The nature of earthquake prediction

Earthquake prediction is inherently statistical. Although some people continue to think of earthquake prediction as the specification of the time, place, and magnitude of a future earthquake, it has been clear for at least two decades that this is an unrealistic and unreasonable definition. The reality is that earthquake prediction starts from long-term forecasts of place and magnitude, with very approximate time constraints, and progresses, at least in principle, to a gradual narrowing of the time window as data and understanding permit. The analogy to catching a rabbit in an overgrown confined field may be appropriate. You do not just start out looking for the rabbit, you instead build a fence dividing the field in two and then decide which half the rabbit is in, thereby gaining one bit of information. You iterate this process until you have located the rabbit “close enough for practical purposes.” This is approximately how earthquake prediction proceeds in the real world, with time and position along a fault comprising the two dimensions of the search. (I assume here that we are considering for the moment only large earthquakes, that is those capable of inflicting serious damage on a regional scale; in California this means events of about M 6.7 and larger.) This more realistic perspective on the problem lays to rest the “red herring” that “earthquake predictions might do more harm than earthquakes.” These imaginary concerns are predicated on the fantasy of a prediction that precisely specifies time, place, and magnitude; in the real world a progression of probabilities that narrows the space-time window in small steps clearly carries no such threat.