The disruption, manipulation, and most of all erosion of trust which bad-faith actors create (AI or otherwise) are the most serious issues in my experience. The concept of "evaporative cooling", in which a platform starting to go bad sees an accelerated departure of well-intentioned, high-value participants.
See the original article from 2010: <https://web.archive.org/web/20101012105003/https://blog.bumb...> (HN discussion: <https://news.ycombinator.com/item?id=1777665>).
And an interesting follow-up, 2015: <https://blogs.cornell.edu/info2040/2015/10/14/the-evaporativ...> (HN discussion, from 2025: <https://news.ycombinator.com/item?id=42597962>).
Also, uh, how does speeding it up that help with the AI spam? (Sorry, I should have emphasized that part of my most more, since that's mostly the topic of this sub-thread I was replying to.)