We can't reliably predict earthquakes. If verifiable predictions are possible, it will be easy to present them and give the predictions to the public.

I checked this with the UCLA lawyers when we were considering a prediction web page, which stalled for other reasons, and they said to go ahead and predict anything we honestly believed with appropriate disclaimers.

However, no methods with probability gains worth noting are yet documented, and thus none have been given to the public. Methods like M8 and its derivatives are moving in this direction, but are not yet impressive enough in their performance. The statistics are straight-forward when one is using information one understands.

So all that is lacking so far is some method with a sufficiently promising trial run.

John