How Much of the Past Must a Quantum Memory Keep? Delayed Erasures and Syndrome Pruning in Streaming Surface Codes

A streaming surface-code decoder cannot keep every detector round. Old rounds are committed — compressed into a boundary “skin” of live detectors plus a one-bit logical frame verdict — and pruned. That is enough to keep correcting forward. It is not obviously enough to re-judge the past: when hardware reports an erasure late, after the rounds it implicates are already gone, the decoder must reconsider a settled verdict using evidence it no longer holds. We measure what four memory architectures — full record, skin-plus-frame, a sparse firedcheck archive, and a local onset-lip bracket — can still recover, scoring agreement, evidence, and truth as three separate questions. One law organizes the results: the living skin outlives the archive. On a repetition chain, the information pruning sheds is exactly a four-bit bracket at the onset lip; retain it and the pruned decoder becomes deadline-optimal. At distance 3, a local bracket recovers 99.6% of the full archive’s advantage, and re-judgment difficulty is set by position on the code, not bracket size — the two-check corner beats the three-check edges. At distance 5, the archive’s accuracy advantage over the skin collapses by two orders of magnitude (≈ 126×), and what survives is mostly epistemic: the archive knowing when to declare an exact 50/50 tie where the skin can only guess. Two measured boundaries set the law’s scope. Past the two-fault seam, the archive’s decisive advantage re-awakens (+0.234 pp within triple-fault histories) — the collapse is a law about magnitude, not a march to zero. And an independent decoder-cap audit shows that improving the skin’s one-line verdict can close the archive’s remaining advantage on a diagnostic panel entirely: how much the past is worth keeping depends on how well it was summarized. We close with this study’s classical precursor on silent corruption in aperiodic substrates, where naming what does the looking is likewise what turns “silent” into a rigorous statement.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-19
DOI
https://doi.org/10.5281/zenodo.24358202
Primary Topic
Quantum Computing Algorithms and Architecture
Type
preprint
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
preprint

How Much of the Past Must a Quantum Memory Keep? Delayed Erasures and Syndrome Pruning in Streaming Surface Codes

K. T. Niedzwiecki
Zenodo (CERN European Organization for Nuclear Research)
Quantum Computing Algorithms and Architecture
preprint

How Much of the Past Must a Quantum Memory Keep? Delayed Erasures and Syndrome Pruning in Streaming Surface Codes

K. T. Niedzwiecki
preprint en

Abstract

A streaming surface-code decoder cannot keep every detector round. Old rounds are committed — compressed into a boundary “skin” of live detectors plus a one-bit logical frame verdict — and pruned. That is enough to keep correcting forward. It is not obviously enough to re-judge the past: when hardware reports an erasure late, after the rounds it implicates are already gone, the decoder must reconsider a settled verdict using evidence it no longer holds. We measure what four memory architectures — full record, skin-plus-frame, a sparse firedcheck archive, and a local onset-lip bracket — can still recover, scoring agreement, evidence, and truth as three separate questions. One law organizes the results: the living skin outlives the archive. On a repetition chain, the information pruning sheds is exactly a four-bit bracket at the onset lip; retain it and the pruned decoder becomes deadline-optimal. At distance 3, a local bracket recovers 99.6% of the full archive’s advantage, and re-judgment difficulty is set by position on the code, not bracket size — the two-check corner beats the three-check edges. At distance 5, the archive’s accuracy advantage over the skin collapses by two orders of magnitude (≈ 126×), and what survives is mostly epistemic: the archive knowing when to declare an exact 50/50 tie where the skin can only guess. Two measured boundaries set the law’s scope. Past the two-fault seam, the archive’s decisive advantage re-awakens (+0.234 pp within triple-fault histories) — the collapse is a law about magnitude, not a march to zero. And an independent decoder-cap audit shows that improving the skin’s one-line verdict can close the archive’s remaining advantage on a diagnostic panel entirely: how much the past is worth keeping depends on how well it was summarized. We close with this study’s classical precursor on silent corruption in aperiodic substrates, where naming what does the looking is likewise what turns “silent” into a rigorous statement.

Zenodo (CERN European Organization for Nuclear Research)
Quantum Computing Algorithms and Architecture
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon