DisprovedChanged2024

Adding more qubits only makes a quantum computer more error-prone

The old claim turned out to be wrong and was rejected by evidence.

TaughtQubits are so fragile that the more of them you string together, the more errors pile up. Error correction was a beautiful theory, but in practice every extra qubit added more noise than it removed, so a useful, scaled quantum computer stayed permanently over the horizon.

NowIn December 2024 Google reported in Nature that its Willow chip crossed the error-correction threshold: as it went from a 3×3 to a 5×5 to a 7×7 grid of physical qubits, the logical error rate was cut roughly in half at each step. For the first time, adding qubits made the encoded information more reliable, not less.

What actually happened

This is the result the whole field had been waiting decades for. Error correction says that if your physical qubits are good enough, grouping many of them into one 'logical' qubit should suppress errors exponentially as the group grows. The catch was always the 'if': below a certain quality threshold, bigger groups just accumulate more faults.

Willow demonstrated operation below that threshold: each time the code grew, the error rate fell by a factor of about 2.1, and the logical qubit outlived its best individual physical qubit. That is 'below threshold,' and it is the hinge the theory hangs on.

The honest caveats: this is one logical qubit's memory, not a working algorithm, and a practical machine needs many such qubits held stable together. Nobody at Google claims the finish line. What was disproved is narrower and real: the belief that scaling up necessarily makes things worse.

Newsletter