A new study published in Physical Review Letters demonstrates that robust information storage is more complex than previously understood. Researchers used machine learning to discover multiple new classes of two-dimensional memories capable of reliably storing information despite constant environmental noise, moving beyond the traditionally known Toom's rule. The research reveals that noise can sometimes *stabilize* memories, and that standard theoretical models often fail to predict the behavior of these systems, highlighting the importance of fluctuations. This work has implications for quantum error correction and understanding how robust behavior emerges in complex systems.
