OPINION OF THE COURT
GREENBERG, Circuit Judge.
I. INTRODUCTION
This matter comes on before this Court on Antonio Vazquez’s appeal from a final order of the District Court denying his petition for a writ of habeas corpus following Pennsylvania state court proceedings. In July 2000 the Common Pleas Court tried Vazquez jointly with Gilbert Santiago on first-degree murder and certain other charges. The jury convicted Vazquez on all of the charges against him, following which it sentenced him to life in prison.
The jury, however, found Santiago not guilty. Vazquez appealed but the Pennsylvania Superior Court affirmed his conviction and sentence in an unpublished opinion on February 22, 2002, that was the only appellate state court opinion in this case dealing with the issues that we consider on this appeal. Vazquez subsequently unsuccessfully sought relief in the Supreme Court of Pennsylvania and the Supreme Court of the United States.
On May 14, 2003, Vazquez filed a petition in the Common Pleas Court for post-conviction relief under Pennsylvania’s Post Conviction Relief Act, 42 Pa. Cons.Stat. Ann. § 9541
et seq.
(West 1998), but that court denied the petition on July 14, 2004. Vazquez appealed, but the Pennsylvania Superior Court affirmed, and the Supreme Court of Pennsylvania denied review on December 29, 2005.
On June 19, 2006, Vazquez filed a petition for a writ of habeas corpus in the District Court under 28 U.S.C. § 2254. After that Court denied the petition
Vazquez appealed to this Court and sought a certifícate of appealability, which we granted on October 11, 2007. As we will explain, the outcome of this case turns on the application of
Bruton v. United States,
391 U.S. 123, 88 S.Ct. 1620, 20 L.Ed.2d 476 (1968), and subsequent Supreme Court cases building on
Bruton.
II. FACTS
A. The Shooting and its Aftermath
At about 3:00 a.m. on January 31, 1999, Melvin Coleman, the murder victim, hired Matthew Caldwell, an unlicensed taxicab driver, to drive him to the corner of Third Street and Allegheny Avenue in Philadelphia. There, Coleman spoke briefly with three men, Vazquez, Santiago, and George Rivera, who were in a gray Buick LeSa-bre.
After the three men in the Buick departed, Coleman asked Caldwell to drive him to a different location in Philadelphia. On the way to that location Coleman and Caldwell saw the gray Buick parked near a payphone. Coleman rolled down his window and asked the three men if there was any “hydro around,” to which one of the men responded “in about five minutes.” App. at 188. Caldwell and Coleman then continued driving. When they stopped at a traffic light a few blocks later, the gray Buick approached the taxi from behind whereupon one or more of its occupants began shooting at the taxi shattering its rear window. As Caldwell pulled his vehicle around the corner, he heard another shot, following which Coleman told him that he had been hit. After Caldwell heard two more shots, he drove Coleman to Temple University Hospital where he died of a single gunshot wound to the upper back.
A few minutes after the shooting two Philadelphia police officers on routine patrol who were unaware of the shooting spotted the gray Buick making an abrupt right turn onto Sixth Street. The officers were concerned with the Buick’s operation and consequently followed it. Then, when the officers attempted to initiate a traffic stop, the driver of the Buick, Santiago, ran a red light and its occupants fled. During the ensuing pursuit, one of the Buick’s occupants, who Vazquez later acknowledged had been he, threw a gun out of a window of the car. At the trial there was evidence supporting a finding that the gun, which the police recovered, was the murder weapon. After the vehicle covered a few additional blocks Vazquez jumped from it on the passenger side and rolled along the ground. Following a brief stop during which one officer took a good look at Vazquez, who escaped and avoided apprehension on the night of the murder, the police continued to pursue the Buick, but they lost track of it after a few more turns.
About 20 minutes later different officers in another patrol car spotted the Buick in which Santiago now was the sole occupant. When Santiago saw the police he fled, first in the vehicle and then on foot. Unlike Vazquez, however, he did not escape as the
police apprehended him shortly into his flight.
Following his arrest, Santiago gave a statement to Philadelphia Detective Will Egenlauf in which he admitted that he had been the driver of the Buick at the time of the shooting and identified Vazquez and Rivera as its other occupants at that time. Santiago said that Vazquez was the shooter and that he and Rivera were surprised when Vazquez opened fire. Santiago explained that he fled from the police solely out of fear and agreed to help them identify and apprehend Vazquez and Rivera.
