Newcomb's Paradox is sometimes considered a contrived scenario designed to make CDT perform "badly", since most real-world interactions do not involve near-perfect predictors like Omega[citation needed]. However, the basic conclusions resulting from it it still hold even if Omega has a nontrivial advantage over random chance. This is realistic: when interacting with each other, humans frequently attempt to predict the actions of other agents and take different actions depending on those predictions. Humans use crude special-case approximations to the underlying accursed decision theories, such as trust, anger, vengeance and honor, but decision-theoretic frameworks allow for a useful conception and generalization cleaner than the notion of "useful irrationality".
A screaming bird of truth.