It turns out most commentors are of a rather different opinion on 1. and 2., so ... update time!

New evidence includes reports that use of the involved drugs does significantly impair memory, a reminder that absence of evidence is evidence of absence, and fairly damning judgements about the Italien legal system. I also realized that my priors for P(group rape | rape) and P(group murder | murder) were insufficiently extreme (i.e. too high).

All of this is fairly general evidence about such cases, so the jury could well have had access to it, as well. They're still in a position of having much better access to evidence, and the real question becomes how rational they are in evaluating it.

Updated probabilities are:

1. 0.35
2. 0.35
3. 0.9