First, Canie, congratulations on the speedy recovery of the Forum!

On the off-chance that Cal's chosen title for this thread was an attempt to refute my posts (which were narrowly focused on the subject of the value of prediction), I would like to point out that the operative information in Cal's post (the death toll figures, since those are where the effects of a successful prediction would be manifested) seem to support my position. These figures, which are fully in line with those from other sources, out for years, support an absolutely worst-case-scenario death toll of barely one-tenth of one percent of the population in the meizoseismal region. Since evacuation of the entire region would be impossible, a successful prediction, then, might result in a significant reduction in deaths in absolute numbers, but a tiny one in relative numbers. Probabilistic prediction, the type pursued by the oft-maligned seismologists, meanwhile, would have saved vastly more lives, in both absolute and relative terms.

Cal provides additional supportive evidence (for my position) in her post which references the death toll of 33 in the Mw 6.7 Northridge quake. In December of 2003 an Mw 6.5 quake in Bam, Iran killed 43,000 people. Now, I concede that the two quakes aren't perfect matches (Northridge, though bigger, did have a good deal of its energy directed away from the largest concentration of population), but you have to admit that that is one heck of a contrast - 33 versus 43,000! I'll leave it the reader to figure out what saved all those lives in Northridge. But one hint is that it wasn't deterministic (short term) prediction.

Were there not very good reasons to believe deterministic prediction to be inherently impossible, or, had it at least shown some promise over the decades it has been pursued, I would certainly support a greater effort in that direction. In any event, even if successful, it would not be as important as other steps society takes to reduce earthquake death tolls. On the whole, I think the modest investment currently being made in the field of deterministic prediction is probably appropriate.

Mike Williams in Arroyo Grande, CA