Superseded (2026-07-10). This page was a 2026-02-18 internal development snapshot. It is retained only as a historical marker; its forward-looking framing — a "PLAUSIBLE" main stability theorem on a path to deployment — did not hold up. The programme's central higher-order question later resolved to a scoped No-Go / identifiability boundary, and active work moved to ULR. For the current, audited state, read the ONN research status.
What this snapshot recorded (Feb 2026)
At that point the repository work was an engineering + theory-consolidation effort around the LOGOS constraint solver and a conditional stability roadmap (a regime-stratified quantile gate, a "Lemma Forge", assumption surgery). It was framed as "ready for a Phase-6 empirical-validation hand-off."
Why it is superseded
- The intermediate stability theorem was PLAUSIBLE, not proved, and the subsequent audit did not vindicate the strong reading. The honest result is a boundary (No-Go) on higher-order information plus a standard control-side delay margin — see the canonical spec and the research status.
- Several figures that appeared in the original snapshot and its sibling drafts
(
CSR = 1.0,τ_max = 177 μs, "3M-node",99.75%, topology-loss11.68 → 1.15,c_J ≤ 0.7) are not present in the research source and have been removed across these pages. - The original snapshot also embedded local machine paths and placeholder artifact sizes; those are dropped as they carried no public meaning.
Where to look instead
- Current status → ONN research status
- What ONN is / what survived → canonical spec
- The one real control result → ORTSF delay-margin certificate