We present a novel methodology for calculating the improvements obtained in successive versions of biomedical ontologies. The theory takes into account changes both in reality itself and in our understanding of this reality. The successful application of the theory rests on the willingness of ontology authors to document changes they make by following a number of simple rules. The theory provides a pathway by which ontology authoring can become a science rather than an art, following principles analogous to those that have fostered the growth of modern evidence-based medicine. Although in this paper we focus on ontologies, the methodology can be generalized to other sorts of terminology-based artifacts, including Electronic Patient Records.