Half-Life Theory of Learning Curves for System Performance Analysis

Learning curves are used extensively in business, science, technology, engineering, and industry to predict system performance over time. Most of the early development and applications have been in the area of production engineering. Over the past several decades, there has been an increasing interest in the behavior of learning curves. This paper introduces the concept of half-life of learning curves as a predictive measure of system performance, which is an intrinsic indicator of the system's resilience. Half-life is the amount of time it takes for a quantity to diminish to half of its original size through natural processes. The common application of half-life is in natural sciences. The longer the half-life of a substance, the more stable it is. Consequently, the more resilient it is. This approach adds another perspective to the large body of literature on learning curves. Derivation of the half-life equations of learning curves can reveal more about the properties of the various curves. This paper presents half-life derivations for some of the classical learning curve models available in the literature.