News for hackers. Stuff that matters.
Re:4.5M (Suspected) Fake Stars in GitHub by halamadrid on Thursday January 02, 2025 @04:04PM (#42578888)
Another interesting way - and I personally think its fraudulent. This is how it goes - run hackathons or sponsor events in Universities. There are a ton of colleges who are constantly seeking support to run events.

Some companies take advantage of this by asking for stars in return of sponsorship. I have seen proposals that say for a $2000 sponsorship - 2000 stars guaranteed. The way it works is if a participant registers in the event they also have to show proof that they starred a specific repo that belongs to the company.

Re:4.5M (Suspected) Fake Stars in GitHub by SOLAR_FIELDS on Thursday January 02, 2025 @07:32PM (#42580715)
Ah, so is the quid pro quo here that if I want to run a hackathon in my city I can get Acme Corp to toss their brand name on it and fund it, but only if I get my participants to also star the repo? That does sound fraudulent. Do they put these agreements in writing?
Re:4.5M (Suspected) Fake Stars in GitHub by kortilla on Thursday January 02, 2025 @07:54PM (#42580915)
It’s sleazy but not fraudulent.
Re:4.5M (Suspected) Fake Stars in GitHub by verst on Thursday January 02, 2025 @07:58PM (#42580945)
At many hackathons I attended (as a mentor, company sponsor, or judge) approximately a quarter of participants had never used GitHub previously. Such participants often thought of GitHub stars as bookmarks. Many hackathon sponsors provide Hackathon samples or references on GitHub - or guidelines for winning the specific sponsor's prize. As expected, those repos were "bookmarked" by the students. In this case this is a misunderstanding of the purpose of GitHub stars by these new users, but certainly not a fraudulent action.
Re:4.5M (Suspected) Fake Stars in GitHub by mintplant on Thursday January 02, 2025 @08:16PM (#42581080)
I mean, that's not a misunderstanding, that is what they're for. Conferring clout is a secondary effect.

https://docs.github.com/en/get-started/exploring-projects-on...

Re:4.5M (Suspected) Fake Stars in GitHub by krick on Thursday January 02, 2025 @08:19PM (#42581101)
Huh? These are bookmarks. I never actually use starts for any other purpose.
Re:4.5M (Suspected) Fake Stars in GitHub by HEYGRANDMA on Thursday January 02, 2025 @08:33PM (#42581193)
I've been on github for 10+ years and that's what I use stars for.
Re:4.5M (Suspected) Fake Stars in GitHub by diath on Thursday January 02, 2025 @09:49PM (#42581765)
What? That's exactly what I use stars for - to bookmark repositories so I can go through things I've found interesting at a later date, or to keep being updated on their releases.
Re:4.5M (Suspected) Fake Stars in GitHub by cortesoft on Thursday January 02, 2025 @10:11PM (#42581921)
Been using GitHub since literally the year it started, and stars have always been bookmarks.
Re:4.5M (Suspected) Fake Stars in GitHub by Asraelite on Friday January 03, 2025 @02:39AM (#42583466)
It reminds me of how subscribing on YouTube and other platforms turned from "inform me when new videos come out" into "I like this channel".
Re:4.5M (Suspected) Fake Stars in GitHub by lenkite on Thursday January 02, 2025 @11:02PM (#42582277)
> In this case this is a misunderstanding of the purpose of GitHub stars by these new users

Were we supposed to use stars some other way ?

Re:4.5M (Suspected) Fake Stars in GitHub by jasonjmcghee on Friday January 03, 2025 @11:47AM (#42587168)
No. They are bookmarks.
Re:4.5M (Suspected) Fake Stars in GitHub by verst on Friday January 03, 2025 @01:16AM (#42583041)
Let me expand on what I meant: Some folks treat GitHub stars as a repo quality or credibility endorsement.

