Conversation

Notices

1. Embed this notice
   Kyle Davis (linux_mclinuxface@fosstodon.org)'s status on Thursday, 02-Jan-2025 23:38:46 JST Kyle Davis

   Another reason GitHub stars are a garbage metric, they are actively gamed and exploited.

   If you get request for measuring #oss growth with them: push back hard.

   https://www.bleepingcomputer.com/news/security/over-31-million-fake-stars-on-github-projects-used-to-boost-rankings/

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Thursday, 02-Jan-2025 23:40:29 JST Rich Felker

     in reply to

     @linux_mclinuxface So stupid that they're used for rankings. I use stars to flag projects I want to come back to visit, and let the authors know I'm interested. Not to play stupid recommendation algorithm games.

   • Embed this notice
     Kyle Davis (linux_mclinuxface@fosstodon.org)'s status on Thursday, 02-Jan-2025 23:46:18 JST Kyle Davis

     in reply to

     • Rich Felker

     @dalias I've heard bonkers stories of people selecting dependencies based on number of stars.That should alarm basically anyone.

     OTOH: there is a disease of being data driven to the point that bigco leadership people can't operate without a number that indicates up-and-to-the-right. Consequently, they latch their jaws into anything that looks like that type of metric. Then people start playing games...

     No metrics is a _feature_ of OSS, not a bug.

   • Embed this notice
     Rich Felker (dalias@hachyderm.io)'s status on Thursday, 02-Jan-2025 23:48:36 JST Rich Felker

     in reply to

     @linux_mclinuxface Wow 🤦

     From my perspective, no-star means "I'm ready to clone and use this now, seems mature, no need to mark it to follow development later."

     A project I star is more likely to be not-ready-for-primetime or have integration difficulties I haven't yet worked out.

   • Embed this notice
     Kyle Davis (linux_mclinuxface@fosstodon.org)'s status on Friday, 03-Jan-2025 08:45:01 JST Kyle Davis

     in reply to

     • Jonathan Yu

     @jawnsy fork count is probably a better metric, but not by much.

     Staring has no agreed use. Is it a save-for-later? Is it a I-use-this? Is it functionally a read receipt? No one agrees.

     (On top of this, it’s often an empty call-to-action “star the repo on GitHub” y tho?)

     The other insidious thing about stars as a social proof is that picking something because it is popular is anti-engineering.

     Look at requirements, evaluate component. No short cut to that.

   • Embed this notice
     Jonathan Yu (jawnsy@mastodon.social)'s status on Friday, 03-Jan-2025 08:45:02 JST Jonathan Yu

     in reply to

     @linux_mclinuxface I'd be lying if I said that I didn't give preference to projects with more stars, there's a social proof element to it, but I agree that it's a dumb metric. It's also just super prominent and there's not that much to go off of when selecting a project. Otherwise, I look at the readme and docs.

     Biggest factor for me is how I come to know the project. Often someone mentions it and if they've used it then it seems more legit to me.

     I should probably pay more attention tbh

     Rich Felker repeated this.