Why Do Some Software Programmers Work for Free?

On the face of it, the open source movement is an economic oddity. Thousands of software developers devoting hundreds of unpaid hours to create free programs for the masses.

Linux, the operating system which powers Google, and Firefox, a web browser which has a taken nice chunk of market share from Microsoft's Internet Explorer (especially in Europe), are two of the more successful examples of the OS movement.

As for the motivation behind working for free, open source developers say that the recognition of peers and believing in the spirit of the project are big factors.

New research published in International Economics and Economic Policy, an academic journal, suggests that a less altruistic motivation may also be at play. (PDF, free version)

Two Israeli economists, Chaim Fershtman and Neil Gandal, downloaded data on 71 of the most active open source programs in 2000 from SourceForge, a web site that hosts over 100,000 open source projects.

These projects are usually licensed under a variety of schemes which dictate any potential commercial use. Fershtman and Gandal broke up their sample into three groups ranging from the most restrictive to the least.

In the unsentimental mind of an economist it would make sense that if a project had a less restrictive license -- and hence a potential monetary payoff -- a developer would have more incentive to contribute.

As for the anti-commercial projects, the "payment" in these cases is likely related to contributing enough lines of code to be listed as an official developer, thus gaining public recognition -- and perhaps a future paying gig. Here is an example of a developer's list on SourceForge.

Fershtman and Gandal used information on the number of contributors and the lines of source code in each project to get a measure of the average productivity for each developer. (In other words: Lines of code per developer)

And perhaps not surprisingly they found that projects that had the opportunity to go commercial were more productive than those that didn't.

Here are some more of their findings:

• The average output per contributor is 6,642 lines of code for non-restrictive licenses versus 2,319.
• Meanwhile, the median output per contributor is 2,125 lines of code for non-restrictive licenses versus 1,367. The median, or the midpoint in a series of numbers, can give a better sense of the "typical" contributor. These numbers suggest that it wasn't the case that the final results were skewed by restrictive projects having more contributors or less code per project.
• Interestingly, the typical restrictive project had 35 contributors while the non-restrictive project had 13 contributors, implying that peer recognition or ideology play large motivating roles for many open source developers.

In the end, the results suggest that a potential monetary payoff down the line does motivate software developers to be more productive, but many developers are still attracted to the movement for things besides pecuniary rewards. They just don't seem to work as hard.