5 N = ∀X: Probabilization and the Taming of Machines

In the final pages of The Taming of Chance, Ian Hacking sums up the work of the philosopher Charles Sanders Peirce as a two-way affirmation of chance. First, Peirce, following the work of the psychophysicist Gustav Fechner and before him the astronomersociologist Adolphe Quetelet, ontologically reconfigured the normal curve.1 The “personal equation,” the variation in measurements made by any observer, becomes “a reality underneath the phenomena of consciousness” (Hacking 1990, 205). Peirce's belief in absolute chance or a stochastic ontology, “a universe of chance” as Hacking puts it, continued a series of realizations of curves, in which astronomical, social, biological, and finally psychological variations were all understood as generated by processes of chance. Second, and to show the underlying reality of the normal curve, “Peirce deliberately used the properties of chance devices to introduce a new level of control into his experimentation. Control not by getting rid of chance fluctuations, but by adding some more” (Hacking 1990, 205). In the century or so since, what happened to the thorough-going affirmation of statistical thought and probabilistic practice epitomized by Peirce? Hacking stresses that he does not understand Peirce as the precursor or innovator of 20th-century statistical thought (Hacking's Taming of Chance ends at 1900) but rather as “the first philosopher to conceptually internalize the way chance had been tamed in the nineteenth century” (215). What would the equivalent philosopher-machine learner internalize today? What would such persons, working in science or media or government, hold firm in relation to chance, probability, and statistics?