Good morning Don...heavy rain in eastern Suburbia from Ivan, and tomorrow early I fly to Palm Springs for the SCEC meeting...where there could be rain from remnants of Javier...Hopefully John answers this also, but I'll take a shot. Great earthquakes over M8 are more likely on subduction zones...which are megathrust faults, than on transform faults....because earthquakes Moment Magnitude is related to area of the fault, and a gently-dipping (low-angle) fault has a much bigger area than a steep fault...because strike-slip earthquakes in continental areas only occur in the brittle part of the crust....and so extend down only 15 or 20 km. There may be something to what you say about obstructions on faults....they are called "asperities"...places where the faults are locked a little harder than elsewhere...which eventually become the hypocenters of earthquakes. Actually, what I just wrote does not make sense, because if it is locked harder there, it should fail somewhere weaker and just rupture through the asperity...John? Anyway...they find that in many areas there are small indentical earthquakes....say. M3 or M4, (or smaller), where the same patch of fault keeps rupturing. But, eventually as the blocks on either side move along, the geometry should change and the asperity should change...if it is a little bump or something, this should not take too long (at least geologically), but is it is some kind of rock that is strong, or a large-scale (kms or 10s of km) irregularity, it could take 10s of thousands of years...
Chris