(Whether it's a good idea or not,) people often use stars as a quick-and-dirty proxy for "is this repo reasonably well trusted?" and "how big is the community for this project?". All else being equal, many people would default to using a repo with 20,000 stars over a repo with 20.

Starring a repo is a way of tracking it. So now there are 55k people who were trying to keep an eye on the project who have silently lost that.

It's also an important indicator of community size and authenticity, important factors in the decision to use the code.

From personal experience and from reading the other comments, some people see open source as a popularity contest and more stars will get you more users (which in turn gets you more stars).

Choosing one library over another just because it has more stars is bad decision making.

Consider choosing a command line tool for making http requests. If you consider GitHub stars: - curl has 24.7k stars - httpie has 4.3k stars as of writing this - wget only has a mirror on GitHub with 264 stars

Does this mean wget is a bad tool? Does it mean httpie used to be vastly superior to curl with over 2x the amount of stars? I don't think you can really say anything meaningful about these projects based on the amount of stars, especially relatively to each other. Except for how many people clicked on the star button.

I sometimes decide what to use based on number of stars. It's a trust in the quality of the software.

Now httpie is on 2.5k stars and if I only knew about it today, I'd consider using an alternative with 10k stars if there's any.

A practical consequence (not necessarily the only consequence) would be for the people who starred it who use stars as a bookmark feature: they have now lost their bookmark.