Active Developer Count

This counts unique contributor identities that authored at least one commit to the tracked repositories during the window. It is generally more robust than a raw commit count, because it is insensitive to whether a team squashes commits or pushes them individually. It measures participation in public code, which is one input into how a protocol is maintained and not a complete picture of the effort behind it.

How it is computed

distinct commit-author identities in tracked public repositories over the last 30 days

Where it misleads

Counting is by author identity, so one person using several accounts counts several times, while a shared organizational or bot account collapses many contributors into one. Chains that began as a fork of another project inherit that project's contributor history, and trackers that do not filter for it will report contributors who never worked on the fork. The measure cannot distinguish a core protocol engineer from someone correcting a typo in the documentation, since both produce one authored commit. Work on closed-source components, on a competing client maintained by a separate organization, or on private security fixes is invisible.

The equity comparison, and why it fails

Engineering headcount. The analogy fails because headcount is a payroll figure for a defined employer, whereas contributor counts are self-identified accounts on a public code host, employed by many organizations or none, and contributing amounts of work that differ by orders of magnitude.

Reading a high or a low value

A higher count means more distinct identities contributed code over the window; a lower one means fewer. Whether those contributions are substantial or trivial is not captured, so the count should be read alongside the repositories being tracked.