SOME PRINCIPLES FOR DESIGN OF DECISION SYSTEMS: A REVIEW OF THE FINAL PHASE OF RESEARCH ON A COMMAND-CONTROL SYSTEM SIMULATION.
Abstract
The chief objective of the present research was to anticipate a number of issues which would arise if an automated aid to decision making were actually implemented. Questions asked included: (1) What happens to system performance if probabilistic information is reduced to an all-or-none form at some point in processing. (2) Can a hierarchical (specialist-nonspecialist) system use limited resources effectively to gather predictive data. (3) Can a system in which aggregation of predictive information is automated benefit from a manual supplement (to handle unanticipated data). Results summarized in the nine principles suggest that (1) all-none transformation of probabilistic data can seriously degrade system performance, especially if system response is in any way dependent upon likelihood of alternative states (but there are several important exceptions to this rule); (2) potential deficiencies in allocation of resources by the system should be guarded against in future system designs; (3) an automated aggregation design can be enhanced by a manual supplement to deal with unanticipated data; (4) a voice communication capability does not offset system performance deficit attributable to degradation of other processing modes; and (5) the value of experience in certain aspects of human decision performance may not extend far beyond the specific conditions under which the experience was acquired. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1968
- Accession Number
- AD0684548
Entities
People
- Charles F. Gettys
- William C. Howell
Organizations
- Ohio State University