Evolution, feedback and software technology

Despite many technical innovations in methods and techniques for software development and in the provision of tool support the current industrial software development process is still considered inadequate. Systems that are delivered are often inadequate in their function, performance or quality. There are still far too many delayed or failed projects. Cost overruns and delivery delays are the norm. Major efforts at process improvement over the past years have not produced major widespread gains. The paper discusses a conjecture that the difficulty in achieving major improvement may be due to negative feedback controls in the global software evolution process and introduces a project, FEAST, that will seek to validate the conjecture in theory and practice. If it can be shown to be correct and the negative feedback stability property exploitable this should lead to methods and tools to facilitate its exploitation in process improvement activity.