The Census Is a Distribution of Power

A census presents itself as the least political act a state performs. It counts people. There is a true number and the task is to find it.

The statute directing it does read that way, and the framing is what makes the exercise so consequential, because a count that is understood as measurement is not argued about the way an allocation would be.

Every allocation downstream is arithmetic on the count, and arithmetic on a contested number inherits the contest without appearing to.

The apportionment rule converts population directly into legislative seats. Nothing intervenes. A change in where people are counted is a change in where power sits.

The formula is fixed and the inputs are not, which locates the entire political question upstream of a calculation nobody disputes.

Undercounting is the mechanism, and it is not random. The populations hardest to count are consistently the same ones: people who move often, who share housing, who have reason to avoid official contact.

A systematic undercount of a particular kind of household is a systematic transfer of representation away from the places those households live, effected without anyone deciding it.

The programme documentation is candid about the difficulty and about the methods used to address it, which is more than can be said for the political discussion around it.

The technical debate about statistical adjustment is genuinely hard, and it has been conducted almost entirely as a proxy for the distributional question underneath.

Adjustment corrects a known bias using established methods. It also, unavoidably, shifts seats and money, which is why arguments about its statistical validity track political interest so closely.

Both sides are being sincere about statistics and both are aware of the consequences, and the debate is unresolvable because it is not really about statistics.

The residence rule is the same problem in a different form. Where a person is counted, rather than whether, decides which jurisdiction receives their representation.

Students, seasonal workers and incarcerated people are each counted somewhere by a rule that is defensible and that produces very different maps depending on which defensible rule is chosen.

The count is also a single event with no second chance. A survey that goes wrong can be repeated; a decennial census that goes wrong is the operative number until the next one, and every institution downstream treats it as settled fact.

The implementing regulations settle these questions in language that reads as administrative housekeeping, which is precisely how consequential choices survive without scrutiny.

Funding follows the same path. A large volume of federal money is distributed by formulas that take population as an input, so an undercount is a permanent budget cut nobody voted for.

The cut lasts a decade, because the number is fixed for a decade, and it compounds: fewer resources in an area make the next count harder, which is a feedback loop with no correcting force.

What makes all of this durable is the ten-year interval. A decision taken once and binding for a decade attracts a fraction of the attention that an annual decision of the same magnitude would.

By the time the consequences are legible, the next count is closer than the last one, and the argument restarts from the beginning with no accumulated record of what the previous choices produced.

There is a further asymmetry in who notices. A jurisdiction that gains from a methodological choice has no reason to examine it, and a jurisdiction that loses often lacks the technical capacity to demonstrate that it lost.

The reform worth having is not a change to any rule. It is a published statement, before each count, of what each methodological choice is expected to do to representation and to money.

That would not resolve the disagreements. It would make them arguments about distribution, which is what they are, rather than arguments about measurement, which is what they are conducted as.

A count is a fact. What is done with it is a decision, and a decade of allocation currently rests on the pretence that only the first of those exists.