Collaboration Quantified: Lessons from Open Source Developer Networks
summary
Analyzes ~500k Jira artifacts across ~42 open-source projects using Python to build interaction graphs with a gravity layout and finds a persistent core contributor network (median ≈7% of contributors, cores as small as ~15–20 people even in projects with 2,500+ contributors), widespread articulation points that disconnect the graph when removed, and an empirical cost model where collaboration grows ~n² so the optimal team size formula ≈1/(fraction of time spent collaborating) (10%→10 people, 30%→3), advocating SLAs/as-a-service to reduce peer-to-peer overhead.
AI-written summary · about
chapters (6)
- 0:00Introduction to collaboration
- 1:04Collaboration costs and trade-offs
- 10:11Professional networks in business
- 17:54Open source collaboration networks
- 32:36Building robust operational networks
- 42:05Q&A and closing thoughts
related talks
39mCollaboration as a Key Health Indicator for OSS Communities
43mWhat is Open Source Without Collaboration?
42mEnhancing Open Source Collaboration: From Incentive Programs to Data-Driven Contribu...
40mAnalyzing code contributions to the CPython project using NetworkX and Matplotlib
10mCollaboration Is Evolving
24mBest practices and (very) small projects
39mA Data Analysis Approach for Contribution-Centric Open Source Strategy Planning
40mDiscovering Emerging Open Source Communities Through...