The cosmic microwave background is a faint glow left over from the Big Bang, like the cooling embers of a campfire. It encodes gravitational waves that may have arisen in the first fractions of a second, when the universe inflated faster than light. But between us and this ancient light hangs the dust of our galaxy. It, too, emits polarized radiation — light whose waves oscillate in a particular direction. And the result is a mess: you look at the sky, and there you have a mix of the dust signal and the cosmic microwave background signal.

Scientists have tried removing the dust before, but the maps came out noisy: where there is little dust, random noise overwhelmed the weak signal. The new method works differently. Instead of looking at each point individually, it looks at how dust forms filaments and clumps, like a spider web in the corner of a room. To do this, it uses a mathematical tool that decomposes the image into textures of different sizes and orientations. Then it fits a dust map whose statistics match the Planck telescope observations. It's like assembling a puzzle without knowing the picture, but knowing the pieces must form a certain pattern.

A test on a bright patch of sky showed that the new method recovers fine details of dust clouds better than older maps. This matters because fine details are exactly what could masquerade as a gravitational-wave signal. If you confuse them, you might mistakenly think you've found traces of the Big Bang when it's just dust.

Now astrophysicists have two sets of maps: one that accounts for random noise, and another that is deterministic, showing only reliable structures. Both will help subtract the dust from the data and get to the clean cosmic microwave background. And then, perhaps, we'll be able to see how the universe inflated in its first moments — faster than light could cross it.