dissenting.
INTRODUCTION
I respectfully dissent from today's order dismissing Kevin Pease's petition for hearing as improvidently granted. During its deliberations, the jury that found Pease guilty of two counts of murder and other serious felonies conducted an unauthorized experiment to test the eyewitness identification evidence that tied Pease to the crimes charged. When the experiment was revealed some years later, the superior court judge who had presided over Pease's criminal trial granted him a new trial. In my view, we must defer to the superior court's fact findings. The identification evidence was critical to Pease's convictions. The experiment was flawed. And there was at least a "reasonable possibility" it affected the outcome. We should therefore reverse the ruling of the court of appeals1 and remand for reinstatement of the superior court order granting Pease a new trial.
FACTS AND PROCEEDINGS
Kevin Pease and three others (Marvin Roberts, Eugene Vent, and George Frese) were charged with the robbery and assault of Franklin Dayton and the separate robbery, sexual assault, and murder of a juvenile, J.H., all on October 11, 1997 in Fairbanks.2 After a change of venue, Pease was tried before an Anchorage jury in the summer of 1999.
A prosecution witness critical to all of the charges against Pease was Arlo Olson. Olson testified at trial that while he was smoking outside a Fairbanks reception hall at around one o'clock in the morning, when only streetlights lit the attack seene, he saw Pease, Frese, Roberts, and Vent attack Dayton. He also testified that he had gotten "stoned" earlier that day, consumed alcohol that evening, and was "feeling a buzz" around the time of the attack. Although Olson did not approach Dayton after the attack, Olson testified that "[Jooking down the street, I recognized [Dayton]"; he also testified that he had known Frese and Vent "for a while" and "recognized them right off" and "recognized the car they were in." He told the jury he had seen the four men in the car outside the reception hall earlier that night. Olson's testimony tying Pease to the Dayton assault also linked Pease to the crimes against J.H. by placing the four assailants together the night of the attacks.
A police officer testified that he had measured the distance between the reception hall steps and the place where Dayton reported he was attacked as 550 feet.
A defense expert, Dr. Geoffrey Loftus, testified at trial about the difficulty of identifying people at significant distances. He told the jury that "a hundred to 200 feet is ... about the limit in terms of ... your ability to recognize a person, distinguish an individual's features that would allow you to recognize the person later on." This limit applies, *306he said, even under "the best cireumstances . in which you had unlimited time, good lighting and ability to pay attention to the person and be in an optimal physiological state ... [including being] awake, alert, not ill, not drunk, not under the influence of any drugs and so on."
The Anchorage jury found Pease guilty of two counts of second-degree murder, one count of second-degree assault, and one count of first-degree robbery. Pease's conviction was affirmed by the court of appeals in 2002.3
In 2008 Pease sought post-conviction relief on the newly discovered ground his trial jury had conducted an unauthorized experiment during its deliberations. The experiment appeared to address the ability to identify persons at significant distances. Superior Court Judge Ben Esch, the same judge who had presided over Pease's criminal trial, was assigned Pease's post-conviction relief application. The superior court ordered that an evidentiary proceeding be conducted.
In 2004 Pease took depositions of nine jurors and of the bailiff who took the jury outside the court building when it conducted the alleged experiment. The depositions established that the entire jury left the court building with the bailiff and that one or two jurors walked down the street from the group while the remaining jurors watched. The jurors did not all describe the purpose of the experiment in the same way. At least one discussed ascertaining what the distances mentioned at trial looked like. Others talked about assessing whether they could recognize at a distance the jurors who had walked away. Some indicated that they were testing Olson's testimony in particular: one juror stated that "one of the most credible witnesses [the state] had was a gentleman who believed he saw one of the defendants, and all the way through [the defendants were] questioning it, that distance, can you see"; another said the experiment "[wlas just to see what someone would look like at 500 feet away" because "one of the most critical two or three elements of the whole thing was-was [Olson's testimony]"; a third said "part of the testimony was that an incident had happened at that distance from the wedding reception" and the jurors were making their own determination as to whether it was true that "it would be difficult to see accurately at that distance." Several jurors remembered that it was daylight, "really bright," or sunny outside when they made their observations. Most seemed to recall or assumed that the bailiff had given permission for the jury to go outside, and several seemed to think or assume the bailiff or even the trial judge had permitted the jury to go out to make its observations. The bailiff testified that he assumed he had permitted the one juror to walk down the street so long as he kept in sight, but testified he would have told the judge if the jurors had told the bailiff what they were doing. The bailiff remained nearby, and one juror recalled the bailiff watching them.
