Pragmatic FDT, and predictors as game theory

Short summary

Author proposes pragmatic FDT (p-FDT), a framework avoiding formal definition pitfalls through constructible algorithm equivalences. Argues that predictor-based counterfactual reasoning blurs decision theory into game theory, explaining divergences between EDT/TDT/UDT/FDT. Reframes classical decision-theory debates as fundamentally game-theoretic problems of responding to predictors.

• •Proposes pragmatic FDT sidestepping formal definition problems via constructible equivalences instead of abstract algorithm matching
• •Shows predictor counterfactuals transform decision theory into game theory, resolving why different frameworks appear irrational in edge cases
• •Connects historical EDT/TDT/UDT/FDT debates to the blackmailer-payment problem framed in predictor terms

Generated with AI, which can make mistakes.