"The biggest surprise to me was the level of involvement that firms engage in with community forms on software development and standard setting in general. That is, forms that are not government sponsored nor formally constituted by partnership, alliance, or consortia agreements."
Siobhán O'Mahony, Assistant Professor of Management, UCDavis
Communities that develop open source software (OSS) are virtual entities on the Internet, not legal entities. Some open source communities establish open source software foundations (OSSF) in order to protect their intellectual property and carry out contractual arrangements. As legal entities, OSSF help communities attain their long-term goals, hold community assets, provide resources to communities, and balance interests amongst different stakeholders.
When OSS started to draw more business interests, commercial companies became involved with open source communities. The emergence of OSSF provides a good platform and opportunities for companies to exert their influence in a more direct way.
This article summarizes our recent research regarding the relationships between company involvement, governance, revenue, and OSSF.
Typically when an open source community incorporates as an OSSF, they seek to gain financial advantages from donations and tax exemptions. OSSF is a non-profit organization (NPO) with a primary objective to support or to actively engage in activities of public or private interest without any commercial or monetary profit purposes. A NPO is a corporation that can handle business dealings, sign contracts, and own property as any other individual or for-profit corporation. NPOs differ from for-profits in terms of taxes and governance. A NPO's governance structures preclude private financial gain. OSSF in the US are registered under tax-exempt 501(c)(3) or 501(c)(6). 501(c)(6) is reserved for "Business Leagues and groups such as Chambers of Commerce" and is a form of a business network in favour of pursuing members' own business interests whereas 501(c)(3) organizations provide public benefits. Most OSSF registered in the US are 501(c)(3) non-profit organizations. In the US, both 501(c)(3) and 501(c)(6) NPOs are exempt from most federal income taxes. OSSFs registered outside of the US usually have similar benefits. From a legal perspective, an OSSF is a legacy NPO.
An OSSF must have at least one software project in which individuals are contributing to a new software release. The Apache Software Foundation is an OSSF while the Open Source Initiative (OSI) is not as it focuses on standards and does not hold any software projects.
OSSF in the Sample
Six OSSF registered in the US were selected for our research:
- The Apache Software Foundation (ASF) was incorporated in Delaware in 1999 to support the Apache HTTP server project. As of 2008, the ASF hosts 65 OSS projects, with 1765 committers. The ASF has a meritocracy based membership structure supporting a large and mature open source community. Its organizational processes serve as examples for other OSSF and it has earned a reputation for incubating OSS projects.
- IBM released Eclipse code as open source and formed the Eclipse consortium in 2001. In 2004, the Eclipse consortium was reorganized into the Eclipse Foundation, a 501(c)(6) non-profit. In 2008, the Eclipse Foundation hosted 11 top-level projects with 21 strategic members, 179 organizational members, and 942 committers. Eclipse is dominant in the market for Java Integrated Development Environments (IDE). Unlike most other OSSF, the Eclipse Foundation supports its members' business interests with a large and established community sponsored and controlled by companies. It is a known example of the commercial open source phenomenon known as OSS 2.0.
- The GNOME Foundation was founded in 1997 to support the GNOME project and related projects which provide a desktop environment and development platform. The GNOME desktop environment is one of the dominant desktop environments for Linux and the GNOME Foundation has a large and established community.
- The Plone Foundation was founded in 2004 to support the development of Plone, an open source content management system (CMS). Plone has a rather small share of the CMS market, but it is deemed as one of the best in the market. The Plone Foundation has a meritocracy based membership structure with many small open source consultants and service vendors within the Plone community.
- The Python Software Foundation was founded in 2001 as an organization devoted to the Python programming language, one of the leading script programming languages.
- Software in the Public Interest (SPI) was formed by members of the Debian project in 1997. Its mission is to "help organizations develop and distribute open hardware and software". It hosts several open source projects, such as Debian, freedesktop.org, and openOffice.org. GNOME was a sub-project hosted by SPI before the GNOME Foundation was founded. SPI seldom intervenes in the affairs of its member projects, but it does hold their common assets.
