Peer Review

In order to ensure the quality and utility of its libraries, Boost adopted the model used to select worthy ideas in science. By subjecting proposed library submissions to careful review by interested users and library authors, a broad range of application domains and design techniques are considered before any library is accepted into Boost. This approach helps to insulate Boost users against both interface obsolescence and the convulsive interface changes that are required when a design must be altered to account for previously unanticipated requirements.

Boost’s use of peer review sets it apart from other open source projects: the code in Boost libraries is neither completely unmediated, nor the product of the ideas of a single visionary leader, nor the result of free-for-all “design by committee.” Acceptance of every library is subject to a vote by reviewers, but the final decision is up to a review manager, and the author of an accepted library retains control over its maintenance and evolution. By combining broad input with individual accountability and ownership, Boost hits a “sweet spot” that maximizes the effectiveness of the review process. Although reviews can be extremely rigorous, many programmers actually crave the application of such detailed scrutiny to their best work.