I think some quakes are predictable and many aren't (i think that earthquakes on hidden thrust faults in California as of now are not predictable).
re; Crustal deformation site:I have not heard of many people using the data.
The site even has a statement by the USGS saying the data is not to be used for engineering etc purposes.
Interestingly - a day or two before the last 6 in Parkfield(2004) I saw a large increase on at least two of the tensor strain and dilameter sites at the Parkfield sites. I did not save the screen or the data and after the earthquake I went back to the site and the strain changes from the earthquake had changed the graph readings. I also had seen somewhat similar but less strain increases before that that were not followed by an earthquake.
I have done web searches occasionally to see if I could find the readings from a few days before the 2004 earthquake but I have not been able to find them. I have not approached the USGS in Menlo Park to see if they have the records available.
Steve
p.s.- near Chalome a little to the south of Parkfield they have discovered slow seismic tremors.