Debates

In CF, a debate is a focused discussion aimed at a specific point of disagreement. When two people hold contradictory beliefs, at least one is wrong, and (unless someone is overconfident) both should want to learn which. Because its scope is narrow, debate needs less shared world-modeling and terminology-syncing than open-ended collaborative truth-seeking, so it works across larger cultural gaps and lower trust.

CF’s distinctive claim is that debate is the prime social mechanism for error correction. A public intellectual open to collaboration but not debate can stay wrong his whole career; a debate policy — a law-like document specifying who, what, and how one will debate — gives critics a guaranteed, bias-resistant path forward and lets the answer to “if I’m wrong and someone knows it, how will I be corrected?” be a real one.

Debate’s effectiveness depends on epistemology. Those who prefer collaboration often implicitly hold an indecisive view — e.g. Bayesians incrementally updating credences. CF’s decisive, pass/fail epistemology makes it realistic to reach conclusions fairly quickly: trace disagreements to premises, build discussion trees, and let small quantitative differences fall within a margin of error rather than forcing distinct conclusions.

CF does not glorify actual debate; much of it is toxic and time-wasting. The remedy is structure — methodology and impasse chains for resolving stalemates — not abandoning debate for collaboration alone. Crucially: if you cannot resolve a debate decisively in your favor, you shouldn’t be confident you’re right.