The appellees do not accept Santiago’s statement to Egenlauf as having been completely accurate as they recite in their brief that “[t]he ballistics evidence ... indicated that two different guns were fired at the cab [and][t]he Commonwealth argued that there was never an ‘innocent’ passenger in the car, but instead that both Rivera and Vazquez were shooters.” Ap-pellees’ br. at 7. Thus appellees assert that the prosecutor “effectively undermin[ed] both Santiago’s statement and Vazquez’s testimony.”
Id.
Nevertheless, the prosecutor partially accepted Santiago’s statement because she argued that Vazquez fired the fatal shot.
Notwithstanding Santiago’s identification and offer to help in Vazquez’s apprehension, the police did not arrest Vazquez for several months until they found him at his wife’s house asleep next to a police scanner. For reasons that the parties do not explain in their briefs or suggest are explained in the record at a place to which they direct our attention, Rivera, who was not a defendant at the trial, was not present at it. Indeed, when we study the parties’ briefs we almost sense that they do not want us to know why Rivera was not at the trial, for appellees cryptically tell us only that “George Rivera, who was also in the car at the time of the shooting, was not brought to trial,” Appellee’s br. at 3, and Vazquez tells us only that “Mr. Rivera was unavailable for trial.” Appellant’s br. at 11.
B. The Trial
A grand jury charged Vazquez and Santiago with first-degree murder, aggravated assault, two firearms-related charges, and conspiracy to commit the offenses, and they were the two defendants in the Common Pleas Court at the trial in the proceedings which we are now examining. Not surprisingly, the identity of the person who fired the fatal shot was the critical issue at the trial inasmuch as Caldwell did not identify the shooter and no one present at the trial except for Vazquez, who denied being a shooter, and Santiago, who did not testify, could have seen any of the shots fired.
The prosecutor, relying primarily on the fact that the murder weapon had three fingerprints, two of which were too smudged to identify but one of which on its barrel matched Vazquez’s left ring finger, contended that Vazquez fired the fatal shot. This fingerprint testimony, however, though supporting a conclusion that Vazquez fired the murder weapon, hardly was conclusive because Vazquez testified that Rivera fired the weapon and then passed it to him in the back seat, telling him to get rid of it, which he did. Accordingly, Vazquez’s testimony could explain why his fingerprint was on the weapon even if he had not fired it.
Though Santiago did not testify, the statement that he
gave Detective Egenlauf which contradicted Vazquez’s testimony on the critical question of who was the shooter spoke for him. Thus, the jury if it did not credit Santiago’s statement — though we can see no reason why it would have discredited the statement to the extent that Santiago said that he was not a shooter, unless it believed Vazquez’s testimony that he, Vazquez, was not a shooter-could have concluded that Santiago, Vazquez, or Rivera fired the fatal shot. But if, as its verdict demonstrated apparently happened, the jury accepted Santiago’s statement, at least to the extent that it believed that he did not fire the fatal shot, then its choice for the shooter was between Vazquez and Rivera.
Inasmuch as the parties realized that there would not be any direct evidence other than Santiago’s statement identifying Vazquez as a shooter, they understood before the trial that the question of how, if at all, the prosecutor could use the statement at the trial would be of the utmost importance. Santiago believed that it was vital to his defense to use an unredacted version of the statement because it showed that when the police arrested him he immediately fully cooperated by identifying Rivera and Vazquez as occupants of the vehicle and by offering to help the police apprehend them. On the other hand, Vazquez believed that even a redacted form of the statement should not be used at the trial because it plainly would identify him as the shooter in violation of his Sixth Amendment right to cross-examine Santiago who did not indicate that he would testify and, in fact, did not do so.
See Bruton v. United States,
391 U.S. at 135-36, 88 S.Ct. at 1627-28. Consequently, both defendants moved for severance of their trials, but the court denied their motions.
