Studying donations and their expenses in open source projects: a case study of GitHub projects collecting donations through open collectives

Operating an open source project requires not only intrinsic motivation (e.g., the joy of participation) but also extrinsic motivation (e.g., financial incentives). Almost 95% of open source projects are no longer maintained after a year. Nowadays, although donations start to play an important role in operating open source projects, there is little knowledge about the characteristics of donors and the usage of donations. A better understanding of the characteristics of donations, their donors, and the usage of donations in open source projects is needed to provide insights to the stakeholders of open source projects to help them operate their projects more sustainably. In this paper, we study the donations that are received through the Open Collective platform (i.e., an online crowdfunding platform) to support open source projects, to understand the characteristics of these donations, their donors, and the usage of these donations. To do so, we investigate 225 GitHub open source projects that received 54,889 donations with a total value of $2,537,281 through the Open Collective platform. We find that: 1) In general, corporate donors tend to donate more money than individual donors in a single donation. However, in a collective, the total donation amount from individual donors is more than corporate donors, suggesting the importance of individual donors. Moreover, individual donors are more likely to redonate to the same collective compared to corporate donors. 2) Non-engineering-related expenses take up to 54.0% of the total number of all expenses that are filtered against donation. For instance, "Web services", "marketing", and "travel" expenses are the three most frequent and costly non-engineering-related expense types. For engineering-related expenses, the most frequent expenses are related to development and maintenance. Interestingly, we also observe that 18% of the engineering expenses were spent to propose bounties for addressing issues with a median cost of $95 per proposed bounty. We further analyze the differences between individual-supported collectives (i.e., collectives where more than 80% of their donation amount is from individual donors) and corporate-supported collectives (i.e., collectives where more than 80% of their donation amount is from corporate donors). We observe that corporate-supported collectives tend to receive a higher donation amount than individual-supported collectives and the monthly received donation amounts are positively associated with the levels of community and maintenance activities in corporate-supported collectives. They have no significant difference in terms of popularity (e.g., the number of pull requests) of their associated GitHub projects. Our findings suggest that the stakeholders of GitHub open source projects should try to attract more individual donors. Collectives should not expect to receive a large amount of funds overall from donations unless their projects are very popular or are mainly supported by corporations. Projects should budget for a reasonable amount (e.g., 13% of total funds) of non-engineering expenses (e.g., marketing and traveling).