Quantum error correction normally runs on a fixed clock: stabilizer checks, called syndrome measurements, repeat at a steady rate. This paper shows that, for quantum memories whose job is to hold quantum information, the time between those measurements is a tunable control. Measure too rarely and passive “idling” errors build up. Measure too often and noisy measurement circuits create faults. The authors find a sweet spot for the measurement interval and show that choosing it correctly can reduce logical errors very strongly as the code size grows.

To make the trade-off tractable, the researchers introduce a simple, phenomenological noise model. In it, idling errors grow with the waiting time Δt roughly like p_idle ≈ p λ Δt for small base error rate p and a rate multiplier λ, while each measurement round itself causes faults with probability p_stab ≈ p. Readout flips are modeled as p_read = b_read p. Using this input they propose a compact formula (an ansatz) for the logical-error rate per unit time and show analytically that the best measurement interval scales inversely with the code distance d (roughly Δt* ∝ 1/d). Choosing Δt in this distance-dependent way produces an exponential reduction of logical-error rates compared with any fixed-interval recipe.

They also treat time-dependent idling noise, such as short bursts where the idling rate jumps briefly. For that case they design an adaptive timing strategy that changes Δt on the fly based on measured syndrome activity. Practically, the method monitors how often syndrome bits flip and uses a simple log-likelihood test (smoothed by a moving average) to declare whether a burst is occurring. Analytically and in simulations of rotated surface codes with a matching decoder, the adaptive scheme beats every fixed-interval protocol. The gains are largest for short, strong bursts: for example, the paper reports improvement factors like about 1.86 for a burst-ratio r=10 and 2.73 for r=20, and nearly a twofold reduction in logical failure for a distance-15 memory under realistic time-dependent noise.