*227*212It is not an easy decision to grant a new trial to a man after he has been convicted of killing another human being and that conviction has been upheld on appeal. But more important than the severity of a crime is the fundamental principle of American law-all accused must receive a fair trial, even those accused of setting an apartment house fire that caused the death of a tenant. That legal principle led the judge here to make such a decision and order a new trial for Frank Robinson. Based on the judge's thoughtful and thorough written opinion, and our review of the record, we agree.
The facts are reported in the direct criminal appeal.
After an apartment house in Topeka burned and one of the tenants died, the United States government charged Robinson with arson resulting in death. When the federal government dismissed the charge before trial, the State of Kansas charged Robinson with felony murder and aggravated arson. All pertinent details of the state prosecution may be read in the direct criminal appeal reported in State v. Robinson , No. 105,281, 2012 WL 4794455 (Kan. App. 2012) (unpublished opinion), which upheld Robinson's convictions for reckless second-degree murder and aggravated arson. But to provide a framework for our discussion, we will give a brief summary.
On the day the apartment house burned down, there were two fires. The first, discovered in the basement, was extinguished by one of the tenants. A second fire, near the stairs, soon engulfed the building in flames. The upstairs tenant did not escape, and she died in the fire.
The government's fire investigator, Agent Douglas Monty, testified that in his opinion, the fire was intentionally set on the stairs *213using an open flame, such as a lighter, and a flammable liquid. In statements to detectives, Robinson denied any responsibility for the fire but did say that he could have been smoking crack cocaine in the hall of the apartment and could have thrown lighted matches on the floor, but he never intended to hurt anyone. "I can't say whether I did it [caused the fire] or didn't, but I know I was smoking right there."
The district court received evidence on the habeas corpus motion.
In his motion for habeas corpus relief, Robinson alleged several grounds for relief. He contended that he was denied his constitutional right to effective assistance of counsel because his lawyer failed to:
• move to suppress some statements he made to law enforcement officers;
• present expert testimony refuting the claims made by the State's fire investigators;
• impeach certain trial witnesses-Detective Bryan Wheeles, Detective Brian Hill, and Fire Marshal Wally Roberts-with contradictory testimony from other hearings;
• present alibi witnesses;
• present exculpatory evidence contradicting Ernest Brown Sr.'s testimony that he saw Robinson fleeing the fire;
• allow Robinson to testify at trial; and *228• consult with Robinson before making a written stipulation that the owner of the apartment house did not consent to having it burned.
Finally, he claimed cumulative errors by trial counsel should lead to a new trial.
At the habeas corpus hearing, Robinson presented several witnesses:
• Joseph Huerter, Robinson's court-appointed trial attorney;
• Gene Gietzen, a forensic scientist with whom Huerter consulted;
• Kirk Redmond, Robinson's federal public defender;
• Melody Brannon, Robinson's federal public defender; and
• Paul Bieber, a fire investigator.
*214For its part, the State called Jason Belveal, one of Robinson's attorneys, to testify.
Eventually, the court ruled that Robinson's defense attorneys were ineffective for failing to investigate and then present sufficient expert testimony to refute the claims of the State's expert. The court also decided that this failure prejudiced the defense. The court also ruled that cumulative errors by the defense counsel called for habeas corpus relief. After concluding that Robinson's convictions were vulnerable to collateral attack because of ineffective assistance of counsel, the court ordered a new trial. But the court remained unmoved by Robinson's other contentions.
The State appeals and Robinson cross-appeals.
To us, the State argues that the district court erred in finding Robinson's trial counsel ineffective. It contends defense counsel's actions were from trial strategy and the court should not, contrary to longstanding caselaw, substitute its judgment for that of counsel. Robinson cross-appeals, arguing that the district court erred in finding that the remainder of his claims for relief had no merit.
We concentrate on the expert evidence issue and will show that the court's holding is supported by substantial competent evidence and is legally correct. The court's finding of cumulative error is insignificant given the expert evidence deficiency. In turn, we agree with the district court about the issues Robinson raises in his cross-appeal. Those issues we will deal with summarily since we agree that he is entitled to a new trial.
