Hacker News Reader

Comment by zarzavat | Hacker News Reader

It's hypocrisy. People will defend entire books and research papers being shared on libgen/scihub, which is unarguably actual copyright infringement on a massive scale, but training an AI on open source code is somehow the worst thing ever even if there's no case law to say that this constitutes infringement at all.

View on HN

It's not really that hypocritical when you understand peoples perspective on it. In general people (here on HN/OS-community) care about sharing experiences and knowledge. Copyright on opensource content does not inhibit peoples ability to learn from it. Theres also the whole "big corp vs little guy" mentality at play here. If copilot was opensource then I don't think that anyone would have an issue with it, I actually think people would respond well to it if that were the case.

It's not hypocrisy to say that different types of work should have different copyright lengths, and that for some types the answer is zero.

Especially if you remember the phrase "to promote the progress of science and useful arts".

It's self-serving hypocrisy when programmers argue for a very strict reading of copyright for their own works (source code) while at the same time arguing for unbounded sharing of other people's works (books and papers).

If you agree with libgen and scihub then you should definitely not see anything wrong with copilot, because what copilot is doing is considerably milder, they are not reproducing anybody's works.

If you disagree with libgen and scihub because they are illegal only then can the argument that copilot is infringing be made.

> It's self-serving hypocrisy when programmers argue for a very strict reading of copyright for their own works (source code) while at the same time arguing for unbounded sharing of other people's works (books and papers).

Yes, if those were the only two things being discussed, it would be.

But almost all of these people think a certain amount of copyright for songs and movies is good too. And I bet a vast majority of them support copyright for books too, or at least most books.

The vast majority of research authors want their work to be freely available. The industry is just organized in a way that prevents that from happening without harming your research career.

I'm all for copilot if microsoft gets treated the same way libgen/scihub are for creating it. or if we abolish copyright, but the fact that they can just decide to do this and it's fine, but scihub gets DNS-blocked reveals the asymmetry at hand here.

Reply on news.ycombinator.com