Rohan N Rajmohan of Northwestern University and colleagues from University of Chicago, Oak Ridge National Laboratory and IBM Quantum have developed a new framework for analyzing quantum noise that moves beyond standard models, revealing that the induced channel depends on which stabilizer eigenspace is chosen as the codespace. The researchers derive a tractable expression for the noise-averaged logical infidelity, accurately modeling error even when noise levels are high. This work reveals that, unlike traditional stochastic Pauli error models, the induced channel is affected by the selected codespace for encoding quantum information. Exploiting this discovery, the team introduces “PROSE” (Protected Stabilizer Eigenspace) encoding, a strategy for actively selecting the optimal codespace to suppress errors, and demonstrates that this eigenspace can be efficiently identified in many relevant situations; the results offer a new, broadly applicable lens on correlated coherent noise in stabilizer codes. Accurately predicting quantum error rates, even with substantial noise, represents a major step forward in building practical quantum computers. Researchers affiliated with the Department of Physics and Astronomy at Northwestern University, the University of Chicago, and IBM Quantum have derived a tractable expression for characterizing how correlated coherent errors impact stabilizer codes, offering a means to assess logical infidelity, a measure of how faithfully quantum information is preserved, without relying on approximations valid only for weak noise. This expression is non-perturbative, remaining accurate even as noise levels increase, a significant improvement over existing models. Demonstrating the practicality of PROSE, the team showed that this eigenspace can be efficiently identified in many relevant situations. Further analysis revealed that noise correlations, often assumed to be detrimental, can actually be harnessed; with the right encoding, even positive correlations reduce the logical infidelity below the uncorrelated baseline. This suggests a potential pathway for mitigating noise by strategically leveraging its characteristics, rather than simply