Quantum error correction: a reading path

Three source-based readings, selected to compare experimental questions rather than rank hardware.

Start with the experiment you need to understand

This reading path is for readers who encounter claims about logical qubits and want to know which experimental task supports them. We select a memory demonstration, an encoded-circuit and sampling study, and an encoded-versus-unencoded circuit comparison. The selection is editorial: these are useful examples of different questions, not a complete survey, a list of the newest papers, or a quality ranking.

Before following a headline number, choose your own question. Are you studying how a stored state survives repeated correction, how a particular encoded circuit performs, or how a logical task compares with its physical baseline? Write the question down. Then use the corresponding reading below as a guide to the original paper’s methods.

A comparison worksheet

Each row identifies a reading question and the conditions we recommend carrying into a comparison. Open the analysis to find the original source, manuscript version, reported findings, and our interpretation.

Swipe the worksheet sideways to see all three columns. With a keyboard, focus the worksheet and use the left and right arrow keys.

Selected readings and comparison questions
Paper authors	Reading question	Record alongside the claim
Rajeev Acharya et al.	Repeated logical storage	Task, code distance, decoder conditions, and error denominator
Dolev Bluvstein et al.	Encoded circuits and sampling	Encoding, output metric, acceptance rule, and attempted trials
M. P. da Silva et al.	Encoded versus unencoded circuit outputs	Physical baseline, failure definition, and selection conditions

Use the worksheet instead of a single score

For each candidate paper, record the measured output, error denominator, code or encoding, decoder, trial acceptance rule, and the baseline used. Preserve unknown fields. If two papers evaluate different tasks, explain the difference before calculating a ratio or comparing the numbers. Our recommended approach is to treat such studies as complementary evidence until a shared comparison is justified.

For example, a research meeting about memory scaling should begin with the storage protocol and its error definition. A meeting about retained sampling outputs should begin with the output metric and trial accounting. These suggested review workflows are our interpretation; they are not claims of independent reproduction or additional experimental capabilities.

Explore beyond these selected readings

The collected research library covers selected public sources and a recent collection window. These earlier papers provide context and may fall outside that window. Use publication years to locate periods in the library and citation order to trace indexed discussion. Missing coverage or a missing citation measurement should remain an uncertainty in your notes.

Browse collected error-correction papers · Understand year and citation order · Compare error-correction results

Read beyond the abstract

Our research notes separate reported results from interpretation, and explain what to check before comparing claims.

Below threshold: what a surface-code memory proves
Increasing code distance can improve a logical memory while leaving important computational capabilities untested. Use the measured task, denominator, and decoder conditions to assess the claim.
48 logical qubits: read the acceptance conditions too
A logical-qubit count becomes informative when paired with encoding, circuit, evaluation metric, and retained trials. This note offers a worksheet for comparing demonstrations with different selection conditions.
Better than physical: which circuit is the baseline?
An improvement factor needs its physical baseline and selection rule. Use this paper to practice comparing an encoded circuit with the actual unencoded task, rather than a headline gate fidelity.