Reassessing Brooks’ Law for the Free Software Community

Proponents of Free Software have argued that some of the most established software engineering principles do not fully apply when considered in an open, distributed approach. Among these principles, “Brooks’ Law” has been questioned in the Free Software context: large teams of developers, contrary to the law, will not need an increasingly growing number of communication channels. As advocates claim, this is due to the internal haracteristics of the Free Software process: the high modularity of the code helps developers to work on comparted sections, without the need to coordinate with all other contriutors.
This paper examines Brooks’ Law in a Free Software context, and it studies the interaction of contributors to a large Free Software project, KDE. The network of interactions is analyzed and a summary term, the “compaction”, is dynamically evaluated to test how the coordination mechanism evolves over time in the project. This paper argues that the claim of advocates holds true, but with limitations: in the KDE project, the few initial developers needed a significant amount of communication. The growth of KDE brought the need to break the number of overall communication channels to a significant extent. Finally, an established amount of 300 developers currently needs the same amount of communication as when the developers were only 10. We interpret this result by arguing that Brooks’ Law holds true among the core developers of any large Free Software project.