This habeas appeal presents several challenging questions. Two of these are particularly intriguing. The first involves the effect of an eyewitness’s expression of less than complete certitude about a crucial out-of-court identification that he previously made. The second is a multi-part question. The initial part deals with the *420cut-off point for determining what constitutes “clearly established Federal law” within the purview of 28 U.S.C. § 2254(d)(1), a provision of the Antiterrorism and Effective Death Penalty Act of 1996 (AEDPA), Pub.L. No. 104-132, 110 Stat. 1214. The next part of that question concerns the selection of the relevant precedent from the Supreme Court’s evolving Confrontation Clause jurisprudence—a question made relevant by the Supreme Court’s decision, in 1998, of Gray v. Maryland, 523 U.S. 185, 118 S.Ct. 1151, 140 L.Ed.2d 294 (1998). Because Gray has not been made retroactive to cases under collateral review, this case turns on the applicability vel non of that precedent to the redacted statement of a nontestifying co-defendant under circumstances in which his objecting codefendant’s name has been replaced with a cryptic designation (“Mr. X”). As matters turn out, the answer to this part of the inquiry depends on the answer to the initial part.
These questions arise in the context of a state-court conviction for second-degree murder. In the proceedings below the district court, acting under habeas jurisdiction, granted relief because it deemed the evidence insufficient to support the conviction and, secondarily, because it deemed the admission of the nontestifying codefendant’s statement violative of the petitioner’s rights under the Confrontation Clause of the Sixth Amendment. See Foxworth v. Massachusetts (Foxworth III), No. 03-11844, slip op. at 27 (D.Mass. May 14, 2008) (unpublished); Foxworth v. Maloney (Foxworth I), No. 03-11844, slip op. at 19 (D.Mass. Aug. 17, 2006) (unpublished).
After a lengthy exegesis through this maze of problems, we reverse in part, retain jurisdiction, and certify a critical question of state law to the Massachusetts Supreme Judicial Court (SJC).
I. BACKGROUND
Because this appeal involves a challenge to evidentiary sufficiency, we rehearse the facts in the light most compatible with the verdict rendered by the state-court jury, consistent with record support. See Jackson v. Virginia, 443 U.S. 307, 319, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979). In a wrinkle peculiar to the exercise of federal habeas jurisdiction we grant a presumption of correctness to the state courts’ factual determinations. See 28 U.S.C. § 2254(e)(1).
Roxbury is an enclave in Boston, Massachusetts. On May 23, 1991, a group of men went to the Roxbury home of Kenneth McLean bent on buying drugs. After they arrived, matters got out of hand. Apparently, McLean was beaten. He then broke free and ran. At least one of the men chased McLean and shot him as he fled. McLean’s wounds proved fatal.
In July of 1991, a Suffolk County grand jury indicted petitioner-appellee Robert Foxworth for the murder. By way of a superceding indictment, the grand jury also charged two other men, Troy Logan and Ronald Christian, with the murder. As the case unfolded, many of the facts were undisputed; the shooter’s identity, however, was hotly contested. This factual dispute became the focal point of the ensuing trial.
The prosecution’s case against the petitioner hinged on an eyewitness identification by Derek Hobson. The petitioner filed a pretrial motion to suppress the identification testimony. His motion was denied.
At trial, Hobson testified that, at approximately 6:00 p.m. on May 23, he was walking down Brookford Street in Roxbury. He observed a man run out of a building located at 5 Brookford Street. The man yelled: “Those people are crazy.” Then, another man (later identified as *421Kenneth McLean) exited the premises and ran pell-mell down the street. He had blood on his mouth and tape on one arm.
Not heroically inclined, Hobson hid behind a car parked directly across from 5 Brookford Street. From that vantage point, he observed another man emerge from the building with a gun. That man fired three or four shots at McLean, who collapsed.
Hobson saw the shooter for at least forty seconds from a distance of approximately fifteen to twenty feet. When the police arrived, he described the shooter as a medium-complected black male, six feet or six feet one inch in height, weighing one hundred forty pounds, and sporting a one inch “tail” that protruded from the back of his head. At trial, Hobson added that the shooter was wearing a black baseball cap and that the “tail” stuck out from under the back of the cap.
On June 17, Detective Daniel Flynn, the officer in charge of the investigation, visited Hobson and presented him with a photo array. The array contained twenty photographs, including photographs of all three men eventually accused of the murder. Hobson selected the petitioner’s photograph from the array, identifying him as the shooter. Flynn testified that Hobson acted “without hesitation.”
Later that month, the police showed Hobson another photo array. Once more, Hobson selected the petitioner’s photograph from the array and identified him as the shooter.
At trial, Hobson did not make a live ineourt identification but confirmed that the petitioner was the person he had identified from the photo arrays. On cross-examination, he acknowledged that he had not seen the shooter head-on but had “seen the whole like side of his face.” He also admitted that he had based his selections from the photo arrays in part on the fact that the man he remembered had a “tail” (and the petitioner was the only person with a “tail” whose picture was displayed). When pressed by defense counsel to gauge his confidence in the identification, Hobson stated that he was “eighty percent sure.”
Anthony McAfee, who was strolling along with Hobson immediately before the shooting, testified that he first observed a man running from the house, yelling “[t]hose people are crazy.” Then, McLean scampered from the building. Two other men followed. One of that pair got into a car parked at the curb on Brookford Street and passed a black object to the other man. The latter proceeded to fire five or six shots at McLean. When Flynn presented McAfee with the same photo array that he initially had presented to Hobson, McAfee identified Logan as the man who handed over the black object. McAfee could not identify the shooter.
The petitioner had filed a pretrial motion to sever his trial from that of his codefendants. He premised that motion on the potential prejudice inherent in a statement made by Logan to the police. The motion was unsuccessful, and the statement was prominently featured at the joint trial.
Detective Flynn read the statement into evidence over the petitioner’s timely objection. In an effort to avert the feared prejudice, the trial justice ordered redaction of the statement. In its redacted form, the petitioner’s name was replaced with the pseudonym “Mr. X” and the statement was altered so that, after the first reference to “Mr. X,” it appeared (falsely) that Logan had stated: “Cause that’s how I know him by. I don’t know his real name.” The trial justice also admitted a copy of the redacted statement into evidence.
In the statement, Logan related that, on the evening in question, he met two men named “Tea Lover” and “Mr. X” at a *422sandwich shop. “Tea Lover” and Logan resolved to go to McLean’s apartment to buy cocaine. “Mr. X” accompanied them because McLean owed him money as a result of a prior sale of “bad” cocaine. Shortly after the group’s arrival, “Mr. X” and McLean argued. “Mr. X” left the premises. Logan opted to do likewise. As he made his way downstairs, “Mr. X” was coming back upstairs. Logan noticed that “Mr. X” had a gun tucked into his waistband. Logan continued his descent and, as he was getting into a cab, heard shots.
The statement indicated that Logan had identified “Mr. X” from a photo array. Flynn testified that the photo array was the same one that had been employed earlier in the investigation. On cross-examination by Christian’s counsel, Flynn revealed that the designation “Mr. X” did not refer to Christian. Moreover, it was a replacement for the name Logan actually had used in his statement.
