It's a bit amazing to me how quickly we have all become experts on what is generated text and what is not. Part of it seems to be that the (extreme?) politeness of much generated text seems inhuman. Do humans have an edginess that LLMs cannot impersonate well?

I know that HN has a policy against generated text, and I learned about the policy by violating it unknowingly. A comment of mine voted to about 80 was then rapidly downvoted after @dang explained the policy to me.

On the one hand, I often prefer a good summary by an LLM to reading a lot of comments, and I like to share what I learn. On the other hand, banning generated text helps prevent a takeover by bots, and keeps the mission of HN focused on humans. After reflection, I found myself in favor of the policy. I can take what I learn from an LLM and write what I've learned in my own voice.