KAYATTA, Circuit Judge.

Kathryn Hayne Barnwell for appellant.

Margaret D. McGaughey, Assistant United States Attorney, with whom Thomas E. Delahanty II, United States Attorney, was on brief, for appellee.

Opinion

KAYATTA, Circuit Judge.

A jury convicted Brunel Constant of illegally possessing a firearm after he had been previously convicted of a crime punishable by imprisonment for a term of more than one year. See 18 U.S.C. §§ 922(g)(1), 924(a)(2). Sentenced to 74 months of imprisonment, Constant now appeals. His principal challenge is to the denial of his pretrial motion to bar a government witness, Adam Dennis, from identifying Constant at trial. He also challenges (1) trial defense counsel's pretrial advice concerning application of the Armed Career Criminal Act (“ACCA”) to his case; (2) the district court's denial of an acceptance of responsibility reduction; (3) the presence of the government's testifying case agent at government counsel's table during trial; and (4) the preponderance of the evidence standard employed by the district court at sentencing to find that Constant had used the firearm to commit another felony offense.

For the following reasons, including the fact that the police videotaped Dennis's pretrial identification of Constant in an otherwise problematic photo array, we affirm the conviction, but vacate the sentence.

I. Background

The charge in this case emanates from a shooting that occurred in the pre-dawn hours of August 19, 2011, in Lewiston, Maine. Earlier that morning, Adam Dennis and Alan Roy had been hanging out in a three-story, six-unit apartment building on Walnut Street. As Dennis began to head home at roughly 4 a.m., he and Roy encountered a man sleeping on the back porch of the first-floor apartment belonging to Roy's mother, Jeannette Cloutier, and his sister, Nancy Cote. Dennis roused the man, and an argument between the two ensued.

The argument, which lasted approximately five to ten minutes, moved from the back porch of the apartment through a lit hallway to the front porch. During this time, the man mentioned that Cote owed him money. Ronald Coleman, another first-floor resident at the apartment building, did not view the encounter but heard someone swear “I'll be back.” Sometime thereafter, still pre-dawn, Coleman heard two gunshots and observed that the shooter was a black man with long braids who was wearing a white tank top. The two bullets had been fired into Cloutier and Cote's first-floor apartment. Because he had heard a loud sound, saw a flame, and did not hear casings fall to the ground, Coleman described the gun as a revolver.

Later that morning, Detective Derrick St. Laurent—then an officer with the Lewiston Police Department—began investigating the shooting. He interviewed Roy, who, after making some inquiries, gave St. Laurent the name “Jamal” and a general location on Bradley Street. St. Laurent subsequently canvassed that area and identified a particular apartment in which a man fitting the general description of the shooter lived with his girlfriend. When St. Laurent and other law enforcement personnel approached this apartment, they encountered Constant, a black man with long dreadlocks wearing a white tank top. Detective James Theiss, who was investigating the matter with St. Laurent, obtained Constant's identification. After determining that there existed a warrant for Constant's arrest, the officers placed Constant under arrest and transported him back to the police station.

During this time, St. Laurent spoke to Constant's girlfriend, who consented to a search of the apartment. The search uncovered a revolver hidden underneath the back porch rafters. Later testing of a slug found inside Cloutier's microwave confirmed that that revolver could have been the firearm from which the slug was shot. St. Laurent and Theiss then returned to the police department and interviewed Constant. Constant denied being involved in the shooting, but eventually confessed to holding for an acquaintance the gun found under the porch rafters.

St. Laurent thereafter requested and received six photographs for a photo array identification. While all of the individuals depicted in the array were black men with dreadlocks who appeared to be between the ages of 18 and 40, no one other than Constant had dreadlocks extending well below his shoulders or was wearing a white tank-top. St. Laurent presented the array to a number of persons, including Dennis. Dennis viewed the array on the afternoon following the shooting. St. Laurent recorded the entire viewing and the associated interview of Dennis on video.

As depicted in the video recording, at the start of the interview St. Laurent informed Dennis that “it looks like we got the guy that did it.” St. Laurent then asked Dennis a series of questions about that night. Dennis described the man on the porch as a “black guy” with “dreadlocks” who was wearing a baseball hat and jewelry. St. Laurent asked Dennis if he could identify the man from a photo lineup; Dennis replied “I guess so.” Just before Dennis viewed the array, St. Laurent, holding two manila folders in his hand, removed Constant's photograph from one folder and transferred it without apparent concealment to a second folder as Dennis looked on closely. St. Laurent then removed six photos from the second folder, placing them on the table, centering Constant's photo directly in front of Dennis.