The prosecution offered the redacted statement as evidence only against Logan. The trial justice gave a limiting instruction both after the reading of the redacted statement and before jury deliberations began. The gist of the instruction was twofold: that the statement could only be considered against Logan (who was being tried on a “joint venture” theory) and that the jury should not speculate about who “Mr. X” might be.
At the close of the prosecution’s case, each of the three defendants moved for a required finding of not guilty. The trial justice granted Christian’s motion in its entirety and granted the other defendants’ motions with respect to the charge of first-degree murder. These rulings preserved the second-degree murder charges against the petitioner and Logan. On March 31, 1992, the jury convicted the petitioner and acquitted Logan. In due course, the trial justice sentenced the petitioner to life imprisonment.
The petitioner filed a new trial motion under Mass. R.Crim. P. 30. The motion alleged a Bruton violation based on admission of Logan’s statement. It also cited purported newly discovered evidence, alleged prosecutorial misconduct, and challenged the photo spread used to identify the petitioner. The trial justice denied the motion on August 8, 1994. The petitioner’s timely direct appeal to the Massachusetts Appeals Court from the conviction, which advanced similar claims, was consolidated with his subsequent appeal from the denial of the motion. On October 21, 1996, the Appeals Court, in an unpublished decision, affirmed both the conviction and the denial of the new trial motion.
The petitioner had twenty days in which to seek further review by the SJC by filing an application for leave to obtain further appellate review (ALOFAR). Mass. R.App. P. 27.1(a). The petitioner made no such filing within the prescribed period.
Ordinarily, then, the conviction would have become final in 1996. Here, however, the petitioner some four years later (on October 25, 2000) filed a pro se ALOFAR in the SJC seeking further review of the Appeals Court decision. The pro se ALO-FAR included a “motion to file late application for further appellate review,” which asked the SJC to excuse his untimely filing on two grounds.
The pro se ALOFAR argued that the petitioner’s Sixth Amendment rights under Bruton had been violated by the introduction of Logan’s statement but made no mention of Gray (which had been decided on March 9, 1998). The ALOFAR also raised several other challenges to the 1996 Appeals Court ruling, none of which are material here.
The SJC did not rule either on the ALOFAR or on the petitioner’s motion to file out of time because, on November 15, 2000, the petitioner asked the SJC to stay *423consideration of his ALOFAR pending the resolution of a second new trial motion. The SJC granted the petitioner’s request for a stay on February 1, 2001, and ordered petitioner to file regular status reports on the progress of his second new trial motion.
The petitioner filed his second new trial motion before the trial court on December 11, 2000. That motion, filed pursuant to Mass. R.Crim. P. 30, attacked the conviction on four grounds. First, it claimed, based on a later affidavit by McAfee, that the prosecutor knowingly and intentionally used perjured testimony and allowed false evidence to go uncorrected. Second, it claimed that the petitioner’s previous counsel provided ineffective assistance because he failed to submit affidavits anent newly discovered evidence. Third, it claimed that the trial justice erred in not instructing the jury on manslaughter. Fourth, it claimed that the judge gave an incorrect instruction as to malice.
After the trial court denied the second new trial motion, the Appeals Court, on April 17, 2002, affirmed the decision, holding that all the claims in the second new trial motion were either previously decided or waived.
The petitioner then moved to consolidate his appeal from the denial of the second new trial motion with his previous ALO-FAR. The SJC appointed counsel on May 22, 2002, and on May 28 granted the petitioner’s motion to extend the time for filing to June 24, 2002. On June 27, 2002, counsel informed the SJC that an amended ALOFAR would be filed.
On July 22, 2002, the petitioner filed a “motion to file an amended FAR application late,” which was allowed by a docket notation of the same day, without comment. It appears from the record that the allowance of this motion related to the petitioner’s non-observance of the June 24 deadline for filing briefs. It is unclear whether that order may have applied to the petitioner’s failure, under Mass. R.App. P. 27.1(a), to file and ALOFAR within twenty days of either the October 21, 1996 Appeals Court decision affirming the conviction and the denial of the first new trial motion or the April 17, 2002 Appeals Court decision affirming the denial of the second new trial motion.
Also on July 22, the petitioner, through appointed counsel, filed his amended ALO-FAR. The July 22 amended ALOFAR was based primarily on the Bruton claim; here, for the first time, the petitioner raised Gray as an explicit basis for his claim. The amended ALOFAR also raised challenges to the photo array and to the sufficiency of the evidence. Although the amended ALOFAR was consolidated with the appeal from the denial of the second new trial motion, it did not raise any challenges to the Appeals Court’s 2002 decision.
On September 6, 2002, the SJC denied the ALOFAR without comment.
The petitioner repaired to the federal district court in search of a writ of habeas corpus. See 28 U.S.C. §§ 2244-2254. Naming an appropriate correctional official as the respondent, his habeas petition raised three fully exhausted claims: (i) that his conviction violated due process because the evidence was constitutionally insufficient under the rule of Jackson, 443 U.S. at 319, 99 S.Ct. 2781; (ii) that the use of an unreliable eyewitness identification violated due process, see Neil v. Biggers, 409 U.S. 188, 198, 93 S.Ct. 375, 34 L.Ed.2d 401 (1972); and (iii) that the admission at trial of Logan’s incriminating statement violated his Sixth Amendment rights as announced in Bruton v. United States, 391 U.S. 123, 126, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968), and elaborated in Gray.
On August 17, 2006, the district court ruled in favor of the petitioner on his *424Bruton claim. The court regarded the two-part inference that “Mr. X” was the shooter and that the petitioner was “Mr. X” as obvious; therefore, “[t]he statement-given by a codefendant with powerful motive to incriminate petitioner, and unchallenged by cross-examination—violated petitioner’s Sixth Amendment rights.” Foxworth I, slip op. at 19. The district court deemed consideration of the petitioner’s two remaining claims unnecessary, vacated the conviction, and ordered the Commonwealth either to retry or to release the petitioner. Id. at 23.
The respondent appealed. We remanded the case to the district court to address the sufficiency of the evidence claim, noting that a finding in the petitioner’s favor on that claim would preclude a retrial. See Foxworth v. Maloney (Foxworth II), 515 F.3d 1, 4 (1st Cir.2008).
The district court sensibly ordered the petitioner’s release on bail and proceeded to address the unadjudicated claims. With respect to the due proeess/eyewitness identification claim, the court determined that although the identification procedure was suggestive to a degree, the state court’s conclusion that any suggestiveness was outweighed by Hobson’s ability to observe at the time of the incident was not objectively unreasonable. Foxworth III, slip op. at 17. Thus, the admission of Hobson’s testimony did not transgress due process. Id. As to the insufficiency claim, the court concluded that “it was objectively unreasonable for the [Massachusetts] Appeals Court to do no more than cite the general standard when, even after crediting the entire testimony of the witness, there was still significant doubt in his identification of Foxworth.” Id. at 26 (emphasis in original). On that basis, the court set aside the conviction and ordered the petitioner’s release. Id. at 28.
