Companion to the tour-de-force IRB review From Oversight To Overkill (ACX-441, 83), and one of the richest companions of the run. The book's author chimes in (IRBs are a data-free zone; small-science vs big-pharma). Clinician horror stories land hard — the 6-page consent form for a zero-risk 3mm skin sample, and a devastating account of a 'strict' local-IRB site whose PI ignored dangerous liver labs for months while the IRB merely acknowledged receipt (strict review = illusion of safety, not safety). Parallel vetocracy from tech/defense (Google SRE error budgets as a two-sided safety/velocity mechanism). Strong analytical threads: CatCube on decision/wrong-decision inseparability, why safetyism expanded as neoliberalism deregulated, the can't-sue-an-IRB question, the PLMNS-vs-utilitarian act/omission analysis (and its Mengele-signoff dark corner), and Scott's sharp rebuttals of JDK's 'blocking research doesn't CAUSE death' quibble (the surgeon-clamp and food-ban hypotheticals). Closes with the NIMBY-analogy correction and IRBs-for-AI.

Quality 73 · Strong. Strong (73, top of the companion band, above the 72 companions). Exceptional density of original Scott reasoning (the causation rebuttals, the vetocracy/act-omission analysis) plus vivid, load-bearing field testimony; a comment-response post riding on a landmark parent, which caps it below the parent's tier.

Claude’s paradigm shift 54 · Moderate. Moderate (54). Sharpens several real frames (illusion-of-safety, act/omission, causation-by-prevention) but assembles them from the discussion.

Real-world impact 1 · Negligible. 1 — a companion commentary; negligible independent real-world effect.

Highlights From The Comments On IRBs