The most unshakeable hypothesis has been that all we can easily learn about future mainshocks from an earthquake is that aftershocks will, on average, follow an Omori law in timing and Gutenberg-Richter distribution in magnitude .

We test our ideas of what the earthquake patterns mean, and we have vast catalogs to test for subtle patterns, and not much beyond the occurrence of aftershocks has emerged. Seismicity patterns are not likely to tell us when the next earthquake will strike.

Maybe once we've first observed several earthquake cycles on a given fault, we can do more, but the Parkfield experiment is an example that it is easy to overestimate our prediction prowess.