A self-hosted star is worth a lot more than a GitHub star

I noticed Xorm on GitHub got ~6,000 over 6 years (2013->2019) that’s an average of 1000 stars per year. It then migrated to being self-hosted Xorm on Gitea got ~400 stars over 4 years (2019->2023) that’s an average of 100 stars per year. Since that’s the exact same software, it could be seen as a star conversion rate: one star on a self hosted forge is worth 10 stars on GitHub.

I’ve heard on occasion that one of the value of hosting a software GitHub is that it allows it to get more contributors. I’ve never seen any actual proof or study (even from GitHub) that this is true. And maybe this belief is based on the fact that self-hosted projects have less stars compared to others hosted on GitHub. I believe that is relatively easy to verify to be true. But assuming that it translates to getting more contributors is a stretch because it assumes there is an absolute value attached to a star. It assumes that a star on GitHub has exactly the same value as a star on a self-hosted platform.

There also is a general consensus that since the vast majority of developers have a GitHub account, it is less friction for them to contribute to a software hosted on GitHub than one that is on a self-hosted platform. This is hardly debatable: creating a new account is necessary and this is a hard fact. It follows that a star on a self-hosted platform requires more effort for most people than a star on GitHub. A person granting a star on a self-hosted platform is more invested in the project than a person granting a star on GitHub.

In addition to the relative value of stars depending on the software hosting platform, there is the ownership of those stars. When a star is granted on GitHub, it does not belong to the project. It belongs to GitHub. If the project migrates, it won’t be allowed to take the stars with them. It won’t even be allowed to know who are the users who granted the stars. As those stars became a value that is generally recognized to help with fundraising a startup or improving a CV, they create a dependency to GitHub that increases over time.

A star on a self-hosted platform has none of these inconvenient: a star on GitHub does not belong to the project, it belongs to GitHub, a star on a self-hosted platform belongs to the project.

I think comparing stars on GitHub with stars on a self-hosted platform should take into account:

  • a conversion rate by which a star on a self-hosted platform is worth more than a star on GitHub (the Xorm example suggests 10 times more)
  • a star on self-hosted platform belongs to the project, a star on GitHub belongs to GitHub
1 Like

Interesting idea, thanks for taking the time to elaborating it on here!

While I agree that stars on self-hosted forges are of higher value than something on GitHub, I’m of the opinion that we can come up with better, more human-centric metrics than stars to measure of a project’s popularity.

I have a few ides of such metrics, but they have more friction than stars. One of them is endorsement from fellow hackers: something like bookauthority.org but for software.

2 Likes