Arson prosecutions are unique because often they are based on opinions. Fires may start from natural causes, by accidental means, or even spontaneously, so if the State seeks to prove someone guilty of arson, most often it must rely on expert opinions to prove the cause and origin of that fire. In other words, it must show that someone intentionally started the fire. That cause and origin evidence is the heart of the State's case, and it follows, then, that an obvious line of defense for someone charged with arson is to attack and try to counter that cause and origin evidence. To mount such an attack, the defense must at least explore the possibility of using its own experts. By obtaining its own expert, the defense can offer *215the jury its view of the cause and origin evidence and not have it rely just on the opinions of the State's expert.
This need for expertise is readily apparent in this appeal, where the district court ruled the defense counsel's performance in this arson prosecution was deficient, prejudicial, and required a new trial. The court questioned the wisdom of the defense counsel waiting until less than two weeks before the jury trial to consult with someone who, as it turns out, had no actual expertise in determining the cause and origin of fires. He was a consultant who merely provided the defense team with some notes for use in the cross-examination of the State's expert witness. The witness offered no opinions about the cause and origin of this fire.
In vivid contrast to the small efforts of Robinson's defense counsel, expert testimony at Robinson's habeas corpus hearing revealed that the State's expert's opinions were vulnerable to attack because part of them were based on a process of elimination of causes and not on any actual physical evidence. This conclusion was not a product of the scientific method. This claim of flawed reasoning is something a jury could have pondered when deciding Robinson's guilt or innocence. Thus, Robinson did show the habeas *229corpus court not only deficient performance by his lawyer through lack of preparation, but prejudice to his defense as well.
This case is a study in contrasts: between lawyers and experts.
Contrasts between the preparation of Robinson's state defense attorneys and his federal public defenders are notable and significant. Contrasts, too, between the experts his state and federal defense counsel consulted, as well as the prosecution's expert who testified at his trial, and the expert who testified at the habeas corpus hearing are startling and unsettling. We can see why the court felt compelled to grant a new trial.
The first witness was Joe Huerter, the attorney appointed to represent Robinson at his trial. Huerter testified that he began his law practice in Topeka in 1984 and had been continuously practicing law since then. His legal practice consists of an almost equal split between criminal defense and family law, with occasional work on other types of law. He estimated that in 2009, he had represented *216defendants in about 12 homicide cases, and almost all of those were first-degree homicide charges.
Huerter testified that his contract provided his firm a flat fee for representing Robinson, but that fee did not include fees for an investigator. While he could have obtained a licensed investigator, he preferred to handle the investigation within the firm. Huerter testified that he did some investigation himself and other parts of it were handled by Belveal, an associate with his firm, and David McDonald, a law clerk who later became an associate in the firm.
Huerter hired no arson expert to testify at Robinson's trial, but he did contact an "arson investigation-type expert, cause and origin person," Gene Gietzen. Huerter stated that after the initial feedback from Gietzen, he decided that "it would not be beneficial to the defense to have him testify but rather to end up using him as a consulting expert to help develop the questions we might ask of the State's expert."
Huerter stated that Gietzen never provided any written report; he only talked to the attorneys on the phone. Huerter testified that Gietzen did not point out any flaw in the State's evidence:
"It was more just a discussion of, since he didn't think he could get up and say, with any clarity, that the cause and origin report was specifically wrong in certain areas, that he could provide us areas where we could ask questions where the investigator might have to admit that, well, the opinion on that was maybe not as rock solid as would appear from the first reading of the report."
At this point in the hearing, Robinson's habeas corpus counsel asked Huerter why he did not seek another expert after Gietzen did not give him ideal answers. Huerter answered: "Well, for one thing I generally don't go around trying to find people that are just opinions for hire because they generally, you know, just blow up in your face." Huerter testified that before he talks to an expert, he generally would obtain the person's curriculum vitae, and then he would ask the person how many times he or she had testified and whether a court had ever refused to recognize him or her as an expert. He stated it was safe to assume that if there was anything on a person's curriculum vitae that showed the person was unqualified, he would not have continued to talk to the person.
Huerter admitted that he consulted with Gietzen less than two *217weeks before trial. He stated that if Gietzen had "told us that he could do something more than what [he] ended up doing," Huerter would have requested additional time from the court. He also testified that Gietzen could not contradict the other expert's determination of the cause and origin of the fire, so he did not want him to make a written report or testify in a way that could potentially bolster the State's case.
