← Back to Home
The Semantic Validity Engine
How Ten patent Families Constitute a Structural Alternative to Editorial Fact-Checking
What we claim: The Semantic Validity Architecture can determine whether a claim is structurally complete—whether it satisfies the six constraints necessary for semantic validity. This operates upstream of empirical verification. A structurally incomplete claim cannot be valid regardless of its factual content.
What we do not claim: The architecture does not replace empirical verification. It does not know whether the Eiffel Tower was built in 1887 or 1923. What it knows is whether the claim asserting a date satisfies the structural conditions that would make it verifiable. It catches the architecture of error, not the content of error.
Why this is stronger: Traditional fact-checkers work downstream—they correct errors after emission. The Hexis works upstream—it prevents structurally incomplete claims from being emitted at all. Prevention, not correction.
Traditional Fact-Checking vs. Semantic Validity Engine
DIMENSION
TRADITIONAL FACT-CHECKERS
SEMANTIC VALIDITY ENGINE
Criterion
None formal. Editorial judgment applied case-by-case.
Six structural constraints constituting the minimum for semantic closure.
Selection
Human editors choose which claims to check. ~0.001% of public claims examined.
Every claim passing through the system is validated against all six constraints. No editorial gatekeeping.
Output
Gradational ratings (Pinocchios, Truth-O-Meter) requiring subjective interpretation.
Categorical signal: COMPLETE or INCOMPLETE. Binary. Operator-independent.
Timing
Reactive. Hours to days after the claim has propagated.
Real-time. Validation occurs before emission, not after publication.
Inference Types
All claims rated on the same scale. No distinction between deduction, prediction, or opinion.
Four inference types (deductive, inductive, abductive, interpolative) discriminated and routed to appropriate closure.
Closure Authority
Editor closes every verdict. No structural routing to human judgment for value claims.
System routes factual claims to system closure, value/ethical claims to human closure. Refuses when constraints unsatisfiable.
Reproducibility
Two editors can reach different ratings from the same facts. Not reproducible.
Same input → same output. Structural determination independent of operator.
Scope
Checks claims against external sources. Downstream of error.
Checks claims against structural completeness. Upstream of error.
Engine Architecture — Claim Processing Pipeline
ANY CLAIM ENTERS THE SYSTEM
Public statement, AI output, document assertion, user query, policy claim, news report
STAGE 1
↓
STAGE 1
Tetrahedral Decomposition
PATENT A
The claim is decomposed into its four components (Claimant, Subject, Grounds, Purpose) mapped onto the tetrahedron’s vertices, and the six relational constraints are identified across the edges. The system asks: Are all four components present? Are all six relations instantiated?
Referential — What is being claimed?
Contextual — Under what conditions?
Premissive — On what grounds?
Inferential — Why does this follow?
Constraining — What are the limits?
Teleological — What is this for?
STAGE 2
↓
STAGE 2
Inference Type Discrimination
PATENT C
The system tags the inference type that produced the claim. Deductive (necessary conclusion) ≠ Inductive (generalization) ≠ Abductive (best explanation) ≠ Interpolative (pattern completion). Each type warrants different confidence and requires different closure. Traditional fact-checkers conflate all four on a single scale.
STAGE 3
↓
STAGE 3
Semantic State Signaling
PATENT D
The system tracks which constraints are satisfied and which are absent, in real time. This is the “semantic spell-check”—a passive diagnostic that shadows the claim as it is being formed, not after publication. Missing constraints are flagged before emission.
STAGE 4
↓
STAGE 4
Constraint Governance
PATENT E
Domain-specific constraint rules are applied. Medical claims require higher Premissive standards (peer-reviewed evidence). Legal claims require verified Premissive grounding (real case citations). Policy claims require explicit Constraining acknowledgment (limits of projections). The constraints adapt their rigor to the domain without changing their structure.
STAGE 5
↓
STAGE 5
Closure Authority Routing
PATENT F
The system determines who should close the claim. Factual lookups: system closes. Calculations: system closes. Value judgments, ethical claims, safety-critical decisions: routed to human closure. The system never finalizes what humans should decide. No fact-checking organization has this structural boundary.
