REACTOR
R15 Sociology Sociology · REV.3

SELF-TEST OK · REACTOR v3 · LOADING [ R15 ]…

The Birth of Statistical Thinking

The normal person is a nineteenth-century invention.

You probably think "normal" and "abnormal" are a line that was always there, waiting to be discovered. The counterintuitive truth is that the idea of the normal person was manufactured by nineteenth-century statistics. How so? For a line called "normal" to work, you first need a ruler: an average, so you can measure how far you have strayed from it. Before the nineteenth century, people had no notion of average height, average lifespan or average crime rate. Without those averages, there was no "normal" line to measure deviation against in the first place. Almost every ranking, indicator, average and question of the form "is this reading normal?" rests on that nineteenth-century invention. It is the first layer of foundation for seeing through any number. N00 has a line: you think you are reading the number, but the number is reading you. Where the number came from is the first half of the answer.

The story starts with a flood of numbers. The philosopher Hacking named it the avalanche of printed numbers: over roughly the two decades from 1820 to 1840, European governments suddenly began printing tables obsessively, counting and printing everything, population, births, deaths, disease, suicide, crime. In twenty years the stock of countable social facts exploded from a scattered handful to hundreds. This was not because people had fallen in love with arithmetic, but because states had a new thing to govern, called population. But a state cannot see population directly: what it faces is tens of millions of individual people, too many for any official to look over one by one. Foucault pointed out that a state is therefore forced to switch to a different way of seeing: seeing through statistics. Birth rates, death rates, incidence rates: once those numbers are laid out, population becomes, for the first time, an object with its own regularities that can be managed. The order matters here: it was not that a clearly delineated society sat waiting to be counted, but that once these numbers existed a governable society became visible. Counting something is itself an act of power, because what you count and what you do not decides what becomes visible and what stays in the dark.

Who first aimed all this at people? Hacking's The Taming of Chance says it was the Belgian Quetelet. He did something bold for his time: he took the bell curve that astronomy used to describe measurement error and applied it to human beings. Astronomers had long known that measuring the same star many times gives results scattered in a bell shape, high in the middle and low at the tails, with the true value in the centre and error on either side. The true value here has a very concrete meaning: it is the star's most likely actual position, an objective fact. Quetelet found that human height and weight scatter into that same bell shape. He carried over astronomy's meaning wholesale: if the centre stands for the true value for a star, then the centre of this curve for people should also be something real, and he named it the average man. The weight of that step lies in a quiet inversion: the bell curve originally described error, the noise we did not want; in Quetelet's hands the average man at the centre became the ideal and the standard, and deviation on either side became what needed explaining, or even correcting. The line between normal and abnormal was drawn by a mathematical curve that had originally described error.

Something else shocked people at the time even more. Once the suicide and crime rates printed in that avalanche of numbers were laid out year by year, they turned out to be frighteningly stable: in a society of strangers, the number of people who killed themselves or committed crimes each year barely changed, as if an invisible hand were setting quotas. So people came to believe that society, like nature, has iron laws that can be studied as physics is, and Quetelet even called it social physics. The belief that numbers reveal hidden laws is seductive, and dangerous: it lets people forget that these regularities only appeared once there were unified statistical definitions that filed wildly different ways of dying into the single slot "suicide." How much of a regularity belongs to the world and how much to the slot has been tangled from the very start.

Once a category is up, it quietly reshapes the people it holds. The information scholars Bowker and Star studied classification systems and called them the scaffolding of information infrastructure, invisible in normal times yet twisting people's lives: every standard privileges one perspective and suppresses another, and the racial classifications of apartheid South Africa are the starkest example, with real people forced into slots that did not fit them and violent consequences. Hacking described a more active side too, called making up people: a new classification opens new ways of being a person, and some ways of living simply cannot be lived until they are named. The average man works the same way. It was manufactured first, and then the line called "normal" began disciplining people: you compare yourself against the average, you feel anxious about deviating, you find ways to move yourself toward the middle. A point that was merely the mathematical centre became a ruler pressing on everyone.

One open question: in the case of the average man, Quetelet started from a real distribution of heights, and the centre really does correspond to something. But when we force an average and a "normal" onto universities with different missions or models with multidimensional abilities, does the centre still correspond to anything real, or is it purely the result of forcing incomparable things onto one ruler and then pretending the midpoint means something? That question goes straight to commensuration (R03).

The one-line takeaway: normal was not discovered, it was computed by nineteenth-century statistics and then handed to us to live by.

Sources / further reading
  • Hacking, I. (1990). The Taming of Chance. Cambridge UP (Quetelet and the average man, the normal curve from error to people); Hacking, I. (1982). "Biopower and the Avalanche of Printed Numbers." Humanities in Society 5:279–295 (the avalanche of printed numbers).
  • Desrosières, A. (1998). The Politics of Large Numbers: A History of Statistical Reasoning. Harvard UP (statistical categories constructed and stabilised first, organising reality after).
  • Foucault, M. Security, Territory, Population (the 1978 "Governmentality" lecture; statistics making population visible).
  • Hacking, I. (1986). "Making Up People"; (1999). The Social Construction of What? Harvard UP (interactive kinds are both constructed and real); Cooper, R. (2004). "Why Hacking is Wrong about Human Kinds." BJPS 55(1) (the other side).
  • Bowker, G. & Star, S. L. (1999). Sorting Things Out: Classification and Its Consequences. MIT Press (classification twisting lives).
  • Intellectual history: research/02 §3, §6, §8, §9; the looping controversy and the open question about AI as a new interactive kind: research/deep/D1 §4.
Where to next