Nothing here is a black box.

On the limits of counting

A number is a kind of light. In a dark room it gives you the shape of a thing: its size, its direction, the speed it moves. It gives you none of its colour, and none of its grief.

This atlas is built almost entirely from counts, so it is worth saying plainly what a count cannot hold. “Not everything that can be counted counts, and not everything that counts can be counted,” wrote the sociologist William Bruce Cameron in 1963. The line is so reliably hung on Einstein, who never said it, that its own false attribution has become a small lesson in measuring the wrong thing. A death becomes a rate per 100,000; a life becomes a point on a line that does not know its name. The chart buys its clarity at exactly that price, and the price is never printed on the chart.

Counts deceive upward as readily as down. A global average is a mercy and a lie at once: it can climb serenely while a single country burns beneath it, can report a comfort assembled out of someone else’s catastrophe. “The map is not the territory,” Alfred Korzybski warned, and one world line is the smallest, palest map into which eight billion separate lives have ever been folded.

And still we count, because the alternative is worse. “All models are wrong, but some are useful,” the statistician George Box allowed, and the whole of this site lives inside that second clause. Take the numbers away and only anecdote and mood remain, and mood is the exact product a decade of headlines is built to sell. A count is a lamp raised in a large dark room. It will never show you the room. It will keep you from walking into the walls, and now and then it catches, for a moment, the shape of a door.

So we try to keep faith with both at once: to count carefully, cite everything, show the working, and to leave confusion standing as a reading in its own right, never a footnote. The most honest thing a measurement can offer is sometimes only this — that here, at the edge of the lamplight, it can no longer see.