This timely appeal ensued. In it, the respondent seeks to test the mettle of the petitioner’s insufficiency and Sixth Amendment claims. The due proeess/eyewitness identification ruling has not been challenged.
II. ANALYSIS
We begin our substantive discussion by laying out the ground rules for federal habeas review of state-court convictions. We then address sequentially the two constitutional claims that are before us.
A. Standards of Review.
Federal habeas review of a state-court conviction is governed by the AED-PA, which permits federal courts to grant habeas relief after a final state-court adjudication of a federal constitutional claim if that adjudication can be shown to be “contrary to,” or to have involved, “an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States” or in the alternative, to have been “based on an unreasonable determination of the facts in light of the evidence presented in the State court proceedings.” 28 U.S.C. § 2254(d). In administering these standards, the state court’s factual findings are presumed to be correct, and they can be overcome only by clear and convincing evidence. See id. § 2254(e)(1); see also Ouber v. Guarino, 293 F.3d 19, 27 (1st Cir.2002).
A state-court decision is “contrary to” clearly established Federal law if “the state court arrives at a conclusion opposite from that reached by the U.S. Supreme Court on a question of law, or if the state court decides the case differently than the U.S. Supreme Court has on a set of materially indistinguishable facts.” Sleeper v. Spencer, 510 F.3d 32, 37-38 (1st Cir.2007) (citing Williams v. Taylor, 529 U.S. 362, 405, 120 S.Ct. 1495, 146 L.Ed.2d 389 (2000)). This is an “unreasonable application” case; no colorable arguments are *425made suggesting that the relevant state-court rulings are directly “contrary to” clearly established Supreme Court precedent.
An unreasonable application of clearly established Federal law occurs when
the state court correctly identifies the governing legal principles, but (i) applies those principles to the facts of the case in an objectively unreasonable manner; (ii) unreasonably extends clearly established legal principles to a new context where they should not apply; or (iii) unreasonably refuses to extend established principles to a new context where they should apply.
Id. at 38. “The habeas question of whether the state court decision is objectively unreasonable is layered on top of the underlying standard governing the constitutional right asserted.” Hurtado v. Tucker, 245 F.3d 7, 16 (1st Cir.2001). So viewed, the state court’s decision is not vulnerable unless it evinces some increment of incorrectness beyond mere error. McCarnbridge v. Hall, 303 F.3d 24, 36 (1st Cir.2002) (en banc).
In probing whether a state-court decision is an unreasonable application of clearly established Federal law, the nature of the legal rule—that is, the extent to which the rule is specific rather than general—makes a substantial difference. If the legal rule is specific, the range of reasonable judgments is correspondingly narrow. Conversely, if the legal rule is general, the range of reasonable judgments is likely to be broad. Yarborough v. Alvarado, 541 U.S. 652, 664, 124 S.Ct. 2140, 158 L.Ed.2d 938 (2004). In other words, “[t]he more general the rule, the more leeway courts have in reaching outcomes in case-by-case determinations.” Id.
Even if a state-court decision is determined to involve an unreasonable application of clearly established Federal law, habeas relief will not follow automatically. The error must be shown to have “had a substantial and injurious effect or influence in determining the jury’s verdict.” Delaney v. Bartee, 522 F.3d 100, 105 (1st Cir.2008) (quoting Brecht v. Abrahamson, 507 U.S. 619, 631, 113 S.Ct. 1710, 123 L.Ed.2d 353 (1993)).
Finally, we note that these ground rules apply equally to the district court and to the court of appeals alike. Consequently, we review a district court’s grant or denial of a state prisoner’s habeas corpus petition de novo. Lynch v. Ficco, 438 F.3d 35, 44 (1st Cir.2006). With these background rules in place, we turn next to the sufficiency of the evidence.
B. Due Process: Sufficiency of the Evidence.
In criminal eases, the constitutional benchmark for evidentiary sufficiency is familiar: “If the evidence presented, taken in the light most flattering to the prosecution, together with all reasonable inferences favorable to it, permits a rational jury to find each essential element of the crime charged beyond a reasonable doubt, then the evidence is legally sufficient.” United States v. Olbres, 61 F.3d 967, 970 (1st Cir.1995) (citing Jackson, 443 U.S. at 319, 99 S.Ct. 2781). Due largely to the expansiveness of this standard, a sufficiency analysis requires a degree of intellectual rigor. In particular, a reviewing court must refrain from giving credence to “evidentiary interpretations and illations that are unreasonable, insupportable, or overly speculative.” United States v. Spinney, 65 F.3d 231, 234 (1st Cir.1995).
In exercising federal habeas jurisdiction we typically look to the rationale of the intermediate appellate court where, as here, the state’s highest court has summarily denied further appellate re*426view. See, e.g., Niland v. Hall, 280 F.3d 6, 11-12 (1st Cir.2002). In this instance, the measuring stick that the Massachusetts Appeals Court applied to the sufficiency analysis meets federal constitutional criteria. Although that court relied on a state case, Commonwealth v. Latimore, 378 Mass. 671, 393 N.E.2d 370 (1979), in reaching a conclusion about evidentiary sufficiency, a state-court adjudication of an issue framed in terms of state law is nonetheless entitled to deference under section 2254(d)(1) as long as the state and federal issues are for all practical purposes synonymous and the state standard is at least as protective of the defendant’s rights as its federal counterpart. See McCambridge, 303 F.3d at 35. Since the sufficiency issue is essentially the same under both federal and state law and the Latimore court transplanted the appropriate federal constitutional standard into the jurisprudence of Massachusetts, see Latimore, 393 N.E.2d at 374 (citing Jackson, 443 U.S. at 318-19, 99 S.Ct. 2781), we can confidently apply the AEDPA standard to this issue.
The Appeals Court concluded that:
The judge properly denied Foxworth’s motion for a required finding of not guilty, on the basis, if no other, of the eyewitness testimony identifying Fox-worth as the person who shot the fleeing McLean several times and then escaped in a car—testimony that was corroborated in all respects, except the identification of Foxworth as the shooter, by a second eyewitness. That evidence, taken in the light most favorable to the Commonwealth (a premise that Fox-worth fails to recognize), was sufficient for a rational jury to find beyond a reasonable doubt that Foxworth was a principal actor in the second degree murder, with malice inferable from the intentional use of a deadly weapon. Foxworth’s presentation of an alibi defense did not serve to take the issue from the jury.
The question, then, is whether this determination constitutes an unreasonable application of the Jackson standard.
The petitioner asserts that the only evidence of his involvement was the testimony of a single eyewitness and that this testimony lacked the force necessary to prove guilt beyond a reasonable doubt. The district court agreed; it found multiple weaknesses in Hobson’s testimony, including discrepancies between his account and McAfee’s, inconsistencies in his description of the shooter, the lack of an in-court identification, flaws in the photo array, and the witness’s admission that he was less than certain about the identification. See Foxworth III, slip op. at 22. The respondent assigns error to this ruling.
