ADAMES GARCIA v. DIVRIS (2026)

Jorge Adames-Garcia (“Petitioner”), a state prisoner, challenges the dismissal of his federal habeas petition. He argues that the jury that convicted him was tainted by a Facebook post and a courthouse demonstration, in violation of his constitutional right to an impartial jury. Because the state court, after a hearing at which both parties had an opportunity to question the jury, found no credible showing that Petitioner was prejudiced by any such exposure, we affirm the denial of habeas relief.

I. Background

We briefly rehearse the relevant facts and travel of the case. In the process, we draw upon the facts recited by the Massachusetts Appeals Court (MAC), supplemented by other facts in the record consistent with that recitation. See Porter v. Coyne-Fague, 35 F.4th 68, 71 (1st Cir. 2022).

A. Crime of Conviction and Jury Trial

Petitioner was indicted on ten counts in connection with a rape that occurred on July 5, 2013: four counts of aggravated rape,¹ two counts of assault and battery, and one count each of kidnapping, unarmed robbery, assault with intent to commit rape, and malicious destruction of property. At trial, the government's case rested on testimony from the victim, evidence from a rape kit, DNA tests, photographs, and police and witness testimony.

The victim, K.T., testified that after a late night out on July 4, 2013, she decided to walk home alone from her friend's house, but next remembers waking up in a ditch on the morning of July 5, 2013. Petitioner ordered her into his car and drove her to a nearby beach parking lot, where he raped her. When he was unable to ejaculate, he drove K.T. to a second location and raped her again. After allowing her out of the car, he strangled her until she lost consciousness. When K.T. regained consciousness, she made her way to a nearby residence for help. About ten minutes later, the police and an ambulance arrived. K.T. reported that she had been raped and was taken to a hospital where she was examined and photographed.

Petitioner testified to a different account. He said that, on the same morning, he had been driving when he spotted K.T. walking and offered her a ride. She accepted, they talked in Spanish, and, eventually, had consensual sex at the beach parking lot. At some point, his condom broke, and he asked K.T. whether she had any diseases. He said she was upset by the question and left.

The case went to the jury on May 10, 2017. Two days into deliberation, on May 12, the jury had reached unanimous verdicts on seven of the ten charges but remained at an impasse on the rest. The court instructed the jurors to continue to deliberate until they reached unanimity on all counts. But after the weekend recess, on May 15, the court received the seven unanimous verdicts -- each one a not guilty verdict.

The jurors were then excused from jury service on May 16 because the court would not be in session. That day, former police officer Steve Tornovish (“Tornovish”) posted on Facebook:

Rape is a horrible crime. This current case is still going on. The trial is held in open court. Citizens may attend. I urge all interested parties to come to the Nantucket courtroom on Wednesday (tomorrow) at 9:15 a.m. Be orderly, be respectful and don't bring items that you wouldn't bring on an airplane. A show of support for the victim would sure be a good thing. Hope that you all can spare 1/2 hour to make our community stronger. Thanks.

There were several comments on the post. Some commenters pledged to wear teal sexual assault awareness ribbons to court the next morning in solidarity with the victim. One comment described those prosecuted for rape as “scum” from whom the community needed protection. And, in a follow-up comment, Tornovish described K.T. as a “bright and thoughtful person.”

The court reconvened on May 17, 2017. Defense counsel brought the Facebook post to the court's attention and requested an individual inquiry of each juror about the post. The trial judge denied the request but agreed to direct the presiding judge ² to ask the jurors, when they came in, whether anyone had difficulty following the jury instructions ³ and whether anyone had concerns bearing on their continued service; if a juror raised a concern, the court could address it at sidebar. The presiding judge also instructed members of the audience that they were not permitted to wear ribbons or symbols in the courtroom. The jury was brought in, and the presiding judge posed the agreed-upon questions. No juror reported any concern. The court then dispatched the jury to deliberate, and it ultimately returned guilty verdicts on the remaining three counts that same day.

Petitioner was sentenced to fifteen to eighteen years of imprisonment on the first count of aggravated rape, and to ten years of probation on the second, to run consecutively. The court dismissed the kidnapping count as duplicative, having served as the underlying felony for the aggravated rape convictions.

B. State Post-Conviction Proceedings

On March 21, 2018, defense counsel sent a letter to the jurors asking whether extraneous influences had affected the verdict. Three jurors responded. Jurors 13 and 8 said they noticed more people in the courtroom on the last day of deliberations but were unaware of the Facebook post. Juror 7 replied that she had seen the Facebook post the day before the final day of deliberations and, on that day, received “constant[ ]” notifications of comments related to the post; she saw a large number of people at the courthouse, some wearing teal ribbons;⁴ jurors “brought up” the Facebook post on the last day of deliberations; the jurors were all aware of the community response to their not-guilty verdicts; and that “[e]ven though the evidence was not there,” the Facebook post and the community's reaction caused the jurors to give “more weight to [the victim's] testimony” on the final day of deliberations and to return the final three guilty verdicts.

Based on Juror 7's response, defense counsel moved for a new trial. The trial court thereafter convened a two-day evidentiary hearing where all living jurors testified.⁵ The jurors were asked whether they were aware of any Facebook post before the final day of deliberations, whether they noticed increased presence outside or in the courtroom, whether they noticed any teal ribbons or shirts, and whether any of these matters had been discussed in the jury room.

Based on the jurors’ testimonies, the court made the following findings:

Juror 7 and Juror 9 saw Steve Tornovich's Facebook post[ ] and some responsive comments to [that post] before coming to court for the final day of deliberations. These two jurors were aware that the post[ ] and comments expressed support for the victim and disappointment with the seven not guilty verdicts.

There were approximately [ten] people gathered outside of the courthouse that morning, some of whom were wearing the color teal. Nine of the [twelve] jurors (including Juror 1, now deceased), either saw those people or heard about them from other jurors before deliberations resumed that morning. The jurors who were aware of the protestors understood that they were there to support the victim and to protest the not guilty verdicts. No juror reported being approached or confronted by any protestor, or observing any threatening or other concerning conduct by any protestor.

Before the jurors assembled in the courtroom that day and were greeted by Judge Barrett and sent out to resume deliberations, there was a brief discussion among some of them about the community's support for the victim, its reaction to the not guilty verdicts, and the wearing of the color teal. No specific details about the content of the Facebook [post was] mentioned. The discussion was quickly cut off by two jurors who recognized that it was inappropriate and admonished the others. No juror reported any exposure to extraneous information to Judge Barrett when he inquired of them about following the cautionary instructions. Neither the Facebook post[ ] and comments nor the presence of protestors were discussed during the jury's deliberations.

The court specifically discredited Juror 7's testimony that “the community's emotional reaction was intimidating, and that in response, a few jurors indicated that perhaps they rushed their earlier decision and the not guilty verdicts had been wrongly decided.” No other juror testified that this conversation took place, and the court found that the community reaction was not prolonged, intense, or highly-charged.

The court then applied a two-step framework -- developed by the Supreme Judicial Court of Massachusetts (SJC) in Commonwealth v. Fidler, 377 Mass. 192, 385 N.E.2d 513, 519 (1979) -- to assess whether Petitioner's conviction had been prejudiced by the extraneous material. Under the framework, a petitioner first bears the burden to prove by a preponderance of the evidence that extraneous material was introduced to the jury, after which the burden shifts to the Commonwealth to prove beyond a reasonable doubt that the material did not prejudice the petitioner. Id. In determining whether the Commonwealth met its burden, the judge is not permitted to inquire into the “subjective mental processes of jurors.” Id. at 517. Rather, “the judge must focus on the probable effect of the extraneous facts on a hypothetical average jury.” Id. at 519 (citation omitted). Nonetheless, where the evidence of subjective mental processes inadvertently comes to the court's attention, it cannot be ignored. Commonwealth v. Kincaid, 444 Mass. 381, 828 N.E.2d 45, 53 (2005).

Here, the court concluded that the first step was satisfied (Petitioner met his burden) because at least two jurors reported seeing the Facebook post, and others learned of it through discussions with fellow jurors. It then proceeded to the second step. At that point, the court explained that the content of the Facebook post (which contained no invective or intimidating language), the discussion surrounding them (brief, drawing immediate reprimands, not occurring during deliberations), the presence of protestors (relatively few in number, orderly and peaceful, with no attempts to contact jurors), and the strength of the Commonwealth's evidence (specific testimony corroborated by physical examination and witness accounts) sufficed to show that the extraneous matter would not have prejudiced the average hypothetical jury.

