Vite was an outlier in an ownership concentration analysis of 26 OSS projects

[SushantVerma7969]

I built a CLI called git-archaeologist that measures ownership
concentration from git history. I ran it across 26 major OSS
projects as a benchmark.

Vite stood out in the dataset: despite having 236 contributors, only 1 module crossed the ownership concentration threshold used in the report. Most projects in the dataset
had several.

Full report with methodology and raw data:
https://sushantverma7969.github.io/git-archaeologist/

I'm aware the metric measures historical commit activity
concentration, not who actually understands the code — that
limitation is documented in the report.

Curious whether this matches what you observe maintaining Vite, and what important factors this methodology might be missing.

[RuntimeTerror001]

Interesting work. The result doesn't surprise me too much, since Vite has a fairly modular architecture and a review culture that spreads knowledge across contributors.

That said, I think commit history alone misses a few important aspects of ownership:

• Code review activity and design discussions often create understanding without showing up as commits.
• Some maintainers have deep expertise in specific subsystems but touch them infrequently.
• Large refactors, file moves, and generated code can skew historical ownership.
• Historical activity doesn't necessarily reflect who is actively maintaining or understands the code today.
• Non-code work (triage, releases, docs, ecosystem coordination) is an important part of project ownership that Git history can't capture.

So I'd view ownership concentration as a useful signal rather than a proxy for bus factor or code understanding. It would be interesting to combine commit history with review participation and recency weighting to get a more complete picture.

[SushantVerma7969]

Thanks, that's a fair point. The study measures commit-touch concentration rather than ownership, expertise, or maintainership, so I treat it as a historical activity signal rather than a complete ownership model.
I agree that review activity, design discussions, and other non-code work would provide useful additional context. One reason Vite stood out in the dataset is that its concentration remained unusually distributed compared to most other projects, which made it an interesting outlier to investigate.