Comment by globuous | Hacker News Reader

Comment by globuous | Hacker News Reader

Parent Full thread

I mean, github does too. Some projects have 0 stars, some have 16k; some have 0 contribs, some have thousands; some have no forks, some have hundred. You know what I mean ? All these variables are a pretty good measure of recognition, I check them every time I add dependencies to my projects. And just like any other measure, it isn't perfect and is hackable in ways I haven't imagined yet.

Forks are similar to citations in a way. It's not the same, there are many differences, but there are some similarities in my opinion. By the way, wouldn't it be great to be able to contribute to papers, and their assets (figures, code, equations) ? I don't know how a system like that might work in practice, it must be non trivial to govern, like any other sizable community - but I can see how it could work.

Does science really need to pay for recognition and prestige today ? It used to be the these journals allowed more people to have access to quality research, now it restricts them. Non free journals are dead a parellel universe Nietzsche might write.

The number of stars on GitHub is significant because most people whose opinion you’d care about are on GitHub, they wouldn’t star project that they were not interested in and there aren’t widespread fake accounts.

I don’t believe that there is an equivalent in academia, one that would capture the h-index dynamic (copied by PageRank) that a star from a prestigious professor is worth a lot more. It’s trivial to build, but “growth hacking” for lack of a better word, is hard, especially in areas where actual growth hacking would be frowned upon.

That’s why, for instance, Facebook celebrates its billion of users so much: it’s genuinely hard to make that. Hosting them too, but not in the same way.

In case you're interested: I'm working with eLife and the Center for Open Science on creating something just like that. Having those two organisations as partners has really boosted the profile, leading to others jumping on board.

There's still ways to go, but I think we're on the right track. But you're right: it's really hard, and it took me a year to even get here.

You've made an argument for looking through the projects which people whose opinion you care about have starred, not an argument for why GitHub stars are significant. Simply looking at number of stars doesn't provide you with much meaningful information.

It's also worth considering when someone starred a project. In my experience, people don't usually won't go and update their old stars, so it shouldn't be taken as endorsement. You don't know what state the project was in at the time it was starred. It's common for project to evolve over time, and it's not always for the positive.

The significance of starring a project is also poorly defined. Many people use GitHub stars as a form of bookmarking. You generally don't know the reason why someone starred a project. For example, just because someone has starred a library doesn't mean that they'd use it in production.

> The number of stars on GitHub is significant because most people whose opinion you’d care about are on GitHub, they wouldn’t star project that they were not interested in and there aren’t widespread fake accounts.

I am on Github and have been active in FOSS for years and yet I rarely star any projects at all, Github stars to me seem like vanity (just like likes on Facebook) indicative of perhaps hype but not of value.

Reply on news.ycombinator.com