More thoughts on assertions
Read the original on LessWrong →
Summary
A careful Bayesian analysis of when an assertion is evidence. On a non-controversial topic, an unsupported assertion is strong evidence (no reason to lie that Uriel Sebree is today's Wikipedia feature). On a controversial topic where you already know the base rate of belief, an average person's assertion adds nothing — it's already priced in. On an extremely unusual topic (otter-people from Neptune control the World Cup), one sincere endorsement slightly raises your estimate of how many believe it (though most of the probability mass leaks into 'the speaker is insane'). Authority sharpens this — the 'Robin Hanson Effect': a credentialed person endorsing an idea you'd thought too fringe for even one PhD shifts you, and HOW you found them matters (Googling 'PhD who supports X' tells you only that one exists; randomly meeting one at a bar tells you they're common) — with the World's-Top-Expert / Watson case as the limit. Finally, 'and I have evidence' adds no factual weight but is a valuable sanity signal (it marks the speaker as the kind of person who knows beliefs need evidence — the physicist-Christian friend; the Dilbert crystal-healing counter-signal).
Why this score
Quality 74 · Strong. Strong — an idea-dense, careful treatment of assertion-as-evidence with several genuinely useful, exportable concepts (base-rate-already-priced-in, the Robin Hanson Effect, how-you-found-them, evidence-claim-as-sanity-signal). Substantive and original; a touch list-like and dated by its race-IQ running example, but sharp throughout.
Claude’s paradigm shift 48 · Moderate. Notable — the case-by-case Bayesian decomposition of testimony, the Hanson Effect, and the evidence-claim-as-signal are fresh framings building on signaling theory.
Real-world impact 2 · Minor. An idea-dense Bayesian treatment of when an assertion is evidence, with several exportable concepts (base-rate-already-priced-in; the Robin Hanson Effect; how-you-found-them; assertion-as-sanity-signal). Conceptual influence within rationalist discourse, no material change — low RWI.