Light from SIMP 0136, a free-floating planet, if you split it into colours, a technique called spectroscopy, looks like a tangle of threads. Each thread quivers as the object spins. Its overall photometry waxes and wanes. Machine Learning (specifically, principal component analysis, which finds the directions of strongest variation in data) teased out two main movements from this flicker.

One movement is a uniform warming or cooling across all colours. What changes is the exoplanet atmosphere temperature. The other: some colours brighten while others dim. That’s clouds: as they rise, certain gases hide; as they sink, they emerge into the light. cosmic dust particles in those clouds act like a curtain.

This simple machinery, just two factors, was hiding behind variability once thought complex. Computer numerical simulation models of atmospheres confirm: the same two factors work in theory, too. Now it’s clear: weather on these solitary worlds obeys only a few rules, so it’s easier to predict. JWST let us peer into the heart of this kitchen, and it’s surprisingly tidy.