PSYCHOLOGY AND THE USER INTERFACE: SCIENCE IS SOFT AT THE FRONTIER.

It is elementary in the history of science that one cannot legislate the quality of the conceptual and empirical content of science merely by legislating the methodological form. In fact, if history is any gauge, a priori limitations on acceptable methods usually have an undermining effect on conceptual and empirical quality. Newell and Card are mistaken in their attempt to confine the psychology of human-computer interaction. Their view of hard science is arbitrary and in particular has been a fairly well-documented failure in providing real leverage in interface design, conceptually and empirically. Their view of AI as hard is similarly inaccurate, as evidenced by the subfield of advisory expert systems.