A developer can spend every evening contributing to a project and look almost inactive from the public web.
Private repositories make that possible by design.
GitHub lets repository owners invite collaborators to private repositories while keeping the code, issues, pull requests, and detailed activity unavailable to people without access. The work is real. The public evidence is intentionally limited. See GitHub’s documentation on repository access and collaboration.
That matters when public contribution graphs are treated as a rough census of technical participation.
The green squares can hide the project
GitHub’s profile system illustrates the problem unusually clearly.
By default, contribution graphs show public repository activity. A user can choose to display activity from private repositories too, but GitHub only exposes the contribution counts to outsiders; the specific private repository and details remain hidden. See Contributions on your profile.
So an observer might see that somebody made twenty contributions on Tuesday without learning whether those were documentation fixes, security work, code reviews, issue triage, or commits to a volunteer project nobody outside the team can inspect.
If the user chooses not to show private contributions at all, even that signal disappears.
Why volunteers work privately
Private does not automatically mean commercial or secretive.
Volunteer teams may use private repositories while preparing a release, handling security-sensitive information, working with unreleased assets, protecting personal data, coordinating an event, or simply deciding that unfinished work does not need a public audience yet.
Some projects later publish the result. Others never do.
The key point is that public Git history measures publicly visible collaboration, not all collaboration.
Even inside public projects, contribution graphs have rules about which commits and activities count. GitHub’s own profile contribution reference documents those conditions and notes that not every event necessarily appears.
Invisible work is still work
This becomes another Human Retreat measurement problem.
If fewer hobbyists appear to be posting code publicly, several explanations are possible. They may truly have stopped. They may have moved into company repositories. They may be contributing privately to volunteer groups. They may be working in self-hosted systems that a GitHub profile never sees.
You cannot distinguish those explanations from a public contribution graph alone.
That does not make public open-source metrics useless. Public commits, issues, and reviews remain excellent evidence of visible collaboration.
They are just not a complete census of human technical effort.
Sometimes the coder did not leave.
The repository closed the blinds.