A knowledge process perspective on the improvement of software processes

Sammanfattning: The number of software-based products is increasing, which is why the importance of an efficient software development process is also increasing, from a competitive advantage perspective. For years, software development organizations have been plagued by poor product quality and high development costs. Studies in software process improvement (SPI) have concerned themselves with improving the software process as a way of improving software product quality and reducing development costs. The basic tenet of SPI is that a relationship exists between process maturity and product quality. The departure point of this thesis is the fact that, despite the best efforts of SPI programs, organizations are still struggling with software product quality and runaway budgets. The development of software is a problem-solving activity and this thesis argues that the knowledge process perspective on software development is only addressed to a limited degree in existing SPI methods and frameworks. The goal of this thesis is to add theoretical knowledge to the field of SPI by using theories from IS (information systems) and KM (knowledge management) in the context of SPI, and to empirically study how software development processes are being, or can be, improved using a knowledge process perspective, hence a KM approach to SPI. In order to meet the research goals, six studies employing case and action research were performed at two major high-tech companies. The results of the different studies are presented in six different articles. In particular, this thesis studied how development methods and knowledge networks have supported knowledge processes such as knowledge creation and sharing in large software organizations. During the research, data was collected by means of interviews and observations. This thesis argues, as a complement to prescriptive improvement models such as the CMM, which focuses on software processes, that software organizations should pay more attention to understanding and improving the knowledge processes of software development. The research conducted recognises the various levels of innovation in relation to product lifecycles. The research results suggest that the focus of a method on explicit knowledge and best practice makes it suitable for supporting knowledge creation and sharing when mature products are being developed. Furthermore, development methods support knowledge creation and sharing in large organizations by means of their ability to provide a common language as well as shared mental models. Knowledge networks, for their part, are better suited to supporting knowledge creation and sharing when radically new products are being developed. The network has the characteristics enabling it to integrate knowledge originating from different units, which is important during problem solving.