Analyzes the methodology of computer simulation in the social sciences and its ability to manage complexly organized systems. The "in principle" arguments advanced in support of the method have been insufficiently tempered by honest appraisal of past "in practice" experiences. A continuation of the existing trends could harm both the method and the processes of problemsolving in a social context. An initial problem is the development of an appraisal function--i.e., a series of questions or criteria against which a computer simulation may be judged. The experiences from two empirical contexts are related in efforts to construct a so-called policy-assisting class of simulation models. Deficiencies in the appraisal included the vastly different orientations of participants interviewed, salesmanship, and various technical and theoretical shortcomings. 47 pp.