The pretrial rulings did not put the question of how the statement could be used to rest as it arose again at trial when the prosecutor sought to use it. At that time the court, over Vazquez’s strenuous objection, ruled that Detective Egenlauf could read a redacted version of the statement to the jury, substituting “my boy” or “the other guy” for the names of Vazquez or Rivera. Nevertheless, both before and after Egenlauf read the statement, which contained more than 20 substitutions of a generic term for Vazquez’s and Rivera’s names, the court permitted Santiago’s attorney to emphasize that Santiago had identified the two other men in the car and had offered to take the police to their homes. Accordingly, the jury almost certainly knew from the evidence at the trial that the someone had redacted Santiago’s statement so as to excise the names of the persons Santiago had identified. We think that this point is clear beyond doubt as we can perceive of no reason why the jury would have believed that Santiago had identified his two passengers but nevertheless had used the generic terms in his statement used at trial. Furthermore, as we explain below, during its deliberations the jury asked the court a question that included the jury’s conclusion that Santiago’s statement identified Vazquez as the shooter.
Santiago called his wife, Nancy Rosado, as a character witness at the trial. During her cross-examination, notwithstanding her lack of personal knowledge of the details of the crime, in response to a question of the prosecutor, Rosado stated that Vazquez was “[t]he guy that murdered” Coleman. App. at 533. Her expression of this view of the ease incriminating Vazquez understandably caused him to move for a mistrial, but the court denied the motion, and, instead, instructed the jury to disregard Rosado’s comments.
At the closing argument, after Santiago’s counsel again emphasized that his client had identified the other occupants of the Buick, the prosecutor effectively eliminated the redaction of Vazquez’s name and reinserted it in Santiago’s statement when she referred to “Mr. Santiago’s statement that he and the other man George, excuse me, the man who’s not the shooter, he said, [had] jumped out of the car.” App. at 714. Obviously, the prosecutor’s statement identified Vazquez as the shooter because in his statement Santiago .had claimed that neither he nor “George,” a name that could mean only Rivera, had not been the shooter. Thus, Santiago’s statement assigned that unwanted role to Vazquez as he was the only person left to fill it. As might be expected, the prosecutor’s comment led Vazquez to move again for a mistrial, which the court denied, even though the prosecutor admitted that “I clearly misspoke ... and I did say it, without question.” App. at 718.
Following closing arguments, the court instructed the jury that it should not consider Santiago’s statement as evidence against Vazquez. Clearly, however, this instruction was not completely effective, if effective at all, because during its deliberations the jury requested that the court have Santiago’s statement read back and asked, “Are we supposed to not consider Santiago’s statement that Vazquez was the shooter?” App. at 805. After considering whether or not to declare a mistrial in recognition of the reality that the jury had concluded that Santiago’s statement had been redacted but originally had named Vazquez as the shooter, the court read the statement to the jury in its redacted form and repeated its instruction that it was not to consider Santiago’s statement as evidence against Vazquez.
The jury convicted Vazquez of first-degree murder, aggravated assault, and two firearms-related charges but acquitted Santiago on all counts.
III. PROCEDURAL HISTORY, JURISDICTION, and STANDARD OF REVIEW
As we previously explained, following his conviction Vazquez filed unsuccessful direct appellate and post-conviction relief proceedings in the trial court, the Pennsylvania Superior and Supreme Courts, and the Supreme Court of the United States. Thereafter, he filed the habeas corpus petition in the District Court, the denial of which we now review. The District Court referred the case to a magistrate judge who, without holding a hearing, made a Report and Recommendation dated January 30, 2007, recommending that the Court deny the petition. The District Court adopted the Report and Recommendation, and on March 20, 2007, executed an order that was filed March 21, 2007, denying the petition.
After Vazquez filed a timely appeal to this Court he sought a certificate of ap-pealability, which we granted on the following issues:
(1) whether the District Court erred in denying Appellant’s claim that the trial court violated his right to a fair trial by admitting the statement of his non-testifying co-defendant,
see Bruton v. United States,
391 U.S. 123 [88 S.Ct. 1620, 20 L.Ed.2d 476] (1968); (2) whether the District Court erred in denying Appellant’s claim that the trial court violated his right to a fair trial by denying his motion to sever; (3)(a) whether Appellant exhausted his prosecutorial misconduct claim as a federal constitutional claim before the state courts pursuant to 28 U.S.C. § 2254(b)(1)(A); and (3)(b) whether the District Court erred in denying Appellant’s prosecutorial misconduct claim on the merits.[
]
Inasmuch as we are granting Vazquez relief on his
Bruton
contention, we need not decide the other issues that we set forth in our certificate of appealability.