The habeas corpus court was then presented with a night-and-day contrast of this last minute search for background information and cross-examination material to the in-depth preparation conducted by the federal public defenders. The court was especially impressed by the quality of the consulting experts.
*230Redmond, Robinson's federal public defender, testified that he and Brannon were Robinson's attorneys when the case was first filed in federal court. He told the court about what they had done in that case before it was dismissed.
At the start, Redmond sought help from an expert named John Lentini to help him evaluate the prosecution's cause and origin report. He sent Lentini all that he had discovered along with the photographs and the cause and origin report. He pointed out that Lentini was on the board that put together the Guide for Fire and Explosion Investigations published by the National Fire Protection Association, called the NFPA 921. Lentini has also written a foundational text on arson investigation.
This NFPA 921 publication was important to Redmond. He studied it and used it in the preparation of his defense. It gave him the technical background he needed to develop insight into where possible lines of defense could be effective. He moved to have it recognized as a learned treatise by the court so the defense could use it as an aid in the cross-examination of Agent Monty, the government's expert. Redmond testified that he did not rely on his own efforts to interview witnesses but he had the chief investigator for his office, Cindy Johnson, interview as many witnesses from the scene of the fire as possible.
Redmond testified that he moved for a " Daubert " hearing, see Daubert v. Merrell Dow Pharmaceuticals, Inc. , 509 U.S. 579, 113 S.Ct. 2786, 125 L.Ed. 2d. 469 (1993), on the cause and origin *218determination and on the accelerant-detection canines used by Agent Monty in his attempt to detect the presence of flammable accelerants in debris taken from the fire scene. Unlike Huerter, at this early point in his preparation Redmond was already preparing several lines of attack on the prosecution's expert opinions.
One of these attacks centered on the cause and origin report. In Redmond's view, it did not fit with the recommendations in the NFPA 921 publication he had read. Redmond contended that Agent Monty's cause and origin opinion was based on a "negative corpus finding." This finding is a conclusion reached through a process of elimination that a fire was an arson because it could not be caused in any other way.
"You have to be able to prove that there was no other way that this fire could have started except for an incendiary determination. And the NFPA says that really you can't-you have [an] extremely difficult time doing that in cases where there's a great deal of destruction ...."
There was much destruction here, because the second floor of the apartment house collapsed onto the first floor during the fire.
In another line of attack, Redmond pointed out that he did not think Agent Monty's reliance on the gas can in the area was valid. He stated that there are "78 million gas cans in the United States" and that particular gas can was never linked to Robinson either by witness testimony, fingerprints, or DNA evidence.
His major line for the defense, according to Redmond, was that the laboratory analysis revealed there was no evidence of any accelerants used at the fire scene. Agent Monty had tried to explain this lack of evidence away by stating any accelerant could have burned up in the fire. But Redmond pointed out that the literature he had studied states that although it is possible, it is extremely unlikely that an accelerant would "get burned up." As an example, he explained that anytime gasoline is "poured on [a] porous surface, which would include wood floors like we had in this case, that you'll almost always find that it's too strong. But that you will regularly find, even after a fire, evidence of that accelerant." From this testimony, Redmond had already been reading deeply in the technical literature to prepare himself for trial.
Redmond also challenged Agent Monty's conclusion that *219Robinson was not flicking matches as reported by Robinson because Monty relied on his personal belief that Robinson was not being truthful about that. Redmond stated that this was not a conclusion Agent Monty was entitled to make because as he testified, he is not an expert on how people smoke crack. *231The federal district court never ruled on the Daubert motion because the Government dismissed the case before the court could rule on it.
The next attorney to testify was Brannon, another Kansas federal public defender. She had been practicing law for almost 26 years, 18 of them as a public defender. She was also appointed to represent Robinson along with Redmond when he was first charged in federal court. She testified that because Robinson was charged as a capital case, in addition to Redmond and she being assigned because of their significant capital experience, they also assigned two investigators and a paralegal to the case.
When Robinson's habeas corpus counsel asked Brannon how a reasonable lawyer would handle an arson case, she answered, "You get the experts." She stated that she has routinely used an expert just to assist her in preparation for questioning. She testified that all the work the experts had done in preparation for Robinson's federal defense would have been available to Huerter. She stated that at one point, she and Redmond had a phone conversation with one of Huerter's associates. It was a short call, and she could not remember all the topics they discussed.
We turn now to experts.
