San FranciscoAI generates plausible posts at nearly zero cost.
Volunteer moderators reviewing each post manually cannot match the throughput, burning out the human filter these communities depend on.
Developers warn no platform has deployed AI-volume detection; niche technical forums are losing maintainers.
Sources: Hacker News
THE ASYMMETRY
Generating a plausible forum post with an LLM costs fractions of a cent and takes seconds. Reading it carefully, checking sources, and deciding whether to remove it takes a human moderator minutes. The cost ratio between producing slop and filtering it has never been this lopsided.
WHO ACTUALLY MODERATES
Reddit, Stack Overflow, Wikipedia, and most large open-source forums run on unpaid volunteers. Reddit's 2023 IPO filing acknowledged the platform depends on roughly 60,000 unpaid moderators; Wikipedia's English edition runs on a few hundred active administrators. The labor that keeps these sites usable is invisible in their revenue.
THE EVAPORATIVE COOLING EFFECT
Communities decay through a pattern Eliezer Yudkowsky named in 2009: when quality drops, the highest-contributing members leave first because they have the most options elsewhere. Each departure lowers the average further, accelerating the next departure. AI slop triggers this loop directly — experts leave before novices notice anything is wrong.
THE STACK OVERFLOW PRECEDENT
Stack Overflow banned ChatGPT-generated answers in December 2022, six days after the model's release. Question volume still fell roughly 50% over the following two years. The site demonstrated that even successful detection cannot restore a community once contributors lose confidence the audience is human.
WHY DETECTION DOESN'T SCALE
Statistical AI-text detectors achieve roughly 80% accuracy on long passages and degrade sharply on short ones. At forum scale — millions of posts — even a 5% false-positive rate flags more legitimate users than a volunteer team can appeal-review. The math forces moderators toward heuristics (account age, posting cadence) that the next generation of bots routes around.
THE COMMONS PROBLEM
Open-source forums, Wikipedia, and technical Q&A sites are digital commons — non-excludable resources maintained by a small share of users for everyone's benefit. Garrett Hardin's 1968 framing predicted such commons collapse under unrestricted extraction. LLM training scraped these commons for free, and the resulting models now flood them with low-effort output, completing the cycle Hardin described.