I really liked The Social Construction of What? and this sounded pretty neat. It was, but it was hard, certainly harder to follow than TSCOW. Like that book what is fun here is the teasing apart of normally bound-together strands (astronomy and measurement error, physiology and the "normal" state of organs, positivism and its social physics, etc. etc.). Hacking's historigraphic method, which is to focus mainly on exemplary but obscure 19th-century figures in European state data-collecting and their infights and grand schemes, is good for lots of laughs. And I got a clearer sense of the "state" in statistics: as Hacking memorably points out, statistics are typically collected by US about THEM: they, the workers, the poor, the peasantry, etc., need us, the enlightened administrators, to study them and assess what can and can't be done for them. This leads in one direction to the idea of the "average man" or the normal and in the other to a eugenicist interest in improving the distribution so as to fight regression to the mean. There is a sinister episode from an 1870s Prussian debate about the statistics of Jewish immigration. Etc.

The final chapters, starring Durkheim, Galton, and CS Peirce, seemed rather bizarre to me. Hacking seemed to be saying that they exemplify a late-19th-c convergence on the idea of autonomous statistical laws of the social and even physical world, laws that were not reducible to underlying deterministic processes. Many things are normally distributed just because: the "taming of chance" is the historical development by which such things are accepted as law-like rather than as contingent violations of law. I guess it's interesting to see how Durkheim the quantitative sociologist of suicide can be connected to Durkheim the theorist of the "social fact" through this particular idea of the statistical. But Hacking claims this is the Birth of the Modern Era (as far as thinking about probability and chance goes). Do I think about probability that way? Does anyone? Except at the quantum level where Bell tells us things really are stochastic, I figure "random" error is just a heuristic assumption for analyzing evidence using statistical tools.

Oh well. I guess it served the Socratic purpose of making me even more uncertain about what I know, and statistics is the science of uncertainty…