Next, Gietzen testified. He has a forensic consulting business in Springfield, Missouri, that he began in 1992. Gietzen testified that McDonald, the law clerk, first talked with him about this case. McDonald asked him to review the reports and assist him as a consulting expert on what to ask about the scene and the arson-related analysis. McDonald provided Gietzen with a 21-page report from the State's expert, two diagrams, laboratory reports, evidence chain of custody documents, and a transcript from the grand jury proceeding.
Gietzen later gave McDonald a timeline and 21 pages of *220recommended cross-examination questions. He also had about an hour-long phone conversation with either McDonald or Huerter. He made no finding on either the cause or origin of the fire.
Gietzen was not a member of the National Association of Fire Investigators (NAFI) or the International Association of Arson Investigators (IAAI). He had once attended an FBI arson analysis class in Quantico, Virginia, for the laboratory analysis of debris from arson cases, and he had testified in court on the subject of that training. Gietzen had also assisted in homicide investigations that involved fire.
Paul Bieber testified. He is a fire investigator who assists innocence projects and public defender offices with case review and consultation on arson cases. He was a firefighter for 15 years, and toward the end of his career, he obtained national certification through the National Association of Fire Investigators. He left the fire department in 2003 and worked for two years as a private consultant performing fire cause and origin determinations. After obtaining his bachelor's degree, he became involved in a post-conviction review of an arson case, which exposed him to some of the untested and unreliable methods used by fire investigators. He is a member of the National Association of Fire Investigators and the International Association of Arson Investigators.
In preparing to testify at the habeas corpus hearing, Bieber reviewed Agent Monty's fire investigation report. He also read the report of the dog handler who was involved in the investigation. He read courtroom testimony from the original trial, viewed several photographs of the scene, and saw some witness statement reports. Bieber testified that the Guide for Fire and Explosion Investigations published by the National Fire Protection Association-the NFPA 921-tells a fire investigator what investigation methods are approved by the industry. Both professional associations, the NAFI and the IAAI, have endorsed using the NFPA 921.
Bieber stated that the NFPA 921 codifies the four aspects to a forensic fire investigation. Bieber described the negative corpus theory as using a process of elimination alone :
"Negative corpus is the methodology of the process of using this process of elimination alone to draw a conclusion. So the example I said was that there are *221several *232potential ignition sources. Let's say there are three. Negative corpus would support examining and eliminating ignition source number one and two, but then in the absence of any evidence to support a final conclusion that number three did start the fire, to then conclude it simply on the basis of eliminating A and B."
Before 2011, the NFPA 921 noted that the negative corpus method was an allowable consideration. But even before 2011, Bieber stated that if the area of origin was not defined, then negative corpus would be inappropriate. In 2011, the NFPA 921 updated its text to explain that the methodology of negative corpus violated the scientific method because it allowed forensic examiners to draw conclusions with no physical or empirical evidence to support the conclusion.
Bieber testified that he believed Agent Monty used the negative corpus method. He focused on Monty's "handheld open flame device" conclusion:
"[Agent Monty] drew cause determinations regarding two specific or separate fires. There's a fire in the basement and a fire in the stairway. His examination of the area of origin in the basement, if I'm understanding his conclusions correctly, is that the fire started in a specific area in the basement, the first fuel ignited was very clear, and there were a couple of potential ignition sources that he was able to very quickly and apparently, accurately eliminate. There [were] some appliances down there but they weren't plugged in. I don't remember what else there might have been. From that, he determined or he concluded that the ignition source for the basement fire was a, if I'm using the proper terminology that he used, a handheld open flame device. I recognize that under the circumstances presented in this fact pattern that a handheld open flame device cannot be eliminated because there is no evidence to support or eliminate it. However, his conclusion is not that-and when he uses the term handheld open flame device he means a match or a lighter. There's no evidence to support that that was the ignition source yet that is his expert conclusion. And the same could be said for the fire in the stairwell."
According to Bieber, Agent Monty's conclusion did not comply with the scientific method. Using scientific terms, Bieber believed that Agent Monty's conclusion should have been that the ignition source was undetermined since there was no physical evidence to support his conclusion that a match or lighter started this fire.