Pease also submitted an affidavit from an investigator who measured the distance from the place where the jurors had been standing during the experiment to the place where the juror or jurors who walked away stood; the investigator stated that the distance was approximately 484 feet.
In August 2004 the superior court ruled that Pease was entitled to summary judgment and granted him post-conviction relief and a new trial. The court found that the jury was not just "getting an appreciation of the distance," but was also "trying to test the validity of [Dr.] Loftug'[s] testimony" and "impliedly, assessing the validity of the testimony of [eyewitness] Arlo Olson." The court relied on Gorz v. State4 in apparently concluding that the experiment violated Pease's rights to confrontation, cross-examination, and counsel. It reasoned that Pease was not present to object to the jurors' potentially inaccurate "understanding of the relevant distance," to any problems with the viewing conditions, or to the differences between the cireumstances of the experiment and the circumstances prevailing when Olson made his observations. The court stated that the Ic, & jury's "range of recollections could have been *307eliminated i#f ... the court made sure that the actual conditions were more similar to those testified to by Olson."
The superior court next considered "the effect, if any, of this misconduct." It quoted West v. State5 for the proposition that a "verdict should stand unless the evidence clearly establishes a serious violation of the juror's duty and deprives a party of a fair trial" 6 and Swain v. State7 for the proposition that "[the ultimate issue of influence on the juror is resolved by reference to the substantial likelihood test, an objective standard."8 The superior court found "that resort to an unapproved demonstration was likely to have influenced the jurors' deliberations." Finally, it noted that the state had not offered any admissible evidence showing that there was a genuine issue of material fact regarding the experiment, so summary judgment on Pease's application was appropriate. The superior court therefore granted Pease a new trial.
A divided court of appeals reversed, explaining that the superior court had found that "the cireumstances surrounding the experiment show[ed] that the jurors were not attempting to re-create Arlo Olson's nighttime observation of the assault on Franklin Dayton. Rather, the jurors were attempting to test Dr. Loftus's general assertion about the limits of human perception.9 It explained that the experiment was "substantially similar" to the event because the "obvious dissimilarities" in the cireamstances were "irrelevant" to the purpose of the experiment.10
We granted Pease's petition for hearing and ordered full briefing. This court today dismisses Pease's petition for hearing as improvidently granted. I would reverse the decision of the court of appeals and remand for reinstatement of the superior court's new trial order.
DISCUSSION
This petition requires us to review the superior court's assessment of Pease's post-conviction relief application in light of the same judge's first-hand understanding of the importance of Olson's eyewitness identification testimony to all the crimes charged.11 There are three essential points about the experiment: it was conducted to evaluate evidence critical to Pease's participation in the crimes charged; it was both inherently flawed and potentially persuasive; and it was not authorized by the superior court in advance nor was it acceptable to the superior court when it learned of the experiment. Given the critical importance of Olson's testimony to the charges against Pease, there is a reasonable possibility the experiment affected the verdict.
A. The Experiment Tested Eyewitness Identification Testimony Critical to Pease's Convictions.
Pease argues that the court of appeals mischaracterized the superior court's conclu*308sions by relying on an incorrect premise that the jury was only testing Dr. Loftus's testimony; he also argues that the superior court correctly described the jurors' experiment as testing the testimony of both Olson and Dr. Loftus. The state responds that the court of appeals did not inaccurately describe the superior court's ruling.