Because Juror 7 revealed information about her and the other jurors’ decision-making process, the trial court properly considered it too. Ultimately, the court discredited Juror 7's testimony about other jurors’ alleged prejudice (because the Facebook post and community protests were not discussed during deliberations, Juror 7 “would have no way of knowing” whether it prejudiced other jurors; and others unequivocally denied being influenced) and her own (the judge found that it appeared the juror had either a change of heart or perception after the trial).

Accordingly, the court denied Petitioner's motion for a new trial. He appealed, but the MAC affirmed and the SJC declined further review.

C. Petitioner's Habeas Petition

Petitioner subsequently filed a habeas corpus petition under 28 U.S.C. § 2254 alleging that the jury had been exposed to harmful, extraneous material in violation of his Sixth Amendment right to an impartial jury. The petition was referred to a magistrate judge, who, after a non-evidentiary hearing, recommended the petition be granted. The magistrate judge concluded that the MAC's decision was contrary to, and rested on an unreasonable application of, Supreme Court precedent, and that the MAC's decision was based on an unreasonable determination of the facts. The district court rejected the magistrate's recommendations, reasoning that the state court carefully investigated the extraneous-influence claim by following a two-step framework that matched one set forth by the Supreme Court and that special deference was owed to its factual findings. Petitioner timely appealed.

II. Discussion

Petitioner's habeas petition regarding his state court conviction is governed by the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”). See 28 U.S.C. § 2254. AEDPA “demands that a federal habeas court measure a state court's decision on the merits against a series of peculiarly deferential standards.” Quintanilla v. Marchilli, 86 F.4th 1, 15 (1st Cir. 2023) (citation modified). For claims adjudicated on the merits in state court, as is the case here, AEDPA provides that “a writ of habeas corpus ․ shall not be granted ․ unless” the challenged state court decision was:

(1) ․ contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States; or

(2) ․ based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.

28 U.S.C. § 2254(d)(1)-(2). “The first of these two bases for granting habeas relief -- subsection (d)(1) -- itself ‘splits into two distinct avenues for relief: the “contrary to” clause and the “unreasonable application” clause.’ ” Quintanilla, 86 F.4th at 16 (quoting Porter, 35 F.4th at 74). Petitioner maintains that the MAC's affirmance of his conviction warrants habeas relief under any of these three avenues. We review the district court's denial of habeas relief de novo, Scott v. Gelb, 810 F.3d 94, 98 (1st Cir. 2016), and address each ground for relief seriatim.

A. Contrary to clearly established Federal law

A state court decision is “contrary to” clearly established federal law if it: (1) “arrives at a conclusion opposite to that reached by [the Supreme Court] on a question of law” or (2) “decides a case differently than [the Supreme Court] on a set of materially indistinguishable facts.” Williams v. Taylor, 529 U.S. 362, 412-13, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000). The phrase “clearly established Federal law” refers to the holdings, not the dicta, of the Court's decisions. Id. at 412, 120 S.Ct. 1495.

Naturally, then, we turn to the relevant “clearly established Federal law” at play in this appeal. The Sixth Amendment guarantees a criminal defendant the right to trial by an impartial jury. U.S. Const. amend. VI; see Parker v. Gladden, 385 U.S. 363, 364, 87 S.Ct. 468, 17 L.Ed.2d 420 (1966). This right is made applicable to the states (with some exceptions) through the Fourteenth Amendment's due process clause. Bebo v. Medeiros, 906 F.3d 129, 135 & n.1 (1st Cir. 2018). To honor this guarantee, a jury's verdict must rest on trial evidence alone, remaining “free from [extraneous material ⁶ ] tending to disturb the exercise of deliberate and unbiassed [sic] judgment.” Id. (quoting Mattox v. United States, 146 U.S. 140, 149, 13 S.Ct. 50, 36 L.Ed. 917 (1892)) (second alteration in original).

As such, when a petitioner shows that extraneous information may have tainted the jury, due process requires that the court hold a hearing (often referred to as a “Remmer hearing”) with all interested parties to “determine the circumstances, the impact thereof upon the juror, and whether or not it was prejudicial.” Id. at 136 (quoting Remmer v. United States, 347 U.S. 227, 230, 74 S.Ct. 450, 98 L.Ed. 654 (1954)); Smith v. Phillips, 455 U.S. 209, 216, 102 S.Ct. 940, 71 L.Ed.2d 78 (1982). “[D]ue process does not require a new trial every time a juror has been placed in a potentially compromising situation.” Smith, 455 U.S. at 217, 102 S.Ct. 940; see also United States v. Tejeda, 481 F.3d 44, 51 (1st Cir. 2007) (explaining that a petitioner claiming that extraneous influence prejudiced his conviction is “not entitled to an automatic reversal, but rather to a hearing before the trial court”). Rather, a new trial is warranted only if the hearing showed that extraneous material prejudiced the verdict. See id.; Remmer, 347 U.S. at 230, 74 S.Ct. 450.

The Supreme Court has addressed extraneous influence claims in a handful of cases. See, e.g., Bebo, 906 F.3d at 135-36 (collecting cases). Most relevant here, in Parker v. Gladden, a post-conviction hearing revealed that a bailiff had told a juror (referring to the defendant) “Oh, that wicked fellow, he is guilty” and that “[i]f there [was] anything wrong [in finding the defendant guilty] the Supreme Court will correct it.” 385 U.S. at 363-64, 87 S.Ct. 468. The state trial court concluded that this had prejudiced the defendant's conviction, and the state supreme court reversed. On appeal, the Supreme Court reversed the state supreme court, explaining that the state's no-prejudice argument “overlook[ed] ․ the official character of the bailiff -- as an officer of the Court as well as the State” which “carries great weight with a jury which he had been shepherding for eight days and nights”; the jury's extended deliberations, which suggested disagreement on the defendant's guilt; and the post-conviction testimony of a juror that she was prejudiced by these statements. Id. at 365, 87 S.Ct. 468. All of these factors, taken together, “support[ed] the trial court's finding ‘that the unauthorized communication was prejudicial ․’ ” See id.

Petitioner also relies on Turner v. Louisiana, where two of the government's principal witnesses -- deputy sheriffs -- were similarly assigned by the court to manage the sequestered jury during trial. 379 U.S. 466, 468, 85 S.Ct. 546, 13 L.Ed.2d 424 (1965). These officers “drove the jurors to a restaurant for each meal, and to their lodgings each night,” and “ate with them, conversed with them, and did errands for them” such that the jurors were “continuously in the company of the” officers over the course of the three-day trial. Id. at 467–68, 85 S.Ct. 546. Though the officers testified that they never spoke about the case with the jurors, the Court nonetheless held that “extreme prejudice [was] inherent in this continual association” between the testifying officers and the jurors given the deputies’ role as “official guardians” of the jury. Id. at 466, 473–74, 85 S.Ct. 546.

With the relevant law explained, we address Petitioner's two distinct “contrary to” arguments. First, Petitioner contends that the state court's no-prejudice conclusion was “contrary to” clearly established law under Parker because, in his view, Parker holds that “where a juror testifies to having been influenced by outside information,” prejudice cannot be dismissed as harmless.⁷ He makes largely the same argument with respect to Turner. We disagree.

To begin, Petitioner's proffered reading of Parker and Turner plucks a single factor that the Supreme Court considered from its context and treats it as a per se rule. But Parker’s holding rested on three factors operating together: the bailiff's official role (and relationship of trust with the jurors), supplemented by the jury's extended deliberations and a juror's testimony that the statements affected her. 385 U.S. at 365, 87 S.Ct. 468. Similarly, in Turner, the jury's “continuous and intimate association” with the deputies who served as the jury's “official guardians” led to the conclusion that there was “prejudice inherent.” 379 U.S. at 473-74, 85 S.Ct. 546. No analogous official role or association is implicated here. And secondarily, Petitioner's argument ignores that the state court in Parker credited the juror's testimony only after a hearing tested and confirmed the juror's credibility. Id. In other words, Parker did not foreclose a trial court's assessment of a juror's credibility or its determination that extraneous information, even if present, did not in fact prejudice the verdict. Rather, it presupposed them.

