@faithisleaping fucking preach. Genuinely it’s so frustrating that entire projects give into this thoughtless dogma.
I remember a time when #FOSS was a break from the corporate bullshit. Now? It feels like the very enshittification we’re trying to escape from big tech is finding its way into FOSS. I despair.
Trouble is, it’s not like this is particularly easy to undo once it takes hold. Idk what the fuck these people are doing at this point.
@QueerMatters The relationship between FOSS and corporate bullshit has always been a complicated one. In many ways, I don't see that as having fundamentally changed. I've been dancing that tango for a long time and it's still going.
But what gets me is the people who think this is some great gift to open-source. Like the ability to have code on GitHub was the goal. Putting the code on GitHub was never the goal of FOSS, as anyone who's had the displeasure of working with an Android kernel tree can attest. Just because someone pushed it to GitHub doesn't mean it's high quality, well-tested, or well-integrated with upstream.
I also have very little patience with the people who think it's somehow going to solve long-term maintenance issues or maintainer burnout. (see also Linus and lots of Linux kernel maintainers.) It's not. If no one is digging the hardware out of their box and testing and fixing things now, having an LLM isn't going to fix that. It just makes it easier to do massive, probably sketchy refactors of questionable quality.
@QueerMatters Yeah, I've never bought that argument. Just because someone says a policy isn't enforceable or because it's difficult to enforce doesn't mean you can't put the policy on the books. Even having it there is often enough to deter the worst of the slop contributors. And if you have it there and are able to determine through the review process that they're just throwing slop at you, you can justify closing the PR on sight.
And if an LLM PR is genuinely indistinguishable from a human-written PR with equal code quality, etc. then accidentally accepting a few isn't going to trash your project. (And if it does then you clearly don't have high enough review standards.) But I've yet to see an LLM PR that's indistinguishable from a human or is at the level of coding that I expect from contributors. It's a theoretical loop-hole, but keep trying, guys!
But neither of those should stop anyone from setting reasonable policy. The "We can't do anything. Don't try." is a psyop.