People have started to value things like GitHub forks and stars as though they mean something. Create a metric and it will get abused, this is how the world works.
Stars usually mean your project is useful to someone, so they are indeed nice to see. They're also an endorsement of people who use a project and are apparently happy, so the project is more likely to do what it says and do it well.
These metrics aren't usually a problem until people attempt to game the system like it's done on platforms like youtube, but I'm not aware of this happening on github (at least not often).
Not surprising to me (I'm in India), whenever I've reviewed someone's resume and they have a GitHub it's always a few dark green dots on a few days in an entire year for several years (and often just 2 years).
Pretty sure it's done for no reason other than to check a box on some recruiting agency's form "I contribute to open source"
In the Wine project, which dates back to the mid-90s, we still use a mailing list for patch submission[1]. Every year we get some complaints that it's hard to use, and that we should switch to GitHub or set up a GitLab instance or something to encourage more submissions. And maybe we should. But it does keep out crap like this. Wine's a fairly high profile project, especially among "low-information" users. Would the increase in useful submissions justify the cost of having to deal with low-quality submissions? Are there any devs out there who both can hack on Wine and can't figure out an email client? I don't know.
basically: he has high expectations for the raw text formatting of a commit message and discussion threads, and github's web interface lacks the necessary controls.
Using something like GitHub or GitLab brings transparency to the process and that encourages contribution. You can still prevent low-quality contributions with some bot automation (like ensuring a template is followed etc). GNOME project has become more approachable for a lot of people since they moved to GitLab: https://about.gitlab.com/blog/2020/09/08/gnome-follow-up/ (Even though I don't have the right skills to "hack" into most of GNOME projects, I can read C code and can provide useful issue submissions, but would not do any of that if I had to subscribe to a mailing list.
I haven't done much open source contribution, but it seems like we should be able to write some kind of flagging/filter system to check if PR's are quality. One-liners and comment modification? Highly suspect. Although I am not sure what GitHub offers as far as PR management API.
20 comments
[ 2.7 ms ] story [ 39.9 ms ] threadhttps://github.com/search?p=5&q=is%3Apr+is%3Aclosed+label%3A...
Scanning through the usernames and clicking on a few accounts, over 90% seem to be from college students in India :(
People have started to value things like GitHub forks and stars as though they mean something. Create a metric and it will get abused, this is how the world works.
[1] https://en.wikipedia.org/wiki/Cobra_effect
These metrics aren't usually a problem until people attempt to game the system like it's done on platforms like youtube, but I'm not aware of this happening on github (at least not often).
[1] https://www.winehq.org/pipermail/wine-devel/2020-October/thr...
https://github.com/torvalds/linux/pull/17#issuecomment-56546...
https://github.com/torvalds/linux/pull/17#issuecomment-56599...
basically: he has high expectations for the raw text formatting of a commit message and discussion threads, and github's web interface lacks the necessary controls.
Just from what I am seeing on this Twitter, it seems pretty clear that there are some easily identifiable patterns. i.e. "Amazing". - https://twitter.com/ryan_southgate/status/131164593596180480...
I could be just completely out of touch with opensource.