LEXIS 17218 at *9. Three other government expert witnesses concluded that the fire was not electrical, and “[biased entirely on their elimination of other causes, ... deduced that defendant must have set the fire.” Id. at *3, 2002 U.S. Dist. LEXIS 17218 at *10. However, defense counsel did not seize upon the primary investigator’s admission “that the cause and origin of 20 percent of all fires remain unknown.” Id. The district court found constitutionally ineffective assistance of counsel:

As an objective matter, [the state expert’s] testimony that the cause and origin of 20 percent of all fires remain unknown, coupled with the dearth of direct evidence of arson by defendant, would have permitted the jury to find reasonable doubt as to defendant’s culpability .... Counsel’s fundamentally flawed trial strategy ignored the one easily available and logical defense to causation and in essence forced the jury to accept the result of the government’s tenuous process of elimination. Such conduct fell below the standard for professionally competent assistance that is safeguarded by the Sixth Amendment.

Id. at *3, 2002 U.S. Dist. LEXIS 17218 at *10-11 (citation and quotation marks omitted). The district court granted the defendant’s motion for a new trial. We affirmed and noted that “given the absence of direct evidence of arson and the fact that the government’s experts were accordingly compelled to attempt to prove the government’s case by eliminating other causes, the frequency of cases of undetermined cause was an obvious theme to have stressed.” Correia, 77 Fed.Appx. at 15.

Here, too, an uninformed cross-examination of the state’s fire experts leads to, at minimum, serious concerns about the effect on the jury verdict of Raimo’s failure to consult an expert. Put in Strickland terms, in a case as close as this, the likelihood of a more effective cross-examination with the use of an expert, and the effect of such cross-examination on the jurors, generates concerns that may reach “a probability sufficient to undermine confidence in the outcome.” Strickland, 466 U.S. at 694, 104 S.Ct. 2052.

IV.

We perceive a distinct possibility that, if Raimo had consulted an arson expert, the outcome of the trial would have been different. We cannot say, as a matter of law, that this possibility does not rise to the level of “a probability sufficient to undermine confidence in the outcome,” Strickland, 466 U.S. at 694, 104 S.Ct. 2052. Neither can we say, as a matter of law, that it does rise to such a level. We simply hold on this record that there is a genuine factual issue precluding summary judgment. Cf. Scott, 1995 WL 808883 at *6, *7, *8, 3 Mass. L. Rep. 309, at *16, *21, *23-24.³⁶ While Higgins’s preliminary report is hardly a coup de grace, the state presented a thin circumstantial case, and the jury only agreed to convict after three days of deliberations.

Fortunately, “[h]abeas doctrine is flexible enough for us to condition a grant of the writ on the outcome of a further inquiry” into the nature of the evidence. Ells-worth, 333 F.3d at 6. Accordingly, we vacate the grant of summary judgment and remand this case to the district court for limited further proceedings.

The district court should order the state to produce all Are debris samples, carbon strips, laboratory reports, chromatograms, and other evidence in its possession that was collected from the fire scene or was the result of analysis of such material, whether obtained by local, state, or private fire investigators. The court should give Higgins an opportunity to analyze this evidence to identify precisely (i.e., not merely detect, but name) which flammable substances were present in the debris, and to determine whether the ignitable liquids found and described by Boudreau were innocuous, as opposed to deliberately applied accelerants. The state should have an opportunity to respond to Higgins’s analysis. The court should consider whether Higgins’s analysis would meaningfully challenge the state’s view of the evidence. If Raimo’s failure to challenge the state’s chemical analysis and to cross-examine the fire investigators on the evidence of smoke shadows and partial ventilation demonstrates that “there is a reasonable probability that, but for counsel’s unprofessional errors, the result of the proceeding would have been different,” Strickland, 466 U.S. at 694, 104 S.Ct. 2052, then the court should find that the defendant was prejudiced by Raimo’s deficient performance and should grant the petition.³⁷

While we are remanding for a specific determination of whether Higgins’s chemical analysis and the evidence of smoke shadows and ventilation could establish prejudice within the meaning of Strickland, the district court has some flexibility in both the procedure and the scope of further proceedings. For example, the district court may choose to receive written submissions only, or it may choose to conduct a hearing. It can, if it chooses, seek additional testimony from the experts that the state presented at trial, subject, of course, to cross-examination by Dugas’s counsel. Finally, if some or all of the relevant evidence has been destroyed or degraded, or if the district court sees other appropriate reasons, the district court has the discretion to expand the inquiry to encompass any other prejudice issues. While we seek the answer to a specific question — is there a reasonable probability that Higgins’s analysis of the chemical evidence and the use of evidence of smoke shadows and ventilation in cross-examining the fire investigators could have affected the outcome of the trial? — we do not constrain the district court in how it approaches that question or preclude it from exploring other factual questions that it may find necessary to answer the Strickland prejudice inquiry.

We emphasize one final point. The case for prejudice here is close; we do not conclude that there was prejudice, but only that, in the circumstances of this appeal, Dugas has raised sufficient doubts about the outcome to avoid summary judgment. The peculiar circumstances of this appeal play a significant role in that decision. In most § 2254 petitions alleging ineffective assistance of counsel, the federal court must defer to the state court’s conclusion regarding prejudice unless it “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law,” 28 U.S.C. § 2254(d)(1). Here, however, the state court never reached the question of prejudice, and we must review the issue de novo. See Ellsworth, 333 F.3d at 4; Fortini, 257 F.3d at 47. Consequently, with respect to the' prejudice issue, we are effectively in the same position as if we were considering a § 2255 petition after a federal criminal conviction. We conclude only that, on the facts of this closely contested case, in the unusual context wherein we review a § 2254 petition without deference to a state court decision on the issue presented, more evidentiary development is necessary.

