Glen....you would have to do the the area-magnitude calculation: these areas are almost oceanic spreading centers that are different mainly because they are drowned in sediment. So, the crust is thin and so you need an extremely long rupture to get high magnitude. I suspect if you use 10 km as width you might need 1000+ km for length to come up with M8.7. I think M8 is also unlikely.
There are large stepovers between San Andreas through Brawley to next one (Superstition hills?) and there is Imperial somewhere and then Cerro Prieto. I suspect this mould be a M7.5 or 7.7. But, if all were primed to go, one could rupture the next and it could get through that. But, I think Imperial broke in 1940(?) so probably is not ready to go again.
There was some talk about "wall to wall" at the SCEC meeting...mainly outside of formal talks. Wall to wall is starting at Coachella and breaking all the way to Parkfield. Related to that is whether the Coachella could break through the area near Palm Springs: I think the Banning and Mission Ridge or is it Mission Creek?....at least one of these faults is moderately dipping. The kim Olsen et al simulation of 3D ground motion uses such a rupture through and on farther north. So, the likelihood of such a rupture is pretty critical: if the prediction of these simulation for ground motion in LA is correct than over certain areas tall building will collapse..I hear. Another seismologist (CN) pointed out that the similations are only for periods of a couple s because the shorter the period (the higher the frequency) the more computing power they need. They are already using 2000 processors and they are going to increase this to 10,000 processors. The higher frequency content that is not simulated may destructively interfere and knock down the amplitudes.
John V.: technical enough and important enough that a comment would be good here...
Chris