Correlation is not causation …
… smart-asses often point out when someone mis-uses some statistical or perceived correlation. That’s true, and easy enough when potential common causal connections are obviously available, but that’s quite unusual in the “real world”.
Causation is fundamentally mysterious even in simple Newtonian billiard-ball cases, or our daily expectation of sunrise, but at the common sense level causation is really about empirical certainty of expectation and prediction – if this then that, whether precise “physical” mechanisms are or are not clearly available.
In the real world however mechanisms are rarely simple or precise, so causation is really a testable theory of why if this then that, where the why is controllable. But real life is never a repeatable experiment, except maybe with drosophilla, and rarely with humans. And in real life caustion may involve many variables over many space-time-scales. It’s complicated.
Correlation is useful even with negligible knowledge of causation. More knowledge of causation always helps. (Now, let’s read this:
“Big Data, epistemology and causality: Knowledge in and knowledge out in EXPOsOMICS” by Stefano Canali
(Hat tip Timandra Harkness.)