With its grant of certiorari in the case of Chrones v. Pulido, No. 07-544, the U.S. Supreme Court ventures once more into the questions of habeas corpus, harmless error, and deference to the state court's decision on direct appeal. The Court addressed related issues last June in Fry v. Pliler, No. 06-5247. The new case deals with the situation where a jury is given more than one path to a conviction, of which one is right and the other wrong. The Supreme Court addressed that situation in Stromberg v. California, 283 U.S. 359 (1931).
Much water has passed under the harmless-error bridge since Stromberg. In more recent cases, the Court has divided errors into "structural" and "trial." Structural errors are reversible without any inquiry as to whether the error was prejudicial in the individual case. Such errors are rare. Most errors are judged under a harmless error standard. On direct appeal, an appellate court asks if a federal constitutional error was "harmless beyond a reasonable doubt," the Chapman test. On habeas corpus, and for nonconstitutional errors at any phase of review, "an error is harmless unless it had substantial and injurious effect or influence in determining the jury’s verdict," the Kotteakos/Brecht test.(Fry, slip op., at 3, internal quote marks omitted.)
On May 24, 1992, Ramon Flores was shot to death during a robbery of the Shell gas station where he worked in San Mateo, California (about 15 miles south of San Francisco). Michael Pulido was convicted of first-degree murder and sentenced to life in prison for this crime. After telling several different stories, he finally settled on claiming that his uncle, Michael Aragon, committed both the murder and the robbery, and that he (Pulido) only joined the crime after Mr. Flores was already dead. Pulido's version of what happened is inconsistent with the fact that his fingerprints are on a can of Coke left on the store counter. Also, his fingerprints and not Aragon's are on the stolen cash register.
On appeal, the California Supreme Court decided that if Pulido's version were believed, that would not be felony murder under California law. However, the state court applied harmless error analysis and found no need to reverse because the jury had rejected Pulido's claim under another finding -- the "special circumstance" finding that the murder was committed while the defendant was engaged in or was an accomplice to a robbery. See People v. Pulido, 15 Cal. 4th 713, 936 P. 2d 1235 (1997).
On federal habeas, the Ninth Circuit applied its precedent in Lara v. Ryan, 455 F.3d 1080 (CA9 2006), which "held that such error was structural and that 'where a reviewing court cannot determine with absolute certainty whether a defendant was convicted under an erroneous theory' reversal is required." Pulido v. Chrones, 487 F. 3d 669 (CA9 2007). The requisite certainty was not present, said the Ninth, because the special circumstance instruction itself had an error that made reliance on an invalid theory possible.
I suspect the Supreme Court's answer will be that the habeas court should have simply applied the straight Brecht test. See Fry, slip op., at 7. Introducing (or reintroducing) a third standard into the harmless error mix has little to recommend it. The kind of error involved here is not so different from other kinds of errors as to call for a different standard. Whether Pulido would get relief under that test is a question they are likely to leave for the Ninth Circuit on remand.
That may well be what the Supreme Court does. But the Brecht standard makes no sense in the erroneous theory context. It is impossible to judge the effect on the jury's verdict when a case has been submitted on an alternative, erroneous theory.
It actually makes much more sense to treat an erroneous theory case as structural error a la Sullivan v. Louisiana. It's a little different, because one simply does not know whether there is the erroneous theory vitiates the verdict in the same way the reasonable doubt instruction in Sullivan did.
Brecht also has not been applied in any ineffective assistance habeas case that I have seen. Of course, I haven't seen all the cases. The interesting thing about the IAC cases in this context is that the constitutional violation, once found, leads to an automatic reversal. The weighing goes into the prejudice prong of the Strickland analysis, not into whether relief is warranted.
Life in Habeas Land is not simple. Maybe it should be. But simplifying by applying the Brecht standard, which on its face has no application to the situation, cannot help
I would not characterize Strickland as an "automatic reversal" rule. I would characterize it as a rule with a "harmfulness" component built in. That is why no separate harmless error analysis is needed, because the petitioner has already cleared a hurdle higher than the one he would have needed to clear to defeat a harmless error claim by the state. The Supreme Court so held for Brady claims in Kyles v. Whitley, 514 U.S. 419, 436 (1996), and the prejudice prong of Strickland is essentially the same. However, for Strickland, unlike Brady, I think this is an inquiry into "whether relief is warranted."
In any case where it is indeed "impossible to judge the effect on the jury's verdict," then the Brecht test is satisfied, and applying it does no harm. I do not think that is invariably the case, though. Hypothetically, a judge might read a jury a form instruction with two theories, but all the evidence and argument is directed to one and there is no other mention of the other. In that case, there would be no "grave doubt," O'Neal v. McAninch, 513 U.S. 432, 445 (1995), and habeas relief should be denied.