I also think stars should just be bookmarks - but some companies are obsessed with the star count (stargazers) as a sign of importance of their repo. Since that is obviously what sparked this research we are discussing to begin with.. just meant to point out that a lot of folks new to GitHub have no idea anyone could possibly care about the repo star count! :)

Re:4.5M (Suspected) Fake Stars in GitHub by saurik on Friday January 03, 2025 @12:20PM (#42587507)
I do not understand how these two ideas are somehow different (and I even say that as someone who thinks caring about star counts is harmful to both sides): if there were a way to track/surveil how many users had bookmarked your website in their browser, it seems like it would be a sane and even rational (if also annoying and harmful and short-termist) thing to use that as a signal on both sides, to both optimize for as a website as well as as a potential user to judge the popular interest in a website... it certainly is a shitty heuristic, but even I will admit that, usually, a website that is bookmarked by a billion users is likely more "important" than one that is bookmarked by only a few users?

But like, any indirect signal, by its very nature, both can be gamed and will be gamed once people care about it, and so you need to be extremely careful ever using such signals on an ongoing basis, certainly once too many people know about them; and, frankly, if you make such a website, you should be doing everything in your power to prevent such signals from existing in the first place (such as by not tracking what users bookmark in the first place; or, if you somehow must, then keeping that information to yourself and not exposing it to users... this is hard to have fortitude on, as it also requires not trying to use it yourself as a signal, as people will figure that out and optimize for it even if they can't see it).

Re:4.5M (Suspected) Fake Stars in GitHub by wslh on Thursday January 02, 2025 @08:12PM (#42581049)
That issue you are exposing is everywhere so if fraudulent is the word we can see that a lot of social media and Internet, in general, is also fraudulent. It is call-to-action and conversion everywhere.
Re:4.5M (Suspected) Fake Stars in GitHub by est on Friday January 03, 2025 @12:39AM (#42582849)
> I personally think its fraudulent

It's fraudulent only because Github use "stars" to rank popular repos.

I think a repo should be ranked by code quality and update activity rather than "stars"

Re:4.5M (Suspected) Fake Stars in GitHub by demaga on Friday January 03, 2025 @12:55AM (#42582940)
And which method should be used to determine code quality?
Re:4.5M (Suspected) Fake Stars in GitHub by est on Friday January 03, 2025 @01:39AM (#42583166)
If that's a problem, then what method should be used to rank repos?

If "stars" were the only metric we can find, then people should stop complaining "fake" stars.

Re:4.5M (Suspected) Fake Stars in GitHub by demaga on Friday January 03, 2025 @01:49AM (#42583209)
I'm with the other commenters on this issue. Stars are good enough as a marker, since you can use other markers (like # of contributors and recent commits) to make your own mind regarding quality of a repo.
Re:4.5M (Suspected) Fake Stars in GitHub by est on Friday January 03, 2025 @02:51AM (#42583547)
> since you can use other markers

Yesh, surely I can, but how about https://github.com/trending ?

It still uses "stars" are a sole marker, no? I don't think it's "good enough"?

Re:4.5M (Suspected) Fake Stars in GitHub by bn-l on Friday January 03, 2025 @02:29AM (#42583408)
Also tests, coverage and tests passing!
Re:4.5M (Suspected) Fake Stars in GitHub by PeeMcGee on Friday January 03, 2025 @01:04AM (#42582974)
An easy one would be to promote projects that look like actual useful projects that contain code. Most of the ultra-starred things I see are awesome lists or otherwise non-useful markdown listicles and blogs.
Re:4.5M (Suspected) Fake Stars in GitHub by saagarjha on Friday January 03, 2025 @06:29AM (#42584705)
Definitely not stars lmao. Why would you want to use a measure of publicity as a measure of quality?
Re:4.5M (Suspected) Fake Stars in GitHub by hulitu on Sunday January 05, 2025 @04:01PM (#42605000)
> Definitely not stars lmao. Why would you want to use a measure of publicity as a measure of quality?

It is Microsoft.

Re:4.5M (Suspected) Fake Stars in GitHub by michaelt on Friday January 03, 2025 @08:44AM (#42585585)
It's simple.

Code quality is (Number of stars + number of forks * median number of commits per fork + 10 * number of closed issues + 100 * number of open issues + 3 * number of dependent packages + 0.01 * number of installs in the last 30 days) * number of commits in the last 100 days * number of core contributors active in the last 7 days / percentage of lines of code with no test coverage / (1+the number of open CVEs in dependencies) * the vibe factor