I like to think of how much strain is accumulated on a fault (which is related to stress, but not the same thing). We do know this well for a fair number of faults-especially the San Andreas. Let's take the central San Andreas in Carizzo (sp?) plain. The last earthquake there was 1857, and the slip was 9 or 10 meters. It is slipping at 3.3 or 3.5 cm/yr-so 4.8 meters of strain has accumulated. It would be a long time before 9 or 10 meters are accumulated, but despite evidence that the last couple EQs were 9 or 10 meters (Kerry Sieh was involved in this work), I would think that an earthquake might be able to propagate into this area from the edges and 5 meters of slip is nothing to sneeze at. I personally think that we may be able to predict quakes in some cases-maybe by combining tracking in detail occurrence of seismicity in 3D in near real time, modeled against continuous GPS data in special ways..with detailed knowledge of the deep 3D structure of faults (from work with microearthquakes.....).....Chris