Other than that, I have no particular idea as to how we might predict earthquakes. Instead, I focus on how to test other people's earthquake prediction ideas. Earthquakes have some peculiar statistical behavior, for instance they cluster together in space and time, and if we aren't careful to include these pecularities in our tests then the tests will be wrong. It turns out that not only will the tests be wrong, but they will tend to be too optimistic and could get us to believe in ideas that we shouldn't believe in.
Here is an example of a test that overlaps with my interest in education: "Do Lost Pet Ads Predict Earthquakes?"