Vacated and remanded for further proceedings consistent with this opinion. Each party is to bear its own costs on appeal.

1 . We use the term "Dugas” to refer to petitioner-appellant Peter Dugas, and specify "Edgar” when we mean his father.

2 . Medium petroleum distillates are typically found in paint thinners, mineral spirits, and charcoal lighter fluid. Normal alkanes, which are further refinements of medium or heavy petroleum distillates, are typically found in solvents or lamp oils.

3 . As he later stated, "I wouldn’t know a hydrocarbon if I fell over one.”

4 . Raimo intimated, but no witness ever directly testified, that pesticides had recently been applied in the basement.

5 . For example, during his cross-examination of Boudreau, Raimo largely conceded that the results of Boudreau's analysis were correct, even suggesting that any confusion he might have in looking at Boudreau's results would be due to his own lack of knowledge as a layman:

"Q Well, I take it that the results are absolutely clear if you know what you're looking for and you know how to drive one of those mass spectrometers?”

A Yes, with — with training- — it obviously takes training, but if I were to show my results to a person in another lab doing forensic arson analysis, I'm confident they would agree with me.

Q But if I got a C-minus in high school chemistry, and I haven't done anything since then, and went up to your lab, and started messing around with your spectrograph and your gas — and your gas chroma-tograph, it probably wouldn't be real apparent to me?

"A No, it probably wouldn’t.”

6 .Dugas testified that he had almost driven out of the parking lot when he remembered that he might have left a cash drawer on a desk in the office, and so returned to the store to put the drawer in its proper place; while he was there, he did some miscellaneous tasks in the kitchen. He also stated that, as he armed the alarm system when he was about to leave, he noted that a light was on; he "scooted over” to turn that light off, and then finally left for the bank.

7 .Raimo proposed to cross-examine Julie Ku-las, P.J. Kulas’s wife, about "whether P.J. had been on trial for rape on the day of the fire, was worried about getting a long prison sentence, was drinking heavily, was upset with the Dugas’ for not providing him with á good job which would have kept him out of trouble, and whether he realized that his time for settling old scores was running out.” The state objected, arguing that the testimony about the rape trial was irrelevant and prejudicial, and Raimo offered to omit the reference to the nature of the trial in his questioning. The trial court rejected the entire line of questioning as more prejudicial than probative. Raimo sought to cross-examine the investigating police detectives regarding their knowledge of Kulas’s legal problems, but the state objected and the trial court again found that the questioning would be more prejudicial than probative. On appeal, the state court affirmed the trial court’s determinations.

8 . Dugas also advanced other claims that are not before us.

9 . See infra Part III.B.

10 . Similarly, Raimo's subjective impression that his representation was inadequate plays no role in our decision. The Strickland test requires us to assess whether "counsel's representation fell below an objective standard of reasonableness." 466 U.S. at 688, 104 S.Ct. 2052 (emphasis added). If "counsel performed as a competent lawyer would, his ... detailed subjective reasoning [would be] beside the point." Cofske v. United States, 290 F.3d 437, 444 (1st Cir.2002).

11 . Even though many post-Strickland ineffective assistance cases, including Wiggins, involve capital offenses, the Strickland standard applies to both capital cases and noncapital cases involving a sentence of imprisonment. See Glover v. United States, 531 U.S. 198, 203, 121 S.Ct. 696, 148 L.Ed.2d 604 (2001) (explaining, in noncapital case on ineffective assistance of counsel, that "any amount of actual jail time has Sixth Amendment significance”).

12 . See Rompilla, 125 S.Ct. at 2469-70 (O’Connor, concurring) (finding ineffective assistance and noting defense counsel failed to properly investigate an issue that "threatened to eviscerate one of the defense’s primary [ ] arguments”); Miller, 255 F.3d at 459 (finding "no excuse for the lawyer’s failure to consult experts on hair, DNA, treadmarks, and footprints” when such factors went to critical defense argument that defendant was not at the scene of the crime); Foster v. Lockhart, 9 F.3d 722, 726 (8th Cir.1993) (noting that where the attorney offered a two-pronged defense to rape and evidentiary weaknesses existed in one prong, "[t]he vulnerability of the alibi evidence shows the unreasonableness of the attorney's failure to investigate further and present the impotency defense”); State v. Hicks, 195 Wis.2d 620, 536 N.W.2d 487, 492 (1995) (noting that "Hicks’s trial counsel understood that the hair samples were going to be a major issue in the case. But he has provided no reasoned basis for failing to pursue a testing process that he knew had the potential to provide exculpatory evidence on this major issue.”), aff'd on other grounds, 202 Wis.2d 150, 549 N.W.2d 435 (1996).