Dennis, after viewing the array for a few seconds, singled out Constant's photo and stated, “I'm guessing it's him, that would be the one I'd be putting my money on, either him or him [indicating another photo].” St. Laurent thereupon cut off any further consideration by tapping Constant's photo with his finger and asking, “So you think it's him right here?” When Dennis reflexively answered, “Yeah,” St. Laurent grabbed a pen and quickly had Dennis sign and date Constant's photograph. Immediately after Dennis signed, St. Laurent told Dennis that the individual he chose was the suspect police had in custody.

After two pretrial hearings and a viewing of Dennis's recorded identification, the district court denied Constant's motion to suppress Dennis's in-court identification of Constant as the man with whom he had argued on the night of the shooting. The district court found that while the photo array shown to Dennis was unduly suggestive, his identification was nevertheless not so unreliable as to require its exclusion.

At trial, the government called only four witnesses: Dennis, Coleman, Theiss, and St. Laurent. The government also introduced a video recording of Constant's interview with St. Laurent and Theiss, which it played to the jury. On direct examination, Dennis identified Constant as the man with whom he argued on the porch on the evening in question. After the defense cross-examined St. Laurent, it introduced the video recording of Dennis's interview with St. Laurent, including the photo array procedure, and played it to the jury.

The jury found Constant guilty of being a felon in possession of a firearm, and the district court sentenced him to 74 months of imprisonment. This timely appeal followed.

II. Discussion

A. Dennis's In–Court Identification

We agree with the district court that the identification procedure used in this case was impermissibly suggestive. Nor does the government now claim otherwise. This leaves us to answer the pivotal question of whether Dennis's subsequent in-court identification should have been excluded as the unreliable artifact of the impermissibly suggestive pretrial identification procedure.

The Supreme Court has several times considered whether witness identifications that follow impermissibly suggestive police conduct must be excluded in order to maintain due process. See Perry v. New Hampshire, –––U.S. ––––, 132 S.Ct. 716, 723–24, 181 L.Ed.2d 694 (2012) (summarizing cases). In a nutshell, whether the identification evidence must be excluded turns on a case-by-case assessment of the reliability of the identification notwithstanding the suggestive actions of the police. Manson v. Brathwaite, 432 U.S. 98, 114, 97 S.Ct. 2243, 53 L.Ed.2d 140 (1977) (“[R]eliability is the linchpin in determining the admissibility of identification testimony....”). In plain terms, we distinguish between, at one end of the spectrum, a witness who would have easily identified the defendant without the suggestive police misconduct and, at the other end of the spectrum, a witness whose identification is very likely simply a product of that suggestion. 3 The factors to be considered include:

[1] the opportunity of the witness to view the criminal at the time of the crime, [2] the witness' degree of attention, [3] the accuracy of the witness' prior description of the criminal, [4] the level of certainty demonstrated by the witness at the confrontation, and [5] the length of time between the crime and the confrontation.

Neil v. Biggers, 409 U.S. 188, 199–200, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972). The corrupting effect of the unduly suggestive procedure is then weighed against an analysis of these factors. Manson, 432 U.S. at 114, 97 S.Ct. 2243.

If this weighing points to “a very substantial likelihood of irreparable misidentification,” the identification evidence must be suppressed. Perry, 132 S.Ct. at 720 (quoting Simmons v. United States, 390 U.S. 377, 384, 88 S.Ct. 967, 19 L.Ed.2d 1247 (1968)). “But if the indicia of reliability are strong enough to outweigh the corrupting effect of the police-arranged suggestive circumstances, the identification evidence ordinarily will be admitted, and the jury will ultimately determine its worth.” Id. And because we usually entrust the jury with the responsibility of determining whether lay witness testimony is reliable, we have said that only extraordinary circumstances warrant the withholding of identification evidence from it. United States v. de Jesus–Rios, 990 F.2d 672, 677 (1st Cir.1993); see also Perry, 132 S.Ct. at 723 (“Only when evidence is so extremely unfair that its admission violates fundamental conceptions of justice have we imposed a constraint tied to the Due Process Clause.” (internal quotation marks and citation omitted)).

Some tension exists in our case law discussing the standard of review brought to bear in considering a district court's decision not to exclude identification evidence. In United States v. Jones, 689 F.3d 12 (1st Cir.2012), we looked for an “abuse of discretion,” thereby signifying “reasonable latitude for case-specific decisions” of this type. Id. at 18; see also United States v. Brown, 510 F.3d 57, 66 (1st Cir.2007) (reviewing a district court's decision to admit voice identification “for abuse of discretion”). More recently, we spoke in terms of “de novo” review, albeit while assaying factual findings for clear error. United States v. Espinal–Almeida, 699 F.3d 588, 602 (1st Cir.2012); see also United States v. De León–Quiñones, 588 F.3d 748, 753 (1st Cir.2009) (“Typically, the district court's ultimate decision to admit or suppress identification evidence is subject to a plenary, de novo standard of review, with underlying findings of fact reviewed for clear error.”).