The question is close, and the district court is correct that Hobson’s testimony is the only evidence that directly implicates the petitioner as the shooter.1 Nevertheless, a criminal conviction can rest on the testimony of a single eyewitness. Even if the eyewitness’s testimony is uncorroborated and comes from an individual of dubious veracity, it can suffice to ground a conviction. See, e.g., Hayes v. Battaglia, 403 F.3d 935, 938 (7th Cir.2005). Hobson’s first-hand testimony was neither inherently improbable nor materially undermined by any other unimpeachable proof. It placed the petitioner at the scene of the murder and identified him as the shooter. No more was exigible to satisfy the Jackson standard.
*427That is not to say that a rational jury had to accept the eyewitness identification. Hobson was exposed to a withering cross-examination, which disclosed a number of weaknesses in his testimony. It is well-established, though, that determining a witness’s credibility, even in the face of a furious attack, is a function that falls squarely within the province of the jury. See, e.g., United States v. Lipscomb, 539 F.3d 32, 40 (1st Cir.2008); Ramonez v. Berghuis, 490 F.3d 482, 490 (6th Cir.2007); United States v. Calderon, 77 F.3d 6, 10 (1st Cir.1996).
In defending the district court’s sufficiency ruling, the petitioner notes that some of the details to which Hobson testified did not match his original statements to the police. In a similar vein, he points out that Hobson’s testimony was not entirely congruent with McAfee’s testimony. But such discrepancies do not as a matter of law render a witness’s testimony unworthy of belief. Rather, they are for the jury to sort out and weigh. “The actual resolution of the conflicting evidence, the credibility of witnesses, and the plausibility of competing explanations is exactly the task to be performed by a rational jury, considering a case presented by competent counsel on both sides.” Matthews v. Abramajtys, 319 F.3d 780, 790 (6th Cir.2003). It is rare that a key witness survives a murder trial unscathed, and the jurors in this case were free to credit Hobson’s version of the events or not, as they saw fit.
The absence of an in-court identification does not tip the sufficiency scales. Extrajudicial witness identifications are routinely used as substantive evidence of guilt. See, e.g., Samuels v. Mann, 13 F.3d 522, 527 (2d Cir.1993); see also Fed. R.Evid. 801(d)(1)(C) (excluding prior out-of-court identifications from the definition of hearsay). There is no requirement, either in the Constitution or in the usual rules that apply to the admission of evidence, that a witness who makes an extrajudicial identification must repeat the identification in the courtroom.2 Consequently, the fact that Hobson never identified the petitioner at trial does not alter the constitutional calculus.
The petitioner offers a closely related argument. He maintains that Hobson’s out-of-court identification was itself so unreliable that a rational jury could not rely on it. In this regard, he stresses that the district court characterized the photo array as suggestive. See Foxworth III, slip op. at 16.
The bottom line, however, is that the district court concluded that it was not objectively unreasonable for the state court to find enough indicia of reliability to allow the introduction of the identification testimony into evidence without offending any constitutional safeguard. See id. at 16-17. The petitioner has not challenged that conclusion on appeal.
Moreover, the alleged flaws in the photo array were fully vetted at the trial; the cross-examiner dwelt on them and argued the point vociferously to the jury. He also exploited the testimony about viewing angles (i.e., that the view Hobson had of the petitioner was in profile). Despite the force of this multi-pronged attack, the jury credited Hobson’s identification. On collateral review, we cannot disturb its judgment. Ramonee, 490 F.3d at 490.
Finally, the district court gave significant weight to the fact that Hobson admitted that he was only “eighty percent sure” of his identification. The court reasoned that even if the testimony was fully credit*428ed by the jury, it carried with it a significant measure of doubt and, therefore, “it would be irrational to find that fact beyond a reasonable doubt [when] the eyewitness himself was not sure of the fact....” Foxworth III, slip op. at 24-25. On appeal, the respondent savages this analysis while the petitioner embraces it.
Although the district court’s reasoning has a patina of plausibility, it does not withstand close scrutiny. A prior identification is not stripped of probative force merely because the witness confesses that he harbors some doubt about it. Cf. Samuels, 13 F.3d at 527 (holding that a jury could credit a witness’s earlier identification of the defendant over the witness’s in-court identification of a different person). The witness’s certainty vel non is properly viewed as a factor that must be evaluated by the jury as the trier of the facts—but a jury is no more bound to find reasonable doubt based on an eyewitness’s profession that he is “eighty percent certain” than it would be bound to accept the identification if the witness professed to being “one hundred percent certain.”
This conclusion is rooted in human experience and common sense. Self-esteem is a wild card. Some witnesses may be over-confident about their powers of perception or recall; others may be diffident or overly cautious. Moreover, “some witnesses verbalize their assurances of being correct with more positiveness than others.” United States v. Smith, 563 F.2d 1361, 1363 (9th Cir.1977). And, finally, there is no necessary correlation between a witness’s self-confidence and the correctness of his identification. In the last analysis, it is up to the jury to gauge the accuracy of an identification, factoring in any protestations of assurance or self-doubt.3
In this case, there is no principled way to label the jury’s choice as irrational. Hobson selected the petitioner’s picture from each of two separate (constitutionally adequate) photo arrays. The first time, he signed the back of the photograph to confirm the identification. Detective Flynn testified that Hobson acted “without hesitation.” He later made a second identification from another photo array. These identifications were made in close proximity to the time of the murder. In addition, the petitioner resembled the general description given by Hobson at that time. Hobson’s subsequent expression of less than complete certitude about the identifications came nearly a year later (after his memory arguably had dulled). Given the totality of the circumstances, we conclude that Hobson’s eyewitness testimony could reasonably be thought to comprise part of a constitutionally adequate foundation for the ensuing conviction.
Of course, the fact that each weakness in Hobson’s testimony can be overcome does not mean that, in the aggregate, they can be overcome. Cf. Bourjaily v. United States, 483 U.S. 171, 179-80, 107 S.Ct. 2775, 97 L.Ed.2d 144 (1987) (explaining that “individual pieces of evidence, insufficient in themselves to prove a point, may in cumulation prove it”). The district court made this point quite forcefully. See Foxworth III, slip op. at 25-26. Under the AEDPA, however, the question is not whether the outcome reached by the state court is correct in an absolute sense but, rather, the reasonableness of that court’s application of clearly established Federal law in reaching that outcome. See Hurtado, 245 F.3d at 20.
We mention one last point. We have said that the underlying constitution*429al question is close. The very closeness of the question solidifies the result that we must reach in this case. Habeas relief is precluded when “it is a close question whether the state decision is in error.” Evans v. Thompson, 518 F.3d 1, 7 (1st Cir.2008) (quoting McCambridge, 303 F.3d at 36). That is especially so when the state court is tasked with applying a general standard, such as that contemplated by Jackson. See Yarborough, 541 U.S. at 664, 124 S.Ct. 2140; see also Wright v. West, 505 U.S. 277, 308, 112 S.Ct. 2482, 120 L.Ed.2d 225 (Kennedy, J., concurring) (identifying Jackson as enunciating a general standard). The generalized nature of the Jackson standard reinforces our conclusion that it was not unreasonable for the Appeals Court to hold that Hobson’s testimony, together with the other evidence of record, was sufficient to under-gird the conviction.