The District Court had jurisdiction under 28 U.S.C. § 2254, and we have jurisdiction under 28 U.S.C. §§ 1291 and 2253. Insofar as we review the order of the District Court our review is plenary.
See Lambert v. Blackwell,
387 F.3d 210, 231 (3d Cir.2004). But this case arises under the Antiterrorism and Effective Death Penalty Act of 1996 (“AEDPA”), 28 U.S.C. § 2254, and thus our standard of review is not simply plenary. Under the AEDPA we must review the state court proceedings and affirm the denial of the petition unless we are satisfied that Vazquez has demonstrated that the Pennsylvania Superior Court, the highest-level state court to review the admission into evidence of Santiago’s statement on the merits, made a determination that “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1).
We are aware that in considering whether Vazquez met the AEDPA’s standards for relief under 28 U.S.C. § 2254(e)(1), we must presume that the state court’s factual determinations were correct. This rule of deference, however, is not significant here because the outcome of the appeal does not depend on state court factual determinations. Vazquez contends in accordance with the AEDPA’s dual bases for relief that the admission of Santiago’s statement was both contrary to and an unreasonable application of clearly established federal law. As we will explain, we agree that he is entitled to relief on the unreasonable application basis, though we doubt that the Superior Court’s decision was contrary to “clearly established Federal law, as determined by the Supreme Court of the United States.”
IV. DISCUSSION
We start our analysis of the Superior Court’s determination that the Common Pleas Court properly admitted Santiago’s redacted statement by first deciding “what constitutes ‘clearly established Federal law, as determined by the Supreme Court of the United State.’ ”
Lockyer v. Andrade,
538 U.S. 63, 71, 123 S.Ct. 1166, 1172, 155 L.Ed.2d 144 (2003). Moreover, in making our analysis of whether the Superior Court’s decision survives an inquiry under 28 U.S.C. § 2254(d)(1) so that Vazquez is not entitled to federal habeas corpus relief, we give its decision the benefit of any doubt,
Woodford v. Visciotti,
537 U.S. 19, 24, 123 S.Ct. 357, 360, 154 L.Ed.2d 279 (2002), even though the Superior Court relied on three Pennsylvania Supreme and Superior Court cases in reaching its result:
Commonwealth v. Rivera,
565 Pa. 289, 773 A.2d 131 (2001);
Commonwealth v. Travers,
564 Pa. 362, 768 A.2d 845 (2001); and
Commonwealth v. McGlone,
716 A.2d 1280 (Pa.Super.Ct.1998).
See Early v. Packer,
537 U.S. 3, 11, 123 S.Ct. 362, 366, 154 L.Ed.2d 263 (2002) (per curiam). Then, if we conclude that the Superior Court’s decision was not contrary to “clearly established Federal law as determined by the Supreme Court of the United States,” we must determine if it was an unreasonable application of that law.
We also point out that there is precedent supporting a rule that a reviewing court in later proceedings, when considering whether a trial court erred in overruling a
Bruton
objection, should review the state court proceedings from the perspective of the state trial court on the record before the trial court when it overruled an objecting defendant’s contention that a co-defendant’s statement should have been excluded even in a redacted form. This point seems evident because if a trial court errs with respect to a statement’s admission it does so when it makes its rulings. Thus, it is difficult to understand how the events following the admission of Santiago’s statement,
i.e.,
Rosado’s testimony identifying Vazquez as the shooter and the prosecutor’s statement to the jury in her closing argument that Rivera was not the shooter and, thus, by unmistakable inference, that Vazquez was the shooter, no matter how prejudicial to Vazquez, could have any bearing on the question of whether the trial court previously had erred in admitting the statement or allowing its use in a redacted form.
See United States v. Sandini,
888 F.2d 300, 305-06 (3d Cir. 1989) (distinguishing the question whether a court abused its discretion in denying a motion for a severance and whether if the court erred its ruling was prejudicial).
As it happens, 19 years ago we set forth, though in a dictum, the importance of viewing a
Bruton
determination on the record before the trial court at the time when it made its determination. In
United States v. Sandini
we entertained a direct appeal from a conviction and sentence in a district court in which the appellant argued that the court erred in denying his pretrial motion for a severance.
