very, very interesting. that must have taken (and probably will still take) a lot of computer modeling to use the accelerometers just right so as to not pay attention to bad readings and to identify actual trends.
Ars Technica has an article on the Quake-Catcher Network (http://qcn.ucr.edu/index.html). Having an IBM ThinkPad myself, it is a fascinating prospect to use the computer's accelerometer to create an early warning system for earthquakes. Hopefully this project will get off of the ground.