I think that someone with a good knowledge of reinforcement learning should give his opinion about exploitation versus exploration. First comment is about being strangled into a local maximum.