Thanks for the compliment, Russel. No, I do not think there are ANY precursory indicators or predictive models that have any chance of providing useful earthquake predictions - other than the case of phenomena, such as the foreshock sequence mentioned in my post, that occur in an area of known major seismicity and are unmistakably well beyond any ordinary fluctuations in the phenomenon being observed. Even in those cases, the likelihood of a false alarm is great, but precautionary measures are reasonable. Generally, any useful prediction method must have a very low alarm rate.

I base my low opinion of the likelihood of succesful (useful) deterministic earthquake prediction on the chaotic nature of the nucleation process. Many, likely most, seismologists have concluded that this process is chaotic in nature and is thus INHERENTLY unpredictable. As one seismologist has stated, even the earthquake doesn't know it's going to happen until it does. This opinion is not countered by the example above. In that case, the obvious large static stress (indicated by the major, damaging foreshocks), combined with the dynamic stress from the shocks themselves, combine to produce a high likelihood of a major event. I guess a more accurate description of my belief is that short-term prediction of a small subset of major earthquakes IS possible, but that those predictions will have their usefulness diminished by a significantly high false-alarm rate.

Mike Williams in Arroyo Grande, CA