Clearly established law requires a new trial only if the trial court, following a Remmer hearing, concludes that the verdict was actually prejudiced by the extraneous material. See Bebo, 906 F.3d at 136. The state court in this case did not act contrary to Parker’s holding.⁸

Petitioner next contends that even if Parker established no per se rule, this case is materially indistinguishable from Parker and thus demands the same result. Again, we disagree. It is true that the juries respectively deliberated for nearly identical periods: 26 hours in Parker, 25 hours here. 385 U.S. at 365, 87 S.Ct. 468. But that is the extent of their similarity. In Parker, the extraneous influence derived from an official court officer who maintained close and continuous contact with the sequestered jury over an eight-day period, and whose official position loaned him considerable weight with the jurors. Id. Here, the extraneous influence consisted of a Facebook post made by a former police chief who had no role in the trial, and no special position of trust or power with the jury -- he did not even know the jury. Moreover, in Parker, the trial court had credited a single juror's testimony that she was prejudiced, and the trial court's resulting finding of prejudice was not upset by the appellate courts. See id. Here, by contrast, the trial court found the single juror's testimony of prejudice to be not credible, and the MAC did not disturb that finding. In short, Petitioner's case is distinguishable from Parker and the state court's conclusion therefore was not contrary to clearly established federal law.

B. Unreasonable application of clearly established Federal law

Under AEDPA, an application of Supreme Court precedent is unreasonable “if, and only if, it is so obvious that a clearly established rule applies to a given set of facts that there could be no ‘fairminded disagreement’ on the question.” Watson v. Edmark, 118 F.4th 456, 461 (1st Cir. 2024) (quoting White v. Woodall, 572 U.S. 415, 427, 134 S.Ct. 1697, 188 L.Ed.2d 698 (2014)). In applying this standard, we must “determine what arguments or theories supported,” or “could have supported, the state court's decision” and then “ask whether it is possible fairminded jurists could disagree that those arguments or theories are inconsistent” with Supreme Court precedent. Harrington v. Richter, 562 U.S. 86, 102, 131 S.Ct. 770, 178 L.Ed.2d 624 (2011). We must keep in mind that “an unreasonable application of federal law is different from an incorrect application of federal law,” and that “[t]he more general the rule, the more leeway courts have in reaching outcomes in case-by-case determinations.” Id. (quoting Yarborough v. Alvarado, 541 U.S. 652, 664, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004)).

First, Petitioner maintains that the state court unreasonably applied clearly established federal law because, in his view, the governing rule is highly specific and leaves little room for disagreement: a criminal defendant is entitled to twelve impartial jurors, and when a juror affirmatively testifies that extraneous information affected deliberations, prejudice cannot be dismissed as harmless. But we have already rejected Petitioner's framing of the governing rule. That leaves his narrower argument that even under the proper legal framework, the state court unreasonably assessed prejudice. Specifically, Petitioner argues that the state court improperly relied on the following facts in its analysis: that the extraneous material was not discussed during deliberations, the exposure was brief, and the finding that a hypothetical juror would not have been influenced by the Facebook post and public support for the victim.⁹ We disagree.

To be sure, Petitioner is correct that a state court does not satisfy its constitutional obligation merely by convening a hearing to consider alleged juror taint. A prejudice determination may still be unreasonable if the court applies the wrong legal standard, ignores materially relevant evidence, or reaches a conclusion no fairminded jurist could accept. We also agree with him that none of these factors are independently required to establish prejudice. Remmer directs courts to “determine the circumstances, the impact thereof upon the juror, and whether or not it was prejudicial.” 347 U.S. at 230, 74 S.Ct. 450. But we think the state court reasonably conducted its Remmer duty, considering reasonable factors in the process.

Start with the absence of discussion during deliberations. It is true that extraneous information need not be discussed during deliberations to be prejudicial. Parker demonstrates as much. But it does not follow that the absence of discussion is irrelevant. If extraneous material had genuinely infected the jury's reasoning, a court could rationally expect some trace of that influence to appear in the deliberations themselves.¹⁰ See, e.g., Cox v. Ayers, 414 F. App'x 80, 85 (9th Cir. 2011) (“None of the jurors stated with certainty that the letter was discussed during deliberations. Under these circumstances, Petitioner cannot show prejudice.”). The same logic applies when considering the duration of jurors’ exposure to extraneous material. A brief exposure can certainly be prejudicial. But all else equal, a shorter exposure will be less likely to prejudice a juror than prolonged or repeated contact. Cf. Turner v. Louisiana, 379 U.S. 466, 473, 85 S.Ct. 546, 13 L.Ed.2d 424 (1965) (“We deal here not with a brief encounter, but with a continuous and intimate association throughout a three-day trial ․”).

What is more, the state court did not consider those two factors alone. It also considered the content of the Facebook post, the substance of the discussion surrounding it, the presence of the protestors, the jurors’ testimony at the Remmer hearing, and the strength of the Commonwealth's case against Petitioner. That multifactor approach is precisely what Remmer contemplates and is consonant with that of other courts. See, e.g., Sassounian v. Roe, 230 F.3d 1097, 1109 (9th Cir. 2000) (listing factors for courts to consider, including “the length of time [extraneous information] was available to the jury” and “whether the material was introduced before a verdict was reached, and if so at what point in the deliberations” (citation modified)); United States v. Lloyd, 269 F.3d 228, 240 (3d Cir. 2001) (considering the “extent of the jury's exposure to the extraneous information” and “the time at which the jury receives the extraneous information”). This “adds further force to the conclusion that” the state court's approach here “is not one with which ‘fairminded jurists’ could not agree.” Linton v. Saba, 812 F.3d 112, 126 (1st Cir. 2016) (quoting Richter, 562 U.S. at 88, 131 S.Ct. 770).

Lastly, Petitioner argues that “[i]n a case with direct evidence of actual prejudice, the court should not have prioritized the average hypothetical juror.” Two responses. First, the court found Juror 7's testimony not credible, so there was no evidence of actual prejudice at the time of deliberations to begin with. Second, and relatedly, the state court's inquiry into an objective “hypothetical juror” was not unreasonable. Because courts generally may not probe the actual influence of extraneous information on a jury's deliberative process, prejudice is ordinarily assessed objectively: whether the extraneous information would likely influence a reasonable juror. Where a juror nevertheless offers testimony touching on deliberations, the court may assess that testimony, weigh its credibility, and then consider it alongside the objective circumstances. That is exactly what happened here.

Petitioner has not shown that the state court's actual prejudice analysis was beyond the bounds of fairminded disagreement. Therefore, we reject his second claim of error.

C. Unreasonable determination of the facts

Under 28 U.S.C. § 2254(d)(2), a federal court may issue a writ if the state court decision “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.”¹¹ A § 2254(d)(2) claim requires a “ ‘demanding showing’ that ‘cannot be made when “reasonable minds reviewing the record might disagree” about the finding in question.’ ” Garrey v. Kelly, 162 F.4th 27, 39 (1st Cir. 2025) (quoting Porter, 35 F.4th at 75). “[A] state-court factual determination is not unreasonable merely because the federal habeas court would have reached a different conclusion in the first instance.” Wood v. Allen, 558 U.S. 290, 301, 130 S.Ct. 841, 175 L.Ed.2d 738 (2010).

Credibility determinations warrant particular deference. While it is true that we usually think jurors are “well qualified to say whether [they] ha[ve] an unbiased mind in a certain matter,” Dennis v. United States, 339 U.S. 162, 171, 70 S.Ct. 519, 94 L.Ed. 734 (1950), the weight to be given to a juror's account, including whether a juror's self-report of bias reflects genuine partiality or some other motivation, is a factual determination that falls squarely within the trial judge's traditional role as factfinder, see Thompson v. Keohane, 516 U.S. 99, 109-10, 116 S.Ct. 457, 133 L.Ed.2d 383 (1995) (pre-AEDPA). After all, “[t]he trial judge is in the best position to assess” credibility “by observing [a juror's] demeanor, reaction to questioning, and overall behavior on the stand.” United States v. Lowe, 145 F.3d 45, 49 (1st Cir. 1998) (direct appeal).¹²

On habeas review, that determination is not ours to second-guess freely. The question before us is not whether we would have credited Juror 7's testimony ourselves; it is whether the trial court's decision not to do so lacked any fair support in the record. See, e.g., Rice v. Collins, 546 U.S. 333, 341-42, 126 S.Ct. 969, 163 L.Ed.2d 824 (2006) (declaring that habeas courts may not “supersede the trial court's credibility determination” simply because “[r]easonable minds reviewing the record might disagree”).

