FDT debate: are dependency hypotheses 'logical'? Two researchers clash
OrionJohnston · x · 2026-11-10
A technical debate over functional decision theory (FDT): the quoted poster argues FDT's dependency hypotheses are inherently "logical" — the causal graph contains logical nodes, and interventions target those nodes to model subjunctive dependence on logical facts, unlike CDT graphs which can omit logical nodes entirely.
The author pushes back: the hypothesis that both his and Omega's actions depend on his "choice" is not logical in any meaningful sense. He models choice as specifying a function X->Y, assumes Omega instantiates another function with a particular dependence on it, and only then derives their match — the dependency hypothesis is pre-logical, with logical derivation downstream of it.