Bieber next testified about Agent Monty's conclusion that an accelerant in the form of an ignitable liquid was used because he thought the fire progressed rapidly. But, according to Bieber, by *222the time the first witness observed the fire, it was already huge. When it was first observed, the fire was already beyond what any ignitable liquid would have produced, so it was impossible to tell at that point how quickly it had progressed.
There are other causes for this fire to spread rapidly. Agent Monty had noted that the apartment house had a balloon frame construction, which would make the fire progress rapidly once it got into the walls or frame of the apartment. An important point according to Bieber was that the fire could have progressed rapidly because there was no sheetrock underneath the stairs. So once the stairs caught fire, the stairway would become like a chimney and quickly take the fire to the second floor.
Bieber went on to say that he did not agree with the fire chief's statement that an ignitable liquid was present in the fire because when the firefighters put so much water on it, it was still not extinguished. According to Bieber, however, the level of effort it took to extinguish a fire is not a helpful factor in determining whether there was an ignitable liquid used here. By the time the fire department got there, whatever started the fire had nothing to do with the size of the fire since that initial fuel was already used up by the time the fire department arrived. Bieber stated that the reason the fire did not get extinguished was probably that the water was not getting to the seat of the flame. The fire was burning underneath where the water was hitting.
*233Bieber also did not agree with relying on the stairs collapsing as support for the conclusion that an ignitable fluid was used. He also did not agree with Agent Monty's belief that there was an insufficient fuel load on the stairway to develop into such a large fire without an ignitable liquid. He pointed to the witness statements about trash and debris as well as the carpet and padding on the stairs which could have been a sufficient fuel source.
Bieber concluded that being unable to pinpoint the area of origin of the fire is significant:
"Within the separate fire in the stairway, the specific ignition-the specific area of origin is not known, but we know the area of origin was somewhere in the stairwell. The ignition source, under these circumstances, is undetermined. And the first fuel ignited is unknown and is undetermined.
*223....
"In the absence of a determination of ignition source and in the case of the stairway first fuel ignited, the forensic fire scene examiner would be unable to determine the cause of the fire. The intentional or accidental nature which is needed-which you need to understand and conclude in order to make a conclusion of incendiary, is not an expert conclusion. It's not a conclusion of forensic science. It's a conclusion made by a fact finder in the-considering the totality of the circumstances, witness statements, other circumstantial information. The only area that I have an opinion on are the conclusions based on forensic fire scene examination. The cause of the fire. The area of origin. How the fire developed. I don't have an opinion much less an expert opinion on the accidental or intentional nature of how the fire began."
Bieber testified that "[t]he cause of these fires, through the forensic science process, is undetermined."
On cross-examination, Bieber testified that the negative corpus methodology was still recognized in the edition of the NFPA 921 in effect in 2009 when this case was tried. And he noted that Agent Monty was more than just a fire expert; he was also a criminal investigator.
The State offered testimony.
The State called Belveal to testify. He testified that he, Huerter, and McDonald looked through the entire federal case file. They believed Robinson's statements were admissible since the federal district court had denied the challenges to those statements. He stated that they also thought the statements helped their case in some respects.
Belveal could not remember Gietzen's name, who they employed to give them some expert advice about the fire investigators. He decided to use Gietzen as a consulting expert rather than to call him as a witness. He consulted with Robinson when making this decision, and the things Gietzen told them were helpful at trial.
Belveal believed that they were prepared to cross-examine Brown and Detectives Wheeles and Hill. Belveal testified that although they would have given Robinson their opinion about whether he should testify at trial, they would have told him it was his decision. He testified that they talked to Robinson before agreeing to stipulate to the fact that the owner of the building did not give Robinson permission to burn it down.
*224On cross-examination, Belveal testified that they talked to Brannon and Redmond when they went to the federal public defender's office to look through their file. According to Belveal, the attorneys walked them through the different things in the file and talked about what they had done. They pointed out motions they thought were important and answered some questions.
Belveal testified that Robinson had two or three people who could be alibi witnesses, and Belveal tried to locate them. He contacted at least one of those people, and what that person told him was not helpful. Belveal could not remember whether he had been able to talk to one or two of the three alleged alibi witnesses, but he knew they could not find all three of them. He also tried to locate Chuckie Praylow, who he believed might be another possible suspect who committed the crime.
*234Belveal testified that he thought Robinson wanted to testify at trial, but counsel suggested that he not do so because they did not think it would be good for him. He had too many inconsistent prior statements. His memory was poor because of drug usage. But ultimately, they told him it was up to him to decide.
