Abstract. Our research focuses on computational techniques to understand the dynamics of space-time patterns in driven threshold systems, particularly on earthquake fault systems. We discuss the scientific and computational formulation of strategies for understanding such spacetime patterns, leading optimistically to earthquake forecasting and prediction. In particular, we describe pattern dynamics techniques that can be used to relate the observable processes on earthquake fault systems to the fundamentally unobservable dynamical processes. To illustrate our results, we study the emergent modes of the earthquake fault system in southern California, both with models (Virtual California) and with data.