13 . See Driscoll v. Delo, 71 F.3d 701, 709 (8th Cir.1995) (finding that “defense counsel’s failures to prepare for the introduction of [state's scientific evidence]” and "to subject the state's theories to the rigors of adversarial testing” involving “an issue of the utmost importance” in the state’s case constituted ineffective assistance); Troedel v. Wainwright, 667 F.Supp. 1456, 1461 (S.D.Fla.1986) (finding ineffective assistance where counsel "neither deposed ... the State's expert witness [on gunpowder residue], nor bothered to consult with an expert in the field prior to trial” despite the fact that counsel "knew pretrial this issue would be critical”), aff'd, 828 F.2d 670 (11th Cir.1987).

14 . See United States v. Tucker, 716 F.2d 576, 581 (9th Cir.1983) (noting that "it should have been obvious to a competent lawyer that the assistance of an accountant [was] necessary” as part of pretrial defense investigation of complex fraud case); Troedel, 667 F.Supp. at 1461 (finding counsel's failure to consult an expert, depose witnesses, and conduct an independent investigation despite the fact that "counsel himself had no special knowledge in the field” to constitute deficient performance).

15 . See Sims v. Livesay, 970 F.2d 1575, 1580-81 (6th Cir.1992) (finding that defense counsel "did not make a reasonable decision that further investigation of the physical evidence was unnecessary,” and noting that he did not make an " ‘independent investigation’ ” and failed to ask a defense expert to examine the evidence); see also William W. Turner, Investigating Particular Crimes, 2 Am.Jur. Trials 171 § II.8 ("[C]ounsel for an accused arsonist ... should, where possible, use discovery methods to determine the evidence on which the prosecution hinges its case.... Findings based on laboratory analysis or expert opinion should be submitted to his own experts to determine their validity.”).

16 . See Paine v. Massie, 339 F.3d 1194, 1202 (10th Cir.2003)(finding defense's failure to elicit expert testimony on Battered Women’s Syndrome (BWS) demonstrated ineffectiveness where such testimony "was necessary to mount an effective self-defense claim given the jury's likely misconceptions about BWS”) (emphasis in die original).

17 .The dissent acknowledges that it would be "an inexplicable blunder” if Raimo told the jury that "he needed to establish a reasonable doubt whether the fire was an arson in order to secure an acquittal for his client....” Post at 3 n. 1. Raimo did not put the proposition to the jury in the blunt terms used by the dissent. We did not say that he did. But the dissent's suggestion that Raimo’s statement to the jury ("we think they're wrong”) did not refer to the state's evidence of arson is insupportable. In his statement to the jury, Raimo specifically referred to the fire investigation when he said to the jurors that "you’re going to hear a lot from the fire investigators in this case, but a lot of the fire investigation, you know, we're going to suggest to you is not rocket science.” He also referred to the "clues there in the physical evidence of the fire.” Raimo’s statement to the jury that "we think they're wrong” was an unmistakable challenge to the state's evidence of arson.

18 . See Wiggins, 539 U.S. at 525-26, 123 S.Ct. 2527 (finding that counsel presented "a halfhearted mitigation case” and noting that "any reasonably competent attorney would have realized that pursing these leads [found in the initial mitigation investigation] was necessary to making an informed choice among possible defenses”); see also Rogers v. Israel, 746 F.2d 1288, 1295 (7th Cir.1984) (holding that "the defense counsel owed a duty to the petitioner to ask a qualified expert whether [the victim] would have been immediately incapacitated by his wound,” noting that counsel had recognized the potential value of such expert assistance during his defense preparations).

19 . See also Pavel v. Hollins, 261 F.3d 210, 223 (2d Cir.2001) (finding that attorney’s de-cisión not to prepare a defense because he believed that the trial court would grant his motion to dismiss was "not the sort of conscious, reasonably informed decision made by an attorney with an eye to benefitting his client”); Austin v. Bell, 126 F.3d 843, 849 (6th Cir.1997) (rejecting counsel’s argument that he "did not present any mitigating evidence because he did not think that it would do any good” despite the availability of witnesses, and finding that counsel's "reasoning does not reflect a strategic decision, but rather an abdication of advocacy”).

20 . See Lindstadt v. Keane, 239 F.3d 191, 201-02 (2d Cir.2001) (rejecting the prosecution's argument that defense counsel’s cross-examination of the expert was sufficient, noting that his "effort was hamstrung, however, by counsel's lack of familiarity with the studies upon which [the expert] was presumably relying; the effect was ruinous because [the expert] testified that the (unidentified) studies ruled out every theory of innocent injury ... posited by the defense”); Spencer v. Donnelly, 193 F.Supp.2d 718, 734 (W.D.N.Y.2002) (finding counsel’s performance deficient despite cross-examination of expert, and noting that "[a]t a minimum, the use of a child psychologist or similar expert would have been most useful and helpful to trial counsel in preparing for the cross-examination of [the state's expert]”).

21 . The dissent worries about the implications of this decision, predicting that defense attorneys and trial judges will read it to require attorneys to hire experts in every case where the prosecution uses an expert, and to require judges in court-appointed cases to grant funds to hire such experts whenever such funds are requested. See post at 319-20, 323-24. As our references to the ample precedents indicate, this is far from the first case where the failure of defense counsel to use experts in preparation for trial was the basis for a finding of ineffective assistance of counsel. Despite the insistence of the dissent to the contrary, this decision, like those precedents, is grounded in the specifics of this case. It is true that attorneys in future cases will try to make use of this precedent to argue that there has been ineffective assistance of counsel, and that they are entitled to funds for expert witnesses in cases involving indigent defendants. That phenomenon is a normal part of the legal process. It is not a reason to avoid applying well-established legal principles to the facts of this case.

