gwern comments on Causal Diagrams and Causal Models