The petitioner, ably represented, tries yet another tack. Arguing for a finding of unreasonableness, he emphasizes the paucity of reasoning in the Appeals Court’s decision. In his view, this perceived flaw is aggravated by that court’s ostensible overstatement of the corrobative effect of McAfee’s testimony. These arguments have some bite but, on habeas review, the ultimate inquiry is not the degree to which the state court’s decision is or is not smoothly reasoned; the ultimate inquiry is whether the outcome is reasonable. See Hurtado, 245 F.3d at 20. A sparsely reasoned state-court decision may set off warning bells, but such a decision does not necessarily mean that the outcome represents an unreasonable application of clearly established Federal law. Id.
So it is here. Although the Appeals Court’s reasoning on this issue seems perfunctory (a single paragraph with scant analysis), the outcome reached by the court appears to be both plausible and adequately supported. Phrased in the idiom of the AEDPA, we cannot say on this record that the outcome was an unreasonable application of the Jackson standard. Accordingly, we reverse the district court’s decision insofar as it purposes to grant the petitioner relief on this claim.
C. Sixth Amendment: Confrontation.
The petitioner argued below that the trial justice’s affirmed ruling allowing Logan’s redacted statement into evidence violated his Sixth Amendment confrontation rights. The district court concurred with the petitioner’s appraisal. Foxworth I, slip op. at 19. The respondent takes exception.
To begin, the panel has identified a pivotal issue concerning the date upon which the petitioner’s conviction became final; that is, the date when the process of direct review of the conviction in the state courts was exhausted. The district court, believing that the respondent had conceded the point, assumed that the conviction became final in 2002 (when the SJC denied the amended ALOFAR). See Foxworth I, slip op. at 4 n. 3. The parties have not explicitly addressed the date of finality in their briefs in this court (although the respondent consistently has argued that Gray should not be applied retroactively to this case). For our part, we have been unable either to locate an outright concession or to agree upon an answer to the finality question of when direct review ended. As we explain below, the issue turns out to be a dispositive one.4
*4301. The Finality Question. We set the stage. In federal habeas cases, there is a requirement, spelled out in 28 U.S.C. § 2254(d), that a state-court decision be measured against “clearly established Federal law, as determined by Supreme Court of the United States.” In most cases (and, specifically, in this case), the date of finality of the state court conviction determines the time line to be used for determining what Supreme Court decisions comprise the corpus of this “clearly established Federal law.”
Here, the jury convicted the petitioner on March 31, 1992; the petitioner’s first new trial motion was denied on August 8, 1994; and both the conviction and the denial of the first new trial motion were affirmed by the Appeals Court on October 21, 1996. The normal time for seeking direct review of that decision expired twenty days later. See Mass. R.App. P. 27.1(a). The petitioner took no action; and his conviction became final. At that time, the Supreme Court’s Confrontation Clause jurisprudence included Bruton and Richardson v. Marsh, 481 U.S. 200, 107 S.Ct. 1702, 95 L.Ed.2d 176 (1987).
The petitioner applied for leave to obtain further appellate review four years later (on October 25, 2000). In that temporal gap, the Supreme Court decided Gray v. Maryland, 523 U.S. 185, 118 S.Ct. 1151, 140 L.Ed.2d 294 (1998). That opinion refined and extended the Bruton rule.
The petitioner’s 2000 pro se ALOFAR did not cite Gray. The proceedings were stayed while petitioner filed and litigated a second new trial motion. On July 22, 2002, the petitioner submitted an amended ALO-FAR which cited Gray and argued that, under Griffith v. Kentucky, 479 U.S. 314, 328, 107 S.Ct. 708, 93 L.Ed.2d 649 (1987), he was entitled to its benefit. The amended ALOFAR was accompanied by a “motion to file an amended FAR application late,” which the SJC allowed. The SJC denied further appellate review without comment on September 6, 2002.
Given this chronology, there are two possible dates of finality: November 18, 1996 or September 6, 2002.5 We know that the SJC has the power to overlook the untimeliness of an ALOFAR and sometimes considers a late application on the merits. See, e.g., Commonwealth v. O’Neil, 436 Mass. 1007, 765 N.E.2d 767, 767 n. 1 (2002). But the question is whether the SJC’s actions here reopened the finality of the criminal conviction in 2002. See 2 R. Hertz & J.S. Liebman, Federal Habeas Corpus Practice & Procedure § 25.6 (5th ed.2001) (suggesting that a state court may choose to delay the point of finality of a criminal judgment and that, since the non-retroactivity doctrine serves the interests of the states, a federal court may be required to respect this decision on habeas review).
The answer to these questions will be conclusive here. After all, if the petitioner’s conviction became final in 2002, then Gray applies because it was decided beforehand. Otherwise, Richardson controls. As we explain below, the result of our analysis of the confrontation issue hinges on which of these precedents governs.
*4312. The Effect if 2002 is the Date of Finality. We turn next to the effect of a 2002 date of finality on the constitutional propriety of the trial justice’s evidentiary ruling. The respondent consistently has argued that the date for determining clearly established law was in 1996. But the respondent gave a particular reason, and that reason is wrong: the respondent posits that because the SJC summarily denied further appellate review, the contours of “clearly established Federal law” should be defined with reference to the date of the Appeals Court decision (which antedated Gray). The district court found this theory unconvincing, see Foxworth I, slip op. at 5, and so do we.
The respondent’s theory is based on an incorrect premise. To support it, the respondent cites Justice O’Connor’s locution that, under the AEDPA, the term “clearly established Federal law” means “the holdings, as opposed to the dicta, of this Court’s decisions as of the time of the relevant state-court decision.” Williams, 529 U.S. at 412, 120 S.Ct. 1495. The respondent reads the phrase “relevant state-court decision” as referring for all purposes to the last reasoned state-court decision.
This reading is untenable. Close perscrutation of Williams discloses nothing in the Court’s constituent opinions that indicates any intention on Justice O’Connor’s part either to modify or to undercut the bright-line rule of Teague v. Lane, 489 U.S. 288, 310, 109 S.Ct. 1060, 103 L.Ed.2d 334 (1989) (effectively limiting the consideration of new constitutional rules of criminal procedure in cases on collateral review to those rules announced before the petitioner’s conviction became final). The opposite is true. Justice O’Connor’s opinion stated that “whatever would qualify as an old rule under our Teague jurisprudence will constitute ‘clearly established Federal law, as determined by the Supreme Court of the United States’ under § 2254(d)(1).” 529 U.S. at 412, 120 S.Ct. 1495. That is a frank recognition that the AEDPA has neither altered nor eroded the marker laid down by Teague. This recognition is fully consistent with Part III of Justice Stevens’s lead opinion, joined by Justice O’Connor; there, Justice Stevens stated that “[t]he threshold question under AED-PA is whether [the petitioner] seeks to apply a rule of law that was clearly established at the time his state-court conviction became final.” Williams, 529 U.S. at 390, 120 S.Ct. 1495 (emphasis supplied).
