AND

This paper reviews work completed oil a value-judgment-based tactical air command system intended to dispatch missions from a limited supply in response to requests for immediate close air support. A Judged Utility Decision Generator (JUDGE) assumes tliat value judg-ments can be made explicitly and in real time by appropriately trained per-sonnel, and that decision systems hould maximize xpected utility. Its inputs include demand forecasts, numbers of aircraft available, and turnaround time distributions. As each request is received, JUDGE makes a dispatching decision based on the judged utility of destroying the target named in the request, kill probability data, number of sorties remaining before resupply of aircraft, and the time. In a war-game situation using experienced military subjects, JUDGE was considerably superior to a simulated current system in amount of expected utility gained. Concluding sections of the paper identify characteristics of judgment-based command systems, and discuss logical im-plications of such characteristics for system evaluation. Algorithms for making decisions in real situations are usually extremely complicated. The search for such algorithms defines much of the work of operations researchers, ystems analysts, and similar experts. But that search is often unsuccessful; many real decision tasks are intractable to full formal analysis. The designers of decision-making systems intended to cope with such tasks usually allocate the functions of information processing and decision making in such systems to men, presumably hoping that human intuition can deal effectively with problems too com