Problem w the Q* algorithm is that it always solves for the greatest reward.

No consideration of tangential, possibly cascading consequences outside the parameters of the equation's focus.