Eighty-five percent of the map, half of the votes

The county map is the most reproduced political image in the United States and one of the least informative. Here is what is actually in it, from the same certified file the desk's county layer runs on: of 3,160 counties and county equivalents, 2,685 were carried by the Republican column and 475 by the Democratic one. That is eighty-five percent of the units on the map. Those 475 counties cast 49.2 percent of every vote counted in the country.

Nothing in that is a scandal, a fraud or a paradox. It is a description of where people live. A county is an administrative container drawn mostly in the nineteenth century for the convenience of a courthouse, and the containers were never sized by population. Loving County, Texas and Los Angeles County are the same object in the legend. One of them cast about sixty ballots.

The arithmetic of concentration is steeper than almost anyone guesses. Rank all 3,160 counties by turnout and start adding. You reach half of every vote cast in America after 162 of them. Start from the other end and the 1,482 smallest counties — very nearly half the map by count — add up to five percent of the electorate between them. An area map does not merely blur this. It inverts it, handing the loudest visual voice to the places with the fewest people in them, at a ratio of roughly ten to one.

The second thing the map hides is worse, because it hides voters rather than acreage. Every county is painted one colour, so everyone who lost inside it disappears. In the counties carried by the Republican column, 28.1 million people voted for the Democratic candidate. In the counties carried by the Democratic column, 27.8 million voted Republican. That is roughly 56 million Americans — more than a third of the entire electorate — rendered as the colour of their neighbours. There is no rural monolith and no urban monolith on that file. There is a country where nearly every county is a mixture and no map ever admits it.

This is not a partisan complaint, and it is important that it is not, because both columns have run the trick. Colour by land and you get the sea of red. Colour by county at national zoom and you erase the small-town half of one coalition; colour by state and you erase the metropolitan half of the other. Every one of those images has been posted triumphantly by people who would fail a five-question quiz about what it depicts.

The fix is old, cheap and sitting in every cartography textbook. Give every unit the same area and write its weight inside it — the tile cartogram the desk uses for the state board. Or keep the geography and scale the ink to population rather than to land. Or, best where the data supports it, drop the fill entirely and shade by margin so a county carried 51–49 stops looking identical to one carried 80–20. None of this is new and none of it is expensive. It is simply less dramatic, and drama is what an area map is actually for.

So the desk publishes the county layer with the margin shading and the turnout on the tooltip, and it publishes the district layer beside it, because 435 congressional seats are the units where legislative power is genuinely allocated and they are drawn — for better and frequently for worse — around people rather than around terrain. Drill into either one and the panel tells you how many live there before it tells you how they voted. That ordering is deliberate. The population is the denominator of every honest sentence you can write about a place.

None of which changes who won. The certified national margin is what it is, the electoral college is what it is, and no amount of better cartography moves a single vote. The claim is narrower and, we think, harder to argue with: an image that misleads in your favour is still an image that misleads, and a reader who understands the shape of their own country is worth more to any political project than a reader who has been handed a flattering picture of it.