Much has been written regarding the similarities and differences between the SAF and the North Anatolian Fault. The two seem like twins until closer inspection. Here is a link to a good article from Scientific American.com:

http://www.sciam.com/article.cfm?articleID=000C0432-2EBC-1C74-9B81809EC588EF21

To me, the occurrence of any two quakes along the SAF does not suggest a pattern of any sort (i.e. - no occurrences proceeding "in an orderly fashion"). The relatively high occurrence rate for M5.0+ quakes in the Bay Area in the 60 or 70 (?) years preceding 1906 is usually considered a normal part of the quake-aftershocks-quiescence(shadow)-slow increase-quake cycle. Though it certainly isn't unreasonable to consider stress transfer from 1857 to be involved.

It is important to note that nowhere on Earth do we have good observational understanding of even one full earthquake cycle (if one exists), and that includes paleoseismological data. Seismologist Seth Stein of Northwestern University wrote a very interesting Opinion piece in the latest issue of Seismological Research Letters. He points out the enormous strides seismology has made in understanding much of the earthquake process, yet seismologists still do not understand "the most basic question about earthquake recurrence: whether it's time-dependent or time independent." In fact, much in that article bears on Todd's questions. I have the hard-copy of Stein's article, and could probably access the electronic version. I'll check it out.

The apparent undisciplined bahavior of earthquake occurrence and re-occurrence is one of many reasons I am such a prediction skeptic.

(disclaimer: I am not a professional seismologist)

Michael Williams in Arroyo Grande, CA
"Quantify it before it disappears" - Stephen Jay Gould