22 . "Q. [W]as this a mistake where you made a tac ... — you made a tactical decision that there was a risk in hiring an expert to advise you, and rather than take that risk, you were not going to hire an expert.”

"A. No.”

23 . The dissent notes that Raimo "took into account” another possible peril — "that, in rare cases, hiring one’s own expert can have the effect of precluding an attorney from pursuing in good faith lines of argument designed to raise doubts about the State’s experts’ methods and conclusions.” Post at 321. As we understand it, the theory is that it was important for Raimo to preserve, by not consulting an expert, the option of pursuing a defense that could almost certainly never succeed without consulting an expert. Put differently, the dissent appears to argue that Raimo acted reasonably by refusing to consult an expert in an effort to preserve the option of proceeding in ignorance with a weak defense. We find this unpersuasive.

24 . The dissent says that we have come to this conclusion because we have accepted Rai-mo's characterization of his own performance as inadequate. See post at 321 n. 2, 322. Indeed, the dissent goes so far as to say that Raimo’s own testimony "forms the basis for much of the majority's analysis." Id. at 322. That is. simply not so. We have only noted Raimo’s acknowledgment that, contrary to the finding of the state court, he did not engage in a weighing of the benefits and perils of consulting an arson expert prior to trial. We have made clear, however, that even if such a benefit-peril balancing had been engaged in by Raimo, neither he nor the state court identified a plausible peril. We also have stated explicitly that Raimo's own judgment that his representation was inadequate has played no role in our decision.

25 . The dissent argues that Higgins's letter of October 20, 2000 (sent to Raimo while the conviction was on appeal), and presumably as well a followup letter sent on April 24, 2002 to Dugas’s post-conviction counsel, are inadmissible because they have not been sworn or authenticated. See post at 10-11. However, the state has never challenged the authenticity or admissibility of Higgins's letters at any phase of the proceeding in any court, and has probably forfeited any such challenge. See D.N.H. R. 7.2(c). ( Moreover, Higgins testified under oath as to the issues he identified in his October 20, 2000 letter during the state court hearing on Dugas's motion for discovery and was cross-examined on his analysis by the state. See Fed R. Evid. 901(b)(1) (authentication by testimony of witness with knowledge).

The dissent also leaps ahead of the limited evidentiary remand that we order, see infra Part IV, by arguing that Higgins misinterprets the applicable technical standard. See post at 10 n. 3. Undoubtedly, the exact requirements of the applicable standards will be a subject of dispute on remand. Assuming that the state chooses to raise this point on remand, the district court is in the best position to decide their merits.

26 . We find that one of the issues raised by Higgins — the state's failure to provide the video equipment and original tape to the defense' — does not affect the prejudice analysis in this case. Higgins suggests that the video equipment and the original tape should have been made available to the defense, and that, as an expert, he would have insisted upon this. He argues that this might have enabled Dugas to show that the tape's timing and sequence were in error. However, in light of Dugas’s admission that he returned to the store and acted in a manner consistent with what appeared on the enhanced videotape, we cannot perceive any impact on the outcome of the trial.

We cannot determine whether the other questionable issue raised by Higgins — his critique of the electrical equipment and wiring analysis — is relevant to the prejudice analysis. Higgins noted that the state's experts should carefully document and diagram the electrical equipment and wiring prior to removing them from the scene, and that the failure to do so calls into question the integrity of the evidence. The district court concluded that "Higgins's opinion about preserving evidence of the electrical system, even if true, does not contradict the state's experts' opinions excluding that system as a cause of fire." Du-gas III, slip op. at 15. Of course, Higgins cannot demonstrate a specific contradiction here because Raimo never conducted an independent investigation and thus never obtained the evidence from the state’s experts. As such, Higgins can only speculate as to whether an independent analysis would have produced different results, and speculation about the existence of missing contradictory evidence alone is not a ground for finding prejudice. However, a critique that undermines the methodology of the state's experts would be relevant in a prejudice determination in this case. We cannot determine, based on Higgins’s report or the record as a whole, whether the state actually failed to fully diagram and document the electrical equipment and wiring in its initial state prior to removal (it is clear only that some diagrams were created for the state's case at some point prior to trial). Nor is it clear how, as Higgins suggests, any failure to do so casts doubt on the integrity of the evidence and the electrical expert's methodology, particularly given the expert’s personal inspection of the scene prior to the removal of the equipment and wiring. If the state’s experts failed to document the initial scene and this failure does cast doubt on the methodology, this issue could constitute another unexplored line of questioning that Raimo could have used during cross-examination. The district court may seek to clarify the factual basis for this critique on remand, but we need not and do not consider this issue in determining that a remand is warranted in this case.

27 . Yet some jurors may have credited Dugas’s testimony about the video in his favor. If Dugas set the fire and was lying about his innocence, he could have chosen to also lie about his whereabouts and maintain the plausible argument that the video showed events from another date. Instead, Dugas testified that, contrary to his earlier statements, he had returned to the store. Dugas only stood to lose from this admission, and some of the jurors may have been inclined to believe Du-gas and his claims of innocence. This inclination may have been stronger if the jurors had also entertained doubts about whether the fire had been an arson at all.