Governance refers to the overall processes and structures used to direct and manage an organization's operations and activities. Three aspects of governance were the focus of our research: i) governance structure; ii) activities of the Board of Directors (BOD); and iii) occupations of the members of the BOD.
Most of the sample OSSF are membership based NPOs with the elected BOD bearing the major responsibilities for the organization. Members of the OSSF can be either a merit member or a sponsor member. An individual becomes a merit member after being recognized as making non-trivial contributions to the foundation. A sponsor member is a company that donates resources, such as money and developers, and in return is admitted as a member of the foundation. Typically, individuals are merit members, organizations are sponsor members, and employees of organizations represent sponsor members.
Based on "the extent to which the sponsor members can participate in the decision making of OSSF", we found three types of OSSF governance structures:
- Merit: all members are merit members with full voting rights.
- Merit dominated: merit members are in a majority and it is difficult for sponsor members as a group to affect outcomes.
- Sponsor dominated: sponsor members are in a majority and may be classified into tiers where the size of the payment determines tier membership. Merit members typically work for a company or research centre.
The ASF, SPI, and the Plone Foundation are merit type foundations. The GNOME Foundation and the Python Software Foundation are Merit dominated. The Eclipse Foundation is sponsor dominated.
Board Activities and Occupations
Twelve categories of governance activities were defined as being carried out by the BODs of OSSF. The twelve activity categories are: i) strategic planning and common vision development; ii) development of policy and guidelines; iii) project governance; iv) financial governance; v) primary resources governance; vi) human resources governance; vii) fund-raising; viii) external relation management; ix) BOD self development; x) governance structure management; xi) community development; and xii) conference governance.
All six BODs in the sample allocated more than 60% of their efforts carrying out activities that fall into the following four categories: i) project governance; ii) development of policy and guidelines; iii) external relations; and iv) governance structure management. All six BODs allocated the least amount of effort to carrying out the following three activity categories: i) strategic planning and common vision development; ii) human resources governance; and iii) BOD self development. Effort was calculated by counting BOD activities per month in that category within the research period. Percentage of activities is the ratio of activities per month in that category to total number of activities per month. That the BODs in the sample did not spend more effort raising funds was a surprise. The BOD of one foundation did not undertake any activities to raise funds; only 3% of another's activities were carried out to raise funds. One reason may be that the virtual nature of OSSF does not require large operational budgets.
Among the six OSSF in the sample, the GNOME BOD is the most active as it carries out more activities per month than the other OSSF, while the BOD for SPI is the least active.
We believe that the occupation and status of BOD members affects the BOD's ability to carry out tasks as well as its behaviour. Eclipse and Plone have a greater proportion of BOD members with a background in management, business development, and strategic management, while the proportion of engineers and scientists is small. More than 50% of the BOD of the Plone and Eclipse Foundations are part of their companies' top management teams.
SPI has the largest proportion of BOD members with an engineering or scientific background (67%) and a small proportion of BOD members with a background in management, business development, and strategic management (11%). SPI and ASF have the largest proportion of OSS developers in their BOD, while Plone and Eclipse have none on their BOD.
Company Involvement in Governance
Company involvement refers to when the company invests resources in an OSSF and influences the mission, primary organizational activities, and relationships of the OSSF. A company may directly influence the decision-making processes of an OSSF by having its employees hold seats in the BOD and other governance groups of the OSSF. A company may also indirectly influence decisions made by the members of an OSSF by providing resources to affect a decision. All six OSSF in the sample receive monetary donations or membership fees from companies. All six BOD in the sample have employees from companies that have business interests in an OSSF's projects.
