Program Verification, Defeasible Reasoning, and Two Views of Computer Science

Abstract

The current debate over the limits of program verification (Fetzer, 1988; Letters, 1989; Technical, 1989; Fetzer et al., 1990) seems to have its roots in an apparent fundamental philosophical difference concerning the methods and goals of computer science. On the one hand, there is the view that computer science is, as its name implies, a science, but more importantly, an empirical science in the sense which de-emphasizes pure mathematics or logic. This sense is meant to cover all and only those experimental disciplines included in the ‘natural’ and ‘social’ sciences. This view is expounded implicitly and explicitly in many current standard computer science texts, for example, by Thomas L. Naps et al.:

Perhaps nothing is as intrinsic to the scientific method as the formulation and testing of hypotheses to explain phenomena. This same process plays an integral role in the way computer scientists work. [Naps et al. (1989) p. 5]

This view is also exemplified by the curricula and attitudes of many current academic computer science departments, in which computer science is put forth as the science of problem solving using computers, and not as ‘mere computer programming’.