The Role of Object-Oriented Metrics

P erhaps the most common concern of project managers who use or who are about to use object technology is for more measurement tools. I suspect that some of these people would kill for anything that could give them some kind of quantitative grasp on the software development process. There is, in fact, an extensive literature on software metrics, including much that pertains to object-oriented development. But surprisingly little of it is of direct use to actual projects. The publications that directly address object-oriented metrics often go back quite a while. An example is Barry Boehm's Software Engineering Economics (Prentice Hall, 1981), with its Cocomo cost-prediction model. Despite the existence of more recent works on the subject, Boehm's is still among the most practical sources of quantitative information and methodology. Metrics are not everything, of course. Lord Kelvin's famous observation is exaggerated: When you cannot measure, when you cannot express [what you are speaking about] in numbers, your knowledge is of a meager and unsatisfactory kind: You have scarcely, in your thoughts, advanced to the stage of a science. Much of math—and most of logic—is not quantitative, but we don't dismiss those elements of science as nonscientific. These considerations also put in perspective some of the comments published recently in Computer by Walter Tichy (" Should Computer Scientists Experiment More? " May 1998, pp. 32-40) on the need for more experimentation; his article was largely a plea for more quantitative data. I agree with Tichy's central argument: We need to submit our hypotheses to the test of experience. But when Tichy writes Zelkowitz and Wallace also surveyed journals in physics, psychology, and anthropology and again found much smaller percentages of unvalidated papers [that is, papers not supported by quantitative evaluation] than in computer science… I cannot help but think, " Physics, OK, but do we really want to take psychology as the paragon of how scientific computer science should be? " I don't think so. In an engineering discipline, we cannot tolerate the fuzziness that is almost inevitable in social sciences, in spite of all their numbers. If we are looking for rigor, the tools of mathematical logic and formal reasoning are crucial, even though they are not quantitative. Still, we need better quantitative tools. In this column I present a classification of software metrics and five basic rules for their application. The first rule of quantitative software evaluation is …