Id.
at 304-05. In considering the severance contention we explained that:
It is important to recognize that there are two separate determinations to be made when a defendant on appeal urges that he is entitled to a reversal because the district court denied a pretrial severance motion. Since the district court acted on the basis of the record before it at the time of the motion, we must first determine from that record whether the
court abused its discretion in denying the severance. Then, if there was an abuse of discretion, we must consider whether the defendant was prejudiced by the order denying severance.
Id.
at 305. In support of the foregoing explanation we indicated by way of analogy that:
The mere fact that the court errs in not granting severance does not mean that a defendant is prejudiced by the ruling. For example, the court might err in refusing to grant a severance as required by
Bruton
[citation omitted], but if the government then does not offer the codefendant’s confession, there would be no prejudice.
Id.
at 305 n. 2 (citation partially omitted). The inverse also is true. If a court does not err in denying an exclusion motion under
Bruton,
then subsequent events should not render its ruling retroactively erroneous.
Yet we pause in considering how events after the
Bruton
rulings should affect our analysis because we recognize that
United States v. Hardwick,
544 F.3d 565 (3d Cir. 2008), may be germane in our consideration of this question. In
Hardwick
we discussed
Bruton
and four other cases that we address below,
Richardson v. Marsh,
481 U.S. 200, 107 S.Ct. 1702, 95 L.Ed.2d 176 (1987);
Gray v. Maryland,
523 U.S. 185, 118 S.Ct. 1151, 140 L.Ed.2d 294 (1998);
United States v. Richards,
241 F.3d 335 (3d Cir.2001); and
Priester v. Vaughn,
382 F.3d 394 (3d Cir.2004). We then indicated that “[w]hat these decisions underscore is that the nature of the linkage between the redacted statement and the other evidence in the record is vitally important in determining whether a defendant’s Confrontation Clause right has been violated.” 544 F.3d at 573.
Yet in
Richardson
the Supreme Court indicated that it had granted certiorari because of a conflict between the decision of the Court of Appeals for the Sixth Circuit in that case and other courts of appeals’ decisions regarding the relationship of a challenged statement and other testimony.
Richardson
said that the court of appeals in the case it was reviewing believed that a “court must assess the confession’s ‘inculpatory value’ by examining not only the face of the confession, but also all of the evidence introduced at trial,” 481 U.S. at 205-06, 107 S.Ct. at 1706, but that decisions “of other Courts of Appeals ... have declined to adopt the ‘evidentiary linkage’ or ‘contextual implication’ approach to
Bruton
questions.”
Id.
at 206, 107 S.Ct. at 1706 (citation omitted).
The Supreme Court referred to this Court as such another court of appeals rejecting a “contextual implication” approach, citing
United States v. Belle,
593 F.2d 487 (3d Cir.1979) (en banc). In
Belle
we approved the approach of the Court of Appeals for the Second Circuit that “evi-dentiary linkage or contextual implication may not be utilized to convert a
non-Bru-ton
admissible statement into a
Bruton
inadmissible statement.”
Id.
at 494.
Belle
remains good law in this Court as the Supreme Court reversed the decision of the Court of Appeals for the Sixth Circuit on the appeal before it and in doing so certainly did not overrule
Belle,
and we have not overruled it either.
After it decided
Richardson
the Supreme Court decided
Gray,
which we understand to read
Richardson
to hold that in
Richardson
there had not been a
Bru-ton
violation because the challenged state-
merit incriminated the objecting defendant only when linked with evidence introduced later at the trial.
Gray,
528 U.S. at 191, 118 S.Ct. at 1154.
Richardson
and
Gray
reinforce us in our conclusion that Belle remains good law because our reading of those cases indicates that the correct
Bru-ton
approach in considering evidence extrinsic to the challenged statement is completely consistent with
Belle.
Yet the teaching of these cases dealing with the extrinsic evidence issue culminating in
Hardwick
may be that there can be a
Bruton
violation in either of two situations. The first basis for a violation would be if the trial court erroneously admitted into evidence or allowed the use at trial of a statement that on its face incriminated the objecting defendant. The second basis for a violation would be if the court admitted into evidence or allowed the use at trial of a statement that became incriminating when linked with other evidence in the case. Here, however, whichever approach we take,
ie.,
limiting a
Bruton
violation analysis to the statement itself or considering the asserted
Bruton
violation in the context of the entire trial, our result would be the same, and thus we need not decide which approach is correct. Either way, there was a
Bruton
violation in this case.