Petitioner argues that the trial court's credibility findings regarding Juror 7 -- both as to her own prejudice and as to her testimony regarding other jurors’ prejudice -- rest on an unreasonable determination of the facts. 28 U.S.C. § 2254(d)(2). We disagree and explain why in turn.

1. Juror 7's credibility regarding her own prejudice

Petitioner maintains that the trial court found that Juror 7 was not credible with respect to her own prejudice based on three facts: “the swift admonitions of other jurors that the extraneous information should not be considered, the jury's collective discussion of only the law and the facts during deliberations, and Juror 7's credible averment that she recognized her duty to follow the law.” He contends that the record does not support these findings. Our own review, however, leads us to conclude that the record fairly supports them, or that they are at least findings about which “reasonable minds reviewing the record might disagree.” Brumfield v. Cain, 576 U.S. 305, 314, 135 S.Ct. 2269, 192 L.Ed.2d 356 (2015) (quoting Wood, 558 U.S. at 301, 130 S.Ct. 841).

First, the record fairly supports that the discussion about outside materials immediately ceased after other jurors reprimanded those who brought it up. Juror 7 testified that when the Facebook post was first mentioned, “a person who cut the conversation was, like, we can't discuss that,” and regarding teal ribbons, “another juror member cut that conversation and said that wasn't what we were supposed to be discussing.” Juror 7 herself stated “the conversation wasn't like a lengthy discussion” and that “immediately, the conversation cut and we went back to just the evidence and not discussing outside things.”

Second, the testimony from other jurors contradicts Juror 7's assertations that the Facebook post was discussed during the deliberations. Only Jurors 4, 7, and 13 reported the Facebook post as discussed in the jury room. Of those three, Jurors 4 and 13 testified that they recalled that discussions occurred outside deliberation time. Only Juror 7 implied the conversation occurred during deliberations and played a central role in shaping the jury's verdict on the final three counts. All eight other jurors testified unequivocally that the Facebook post was not discussed at all in the jury room. A single juror's isolated and unanimously contradicted account does not establish that the trial court's factfinding was unreasonable; rather it supports the trial court's finding that Juror 7 was not credible.

Lastly, Petitioner contends that Juror 7's recognition of her duty to follow the law gave credence, rather than damaged, her credibility. He further argues that these facts cannot reasonably support the trial court's inference that “it appears likely that Juror 7 had either a change of heart, or a change of perception, after the trial.”¹³ But that framing misunderstands the standard of review. The question is not whether these were the only permissible inferences, or even the ones we might have drawn in the first instance. It is whether the state court was unreasonable in drawing them. Given the record support for the underlying factual predicates -- immediate admonitions against discussing outside information, the lack of meaningful deliberative discussion of the extraneous material, and Juror 7's acknowledgment of her obligation to decide the case based only on the evidence -- the trial court's inferences were at least permissible ones. That Petitioner can offer a competing interpretation, and even that he can offer a reasonable one, does not establish unreasonableness. See Collins, 546 U.S. at 341-42, 126 S.Ct. 969 (“Reasonable minds reviewing the record might disagree about the prosecutor's credibility, but on habeas review that does not suffice to supersede the trial court's credibility determination.”); Weisheit v. Neal, 151 F.4th 855, 876 (7th Cir. 2025) (“Differing interpretations can be reasonable.”). Thus, we decline to disturb the state court's factual finding that Juror 7's testimony that she was prejudiced was not credible.

2. Juror 7's credibility regarding other jurors’ prejudice

The state court found Juror 7's statement that other jurors gave more weight to the victim's testimony after exposure to the extraneous influences not credible. Its conclusion rested upon five factual findings: (1) the jury's exposure to the extraneous matter was brief and drew immediate reprimands that caused the jurors to promptly recognize their impropriety and immediately stop considering the information; (2) the outside information was not discussed during deliberations; (3) the evidence against Petitioner was strong; (4) the community protest was not intimidating, prolonged, intense, or highly-charged; and (5) the jurors’ discussion did not reference any specific details contained in the Facebook post. Petitioner maintains the record does not support these findings. We again disagree.

With respect to the first two findings of fact, we have already explained that the record supports a finding that the discussion of extraneous material drew immediate reprimands and quickly ceased, and that the extraneous material was not discussed during deliberation. That leaves the findings that the jury's exposure to extraneous materials was brief. On this, eight jurors characterized any discussion as momentary: “mention,” “couple of comments,” “minimal,” discussions that “ended there,” and “wasn't a lot of discussion.” Only Jurors 7 and 9 had seen the Facebook post before the last day of deliberations. But Juror 9 testified that he saw the initial post and “didn't look at any comments.” He also could not recall many specifics about the content of the post and was unaware of any discussion about them in the jury room. From all of this, the trial court could reasonably find that the rest of the jurors’ exposure to extraneous materials was brief.

We move on to the state court's third finding that the evidence against Petitioner was strong (and thus, that it undercut Juror 7's testimony that “the evidence was not there”). Petitioner argues that the seven not-guilty verdicts and the length of the deliberation undercut this finding. But the record shows that the jury was presented with K.T.’s testimony, as well as corroborating DNA evidence, testimony about the extensive nature of K.T.’s injuries, and photographic documentation -- all of which support the finding that the evidence was strong with respect to the counts of conviction.

In any case, Petitioner's argument is unconvincing on its merits. That the jury found insufficient evidence for the acquitted counts does not call into question the strength of the evidence supporting the counts of conviction. The acquitted counts either have different elements (assault, robbery, and malicious destruction of property) or rest on the assaults having happened more than once (recall that the Commonwealth charged four aggravated-rape counts -- two for the first assault at the beach parking lot and two for the second assault at the second location). And the length of deliberations alone does not necessarily undermine the strength of the evidence. Cf. United States v. Powell, 469 U.S. 57, 65, 105 S.Ct. 471, 83 L.Ed.2d 461 (1984) (verdicts may result from “compromise” or “lenity”).

On to the fourth factor relied on by the state court -- that the community protest was not intimidating, prolonged, intense, or highly-charged. The record also supports this. The demonstration occurred on a single day, and the trial court immediately limited any exposure by prohibiting teal ribbons and symbols in the courtroom before it allowed the jury in. There was no evidence that the jurors were approached, addressed, or targeted by the protestors. And, again, as to the Facebook post, only two jurors saw the Facebook post at all, and only one -- Juror 7 -- read the comments. The record thus demonstrates that the rest of the jury had no prolonged exposure as to whatever sentiment the post and its comments may have reflected. Petitioner relies on the Facebook comments’ references to Petitioner as “scum,” K.T. as “a bright and thoughtful person,” and on the use of exclamation points and caps as demonstrating intensity.¹⁴ But these comments were not part of the original post -- and, again, were not seen or discussed by any juror except Juror 7. And, more importantly, reasonable minds reviewing the record could find that this social media engagement and a single day of orderly community attendance did not constitute a prolonged, intense, or highly-charged demonstration.

Lastly, the record supports the trial court's finding that the jurors’ discussion did not reference any specific details contained in the Facebook post or comments. The jurors that testified about discussions of the Facebook post merely acknowledged the general existence of it and that the community was upset about the prior verdicts.

At bottom, the record admits that jurors were not consistent in all aspects of their testimony. The state court necessarily found some of their testimony credible and some of it not credible. But the record fairly supports all of the underlying facts relied upon in the court's credibility determination. We therefore decline to upset its findings.

III. Conclusion

For the foregoing reasons, we affirm the district court's denial of habeas relief.

The Sixth Amendment required Adames to “be tried by 12, not 9 or even 10, impartial and unprejudiced jurors.” Parker v. Gladden, 385 U.S. 363, 366, 87 S.Ct. 468, 17 L.Ed.2d 420 (1966).¹⁵ And the jurors’ “exercise of calm and informed judgment ․ is essential to proper enforcement of law.” Sinclair v. United States, 279 U.S. 749, 765, 49 S.Ct. 471, 73 L.Ed. 938 (1929). Remember that while you read this:

[T]he presence of all those people from the community shook up the jurors․ We [the jury] all wanted to point our finger at [Adames], but the evidence was not there. We could not say he was guilty beyond a reasonable doubt, but we ended up making a decision in line with the community's response․ After the community's reaction, we could not continue saying “not guilty” despite being mostly decided on that verdict two days prior.

