[beakachu]

Let me say first that I don't particularly have a problem with the idea of putting a (more or less precise) number on the likelihood of an event having happened. But I don't believe that putting a probability on guilt fits with many people's ideas about knowledge.

The problem here is much-discussed in philosophy: on an intuitive basis, it seems that you can know A will happen, but not know B will happen, despite B being more likely than A. The typical example involves B being the outcome of a lottery. Having bought a lottery ticket, I can't say I know I won't win, or else I clearly wouldn't have bothered buying it. But people can reasonably claim to know much less likely things (or else we'd never have knowledge).

It could just be that intuition is off in these cases, but they're taken fairly seriously by philosophers. We can imagine similar criminal cases. What if a man is on trial for murder as part of a 10-man firing squad, one of whose guns fired a blank? Surely he can't be convicted (although presumably he could be convicted of attempted murder), but I don't think it's on the basis of his 1/10 chance of innocence. Personally, I'm not sure you could convict him if the squad was 100 or 1000 or more men.

Or maybe you could. Intuition only goes so far.