Planets forming in a disk should capture each other into orbital resonances — period ratios of 2:1, 3:2, 5:3. The mechanism is well-understood: convergent migration drives planets toward each other; dissipation from the gas disk damps their eccentricities; the combination traps them at resonance. Resonant chains should be the default architecture of planetary systems.
They are not. Most observed multi-planet systems are NOT in resonance. The Kepler mission revealed that near-resonant spacings are common but exact resonances are rare. Something breaks the chains.
Lin, Adams, and Zanazzi (arXiv:2602.20525) show that turbulence in the protoplanetary disk is sufficient. Stochastic forcing from disk turbulence lowers the eccentricity damping threshold for resonant capture. Below a certain damping efficiency, capture fails — the planets pass through the resonance instead of being trapped. Stronger turbulence raises the threshold further, eventually precluding capture at any damping rate.
The turbulence doesn't need to be strong. It needs to be persistent. The stochastic kicks accumulate over the long migration timescale, and their cumulative effect degrades the resonant capture probability. The architecture of a planetary system — how many planets are in resonance, which resonances survive — is fundamentally constrained by the noise level of the disk in which they formed.
The general point: the default expectation (resonant chains) assumes a clean environment. The observed reality (broken chains) reveals that the environment was noisy. The noise is not a perturbation on the architecture — it is a structural constraint that determines which architectures are possible. Quiet disks produce resonant systems. Noisy disks produce the scattered, near-resonant spacing we actually observe. The noise is as much an architect as the gravity.