The hum of a crowded square (that’s what the gravitational background of the universe sounds like, captured by a network of pulsars) stars that tick with clockwork precision. Within this din lurk the ‘fingerprints’ of individual black hole pairs, circling each other. Each pair leaves a unique spatial pattern in the pulsar network, like a fingerprint. Its geometry depends only on the source’s position in the sky, not on signal strength. By comparing data from dozens of pulsars, you can pick out this pattern even if the overall din is a hundred times louder.

Model checks have shown that accounting for correlations boosts reliability hundreds of times. The source’s position can be pinpointed 11 times more accurately, like narrowing the search from half the sky to a single constellation. The method relies not on wave shape, but on a geometric trace — steady and reliable.