Special Section: Open Source Software

While many successful and well-known open source projects produce output of high quality, a general assessment of this development paradigm is still missing. In this paper, an online community of both small and large, successful and failed projects and their programmers is analysed mainly using the version-control data of each project, also according to their productivity and estimation of expended effort. As the results show, there are indeed significant differences between this cooperative development model and the commercial organization of work in the areas explored. Both open source software projects in their size and their programmers’ effort differ significantly, and the evolution of projects’ size over time seems in part to contradict the laws of software evolution proposed for commercial systems. Both the inequality of effort distribution between programmers and an increasing number of developers in a project do not lead to a decrease in productivity, opposing Brooks’s Law. Effort estimation based on the COCOMO model for commercial organizations shows a large amount of effort expended for the projects, while a more general Norden–Rayleigh modeling shows a distinctly smaller expenditure. This proposes that either a highly efficient development is achieved by this self-organizing cooperative and highly decentralized form of work, or that the participation of users besides programming tasks is enormous and constitutes an economic factor of large proportions.