Dagon comments on Probabilities, weights, sums: pretty much the same for reward functions