Review for NeurIPS paper: Forethought and Hindsight in Credit Assignment

Additional Feedback: major points: lines 234-246: what is "fan in" and "fan out", this paragraph doesn't explain what it is? Does it relate to neural network architecture (shown in Figure 1)? I have no idea what "large fan-in" and "small fan-out" means. Is channeling a bottleneck in the state space? If so, where is this?