I think the court of appeals made two related errors that misdirected its analysis. First, its description of the superior court's findings was incomplete in an important respect. Second, it drew a distinction between permissible and impermissible experiments that is unsustainable and irrelevant. And as to both errors, it failed to defer to the superi- or court's findings concerning the relationship between the experiment and critical trial evidence, and concerning the experiment's possible effect.
The court of appeals described the jurors' purpose as follows: "As Judge Esch found, the circumstances surrounding the experiment show that the jurors were not attempting to re-create Arlo Olson's nighttime observation of the assault on Franklin Dayton. Rather, the jurors were attempting to test Dr. Loftus's general assertion about the limits of human perception."12
That description is incomplete because it does not acknowledge the more-nuanced findings the superior court actually made. The superior court found that: "[what the jury was doing, in addition to getting an appreciation of the distance, was trying to test the validity of Loftus'[s] testimony.... They were also, impliedly, assessing the validity of the testimony of Arlo Olson when he stated that he could recognize two individuals at that distance." The superior court repeated this finding-stating that "the jury was using the mutual observation experience to attempt to check Olson's testimony"-in discussing its legal significance. In effect, the superior court found that the jurors were evaluating the eredibility of both Olson's testimony that he was able to identify Dayton's attackers and Dr. Loftus's testimony that it is impossible to recognize a person from more than 200 feet away.
These findings are at least permitted, if not required, by the jurors' testimony and are the only reasonable explanation of what the jurors said they were doing.13 At least one juror referred in his deposition to Olson by name, and others clearly indicated that they had been testing Olson's identification testimony. That the jurors, in conducting their experiment, far exceeded the 200-foot distance discussed by Dr. Loftus confirms that they were directly testing Olson's identification testimony. Some jurors did not mention Olson and some described potentially more limited purposes. But likewise, few mentioned or described Dr. Loftus, either, and the state offered no contrary admissible evidence and has not argued that there was a genuine factual dispute material to Pease's post-conviction relief application.
The appellate court's incomplete description caused it to misfocus on the issue of the experiment's validity in testing Dr. Loftus's opinions, and to overlook the issue of its invalidity in testing Olson's eredibility.
The court of appeals relatedly drew a distinction that is both unsustainable and irrelevant. It implicitly reasoned that the testimony of prosecution eyewitness Olson could be isolated from the testimony of the defense expert, and that jurors could test the expert's categorical opinions without simultaneously testing Olson's long-distance identification.14 Thus, the court of appeals stated that "the main objective [of the experiment] was to test Dr. Loftus's assertion about human perception under optimal conditions."15 But Dr. Loftus was a defense witness whose expert opinions squarely challenged Olson's ability to identify Pease and Dayton. The superior court correctly observed that "[the primary focus of Dr. Loftus'[s] testimony was the impeachment of Olson." As Pease asserts, the testimony of the two witnesses was "inextricably linked." Their testimony was *309also inherently inconsistent. There is no basis in the record for thinking the jurors were testing the testimony of only one of these two witnesses and not the fundamentally inconsistent testimony of the other.
It should also be important that the trial judge, who was best able to understand the relationship between the Olson testimony and the Loftus testimony, did not consider any such exacting distinetion important when he granted a new trial. To the extent such a distinction affected the analysis of the court of appeals, the distinction is chimerical and the resulting analysis wrong.
There is no legitimate or material factual dispute about why the jurors conducted the experiment or the importance of Olson's identification testimony.
B. The Experiment Was Both Inherently Flawed and Potentially Persuasive.
The conditions under which the experiment was carried out were gravely dissimilar to the conditions during the attack on Dayton. The superior court stated that "the differences between the night setting with artificial light in Fairbanks and the conditions in Anchorage are very troubling." The record confirms that the lighting during the experiment was nothing like the lighting during the attack. The attack occurred at around 1 a.m. in Fairbanks on October 11. Streetlights apparently provided some lighting. The jury's experiment took place during daylight hours in Anchorage in mid-August, and several jurors testified that it was daylight, "really bright," or sunny.
