This is exactly how are why statistical estimation is bullshit.

Correlation, if any, does not imply causation, and the causality principle has nothing to do with statistics.

Observations (counting), however many, will never reveal causality, because they are orthogonal.