Foundations and experiments in software science

Abstract

A number of papers have appeared on the subject of software science; claiming the existence of laws relating the size of a program and the number of operands and operators used. The pre-eminent theory was developed by Halstead in 1972. The thesis work focuses on the examination of Halstead's theory; with an emphasis on his fundamental assumptions. In particular, the length estimator was analyzed to determine why it yields such a high variance; the theoretical foundations of software science have been extended to improve the applicability of the critical length estimator. This elaboration of the basic theory will result in guidelines for the creation of counting rules applicable to specific classes of programs, so that it is possible to determine both when and how software science can be applied in practice.