That's Juror 7 explaining under penalty of perjury how community outcry across Nantucket changed the course of jury deliberations in Adames's case. Pretty damning prejudice, right?

Well, not to the state trial court -- and, apparently, not to my colleagues, either. After a brief bout of questioning on a Zoom hearing a year-and-a-half after Adames's trial, the state trial court somehow concluded that Juror 7 wasn't even credible enough to testify that she herself was unduly pressured to reach a “guilty” verdict. Instead, the above-described concerns could just be chocked up to Juror 7 “likely” having “either a change of heart, or a change of perception” after the trial.

That credibility call doesn't make any sense. Generally, if a juror comes right out and says an extraneous influence prejudiced them, the court should probably believe them -- and, really, commend them for their self-awareness. Cf. Smith v. Phillips, 455 U.S. 209, 217 n.7, 102 S.Ct. 940, 71 L.Ed.2d 78 (1982) (“Surely one who is trying as an honest man to live up to the sanctity of his oath is well qualified to say whether he has an unbiased mind in a certain matter.”). And, anyway, the specific reasons that the state trial judge offered to discredit Juror 7 don't hold up. (I'll walk through them later.)

With Juror 7's written and oral testimony in hand, it'd be especially odd to find that the Commonwealth proved beyond a reasonable doubt that she wasn't compromised -- and thus that the verdict was consistent with the Sixth Amendment's requirement of twelve unprejudiced jurors. Yet that's exactly what the state trial court concluded, and it's essentially what my colleagues now affirm -- albeit under a more deferential standard of review within our habeas framework.

But “even in the context of federal habeas, deference does not imply abandonment or abdication of judicial review.” Miller-El v. Cockrell, 537 U.S. 322, 340, 123 S.Ct. 1029, 154 L.Ed.2d 931 (2003). And meaningful judicial review (i.e., the work done by U.S. Magistrate Judge Judith Dein in this case) reveals that the state trial court's decision rested on “an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(2). So denying Adames post-conviction relief thus was -- and remains -- error. I respectfully (but emphatically!) dissent.

I

Here are the facts most relevant to the point I'm making.

A

Adames's criminal trial on ten weighty counts went on for two days. On a Wednesday -- May 10, 2017, to be specific -- jury deliberations began. That Friday (two days into deliberations), the jury let the judge know that it had unanimously agreed on seven out of the ten charges, without specifying the outcome. The judge told the jury to keep deliberating until it reached unanimity on every count. But the twelve failed to reach a verdict on the last three counts and were thus excused for the weekend.

Then came Monday morning. After some judge-shuffling (described above by the majority), the presiding judge let the jury issue the seven unanimous verdicts decided thus far -- all “not guilty.” Three charges remained: one count of kidnapping and two counts of aggravated rape. The jury went back to deliberate further, but it still was unable to reach a decision by end-of-day.

Court wasn't in session on Tuesday. But that didn't stop news about the trial from buzzing around the small island of Nantucket. Former Nantucket police detective Steve Tornovish (on Facebook under the alias “Steve Tuna”) posted this on the local community Facebook page:

Rape is a horrible crime. This current case is still going on. The trial is held in open court. Citizens may attend. I urge all interested parties to come to the Nantucket courtroom on Wednesday (tomorrow) at 9:15 a.m. Be orderly, be respectful and don't bring items that you wouldn't bring on an airplane. A show of support for the victim would sure be a good thing. Hope that you all can spare [a] 1/2 hour to make our community stronger. Thanks.

Two of the twelve jurors were Facebook friends of “Steve Tuna.”

And Tornovish's post clearly got some traction, based on the commotion at the courthouse on Wednesday. One Facebook comment on the post (from a “Shannon Bennett”) noted, just after 10:00am, that she was “in the courtroom” and “it's PACKED,” along with “the hallway the stairs etc etc.” Adames's counsel told the court about the Facebook post (and comments) and then sought a curative instruction and inquiry of each juror about their knowledge of the post and the crowd's purpose. But the state trial judge denied the request.¹⁶ Still, instructions were given to the new crowd regarding appropriate courtroom decorum, such as not wearing teal ribbons (teal being the color worn in support of survivors of sexual assault).

Jury deliberations -- which, to reiterate, had started last Wednesday and, by now, had been going on for nearly a week -- resumed a bit after 10:30am. Just two hours later, the jury delivered a unanimous “guilty” verdict on the last three charges.

Adames ultimately got fifteen to eighteen years of imprisonment for the first aggravated rape count, and ten years of probation to run consecutively for the second (with the kidnapping charge dismissed as duplicative).

B

About ten months after the verdict, Adames's attorney began contacting jurors with four main inquiries:

(1) if they knew about the “Steve Tuna” Facebook campaign;

(2) if they knew the crowd in the courtroom was there in response to that campaign;

(3) if they were offered teal ribbons that day;

(4) if they noticed people wearing teal ribbons in the courtroom that day (or understood teal's significance).

Three jurors responded. Two (Jurors 13 and 8) said they didn't know about the Facebook campaign, didn't feel any pressure or influence, but did notice more people in the courtroom (in one's words, a “significant increase”) on the last day of deliberations.

But Juror 7 saw things quite differently, as she explained in an affidavit. For starters, she saw the “Steve Tuna” post the day before deliberations ended, and she said it was “blowing up” on Facebook, with people “constantly responding.” And when she got to court the next day, she noticed “the large number of people outside of the courtroom and filling the courthouse area.” But particularly striking were the observations she made about what happened next:

• “The presence of all those people from the community shook up the jurors.”

• The Facebook posts were “brought up by the jurors,” and “everyone in the jury room was aware of the community's response to the verdicts on the first seven counts.”

• “To see a community so outraged by a decision we were struggling to make was very trying to me and to most of the jurors.”

• “We all wanted to point our finger at Mr. Adames-Garcia, but the evidence was not there. We could not say he was guilty beyond a reasonable doubt, but we ended up making a decision in line with the community's response.”

• “Even though the evidence was not there, because of the Facebook posts and the community reaction to those posts, we gave more weight to [the victim's] testimony on the day of the last three verdicts.”

• “Her testimony, the response inside the courtroom, and the community's outcries had a huge impact on the decision we all made.”

• “After the community's reaction, we could not continue saying ‘not guilty’ despite being mostly decided on that verdict two days prior.”

So Adames's counsel sought a new trial. And because Adames claimed extraneous influence on the jury, a two-step inquiry took center stage.

Adames first had to show that the jury was exposed to the “extraneous matter.” Commonwealth v. Kincaid, 444 Mass. 381, 828 N.E.2d 45, 49 (2005). If he made that showing, “and the judge [found] that extraneous matter came to the attention of the jury, the burden” would then shift “to the Commonwealth to show beyond a reasonable doubt that [Adames] was not prejudiced by the extraneous matter.” Id.

The state trial judge held a Zoom hearing to question the eleven living jurors (the foreman had passed since the trial) about the Facebook post and the crowd at the courthouse.¹⁷ She asked questions “about whether information may have been brought into the jury room and, if so, what that information was.” But she specifically said she wouldn't ask (among other things) “how the jurors used any such outside information, or the effect such information had on the thinking of any juror.”

Juror 7 went first. She testified that she received a phone notification about the “Steve Tuna” post, as well as comments that the seven “not guilty” verdicts were “unfair,” but couldn't recall all the details “because it was so long ago.” She also said that she did not inform the judge about her exposure to the Facebook post at the time, yet that it was discussed in the jury room. (But she couldn't recall who among the jurors initiated the discussion.)

She also explained that, in the jury room, someone observed that the crowd in the courtroom were “the people from Facebook in the community that are mad at us for making the verdict that he was not guilty.” To that, someone apparently asked, “[H]ow are we supposed to continue with that verdict[?]” Yet someone else said that such outside information should not be considered (which seems to have happened at least one other time), but the conversation still wandered back to these extraneous topics.

All told, Juror 7 recalled the discussion, particularly about the crowd, being “an intimidating one.” She thought “the emotional response of the community ․ kind of ignited a conversation ․ inside the deliberation room as to how we were going to proceed and how it affected us.” The tension “split” the jury, in her view.

(To keep things brief, I will move past the ten other jurors’ testimony. In short, they largely said they didn't know about the Facebook post and mainly claimed they didn't make much of the crowd. But plenty of awareness and discussion of the crowd existed, as I'll explain later.)

