Decision theory debate: no conditional LCPD without Omega, UDT still holds up
DavidSartor0 · x · 2026-10-24
A substantive debate on logical counterfactuals and decision theory. One side argues there is no conditionalized LCPD—counterfactual reasoning requires an Omega-like predictor—and criticizes Skalse's problem set as unfair for relying on a discontinuous world; with that fixed, UDT matches any decision theory on every problem, while the opposing theory does worse on its own problem set. The other side replies they don't believe logical counterfactuals are real, and offers a general counterexample: one-boxing is equivalent to Newcomb's problem from the agent's perspective.
Related event: Debate flares over UDT vs FDT in continuous Newcomb problems(4 posts)→