Our high time resolution observations of individual pulses from the Crab pulsar show that the main pulse and interpulse differ in temporal behavior, spectral behavior, polarization and dispersion. The main pulse properties are consistent with one current model of pulsar radio emission, namely, soliton collapse in strong plasma turbulence. The high-frequency interpulse is quite another story. Its dynamic spectrum cannot easily be explained by any current emission model; its excess dispersion must come from propagation through the stars magnetosphere. We suspect the high-frequency interpulse does not follow the ``standard model, but rather comes from some unexpected region within the stars magnetosphere. Similar observations of other pulsars will reveal whether the radio emission mechanisms operating in the Crab pulsar are unique to that star, or can be identified in the general population.