Zstandard
facebook.github.io
facebook.github.io
If there are more stars, I would assume that if the owner of the repo dropped off the face of the earth, it's more likely that a new maintainer would step up.
From my personal observation, those two go together and I usually don't take them in isolation. One may be able to game the stars, but they won't be able to game activity (issues, commits, etc.), I'm hoping at least not in a meaningful manner.
Furthermore, I would add the following considerations (in no particular order): documentation existence/quality, possible use of CI, reasonable coverage (if shown/known), use/inclusion in other projects, outward quality/appearance of code (can I read it, could I fix it, could I maintain it), licensing (clear/compatible) — outside of the repo: are there articles about it on the web, are there videos about it on Youtube.
It varies depending on what the repo is, and what I might intend to use it for, as to how in-depth I will actually go with each of these. But it's always worth a quick scan. There is much that can be inferred from examining such things.
Stars are just a single indicator among many.
To me this seems totally reasonable and proportional, given the types of developers that might use each library.
Personally, I don't believe there is much to gain by gaming Github's stars (particularly for a compression library), because there is much more to evaluating a repo and its contents than the count of its stars. Granted, that's just my opinion, and it is somewhat tangential to your original point, just my 2p.
[Disclaimer: I've used Zstandard in several projects, and will happily use it again! :) ]