DETERMINATION
↓
ALL SIX CONSTRAINTS SATISFIED?
Categorical determination — not a score, not a rating
6/6 = COMPLETE
↓
<6 = INCOMPLETE
↓
EMIT AS VALID
Semantically complete. All constraints satisfied. Inference type tagged. Closure authority assigned. The claim has the structural form of knowledge.
DO NOT EMIT
Structurally incomplete. Specific missing constraints identified. The system names what is absent.
SUPPORTING INFRASTRUCTURE
↓
Memory, Condensation & Integration
B
H
G
Patent B (Form-Preserving Memory) maintains validated state across sessions—the system remembers what it has verified. Patent H (Semantic Condensation & Reconstitution) extracts the eidos—the formal structure—of any content, enabling compression without meaning loss and proving that form survives substrate removal. Patent G (Integrated System) governs the coordination of all components into a unified architecture.
Patent Family → Engine Function Map
FAMILY A
Tetrahedral Validity Structure
The core decomposition engine. Every claim mapped onto four vertices (components) and six edges (constraints). The geometry that makes validation possible.
ENGINE ROLE → STAGE 1: DECOMPOSITION
FAMILY B
Form-Preserving Memory
Validated claims stored with their structural signatures. The system retains not just what was verified but the structure of the verification. Persistent validity across sessions.
ENGINE ROLE → PERSISTENCE LAYER
FAMILY C
Inference Discrimination
Tags inference type on every claim. Prevents the conflation that lets fact-checkers rate deductions and speculations on the same scale.
ENGINE ROLE → STAGE 2: INFERENCE TAGGING
FAMILY D
Semantic State Signaling
Real-time constraint status. The “semantic spell-check” that operates before emission, not after publication. Eliminates latency.
ENGINE ROLE → STAGE 3: REAL-TIME DIAGNOSTICS
FAMILY E
Constraint Governance
Domain-adaptive rigor. Medical, legal, military, educational claims each governed by the same six constraints at domain-appropriate stringency.
ENGINE ROLE → STAGE 4: DOMAIN CALIBRATION
FAMILY F
Closure Authority
Routes outputs to appropriate closure. System closes factual lookups; humans close value judgments. The structural boundary no fact-checker has.
ENGINE ROLE → STAGE 5: CLOSURE ROUTING
FAMILY G
Integrated System
Coordinates all components into a unified architecture. Governs inter-module communication and system-level validation orchestration.
ENGINE ROLE → SYSTEM INTEGRATION
FAMILY H
Semantic Condensation & Reconstitution
Extracts the eidos (formal structure) from content, enabling lossless compression, cross-domain transfer, and empirical proof that form survives substrate removal.
ENGINE ROLE → EIDOS EXTRACTION & PROOF
FAMILY I
Hexis-Governed Cognitive Pedagogy
Structures learning sequences through six-constraint scaffolding. Ensures pedagogical interactions develop genuine understanding rather than surface pattern recognition.
ENGINE ROLE → PEDAGOGICAL SCAFFOLDING
FAMILY I-EXT
Extended Pedagogical & Diagnostic Applications
Extends the Hexis framework to failure-mode taxonomy, political discourse analysis, and cross-domain diagnostic applications governed by the six constraints.
ENGINE ROLE → DIAGNOSTIC EXTENSION
What We Claim — and What We Do Not
We claim
- Structural completeness checking against six formal constraints
- Categorical output (complete/incomplete) independent of operator
- Real-time validation before emission, not reactive correction
- Inference type discrimination — deduction tagged differently than speculation
- Closure authority routing — the system knows when humans must decide
- Universal application to every claim, not editorial selection
- Formal operators derived from Aristotelian causal analysis, implemented computationally
We do not claim
- Replacement for empirical verification of factual content
- Omniscience about historical dates, statistics, or events
- Infallibility — a structurally complete claim can still be factually wrong
- Elimination of all error — we eliminate structurally preventable error
- Automated resolution of genuine value disputes or ethical questions
- Independence from all human judgment — closure authority preserves it where needed