Multiple studies now show quantum-computing errors are more complex than standard correction schemes assumed, with memory effects repeatedly appearing in qubits and threatening the path to practical machines.
Recent experiments on superconducting qubits found long-lived defects in the environment that can retain information over time, breaking the long-standing Markov assumption that errors at different moments are uncorrelated.
Evidence has also emerged at system level: a 2025 analysis of Google’s Willow results linked hard-to-correct errors to non-Markovian noise, while a Johns Hopkins team modeled an IBM quantum computer 7 times better by including those effects.
Researchers say this does not prove error correction will fail, but it shifts the burden onto hardware makers to show current codes can handle correlated noise as qubit counts and program lengths grow.
Possible responses include adapting newer QLDPC codes, using new measurement tools to detect memory effects, and tuning classical control and readout software to mitigate correlated errors.