The distance at which the jurors conducted the experiment was also probably dissimilar for testing Olson's testimony. A police officer measured the distance from where Dayton reported being attacked to the place where Olson stood at 550 feet.16 The jurors' testimony and an affidavit submitted with Pease's application together indicated that the two groups of jurors were only 484 feet apart.17 This difference of more than one hundred feet requires a conclusion that the cireumstances of the experiment and the event were not substantially similar. If the jurors were markedly closer to each other than Olson was to Dayton, their experiment was flawed in a way that potentially prejudiced Pease by enhancing the credibility of Olson's testimony.
The experiment's methodology was also flawed. There may be a notion the jurors simply made the same observations they permissibly could have made from the jury room windows. But observations from the windows of the jury room would likely have been of strangers on the streets, and would certainly not have been of fellow jurors who had been sitting and conversing together in close quarters for at least two weeks during selection, trial, and the initial deliberations. Instead, here one or two jurors walked away from their colleagues in broad daylight and then stood at a distant point before returning.
Unlike almost all the experiments discussed in cases cited by the court of appeals,18 this experiment was not conducted in *310the jury room, and in fact could not have been conducted without leaving the court building. And because, as the bailiff explained, the jurors had to stay together, one or two could not have exited the building and walked down the street while the remaining jurors watched them from the jury room windows. One of the unique flaws of this experiment-the continued observation of fellow jurors as they paced off the estimated distance-could not have occurred had the jury restricted itself to making observations only from inside the jury room.
Also, all of the jurors would have recognized that observations from the jury room windows necessarily involved estimated distances and dissimilar conditions, but the formality of the group exercise in attempting to pace off the distance could have influenced jurors by adding a spurious sense of experimental precision.
Such a test was not a valid test even of Dr. Loftus's "optimal conditions" opinion because the test was unacceptably suggestive. The remaining jurors had watched their known colleague or colleagues walk away. It is not surprising one juror would testify that others remarked that "I can tell that is 'so and so.' " Had the prosecutor proposed such an experiment, Pease's trial lawyer could have raised a persuasive objection. And had the court nonetheless allowed the experiment, Dr. Lof-tus surely could have addressed its flaws and defense counsel could have sought a limiting instruction. But because the experiment occurred during deliberations and was not revealed for some years, neither Pease nor his expert had an opportunity to evaluate and confront this juror-developed evidence.
Finally, we should conclude that the jury's collective conduct in going outside and carrying out the test potentially made this self-generated evidence particularly important and relevant to the jurors. The jury's collegial physical excursion for a shared purpose likely increased the experiment's seeming validity and communal importance. It is predictable and likely that the way the jurors conducted the experiment heightened their willingness to give it weight. That is especially so given that several jurors testified that they assumed that the excursion had been permitted by the bailiff or the trial judge, and all of the jurors must have been aware that the bailiff was present during the experiment.
C. The Experiment Was Not Authorized by the Superior Court and Was Not Acceptable to the Superior Court when It Learned of the Experiment.
Of course, the experiment had not been authorized by the superior court.
The court of appeals thought that "given the circumstances, reasonable judges would have allowed the experiment if the jurors had asked."19 Pease argues to the contrary. He contends that under the analysis required by Love v. State, the court would have had to conduct a hearing to determine whether "the experiment's and the event's conditions were substantially similar" before admitting the evidence.20 He reasons that because the conditions of the experiment were unlike those when Olson viewed the attack-"[tlhe scene, line of sight, distance and lighting were all different"-the jurors' "experiment would not pass the test set forth in Love." The state argues that the experiment would have been acceptable because "whether an experiment is 'substantial[ly] similar[ J and therefore admissible under Love hinges on what the experiment is offered to prove," and because the jury was merely testing Dr. Lof-tus's statements about recognition in good conditions. The state does not argue that the experiment would have been appropriate to test Olson's testimony.
