Rule consequentialism - after you get around the collapse objection with Hooker, (see SEP for acceptance vs. compliance vs. following distinction) how can you make sure the rules it prescribes both (a) give determinate answers in all situations, and (b) really do put the world on anything like the maximum EV path(s). The worry is what rules constrain action in cases where the optimal action can’t be achieved - or generally wont be achieved - by agents consciously following rules, or having previously accepted rules. In these situations, it seems the rules will either say nothing about what to do. Or they will be made very wide in scope to cover these cases via a mechanism like precommitment, which necessarily won’t be very sensitive to what actions to take in these situations.