In this study, we have explored some of the key metrics that users look at in a GitHub repository.
207 users participated among which 139 users are active open-source contributors, 57 are looking to hire developers and 28 are not developers.
- People looking to hire developers mainly look at commit history and readme
- A small section of developers dive into the codebase directly without looking at the documentation
- Readme and stars are the first information that most users notice
Which of the following metrics matter to you?
Options given: stars, forks, contributors, commits, watchers, readme, commit history, code
What do you look at to get a better understanding of codebase?
Options given: commit history, current master branch, pull requests, issues, documentation
See previous study details and participate in current study here
Are people really more concerned about the number of stars, forks and contributors a project has, and not the code?
I would love to see this broken down by language.
Maybe most people are just downloading libraries, themes or plugins (VS, VSCode and Vim plugins are almost all hosted on Github) that have the most stars.
This makes sense to me considering that forks and contributors represent support, stars represent interest (eg ‘it appears to solve a problem many people have’) and readme answers ‘does this solve a problem that I have?’
Imagine if people looked at the openssl code before using the library – we might not have anywhere near the https adoption we do today.
Also, for libraries like openssl (eg shared C libraries) the important thing for many people is what the top level api looks like, in other words “how easy is this to call from <my language>?”
People are indeed unpredictable but I think this equally has to do with how GitHub tries to shape repositories too.
I believe this result signifies the fact that open-source development at GitHub has a social touch in it. Personally, I enjoy exploring new projects at GitHub as well.
I wonder how will this study turn out if it considers repositories on GitLab, Do users of GitLab significantly different from users of GitHub?
I believe most GitLab users may be professional developers in contrast to GitHub which serves a wider audience.
I would love to see a study comparing some metric between GitHub and GitLab.
This is interesting
I see 33.81% of users are concerned about code and of it, 18.84% of users take a look at pull requests of which I believe most will be contributors. So, the inference is that:
nearly 6.36% of the users of the code contribute to it
Seems fair enough
But its a sad truth that people are looking at the stars of Repo
I think considering the Readme is the best option
Yes, Bhupesh, it is a sad truth as well as misleading to some extend in the sense, that some popular projects are abandoned or better alternatives rise but the stars give a different impression and sometimes, readmes are not modified as well or we miss it.
Truth be told: Stars do represent that the project is or has been popular and useful
For instance, earlier this year, I spent time getting myself acquainted with the Torch framework and to my surprise, the development of Torch has been paused. It may be for good and thus, I had to make a switch to TensorFlow.
I am not sure how we can avoid such a scenario
This sometimes seems similar to geeting likes on FB
karmas on Reddit and retweets on Twitter
Well said , this reflects that GitHub is a social network.
In fact, I see funny comments on issues and rise of some fun repositories occasionally. GitHub is a funny and productive place
Many people think that star is similar to “like” button on FB. However, its main intension was to bookmark the repository. It’s true that my some the projects on GitHub didn’t get popularity as they were having fewer stars and when I asked some friends to star it, I started receiving the response from many strangers. That’s why I created a repo so that I can help some projects to get some initial stars and popularity. But it is having no star