Actually, you can get away with assigning anything you want. That's hidden in the assumption that the criteria he lists are more-or-less independent.
> You might have a different list of criteria or different probabilities, but the point is to lay them down and give a reason why you think it's unlikely enough not to pay the small sum of money.
You've got the burden of proof backward. The cryonics advocates are the ones trying to argue that cryonics will work with sufficiently high probability to make investing in it worthwhile. The grandparent cited 5% as if that figure had even one significant digit -- which it doesn't, by Hanson's own calculation.
> It also lets others argue with you over your estimates, because they might have information you don't, now or in future years.
It also lets others argue even if they don't have new information, which just wastes everyone's time. The same phenomena also occurs with the Drake equation.
The "non-zero" argument can be fixed simply by comparing relative likelihoods, or noting that while one can't ever rationally believe with 0% probability, you can use epsilon as a substitute for "it's more than 0 but I can't measure how much because it's so small". Personally I think there's enough evidence to put religious ideas of afterlife far below the probability of a single sha256 hash resulting in a new block in the bitcoin blockchain: http://blockexplorer.com/q/probability