The increased industrial software metrics activity seen in the last 10 years may be regarded as a successful symptom of an expanding subject area. Yet, this industrial practice is not related in content to the increased academic and research activity. The metrics being practised are essentially the same basic metrics that were around in the late 1960s. In other words these are metrics based around LOC (or very similar) size counts, defect counts, and crude effort figures (in person months). The mass of academic metrics research has failed almost totally in terms of industrial penetration. This is a particularly devastating indictment given the inherently applied nature of the subject. Having been heavily involved ourselves in academic software metrics research the obvious temptation would be to criticise the short-sightedness of industrial software developers. Instead, we recognise that industrialists have accepted the simple metrics and rejected the esoteric for good reasons. They are easy to understand and relatively simple to collect. Since the esoteric alternatives are not obviously more valid there is little motivation to use them. We are therefore using industrial reality to fashion our research. One of our objectives is to develop metrics-based management decision support tools that build on the relatively simple metrics that we know are already being collected. These tools combine different aspects of software development and testing and enable managers to make many kinds of predictions, assessments and trade-offs during the software life-cycle, without any major new metrics overheads. Most of all our objective is to handle the key factors largely missing from the usual metrics models: uncertainty and combining different (often subjective) evidence. Traditional (largely regression-based) models are inadequate for this purpose. We have shown how the exciting technology of Bayesian nets (whose complexities are hidden to users) helps us meet our objective. The approach does, however, force managers to make explicit those assumptions which were previously hidden in their decision-making process. We regard this as a benefit rather than a drawback. The BBN approach is actually being used on real projects and is receiving highly favourable reviews. We believe it is an important way forward for metrics research.
To go back to our resources section click here.[papers/_private/copyright_notice.html]
Last modified: July 28, 1999.