Patent Cliff Notes
Seven claims, in plain language, each connected to the specific proof result behind it and the real site evidence that demonstrates it — not the filing itself, an honest map of how the pieces connect.
The seven claims, plain language
| Claim | In plain language | Backed by | Real site evidence |
|---|---|---|---|
| 1 — Four-Channel Structural Separation Broadest claim |
Four signal types (environment, internal state, interface, execution/calibration) must be stored as genuinely separate data structures, with no write pathway between them. | Definition 2.1, 2.2 | Four-Channel Framework |
| 2 — Prerequisite-Gated DAG State Engine | Internal state is structured as a graph of nodes with prerequisite dependencies — you can't mark something mastered if what it depends on isn't mastered first. | Definition 2.1 (DSP) | GED Tier Maps, Module Architecture |
| 3 — Circuit Breaker Write-Suspension | When execution-layer anomalies are detected, writes to internal state get suspended at the schema level — not flagged for review, structurally blocked — until the anomaly clears. | Definition 2.4, 2.4a; Theorem 2 | Glossary: Circuit Breaker |
| 4 — Method | The method form of Claim 1 — the same structural separation, described as a process rather than a system. | Definition 2.2, 2.3 | Pipeline Orchestrator |
| 5 — Tamper-Resistant Schema-Level Isolation | The absence of a write pathway from execution-layer observations into internal state is a security property, not just an architecture choice — it's what prevents a bad actor from disguising a manipulated value as legitimate. | Definition 2.2 | Value Laundering — Found & Fixed (the real gap this claim exists to close) |
| 6 — Constitutional Behavioral Unit Architecture | Every behavior a system can take is defined as a structured unit — trigger conditions, execution bounds, non-activation invariants, failure containment, and an audit signal — that can't be bypassed by any role, including an operator. | Formal 5-tuple, see Glossary for related architectural terms | Real implementation exists; not yet published pending review (see note below) |
| 7 — Schema-Level Role Access Control | Access restrictions between roles (student, instructor, parent, operator) are enforced by the structural absence of a pathway, not a permission check — the same non-collapsibility principle applied to roles instead of signal channels. | Co-instantiation of the same mechanism as Claims 1-5 | Estimator Comparison discusses the same architectural-vs-statistical distinction this claim rests on |
Claims 6 and 7's real code implementations exist and have been run and verified — the same standard as everything else on this site — but aren't published here pending review with the attorney handling the filing. Worth knowing this table doesn't hide that gap; it names it.
What's actually original — stated at the level this site keeps everything else
Checked against a real search of the published knowledge-tracing and student-modeling literature (see the Estimator Comparison page for the full result): nothing found does what Claim 1 and Claim 5 require — enforcing non-collapsibility as a hard, schema-level property rather than a statistical estimate that improves with better data. The closest published work (Dynamic LENS, CFGKT) uses learned separation that gets explicitly recombined downstream — the opposite of what these claims require. That's a real, checked distinction, not an assumption.
That's this page's own literature search — real, but not the strongest available evidence. A separate, formal patent-database search has been conducted as part of actual prosecution, identifying a specific closest-prior-art reference and directly addressing it. That analysis is attorney work product and stays private; treat this page as an honest map of the claims and their real backing, not as legal analysis.