Hacker News Reader

Comment by intunderflow | Hacker News Reader

FYI that Codeberg is currently holding a vote to broadly ban projects written mostly using AI, so its not a neutral space for hosting your projects like GitHub: https://codeberg.org/Codeberg/org/pulls/1253 https://tldr.nettime.org/@tante/116880003584050912

View on HN

I've been on Codeberg for my projects since September but will pull them if this happens. I see no definition of what "mostly using AI" means and refuse to be "punished" for the choice of tooling I use. The tone and content of the thread on Mastodon is absolutely toxic.

I'd rather self-host than deal with the ambiguity and threats. Invariably from people who don't even make anything.

I have plenty of concerns around "AI" and project governance and IP law. My projects are copyleft, I'm in favour of free software, bleah bleah bleah.

But this just looks like pitchforks and mob behaviour.

Same here. If this policy is actually implemented, I’ll move my projects elsewhere—perhaps to GitLab or a self-hosted solution. I certainly don’t want a hosting provider policing which development tools I can use.

Same here, i've been on Codeberg since 2022 but if this passes, i'll most likely move my stuff elsewhere (my projects do not even have any AI code, i just dislike the idea of telling people what tools to use to make their own stuff).

I dislike GitHub and i'm using a shared host for my website which wouldn't let me install something like Forgejo (this is by choice as i don't want to bother with sysadmin stuff, my site is almost all static HTML with only a couple PHP scripts for some minor tasks). So i guess i'll be migrating to GitLab, even though i do not really like its UX. Or maybe i'll use some other Forgejo instance like CodeFloe. URLs aside, migrating between Forgejo instances should be easy.

Yep. I have a self-hosted Gerrit instance that I was using for other more private stuff. I could in theory move over to there.

The downside is then having to worry about backups. But also lack of all the other "forge" things like issue tracking and so on. And lack of public exposure.

Frustrating, really. I felt welcome on Codeberg up until now. But that Mastodon thread is... wow.

If I understand it correctly, their target is projects like the recent pgrust (PostgreSQL in rust) which are full conversions of projects that no-one else will ever use. It's just noise.

Proposed language only says:

  7. You must not share projects that mostly consist of code written by "generative AI"-tools (including services such as *Claude*, *OpenAI Codex*).

Who defines "mostly"? And what's the definition of "written"? If I'm heavily prompting, curating, reviewing, and editing after the agent has gone through, is that somehow now "allowed?" Who makes the decision and how can I appeal?

Almost everything I write now has some agentic aspect to it, and I'm by far not the only one. That wasn't the case when I started my projects but it is now.

But I'm a software engineer with 25-30 years of experience and a computer nerd since 1979 when they wheeled the Apple II into my Grade 1 classroom. I'll reserve the right to decide what tooling I use to write my copylefted free software.

Given the Mastodon thread linked above I don't trust the people involved to be the judge jury and executioner. Insults flew freely and moral absolutism with it. I'll just pull my project off their infrastructure rather than be subject to their personal whims.

Well, that's a good thing. Being "neutral" _is_ taking a side. It means you're taking an amoral stance even in big questions.

Every organization has a stance. We're just become used to companies that take a stance of "as long as we get paid".

No, being "neutral" is not "taking a side", that's the definition of the word. If you pre-moderate you open a whole can of worms. What will they ban next? This makes me not want to use Codeberg ever because it's not plannable.

Do you feel this way about everything? What about pre-moderating CSAM?

Clearly some things are harmful enough that they should be banned.

For pre-moderating you need some sort of upload filter and people correctly went to the streets in the EU against article 13.

If you're going to jump ship because they're banning AI slop, then I think the feature is working as intended.

Why do you think you can judge my code and how I program from a comment?

Calling all uses of such tooling "AI slop" right away shows your hand as only holding a blunt, crude, instrument.

Many of us are capable of more nuance that that and don't apply the label "slop" to everything where generative AI is involved. The linked Mastodon shows no nuance, and neither does your reply.

Two things can be true at once: GitHub has degraded into a "slop" cesspit and that agentic tooling has legitimate use cases.

Finally people can also legitimately object to a host mandating what tools can be used in authorship even if they don't particularly care for those tools themselves.

Reply on news.ycombinator.com