28 . How the length of jury deliberations affects the prejudice analysis depends on the precise circumstances. See, e.g., Murtishaw v. Woodford, 255 F.3d 926, 974 (9th Cir.2001) (concluding, in challenge to capital case penalty phase jury instructions, that "the mitigating evidence presented, the jury’s apparent interest in it, and the length of the jury's deliberations,” established that error was not harmless); United States v. Bubar, 567 F.2d 192, 202 (2d Cir.1977) (noting, in ineffective assistance of counsel challenge, that while counsel’s defense was "odd,” nevertheless "some indication of its effectiveness is that it kept the jury out deliberating [defendant's] fate for three days”).

29 . The standard is formally known as ASTM E-1387, "Standard Test Method for Ignitable Liquid Residues in Extracts from Fire Debris Samples by Gas Chromatography.”

30 . To be sure, even if there is a credible innocuous explanation for the presence of ignitable liquids in the debris samples, Dugas has to contend with the fire investigators' unanimous opinions that the fire appeared to be intentionally set. We address that issue later in this section.

31 . Although Raimo intimated in his cross-examination of Boudreau that "J.P. Chemical” had recently applied pesticides to the basement, no witness testified to such an application. See supra note 4.

32 . It is, of course, possible that the toxic active ingredient degrades more rapidly than the solvent and would not show up in testing. Due to Raimo’s failure to explore this issue in greater depth, such issues were not raised at trial.

33 . At this stage, we place no weight on Higgins's statement that he tested some fire debris from the basement and found no ignitable liquids. Boudreau also testified that some of his samples yielded no ignitable liquids, and Eddy testified that, to the best of his knowledge, all of his samples tested negative. Eddy, however, also testified that he was not surprised by this finding, as he did not have the benefit of an accelerant detection dog, and therefore did not know precisely where to look. When a state police expert and an independent expert testifying for the prosecution both admit that at least some of their samples contained no ignitable liquids, a defense expert who also testified that his samples yielded no ignitable liquids would only have a marginal impact on a factfinder’s evaluation of the evidence.

34 . We note that the state's case as a whole was so circumstantial that any additional doubt might have tipped the jury. Thus, even if the only ground for prejudice was the critique of Boudreau's analysis, we would not be convinced that Dugas could not establish prejudice.

35 . Although Correia is an unpublished opinion, we mention it because of the similarity between its facts and those of this case.

36 . Although Scott is a state trial court decision, answering a different question under a different body of law, it is worth summarizing because the basic factual issues are similar. There, a forensic chemist at the Massachusetts State Crime Laboratory found a medium petroleum distillate in a fire debris sample, but inadvertently discarded the sample before trial. 1995 WL 808883 at *1, 6-7, 3 Mass. L. Rep. 309, at *4, *17-19. The defendants moved to dismiss the indictment, and Higgins testified, as a defense expert,

that if he had the actual physical evidence or the remaining liquid extraction, he could have re-run the tests to confirm or disagree with [the state chemist’s] findings.... If the physical evidence had been available, Higgins possibly could have conducted tests to specifically identify the petroleum product as being an accelerant or a natural component of the fire.

Id. at *2, 1995 Mass.Super LEXIS 847 at *6.

The trial court analyzed the motion to dismiss for destruction of evidence under the standard of " ‘a reasonable possibility, based on concrete evidence rather than a fertile imagination, that access to the destroyed evidence would have produced evidence favorable to his cause.’ ” Id. at *5, 1995 Mass.Super. LEXIS 847, at *14-15 (quoting Commonwealth v. Phoenix, 409 Mass. 408, 567 N.E.2d 193, 197 (1991)). It concluded that the samples could have been tested, under existing testing techniques, “to determine whether the petroleum product was an accel-erant or a natural component of the fire.” Id. at *6, 1995 Mass.Super. LEXIS 847 at *15-16. It further held that the evidence was material and could create a reasonable doubt:

The defendants have been charged with arson and an essential element that the Commonwealth must prove is that the defendants intentionally set the fire. The Commonwealth conducted tests which indicated that petroleum products and gasoline were detected in the samples. If the physical evidence had been available to the defendants, they could have performed their own tests, the results of which may have cast reasonable doubt that they set the fire. Essentially, defendants may have been able to prove the petroleum products were already on the premises and were not accelerants intentionally placed to set the fire.

Id. at *8, 1995 Mass.Super. LEXIS 847 at *21-22. Nevertheless, the court denied the motion to dismiss because it found that "[d]e-fendants’ right to a fair trial will be adequately protected by permitting the defendants latitude during their cross examination regarding the destroyed evidence.” Id. at *9, 1995 Mass.Super. LEXIS 847 at *24-25. Here, of course, defense counsel did not have the benefit of Higgins's critique at the time of trial.

37 . Granting the petition does not necessarily mean that the petitioner is immediately released. "Courts usually condition the issuance of a writ, which releases the body of the prisoner from custody obtained through unconstitutional means, upon the state's failure to retry the habeas petitioner within a reasonable time in a way that comports with constitutional dictates.” Henderson v. Frank, 155 F.3d 159, 168 (3d Cir.1998).

---

Dissenting Opinion

HOWARD, Circuit Judge,

dissenting.

