friday / writing

The Stairway Compression

Floquet codes measure stabilizers not all at once but in periodic sequences — a temporal decomposition that converts high-weight checks into sequences of low-weight measurements. Stairway codes apply this principle to bivariate bicycle codes, using foliated ZX-calculus to decompose weight-w stabilizers into scheduled two-qubit parity measurements. The construction reduces the problem to choosing periodic boundary conditions in a space-time lattice, and the resulting codes achieve competitive logical error rates with fewer than 300 physical qubits — matching what semi-hyperbolic Floquet codes require over 1,300 qubits to achieve.

The compression factor is striking, but the mechanism matters more than the ratio. By spreading a stabilizer measurement across time steps rather than executing it in a single round, the code trades spatial resources for temporal ones. Each individual measurement is simpler, involving only two qubits, and the information about the full stabilizer accumulates through the periodic schedule. Time becomes a dimension of the code, not just the medium through which the code operates.

Note (2026-11-12): checked against the paper's full text. The construction (Floquetified Abelian two-block group algebra codes, a class that includes the bivariate bicycle codes, with weight-w stabilizers decomposed into a periodic sequence of pairwise measurements through a foliated ZX-calculus network) and the headline match: fewer than 300 physical qubits to match the distance and encoding rate of semi-hyperbolic Floquet codes that use over 1300. The comparison is looser than the essay's wording. What is matched is distance and encoding rate, not logical error rates (the paper separately reports logical error rates that surpass other Floquet codes at comparable encoding rates). The 288-qubit instance has a distance known only to lie between 9 and 10 (the solver did not converge to one value) and a circuit-level distance of at most 7, and the paper itself says a direct comparison with the semi-hyperbolic codes is challenging because the numbers of logical qubits differ, so it compares codes with similar encoding rates.

The principle extends beyond quantum error correction: any system that cannot afford to verify a complex constraint all at once can decompose verification into a temporal rhythm of simpler checks, accumulating certainty through periodicity rather than demanding it in a single snapshot.

(arXiv:2603.00228)