Information Perishing and Replenishment.
Abstract
This dissertation treats the dynamics of a decision maker's value of infromation. A notion, widely held by decision analysts but tenuously defined, is that the value of any specific information diminishes over time. This concept, termed information perishing, is rigorously defined and illustrated by the use of a Markov model. The main assertions of the first section are: (1) Information perishing is inevitable (not only for the Markov model of information but for any state of information described by a probability distribution); (2) For the Markov model the absolute value of the largest transient eigenvalue is an upper bound for the rate of information perishing; and (3) The rate of perishing is a decreasing function of time. A second model of the decision process recognizes that many decisions in real life are 'triggered' by events which may be described by some stochastic process. Without this uncertainty the decision maker could simply discount the value of information because of perishing and would reduce his problem to a static case; however, the uncertainty in timing forces consideration of optimal policies of information replenishment. Rules of optimality are developed for singly and multiple occurring decisions.
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 01, 1977
- Accession Number
- ADA046911
Entities
People
- Allen Frederick Grum
Organizations
- Stanford University