I might actually be able to say something about the computational complexity of achieving von Neumann-Morgenstern coherent goals