I think that one of the problems is that a lot of the earthquake prediction community is trying to run before it can walk. I know Dennis complains about statistical testing of earthquake predictions, but this is very necessary as a protection against charlatans, and anyone who's had the slightest involvement with the subject will no doubt be able to supply a few names.

Even the phrase "earthquake prediction research" at the moment promises too much too soon. The first stage is "earthquake precursor research". We need a lot more basic data about what precursors exist and what their properties are before we start jumping in with both feet and making predictions. This is why I'm impressed by a lot of what I've seen coming out of Japan in the last two years. These people are not making predictions at all (to the despair of the Japanese govt) but they are quietly assembling the data sets that may one day lead to prediction.

I really don't care what goes on things like the prediction success tables that were on this site before Dennis got them removed. It's so much bingo. What's needed right now is good data gathering and basic research. How it can be used will come later.

Incidentally, that Bayesian argument from Geller et al is flawed IMO. You could use it to prove that anything difficult is impossible.