The Ninth Circuit's certification request here makes facial sense. California law on the issue is indeed unclear. Reasonable minds might differ. Comity, and modesty, thus counsel in favor of certifying the questions.
But there's a minor twist. The California Supreme Court has already granted review of a case (Albillar) that raises very similar questions to those that the Ninth Circuit certifies. The panel argues that the questions are slightly different (and presented in a unique way) and certification thus appropriate notwithstanding the pending case. Maybe. But I could easily see the California Supreme Court declining the certification on the theory that its answers in Albillar are (or will be) "close enough". So that's one way the Ninth Circuit might hear a "No" from its brethren.
But there's also a major twist, and one that I think raises policy (not merely practical) issues. The certified question here arises in a habeas case. That's different. Way different. Federal courts typically certify state law issues when they're presented on direct appeal and in settings in which the state courts have had no opportunity to opine on the particular issue. But that's not at all true in habeas cases, in which the matter began in the state court and in which the state judiciary has already expressed its opinion; namely in affirming the conviction. Moreover, since we require habeas claims to be exhausted, you already have a state court decision on the merits. Further, since we require complete exhaustion, we also have -- as here -- a state supreme court that has already refused to answer the certified questions when it decided to deny review when the case was on direct appeal.
It seems to me that the comity and other concerns that motivate certification might be very, very different in such a case than in the usual case in which the state courts have never had jurisdiction over the case. Moreover, I could imagine at least colorable arguments that federal courts shouldn't certify such cases. After all, if the state's highest court has already denied review, having found the issues unworthy of its time, why should they be again bothered. It might even be viewed as anti-comity for a federal court to essentially say: "Hey, you thought this case wasn't worthy of your time, but we disagree, and want you to hear it."
Moreover, it seems to me that there may be a real downside to giving the state courts two shots at answering a question -- the first on direct appeal and the second on habeas. Practically, when a federal court certifies a question, that's a pretty good indication that the federal court may be thinking about granting relief (e.g., that reasonable minds on the panel might disagree). Were a federal court to certify such a case, it's basically telling the state court: "You decided X, and affirmed the conviction, but we're thinking about reversing you and granting habeas. Do you want to revisit X, or decide Y, to avoid a reversal?" That seems a bit unseemly to me. Especially when the state court might have an understandable interest in the finality of its result and thus have an incentive to answer the certified question in a way that will affirm the validity of its own prior judgment.
You don't have these issues arise in the usual certification cases, which is why I'm generally in favor of that process. But in habeas cases, I don't know. This seems different. In a way that may require a bit more examination. Both from the panel as well as the California Supremes.
So it's an interesting case. As well as an interesting topic: Certification in habeas cases. An issue that I readily concede I hadn't at all thought about before today's opinion.