Quantum Error Correction

Every algorithm and every gate discussed so far has assumed qubits behave perfectly. In reality, qubits are extraordinarily fragile, and errors creep in constantly. Quantum Error Correction (QEC) is the field dedicated to detecting and fixing these errors — and it's widely considered the single biggest barrier between today's experimental machines and tomorrow's transformative ones.

Why errors are a bigger problem in quantum computing

Classical computers experience errors too, but they're extraordinarily rare and easy to fix — a classical bit is either 0 or 1, and "majority voting" across redundant copies works well. Quantum computing faces three compounding challenges:

• Decoherence. Any unwanted interaction between a qubit and its environment — stray electromagnetic fields, vibrations, temperature fluctuations — can cause it to lose its quantum state. The longer a computation takes, the more opportunities decoherence has to strike.
• Continuous errors. A classical bit either flips or doesn't. A qubit's state is described by continuous values (the probability amplitudes), so errors can be tiny, partial shifts — not just clean flips — making them harder to detect and categorize.
• The no-cloning theorem. You can't simply "make a backup copy" of a qubit's state the way you'd copy a file, because quantum mechanics forbids creating an identical copy of an unknown quantum state. Classical error correction's simplest trick — redundancy through copying — isn't directly available.

Physical qubits vs. logical qubits

The solution to all three challenges is to spread the information of one reliable "logical" qubit across many imperfect "physical" qubits, using clever encoding schemes that allow errors to be detected and corrected without ever directly "looking at" (and thus collapsing) the underlying quantum information.

This is why you'll often see two very different numbers discussed in quantum computing news: the number of physical qubits on a chip (which can be in the hundreds or thousands today) and the number of logical qubits that those physical qubits could support after error correction (which, with current overheads, would be a much smaller number).

The basic idea: encoding and syndrome measurement

A simplified illustration: in a classical "repetition code," you might represent a single bit using three copies (0 becomes 000, 1 becomes 111). If one copy flips due to noise (e.g., 010), majority voting reveals the original value was 0 and corrects the error.

Quantum error correction codes use a more sophisticated version of this idea. Multiple physical qubits are entangled together to encode one logical qubit. Periodically, special measurements called syndrome measurements are performed — these reveal whether and where an error occurred, without revealing the actual quantum information being protected. Based on the syndrome, a correction can be applied to fix the error.

Surface codes and the threshold theorem

The most widely studied approach for near-term hardware is the surface code, which arranges physical qubits in a 2D grid with neighboring qubits performing local syndrome measurements. Surface codes are popular because they only require qubits to interact with their immediate neighbors — a good match for how superconducting and other chip-based qubits are physically laid out.

A foundational result called the threshold theorem states that if the error rate of individual physical qubits and gates is below a certain threshold, then increasing the number of physical qubits per logical qubit can drive the logical error rate arbitrarily low — in principle allowing arbitrarily long computations. Getting hardware error rates reliably below this threshold, and then scaling up the number of physical qubits per logical qubit, is the central challenge of the field.

NISQ vs. fault-tolerant quantum computing

Today's quantum computers are often described as being in the NISQ era — "Noisy Intermediate-Scale Quantum." These machines have enough physical qubits to be interesting, but not enough error correction overhead to create many (or any) fully protected logical qubits. Algorithms designed for the NISQ era (like variational algorithms) try to extract useful results despite this noise.

The long-term goal is fault-tolerant quantum computing — machines with enough error-corrected logical qubits, running for long enough, to execute algorithms like Shor's Algorithm at a scale that matters. Closing this gap is widely viewed as requiring continued improvements in both the quality (error rates) and quantity (count) of physical qubits.

What's next?

Error correction is fundamentally about hardware — the physical qubits, their error rates, and how they're wired together. Let's take a tour of how quantum hardware is actually built.

Continue to: Quantum Hardware →

Frequently Asked Questions

How many physical qubits are needed for one logical qubit?

It depends on the error correction code and the physical error rates of the hardware, but commonly discussed estimates for surface codes range from dozens to over a thousand physical qubits per logical qubit. Reducing this overhead is a major research goal.

Can software fix quantum errors without special hardware?

Error correction requires extra physical qubits and specific control sequences, so it's a combination of hardware design and software/control systems working together — not something that can be added purely in software after the fact.

Why don't we just use better-quality qubits instead of error correction?

Improving individual qubit quality is also an active and important area of research, and the two approaches go hand in hand: better physical qubits reduce the error-correction overhead needed, but some form of error correction is expected to remain necessary for large, long-running computations regardless of how good individual qubits become.