i think we need to separate evaluation from whatever we call the "value" judgement.

We have the task of saying "hit or miss" which is straightforward; a quake fits the prediction or not, unless we allow near misses?

But value is a separate issue, involving probability of success for which there are several ways of calculating; Monte Carlo, Jones', simple time-based chance and chi-square tests.

Of these, Jones works best for single predictions, the others are for a large series (or history) of predictions.

Do we agree?

Roger