Rootclaim’s point-by-point rebuttal of Scott Alexander’s ACX piece. Central claim: Scott’s 94%-zoonosis conclusion turns entirely on treating p(Huanan-Seafood-Market-cluster | Lab Leak) as a ~500x Bayes factor for zoonosis, when Rootclaim argues the correct factor is ~2x (markets recurrently form early clusters after any pandemic, lab-leak included); correcting this one number allegedly flips Scott’s own tables to ~94% lab-leak. Also argues the furin-cleavage-site “coincidence” is properly ~3x10^7-to-1 against a natural origin, and that judges’ and other raters’ extreme zoonosis-favoring Bayes factors reflect a systematic failure to “steelman” the lab-leak hypothesis (the piece gives a semi-formal account of this “strawmanning” failure mode). Contains the source passage for the case’s “~23 orders of magnitude” framing: “Peter, Saar, and the two judges all did their own Bayesian analysis… Daniel Filan… did one too” (quoting Scott’s own ACX post) — i.e., the six are Peter Miller, Saar Wilf/Rootclaim, judges Will van Treuren and Eric Stansifer, Scott Alexander, and Daniel Filan; removing Peter’s (self-described “trolling”) numbers still leaves ~7 OOMs, and removing Saar’s leaves a factor of ~50.
relevance_note: the primary source for the case’s “six estimates spanning 23 orders of magnitude” claim, and Rootclaim’s fullest post-loss methodological defense.