The single issue raised in this case is whether Attorney Raimo acted incompetently at the point in time prior to trial when he chose not to hire an arson expert (or, perhaps, multiple arson experts, since the State called six experts at trial), and instead chose to focus his efforts on an identity defense — i.e., a defense directed at raising reasonable doubts in the jurors’ minds about whether Dugas was the arsonist. In concluding that Raimo’s pretrial decision not to hire an arson expert was incompetent — and in concomitantly holding that the New Hampshire Superior Court’s contrary determination was “not only incorrect but objectively unreasonable,” Rompilla v. Beard, — U.S. —, —, 125 S.Ct. 2456, 2462, 162 L.Ed.2d 360 (2005) — the majority portrays the decision as an unusual departure from professional norms, calling only for a modest and ease-specific corrective opinion unlikely to have repercussions beyond this case.

With respect, I think the majority opinion has broader implications. As the attorneys and judges, who work in the state and federal trial courts within our jurisdiction surely will notice, this case involves a professional decision of a type that is not only common and defensible, but necessary to the practical functioning of the criminal justice system. These professionals, who have more than an academic interest in our rulings, read our habeas and ineffective-assistance-of-counsel opinions with background knowledge of the deference federal habeas courts are to give state courts and trial counsel. They understand the broad discretion the Constitution necessarily confers on trial counsel to strategically allocate limited temporal and financial resources. They will appreciate that, even with the benefit of 20/20 hindsight, there is little reason to question whether the fire at the Dugas Superette was an arson. They will notice too that Raimo was supportably found to have deposed or interviewed all the State’s experts, toured the arson scene, read up on arson, and conferred with fellow counsel prior to deciding not to hire an expert. In view of all this, they may well conclude that, under the logic of the majority’s analysis, the hiring (or appointment, in the case of indigent defendants) of consulting “counter” experts is constitutionally required in nearly all cases where the prosecution calls an expert to prove an element of its case. I write separately to express my disagreement with the majority’s ruling and my concern about its implications.

At the outset, there is a need to clarify the scope of this appeal. The majority opinion gives the impression that there is a second issue before us: whether Raimo was incompetent in promising the jury early in the trial that he would establish that the fire was “not arson,” but then failing to deliver on that promise. See, e.g., ante at 324, 328, 330. There is no such issue. First, petitioner has not made an “unfulfilled promise” argument, either before the state court or on collateral review. Second, petitioner did not make opening and closing arguments part of the habeas record, and there is no other evidence of Raimo having made such a promise.

The majority’s conclusion that there is an unfulfilled-promise issue in this case is based entirely on Raimo’s statement to the jury, prior to its taking a view of the fire scene, that “what we’re going to be asking ourselves during this trial is how this fire started and why.... I want to just make it clear ... where the State is bringing in all of these witnesses ... we think they’re wrong.” Ante at 323. Indeed, the majority goes so far as to suggest that this statement was a de facto admission to the jury by Raimo that “[mjueh of Dugas’s defense ... depended on Raimo’s ability to convince the jurors that the State’s experts might be wrong in concluding that the fire was arson.” Ante at 328. No such admission is made or implied by Raimo’s statement, which can as easily be taken to refer to the State’s allegation that Dugas was the arsonist — the allegation Raimo contested vigorously — as to the less — contested allegation that the fire was an arson.³⁸ And in any event, even if the statement is assumed to refer to the State’s anticipated arson evidence, it is hardly the sort of unusual “promise” from defense counsel that the majority makes it out to be.

The record does not contain evidence that Raimo promised to prove that the fire was not an arson. Nor does it suggest that Raimo told the jury that his defense depended on his ability to raise reasonable doubts whether the fire was an arson. Rather, the record reveals that Raimo made a somewhat modest effort through cross-examination to raise doubts about whether the State had proved arson beyond a reasonable doubt, but focused the majority of his attention and energy on challenging the State’s logically unrelated identity case. The record does not bear out the contrary impression left by the sections of the majority opinion just discussed, and the majority’s many references to the “not arson” defense.

This leaves the question actually raised in petitioner’s brief: whether the state court unreasonably determined that Raimo acted competently at the point in time, prior to trial, when he decided not to hire an expert and instead to emphasize his identity defense. A review of how the state court disposed of this argument, conducted in the manner required by the ha-beas statutes, reveals that the answer is no.

Following an evidentiary hearing on Du-gas’s motion for a new trial, the state court issued an opinion rejecting the claim that Raimo’s pretrial decision-making was constitutionally flawed. See State v. Dugas, No. 98-S-1899, slip op. at 3-6 (N.H.Super.Ct. Aug. 12, 2002) (orders on motions for new trial and discovery). The court premised its holding on a finding that Rai-mo’s decision not to hire a consulting expert was tactical, and was made after a diligent and thorough investigation and consideration of the benefits and perils of a contrary path. Id. at 5-6. This finding essentially compelled the conclusion that Raimo’s decision did not constitute ineffec-five assistance of counsel. Id. (applying the standard by which the constitutionality of pretrial investigations are measured— that a strategic decision made after a thorough investigation is almost never incompetent — as articulated in Strickland v. Washington, 466 U.S. 668, 691, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984)). The,majority says that this finding was sufficiently at odds with the record to meet the extraordinarily demanding requirement for upsetting a state court factual finding on federal collateral review: that the finding be “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(2). But this is not so.