The rules we must obey .
A habeas corpus claim alleging ineffective assistance of counsel presents mixed questions of fact and law. When the district court conducts a full evidentiary hearing on such claims, the appellate courts determine whether the district court's findings are supported by substantial competent evidence and whether the factual findings support the court's legal conclusions. The appellate courts apply a de novo standard to the district court's conclusions of law. Fuller v. State , 303 Kan. 478, 485, 363 P.3d 373 (2015).
To prevail on a claim of ineffective assistance of counsel, a criminal defendant must establish (1) that the performance of defense counsel was deficient under all the circumstances; and (2) prejudice-that there is a reasonable probability the jury would have reached a different result absent the deficient performance. Sola-Morales v. State , 300 Kan. 875, 882-83, 335 P.3d 1162 (2014) (relying on Strickland v. Washington , 466 U.S. 668, 687, 104 S.Ct. 2052, 80 L.Ed. 2d 674 [1984] ).
*225Judicial scrutiny of counsel's performance in a claim of ineffective assistance of counsel is highly deferential and requires consideration of all the evidence before the judge or jury. The reviewing court must strongly presume that counsel's conduct fell within the broad range of reasonable professional assistance. State v. Kelly , 298 Kan. 965, 970, 318 P.3d 987 (2014). To establish prejudice, the defendant must show a reasonable probability that, but for counsel's deficient performance, the outcome of the proceeding would have been different, with a reasonable probability meaning a probability sufficient to undermine confidence in the outcome. State v. Sprague , 303 Kan. 418, 426, 362 P.3d 828 (2015).
If counsel has made a strategic decision after making a thorough investigation of the law and the facts relevant to the realistically available options, then counsel's decision is virtually unchallengeable. Strategic decisions made after a less than comprehensive investigation are reasonable exactly to the extent a reasonable professional judgment supports the limitations on the investigation. State v. Cheatham , 296 Kan. 417, 437, 292 P.3d 318 (2013) (citing Strickland , 466 U.S. at 690-91, 104 S.Ct. 2052 ).
The State's arguments are unpersuasive .
The State first argues that the district court erred in finding that Robinson's right to the effective assistance of counsel was violated because his defense attorneys failed to obtain an expert to assist in cross-examining or refuting the claims made by the State's fire investigators. But that is not exactly what the court held.
To obtain relief, the first thing Robinson had to show is that his counsel's representation fell below an objective standard of reasonableness, considering the entire circumstances of the case. See Bledsoe v. State , 283 Kan. 81, 90, 150 P.3d 868 (2007). The court here gave a point-by-point analysis of this deficient performance. First, it found that the defense attorneys were ineffective for failing to present sufficient expert testimony to refute claims made by the State's fire investigators. By not contacting an expert until less than two weeks before trial, there was little time to prepare a defense. Then, when that expert proved not to be an expert, he merely provided some help with possible cross-examination of the *226State's expert. Then, when the alleged expert's assistance proved to be unhelpful, the defense attorneys did not even consider talking to another expert. Thus, the district court held that the defense attorneys performed an insufficient investigation into the use of an arson causation expert.
To us, the State argues that "the initial question is: Was Mr. Huerter's decision to use Mr. Gietzen as a consulting expert objectively reasonable?" It suggests the answer is yes. The State maintains that Huerter decided only to use Gietzen's help with formulating cross-examination questions for the *235State's fire expert because Gietzen could not contradict the State's expert's opinion and the State could use him as a witness if he wrote a report. The State also argues that Huerter gave a sufficient explanation for how he chose Gietzen as a consulting expert. Basically, the State argues all of this resulted from the considered judgment of an experienced defense counsel.
But that is not the correct question to consider at this point. The district court did not find that Huerter was ineffective for failing to use Gietzen as an expert. Instead, the court found Huerter's performance deficient because he contacted Gietzen less than two weeks before trial and then did not talk to another expert when Gietzen proved unhelpful. Huerter still could have cross-examined Agent Monty to try to get him to admit the flaws in his own report even if he had contacted another expert.
Next, the State argues that the defense attorneys did not have to get an expert to testify to provide adequate representation of Robinson simply because the State had an expert. But the district court never said that. Instead, the district court found that Huerter's efforts on trying to procure an expert witness were insufficient in this specific situation.
