Process improvement-the way forward

The software process constitutes a feedback system with both positive and negative feedback, though what constitutes positive and what negative feedback requires clarification. It is plausible that, as a feedback system, the process will possess the external stability property of such systems (it is, in fact, most unlikely not to be true). The involvement of people, however, in most, if not all, of the feedback loops as observers, communicators, decision-makers, evaluators and implementors, raises questions as to whether the system dynamics, its responses to changes in the forward path or feedback mechanisms, can be rigorously and systematically modelled, predicted or controlled. The progress made in system dynamics and in control theory applied to economic and organisation modelling suggests that limited progress can be achieved. There are, in fact, grounds for believing that the software process may be more amenable to control because of moderating and stabilising influences of both the application in its domain of operation and the program code. The FEAST investigation is intended to explore and exploit this insight. What seems clear is that for major process improvement, the total global process, its mechanisms and characteristics must be considered whenever local improvements are investigated. Merely changing forward path technology can only produce limited benefit