PERSON
Karl Pearson
The Victorian mathematician who invented the statistical apparatus that modern machine learning runs on—the correlation coefficient, chi-square test, standard deviation, and the positivist doctrine that description is all there is—and who then applied his methods, with terrible logical consistency, to rank the worth of human beings, making him the most uncomfortable and unavoidable intellectual ancestor of the AI age.
Karl Pearson is the intellectual ancestor nobody wants and everybody has. Between 1893 and 1912 he built, almost single-handedly, the statistical toolkit that every data scientist reaches for without thinking: the product-moment correlation coefficient, multiple regression, the chi-square goodness-of-fit test, the standard deviation, the Pearson distributions. He gave statistics the ambition to be a universal grammar of knowledge—a method that could be turned on heredity, on biology, on society, on anything at all. Contemporary machine learning is, at its mathematical core, a
Pearsonian intelligence: a correlation engine of staggering scale, performing the same basic operation of finding which things vary together and acting on that finding, at a speed Pearson could not have imagined. But Pearson also gave statistics a philosophy that the causal-inference movement, led by
Judea Pearl, now indicts as a hundred-year wrong turn: