The Politics of Large Numbers
Statistics as an instrument of state
Description
In the 1660s, an English shopkeeper named John Graunt sat down with the weekly bills of mortality — the parish lists of who had died in London, and of what — and started adding them up. Nobody had really done this before, not like this. Graunt noticed that the numbers, chaotic week to week, settled into patterns across years: roughly stable ratios of boys to girls at birth, predictable death rates, a rough shape to how a population aged and thinned. From a pile of clerical record-keeping, he pulled something that looked like a law. The individual death was unknowable; the mass of deaths behaved.
That small gesture sits near the start of a long story that the French statistician and historian Alain Desrosières told in his 1993 book, translated into English as The Politics of Large Numbers. His question was deceptively simple. We treat statistics as neutral — a mirror held up to a society that already exists, just waiting to be measured. Desrosières spent the book showing that it was never that. Counting a population required first inventing the categories that make counting possible, and building the administrative machinery to enforce them. The numbers and the state that produced them grew up together.
What makes the argument bite is that these tools didn't merely describe populations. They made populations into something a government could act on at all — a governable object, held still long enough to be taxed, drafted, insured, planned for. And the categories used to do the holding had a strange afterlife. Once built, they stopped feeling like choices and started feeling like facts about the world.
The question we’re asking : How did statistics turn a messy human population into something a state could actually see, hold, and govern — and what did the categories used to do it quietly become?What we’ll see : How counting became a technology of power, from mortality tables to the census, and why the boxes we sort people into end up shaping the reality they claim only to record.
Table of contents
01Chapter 1 — Before the state could count
For most of human history, rulers did not know how many people they ruled. They had a rough sense — enough men for an army, enough households for a tax — but no reliable total, no breakdown, no way to see the population as a single object. Desrosières begins here, in the gap between wanting to govern a territory and being able to actually perceive the people in it. A crowd is not a population. To turn one into the other, you need instruments, and those instruments did not yet exist.
The first push came from the practical needs of early modern states. In seventeenth-century England, the political arithmeticians — Graunt, and after him William Petty — argued that a country could be understood the way a merchant understood a ledger: in figures of births, deaths, trade, land, and labor. Petty even coined the phrase "political arithmetic" for the project of reasoning about the state in numbers rather than in rhetoric. The idea was radical precisely because it was so ordinary-sounding. Run the kingdom like an account book, and the sovereign gains a kind of vision he never had.
02Chapter 2 — Two ways of holding things together
At the heart of Desrosières's book is a distinction that sounds abstract but explains almost everything that follows. A statistic works by treating different things as the same — this death and that death, this worker and that worker, counted together as if they were interchangeable. He asks a blunt question about that move: what actually holds the box together? What makes it legitimate to say these units are equivalent? And his answer is that historically there have been two very different kinds of glue.
The first he calls realist. On this view, the category corresponds to something genuinely out there in the world — a real class of things that the statistician merely discovers and records. When we count deaths, we feel we are counting a real, natural fact; death is death. The realist stance treats the statistician as a describer, whose categories are true or false depending on whether they match a reality that exists prior to the counting. Much of the authority of statistics comes from this posture: the numbers feel objective because they seem to name things that were already there.
03Chapter 3 — The average man and the invention of society
In the 1830s and 1840s, a Belgian astronomer named Adolphe Quetelet took a tool from the study of the stars and pointed it at people. Astronomers had long used the theory of errors: measure the same star many times, and the scattered readings cluster around a true value, with the errors distributed in a predictable bell-shaped curve. Quetelet had the audacious idea that human traits behaved the same way. Measure the chests of thousands of soldiers, the heights of thousands of conscripts, and the figures scattered around a central value in exactly that curve — as though nature were aiming at an ideal and missing by varying amounts.
From this he built one of the most consequential ideas of the century: l'homme moyen, the average man. The average man was not any real person but a statistical composite — the mean of a whole population, treated as its true type. Individual variation became a kind of error around this ideal figure. And crucially, Quetelet extended the method to moral facts. Rates of marriage, of suicide, of crime turned out to be strikingly stable from year to year across a whole society, even though each act seemed like a free individual choice. The mass had regularities the individual did not.
04Chapter 4 — Categories that make the world they describe
Step back from the mortality tables and the average man, and the deeper claim of the book comes into focus. Statistics, for Desrosières, is not a mirror set beside society. It is one of the things that makes society hold together as a shared reality in the first place. To argue, to negotiate, to govern, people need objects they can point at and agree exist — the unemployment rate, the poverty line, the cost of living, the gross national product. These objects are what he calls a common language, and without them collective decisions could not be made. You cannot debate a policy on inequality without a statistical measure of inequality that both sides accept as roughly real.
The point sharpens once you notice how the categories loop back onto the people inside them. Build an official category of "the unemployed," attach benefits and programs to it, and you don't merely record a group that was already there. You create a status people can occupy, apply for, be counted in, and organize around. The socio-professional classifications that French statisticians built in the mid-twentieth century did not just describe existing classes — they gave people a grid through which to understand their own position, and around which unions, parties, and policies then formed. The category becomes a thing people live inside.
05Conclusion
John Graunt's bills of mortality look, from a distance, like the humble start of a technical craft — a shopkeeper tidying up parish records. What Desrosières traced across three centuries is how that craft became something else entirely: the machinery by which a scattered human population is turned into an object a state can see, hold still, and act upon. The average man, the census, the socio-professional grid, the unemployment rate — each was a way of building equivalence, of deciding that these units belonged in the same box, and each hardened over time from a convention into what feels like a fact.
Related books
