Quantum error correction
Also known as: QEC, quantum error-correcting code, surface code, fault-tolerant quantum computing, fault tolerance
Quantum error correction (QEC) is the set of techniques that protect quantum information by encoding one 'logical' qubit across many physical qubits, so errors can be detected and corrected during a computation.
While it sounds like the error correction in your phone or hard drive, quantum versions are much harder because you can't copy qubits or look at them directly without disturbing them. Reaching reliable QEC is widely seen as the turning point for useful quantum computing.
Quantum error correction, explained at three levels
At home
Imagine sending an important message by having five friends each memorize it. If one forgets a word, the others can correct it. Quantum error correction works similarly, spreading information across many qubits so mistakes can be caught and fixed.
At work
Today's quantum computers make too many errors for most useful tasks. Error correction is the technology that could change that, and progress in this area is one of the best signals of when quantum computers will become practical.
At home
You can't just copy a qubit to make backups, because of a rule called the no-cloning theorem. Instead, error correction uses extra qubits to check relationships between data qubits, revealing whether an error happened without revealing the actual data.
At work
QEC requires many physical qubits per logical qubit, often estimated at hundreds to over a thousand for surface codes at realistic error rates. It also needs fast classical processing to decode error signals in real time. Vendors' roadmaps now focus on logical qubits and logical error rates, not just physical qubit counts.
At home
In the surface code, data qubits sit on a grid, with 'measurement' qubits checking groups of neighbours repeatedly. The pattern of check results, called the syndrome, lets a decoder infer where errors likely occurred. In my opinion, it's a remarkable trick: you learn about errors without ever measuring the information you're protecting.
At work
Code choice affects overhead: surface codes need only local connections but many qubits, while quantum low-density parity-check (qLDPC) codes, favoured by IBM's roadmap, can reduce overhead but need longer-range connections. Fault-tolerant computation also requires implementing non-Clifford gates, often through costly 'magic state distillation.' Resource estimates for breaking RSA or simulating chemistry depend heavily on these engineering choices.
Real-world examples
At home
Fatima followed quantum computing news and noticed headlines shifting from 'more qubits' to 'better error correction.' She learned that error correction could make quantum computers reliable enough for real problems, which mattered more than raw qubit numbers.
At work
Marcus is a cybersecurity planner at a bank. When assessing quantum risks to encryption, he tracked error correction milestones and logical qubit counts rather than physical qubits, because breaking encryption requires large numbers of error-corrected qubits.
How it compares
| feature | Quantum error correction | Classical error correction |
|---|---|---|
| Can copy data | No (no-cloning theorem) | Yes |
| Error types | Bit flips and phase flips | Bit flips |
| Overhead | Often hundreds of physical qubits per logical qubit | Small |
| Detection method | Indirect checks (syndromes) without measuring data | Direct parity checks |
| Maturity | Early demonstrations | Ubiquitous |
Quantum error correction solves a much harder problem than classical error correction, with far more overhead.
For analysts
If you're tracking quantum computing for strategy, watch logical qubits and logical error rates. I'd treat those as the real milestones, since physical qubit counts alone say little about useful capability.
History & origin
Many physicists in the early 1990s doubted quantum error correction was possible. In 1995, Peter Shor published a nine-qubit code that could correct arbitrary errors on one qubit, and Andrew Steane introduced another code in 1996. The threshold theorem, developed by several groups in the late 1990s, showed that if physical error rates fall below a certain level, errors can be suppressed arbitrarily by adding more qubits. Alexei Kitaev's work led to the surface code, now a leading approach. In December 2024, Google announced that its Willow chip had shown logical error rates falling as the code grew larger, a key 'below threshold' milestone. In my view, that result marked the field's shift from asking whether QEC works to asking how fast it can scale.