Griffith removes any vestige of doubt. We have read that decision as establishing that “[i]f [a] conviction is not yet final when the Supreme Court announces [a] rule, then inferior courts must apply that rule to the defendant’s case.” Derman v. United States, 298 F.3d 34, 39 (1st Cir.2002) (citing Griffith, 479 U.S. at 322, 107 S.Ct. 708). By contrast, if the conviction is already final, then the defendant ordinarily—there are special circumstances, but none that are relevant here— may not avail himself of the newly announced rule. See Teague, 489 U.S. at 310, 109 S.Ct. 1060. It is, therefore, evident that finality, not the date of the last reasoned decision, is the principal determinant of whether a “new” rule can be applied to an “old” case.
Finality means that “a judgment of conviction has been rendered, the availability of appeal exhausted, and the time for a petition for certiorari [has] elapsed or a petition for certiorari [filed and] finally denied.” Griffith, 479 U.S. at 321 n. 6, 107 S.Ct. 708. Under this definition, if the petitioner’s conviction was final no earlier than September 6, 2002, Gray would control. See Currie v. Matesanz, 281 F.3d 261, 266 (1st Cir.2002).
To recapitulate, both Teague and Griffith remain good law. Those cases mean *432what they say. Together, they lead to the inexorable conclusion that if the petitioner’s conviction was not “final” until after Gray was decided by the Supreme Court, the petitioner is entitled to the benefit of Gray.
We add a coda. While this result is dictated by precedent, it also makes good sense from a policy standpoint. The construct advocated by the respondent would allow a state court to subvert Griffith and deny criminal defendants the benefit of new Supreme Court precedent by the simple expedient of summarily affirming a lower court’s decision. That would give state courts a perverse incentive to avoid addressing constitutional claims in contemporaneous terms while insulating their actions from subsequent federal habeas review. That would be an unattractive prospect.
The respondent has yet another string to his bow. He argues that even if Gray applies, the trial justice’s ruling should be upheld. Initially, that argument requires us to consider the standard of review applicable to the state court’s rejection of the Sixth Amendment claim.
On this point, the respondent urges us to use AEDPA’s customary “unreasonable application” standard. 28 U.S.C. § 2254(d)(1). The petitioner resists. He notes that this standard applies only to a “claim that was adjudicated on the merits in State court proceedings.” Id. § 2254(d). Because Gray was never addressed by the state courts, he proposes de novo review. See Fortini v. Murphy, 257 F.3d 39, 47 (1st Cir.2001) (stating that a federal court “can hardly defer to the state court on an issue that the state court did not address”); cf. Brown v. Maloney, 267 F.3d 36, 40 (1st Cir.2001) (declaring that absent “reasoning on a holding from the state court on [an] issue,” it cannot be said that the claim was adjudicated on the merits).
The precise question raised by the dueling arguments is whether a state-court decision disposing of a federal constitutional claim on the merits without a reasoned application of the most pertinent Supreme Court authority is an adjudication that comes within the AEDPA’s purview. This is a question different from that which we confronted in Fortini. The Appeals Court, citing Bruton, squarely decided the Sixth Amendment claim and elaborated its reasoning. But that decision did not take account of Gray; indeed, it could not have done so; the Supreme Court decided Gray after the Appeals Court issued its decision. In these uncharted waters, ascertainment of the appropriate level of review presents a vexing question.
Deferring the answers to difficult procedural questions until the law develops is often the path of prudence. We have the luxury of taking that path here because nothing turns on the standard of review. Even were we to accept the respondent’s argument and evaluate the state court’s resolution of the Sixth Amendment claim through the AEDPA’s more deferential prism, we would conclude that the petitioner is entitled to relief. Thus, we assume for ease in administration—but do not decide—that the AEDPA’s “unreasonable application” standard pertains to the state courts’ resolution of the Sixth Amendment claim.
This brings us to the merits of the respondent’s contention that even if the petitioner’s conviction did not become final until 2002 and Gray therefore applies, the trial justice’s ruling was neither contrary to nor an unreasonable application of Gray. This argument requires us to exercise the trilogy of Supreme Court cases that collectively constitute the controlling precedents for the petitioner’s Confrontation Clause claim. This trilogy includes *433Bruton, Richardson, and Gray. We start with a brief synopsis of each decision.
In Bruton, the Supreme Court proscribed the introduction of a nontestifying defendant’s extrajudicial statements that are “powerfully incriminating” as to a jointly-tried codefendant. 391 U.S. at 135, 88 S.Ct. 1620. The Court made pellucid that the vice inherent in such a tactic cannot reliably be abated through the use of limiting instructions. Id. (explaining that “the risk that the jury will not, or cannot, follow [such] instructions is so great, and the consequences of failure so vital to the defendant, that the practical and human limitations of the jury system cannot be ignored”).
In Richardson, the Court refined the Bruton rule. It held that “the Confrontation Clause is not violated by the admission of a nontestifying codefendant’s confession with a proper limiting instruction when ... the confession is redacted to eliminate not only the [objecting] defendant’s name, but any reference to his or her existence.” 481 U.S. at 211, 107 S.Ct. 1702. Under this refinement, “[statements that are incriminating only when linked to other evidence in the case do not trigger application of Bruton’s preclusionary rule.” United States v. Vega Molina, 407 F.3d 511, 520 (1st Cir.2005).
Richardson left open the question of whether Bruton’s prophylaxis extended to a statement in which the objecting defendant’s name has been replaced with a symbol or neutral pronoun. Gray answered that question, holding that “redactions that replace a proper name with an obvious blank, the word ‘delete,’ a symbol, or similarly notify the jury that a name has been deleted are similar enough to Bruton’s unredacted confessions as to warrant the same legal results.” 523 U.S. at 195, 118 S.Ct. 1151. Under this regime, an inquiring court must judge the efficacy of redaction on a case-by-case basis, paying careful attention to both a statement’s text and the context in which it is offered. Vega Molina, 407 F.3d at 520.
In this corner of the law, one size does not fit all. One case may involve numerous actors and events, such that no compelling inference can be drawn that a symbol or neutral pronoun refers to a specific defendant. A different case may involve few actors and events, such that a symbol or neutral pronoun becomes transparent and leaves little doubt in the listener’s mind about the identity of the person whose name has been removed.
This brings us to the instant case. In affirming the trial justice’s admission of Logan’s statement, the Appeals Court wrote:
Foxworth’s reliance [on Bruton ] is misplaced, however, because (a) the challenged statement as redacted did not name or necessarily implicate Fox-worth; (b) it was not “powerfully” inculpatory, since Logan concededly had left the premises before and did not personally observe the shooting; (c) the Commonwealth’s case rested primarily on the eyewitness identification of Foxworth as the shooter; (d) the prosecutor did not use the statement as substantive evidence against Foxworth but rather against Logan and in support of a joint venture theory; and (e) the judge provided an appropriate instruction to obviate the Bruton concerns.
