News for hackers. Stuff that matters.
Re:Research recitation: A first look at rote learning in Git... by nullc on Sunday July 04, 2021 @12:19AM (#27727194)
The flip side is that if, from a policy perspective, we decide that the licensing doesn't matter-- then it's extremely likely that anyone else can make a copilot competitor.

If we decide that licensing is required, it's likely that no one except github (or a few other huge players) could ever make something like this just due to access to the code.

(Github would make it a requirement of the TOS that your code is licensed to allow copilot to use it, and require users to indemnify github from third party legal action arising from github using code you posted in copilot.)

The permissive handling levels the playing field.

Re:Research recitation: A first look at rote learning in Git... by dannyw on Sunday July 04, 2021 @04:27AM (#27728112)
no, if github does that the entire open source community will move to GitLab, much like the freenode to libera exodus.
Re:Research recitation: A first look at rote learning in Git... by nullc on Sunday July 04, 2021 @01:40PM (#27731307)
Already their TOS gives them broad permissions, it just stops short of indemnification. Their TOS arguably already turns people posting projects containing third party copylefted code into license violators, -- yet people are still using them.

So I wouldn't be so sure-- but moreover, I think my larger point remains: the entire world being fragmented into separate licensing is a huge moat.