Recall that “extraneous matter” framework we discussed a couple of pages ago; it now comes back into the fold. After hearing all the testimony, the state trial judge concluded first that Adames had proven that the jury was exposed to extraneous influence. Two saw the Facebook post before they got to the courthouse, and several more learned about it (and the broader community response vis-à-vis the crowd) through discussion among the jurors.

So the question became whether the Commonwealth proved beyond a reasonable doubt that the extraneous matters didn't prejudice Adames. See Kincaid, 828 N.E.2d at 49. The state trial court held that the Commonwealth did prove just that. Partly, that was because Juror 7's statements about how all this affected the other jurors were “merely speculative and deserve no weight.” And Juror 7's statements that all the outcry impacted her own ability to be impartial and unprejudiced couldn't be believed either, apparently. Here's what the state trial judge had to say, in full, about that:

Finally, given the swift admonitions of other jurors that the extraneous information should not be considered, the jury's collective discussion of only the law and the facts during deliberations, and Juror 7's credible averment that she recognized her duty to follow the law, I do not find Juror 7's averment that she herself was influenced by the community's reaction in rendering her verdict to be credible. Rather, it appears likely that Juror 7 had either a change of heart, or a change of perception, after the trial.

So the state trial judge denied Adames's post-conviction claim. As did the Massachusetts Court of Appeals (or “MAC” for short). See Commonwealth v. Adames-Garcia, 158 N.E.3d 887 (Table) (Mass. App. Ct. 2020).

But once Adames got to federal court, U.S. Magistrate Judge Judith Dein issued a thorough report that recommended granting Adames's habeas petition. See Adames-Garcia v. Divris (“Adames R&R”), No. 21-11016-NMG, 2024 WL 4958309, at *1 (D. Mass. July 18, 2024), report and recommendation rejected, 759 F. Supp. 3d 189 (D. Mass. 2024). Yet the federal district court declined to adopt that report, see generally Adames-Garcia, 759 F. Supp. 3d 189, and my colleagues share the district court's view.

I don't, though. As I'll explain shortly, Judge Dein's careful analysis holds the keys to this case.

II

One way for a petitioner to succeed under AEDPA is for him to show that the state court's decision “was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(2).

No doubt that this is a “demanding showing,” and the standard isn't satisfied simply where a petitioner can cause reasonable minds to disagree. See Quintanilla v. Marchilli, 86 F.4th 1, 17 (1st Cir. 2023). Instead, we typically assume that “clear and convincing evidence” is necessary.¹⁸ See Porter v. Coyne-Fague, 35 F.4th 68, 79 (1st Cir. 2022). And (as the majority aptly points out) credibility determinations often get particular deference, grounded in the trial court's unique institutional capability for assessing witnesses.

But, importantly, “even in the context of federal habeas, deference does not imply abandonment or abdication of judicial review.” Miller-El, 537 U.S. at 340, 123 S.Ct. 1029. Indeed, the Supreme Court has said that we “can disagree with a state court's credibility determination” in resolving a habeas claim.¹⁹ Id.

III

Now onto the merits.

A

Remember, the state trial court said it wouldn't credit Juror 7's testimony that the community uproar prejudiced her. It offered some reasons to support that finding (as well as the subsequent inference that she likely had either “a change of heart, or a change of perception”). And my colleagues, echoing that reasoning, conclude there is “record support for the ․ factual predicates” undergirding the credibility call and the subsequent “change of heart/change of mind” inference; those “factual predicates” are “immediate admonitions against discussing outside information, the lack of meaningful deliberative discussion of the extraneous material, and Juror 7's acknowledgment of her obligation to decide the case based only on the evidence.”

I cannot agree. I'll take the predicates seriatim.

First, the purported “immediate admonitions against discussing outside information.” For starters, “Juror 7 was the only juror who mentioned anyone saying anything to the effect that the jury should not consider such information.” Adames R&R, 2024 WL 4958309, at *22. Yet the state trial judge offered no reason “why she found Juror 7 credible as to this point only.” Id.

And, anyway, the phrase “immediate admonitions” vastly overstates things. The record reflects “that outside influences were discussed.” Id. at *25. But then (in Juror 7's words), someone “cut the conversation,” and that person “was like, we can't discuss that.” So (as Judge Dein explained in her R&R) “the conversation ‘shifted’ back to a deliberation of the facts and law,” but it “then wandered back to the topic of outside influences again,” at least one more time. Id.

So I don't see where these “immediate admonitions” are in the record, or how the occasional urging to stay on topic banished these topics’ influence from everyone's minds. Nor, of course, would such admonitions (if they even existed) have anything to do with Juror 7's internal response to these external pressures.

Second, the “lack of meaningful deliberative discussion of the extraneous material.” For one, finding legal significance in this dearth of deliberative evidence begs the question, in the phrase's formal logic sense. That's to say: of course there wasn't any such evidence, because the state trial judge said she specifically avoided questioning the jurors about “how the jury's decision was reached.” That's “deliberative discussion.” So (as Judge Dein makes clear) it's metaphysically impossible “to establish that the [extraneous] topics were not discussed” in a meaningful way “during deliberations.”²⁰ Id. at *21.

And setting aside that philosophical quandary, the record provides evidence aplenty of strong influence from the extraneous material. Testimony from eight jurors revealed (in Judge Dein's words) a “pervasive awareness of the community sentiment opposing the not-guilty verdicts.” Id. at *22. And that “sentiment was discussed both before and after the jurors were greeted by the trial judge,” and it “remained on people's minds after the verdict was reached.” Id. For instance, Juror 4 explained that the jury knew about a “group outside” that was “a little up in arms that we had passed [a ‘not guilty’] verdict on seven of the ten charges.” And Juror 13 noted discussion about how “the former police chief was involved in trying to rally people to, you know, kind of put pressure on us” (though Juror 13 couldn't recall if that observation was made before or after deliberations ended).

Lastly, I can see how the presence of evidence about “meaningful deliberative discussion of the extraneous material” would help show that the jury improperly considered the commotion in its decision-making. But the absence of such evidence about the group's deliberative dynamics offers no reason to discredit Juror 7's testimony about the machinations of her own mind and heart. And in both Parker and Turner (Adames's two principal Supreme Court analogues), prejudice existed even absent evidence that the outside influence was discussed in the courtroom. See Parker, 385 U.S. at 363, 87 S.Ct. 468; see generally Turner v. Louisiana, 379 U.S. 466, 85 S.Ct. 546, 13 L.Ed.2d 424 (1965). So that absence can't support the state trial judge's credibility call.

Third and finally, Juror 7's “acknowledgment” that she “would decide the case based only on the evidence.” I think the state trial court drew (and my colleagues draw) that acknowledgment from her affirmation that she understood the jury instructions. But yet again, Judge Dein put it best: based on Juror 7's written and oral testimony, she “struggled with ‘doing her duty’ both to the victim and the Defendant, but in the end felt compelled to follow the community's view.” Adames R&R, 2024 WL 4958309, at *25. But that Juror 7 acknowledged (in the post-trial questioning) that she didn't notify the state trial court about the Facebook post does not diminish her credibility any more than it would diminish the credibility of the other juror who also didn't notify the court, but whose testimony wasn't called into question. Id. at *16.

Likewise, the testimony about Juror 7 not telling the court about the Facebook post the morning of the verdicts doesn't repudiate her sworn statements (made more than a year later) that she knew about the post and the protest and that both influenced her. Instead, “there is no indication that [Juror 7] continued to think about, stress or obsess about the verdict after it was rendered, and she did not contact counsel on her own initiative: she simply responded to an inquiry.” Id. at *21. And so “there is nothing which would support the trial judge's conclusion that [Juror 7's] testimony was in any way buyer's remorse, as opposed to an honest attempt to answer the questions presented.” Id.

Although we “may not know or altogether understand the imponderables which cause one to think what [s]he thinks,” of course “one who is trying as an honest [wo]man to live up to the sanctity of [her] oath is well qualified to say whether [s]he has an unbiased mind in a certain matter.” Smith, 455 U.S. at 217 n.7, 102 S.Ct. 940. That the state trial court set aside her statements about her own mindset is truly baffling. And so my colleagues err seriously in upholding the state trial court's determination that Juror 7 wasn't credible in explaining her own partiality.

