Reinforcement Learning on the Tech Tree