Ten criticisms of PARRY
Much of the Artificial Intelligence community is now aware of a computer simulation of paranoid processes developed by the Colby group at Stanford. The model (called PARRY) has been available for interviewing on the ARPA network and thousands of interviews have been conducted with several versions of the model. During the long period of development of the model, we have been aware of the limitations of various alternative programming approaches to designing an algorithm capable of conducting useful non-trivial dialogue in natural language. Colleagues, associates, and students have volunteered a number of criticisms of the model. Since criticisms can be endless, I shall restrict the discussion to only those which we consider serious, reasoned, and well-founded.