Whatever the experiment's purpose, I think its suggestive methodology altogether foreclosed this experiment. In any event, the experiment did not have the exclusive *311purpose of testing Dr. Loftus's opinions.21 As the superior court found, the jury was at least in part testing Olson's testimony. The experiment would not have satisfied Love if offered in part for that purpose. That also seems to have been the implicit view of the superior court. That differences in the ambient light conditions were obvious to the jurors did not render obvious other potentially significant dissimilarities, such as differences in the distances between the groups of jurors and between Olson and Dayton, and possible measurement errors during the experiment. In my view, a reasonable judge would not have allowed this experiment.
D. There Was a Reasonable Possibility the Misconduct Affected the Ver-diet.
Before granting a new trial, a court considering a post-conviction relief claim that the jury conducted an unauthorized experiment must conclude that there is a reasonable possibility the experiment affected the verdict.22 I conclude that there is here. Olson's testimony was critical in linking Pease to both victims because it connected the four assailants that evening. The superior court explained that "[mJore than sixty witnesses testified and Arlo Olson was the only one who identified the defendants as Dayton's assailants and one of only a few who placed the four together on the night of the erimes." The superior court also found that
[iJn closing argument, all counsel discussed the testimony of [Olson and Dr. Loftus] and attempted to bolster or denigrate its value. The court cannot find that an experiment that attempted to resolve the validity of this difference [between Olson's and Dr. Loftus's testimony] was not likely to have influenced the verdict.
The judge who presided over the criminal trial and heard firsthand the extensive and contradictory witness testimony and the other trial evidence was in the best position to assess the likely effect of the experiment on the jury's deliberations. In assessing the potential effect of the experiment on the verdict, the superior court found that "resort to an unapproved demonstration was likely to have influenced the jurorg' deliberations." It also stated that it could not find that the experiment "was not likely to have influenced the verdict." We should defer to those rulings. Because there is no good reason to reject the superior court's assessment, we should accept it as valid.
The superior court did not apply the "reasonable possibility" standard in assessing the legal significance of its findings.23 But the court's statements plainly indicate that the applicable standard is satisfied. An experiment that "is likely to have influenced" the deliberations likewise necessarily has a "reasonable possibility" of having affected the verdict. There might be some dispute about exactly what particular jurors wished to test, but there is no genuine dispute about what the jurors did and that their goal was to assess the critical identification evidence in some way. There might also be a genuine dispute about the experiment's actual effect on the outcome, but this question is immate*312rial given the superior court's finding about the experiment's likely effect.24
CONCLUSION
Because the jury engaged in misconduct in evaluating the crucial eyewitness testimony, there is a reasonable possibility that the misconduct influenced the verdict. I would consequently reverse the ruling of the court of appeals and remand for reinstatement of the new trial order entered by the superior court.
. State v. Pease, 163 P.3d 985, 995 (Alaska App. 2007).
. Pease v. State, 54 P.3d 316, 319 (Alaska App. 2002).
. Id. at 332.
. Gorz v. State, 749 P.2d 1349 (Alaska App.1988).
. West v. State, 409 P.2d 847 (Alaska 1966).
Id. at 852.
. Swain v. State, 817 P.2d 927 (Alaska App.1991).
. Id. at 932 (quoting ABA Stanparps ror CRIMINAL JUSTICE Commentary to § 8-3.7).
. State v. Pease, 163 P.3d 985, 993 (Alaska App. 2007). Judge Stewart dissented. Id. at 996-1000. In my view, his dissent is highly persuasive, as is the memorandum decision of the superior court.
. Id. at 994 (citing Bierria v. Dickinson Mfg. Co., 36 P.3d 654, 658-59 (Alaska 2001)).
. A post-conviction relief action is a civil proceeding that is separate from the underlying criminal case; we treat this petition as we would other civil appeals or petitions. See State v. Hannagan, 559 P.2d 1059, 1062 (Alaska 1977) (holding that state has right to appeal in post-conviction relief proceeding because such proceedings are separate from original criminal proceeding and are governed by civil procedure).
We "review[ ] a grant of summary judgment de novo and affirm[] only if there are no genuine issues of material fact and the moving party is entitled to judgment as a matter of law. In making this determination, we draw all reasonable inferences in favor of" the party against whom judgment was entered. Barr v. Goldome Realty Credit Corp., 46 P.3d 1004, 1006 (Alaska 2002) (citing Moore v. Allstate Ins. Co., 995 P.2d 231, 233 (Alaska 2000)).
