Eyeballs, Bugs, and Releases in Open Source Software

This study examines two widely cited principles of Linus’ law, namely “given enough eyeballs, all bugs are shallow”, and “release early and release often”. The aim is to understand their relationships and their limits to debugging open source software (OSS) bugs. Anecdotally, most of the successful OSS groups seldom develop in isolation; and their bugs and the underlying debugging processes are likely to be intertwined among multiple groups. We argue that the interrelatedness among software groups is an outcome of the long range contacts established through the boundary spanning activities of their contributors. Long-range contacts can exert an inverted U-shaped impact on releases, that is, initially, as conduits of valuable information, they benefit releases. But too many contacts slow down releases. We also hypothesized the influence of long-range contacts on releases is moderated by the relative location of OSS groups. We tested these intricate relationships using the contributions made by 7078 developers in solving over a million of highest priority bugs from 2343 software groups. Our empirical models are largely supported.