Episode 295

Jiayi Sun & Shurui Zhou on Codes of Conduct in Open Source Communities

October 16th, 2026

34 mins 9 secs

Transcript

Link with Timestamp

Download MP3 (62.8 MB)

Your Host
Special Guests

About this Episode

Guest

Jiayi Sun | Shurui Zhou

Panelist

Richard Littauer

Show Notes

In this episode of Sustain, Richard Littauer is joined by University of Toronto researchers Jiayi Sun and Shurui Zhou to discuss their research on how Codes of Conduct affect open source communities after adoption. Drawing from a dataset of more than 150,000 GitHub repositories, they explore how often Codes of Conduct are updated, whether they attract or discourage contributors, and why simply adding a policy may not be enough to create a healthy community. They also discuss using AI to analyze thousands of Codes of Conduct changes and how the rise of AI-generated contributions is creating new challenges for open source maintainers. Download and listen now!

[00:01:03] Jiayi Sun and Shurui Zhou introduce their research into collaboration, sustainability, and community dynamics within open source software.

[00:03:14] The researchers built a massive dataset covering more than 150,000 GitHub repositories with Codes of Conduct.

[00:04:04] Their paper, Beyond Adoption: Examining the Evolution and Impact of Codes of Conduct on Open Source Communities, examines what happens after a Code of Conduct is adopted and whether it changes community participation.

[00:07:24] Jiayi explains how the team identified Codes of Conduct using GitHub issues, pull requests, commit messages, repository files, and GitHub's community metadata.

[00:10:53] One of the biggest findings: more than 75% of projects never update their Code of Conduct after initially adopting it.

[00:15:10] Richard asks about the tools they used to cluster and understand the objects they were looking at. The researchers used GPT-4 Turbo to help analyze roughly 8,000 Code of Conduct-related commits and categorize the types of changes projects were making.

[00:17:41] Code of Conduct adoption was associated with a short-term increase in new contributors, along with some increase in existing contributors disengaging.

[00:20:21] Long-term effects were much smaller, suggesting that a Code of Conduct can serve as an important signal, but community health requires continued structural and cultural work beyond simply having a policy.

[00:22:01] AI-assisted and agentic software development is creating a new challenge for Codes of Conduct, as communities debate how AI-generated contributions should be handled and what behavior should be allowed.

[00:26:06] Shurui explains how the research grew out of conversations about inclusion and community management, asking a fundamental question: if so many projects already have Codes of Conduct, why do maintainers still spend so much time resolving conflicts?

Quotes

[00:11:27] “Over 75% of projects never touch their code of conduct after initial adoption.”

[00:23:09] “This question of how people engage with contributors has become more important with agentic software engineering and the growing volume of AI assistant contributions.”

Spotlight

  • [00:29:46] Richard's spotlight is Seldo, a.k.a Laurie Voss, and a blog post they wrote called, “Nobody pays for open source. We can force them to.”
  • [00:30:43] Shurui’s spotlight is Arliss Collins from NumFOCUS.
  • [00:31:18] Jiayi’s spotlight is the Scientific Software Community.

Links

Sponsor

CURIOSS

Credits

Support Sustain