Myths and Methods: Is There a Scientific Basis for Y2K Inspections?

Dr. D.L. Parnas


Abstract

Although it is possible to use scientifically based mathematical models in the analysis of software, most programmers rely on their intuitive understanding instead. With complex programs, our intuition is often inadequate and we overlook serious faults. Although most organisations have introduced a systematic management process for Y2K software inspection and repairs, the actual analysis of the software relies on intuitive "eyeballing" of the code. Intuition has given rise to some folklore about the Y2K problem that has no scientific basis. We need to base our analysis on sound science and mathematics. Moreover, sound models suggest procedures that are more systematic and more trustworthy than the intuitive ones. This paper will discuss both some of the unscientific myths and sound inspection methods. If we base our program analysis on unsound methods, we are "building on sand" and what we do will fail.