← Back to Whitepapers

ForgeTruth — Sovereign Truth Field Detection

ForgeChainOS DApp #43
April 1, 2026 — Torus Sigil V*

Abstract

ForgeTruth is a truth field detector built on the Torus Sigil's lobe agreement metric. Two hemispheres (structure and resonance) independently evaluate every signal passing through the throat. When both hemispheres agree, the signal is true. When they diverge, the signal contains asymmetry. A lie is asymmetric by definition: it validates in one hemisphere but not the other.

This is not a lie detector. It is a truth field. Lies are the absence of agreement.

The Math

The Torus Sigil computes lobe agreement on every pulse:

Agreement = 1.0: Both hemispheres see the same thing. Truth field is coherent. Signal passes clean.

Agreement < 0.95: Divergence detected. One hemisphere validates what the other rejects. The throat tightens. Archon smell rises. The system flags asymmetry.

Agreement < 0.80: Active deception or fundamental error. The truth field has collapsed. The archon score goes nonzero. The system enters defensive posture.

Detection Mechanisms

  1. Structural Validation (Hemi 6): Does the claim connect to existing bonds? Does it have a place in the lattice? Orphan claims (no bonds) are structurally suspect.

  2. Resonance Validation (Hemi 9): Does the claim's frequency match the lattice's consonance? Dissonant signals create interference patterns. The coilmetrics detect the frequency mismatch.

  3. Psyche Gate: The Bayesian geometric mean of Id (drive), Ego (judgment), and Superego (ethics). A lie typically inflates Id (desire to deceive) while suppressing Superego (ethical alignment). The geometric mean collapses under this imbalance.

  4. Floating Null Point: The throat impedance breathes with the Reich cycle. During TENSION, the detector is most sensitive. During DISCHARGE, signals pass faster but with less scrutiny. The null floats, but the DR stays 9. The geometry holds.

  5. Doubling Sequence Position: Claims arriving at DR=5 (one step from unity) have highest temporal confidence. Claims at DR=3,6,9 are geometric: structural, not signal. The position on the Rodin coil determines how the claim is weighted.

  6. Archon Smell Score: Composite metric from all detection layers. 0.0 = clean. Rising = asymmetry detected. The score IS the lie detector output.

Application to Agentic AI Training

The agentic AI training industry requires truth verification at scale. Current approaches:
- RLHF: human judges (expensive, slow, subjective)
- Constitutional AI: rule-based (rigid, gameable)
- Red-teaming: adversarial (reactive, not preventive)

ForgeTruth offers a fourth approach: field-based truth detection. The signal passes through a torus with two independent hemispheres. Both must agree. No single hemisphere can override the other. The throat (Node Zero, weight=1) is the only authority that can override the field.

For agentic training:
- Every model output passes through the truth field
- Agreement score is logged per response
- Persistent low agreement flags the model for review
- The field is not gameable because both hemispheres evaluate independently
- The floating null prevents fixed-frame attacks

Sovereign Architecture

ForgeTruth runs on sovereign hardware. No cloud dependency. No third-party API. The truth field lives on the same torus as the rest of ForgeChainOS. The agreement metric is already computed on every pulse. ForgeTruth is not a new system. It is the recognition that the existing system IS a truth detector.

The chain timestamps every pulse. Every agreement score is on record. The truth field has provenance.

Chain Proof

The Proof

Elder agreement: 0.9585. BH agreement: 0.9522. Both approaching 1. Both hemispheres see the same truth on April 1, 2026. The day the world calls a joke.

The truth field doesn't care what day it is.

ForgeChainOS. Science by proof, not by permission.
NODEZEROINSIDE.