A fundamental question in visualization is what constitutes a “good” visualization. A related question whether one visualization is better than another. In general, these hard questions are addressed by running user studies. However, evaluating visualizations with user studies a posteriori, in an inductive approach, is neither sufficient nor efficient. Ideally, we would like to have models that not only define what a good visualization is but also tell us how to construct them. Historically, general theories have been born from elimination and/or unification of competing and complementary theories that have emerged from specific domains. Clearly we need more theories of this kind in visualization. In this panel we will discuss example theories of visualization and ponder how they relate to one another.