B

That's not all, though. Usually, “when the last state court to decide a prisoner's federal claim explains its decision on the merits in a reasoned opinion ․ a federal habeas court simply reviews the specific reasons given by the state court and defers to those reasons if they are reasonable.” Wilson v. Sellers, 584 U.S. 122, 125, 138 S.Ct. 1188, 200 L.Ed.2d 530 (2018). So I also feel obligated to look at what the MAC (“the last state court to decide” Adames's claim “in a reasoned opinion”) had to say about the state trial court's finding. As to Juror 7's self-reported partiality, here's the MAC's take in full (omitting only case cites):

The judge also explicitly discredited juror no. 7's statement that she was in fact influenced by the extraneous matter. The judge instead found that it appeared the juror “had either a change of heart, or a change of perception, after the trial.” “[T]he determination of the weight and credibility of the testimony is the function and responsibility of the judge who saw and heard the witnesses, and not of this court.” We defer to the judge's conclusions because she was in the best position to assess the credibility and the weight of the testimony. Accordingly, we cannot say that the judge committed an error of law, or abused her discretion, in finding that the defendant was not prejudiced by the jury's exposure to the extraneous matter.

Commonwealth v. Adames-Garcia, 158 N.E.3d 887 (Table) (Mass. App. Ct. 2020). More simply, the MAC affirmed because the state trial court should get deference. Id.

But, again, “even in the context of federal habeas, deference does not imply abandonment or abdication of judicial review.” Miller-El, 537 U.S. at 340, 123 S.Ct. 1029. This isn't the usual situation where a state trial court should get its typical deference. “The trial judge did not observe Juror 7 at the time of her exposure to the Facebook posts or to the protesters, and her limited questioning of Juror 7 was done over [Z]oom, more than a year after the events in question,” and so “the court did not observe Juror 7's demeanor in person.” Adames R&R, 2024 WL 4958309, at *21. Plenty of courts have noted concerns about the factfinder's diminished ability to assess credibility over Zoom.²¹ So, for those reasons, the MAC's reason to affirm the state trial court's finding isn't reasonable either.

C

Nor do the reasons that the federal district court offered in denying relief sway me differently. See Adames-Garcia, 759 F. Supp. 3d at 196-97.

First, the federal district court observed that the state trial court was in a “better position” to evaluate credibility. Id. at 196. True enough, but “better” does not mean “correct ex cathedra.” And a state trial court's credibility determinations cannot stand simply because they're a state trial court's credibility determinations -- they've still got to be record-supportable and make some sense. See Miller-El, 537 U.S. at 340, 123 S.Ct. 1029. Here, unfortunately, neither's the case.

Second, it's also true that a witness can be “credible sometimes but not all the time” (a point the district court raises to say it was fine to credit some parts of Juror 7's testimony, but not others). Adames-Garcia, 759 F. Supp. 3d at 196. But we should usually expect a judge acting as factfinder (as compared to a jury) to make some sense of internal discrepancies, even in a credibility-determination context. After all, “juries are not required -- indeed, as a joint lay body are scarcely able -- to give detailed explanations for their decisions; but trial judges in proceedings of this kind,” (i.e., criminal proceedings) “are expected to give some explanation ․ unless the basis is plain from the record.” Cf. United States v. Oquendo-Rivera, 586 F.3d 63, 68 (1st Cir. 2009) (concerning revocation of supervised release). And, given Juror 7's affidavit and testimony to the contrary, the basis for finding her not credible in stating that she herself was prejudiced was hardly “plain from the record.” Id.

And third, while the state trial court's assessment was “informed by the testimony of all of the other jurors interviewed,” that makes little difference for the inner turmoil that Juror 7 said she experienced, particularly given her perspective that the evidence “was not strong.” Adames-Garcia, 759 F. Supp. 3d at 197.

Like the federal district court, I get that it's “easy to be a Monday-morning quarterback with instant replay.” Id. at 198. But it doesn't take a close look at the all-22 film ²² to figure out the problem here -- Juror 7 laid out what happened in a crystal-clear affidavit, supplemented by pretty consistent testimony. Nothing that came up in the post-trial inquiry rebutted Juror 7's statements about her own prejudice. Vibes, hunches, and unsupported assertions from the state trial court, followed by deference for deference's sake from the reviewing courts, cannot alone serve as the basis to deny relief.

IV

Why does all that matter? Well, there's no factual predicates to support the state trial court's finding that Juror 7 was not credible in testifying that she was prejudiced by the community's outcry. (And thus the “change of heart/change of perception” inference rests on nothing but speculation, either.)

So let's set aside that illogical finding. Instead, take Juror 7 -- who (at risk of beating a dead horse) gave us no reason whatsoever to conclude that she “continued to think about, stress or obsess about the verdict after it was rendered” or that “her testimony was in any way buyer's remorse, as opposed to an honest attempt to answer the questions presented” -- at her word. See Adames R&R, 2024 WL 4958309, at *21. That means we have sworn statements from a juror stating she voted to convict based not on the evidence but on community pressure.

Now, remember that the Commonwealth, in the post-conviction proceedings, had to prove “beyond a reasonable doubt” (no small feat) that Adames wasn't prejudiced by the jury's exposure to the pressure. But how could it? One juror plainly admitted she was prejudiced, and nothing in the record -- beyond an unsupported and, in my view, irrational credibility determination -- contradicts such admission. The Sixth Amendment cannot tolerate a conviction arising out of such circumstances. See U.S. Const. amend VI (guaranteeing the accused “the right to ․ an impartial jury”). So Adames has sufficiently surpassed even the “clear and convincing” hurdle in showing a Sixth Amendment violation (though, as I noted earlier, the height of the hurdle that he must actually clear seems to be an open question). 28 U.S.C. § 2254(e)(1).

Finally, that's true even if Juror 7 was the only one who felt the heat during deliberations. See Parker, 385 U.S. at 366, 87 S.Ct. 468 (“In any event, petitioner was entitled to be tried by 12, not 9 or even 10, impartial and unprejudiced jurors.”); see also Fullwood v. Lee, 290 F.3d 663, 678 (4th Cir. 2002) (“[I]f even a single juror's impartiality is overcome by an improper extraneous influence, the accused has been deprived of the right to an impartial jury.” (citing Parker, 385 U.S. at 366, 87 S.Ct. 468)); cf. Dyas v. Poole, 317 F.3d 934, 937 (9th Cir. 2003) (“[I]f even one juror is biased by the sight of the shackles, prejudice can result.” (citing Parker, 385 U.S. at 366, 87 S.Ct. 468)).

* * * *

All told, the state trial court's explanation for rejecting Juror 7's under-oath statements about her own partiality makes no sense. The state trial court thus made an “unreasonable determination of fact” in holding that the Commonwealth proved beyond a reasonable doubt that the community uproar didn't prejudice Adames. See 28 U.S.C. § 2254(d)(2). So the proceedings tarnished the well-established right to a jury composed entirely of impartial, unprejudiced members. See Parker, 385 U.S. at 366, 87 S.Ct. 468. And thus, Adames is “in custody in violation of the Constitution ․ of the United States.” 28 U.S.C. § 2254(a). So, I say once more, I respectfully dissent.

FOOTNOTES

1.   More specifically, Petitioner was indicted on two counts alleging natural intercourse and two counts alleging unnatural (anal) intercourse.

2.   Judge Kathe Tuttman, sitting by designation for two weeks, presided over the trial and the first three days of deliberations. When her designation period ended, Judge Thomas Barrett assumed the bench for the final two days of deliberations, with Judge Tuttman participating remotely by telephone.

3.   The trial judge had instructed the jurors, among other things, to decide the case only on the evidence admitted at trial, not on anything they may have read, heard, or seen outside the courtroom.

4.   Juror 7 later testified that she had not observed the teal ribbons that day, but that another jury member had mentioned them.

5.   Juror 1, who served as foreperson, passed away after the trial and before Petitioner's new trial motion was filed.

6.   Extraneous material includes “any private communication, contact, or tampering, directly or indirectly, with a juror during a trial about the matter pending before the jury.” Remmer v. United States, 347 U.S. 227, 229, 74 S.Ct. 450, 98 L.Ed. 654 (1954).

7.   During oral argument, counsel for Petitioner stated that it was not his position that a court must automatically credit a juror's testimony about outside influence. Instead, he argued that Juror 7's testimony should have been credited absent sound reasons to discredit it. We understand that clarification to bear on his separate challenge under § 2254(d)(2), addressed below. So we address his briefed argument under § 2254(d)(1) here.

