timtyler comments on Expected utility without the independence axiom