Why Is It so Hard to Nd Feedback Control in Software Processes? Invited Presentation

Early data on the phenomenology of software system evolution suggest that such evolution involves and is, to some extent, governed by feedback. This feedback may take the form of information fed back to individuals or groups as a form of learning from experience or may take the form of observation and data that are used to control some aspect of the process. For the moment, we shall put the former to one side and concentrate only on feedback to explicit control mechanisms. Initial investigations, using a basic model for feedback control, have exposed a variety of reasons why software processes are not amenable to classical feedback control: software processes are design, not production processes; control-directed process changes tend to be step functions, not regulatory ones, and are often as creative as the processes they control; and system development and evolution processes are still immature with little theory to guide the design and application of regulation control mechanisms. Despite these limitations, we have found promising examples of feedback control and, on the basis of more recent phenomenological evidence, believe this area of research to be critically important and vital to understanding and controlling the development and evolution of software systems and improvement of software processes.