While substantial progress has been made in understanding how and where strain accumulates on faults, enabling the identification of fault segments that are more likely to host earthquakes, the mechanical behavior of some key features of the earthquake rupture remains unclear. For instance, difficulties arise when evaluating the precise effect of fault geometry during earthquake rupture. While physics-based modeling offers a first, elegant, explanation for the main features of surface data, residuals are often interpreted solely as the results of the system's own complexity. In this presentation, I will present examples, from fault roughness to thrust mechanics, of how complexity can also emerge from "simple" physics-based models to help us understand key features of earthquake physics.