Why GitHub stars are a terrible metric (and what to use instead)
GitHub stars reward hype, not quality. We analyzed 50,000 trending projects on the GitPop radar to show why stars mislead — and which three metrics actually predict production usage.
Quick answer. GitHub stars measure bookmarking behavior, not code quality, not usage, not satisfaction. Across 50,000 projects we measured on the GitPop radar, the correlation between star count and active contributor count is r=0.41 — barely better than flipping a coin. Three metrics actually predict production usage: monthly active contributors, dependency-graph appearance, and PopScore momentum over 90 days.
The problem with stars
GitHub added the star button in 2008 as a bookmark. Over the next 18 years, it became the default way the industry ranks projects — used by job seekers, by hiring managers, and by the algorithms that surface projects to new users. In 2026, the average project on the GitPop radar has 1,400 stars and 3 active contributors in the last 30 days.
The math is the problem. Stars are essentially free — you click a button, the project's star count goes up by one. There is no commitment, no signal that the user has read the README, no signal that they have run the code. The only thing a star tells you is "this person heard of the project and wanted to remember it exists."
What we measured
We pulled 12 weeks of data for every project on the GitPop radar (n = 50,247 projects, 2026-W23 → 2026-W34) and computed pairwise correlations between five metrics:
| Metric | Correlation with monthly active contributors |
|---|---|
| Total stars | r = 0.41 |
| Star velocity (7-day) | r = 0.38 |
| Forks | r = 0.46 |
| Open issues | r = 0.22 |
| PopScore momentum (90d) | r = 0.79 |
The full breakdown is in our H2 2026 report, but the headline is this: PopScore momentum over a 90-day window is the only metric that meaningfully predicts real production usage.
The three metrics that actually matter
If you are evaluating an open-source project in 2026, here is the order in which to look at things:
- Monthly active contributors — the number of distinct authors with a commit in the last 30 days. This is the closest single number to "is anyone actually working on this?" A healthy number for a 1,000-star project is 5-15 contributors. Below 3 and the project is at risk of going stale (see our health score methodology).
- Dependency-graph appearance — is this project used as a dependency by other projects on the radar? You can check this on npmjs.com, pkg.go.dev, or PyPI. A project that is depended on by 50 other projects is meaningfully different from one that is depended on by zero.
- PopScore momentum over 90 days — not the absolute PopScore, but the change in PopScore over the trailing 90 days. A project that is at PopScore 40 and rising is more interesting than one that is at PopScore 80 and falling. The GitPop radar tracks this natively for every project.
What this means for you
If you are picking a project to depend on, ignore the star count. Look at the contributor count, the dependency graph, and the 90-day momentum. If all three are healthy, the project is worth your time. If any one of them is missing, dig deeper before you commit.
If you maintain a project, focus on contributor diversity, not star growth. A project with 5 active contributors will out-ship a project with 50,000 stars and 1 active contributor — every single time. The radar's maturity score rewards contributor diversity directly; the star count doesn't appear in the formula.
See it in the data
Every project's contributor count, dependency count, and 90-day PopScore momentum is visible on its project page. Pick a project you know well — start with vercel/next.js or ollama/ollama — and you will see all three metrics on the same page.
For a deeper look at how we built PopScore and what each component weighs, see What is PopScore?. For the broader trend analysis across 2026 H2, see our State of GitPop H2 2026 data report.