For years, the quantum computing community has benchmarked its progress using simplified, independent noise models, leaving us to navigate the complexities of large-scale devices largely by feel. This commentary analyzes Google's recent out-of-time-order correlator (OTOC) experiments on the 65-qubit Willow processor (Nature 646, 825–830, 2025) and proposes that the primary value of the technique is as a new diagnostic tool rather than a pure quantum-advantage benchmark. The method, combining OTOC measurements with a Pauli insertion protocol, provides a direct, quantitative probe of many-body quantum interference. Two quantitative findings anchor the argument: (1) a simple independent-error model undershoots the observed SNR by roughly an order of magnitude (χ²/ν ≈ 14.2; experimental SNR 3.9 vs. model 0.4), and (2) the Pauli-insertion protocol differentiates small-loop and large-loop interference, with 1-ρ ≈ 0.003 for OTOC⁽¹⁾ versus 1-ρ ≈ 0.445 for OTOC⁽²⁾. We score these results against Gil Kalai's 2016 and 2023 predictions about correlated noise, finding direct experimental signatures for most of his claims. This follow-up to the author's 2024 analysis of Google Willow reframes a quantum-advantage result as the first scalable stethoscope for correlated noise in large quantum processors.
out-of-time-order correlator · OTOC · correlated noise · quantum error characterization · Willow processor · superconducting qubits · Pauli insertion protocol · Hamiltonian learning · Google · Kalai