Evaluating the Appeals Court’s ruling against Bruton and Richardson, we hold that it was not an unreasonable application of clearly established Federal law. See infra Part 11(C)(3). Thus, the petitioner’s *434claim depends wholly on the refinement and extension of the doctrine that Gray heralded.
Here (as in the district court), the petitioner challenges the Appeals Court’s recitation on several levels, primarily arguing that the redacted statement left no doubt as to his identity and, in the bargain, powerfully incriminated him. He succeeded in convincing the district court that the Appeals Court’s rejection of this challenge constituted an unreasonable application of Gray. See Foxworth I, slip op. at 19.
The respondent assigns error, maintaining that Gray and this case are horses of a different hue. In a sense, that is so. In Gray, the redaction—a blank space—was clear on the face of the statement. Here, a clumsy attempt was made to redact Logan’s statement in a way that might serve to conceal from the jury the fact that a redaction had been made.
Still, this difference seems to be of little moment. As said, this is an “unreasonable application” case; thus, the mere existence of a slight factual distinction does not signify that the statement in question necessarily spins outside of Gray’s precedential orbit. What matters is that the situation at hand fits comfortably within the parameters of Gray’s doctrinal teachings.
We have carefully examined both the text of the redacted statement and the context in which it was admitted. We conclude that its admission was an unreasonable application of Gray. Our reasons follow.
First, the redaction is obvious. In and of itself, the use of the designation “Mr. X” would have caused any rational juror to raise an eyebrow. See United States v. Peterson, 140 F.3d 819, 822 (9th Cir.1998) (finding Bruton error where “person X” was substituted for defendant’s name); see also Malinski v. New York, 324 U.S. 401, 430, 65 S.Ct. 781, 89 L.Ed. 1029 (1945) (Rutledge, J., dissenting) (describing substitution of a name in a confession with “X” or “Y” as “devices ... so obvious as perhaps to emphasize the identity of those they purported to conceal”).
Here, however, we need not muse about raised eyebrows and other such subtleties. Detective Flynn drove this point home, testifying that “Mr. X” was a substitute for the name of the person actually identified by Logan. Furthermore, the prosecutor referred to Logan’s statement as “the redacted statement.” To cinch matters, the copy given to the jury had the phrase “Mr. X” in a different font (which did not exactly fit the spaces left by the excision of the petitioner’s name). It follows inexorably from these facts that the use of the pseudonym “Mr. X” represented an “obvious deletion” that “encourag[ed] the jury to speculate about the reference.” Gray, 523 U.S. at 193,118 S.Ct. 1151.
With knowledge of the redaction, it was child’s play for the jury to identify the petitioner as “Mr. X.” There were only three defendants—-Logan, Christian, and the petitioner—and no other suspects. Logan, as the author of the statement, could not have been “Mr. X.” Flynn’s testimony unambiguously eliminated Christian from consideration. That left the petitioner.
Of course, the fact that it was obvious that the petitioner was “Mr. X” does not necessarily render Logan’s redacted statement powerfully incriminating. In this regard, the respondent points out that the statement did not designate “Mr. X” as the shooter. That disclaimer is true as far as it goes, but it does not take the respondent very far.
To recapitulate, the redacted statement revealed that “Mr. X” had gone to McLean’s apartment because McLean had sold him some bad cocaine; once there, “Mr. X” argued with McLean; then, after “Mr. X” had departed, Logan saw him *435reentering the building with a gun tucked into his waistband; and shots were heard shortly thereafter. Given the chain of events chronicled in Logan’s redacted statement, the inference that the petitioner shot McLean was virtually inescapable. As a practical matter, the statement directly implicated the petitioner and powerfully incriminated him.6 Under Gray, the redacted statement should have been excluded.
That conclusion does not end our odyssey. Even though the admission of the redacted statement violated Gray, the AEDPA standard requires more than a showing of trial error. See McCambridge, 303 F.3d at 36. At a bare minimum, the error must be sufficiently egregious to comprise an unreasonable application of clearly established federal constitutional principles. 28 U.S.C. § 2254(d)(1). That standard is satisfied here.
We need not tarry. The range of reasonable outcomes here is small because the applicable constitutional rule-explained in Gray—is narrow, see United States v. Thayer, 204 F.3d 1352, 1355 (11th Cir.2000) (describing Bruton rule as “narrow”). The outcome reached by the Appeals Court (albeit understandable because Gray had not yet been decided) is patently offensive to that rule. There is simply no way to defend it: the inferences necessary to conclude that the petitioner was “Mr. X” and that “Mr. X” was the shooter are nose-on-the-face plain, and the inculpatory thrust of the statement is potent. We hold, therefore, that if the state conviction was not final until 2002, Gray applies; and, applying Gray to this case, it would run afoul of clearly established Federal law to find no violation of Gray on these facts.
As a rear-guard action, the respondent suggests that any Sixth Amendment violation was harmless. This suggestion lacks merit.
Once again, we pause to clarify the applicable standard of review. Although the Appeals Court did not find a Bruton error, it nevertheless hedged its bets. It added that “[e]ven if there were error, [in connection with the admission of Logan’s statement], it was harmless, because the evidence related to a noncrucial issue ... and was at most merely cumulative of evidence properly before the jury.” The court cited Commonwealth v. Sinnott, 399 Mass. 863, 507 N.E.2d 699 (1987), which holds that a Bruton error does not require reversal if it was “harmless beyond a reasonable doubt.” Id. at 705. In other words, it applied the federal standard for harmless constitutional error applicable on direct review. See Chapman v. California, 386 U.S. 18, 24, 87 S.Ct. 824, 17 L.Ed.2d 705 (1967).
That determination of harmlessness does not engender AEDPA deference. The Supreme Court recently instructed federal habeas courts to perform a straightforward harmless error analysis under Brecht, 507 U.S. at 638, 113 S.Ct. 1710, rather than review a state court’s harmless-beyond-a-reasonable-doubt determination for unreasonableness. Fry v. Pliler, 551 U.S. 112, 127 S.Ct. 2321, 2327, 168 L.Ed.2d 16 (2007); see Delaney, 522 F.3d at 105. This is so because the more forgiving Brecht test subsumes the hybrid *436Chapman/AEDPA test. Fry, 127 S.Ct. at 2327. We proceed accordingly.
On collateral review of trial error, the test for harmless error is whether the error had a “substantial and injurious effect or influence in determining the jury’s verdict.” Sinnott v. Duval, 139 F.3d 12, 14 (1st Cir.1998) (quoting Brecht, 507 U.S. at 637, 113 S.Ct. 1710). The burden of establishing harmlessness rests with the state qua respondent. O’Neal v. McAninch, 513 U.S. 432, 437, 115 S.Ct. 992, 130 L.Ed.2d 947 (1995). If the habeas court entertains “grave doubt as to harmlessness, the petitioner must win.” Id.
The identification of the shooter was the pivotal issue in this case. The prosecution’s case in chief, though constitutionally sufficient, was painfully thin. See supra Part 11(B). Other than Logan’s statement, the only evidence bearing on identification was Hobson’s testimony. That testimony was scarcely unassailable. Logan’s statement bolstered Hobson’s account by placing the petitioner at the murder scene, armed and with a motive, immediately before the shooting. It also supported Hobson’s version of the events leading up to the slaying as it suggested that McLean was chased by one man rather than two.
