Brooks' versus Linus' law: an empirical test of open source projects

findings, recommendations, and opinions expressed are those of the authors and do not reflect the views of the funding agency. Thanks go to Megan Conklin, Kevin Crowston and the FLOSSmole project team (http://ossmole .sourceforge.net/) for making their Sourceforge data available, and for their assistance. We are also grateful to Thomas Folz-Donahue for programming work building our FLOSS project database, and for support from the ABSTRACT Free/Libre and Open Source Software (FOSS) projects are Internet-based collaborations consisting of volunteers and paid professionals who come together to create computer software. Because FOSS is available at no cost and has other advantages over proprietary software, both governments and businesses are increasingly using FOSS products. As a result, understanding what makes FOSS projects succeed has become a crucial question for many information technology (IT) professionals in both the public and private sectors. The specific question addressed in this paper is whether adding additional programming staff to FOSS projects makes those projects more successful. This question is controversial because traditional theory on software development indicated that adding programmers to late projects makes the problem worse (the famous Brooks Law in Software Engineering). We note that a similar theoretical argument regarding group size and public goods was made by Mancur Olson in his book The Logic of Collective Action. However FOSS literature suggests that adding programmers helps solve difficult programming problems (Linus' Law). We test these theories by studying the very large amount of data that is available about FOSS projects on Sourceforge.net (SF), an open source software hosting web site. In previous work, we created a measure of FOSS project success and then used this measure to classify nearly all the projects hosted on SF. In this paper, we examine whether the number of developers on a project correlated with success. The results showed that as the number of developers increases, the likelihood that a project will be successful increases. These results favor Linus' Law over Brooks Law/Olson. Although more research is needed, this finding suggests that there may be advantages in the way FOSS projects are organized and managed, both in software development but also in other areas of public sector work.