ORDER OF MAGNITUDE REASONING

Order of magnitude reasoning is pervasive. It copes with incomplete quantitative information and reduces the burden of calculations. This paper explores the foundations of this form of commonsense reasoning. Intuitively such reasoning is like using a coarse balance which weights quantities with a variable level of precision. The coarse equilibrium is captured by order of magnitude equations, and the precision levels are given by order of magnitude scales. The system, Estimates, introduced to solve such equations, captures some basic engineering skills. Two tasks demonstrate the leverage it provides: the diagnosis of analog circuits and the estimation of the acidity of chemical solutions.