I don't closely follow habeas decisions from the First Circuit (covering most of New England) but the court's decision last week in Deciantis v. Wall, No. 12-2383 (1st Cir. July 1, 2013), caught my attention. While the decision probably reached the right result, it raises the interesting and recurring question of how accurately judges can predict what a jury would have done absent some trial error.
To begin with, the decision is worth noting for the sole reason that it reads like an outtake of the Departed. It was the early 1980s, and the mob still held sway in many big cities. Cities like Providence, Rhode Island, where on December 4, 1981, Anthony DeCiantis and another man forced Dennis Roche into a car and drove him to garbage dump. The next day, Roche's body was found with two gunshot wounds, several stab wounds, and "injuries to the face and head consistent with his having been run over by a car." Witnesses testified that DeCiantis killed Roche for the part he played in the disappearance of DeCiantis' brother Rocco. DeCiantis was convicted of first-degree murder in 1984 and sentenced to life.
The Case of the (Really) Bad Prosecution Witness
Fourteen years later (1998), DeCiantis moved for state postconviction relief on the ground that the state had withheld at trial material information favorable to the defense in violation of Brady v. Maryland. Brady requires prosecutors to disclose to the defense, among other things, their witnesses' prior criminal histories as well as any benefits conferred by the State on a witness that the defense might use to impeach the witness. One of the witnesses against DeCiantis was William Ferle, a mob turncoat with an extensive criminal history. Prior to trial the prosecutor told DeCiantis about several of Ferle's crimes, including conspiracy, bank fraud, robbery, murder, and arson. But, according to DeCiantis' postconviction motion, the State neglected to disclose information about many other crimes that Ferle told the authorities about. This withheld list included twenty-four crimes, including several additional instances of arson, robbery and murder.
The Providence trial court was in no hurry to consider the motion - it waited seven years to hold a hearing (December 2005) only to deny the claim in 2007. Four years after that, the Rhode Island Supreme Court affirmed (2011). Essentially, both courts held that the prosecutor had disclosed to DeCiantis enough dirt on Ferle to allow him to show the jury that Ferle was a shady character, and that the failure to disclose the full list of Ferle's wrongdoing did not, at the end of the day, make a difference to the outcome. In Brady jurisprudnce, this idea is captured by the "materiality" standard: the Rhode Island Supreme Court determined that evidence of Mr. Ferle's additional criminal activity was not material:
"It is clear to us that it could not have escaped the attention of the jury that Mr. Ferle had an extensive criminal background, that he had decided to testify against members of organized crime (and that he was recently such a member himself), and that he hoped to receive protection and avoid incarceration in exchange for his testimony. We are unpersuaded that evidence of additional criminal activity on the part of Mr. Ferle would have materially changed the jury's evaluation of him as a witness or would have changed the ultimate result of the proceeding."
Can Judges Read Juries' Minds?
Ultimately, the federal courts agreed with this conclusion. And, let's face it, in this case it's almost certainly correct. As bad a guy as Ferle was, DeCiantis was no angel (or simply one with a dirty face). Other evidence (not affected by the alleged Brady violation) showed that it wasn't enough for DeCiantis to shoot Roche to death - he had to stab him and run over him with his sister's car for good measure. On facts like those, a jury is unlikely to acquit even where one of the witnesses has to squeeze his testimony into a full schedule of criminal conduct. As the First Circuit put it, "e[v]en if we assume this evidence would have induced the jury to view Ferle differently, there was still the testimony of the other witnesses at trial, who overwhelmingly pegged DeCiantis."
But what about the cases where the crime is not so brutal, nor the evidence so overwhelming? Brady claims in those cases also require reviewing courts to make materiality determinations - that is, they must answer what the jury would likely have done if it had some additional information. Defense lawyers are always telling clients that trials are risky; that "you never know what a jury is going to do." Why are judges as a group any better at this than defense attorneys or prosecutors?
In fact, the evidence is that they are not great at it. In their recent book, "Not Guilty: Are the Acquitted Innocent?," authors Daniel Givelber and Amy Farrell cite a widely-quoted poll of judges concluding that they agree with jury verdicts in their cases only 78% of the time. Of the cases where the judge's personal "verdict" is at odds with the juries' actual one, judges are far more likely to convict where a jury acquits (19%) than to acquit where a jury convicts (3%). These numbers suggest that, as a whole, judges' after-the-fact calls on what a jury "would have done" may be biased in favor of upholding a conviction (the only context in which Brady claims arise).
Brady's materiality standard has been criticized for many reasons, and this may be one more reason to rethink it. In any event, judges should take care to carefully consider what a jury - in all its inscrutable, irrational glory - would have done with additional evidence, and not allow the judge's own preferences and experience to fill the void. While the decision may not be a tough call in the case of a wiseguy like DeCiantis, it is less clear in less Scorcesesque cases.