This tension may be more apparent than real. In none of the cases employing the de novo standard did the standard make any difference; that is to say, in each case we affirmed under that stricter standard, and thus clearly would have affirmed under any other standard. In all cases, too, we reviewed de novo the articulation of the correct legal standard, while examining the underlying findings of fact only for clear error. Tension remains concerning only how we review the application of the correct legal standard to the facts of a specific case. And even that tension is less than it seems, as “abuse of discretion” in this context is “perhaps more misleading than helpful,” representing in substance an assessment of “reasonableness” in the district court's fact-bound application of the law. United States v. Bater, 594 F.3d 51, 54 (1st Cir.2010). All that remains unclear, in sum, is whether we ask whether the district court's application of law to fact was reasonable, or whether we ask whether we would have reached the same conclusion. We opt for the more deferential formulation. Simply put, gauging the reliability of a witness's testimony in a case like this is precisely the type of judgment that trial judges are both well-equipped and well-positioned to make.

Dennis had a significant, face-to-face, five-to-ten minute conversation with the subject less than twenty-four hours before he viewed the photo array. And, unlike the witness in de Jesus–Rios, Dennis's oral, unprompted description of the subject at least matched the defendant in its most salient respects. See de Jesus–Rios, 990 F.2d at 678 (wrong on height and race). On the other hand, even when doubly prompted to pick Constant—first, through the presentation of an unduly suggestive photo array and, second, through St. Laurent's improperly suggestive behavior during the presentation—Dennis clearly hesitated to settle confidently on Constant rather than another individual who did not look much like Constant.

In assessing the reasonableness of the district court's reliability determination, we think it significant that the entire photo array procedure containing all of the suggestive conduct is recorded in a video that allows any viewer to see and hear firsthand both the suggestive prompts and Dennis's response to them. In the ordinary case, the evidence about how the witness first identified the defendant consists largely of oral testimony, usually from the witness and/or the police. See, e.g., de Jesus–Rios, 990 F.2d at 678. Lost in such testimony are important, unspun details of tone, expression, timing, and body language. A jury might therefore quite reasonably and unwittingly assign the in-court identification more value than its actual provenance supports. Here, by contrast, the jurors' ability to see for themselves the original identification and all police prompting empowered them to assess more accurately the extent to which “the evidence should be discounted as unworthy of credit.” Perry, 132 S.Ct. at 723. In short, the existence of the video recording guarded against the harm that the district court might have otherwise guarded against by excluding the identification: misleading a jury into thinking that evidence is more probative than it really is. And the video guarded against that harm in a manner that both avoids overshooting the mark and relies on jurors to do no more than that which we routinely rely on them to do.

We note, too, that the evidence in this case entirely apart from Dennis's identification was very strong. Constant confessed to the only crime for which he was charged: possessing the revolver found under the porch rafters. While his trial counsel argued that the confession was false, that argument was a tough sell considering that the confession was video-recorded and evidence of a motive to falsely admit possession was thin. And while Constant was not charged with the shooting, the rough fit between the descriptions of the shooter and his gun matched that of Constant and the revolver he confessed to possessing, which added cause to regard the confession as very likely true.

For all of these reasons, we find that admission of the identification evidence did not violate Constant's due process rights.

B. Ineffective Assistance of Counsel

Constant next argues that his trial counsel was ineffective in advising him that he would be exposed to the ACCA's mandatory 15–year sentence. This claim is based on a letter from his trial counsel dated February 12, 2013—approximately 3 months before trial commenced—in which counsel advised him that he would face a mandatory minimum 15–year sentence if he pleaded guilty when, in fact, no such mandatory minimum sentence applied. That letter also refers to an earlier October letter, described in the February letter as having first rendered the opinion. But for that erroneous advice, Constant claims that he would have entered a straight or conditional guilty plea, likely triggering a straightforward application of at least a two-level reduction in his total offense level for acceptance of responsibility under U.S. Sentencing Guidelines (“U.S.S.G.”) § 3E1.1(a). Therefore, he reasons that he should be resentenced with the benefit of, at least, the two-level reduction to remedy the ineffective assistance of counsel.