At the trial-court level, this case was close. In close cases, there is often a tipping point. We think that it is probable that Logan’s statement tipped the balance. The statement lent significant weight to the word of the lone witness who identified the petitioner as the shooter and, as the Bruton Court recognized, 391 U.S. at 135, 88 S.Ct. 1620, the trial justice’s limiting instructions were not an effective safeguard against the jury’s exposure to such powerfully incriminating evidence. Indeed, the impact of the statement is adequately evinced by the fact that the jury convicted the petitioner while acquitting Logan.
In sum, we agree with the district court, Foxworbh I, slip op. at 23, that the admission of the redacted statement likely had a substantial and injurious influence on the outcome of the proceedings. At the very least, there is grave doubt. Accordingly, the petitioner is entitled to a new trial if Gray applies.
3. The Effect if 1996 is the Date of Finality. If the petitioner’s conviction became final in 1996 rather than final in 2002, the case plays out differently. Because the Supreme Court had not yet decided Gray,7 the relevant “clearly established Federal law” would be capped by Richardson. Under that precedent-and without the benefit of Gray—the Appeals Court’s decision cannot be said to be either contrary to or an unreasonable application of clearly established Supreme Court precedent. After all, the Richardson Court specifically left open the question, implicated in this case, of whether a Confrontation Clause problem exists when “the defendant’s name has been replaced by a symbol or neutral pronoun.” 481 U.S. at 211 n. 5, 107 S.Ct. 1702. The Court’s acknowledgment that an issue remains fairly debatable precludes any credible argument that a state court’s decision on the issue is contrary to or an unreasonable application of clearly established Supreme Court precedent.8 See L’Abbe v. DiPaolo, 311 F.3d 93, 98 (1st Cir.2002).
*4374. The Decision to Certify. It thus appears that the date of finality, which will determine whether Gray or Richardson is the controlling precedent, will be dispositive of this aspect of the respondent’s appeal. The SJC has offered no real guidance on that issue, and the correct answer remains shrouded in uncertainty. Moreover, the case is obviously important: a murder conviction hangs in the balance. Consequently, we deem it appropriate to certify the question concerning the date of finality to the SJC. See, e.cj., In re Engage, Inc., 544 F.3d 50, 57 n. 10 (1st Cir.2008) (approving sua sponte certification in appropriate cases); Brown v. Crown Equip. Corp., 501 F.3d 75, 77 (1st Cir.2007) (similar).
This case fits the model for certification. We may certify doubtful questions to a state supreme court in cases, such as this, where we find no controlling precedent, where the answer is unclear, and where the answer will be determinative of an issue in the litigation. See Boston Gas Co. v. Century Indem. Co., 529 F.3d 8, 15 (1st Cir.2008); Nieves v. Univ. of P.R., 7 F.3d 270, 274 (1st Cir.1993). And, moreover, the SJC has indicated a willingness, under such circumstances, to answer certified questions. See Mass. S.J.C. R. 1:03.
III. CONCLUSION
We need go no further. For the reasons elucidated above, we reverse the district court’s ukase holding the evidence of the petitioner’s guilt insufficient. At the same time, however, we withhold decision on the district court’s ruling that the petitioner is entitled to relief on the basis of a Sixth Amendment violation. The outcome of that question turns on whether the petitioner’s conviction became final in 1996 or 2002. We certify that question to the SJC as per the order filed in conjunction herewith. Pending that court’s response, we retain appellate jurisdiction.
So Ordered.
ORDER FOR CERTIFICATION OF A QUESTION TO THE SUPREME JUDICIAL COURT OF MASSACHUSETTS
June 29, 2009.
For the reasons stated in our opinion of even date in this case, submitted with this order, we certify the following question to the Supreme Judicial Court of Massachusetts:
In considering the petitioner’s application, filed in 2000, for leave to obtain further appellate review of the 1996 Massachusetts Appeals Court decision, did the Supreme Judicial Court reopen the finality of the petitioner’s 1992 conviction in 2002?
This court certifies that this question is determinative of a claim in this case and that it appears to us that there is no controlling precedent in either the decisions or rules of practice of the Supreme Judicial Court.
In responding to this question, we invite any additional guidance about relevant Massachusetts law or practice that the Supreme Judicial Court may wish to offer.
The clerk of this court is to forward, under the official seal of this court, a copy of the certified question and our opinion in this case, along with the briefs and trial transcripts filed by the parties, to the Supreme Judicial Court of Massachusetts. Pending the receipt of a response, we shall retain appellate jurisdiction.
. Although Logan’s statement also implicated the petitioner, see infra Part 11(C)(2), the statement was admitted solely against Logan.
. This paradigm makes good sense: it is designed to facilitate the introduction of eyewitness identifications made "when memory was fresher and there had been less opportunity for influence to be exerted upon [the witness].” United States v. Lewis, 565 F.2d 1248, 1251 (2d Cir.1977) (quoting United States v. Marchand, 564 F.2d 983, 996 (2d Cir.1977)).
. The trial justice instructed the jurors that the identity of the shooter was of paramount importance in this case and that it was up to them to decide the accuracy of Hobson’s identification.
. There is some room for doubt as to whether the respondent has waived or forfeited the date-of-finality issue by failing to contest the district court's unfounded statement, by failing to raise the issue squarely in his appellate briefs, and/or by his counsel’s statements at oral argument in this court. The panel is divided on this point: Chief Judge Lynch and Judge Siler find neither waiver nor forfeiture; *430Judge Selya would hold the point waived or, at least, forfeited. Since a majority of the panel believes that the question of finality is properly before us, no useful purpose would be served by rehearsing the arguments for and against waiver.
. In fact, the latter date may well be extended by ninety days to encompass the period in which a petition for certiorari to the Supreme Court of the United States could be filed. See, e.g., Griffith, 479 U.S. at 321 n. 6, 107 S.Ct. 708. Because this ninety-day shift makes no difference here, we use September 6, 2002 as a convenient shorthand.
. The Appeals Court made much of the trial justice’s instructions limiting the jury's use of the statement to the case against Logan. Where a nontestifying codefendant’s statement is powerfully incriminating, however, a limiting instruction is cold comfort. See Bruton, 391 U.S. at 135, 88 S.Ct. 1620. That is especially so in this case since Logan was charged with murder under a joint-venture theory, and the petitioner was a likely candidate for membership in any such joint venture.
. We note that the Court has not directed that Gray should be applied retroactively. See Teague, 489 U.S. at 310, 109 S.Ct. 1060; see also Garcia v. United States, 278 F.3d 1210, 1216 (11th Cir.2002) (refusing to apply Gray retroactively).
. The closeness of the question left open in Richardson is emphasized by the fact that in the subsequent decision in Gray four justices dissented. See Gray, 523 U.S. at 200, 118 S.Ct. 1151 (Scalia, J., dissenting).