A Delphi study of the influences on innovation adoption and process evolution in a large open source project: the case of Debian

The Debian Project is possibly the largest FLOSS project, but its processes have not scaled in relation to its growth and complexity. Many aspects of daily tasks require the volunteer contributors to expend time on repetitive, minimallyintegrated tasks, leaving less energy for progress and innovation. Techniques that support collaboration and foster efficiency exist, but they are not being adopted as readily as could be. As a cultural insider and active participant in the Debian Project, I sought to bridge this gap through the identification of the influences that shape the adoption behaviour of contributors to the project. I conducted a Delphi study with a panel of 21 carefully selected Debian Contributors, and we found 24 influences, which are supported by numerous arguments from the participants’ study. We established and defined the role and scope of each influence, explored its facets, and argued as to its effects. For instance, we found that (a) adopters require multiple perspectives of an innovation, as well as active marketing involving their peers, (b) perceptions of elegance and familiarity are closely related, and considerably affect acceptance of a new approach, (c) consensus depends on drivers and executive decisions in due time, and (d) standardisation needs to define interfaces, not processes. From the findings, I crystallised 19 implications for practice, which can help designers and diffusers optimise the adoption of tools and techniques in the Debian Project. I postulate these will help increase competition between ideas, boost efficiency and collaboration, and improve scalability of the project’s processes. This was the first application of the Delphi technique in a FLOSS context, to my knowledge, and the method fit spectacularly the mode of communication of the Debian Project, and the specific research objectives. My research demonstrates the approach’s suitability to FLOSS research, and my second contribution is the meticulous documentation of all design considerations, and details of the entire process, from sampling the panel to processing the data. This work should facilitate the use of the Delphi method for future research, as well as daytoday consensus-finding in FLOSS projects. In the spirit of the Debian Project, and to foster further research into the adoption behaviour of Debian Contributors, I obtained permission from all panelists to publish all data generated during the study under a free licence. The availability of these data also reinforces the validity and reliability of my research.