To succeed on an ineffective assistance of counsel claim, Constant must show that (1) “counsel's representation fell below an objective standard of reasonableness,” and (2) “there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different.” Strickland v. Washington, 466 U.S. 668, 688, 694, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). In assessing prong one, we must make “every effort ... to eliminate the distorting effects of hindsight, to reconstruct the circumstances of counsel's challenged conduct, and to evaluate the conduct from counsel's perspective at the time.” Id. at 689, 104 S.Ct. 2052.

As we have often repeated, this court customarily will not entertain ineffective assistance of counsel claims on direct review, but will instead leave such fact-bound determinations to the trial court to decide in the first instance. See, e.g., United States v. Mala, 7 F.3d 1058, 1063 (1st Cir.1993). This is because “the trial judge, by reason of his familiarity with the case, is usually in the best position to assess both the quality of the legal representation afforded to the defendant in the district court and the impact of any shortfall in that representation.” Id. An exception to this rule exists where “the critical facts are not in dispute and the record is sufficiently developed to allow reasoned consideration of the claim.” Id. Here, though, the record is not sufficiently developed to allow us “to reconstruct the circumstances of counsel's challenged conduct, and to evaluate the conduct from counsel's perspective at the time.” Strickland, 466 U.S. at 689, 104 S.Ct. 2052. As Constant admits in his brief, while the prosecutor stated at sentencing that ACCA's application to this case turned on whether a juvenile adjudication in Constant's criminal history qualified as a predicate offense under the statute, the record is incomplete as to why trial defense counsel anticipated that the ACCA would apply. This information, along with testimony as to the exact advice that trial defense counsel gave to Constant, could be critical in assessing counsel's performance.

That leaves us with two options. As we usually do, we could require “ ‘that such claims “must originally be presented to the district court” as a collateral attack under 28 U.S.C. § 2255’ due to the paucity of the record and the district court's ‘better position to adduce the relevant evidence’ as to whether counsel's performance was deficient and whether such deficiency prejudiced the defendant.” United States v. Kenney, 756 F.3d 36, 48–49 (1st Cir.2014) (quoting United States v. Colón–Torres, 382 F.3d 76, 84–85 (1st Cir.2004)). Alternatively, in “special circumstances,” United States v. Vega Molina, 407 F.3d 511, 531 (1st Cir.2005), we have stated that where “the record is embryonic but ‘contain[s] sufficient indicia of ineffectiveness,’ we may opt to remand for an evidentiary hearing without requiring the defendant to bring a collateral challenge.” Kenney, 756 F.3d at 49 (alteration in original) (quoting Colón–Torres, 382 F.3d at 85). For three reasons, we opt for such a remand in this case.

First, and most importantly, we have significant “indicia of ineffectiveness.” The challenged advice at issue is written and unequivocal. While the government echoes the district court's conclusion that “[trial defense counsel] was basically advising in this letter that [Constant] might be [subject to the ACCA] and that it's possible that [he] wouldn't be,” the February 12 letter was plainly more definitive, and the government points to nothing else in the record that supports its claim that the advice only posited a possible outcome. The written, unequivocal advice was also both incorrect and material. The sentencing range under the Guidelines of 210 to 262 months as calculated by trial defense counsel in the February 12 letter far exceeded Constant's actual Guidelines range of 53 to 78 months. Based on this large disparity, trial defense counsel informed the court at sentencing that Constant's decision to proceed to trial “was driven by the ACCA,” and that “[h]is only hope to avoid [ACCA's 15–year minimum sentence] was to have a trial.”

Second, this is not a case in which the ineffective assistance claim calls into question a broad array of issues. We instead have here “an isolated and easily analyzed trial decision.” See Kenney, 756 F.3d at 49 (requiring the filing of a habeas claim where “the alleged deficiency ... did not consist of an isolated and easily analyzed trial decision”).

Finally, resolution of the Strickland claim may shed light relevant to the district court's exercise of its sentencing discretion and its denial of a downward adjustment for acceptance of responsibility. At sentencing, Constant argued that the circumstances here warranted a discretionary variance even if a reduction for acceptance of responsibility was not permitted under the Guidelines. The district court's rejection of that argument was likely predicated in large part on its understanding that trial defense counsel did not definitively advise Constant that he would be subject to the 15–year mandatory minimum and that Constant's decision to proceed to trial was “a real strategy call.” After considering the record as it now stands—specifically the February 12 letter—it is unclear how and why the district court came to such a conclusion. It may be that the advice contained in the October letter referenced in the February letter was less definitive, and that the district court was referring to that letter. The government, though, points to no such letter in the record before us. Having the Strickland claim resolved on remand will therefore allow the court to simultaneously revisit its discretionary ruling.

