Network Science, Homophily and Who Reviews Who in the Linux Kernel?

TL;DR

This paper analyzes peer review patterns in Linux kernel development using network theory, revealing strong homophily among maintainers but no bias related to organizational affiliation, and highlighting the role of social norms.

Contribution

It introduces an analytical model combining homophily with peer review analysis in open-source software, providing new insights into review network dynamics.

Findings

Strong homophily among maintainers reviewing each other

No organizational bias in peer review despite firm involvement

Network theory effectively explains peer review network evolution

Abstract

In this research, we investigate peer review in the development of Linux by drawing on network theory and network analysis. We frame an analytical model which integrates the sociological principle of homophily (i.e., the relational tendency of individuals to establish relationships with similar others) with prior research on peer-review in general and open-source software in particular. We found a relatively strong homophily tendency for maintainers to review other maintainers, but a comparable tendency is surprisingly absent regarding developers' organizational affiliation. Such results mirror the documented norms, beliefs, values, processes, policies, and social hierarchies that characterize the Linux kernel development. Our results underline the power of generative mechanisms from network theory to explain the evolution of peer review networks. Regarding practitioners' concern over the Linux commercialization trend, no relational bias in peer review was found albeit the increasing involvement of firms.