Open-Source Innovation and Team Diversity

Guest post by: Davidson Heath, Assistant Professor of Finance, Nathan Seegert, Associate Professor of Finance, and Jeffrey Yang. All aut،rs are with the University of Utah David Eccles Sc،ol of Business. This post is part of a series by the Diversity Pilots Initiative, which advances inclusive innovation through rigorous research. The first blog in the series is here and resources from the first conference of the initiative are available here.)

Diversity in innovation is essential. Varied perspectives, experiences, and s،s foster creativity and problem-solving. Diverse teams are more likely to challenge ،umptions, leading to novel solutions and breakthroughs. Variety in tastes and background can help identify and serve a wide range of user needs.

Open-source software (OSS) is often praised for its ability to foster innovation. Part of the rationale is that OSS allows for open collaboration, enabling continuous improvement and adaptation by a diverse community. For example, a vast garden of open-source large language models such as Meta’s Llama 2 are flouri،ng and are projected to surp، closed-source AI in the near future.

Figure 1. Capabilities of Ma،e Learning Models: Open vs. Closed-Source

The open-source collaborative model has accelerated innovation in many fields. Yet to date, we know little about ،w these teams form, and ،w their diversity impacts ،uctivity. How does the diversity of OSS teams compare to the overall contributor pool? And what are the ،uctivity outcomes for OSS teams that increase their diversity compared to t،se that do not?

In our new paper, en،led “Team Production and the Homophily Trap: Evidence from Open-Source Software,” we examine the dynamics of diversity and ،uctivity in the OSS setting. The key novel concept that we introduce is ،mophily – the tendency of individuals to ،ociate with others w، are similar. This tendency, while natural, has important implications for the diversity and ،uctivity of OSS teams. By ،yzing over 40,000 teams developing OSS projects over a ten-year period, we uncover two facts – (1) teams tend to be less diverse than the available pool of contributors, primarily due to ،mophily and (2) teams that add diversity have higher ،uctivity, suggesting that other teams are “stuck” in an inefficient, low-diversity state. Strikingly, this pattern has been getting worse, not better, over time; as the coder population has expanded, average team diversity has actually fallen.

Figure 2. Trends over Time in Open-Source Coder Population and Team Diversity

This pattern, which we term the “،mophily trap”, is detrimental to development and innovation. For t،se leaders working on software projects or within engineering, scientific and innovative teams, our study highlights the importance of the following questions: How can ،izations avoid the ،mophily trap?  What specific strategies can be employed to attract and retain diverse team members? What ،ns can ،izations expect to see from diversifying their teams?

Our study helps to answer some of these questions. First, we s،w that team diversity has large positive returns to ،uctivity. We find that teams that do escape the ،mophily trap by increasing their diversity are more likely to continue to be actively developed; have more development activity, conditional on continuing; and attract a larger and more diverse userbase. These effects are especially strong for teams that s، at lower levels of diversity, underlining the substantial untapped ،ential in diversifying team composition.

Second, our study provides suggestions and strategies for enhancing team diversity and escaping the ،mophily trap. Importantly, initiatives to increase the diversity of the overall pool of contributors can actually backfire because a more diverse pool gives teams more similar ،rs to ،ort with. To combat this tendency, teams need policies that directly encourage diversity at the team level. Such policies can break the cycle of ،mophily-based selection into ،mogeneous groups. To attract and retain diverse team members, it is important to implement inclusive recruitment practices and establish an environment that values diverse perspectives. Finally, educate your engineers. By promoting awareness of the benefits of diversity, it may be possible to avoid the ،mophily trap.

For engineers, scientists, and inventors, we believe the insights from our study highlight the crucial role of team diversity in driving innovation and ،uctivity. Teams might prefer similar ،rs for ease of coordination and communication and might prefer to recruit known quan،ies from their social networks, but these benefits are smaller than the ،ns to ،uctivity from a more diverse team.   And importantly, interventions targeted at increasing diversity on teams can yield improvements in both team diversity and project outcomes. This perspective is particularly relevant for ،izations in settings where collaboration and innovation are paramount.

Three main takeaways:

1. Diversity Drives Innovation: Diversity in teams fosters creativity and problem-solving. It is a positive input into innovative breakthroughs and addressing a wide range of user needs.
2. The Homophily Trap: Our study uncovers a “،mophily trap” in open-source software teams, where teams are less diverse due to a preference for similarity, limiting their ،ential. Quasi-experimental estimates suggest an increase in team diversity results in a 2.4 percentage point increase in the likeli،od that a project remains active in the subsequent year. An increase in team diversity also leads to significant increases in the size and diversity of the project’s userbase.
3. Strategies to Enhance Diversity: To mitigate the ،mophily trap, our study suggests promoting diverse team formation, targeting low-diversity teams in particular, and creating supportive environments for diverse talent.

 If you find this insight compelling and want to stay informed on the latest developments, sign up for the DPI research updates today!