OSSF effectiveness refers to the ability of a foundation to utilize resources to achieve its goals and missions. Effectiveness metrics can be categorized as: i) project effectiveness; ii) community effectiveness; iii) resource acquisition effectiveness; and iv) stakeholders' assessment. From our observations on the OSSF in the sample, we provide the following propositions:
Proposition 1: the proportion of company employees in the BOD positively affects the power a committee of the BOD has over timing and the content of roadmap software releases.
Proposition 2: the proportion of company employees in the BOD is negatively related to the proportion of BOD members who have a reputation in the OSS community.
Proposition 3: the proportion of company employees in the BOD is negatively related to the proportion of engineers and scientists in the BOD.
Proposition 4: the extent to which sponsor members can participate in decision making positively affects the proportion of employees in the BOD from companies that do not depend on the OSSF's projects for significant revenue.
Proposition 5: the proportion of members of the BOD from companies that do not depend on the OSSF's projects for significant revenue positively affect the proportion of BOD effort dedicated to strategic planning and common vision development.
Proposition 6-a: the extent to which sponsor members can participate in decision making positively affects OSSF revenue.
Proposition 6-b: the proportion of employees in the BOD from companies that do not depend on OSSF's projects for significant revenue positively affects OSSF revenue.
It was discovered that company BOD involvement is comprised of two dimensions: i) company employees and ii) employees of companies that do not depend on the OSSF's projects for significant revenue.
OSSF governance is comprised of five dimensions. The first dimension focuses on how much effort the BOD spends on strategic planning and common vision development. The second and third dimensions relate to the composition of the BOD: i) members who have a reputation in the OSS community and ii) members with a background in engineering and science. The fourth dimension deals with power, specifically the power the BOD has over timing and content of the roadmap and software releases and the power of sponsor members. Finally, the fifth dimension focuses on sponsor members' ability to participate in decision-making.
Revenue could be used to assess OSSF effectiveness. Two dimensions, one that relates to company involvement and one that relates to governance, positively affect revenue. The importance of including board members from companies that do not depend on the OSSF's projects for significant revenue should be highlighted. Their presence increases the effort dedicated to strategic planning and common vision development as well as the OSSF's revenue. Governance structure, or the extent to which sponsor members can participate in decision making, affects company involvement and OSSF revenue. This discovery is important for designing an OSSF.
Revenue for a 501(c)(3) OSSF is rarely more than $150,000 a year over the long run. OSSF with sponsor dominated governance structures tend to register as 501(c)(6) non-profits, while OSSF with merit and merit dominated structures register as 501(c)(3) non-profits. Since donations to 501(C)(6) OSSF are not tax deductible, these OSSF must attract revenue from membership fees to survive. OSSF registered as 501(c)(3) can issue tax receipts, must receive at least one third of their support from the general public, and membership fees and revenues from OSS conferences are not deemed as public support. Large donations from BOD members, or repeatedly going back to the same donors for income may not count as public support. In the sampled 501(c)(3) OSSF, long-term public support is generally below US $50,000.
Both "policy governing BOD" and "working BOD" exist in the sample. A BOD may delegate most of the managerial and operational activities to a full-time Executive Director or Chief Executive Officer and specific councils. This working style is considered a policy governing BOD where a full-time management organization in the OSSF is the enabler. The BOD of the Eclipse Foundation is an example of a policy governing BOD. In a working BOD style, BOD members, volunteers and staff members carry out the work of the BOD. The other OSSF in the sample belong to this working style.
Key findings in the research reported in this article contribute to the existing literature on open source and non-profits. Our findings suggest practitioners of OSSF need to select a proper governance structure and corresponding strategies to achieve their goals.
OSSF are powerful institutional tools but are still not well understood by academics and practitioners. The research reported in this article is one of the few to address company involvement, governance, and effectiveness in OSSF. Future research can test the propositions proposed in this research, utilize subjective measurements to assess OSSF effectiveness, or evaluate companies' indirect influence on OSSF not addressed in the research reported.
Comparing Motivations of Individual Programmers and Firms to Take Part in the Open Source Movement