dissenting:

The ultimate question in this case is twofold. First, whether it was “objectively unreasonable” for the California Court of Appeal and Supreme Court to have concluded that Reynoso’s appointed counsel’s performance — with respect to her handling of witnesses Mendoza and Terrones — was within the range of competence demanded *1121 by the Sixth Amendment. See Strickland v. Washington, 466 U.S. 668 , 104 S.Ct. 2052 , 80 L.Ed.2d 674 (1984); Williams v. Taylor, 529 U.S. 362 , 120 S.Ct. 1495 , 146 L.Ed.2d 389 (2000). If the answer to this question is “no,” the case is closed in the Warden’s favor. If the answer is “yes,” then we must decide whether those courts would have been objectively unreasonable in concluding that Reynoso suffered no prejudice from counsel’s deficient performance. Because we have “no reasoned state court decision” on these issues, we must perform an independent review of the record to ascertain whether the state court decision was objectively unreasonable. Pham v. Terhune, 400 F.3d 740, 742 (9th Cir.2005) (quoting Himes v. Thompson, 336 F.3d 848, 853 (9th Cir.2003)).

We now have a record from our district court which partially illuminates these issues and weighs upon the objective reasonableness of tactical decisions made by Rey-noso’s counsel, but herein lies a serious problem that — with all respect to my colleagues — makes it impossible for us at this point to render a valid decision: the district court failed to answer important factual questions to which we need answers as a predicate to our legal analysis.

I

The first question is whether defense counsel knew that witnesses Mendoza and Terrones were aware of the possibility of a reward, and in turn, whether she chose not to pursue that line of questioning with them in favor of another line of attack.

To begin with, the prosecutor — found to be a “particularly credible witness” by the district court — maintained under oath that defense counsel was aware of the existence of the well publicized reward long before the trial and that defense counsel was aware also that Lopez, Terrones, and Mendoza knew about the reward. As developed by the district court,

(Regios Deck ¶ s 4, 7).

So far, so good. Regrettably, however, the district court did not determine which version of the pivotal facts is correct. Did the prosecutor and defense counsel discuss before the trial the “fact that Terrones and Mendoza knew about the reward,” or not? Instead, the district court chose not to resolve this critical factual dispute, saying in a footnote that “[b]ecause no Brady violation occurred, the Court need not decide whether the prosecutor or the defense was more credible, although the Court notes that the prosecutor was a particularly credible witness.” Nevertheless, the magistrate judge then assumed, — contrary to the “credible” prosecutor’s testimony— that defense counsel “failed to determine whether Mendoza and Terrones knew about the reward,” and that this failure was prejudicial Strickland error. What happened to the “credible prosecutor’s” testimony that she and defense counsel discussed the fact that Lopez, Mendoza, and Terrones knew of the reward? In my judgment, the court clearly erred in declining to decide which version of the facts was correct, and then proceeding as though it had made such a finding — • against the prosecutor and the Warden.

With all respect to the district court, one can only wonder why the magistrate judge failed to resolve these pivotal questions when one reads the precise issues requiring resolution identified by the court itself in court minutes entitled “Scheduling of Evidentiary Hearing and Briefing Schedule” dated March 21, 2003. Directing counsel to file “supplemental briefs ... on the following issues,” the court offered these questions for resolution at the evi-dentiary hearing:

The answers to these questions — especially #2 — are crucial, because if the court had found the prosecutor’s version to be true, then we might have a clear ease of a strategic choice not to go into the reward with Mendoza and Terrones, but — as defense counsel said at the hearing — to pursue another line of attack on the validity of the eyewitnesses’ testimony. Counsel described her general method of defense as not wasting time on weaker issues, which would explain why she might have known about the reward issue but chosen not to use it:

(Emphasis added).

The majority’s opinion illustrates why remand is necessary. Instead of being able to work from facts found by the district court, the opinion says, “the most likely explanation of what transpired is that the defense counsel had some discussion about the reward, but for whatever reason — perhaps because the conversation lacked sufficient specificity — Regios either failed to receive or failed to understand the full message; in short, whether the prosecutor or defense counsel was at fault, and it may have been both, Regios did not comprehend that Mendoza and Terrones had acknowledged their awareness of the reward, and she had no knowledge that they had expressed an interest in receiving it.” (Emphasis added). This fanciful rationalization by the majority opinion tactfully admits we are swimming in speculation, which is not appropriate given the seriousness of this case and of the issues it raises. At the very least, the majority’s discussion suggests a failure on the part of the petitioner to satisfy his burden of proof regarding his attorney’s behavior.

Thus, we need to remand this case to the district court pursuant to 28 U.S.C. § 2106 to answer the pivotal questions the court itself understood to be critical. As permitted by the statute, this case cries out for “further proceedings” in the interest of justice. The issue is very simple: Regios either knew that Terrones and Mendoza knew about the reward, or she did not.

II

This mistake was ours, the federal courts. This factual gap was created by *1124 us, not counsel, and before we undo a state criminal conviction, we must be on solid factual ground. Right now, we are trying to come to a legitimate answer without getting to the bottom of a central factual dispute. What good is § 2106 if we are not going to use it? We rely on our district courts to find the facts, and then, we weigh in on the law. Here, the system has failed. Given the record, the prosecutor’s testimony, the defense attorney’s testimony, the defense attorney’s aggressive assault against Lopez because of his interest in the reward, and the apparent minimal interest of both Mendoza and Ter-rones in the possibility of a reward, it is manifestly and demonstrably probable that defense counsel made a reasonable strategic choice to focus on fertile ground, and not dilute her attack by opening up another front. For example, Terrones testified at the evidentiary hearing that he had no interest in the reward, even though he had asked about it. Likewise, the police had to track Mendoza down as a witness, and he told a detective that he, too, was not interested in the reward. How far does a defense attorney get with witnesses saying, “Sure, I knew about the reward, but that’s not why I am testifying” — which is what Lopez said when confronted at trial on this issue. The scenario of awareness- and-choice-not-to-pursue is strongly supported circumstantially by the record, which includes a remarkable number of “I don’t remember” answers by the defense attorney — who did not bother to review her file before the hearing.

With all respect to my colleagues, this appeal is not ripe. The appropriate step at this juncture is to remand the matter to the district court for answers to that court’s own questions. Thus, I register my dissent.