Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

GitHub's fake star economy

awesomeagents.ai

Post reply on HN
Print view View original

11–20 of 403 posts

Re: GitHub's fake star economy

#11

Post by test1235 »

3form wrote: post #9
dafi70 wrote: post #3

Honest question: how can VCs consider the 'star' system reliable? Users who add stars often stop following the project, so poorly maintained projects can have many stars but are effectively outdated. A better system, but certainly not the best, would be to look at how much "life" issues have, opening, closing (not automatic), and response times. My project has 200 stars, and I struggle like crazy to update regularly…

The stars have fallen to the classic problem of becoming a goal and stopping being a good metric. This can apply to your measure just as well: issues can also be gamed to be opened, closed and responded to quickly, especially now with LLMs.

"When a measure becomes a target, it ceases to be a good measure"

https://en.wikipedia.org/wiki/Goodhart%27s_law

Re: GitHub's fake star economy

#12

Post by 3form »

HighlandSpring wrote: post #7
dafi70 wrote: post #3

Honest question: how can VCs consider the 'star' system reliable? Users who add stars often stop following the project, so poorly maintained projects can have many stars but are effectively outdated. A better system, but certainly not the best, would be to look at how much "life" issues have, opening, closing (not automatic), and response times. My project has 200 stars, and I struggle like crazy to update regularly…

I wonder if there's a more graph oriented score that could work well here - something pagerank ish so that a repo scores better if it has issues reported by users who themselves have a good score. So it's at least a little resilient to crude manipulation attempts

It would be more resilient indeed, I think. Definitely needs a way to figure out which users should have a good score, though - otherwise it's just shifting the problem somewhat. Perhaps it could be done with a reputation type of approach, where the initial reputation would be driven by a pool of "trusted" open source contributors from some major projects.

That said, I believe the core problem is that GitHub belongs to Microsoft, and so it will still go more towards operating like a social network than not - i.e. engagement matters. It will still take a good will to get rid of Social Network Disease at scale.

Re: GitHub's fake star economy

#13

Post by Miraltar »

Topfi wrote: post #4

I don't know what is more, for lack of a better word, pathetic, buying stars/upvotes/platform equivalent or thinking of oneself as a serious investor and using something like that as a metric guiding your decision making process. I'd give a lot of credit to Microsoft and the Github team if they went on a major ban/star removal wave of affected repos, akin to how Valve occasionally does a major sweep across CSGO2 bann…

Citing Valve as a model for handling cheating is not what I'd have reached for.

Re: GitHub's fake star economy

#16

Post by foresterre »

dafi70 wrote: post #3

Honest question: how can VCs consider the 'star' system reliable? Users who add stars often stop following the project, so poorly maintained projects can have many stars but are effectively outdated. A better system, but certainly not the best, would be to look at how much "life" issues have, opening, closing (not automatic), and response times. My project has 200 stars, and I struggle like crazy to update regularly…

With the advent of AI, these "life" events are probably even simpler to fake than AI though, and unlike the faking of stars not against the ToS.

Re: GitHub's fake star economy

#17

Post by Se_ba »

dafi70 wrote: post #3

Honest question: how can VCs consider the 'star' system reliable? Users who add stars often stop following the project, so poorly maintained projects can have many stars but are effectively outdated. A better system, but certainly not the best, would be to look at how much "life" issues have, opening, closing (not automatic), and response times. My project has 200 stars, and I struggle like crazy to update regularly…

This is a good idea, but from my experience most VCs (I’m not talking about the big leagues) aren’t technical, they tend to repeat buzzwords, so they don’t really understand how star systems works.

Re: GitHub's fake star economy

#19

Post by sunrunner »

3form wrote: post #9
dafi70 wrote: post #3

Honest question: how can VCs consider the 'star' system reliable? Users who add stars often stop following the project, so poorly maintained projects can have many stars but are effectively outdated. A better system, but certainly not the best, would be to look at how much "life" issues have, opening, closing (not automatic), and response times. My project has 200 stars, and I struggle like crazy to update regularly…

The stars have fallen to the classic problem of becoming a goal and stopping being a good metric. This can apply to your measure just as well: issues can also be gamed to be opened, closed and responded to quickly, especially now with LLMs.

Was it ever a good metric? A star from another account costs nothing and conveys nothing about the sincerity, knowledge, importance or cultural weight of the star giver. As a signal it's as weak as 'hitting that like button'.

If the number of stars are in the thousands, tens of thousands, or hundreds of thousands, that might correlate with a serious project. But that should be visible by real, costly activity such as issues, PRs, discussion and activity.

Post reply on HN