The parties are in agreement, and we concur with their view on this point, that three Supreme Court cases,
Bruton,
391 U.S. 123, 88 S.Ct. 1620;
Richardson,
481 U.S. 200, 107 S.Ct. 1702, 95 L.Ed.2d 176; and
Gray,
523 U.S. 185, 118 S.Ct. 1151, establish the controlling precedent for this case for purposes of 28 U.S.C. § 2254(d)(1). Unquestionably, the leading case on the restrictions on the use of a non-testifying co-defendant’s statement incriminating another defendant is
Bruton,
and thereafter
Richardson
and
Gray
dealt with refinements in
Bruton’s
application.
We sum up
Bruton, Richardson,
and
Gray
as follows. In
Bruton
the Court held that a defendant is deprived of his rights under the Sixth Amendment’s Confrontation Clause when the non-testifying co-defendant’s statement naming him a participant in the crime is introduced at their joint trial, even if the trial court instructs the jury to consider the statement only against the non-testifying co-defendant.
The Court, however, limited
Bru-ton
in
Richardson
when it held that “the Confrontation Clause is not violated by the admission of a non-testifying codefendant’s confession with a proper limiting instruction when ... the confession is redacted to eliminate not only the defendant’s name, but any reference to his or her existence.” 481 U.S. at 211, 107 S.Ct. at 1709. Thus, the Court distinguished the statement in
Richardson
from that challenged in
Bru-ton
as the
Bruton
statement was “incriminating on its face” with respect to the objecting defendant, whereas in
Richardson
the statement did not implicate the objecting defendant.
Id.
at 208, 107 S.Ct. at 1708.
Richardson,
however, also had its limitations because in that case the Court specifically indicated that it “express[ed] no opinion on the admissibility of a confession in which the defendant’s name has been replaced with a symbol or neutral pronoun.”
Id.
at 211 n. 5, 107 S.Ct. at 1709 n. 5.
Finally, in
Gray
the Court addressed the scope of redactions, 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. at 1156.
Applying
Bruton, Richardson,
and
Gray,
we cannot escape from a conclusion
that the Superior Court’s decision upholding the use of Santiago’s statement, even as redacted and subject to an instruction that the jury should not use it against Vazquez, though probably not directly contrary to those cases, plainly was “an unreasonable application” of them. In reaching its result the Superior Court, citing the Pennsylvania cases of
Travers, Rivera,
and
McGlone,
indicated that “the Pennsylvania Supreme Court has held that a non-testifying co-defendant’s statement in which a defendant’s name is replaced with the term ‘other guy’ or a similar term is admissible in a joint trial when coupled with a cautionary instruction.” App. at 108. Of course, the reference to a “cautionary instruction” relates to precluding its use against the incriminated but objecting defendant. The Superior Court then explained that Santiago’s statement was redacted in accordance with those statements, the redaction was neutral and did not facially implicate Vazquez, and on its face the statement as redacted did not reveal that names had been removed from the statement.
The Superior Court then reached the heart of its
Bruton
discussion:
The manner in which the statement was redacted still leaves to the jury to decide which of the three men fired the shots. [Vazquez] also testified at trial and denied that he was the shooter; he claimed that Rivera fired the shots at the victim. Even if the jury credits Santiago’s statement that he was the driver, they still must decide whether Rivera or [Vazquez] fired the shots. Obviously Santiago could identify the shooter as the other passenger in the vehicle as he was admittedly present when the incident occurred.
App. at 108-09.
We think that the problem with the Superior Court’s conclusions under
Bru-ton, Richardson,
and
Gray
is quite clear. To start with, the possibility that the jury could have disbelieved Santiago’s statement entirely and concluded that he was the shooter is immaterial. Certainly, the nature of the judicial process allows a jury in a
Bruton
situation to reject an incriminating statement, but that possibility does not eliminate the chance that the jury will credit the statement and conclude that the statement pointed to the objecting defendant as the offender even though he could not cross-examine the declarant. Indeed, the result at the trial in this case indicates that that is what happened here because the jury acquitted Santiago but convicted Vazquez. Thus, as far as admission or use of the statement is concerned, this is and always has been a two-person case involving Vazquez and Rivera, and the Superior Court’s attempt to expand it into a three-person case was unavailing.