None of this is to preordain the outcome. As we have noted, the record—as is usual on Strickland claims—is not fully developed, containing neither the original October letter nor an explanation of counsel's thought process. Establishing ineffective assistance requires Constant to show that his trial counsel's performance “was not only substandard, but also ‘deficient in some way sufficiently substantial to deny him effective representation.’ ” Logan v. Gelb, 790 F.3d 65, 71 (1st Cir.2015) (quoting Epsom v. Hall, 330 F.3d 49, 53 (1st Cir.2003)). Courts considering such alleged deficiencies “must judge the reasonableness of counsel's challenged conduct on the facts of the particular case, viewed as of the time of counsel's conduct.” Strickland, 466 U.S. at 690, 104 S.Ct. 2052. Only if a court after this consideration finds that counsel's conduct was “outside the wide range of professionally competent assistance” may it grant an ineffective assistance claim. Id.

In remanding, we, nevertheless, expressly reject Constant's argument that his Strickland claim calls into question the conviction itself. Constant was convicted after a trial, while his Strickland claim is that, but for bad advice, he would have pleaded guilty. In either case, he ends up guilty of the same charge. The only question concerns his sentence.

C. Acceptance of Responsibility

Constant next argues that the district court erred by denying him a downward adjustment for acceptance of responsibility or a downward variance based on his uninformed decision to proceed to trial. As a practical matter, this issue will be subsumed by the district court's evaluation of the Strickland issue on remand. If the district court finds that counsel's assistance fell below an objective standard of reasonableness, and that Constant would have pled guilty but for bad advice, it will then be free to determine whether, but for that advice, Constant would have received the adjustment. Conversely, if it determines that Constant was not harmed by any ineffective assistance of counsel, then its ruling on any adjustments will likely stand. In any case, our decision to vacate the sentence to allow consideration of the Strickland issue will allow the district court to evaluate the appropriateness—or not—of this adjustment upon a more complete record.

D. Remaining Claims

Constant makes two other challenges to his conviction and sentence. First, he argues that the district court committed reversible error when Detective St. Laurent sat at government counsel table throughout jury empanelment and trial. At the pretrial suppression hearing, the district court asked trial defense counsel whether he objected to St. Laurent's presence at government counsel table. Trial defense counsel replied that he did not object, but requested that St. Laurent testify as the government's first witness, which he did. At trial, Constant then raised no objections to St. Laurent's presence at government counsel table and did not request that St. Laurent testify first.

The government contends that this acquiescence arguably constituted waiver, but we need not pursue that suggestion because Constant's claim cannot survive the plain error review we apply to unpreserved claims of error. To establish plain error, Constant must show that “(1) an error occurred, (2) the error was obvious, (3) the error affected substantial rights, and (4) the error ‘seriously impaired the fairness, integrity, or public reputation of judicial proceedings.’ ” United States v. LaPlante, 714 F.3d 641, 643 (1st Cir.2013) (quoting United States v. Vargas–De Jesus, 618 F.3d 59, 67 (1st Cir.2010)).

We find no obvious error here. We have previously held that whether to allow a case agent to sit at counsel table is a matter within the discretion of the trial judge. See United States v. Charles, 456 F.3d 249, 259–60 (1st Cir.2006); United States v. Anagnos, 853 F.2d 1, 4 (1st Cir.1988). There is nothing in the facts of this case that points to any obvious abuse of that discretion. And while Constant now argues on appeal that St. Laurent's seat at counsel table unfairly bolstered St. Laurent's credibility to the jury, the record suggests that his presence could have been seen by trial defense counsel as a two-edged sword cutting more sharply against the prosecution by suggesting that the same guy who skewed the witness identification interview was co-piloting the prosecution.

Finally, Constant argues that the district court's relevant conduct determination that he was the shooter, which enhanced his base offense level under the Guidelines by four levels, should have been proven beyond a reasonable doubt to a jury. This argument, as Constant concedes, contravenes this circuit's precedent and must therefore be rejected. See United States v. Leahy, 668 F.3d 18, 22 (1st Cir.2012) (affirming that “sentencing factors affecting a judge's discretion within a statutorily prescribed range may be proved to a judge at sentencing by a preponderance of the evidence” (emphasis omitted)).

III. Conclusion

Based on the foregoing, the conviction is affirmed but the sentence is vacated and the case is remanded to the district court so that it can conduct an evidentiary hearing and make a determination on whether trial defense counsel provided ineffective assistance in advising Constant on his ACCA exposure and whether, in light of that determination, the sentence previously chosen should be re-affirmed or changed.