This paper reframes logical-qubit benchmarking from a scalar error-suppression problem into a branch-resolved logical-state maintenance problem. Conventional quantum error-correction reporting focuses on logical error rates, better-than-physical comparisons, code-distance scaling, decoder performance, and accepted-stream outcomes. These metrics are essential, but they do not by themselves reveal which stabilization mechanism produced the improvement, whether the result depends on active correction or detection-assisted filtering, how much acceptance overhead was paid, or which branches remain underdetermined because raw timing, reset, syndrome, leakage, or feed-forward records are unavailable. The central contribution is QHOME-RETROFIT, a branch-resolved diagnostic framework that operationalizes the Quantum-State Homeostasis condition of Paper 22 on published quantum error-correction records. The framework converts logical-state maintenance into normalized engineering margins, evidence-status labels, tiered verdicts, limiting-branch reports, and a reproducible CSV-first evaluation workflow. It separates restoration capacity, logical-basis sharpness, protected-mode occupation, source-sink coupling, logical gain, acceptance overhead, and repeated-round stability instead of collapsing them into a single logical-error-rate claim. QHOME-RETROFIT is not a new quantum processor, error-correcting code, decoder, or threshold theorem. It does not claim raw shot-level access where such records are not publicly available, and it does not treat detection-assisted post-selection as equivalent to full real-time source-sink closure. The core operating rule is: score only the branch supported by the released evidence, label what is weakly proxied, and leave underdetermined what the public record cannot decide. The manuscript is a systems-code and reproducibility-layer paper, not a hardware demonstration. It defines a diagnostic grammar, evidence classes, tier gates, reproducibility plans, raw-record replay boundaries, and package-level audit artifacts. Its companion package provides fixed CSV inputs, generated outputs, dashboards, tests, and manifest checks so that the branch verdicts can be inspected rather than accepted as prose claims. Before scaling claims about fault-tolerant quantum computing, this paper argues, researchers should test logical-state maintenance: whether a reported improvement is correction-driven, detection-assisted, acceptance-limited, repeated-round stable, source-sink supported, and sufficiently evidenced to justify the engineering claim being made. Keywords: QHOME-RETROFIT, quantum error correction, logical qubits, fault-tolerant quantum computing, logical-state maintenance, branch-resolved diagnostics, source-sink architecture, post-selection, accepted-stream filtering, reproducibility, systems code, evidence-status labels, tiered verdicts, logical error suppression.
Taekyung Lee (Tue,) studied this question.