In any event, even though our outcome does not depend on the point, we can perceive of no way that the jury could have failed to credit Santiago’s statement that he was the driver at the time of the shooting as the statement on that point was completely plausible, the prosecutor did not reject it, and there was no contrary evidence suggesting that anyone else was the driver. Moreover, when the police stopped the Buick, Santiago was its sole occupant, though we acknowledge that because of the time interval between the shooting and the stop it is conceivable that there could have been different drivers at the two times. Inasmuch as the Superior Court recognized that the jury was likely to have concluded that Santiago had been the driver, under the Superior Court’s analysis in reality there were two possible factual candidates for the role of the shooter of the fatal shot: Vazquez, identified in Santiago’s statement as a shooter, and Rivera, as Vazquez testified.
Moreover, and this would be a crucial point if we accepted rather than rejected the Superior Court’s approach that we are dealing with a three-person case, if we view the case in its status when the trial court ruled when making both its pretrial and trial rulings with respect to the admission or use of Santiago’s statement, the trial court knew or surely should have known that it was likely that the jury would conclude that Santiago was the driver and that if the jury accepted the statement at least to that extent, as it clearly did, no one other than Vazquez or Rivera could have fired the fatal shot. In this regard, we point out that when the trial court ruled it was aware of the contents of Santiago’s statement.
The fact that there were only two possible shooters under Santiago’s statement should have made clear to the trial court that, whether or not the jury credited the statement in its entirety, it was almost certain to conclude that the individual Santiago described in his redacted statement as “my boy” or “the other guy” as the shooter was Vazquez because Rivera was not on trial and the Commonwealth argued that Vazquez fired the fatal shot.
See Hardwick,
544 F.3d at 573. Thus, we are constrained to reverse the order of the District Court and grant habeas corpus relief for if this case does not involve “an unreasonable application! ] of clearly established Federal law, as determined by the Supreme Court of the United States,” it is difficult to conceive of any case that could meet that admittedly exacting standard.
We emphasize that because of the deference and respect that we give the Pennsylvania state courts, not only because' of the requirements of the AEDPA but in general, we reach our result reluctantly. Yet we are compelled to recognize, though appel-lees contest this point, that the Supreme Court of Pennsylvania in
Travers
and
Rivera
came close to endorsing a bright-line rule that when terms like “my boy,” the “other guy,” or the “other man” are used to substitute for an actual name in a statement admitted at trial there cannot be a
Bruton
violation. Thus, in
Travers
that court indicated that:
This case ... involves [the] question [of] the viability of a redaction that substitutes a neutral pronoun ... for the defendant’s name. Specifically, the eo-de-fendant’s statement here was redacted to replace references to appellant by name with the term ‘the other man.’ Although this was not the type of redaction at issue in
Gray,
the
Gray
Court’s reasoning, including its distinction of
Richardson,
leaves little question that this sort of redaction is appropriate under the Sixth Amendment. At a minimum, as one Circuit Court has noted, the Supreme Court ‘strongly implied’ in
Gray
that a redaction employing a neutral pronoun such as ‘the other guy’ does not offend the Sixth Amendment.
Travers,
768 A.2d at 850-51 (citation omitted). Then the Supreme Court of Pennsylvania indicated in
Rivera:
In
Commonwealth v. Travers,
564 Pa. 362, 768 A.2d 845 (2001), this Court held that the redaction of a non-testifying co-defendant’s confession, which replaced any direct reference to the defendant with the words ‘other man,’ when accompanied with the appropriate cautionary change, sufficiently protected a defendant’s confrontation clause rights. We held that this method of redaction not only eliminated the name of the defendant, but also eliminated any suggestion of alteration, and thus, eliminated the incriminating nature of the obvious deletion or blank method of redaction used in
Gray.
Rivera,
773 A.2d at 138.
Courts and attorneys cherish bright-line rules as such rules simplify their tasks and
lay out clear paths for them to follow. Furthermore, it certainly is true that ordinarily the use of a term like “the other guy” will satisfy
Bruton.
