Using Networks to Understand the Dynamics of Software Development

Software engineering, being a relatively new field, has struggled to find ways of gauging the success/failure of development projects. The ability to determine which developers are most crucial to the success of a project, which areas in the project contain the most risk, etc. has remained elusive, thus far. Metrics such as SLOC (Source Lines of Code) continue to be used to determine the efficacy of individual developers on a project despite many well-documented deficiencies of this approach. In this work, we propose a new way to look at software development using network science. We examine one large open-source software development project—the Python programming language—using networks to explain and understand the dynamics of the software development process. Past works have focused on the open source community as a whole and the relationships between the members within. This work differs in that it looks at a single project and studies the relationships between the developers using the source code they create or work on. We begin our analysis with a description of the basic characteristics of the networks used in this project. We follow with the main contribution of this work which is to examine the importance of the developer within their organization based on their centrality measures in networks such as degree, betweenness, and closeness.