> Still, I'm confused by the thinking behind policies like this, where it's not a practical problem of AI code creating too much low-quality work to review
That is the problem in these subreddits.
Up until 2024, if someone took the time to write and publish an open source project you could use that as a sign that they were trying to participate in the community and that it could be worthwhile to look at.
In 2026, that assumption no longer holds. You can’t tell if someone put 1000 hours into a project or 15 minutes of asking Claude Code to make a thing and then suggest copy to post to Reddit.
For a while, Claude was even telling people to go post projects to Reddit as a way to get users. They would copy and paste the suggested post into the Reddit.
So this created a backlash against all things LLM. The presence of any LLM tells in a project became a sign that it was not worth considering.
That is the problem in these subreddits.
Up until 2024, if someone took the time to write and publish an open source project you could use that as a sign that they were trying to participate in the community and that it could be worthwhile to look at.
In 2026, that assumption no longer holds. You can’t tell if someone put 1000 hours into a project or 15 minutes of asking Claude Code to make a thing and then suggest copy to post to Reddit.
For a while, Claude was even telling people to go post projects to Reddit as a way to get users. They would copy and paste the suggested post into the Reddit.
So this created a backlash against all things LLM. The presence of any LLM tells in a project became a sign that it was not worth considering.