dissenting.
With respect, I disagree with the majority’s view of the case. Martin’s identification of Perkins as one of the assailants was the central evidence of his guilt. There was no physical evidence tying Perkins to the crime and no other witness testified against him at trial. In addition, because Martin knew Perkins well — he was his drug dealer and had dated Perkins’ sister — there was no issue of the reliability of his eyewitness identification. Rather, the paramount issue in the case was Martin’s honesty. Did Martin identify Perkins as the assailant because Perkins in fact committed the crime or because Martin held a grudge against Perkins related to Martin’s romantic relationship with Perkins’ sister and wanted to frame him for the crime? Against this backdrop, it is now brought to the court’s attention that, according to Martin’s own affidavit, Martin lied at trial when asked whether he received any inducement to testify against Perkins. Because Perkins did not have access to important impeachment information to which he was entitled (that a police officer told Martin that “things [would go] a lot easier for [him]” if he testified against Perkins), Perkins was left to battle the key evidence of his guilt — Martin’s identification — with one hand tied behind his back.
As a threshold matter, I believe we must review Perkins’ claim de novo because, as the majority concedes, “the state court treated Perkins’ disclosure and subornation claims interchangeably and did not cite any cases — state or federal — discuss*121ing the standards governing those claims.” See Pina v. Maloney, 565 F.3d 48, 54 (1st Cir.2009) (“In the absence of a state merits determination, the more deferential habeas review under the Antiterrorism and Effective Death Penalty Act is inapplicable and our review is de novo.”). In addition, on the prejudice question, the Massachusetts Appeals Court made no reference to or consideration of the Supreme Court’s decision in United States v. Agurs, 427 U.S. 97, 96 S.Ct. 2392, 49 L.Ed.2d 342 (1976).
Employing de novo review, I would reach a different answer than the majority on the prejudice question. The test for prejudice when there has been knowing suborning of perjury by the prosecution is whether, “there is any reasonable likelihood that the false testimony could have affected the judgment of the jury.” Agurs, 427 U.S. at 103, 96 S.Ct. 2392. Given the particulars of this case, I think this relatively “defendant-friendly standard” is met. Mastracchio v. Vose, 274 F.3d 590, 601 (1st Cir.2001). The case against Perkins relied almost entirely on the jury’s perception of Martin’s honesty. Yet, Martin’s honesty was already called into question by at least two factors: his status as a drug dealer with pending charges against him and the presentation of evidence suggesting that Martin had a motive for disliking Perkins and wanting to frame him for the crime. Given that the jury had to convict Perkins beyond a reasonable doubt, I believe the additional increment of doubt provided by the information that a police officer had induced Martin to testify against Perkins may well have “affected the judgment of the jury.” Agurs, 427 U.S. at 103, 96 S.Ct. 2392.
The majority dismisses the importance of the inducement evidence by stating that it “had to be obvious” to the jury that Martin had a personal stake in testifying and “anyone with sense” would know that Martin would be better off cooperating. The majority concludes that, “The policeman’s alleged statement only spelled out the obvious.” However, guilt and innocence, and in this case the question of prejudice, should be determined based on the evidence presented at trial. We should not rely on a vague notion that the jury surely must have presumed as much as a substitute for the admission of persuasive impeachment material that could well have tipped the scale in Perkins’ favor. Effective and constitutionally sufficient defenses are not built on hunches and presumptions of general knowledge on the part of the jury; they are built on admissible evidence. To conclude otherwise, I believe, ignores the central crucible of our adversarial system.
In addition, I am not persuaded by the majority’s argument that the impeachment evidence was irrelevant simply because Martin made his identification of Perkins shortly after the incident and before his own arrest in the subsequent drug case. This ignores Perkins’ argument, made at trial, that Martin should not have been believed because, at the time of the identification, he had a contemporaneous reason to dislike and therefore frame Perkins.
Thus, because I would conclude that there is a reasonable likelihood that prejudice would have resulted had there been suborning of perjury regarding the inducement, I would remand the case to the district court for an evidentiary hearing on the subornation issue. The majority dismisses the relevance of what I see as the most important question in this case— whether the prosecutor knew or should have known of Martin’s apparent perjury. Agurs, 427 U.S. at 103, 96 S.Ct. 2392. I would remand for a factual finding on precisely that issue.
In a case where guilt and a significant prison sentence hinged on the testimony of *122one man — an admitted drug dealer facing criminal charges of his own — and where there is a supported claim of suborned perjury regarding material that might well have impeached the credibility of that sole witness, I believe an evidentiary hearing to determine whether subornation actually occurred is appropriate and just.
I therefore respectfully dissent.