The two-layer quasi-geostrophic model of mid-latitude dynamics introduced by Norman Phillips in 1951 has provided generations of meteorology students with a simple picture of baroclinic instability and of large-scale evolution in midlatitudes. It also provided Norm with the framework with which to generate the first statistically steady model of the general circulation incorporating baroclinic instability, i.e., the first atmospheric GCM, in 1956. In this talk, I will touch on what we have learned from the two-layer model over the last half century, and explain why i have devoted a substantial share of my research to this model. I will argue that the model's limitations, such as the critical shear for linear instability that disappears as one adds more vertical resolution, are exaggerated by the linear theory. I will also argue that the two-layer model continues to play an important role in our field, and that, in fact, we will need to understand the statistically steady states of the two-layer model more fully if we are to understand the general circulation of the atmosphere.