Roger, I know this is not very helpful if I don't look for the references, but I am sure there are older papers on this; I've heard that the larger the earthquake the longer until the next one. It even has a name, which I think may be "slip predictable"(?). The other model was "time predictable", amd was used for the unsuccessful Parkfield prediction. But, Weldon, Fumal, Scharer and others work on the San Andreas ar Wrightwood, and many people's work in the Mojave demonstrated rather well that neither of those models are applicable. John Vidale posted here that he thinks that it is more likely that there will be another quake in an area that just had one, but it was unclear if he was talking about the same part of the same fault, or nearby faults. Maybe he will clarify.
Chris