In an odd interpretation of what the district court held, the State argues that the court erred in relying on the fact that Huerter did not spend more time or ask for a continuance to find an expert. The State claims that he needed not spend more time because Gietzen was not helpful. What the district court found, however, was that because Gietzen's assistance was not helpful, Huerter should have spent more time looking for someone who would be helpful.
*227None of the State's arguments shakes the validity of the district court's finding that the defense attorneys simply made an insufficient investigation here. There is no evidence of a comprehensive investigation by defense counsel of a most important aspect of this case-fire cause and origin expert opinions. Huerter started his battle with the State unarmed and unequipped with the expertise Robinson needed for a defense.
We realize that no two attorneys prepare for a trial in the same way. Often the circumstances of a prosecution offer few avenues for a defense. But to realize what factual hurdles exist for the defense requires a thorough investigation of those circumstances by the lawyer charged with the responsibility to craft a defense. Even though small details may often win the day, it takes an understanding of the overall facts to realize this. In a prosecution that must rely on cause and origin expertise, an understanding of why this is so is foundational. The difference in how the federal and state counsel prepared Robinson's defense highlights our point.
The federal public defender expended the effort to read extensively the technical literature of arson investigation before discussing the matter with an arson expert who had expertise in such investigations. This technical information allowed him to ask questions that led to areas that a defense could be built on. The State defense attorney allowed his law clerk to discuss the matter just before trial and settled on a few pages of material for cross-examination from an expert who had no fire cause and origin expertise. It is a small wonder that the defense could not shake the opinions of the State's expert because the defense expert could not render an opinion on how the fire started. This lack of effort by counsel ceded the field to the prosecution.
We acknowledge that if counsel has made a strategic decision after making a thorough investigation of the law and the facts relevant to the realistically available options, then counsel's decision is virtually unchallengeable. But choices made by counsel after a less than complete investigation are reasonable only if the decision to limit or forego certain investigations is reasonable under the circumstances. Cheatham , 296 Kan. at 437, 292 P.3d 318 (citing *228Strickland , 466 U.S. at 690-91, 104 S.Ct. 2052 ). Under the facts of this prosecution, counsel's investigation was not reasonable.
Robinson has shown us that Huerter failed to provide a valid strategic reason for simply relying on notes from Gietzen as a proper preparation for the defense. He also argues *236that Huerter's claim that his use of Gietzen was strategic because Gietzen could not contradict the State's expert is unsupported by the facts. This is true. Robinson validly contends that Gietzen was not qualified to make a cause and origin evaluation, so his opinion could not have been the same as the State's expert. This is also true.
The State once again argues that defense counsel did not have to retain an expert just because the State retained an expert. The district court, however, did not find that Huerter should have found an expert to testify at Robinson's trial. Instead, it found that under all the circumstances, Huerter failed to make a reasonable effort to find an expert witness to assist in his defense and his efforts constituted no comprehensive investigation of the expert witness issue.
Finally, the State argues that the district court's finding that Gietzen "was, in fact, not an expert at all in the essential subject matter of the trial" is unsupported by the facts and is a legally incorrect finding. We do not agree. The only experience or training Gietzen testified about was that he had attended an FBI arson analysis class on laboratory analysis of arson debris and he had worked as a crime scene investigator in homicides that involved fire. He had no training or experience in determining either the cause or origin of a fire. The district court's finding that Gietzen was not a qualified expert is supported by substantial competent evidence, as is the district court's ultimate finding that Huerter's performance was deficient.
Prejudice to the defense
Again, the district court was presented with startling contrasts. After finding that Robinson's trial counsel's representation fell below an objective standard of reasonableness, the court determined that this deficient performance prejudiced Robinson. It based this conclusion largely on Bieber's testimony at the habeas corpus hearing in which he vividly painted the deficiencies in Agent Monty's trial testimony.
*229The district court, in ruling on prejudice to the defense, astutely relied on the words of this court in Robinson's direct appeal. In finding prejudice, the district court underlined the importance of the testimony from the fire expert. In affirming his conviction, this court noted in the appeal: "An arson investigator testified the second fire appeared to have been set on or immediately next to the stairway using a flammable liquid as an accelerant." 2012 WL 4794455, at *3. The court determined that this conclusion permitted the jury to disregard Robinson's "equivocations and qualification to the detective about whether he started the fatal fire." 2012 WL 4794455, at *3. But if that evidence is not to be believed, then there is nothing left in the State's case against Robinson. It is opinion evidence that suggests the fire was intentionally started and opinion evidence that suggested Robinson started it. If those opinions are impeached, a finding of guilt becomes highly questionable.