8.   Petitioner maintains that Juror 7's testimony also makes this case one of implied bias, and thus that he is entitled to habeas relief regardless of the state court's “no actual bias” finding. Assuming without deciding that implied bias is grounds for relief, compare Smith, 455 U.S. at 221, 102 S.Ct. 940 (O'Connor, J., concurring) (“[T]he opinion does not foreclose the use of ‘implied bias’ in appropriate circumstances.”), with Cutts v. Smith, 630 F. App'x 505, 509 (6th Cir. 2015) (“[T]he implied bias doctrine is not clearly established for the purposes of § 2254.”), it does not apply here. “Such a claim requires exceptional or extreme circumstances giving rise to an implication of bias.” United States v. Kuljko, 1 F.4th 87, 93 (1st Cir. 2021) (citation modified). Examples of such circumstances include “the juror is an actual employee of the prosecuting agency, that the juror is a close relative of one of the participants in the trial or the criminal transaction, or that the juror was a witness or somehow involved in the criminal transaction.” Smith, 455 U.S. at 222, 102 S.Ct. 940 (O'Connor, J., concurring). The extraneous influence here falls short of that standard.

9.   Petitioner also characterizes the court as improperly considering whether Juror 7 would have known whether other jurors had been prejudiced. He maintains that fact is irrelevant because a single influenced juror is sufficient to establish prejudice. But the court did not suggest that prejudice to one juror would be insufficient. Rather, the court considered Juror 7's claimed knowledge of other jurors’ mental states as part of its assessment of how much weight to give to her testimony.

10.   It is true that Judge Tuttman specifically avoided questioning the jurors about how the decision was reached. But the Supreme Court has upheld a bar on interrogation of jurors over their internal mental processes even in the face of Sixth Amendment fair jury arguments. See, e.g., Tanner v. United States, 483 U.S. 107, 116-28, 107 S.Ct. 2739, 97 L.Ed.2d 90 (1987).

11.   We note there is “some tension” between this provision and § 2254(e)(1), which provides that “ ‘a determination of a factual issue made by a State court shall be presumed to be correct’ unless rebutted ‘by clear and convincing evidence.’ ” Watson, 118 F.4th at 459 n.2 (1st Cir. 2024) (first quoting Porter, 35 F.4th at 79; and then quoting § 2254(e)(1)). The Supreme Court has declined to resolve how to harmonize these provisions and so have we. Id. As Petitioner does not prevail under § 2254(d)(2), we need not “decide whether that determination should be reviewed under the arguably more deferential standard set out in § 2254(e)(1).” Wood v. Allen, 558 U.S. 290, 301, 130 S.Ct. 841, 175 L.Ed.2d 738 (2010).

12.   The dissent argues that “[t]his isn't the usual situation where a state trial court should get its typical deference” because Juror 7 was questioned over Zoom, so the court did not observe Juror 7's demeanor in person. But the fact that testimony was taken by Zoom does not necessarily diminish the state court's credibility findings. See PRN Real Est. & Invs., Ltd. v. Cole, 85 F.4th 1324, 1339 (11th Cir. 2023). “We generally defer to the trier of fact's credibility determination because the fact finder heard the witness's testimony and saw his demeanor, while we are stuck with a cold paper record.” Id. (citation modified). And even if “in-person testimony is preferable to a live video stream ․ the live video stream gave the ․ court greater insight into [Juror 7's] credibility than the cold paper record gives us.” Id.

13.   In fact, Petitioner relies on the fact that “Juror 7 corrected the state court in ways that made her testimony less helpful to [Petitioner]” to support his theory that the court had no evidence that Juror 7 was inclined to help him because of a change of heart. Specifically, he references the moment the court asked Juror 7 what she had observed about “blue ribbons,” and Juror 7 clarified that she “didn't observe it”; another juror had merely mentioned it.We are not persuaded. That clarification concerned the very issue on which Juror 7 had already given inconsistent accounts. In the affidavit that triggered the hearing in the first place, Juror 7 swore that she personally “saw these people wearing teal sexual assault ribbons and teal shirts on [her] way to court.” See supra note 3. The hearing exchange therefore did not compel the state court to find Juror 7 reliable. If anything, it gave the court another reason to question her consistency and credibility.

14.   Petitioner points to a comment about bringing a sharpened CD into the courtroom, but the record shows said comment was posted after the jury rendered its final verdict.

15.   Unless otherwise indicated, my case quotations omit all internal citations, quotation marks, footnotes, alterations, and subsequent history.

16.   The state trial judge's response to this motion is perhaps surprising, given Massachusetts caselaw about the importance of examining extraneous influence on juries. See, e.g., Commonwealth v. John, 442 Mass. 329, 812 N.E.2d 1218, 1226 (2004) (“Whenever a claim of extraneous influence on the jury arises, the trial judge should determine, within his discretion, whether there exists a serious question of possible prejudice. If the judge determines such a question exists, he should conduct a voir dire of the jurors.”). But Adames doesn't press that point before us, so I won't linger on it.

17.   The hearing was virtual for a few logistical reasons, including complications with transporting Adames to the island for in-person proceedings.

18.   Importantly, our circuit hasn't taken a definitive view about the tension between 28 U.S.C. § 2254(d)(2), which requires showing an “unreasonable determination of the facts,” and § 2254(e)(1), which requires “clear and convincing evidence” to rebut a state court's factual determination. See Porter, 35 F.4th at 79. And our circuit normally assumes that the “arguably more stringent standard” of clear and convincing evidence “applies.” See id. I think Adames (who did identify the discrepancy for us) satisfies either, but certainly clears the lower bar set out by § 2254(d)(2). But our colleagues might do well to sort out this evidentiary-standard confusion in further review, particularly to the extent it matters here.

19.   True, our court (quoting the Supreme Court) once said that a federal habeas court “has no license to redetermine credibility of witnesses whose demeanor has been observed by the state trial court, but not by them.” Caldwell v. Maloney, 159 F.3d 639, 650 (1st Cir. 1998) (quoting Marshall v. Lonberger, 459 U.S. 422, 432-34, 103 S.Ct. 843, 74 L.Ed.2d 646 (1983)). But that preceded the Supreme Court's Miller-El decision, which says precisely the opposite.

20.   True, Massachusetts caselaw seems to prohibit probing the actual effect on the jurors’ thought processes. But that makes the case even clearer: if such questions can't be asked, why would that absence of material drive the outcome?

21.   Concerns about assessing credibility over a screen have been around almost as long as videoconferencing itself. See Rusu v. I.N.S., 296 F.3d 316, 322 (4th Cir. 2002) (explaining that “video conferencing may render it difficult for a factfinder in adjudicative proceedings to make credibility determinations and to gauge demeanor” (collecting cases)).And such concerns remain pertinent in the age of Zoom. See, e.g., United States v. Jenkins, No. 2:24-MJ-00002-EJY, 2023 WL 9062283, at *2 (D. Nev. Dec. 28, 2023) (“The Court must be able to weigh the credibility of witnesses at a bench trial and, as good as video appearances may be, they are not reliable and do not allow for the observations that come with in-person testimony.”); Pinson v. United States, No. CV-19-00422-TUC-RM, 2023 WL 8376209, at *3 (D. Ariz. May 31, 2023) (“Videoconferencing may impede a fact-finder's ability to observe demeanor and assess credibility.”); Howitt v. Massachusetts, No. CV 24-13207, 2025 WL 2146607, at *2 (D. Mass. July 29, 2025) (“The state court judge gave a valid explanation why (after originally commencing the trial via Zoom) conducting the trial remotely would not be a ‘reasonable’ accommodation -- that the limitations of the courtroom computer system ‘seriously compromised’ the judge's ‘ability to assess the credibility of each party’ or to view the dozens of disputed trial exhibits.”).So while it's fair to say (as the majority does) that the live video feed gave the state trial judge more insight into Juror 7's demeanor than the “cold paper record” gives us, I think that misses the point. The deference due to the state trial judge based on institutional capabilities is surely lessened in this scenario, and the state trial judge's greater insight into credibility than ours does not mean it is unquestionably correct.

22.   Sticking with the district court's football imagery, the “all-22 film” is what players and coaches use to break down each snap, “all-22” being a reference to the camera angle that lets the viewer see all twenty-two players on the field at once (a view you don't normally get on your typical Sunday TV broadcast).

GELPÍ, Circuit Judge.