Quantum error correction does not repeatedly measure each data qubit to ask whether it is 0 or 1. It encodes information across several physical qubits, measures selected relationships among them, and uses those measurement results—called a syndrome—to infer likely errors without directly reading the encoded quantum information.
How does quantum error correction work?
A physical qubit is a hardware-level unit of quantum information. It can be disturbed by effects such as unwanted fields or temperature changes, and the gates, measurements, and initialization used to operate it can also be faulty. Quantum error correction (QEC) addresses this by encoding one logical qubit across multiple physical qubits.
The encoding makes the information collective: it is stored in relationships among the physical qubits rather than in any one qubit by itself. Those relationships provide redundancy. A QEC system measures selected properties of the encoded group to check whether it still behaves as expected.
- Encode. Prepare physical data qubits in a code space that represents the logical information. The encoded state is distributed across the group.
- Measure checks. Ancillary measurement qubits interact with groups of data qubits to measure parity or other stabilizer values. The results reveal whether expected relationships have changed, not the logical value itself.
- Repeat the checks. Syndrome extraction is repeated to build a history of changes. That history helps distinguish a data-qubit error from a faulty measurement.
- Decode. A classical decoder processes the syndrome history, often using a model of likely hardware faults, and estimates which error pattern most likely occurred.
- Protect the computation. The system may apply a physical correction, or it may use the decoder’s result to reinterpret later logical measurement outcomes.
The checks do not always identify one certain physical fault. Different faults can produce the same observed syndrome, so decoding is an inference problem: the decoder selects a likely explanation and corresponding logical correction. If the errors are too numerous, correlated, or otherwise difficult to distinguish, that inference can fail.
Recommended Free Tools
#1 Best Overall
How can you detect a qubit error without measuring it?
A direct measurement of a data qubit can disturb or reveal information about the quantum state, including a superposition the computation needs to preserve. A QEC check instead measures an observable shared by several qubits. Its outcome says whether a coded relationship has changed; it does not reveal the full encoded state or directly answer whether an individual data qubit is 0 or 1.
The pattern of check outcomes is the syndrome. A single check result may not say when or where an error happened. Comparing results across repeated rounds gives the decoder more context, including clues that a check itself may have been faulty. For example, Google Research’s repetition-code explainer describes one-microsecond rounds in a particular experiment; that is an experimental detail, not a universal QEC cycle time.
There is an important limit to the analogy with classical majority voting. A classical repetition code can store a bit several times and vote on the result. Quantum codes must preserve both superposition and relative phase, so they cannot simply read every encoded bit and vote without damaging the information they are meant to protect.
What errors do quantum codes detect?
Two basic error types help explain why quantum checks must be designed carefully:
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
- Bit-flip error: changes a qubit in a way analogous to switching 0 and 1.
- Phase-flip error: changes the relative phase in a quantum state. It may not look like a classical bit changing, but it can still corrupt a computation.
A simple repetition code makes bit-flip detection easy to illustrate: parity checks can show that one relationship among the encoded bits has changed. In its simplest form, however, it does not correct both bit-flip and phase-flip errors. A surface code uses complementary stabilizer checks to detect both types. Google Research’s 2023 surface-code explanation describes this approach and reports a demonstration that scaled from 17 to 49 physical qubits; those figures describe that specific experiment, not a general qubit requirement for every surface-code logical qubit.
What is a logical qubit?
A logical qubit is quantum information encoded collectively across physical qubits so that error checks can detect and correct many faults before they become logical failures. It is not a single special hardware qubit, and it is not error-free. Its reliability depends on the code, the physical devices, the operations used to extract syndromes, and the decoder.
Code distance describes the minimum number of errors in a particular code that can cause an undetected logical failure. Increasing distance generally improves protection when the hardware operates in the favorable noise regime, but it also requires more physical resources. The exact number of physical qubits per logical qubit depends on the code definition and layout.
Why correction does not guarantee a better result
Error correction adds operations, and those operations can fail too. Initialization, gates, measurements, and the decoding process all contribute to the system’s behavior. Adding physical qubits without controlling these faults can create more opportunities for errors rather than more protection.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →A code has a threshold: for a specified code and implementation, below a relevant noise boundary, increasing protection can reduce logical error. The boundary is not one universal percentage for all quantum computers; it depends on the code, the noise model, and the gates and measurements available. Above the relevant threshold, making a code larger may not deliver the intended suppression.
Rank #4
Fault tolerance concerns the whole computation, not just the encoded memory. Its design must prevent imperfect operations from spreading faults in ways that overwhelm the code. Correlated errors—faults that affect several qubits together or persist over successive correction rounds—can be especially difficult because their syndrome patterns may be harder to decode. More redundancy alone does not solve that problem.
QEC is also distinct from error mitigation. Correction encodes information and uses error checks to protect logical operations or outcomes. Mitigation instead uses techniques to reduce or estimate the effect of errors in results, without the same error-correcting encoding. Neither label means that the underlying physical operations are perfect.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What experimental results show—and what they do not
Google Quantum AI and collaborators reported a surface-code memory using 101 physical qubits in a distance-7 code in a paper published in Nature on February 27, 2025. In that experiment, the distance-7 memory had a logical error rate of 0.143% ± 0.003% per correction cycle, and its measured lifetime was 2.4 ± 0.3 times that of the best constituent physical qubit. The paper reports this as a below-threshold memory result; it is evidence of improved logical memory in that experiment, not proof that a general-purpose, large-scale fault-tolerant computer is already available.
The same paper reports an average decoder latency of 63 microseconds at distance 5 alongside a 1.1-microsecond correction-cycle time. These are different reported quantities: the decoder latency is not the cycle duration. The paper’s broader claim is conditional: if the demonstrated performance can be scaled, it could meet requirements for large-scale fault-tolerant algorithms.
A separate IBM Research paper, published March 27, 2024, estimated that 12 logical qubits could be preserved for nearly one million syndrome cycles using 288 physical qubits, assuming a 0.1% physical error rate. It reported a 0.7% threshold for the paper’s code family under a standard circuit-based noise model. These are results and estimates under stated assumptions, not specifications for a commercially available processor or a universal threshold.
NIST’s general explainer, whose publication date is not shown on the page, gives the broad comparison that leading quantum devices make an error roughly once per thousand operations. That is explanatory context rather than a current benchmark for every machine or operation.
Quick Recap
What to keep in mind
- QEC protects information by encoding it across physical qubits and checking relationships among them.
- Syndromes provide evidence of errors; a classical decoder uses that evidence to infer likely faults rather than receiving a perfect label for each one.
- Bit-flip and phase-flip errors require complementary checks, which is why a simple repetition code is only an illustration of part of the problem.
- A logical qubit can be more reliable than its constituent physical qubits, but its protection is conditional on the code and hardware noise being suitable.
- Experimental logical-memory improvements are meaningful milestones, but scaling them into large fault-tolerant computations remains a separate challenge.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
