Any simplified description of a brain system that survives 20 years is bound to have some problems. Twenty years of research should have added details that were not known when the now classical description of the basal ganglia circuitry was new. Some examples of such new information about the parts of the original model include: All of these details make it difficult to cling to the model as it was presented, but which of the new information will support a new synthesis? Do we have a coherent replacement for those powerfully heuristic ideas put forward by Albin Young and Penny in 1989? Perhaps in the quantitative anatomical detail there is a plan that may support a new theory. If all of the cortex is not accessible to every striatal cell. If dopamine influences many spines rather than the one it contacts. If we stop looking for a mechanical control system for movement in the basal ganglia and begin to investigate the predictive properties of 'reinforcement learning'. Then maybe a new functional model will emerge that is less tied to anatomical and physiological simplification, but perhaps will still allow us to engineer therapeutic strategies to help patients with movement disorders.