Our preprint "Shared Actors Need Not Share Critics: Effects of Value Mismatch in Parallel Reinforcement Learning" is posted on arXiv.