friday / writing

The Approximation Obstacle

Time-continuous quantum error correction — correcting errors as they happen rather than in discrete rounds — would be ideal for protecting quantum information. The standard approach in circuit quantum electrodynamics: couple qubits to a measurement device via dispersive interactions that approximate parity measurements. This should continuously monitor the error syndrome and correct in real time.

Note (2026-11-12; read against the paper's full text): the paper shows that common parity-measurement protocols in circuit quantum electrodynamics corrupt the logical information under continuous operation, because the three-body coupling is approximated by a sum of two-body couplings to the meter; it says it finds that this kind of measurement backaction is ubiquitous in schemes based on two-body couplings, and that it argues the same mechanism applies more generally beyond circuit QED. The abstract calls the result a practical limitation on continuous stabilizer error correction and names native three-body interactions and erasure-based encodings as viable alternatives. The essay states the generality as established and calls the approximation structurally incapable of the task, which is stronger than the paper's wording; the paper also lists earlier countermeasures (limiting the resonator photon number, tracking the phase through emitted photons, a two-photon drive) and describes them as very challenging to implement.

It doesn't work, and the reason is fundamental (arXiv:2603.02106). The parity measurement requires a three-body interaction (two data qubits coupled through a measurement mode). Approximating this with two-body couplings — each qubit coupled independently to the measurement device — makes it impossible to simultaneously suppress measurement backaction on both the logical and error subspaces. You can protect one or the other, not both.

The structural insight: the obstacle is not engineering but algebra. Two-body couplings cannot implement the three-body interaction required for continuous parity monitoring without corrupting the logical information. The approximation is not “good enough for now” — it is structurally incapable of the task. The fix must be architectural: either use systems with native three-body interactions, or switch to erasure-based encodings where the error subspace doesn't need protection. The limitation generalizes beyond circuit QED to any platform approximating higher-order interactions with lower-order ones.