Comment by BadBadJellyBean
4 days ago
The problem I see with the change in their rules is that they have created an unenforceable or worse arbitrarily enforced rule.
There is no sure way to identify LLM code. I could write a project by hand with all the GPT "mannerisms" for fun. I could sign my commits with "Co-Authored-By: Claude" because my cat's name is Claude and I find it funny.
Now this rule will either be never applied or when it's applied it will be an arbitrary decision made with mostly inconclusive "evidence". This is not a good way to write rules.
Write rules based on merit. Excessive data usage or licensing problems might come to mind. These things are measurable. Otherwise it's really just posturing.
Do the other rules bother you the same way? Like the one about racial discrimination? I mean, there's no surefire way to identify racial discrimination.
Racial discrimination can have fringe cases but is at least something you can often clearly see. You can not detect the intention but you can often see the outcome. But yes. Other rules that are unenforceable without reading tea leaves are also a problem.
It's lucky that the rules aren't enforced by a computer program, isn't it?
2 replies →
Also, the there are hundreds of ways to use these tools. From semi-intelligent autocomplete to fully agent-driven. Where's the line? What if I had an LLM clean up a fairly long stored procedure? Is that single action now forbidding me from ever uploading that code base to Codeberg? What if no other part of my code was written using LLMs?
> Now this rule will either be never applied or when it's applied it will be an arbitrary decision made with mostly inconclusive "evidence".
Or people just lie. Which is the whole problem with not having a strong way to determine code's origin. And it's also a tad ridiculous because if done correctly, the seams between my thoughts and those of the agent shouldn't be apparent. If it's utilizing existing patterns in my codebase that I've written by hand, then you wouldn't even know.