I agree. There is something to be said about autonomous systems producing valid outcomes within the current constraints of RL