Do we have a tool that can detect whether a project is majority LLM-written? This is an unenforceable rule, unless it's just coming down to vibes and they're going to decide what projects to ban based on their feelings.
Are you sure? It's pretty easy to find telltale signs of LLM usage, like the CLAUDE.md file, the '.cursor' directory, the 'vim'-sized codebase done in a week.
That's just vibes. Measuring how much data someone uses is concrete and empirical. They are implementing TOS rules that are based on judgement and require a judge. Very different when implemented practically, especially in terms of effort on the part of the org. Quotas are automated, evicting projects is highly manual.
They are not. You can probably draft some query to get candidates based on those criteria. And so very quickly.
> Quotas are automated, evicting projects is highly manual.
The human world is very chaotic. Automation is not a panacea, which is highlighted by Big Tech abysmal support. Sometimes, you want humans to judge, because the optimal result is based on heuristics and not rules written in stone.
Non-profits like Codeberg are resource-constrained. Apple is delaying app reviews and degrading the reviews that are done because of a slew of AI-generated app. Codeberg is basically adding a "human must review things" step to solve the "we have too much traffic" problem. As a bonus, they are instituting a process that's not only costly, but is vulnerable to being co-opted to ban projects that are not desired for any reason. It's just a naive policy mistake. Time will tell. Let's revisit in 2 years.