Nicolas Lacombe comments on Reward is not the optimization target