Data Tales from an Open Source Research Team
summary
Team analyzes the GitHub Archive and runs queries over >1M events from 2022 (20M distinct actors) and finds bots are <1% of accounts but create ~24% of events and 70M bot pull request events ≈ 50%+ of PR activity, with projects like Kubernetes showing bots mainly closing PRs. Other projects use sentiment analysis and social media pipelines and learn solution mismatch is common — social chatter about infant formula centers on parenting topics (poop) so the alerting idea fails. Practical lesson: measure what your tools actually capture.
AI-written summary · about
chapters (6)
- 0:00Welcome
- 1:09Mission of Google’s OSPO
- 2:58GitHub bot activity
- 12:53Buenos Aires grocery project
- 17:05Data science consulting lessons
- 24:49Key takeaways and final Q&A session
related talks
35mRegulations, Ethics And Messy Data: Tales From An Open Source Analyst
24mA Beginners Guide to Open Source Metrics and Analysis
01h 03mWhat can we learn from 750 billion GitHub events and 42 TB of code
47mUnder the Radar: How we found 0-days in the Build Pipeline of OSS Packages
35mWho What How: Understanding Kubernetes Development through DevStats - Josh Berkus & Łukasz Gryglicki
30mThe rockstar of my community is a bot: Where are the humans?
22mGetting Started with Open Source Project Insights
40mDiscovering Emerging Open Source Communities Through...