Fisher’s main scientific and statistical argument boils down to the possibility that gene variants that contribute to lung cancer are tightly linked with gene variants that make nicotine more or less addicting. By “linked with” he meant “close together on the same chromosome”. This kind of linkage can lead to strong a statistical association but without a mechanistic association (guilt by neighborhood).

Disproving this hypothesis is tricky, and Judea Pearl does a brilliant job of explaining the problem and its solution in his marvelous book: The Book of Why.

Fisher gets “assist points” for debilitating and killing millions, although full horrible credit goes to cigarette companies and their advertising co-conspirators.

Judea Pearl points out one cruel irony: The cholinergic receptor gene CHRNA5 that modulates risk of nicotine addiction also modulates lung cancer risk separately. To sort out the causality we now use Mendelian randomization.

Bottom line: smoking cigarettes does kill even when you tidy up the statistics and genetics.