Seizing upon this holding, the district court also determined that much of Bieber's testimony at the habeas corpus hearing supported a finding that Robinson was prejudiced by the defense attorneys' failure to obtain an arson expert. The court noted that Bieber testified at length about the deficiencies in Agent Monty's trial testimony. Ultimately, the district court ruled that "had trial counsel obtained the services of a qualified expert, be it for cross-examination purposes or testimony purposes, trial counsel would have been able to undermine Agent Monty's conclusions far more effectively than he was ultimately able to [do]."
The State argues that Robinson failed to meet the Strickland standard of proving it is reasonably likely the result would have been different if Robinson's attorneys had consulted an arson expert. It contends that Bieber offered no opinion on the cause and origin of the fire, but simply stated that the cause and origin of the fire was undetermined. In other words, Bieber did not state that Agent Monty's conclusion was incorrect.
And the State points out that although Bieber challenged Agent Monty's use of the negative corpus method of determining cause and origin, the method was a " 'finding consistent with the NFPA 921' " until 2011. We are not persuaded. Even if the NFPA 921 still said that the negative corpus method *237was valid at the time of *230this trial, a knowledgeable expert could have helped the defense attorneys effectively challenge that method even in 2009. After all, it is a methodology of elimination and not a method based on the scientific method. It produces a hypothesis that cannot be proved. There is no physical evidence supporting its conclusion. It is a conclusion because any known alternative causes for the fire cannot account for the cause of the fire.
The State had the burden to prove beyond a reasonable doubt at trial that Robinson set this fire. Since there was no physical evidence of any chemical accelerants at the scene, Agent Monty's conclusion, given Bieber's testimony, is unpersuasive. The significance of the gas can found at the scene is also speculative. The speed of the fire could be caused by the balloon structure of the apartment house and lack of sheetrock on the underside of the stairs. When we consider all of this, we find that the district court correctly determined that there is a reasonable probability the jury would have reached a different result but for the deficient performance of Robinson's trial counsel. See Sola-Morales , 300 Kan. at 882, 335 P.3d 1162.
We summarily deal with the cross-appeal issues.
We agree with the district court that Robinson must receive a new trial based on the deficient and prejudicial performance by his defense counsel based on the lack of effort in developing any insight or strategy on the cause and origin evidence offered by the State. We also agree that the remaining claims raised by Robinson are not worthy of compelling a new trial. A brief review highlights our point.
In the district court's view, the decision not to move to suppress Robinson's statements to the police was a strategy decision by his trial counsel (a questionable strategy according to the judge). But the judge would not look back in hindsight and substitute her judgment for that of Robinson's attorney. We agree.
The judge also found no merit in Robinson's argument that his counsel's failure to impeach trial witnesses Wheeles, Hill, and Roberts was a reason to order a new trial. Likewise, since Robinson never presented any specific facts or information that any possible alibi witness could offer that would have helped his case, the judge *231ruled that counsel's failure to track down alibi witnesses to testify was not grounds for a new trial. We find no error here.
Moving on, the court rejected Robinson's contention that other witnesses could rebut Brown's testimony that he saw Robinson fleeing the fire. After a careful review, the court decided that their testimony did not contradict Brown's testimony. And because the record showed that Robinson waived his right to testify, the court rejected his contention that his counsel was deficient for failing to call him in his own defense.
Finally, the court held that the written stipulation made by counsel-that the owner of the building did not consent to the fire-resulted from trial strategy and not from a deficient performance by the lawyer and was not grounds for a new trial. We have not been shown how this stipulation harmed the defense. After all, this stipulation prevented another witness from testifying to the jury about how devastating this fire was.
Since we find counsel's errors have prejudiced Robinson, we need not consider any of the other ineffective assistance of counsel allegations. See State v. Stinson , 43 Kan. App. 2d 468, 469, 227 P.3d 11 (2010). Stinson held that because the court was reversing and remanding for a new trial, the remaining argument on ineffective assistance of counsel was moot.
Affirmed.