I posted a few months ago here saying that they should increase the probablity of a large earthquake on the two segments where they have assigned very low probability for an earthquake in the 30 years following that report (boy, is the previous sentence bad writing!). Let's try again: it was viewed as very unlikely that the Carrizo Plain and San Francisco-northern segments of the San Andreas would fail in the following 30 years. The logic was that there were very large slips on these segments in 1857 and 1906. I suggested that they did not know enough about how things worked to assign such low probabilities (was it 3%?), or such a high probablity to Parkfield. The abstract that Don linked I read to say that earthquakes on the Carrizo segment are more frequent than they thought. Almost 5 m of slip has accumulated on that segment since 1857, nothing to sneeze at.....

As always, I'm not taking the time to look all of this up and am relying on my increasingly-shaky long-term memory....but I should be basically correct. Chris