How much more likely is ¬correlation -> ¬causation than correlation -> causation?
You can make a MC analysis with Erdös-Renyi directed graphs and linear (polynomial? (linear + reLU?)) functions over this, pick two random nodes, correlation, causation, bam
