Moderation Is Different From Censorship
Read the original on Astral Codex Ten →
Summary
Draws a sharp, clarifying distinction the social-media debate keeps eliding: moderation is the normal business activity of giving customers a product they like (a harassment-free feed), while censorship is a third party blocking an exchange that both sender and receiver want, to satisfy people in power. The racket conflates them ('we must remove offensive content for everyone to keep the platform safe'). His minimum-viable fix: do exactly the same moderation, but add an opt-in 'see banned posts' toggle — banned accounts stay visible to anyone who opts in, and to each other. Applied to China, this is transformative: a Weibo user could click one button for accurate Xinjiang/Tiananmen information, placing a floor on how bad the information ecology can get. He notes fancier versions (per-filter toggles, user-chosen fact-checkers) and that the current one-size compromise satisfies no one. There remain genuine arguments for true censorship (averting false ideas, non-idea content like CSAM/bomb-instructions, bad actors organizing) — but nobody debates those now because conflating censorship with moderation lets proponents argue for the easy case; de-conflate and offer moderation too cheap to meter, then debate real censorship on its merits.
Why this score
Quality 73 · Strong. Strong band, upper. A clean, important conceptual distinction with a concrete, actionable proposal (the opt-in toggle as a moderation/censorship firewall) and a vivid China illustration. Short, so upper-Strong.
Claude’s paradigm shift 48 · Moderate. Moderate. The moderation-versus-censorship framing and the opt-in-toggle MVP are a fresh, exportable contribution to the content-policy debate.
Real-world impact 2 · Minor. A clarifying conceptual distinction the social-media debate keeps eliding (moderation = giving customers a product they want; censorship = a third party blocking a wanted exchange), with a concrete opt-in-toggle proposal. Real topical relevance to the content-moderation debate, but a conceptual contribution informing it — low RWI.