Nevertheless, it is an unreasonable application “of clearly established Federal law under the decisions of the Supreme Court of the United States” to hold that their use always will be sufficient for that purpose. Here, regardless of whether the Pennsylvania Supreme Court established the bright line that we discern in its opinions, the use of a generic name in place of an actual name plainly was not sufficient to satisfy
Bruton
without regard for whether we view this case in accordance with the confined
Belle
or expansive
Hardwick
approach.
In reaching our result we, of course, have not overlooked our opinions under
Bruton,
even though we have indicated that “court of appeals precedent is irrelevant to the ultimate issue.... ”
Wilkerson v. Klem,
412 F.3d 449, 455 (3d Cir.2005). In
Wilkerson
we regarded our opinions as being cabined because we are obliged to ascertain whether the state court decision being examined “was contrary to, or involved an unreasonable application of clearly established Federal law, as determined by the Supreme Court of the United States.” 28 U.S.C. § 2254(d)(1). Nevertheless, we address
United States v. Richards,
241 F.3d 335, and
Priester v. Vaughn,
382 F.3d 394, dealing with
Bru-ton.
In
Richards
we found that the substitution of “my friend” and “inside man” to be facially obvious with respect to the person identified. 241 F.3d at 341. Thus, in
Richards
we held that the use of the those terms in place of the co-perpetrators’ names violated the Sixth Amendment because it “was just as blatant and incriminating” as the word “deleted” in
Gray.
241 F.3d at 341. The situation in
Richards
insofar as it considered whether there had been a
Bruton
violation, cannot be distinguished from that here.
On the other hand, in
Priester
we emphasized that in a
Gray
analysis the number of persons involved is significant. In
Priester
at least 15 persons were involved, so that the use of the terms “the other guy” or “another guy” did not point to any person. 382 F.3d at 399^100. Thus, in
Priester
we distinguished
Richards
by pointing out that because
Richards
involved only three people, in
Richards
the redactions were “tantamount to an explicit reference to the co-defendant.” 241 F.3d at 401.
Finally, in concluding that there was a prejudicial
Bruton
violation in this case, we have not overlooked appellees’ argument that Vazquez cannot demonstrate that the
Bruton
error had a “substantial and injurious effect” on the trial.
See Fry v. Pliler,
551 U.S. 112, 127 S.Ct. 2321, 2328, 168 L.Ed.2d 16 (2007);
Bond v. Beard,
539 F.3d 256, 275-76 (3d Cir.2008). Therefore, we have considered their contentions that Vazquez’s own testimony “greatly benefitted the Commonwealth,” Vazquez made “repeated attempts to flee police,” thus implicating himself, and Vazquez told Santiago that the crime should have been covered up. Appellees’ br. at 26. Moreover, we recognize that fingerprint evidence pointed to Vazquez as the shooter.
In considering this case under the “substantial and injurious effect” standard it is helpful to compare this case to
Bond.
In that cqse the prosecution did not deny that there had been a
Bruton
Confrontation Clause violation but argued, instead, that
the error had been harmless. But in
Bond
there was an eyewitness who testified at the trial and identified the petitioner as the shooter. Furthermore, the petitioner had confessed to the commission of the crime. Here, Vazquez never confessed to being a shooter, and there was no witness at the trial who said that he saw Vazquez fire a weapon. In a narrative or descriptive sense, laying aside ballistic evidence, only Santiago’s statement directly identified Vazquez as the shooter, and, of course, the use of the statement is the problem in this case rather than its solution. Moreover, although there was evidence at the trial incriminating Vazquez beyond Santiago’s statement, it was not so compelling that it overcame the
Bruton
error under the “substantial and injurious effect” standard.
In view of our result under
Bruton,
we need not reach Vazquez’s other points relating to the denial of his motion to sever and the prosecutor’s alleged misconduct attributable to her identification of Vazquez as the person that Santiago said was the shooter. Clearly, inasmuch as Santiago was acquitted the severance issue cannot arise again, and we see no reason to believe that the prosecutor at a new trial will repeat the prosecutor’s closing argument.
V. CONCLUSION
We note that the attorneys representing Vazquez on this appeal have done so on a pro bono basis and we thank them for these services, which have been in the highest tradition of the bar. For the foregoing reasons we will reverse the order of the District Court of March 21.2007, and
will remand the case to that Court for further proceedings consistent with this opinion. In particular, the District Court should order that the state authorities free Vazquez from custody unless he is retried in the state courts with the trial to start within a period of time the District Court fixes.