We review questions of law de novo, using our independent judgment to "adopt the rule of law that is most persuasive in light of precedent, reason, and policy." Univ. of Alaska v. Alaska Cmty. Colls. Fed'n of Teachers, Local 2404, 64 P.3d 823, 825 (Alaska 2003) (quoting Guin v. Ha, 591 P.2d 1281, 1284 n. 6 (Alaska 1979)).
. Pease, 163 P.3d at 993.
. The court of appeals did not hold that these findings were clearly erroneous. It was therefore bound by them.
. Pease, 163 P.3d at 993-94.
. Id. at 994.
. One juror testified: "What I recall was that one of the jurors said that he was some kind of civil engineer or something, or planner, city planner, and that he knew how far these blocks were apart. And he said that that was about 500 feet, so that's why that corner was chosen." Another testified: "[Plart of the testimony was that an incident had happened at that distance from the wedding reception. ... And I think the contingent was that it would be difficult to see accurately at that distance. And so it was just a maatter of satisfying ourselves whether that was, in fact, true or not."
. The state notes that Olson described the attack as occurring in a different location than Dayton identified; the distance from Olson to that alternative position was 396 feet. Although that distance is somewhat less than the distance between the two groups of jurors, it seems that one juror indicated that they intended to be five hundred feet apart, suggesting that the jurors considered Dayton's testimony regarding his location more credible.
. Pease, 163 P.3d at 990-92. Only in People v. Smith, 59 N.Y.2d 988, 466 N.Y.S.2d 662, 453 N.E.2d 1079, 1080 (1983), did a juror's experiment take place outside the jury room. That sequestered juror evaluated a police officer's purported ability to see through the rear windows of cars by attempting to look at car interiors while on the way to dinner or on the bus to the hotel. The court found this to be "everyday experience and, therefore, not misconduct." Id. at 1080. *310All other cases cited involved experimentation inside the jury room.
. Pease, 163 P.3d at 993-94.
. See Love v. State, 457 P.2d 622, 627 (Alaska 1969) ("[Flor such evidence to be admissible, it should have been developed under conditions substantially similar to those surrounding the event at issue.").
. If permission for that exclusive purpose had been sought, a limiting instruction that might have kept the jury from misapplying its observations almost certainly would have been required.
. Pease argues that the appropriate standard for evaluating whether the misconduct mandates reversal of the verdict is the objective "substantial likelihood" test. Swain v. State, 817 P.2d 927, 932 (Alaska App.1991) (quoting ABA Stan-parps For Crimixat Justice, Commentary to § 8-3.7). Alternatively, he argues that the standard should be "whether it can be said beyond a reasonable doubt that the misconduct did not contribute to the verdict." Gorg v. State, 749 P.2d 1349, 1355 (Alaska App.1988). The state argues for adoption of the "actual prejudice" standard articulated in the federal habeas context. Brecht v. Abrahamson, 507 U.S. 619, 637, 113 S.Ct 1710, 123 L.Ed.2d 353 (1993). Because this case is neither a habeas appeal nor a direct appeal such as Gorz and Swain, the "reasonable possibility" standard applies. Grinols v. State, 10 P.3d 600, 618-19 (Alaska App.2000), aff'd, 74 P.3d 889 (Alaska 2003) (ruling on other issues) (recognizing that defendant was not entitled to post-conviction relief without a showing of both error and prejudice, defined by court of appeals as "a reasonable possibility that [the absence of the error] would have affected the outcome of the underlying criminal proceeding" (citing State v. Steffensen, 902 P.2d 340, 343 (Alaska App.1995))).
. Grinols, 10 P.3d 600.
. The state has not argued here that the counts against Pease that arose from the attack on J.H. should stand even if those based on the attack on Dayton are reversed. It is undisputed that OH-son's identification of Pease was critical to proving all of the charges against Pease.