Modeling Uncertainty in a Decision Problem by Externalizing Information

Abstract

This paper deals with decision problems under uncertainty. The solution of a decision problem involves observation, processing, and modeling of statistical data in order to quantify the uncertainty. Better data measurement and estimation of uncertainty add more consistency to the solution of a decision problem. The paper proposes a new way of predicting the Bayesian- Nash equilibrium which uses information sources to measure new information received by information consumers. Thus, the estimation of uncertainty is based on a more solid mathematical foundation, needed (as in the case of artificial intelligence) to produce logical inferences. From another perspective, the externalization of information helps the software designers to produce better software architectures for decision support systems. An theoretical example illustrates a market situation with a small number of firms, each firm’s output being likely to have a large impact on the market price.