The learning curve: a new perspective

Despite the popularity of the “learning” curve, it should be called the “forecasting” curve. Given an industrial process, the traditional learning curve forecasts how costs are expected to decline (or other performance measures improve) in the future. Forecasting is certainly very useful. Nevertheless, the previous literature is notably mute on how the improvement or learning occurs. Without that information about how the process operates, the traditional learning curve cannot improve the rate of learning. This paper transforms the curve so that it not merely forecasts but rather actively creates the learning and knowledge. By extending work of the authors (Zangwill and Kantor, 1998), this paper constructs a new theoretical framework for learning and making improvements based upon learning cycles. This approach allows what is learned in one period to be intelligently applied to the next, improving the rate of learning right as production is occurring. Previous approaches could not do this. Because it should quickly increase the rate of learning, this approach might have important industrial applications.