OSSRank: How We Think About Open Source Project Quality

When evaluating open source projects — whether for adoption, investment, or competition analysis — most people start with GitHub stars. That's a mistake.

The Star Count Trap

Stars measure one thing: how many people clicked a button. They don't tell you:

• Whether the project is actively maintained
• If contributors are diverse or concentrated in one person
• Whether issues get addressed or pile up
• If the project has commercial sustainability

A project with 50,000 stars and one burned-out maintainer is less valuable than a project with 5,000 stars and a funded team shipping weekly.

What to Look At Instead

Velocity

• Commits per week/month
• Time to merge PRs
• Release frequency

Community Health

• Number of unique contributors (not just the maintainer)
• Issue response time
• Ratio of issues closed vs. opened
• Quality of documentation and onboarding

Commercial Viability

• Is there a company behind it?
• What's the business model?
• Are enterprise features gated or is it purely community?

Dependency Risk

• How many other projects depend on this?
• What's the bus factor (how many maintainers would need to leave for the project to die)?

A Quick Checklist

Before adopting an OSS project, I check:

1. Last commit date — If it's been 6+ months, be cautious
2. Open issues trend — Is the backlog growing or shrinking?
3. Contributor count — More than one active maintainer?
4. Funding — Is there a company, foundation, or sponsors?
5. Alternatives — What would you use if this project died tomorrow?

Tools like Star History can show growth trends^[1], and CHAOSS provides comprehensive metrics for community health.^[2]

Why This Matters

Building on unmaintained open source is technical debt in disguise. You'll either fork and maintain it yourself, or scramble to migrate when it breaks.

The best projects combine community enthusiasm with commercial sustainability. That's what we look for.

---

This framework informed how we built Astronomer on Apache Airflow, and later how we evaluated the AI coding agent landscape at Tembo.