COBOL '74
Paul Oliver · 1975
Gerald Weinberg, in his book The Psychology of Computer Programming suggests that we cannot really measure the goodness of programs on an absolute scale, and that we generally cannot even measure them on a relative scale. There is indeed evidence that rapid quantification of software quality is not really feasible, because simple formulas can often be misleading and hence not very credible. One could, for example, measure program complexity in terms of the fraction of program statements which are branch statements. Consider however two programs, one with 10 decision points leading to 10 different tasks, each consisting of one instruction, and one with 3 decision points leading to 3 tasks, each consisting of 30 instructions. The "complexity" rating of the first would be 0.50, while that of the second would be 0.09. Surely two such programs would not differ so greatly in complexity.