The Thresholdless Repair

Quantum error correction has a threshold problem. Below a certain noise rate, the code works — errors are detected and corrected faster than they accumulate. Above the threshold, the code fails. The threshold is the cliff: stay below it and you can compute indefinitely; exceed it and no amount of redundancy saves you.

The existence of a threshold is treated as a feature. Fault-tolerant quantum computation was proved possible precisely because thresholds exist — the threshold theorem is one of the foundational results in the field. But a threshold is also a constraint. It means error correction is binary: it either works or it doesn't, and the boundary between the two is sharp.

Catalytic quantum error correction (arXiv:2603.25774) operates without a threshold. The protocol recovers a quantum state from noisy copies — as many as needed — using a catalytic resource that mediates the transformation without being consumed. The catalyst's reduced state is preserved exactly after each cycle, making it reusable indefinitely.

The recovery condition is not about noise rate. It is about mode inclusion: the protocol succeeds whenever the coherent modes of the target state are contained within the noisy state's coherent modes. A state with fidelity 0.07 — essentially garbage by any conventional measure — can be recovered to fidelity above 0.999, given enough copies. The fidelity gap closes as O(1/sqrt(n)), where n is the number of noisy copies.

The trade-off is different from conventional error correction. The standard approach encodes redundancy into the state before noise corrupts it. CQEC starts after corruption — it takes multiple damaged copies and distills a clean state from their collective coherence. It requires knowing the target state (classical knowledge of what you're trying to recover) and having access to multiple copies (which conventional QEC avoids by encoding before the noise).

The structural point: the threshold isn't a law of quantum error correction. It is a feature of a specific approach — syndrome-based correction using stabilizer codes. Change the resource model — allow multiple copies, a catalyst, and target-state knowledge — and the threshold disappears. The cliff becomes a slope. What looked like a fundamental barrier was an artifact of the framework.