H IMPACT O F ; INFORMATION ON DECISIONS: E /COMMAND ~ld-CONTROL SYSTEM EVALUATION .-
Marvin S. Cohen, and Anthony N.S. Freeling
Sponsored by Defense Advanced Research Projects Agency Under Contract M D A 9 0 3 - 8 0 - C - 0 1 9 4 DARPA Order No. 3 8 3 1
Under Subcontract from Decisions and Designs, Inc.
Februarv 198 1 .
THE VIEWS AND CONCLUSIONS CONTAINED IN THIS DOCUMENT ARE THOSE OF THE AUTHOR AND SHOULD NOT BE INTERPRETED AS NECESSARILY REPRESENTING THE OFFICIAL POLICIES. EITHER EXPRESSED OR IMPLIED, OF THE DEFENSE ADVANCED RESEARCH PROJECTS AGENCY OR THE UNITED STATES GOVERNMENT.
L J _ _ ~ ~
DECISION SCIENCE CONSORTIUM, INC.
7700 Leesburg Pike, Suite 421 Falls.@hurch,-Virginia 22043(703) ---- 10 -- ---- --- -790-05 - --
In work reported here, Decision Science Consortium, Inc. (DSC) has examined the application of the decision analytic concept of value of information to the design of information systems. Automated data base systems play an increasingly prominent role in a variety of areas - including Command, Control, Communications, 3 and Intelligence (C I), Indications and Warning (I & W), and business management. However, a basic problem of data base design has not been solved: what information should be included in the system, and what subset of that information should be presented to a user, so as to best achieve the objectives of the relevant organization? A common characteristic of systems in current use is that they often provide vast quantities of partially relevant data, while failing to identify the information which the decision maker actually needs to solve his problem. Current evaluation techniques for information systems appear to bypass this problem altogether. Evaluation in terms of dataprocessing parameters, like channel capacity or memory size, ignores the ultimate objectives of the system and seems to assume, simply, that more information is better. Direct assessment of information quality, in terms of such attributes as relevance and accuracy, fails to ensure that the actual impact of information on decisions (hence, on ultimate objectives) is considered. Multiattribute utility models similarly have not explicitly required consideration of how information is used in decision making. The concept of Value of Information (VOI) implies that information has value to the extent that it can alter decision and improve payoffs. However, the application of VOI techniques', as they now stand, to information systems is prohibitively complex. These -_ techniques presuppose a highly structured decision problem, in which information, the uncertainties to which it pertains, and
the options available to the decision maker are-all specified in detail. Moreover, they assume that the decision maker will behave optimally in the light of the information he receives. Complex, multipurpose information systems, on the other hand, are expected to operate in a variety of environments, some of which cannot be predicted in advance. And, they must serve users who cannot always conform to established normative ideals. The aim of the present work is to devise modifications of standard VOI techniques which make them simple enough and realistic enough to apply to information system design, while retaining a basic reference to the impact of information on decisions. In doing so, we proceed by steps of (roughly) increasing simplification: (1) Explicit reference to all the information in a data base can be omitted from a VOI analysis if acts are modeled as events (Brown, 1975). Assessment heuristics are presented which facilitate modeling acts as events in this (or any other) context. By means of these heuristics, a system designer can examine tradeoffs between the information value of a system and its usability....
Please join StudyMode to read the full document