An Empirical Study of Donations in Open Source

Even as open source software becomes more widely used, there has been growing concern about its sustainability Eghbal2020. Volunteers can only do so much for so long: thousands of pieces of critical infrastructure only exist because a handful of people are willing to sacrifice their evenings and weekends, and sooner or later, they burn out, become disillusioned, or have to devote their attention to other things.

Overney2020 looks at one model for funding their work: donations through platforms like PayPal and Patreon. The authors found:

…25,885 projects asking for donations on GitHub…typically with the goal of supporting engineering activities. Many of these projects receive donations but rarely enough to fund a full-time engineering position. While we do not find strong evidence that received donations associate with higher levels of activity in a project, we find a multitude of different patterns of how received donations are spent… many projects that are successful at fund-raising do not spend their funds [and] funds are often spent on non-engineering community activities (e.g., travel), web hosting, and personal expenses.

The finding that some projects don't spend the money they raise surprised me, but the authors report that:

…there are big differences regarding spending: Some projects actively spend all raised funds while others barely spend any, sometimes accumulating significant amounts of money… In our sample, 24 projects (40%) spent less than 25% of their raised donations, and 9 projects (15%) spent more than 75%.

The paper's tag line "How to not get rich" is probably its most important finding. Companies are willing to pay people to work on a few high-profile projects like Linux and Python, but they are the exception. Having won the fight for respectability, the biggest struggle for open source is now for support; studies like this that tell us how today's strategies are working are an essential step toward that.