We review our high-time-resolution radio observations of the Crab pulsar and compare our data to a variety of models for the emission physics. The Main Pulse and the Low Frequency Interpulse come from regions somewhere in the high-altitude emission zones (caustics) that also produce pulsed X-ray and γ -ray emission. Although no emission model can fully explain these two components, the most likely models suggest they arise from a combination of beam-driven instabilities, coherent charge bunching and strong electromagnetic turbulence. Because the radio power fluctuates on a wide range of time scales, we know the emission zones are patchy and dynamic. It is tempting to invoke unsteady pair creation in high-altitude gaps as the source of the variability, but current pair cascade models cannot explain the densities required by any of the likely models. It is harder to account for the mysterious High Frequency Interpulse. We understand neither its origin within the magnetosphere nor the striking emission bands in its dynamic spectrum. The most promising models are based on analogies with solar zebra bands, but they require unusual plasma structures which are not part of our standard picture of the magnetosphere. We argue that radio observations can reveal much about the upper magnetosphere, but work is required before the models can address all of the data.