Systems and Information

Abstract

Information theory is essentially a method for retaining some knowledge of cause-effect relations when the causes, and their effects, become so many that detailed knowledge of each pair is abandoned, while an over-all check is retained that at least the number of causes is sufficient to account for the number of effects. Communication theory has been severely restricted by its original concentration on the ergodic and stationary. Adaptive systems, however, by leaving bad ways of behaving and changing permanently to other (and better) ways of behaving are necessarily nonergodic and nonstationary. The methods of uncertainty analysis (McGill and Garner) seem appropriate here. These methods give a new precision to such questions as: What is a 'system'?-a question that must be answered with precision if a proper theory and dynamics of systems in general is to be built. When systems become complex, their theory is practically that of how to simplify them.