But Pearls book explicitly argues for using correlational evidence to infer causation. The trick is to actually have a causal model and design observational studies that rule out as many alternative explanations as possible.
Thank you! That seems to be the issue with many commenter here, they believe you can have a single experiment that somehow proofs causality.
The same scepticism would not be brought against many "hard science" experiments, even though they do the exact same thing, falsify alternative explanations until they have high certainty that they have causality.
