ABSTRACT. A model is an explicit simplified working representation of a physical process. Modeling is the means by which we transform the whirring, buzzing, confusion of data into a coherent story. In this era, models usually take the form of computer code. Despite their essential role, we do not yet have a stable foundation of shared consensus codes to describe the workings of early vision. In this talk I will share some lessons that I have learned about modeling of vision, and offer some ideas about how we might move from a predominantly empirical science to one based on an evolving infrastructure of theory.