The general dependence on large scale systems together with rapidly changing technology require predictive models of the performance of complex systems in order to be able to judge in advance the functionality and safety of new system concepts. Complex systems including human actors, however, cannot be modelled by quantitative, deterministic models and causal models in terms of objects and events have typically been adopted. The paper presents a discussion of several basic difficulties with this approach. Definition of human error during supervisory tasks is becoming increasingly difficult, and post-hoc identification of causes of an accident depends on a pragmatic stop-rule for the termination of the analysis. Empirical verification of the design of a complex system, likewise, raises the question of stop-rules for adjusting the experimental conditions. In addition, there is a need for the development of predictive models of human performance in complex systems. For intellectual, creative tasks, they cannot be in terms of procedural task descriptions but have to be based on higher level 'first principles' which take into account goal oriented, adaptive human characteristics. This approach is discussed for models of individual human actors as well as for models of organizations managing large scale systems.
|Number of pages||13|
|Publication status||Published - 1989|