It is undisputed that Raimo, who at the time of trial had more than two decades of experience trying criminal cases, undertook an investigation prior to deciding whether to hire an arson expert. Raimo conceded on cross-examination by the State³⁹ that he toured the crime scene with his client and certain of the State’s experts, who explained to him their opinions that the physical evidence suggested arson; deposed or interviewed each of the State’s experts; concluded that the State’s experts were formidable and would impress the jury; independently educated himself on arson; and consulted with fellow defense counsel about the likely efficacy of hiring his own expert under the circumstances.

Raimo also mentioned two additional concerns. First, he not only concluded that retaining his own expert would likely be a waste of time and money (presumably because the strength of the State’s evidence and experts led him to conclude that an expert good enough to go toe-to-toe with the State’s experts likely would concur in the conclusion that the fire was an arson), but he also took into account the fact that, in rare cases, hiring one’s own expert can have the effect of precluding an attorney from pursuing in good faith lines of argument designed to raise doubts about the State’s experts’ methods and conclusions. Second, he considered the risk that the state somehow would discover the potential expert’s opinion, which might further damage his client’s defense and limit his options. The majority, following the lead of the district court, emphasizes that a consulting expert need not be disclosed. But the more significant point is that, based on his pretrial investigation, Raimo had legitimate practical litigation concerns that consulting with an expert would result in a net loss to his client. In the specific circumstances of this case, these considerations, whatever their weight, were not wholly insubstantial. Or at the very least, the state court did not act unreasonably in so concluding.

In any event, the totality of this evidence amply supports the state court’s determination that Raimo’s decision not to retain an expert was tactical, considered, and based on a diligent and thorough investigation. The majority is led to conclude otherwise by means of what is, at bottom, a tautology: that, in an area involving scientific knowledge in which one is not well versed, one cannot have sufficiently investigated whether to hire a defense “counter” expert without retaining an expert to see whether an expert is needed. But this line of reasoning ignores the fact that experienced defense counsel and trial judges regularly must make difficult resource-allocation judgments prior to trial. The state court supportably found that Raimo, an experienced criminal trial attorney, took stock of the State’s evidence and multiple expert witnesses — all of whom quickly agreed that the fire was an arson for reasons that they explained to Raimo, presumably to his satisfaction — and concluded that the State would be able to put on an overwhelming case of arson; that he considered the possibility that hiring his own expert would be a waste of time and resources and conceivably could damage his client’s case; and that he therefore decided to invest his time (and Dugas’s money) in what he perceived to be a more fruitful defense theory: that the State could not prove beyond a reasonable doubt that his client was the arsonist because there were others who had the opportunity and motive to commit the crime. Perhaps these choices were unwise, but they simply cannot ground a ruling that the state court acted unreasonably in rejecting Dugas’s argument that Raimo “was not functioning as the ‘counsel’ guaranteed [Dugas] by the Sixth Amendment.” Strickland, 466 U.S. at 687, 104 S.Ct. 2052.

To be sure, Raimo characterized his performance as inadequate on direct examination, and he attributed his decision not to hire an expert to overconfidence in his theory that, if the fire were an arson, somebody else was the perpetrator. This testimony forms the basis for much of the majority’s analysis. But as already noted, see supra note 2, client-friendly, self-incul-patory characterization and attribution testimony of this sort — hardly unprecedented in post-conviction proceedings — should have been disregarded because it is at odds with what the state court supportably found. And in any event, as the majority recognizes, Raimo’s opinions about the law and his conduct have no bearing on our analysis. See Strickland, 466 U.S. at 688, 104 S.Ct. 2052 (emphasizing that counsel’s performance must fall below “an objective standard of reasonableness”) (emphasis supplied); see also Cofske v. United States, 290 F.3d 437, 444 (1st Cir.2002) (stating that, under Strickland, “as long as counsel performed as a competent lawyer would, his or her detailed subjective reasoning is beside the point”).

In sum, we should have upheld as a “[ jreasonable determination of the facts in light of the evidence presented” the state court’s finding that Raimo engaged in a thorough investigation of the pros and cons of hiring a consulting expert prior to making his decision. 28 U.S.C. § 2254(d)(2). And this should have led us to conclude that the state court’s denial of the new trial motion did not involve “an unreasonable application of’ Strickland. 28 U.S.C. § 2254(d)(1); see Strickland, 466 U.S. at 691, 104 S.Ct. 2052 (“strategic choices made after thorough investigation of the law and facts relevant to plausible options are virtually unchallengeable”).

A conclusion that the state court reasonably determined that Raimo provided Du-gas with competent representation would have negated any need to consider the issue of prejudice. But even if the issue of prejudice were to be properly reached, the analysis provided by the majority in support of its remand order is unpersuasive. As an initial matter, the “report” submitted in connection with Dugas’ summary judgment papers — the document which grounds the majority’s prejudice ruling — is not the sort of expert report contemplated by the Federal Rules. It is, in fact, only a brief, unsworn, unauthenticated letter from putative expert Michael K. Higgins, which in its most relevant section contains hearsay declarations about how the State may have violated certain laboratory standards in handling the arson evidence.⁴⁰ The fact that the Higgins letter does not constitute or contain admissible evidence, see, e.g., Gorski v. New Hampshire Dept. of Corrections, 290 F.3d 466, 475 (1st Cir.2002) (party opposing summary judgment must do so with admissible evidence),⁴¹ and that it otherwise does little to undermine the plentiful evidence that the fire was an arson — evidence, that ranges far beyond the evidence that the arsonist placed an accelerant on the papers that were ignited — should have led the court to conclude that a remand is unwarranted.

In conclusion, the majority states that its opinion is narrow and tailored to the facts of this case, and that it has not established a new rule of constitutional law requiring the retention or appointment of double-checking counter experts in the many contexts' — e.g., fingerprints, chemical analyses of narcotics, ballistics, origin of firearms — where prosecutors use experts to explain scientific or otherwise specialized evidence. But law is made less by what an opinion says than by' what the opinion causes practitioners to do as they engage in the predictive processes inherent in legal practice and reasoning. Cf. Holmes, The Path of the Law, 10 Harv. L.Rev. 457, 461 (1897). In view of the facts of this case, there is reason for concern that, notwithstanding the majority’s minimalist assurances, defense attorneys will read today’s opinion as embracing, at the least, a presumption that they must spend precious time and money on constitutionally required double, checks of most prosecution science experts — double checks that never before were required and reasonably may be eschewed in many circumstances. So too is there reason to fear that trial judges will read this opinion as constraining their discretion in deciding, in the case of indigent defendants, whether and when to expend limited public funds on court-appointed defense experts for purposes of double-checking the prosecution. And thus would the already slow and costly criminal trial process unnecessarily become slower and more costly.

38 . Indeed, a de facto admission by Raimo to the jury that he needed to establish a reasonable doubt whether the fire was an arson in order to secure an acquittal for his client, had it actually been made, would have been an inexplicable blunder — one that itself would raise Sixth Amendment concerns. The State's arson case was strong, its identity case was weaker, and the two were logically independent of one another, making it perfectly rational for Raimo to vigorously challenge the State's evidence that Dugas was the arsonist while at the same time leaving its evidence that the fire was an arson largely uncontested. Cf. Haines v. Risley, 412 F.3d 285, 289-90 (1st Cir.2005) (observing that trial counsel sometimes reasonably forego weaker alternative defense theories to avoid detracting from a primary defense theoiy or otherwise losing credibility with the jury). The majority's contrary argument — that Dugas’s overall prospects for an acquittal depended on Raimo’s ability to convince the jurors that the State's arson experts might be wrong, and that Rai-mo knew this, see ante at 327-28 is hard to fathom.

39 . Raimo was called as a witness by Dugas at the hearing on his motion for new trial. On direct examination, Raimo expressed an opinion that he had been constitutionally ineffective and made many of the "concessions” and "admissions” so heavily relied on by the majority. As commonly happens in post-conviction proceedings, Raimo fell on his sword for his former client. Thereafter, however, under cross-examination by the State, Raimo testified to the particulars of his investigation. These were the particulars that led the state trial judge to reject Raimo's self-denigrating testimony on direct and to conclude that the decision not to hire an expert was a tactical one made after a thorough investigation.

In crediting Raimo's direct testimony, the majority opinion does not correctly apply 28 U.S.C. § 2254(d)(2), which requires that we defer to the state court’s factual findings unless they were based on an unreasonable construction of the evidence. As explained momentarily, the state court’s findings that Raimo’s investigation was thorough and his decisions tactical are based on a reasonable construction of the evidence. We thus were obliged to reject those aspects of Raimo's testimony which were at odds with these conclusions. The majority has not done so.

40 . The most significant violation identified by Higgins in his letter is the State’s alleged abridgment of ASTM E-1387-95, Section 8.1.2.2., which, Higgins says, "requires the exact identification” of any flammable liquid identified at trial as an accelerant. (As the majority notes, the State introduced evidence at trial that there was an accelerant on the papers set ablaze in the Superette’s basement, but it did not identify the accelerant). It is worth noting, however, that the text of the relevant standard, submitted with Higgin's letter, does not explicitly call for such an identification. Instead, it merely states that “[ejveiy case file that includes a positive identification of an ignitable liquid or residue must include the standard chromatogram used to confirm the identification.” ASTM E-1387-95, Section 8.1.2.2.

41 . The majority suggests that the State "probably” has forfeited its right to challenge the admissibility of the Higgins letter. Ante note 23. The suggestion is incorrect. Because we should leave undisturbed final judgments that are demonstrably correct, it is long settled that we are free to affirm on any ground supported by the record — whether that ground has been argued by the respondent/appellee or not. See, e.g., Sammartano v. Palmas del Mar Properties, Inc., 161 F.3d 96, 97 n. 2 (1st Cir.1998). The imperative to leave substantively correct final judgments undisturbed is strong in all cases, but it is difficult to conceive of a situation where it would be stronger than where the final judgment under consideration is a state criminal conviction challenged in federal court on collateral review.

The majority further suggests that the inadmissibility of the Higgins report was, in effect, cured by his sworn testimony given in connection with Dugas's motion for discovery before the state trial court, which covered "the issues identified ... in his letter.” Ante at note 23. But the transcript. of Higgins’ testimony shows little overlap with the "issues” identified in the letter and on which the majority relies in ordering a remand. At that hearing, Higgins merely testified on direct examination what he wanted in terms of discovery from the State and why; on cross-examination, he merely strived (unsuccessfully) to explain the unsupported characterization of the requirements of ASTM E-1387-95, Section 8.1.2.2., set forth in his unsworn letter.