Signal Engine V1 — Real-Time Signal vs Narrative Divergence Scoring

EPICON C-253: Signal Engine V1¶

• Layer: PRODUCT > civic-ai-terminal > signal-engine
• Author: ATLAS Agent (+Michael Judan)
• Date: 2026-03-17
• Status: Active

Intent Publication (EPICON-02 Compliance)¶

epicon_id: EPICON_C-253_PRODUCT_signal-engine-v1_v1
title: Signal Engine V1 — Real-Time Signal vs Narrative Divergence Scoring
cycle: C-253
scope: product
mode: normal
issued_at: 2026-03-17T10:06:00Z
expires_at: 2026-09-17T10:06:00Z

justification:
  VALUES INVOKED: integrity, honesty, epistemics
  REASONING: |
    The Iran-Hormuz conflict created a real-time test case where verified
    facts (Hormuz disruption, oil market stress) coexisted with unverified
    narrative claims (yuan-only passage, petrodollar collapse). The terminal
    needed a way to explicitly score and surface this divergence.
  ANCHORS:
    - ZEUS analysis confirmed Hormuz disruption as real while flagging yuan-only claims as unverified
    - No existing tooling distinguished between signal strength and narrative amplification
    - The gap between what is verified and what is amplified is the core civic intelligence problem
  BOUNDARIES:
    - Signal scoring uses heuristic keyword matching in V1
    - Does not replace human judgment — surfaces divergence for operator review
  COUNTERFACTUAL:
    - If keyword matching produces systematic false classifications, replace with model-based scoring
    - If divergence threshold of 10% produces too many alerts, make configurable per category

Context¶

The Iran-Hormuz conflict created a real-time test case for Mobius. The Strait of Hormuz was under severe disruption, with verified shipping constraints and oil market stress coexisting alongside unverified narrative claims (yuan-only passage, immediate petrodollar collapse).

ZEUS analysis on the morning of C-253 established:

• Verified: Hormuz disruption is real, selective negotiated passage occurring, oil markets reacting
• Not verified: formal yuan-only passage regime, universal war tax in CNY

This signal-vs-narrative divergence became the design specification for the Signal Engine.

Changes¶

1. Signal Engine scoring logic (lib/echo/signal-engine.ts)¶

Three scoring dimensions (all 0–1):

Classifications:

• SIGNAL — verified event, narrative proportional to facts
• EMERGING — real event confirmed, narrative elevated but not dominant
• DISTORTION — narrative dominates; amplification exceeds verification depth

Divergence metric: narrative score minus signal score. When positive, narrative is running ahead of verification.

2. Signal Engine Panel (components/terminal/SignalEnginePanel.tsx)¶

• Cycle health summary (signal/emerging/distortion counts + averages)
• Per-event cards with three color-coded score bars
• Classification badges with semantic colors
• Divergence warnings when narrative exceeds signal by >10%

3. Inspector view¶

Full signal score breakdown, divergence analysis, and classification explainer per event.

4. EVE-bot — automated cycle rotation at midnight EST (PR #11)¶

First Real-World Application: Iran-Hormuz Crisis¶

Three EPICON entries recorded against the crisis:

The Insight¶

The Signal Engine doesn't ask "is this true?" — it asks "is the world reacting proportionally to what's actually verified?" That's a fundamentally different question, and it's the question that matters for civic intelligence.

ZEUS Monitoring Conditions (Yuan-Shift Thesis)¶

Conditions that would confirm or deny the yuan-only passage narrative:

• Official policy announcement from Iran
• Multiple nations settling oil in CNY through Hormuz
• Central bank reserve shifts tied to energy trade
• Shipping contracts denominated in non-USD currencies

Impact¶

• First automated signal-vs-narrative scoring in the Mobius ecosystem
• Demonstrated real-time utility against a live geopolitical crisis
• Established the three-dimensional scoring model (signal/narrative/volatility) for future versions
• Created the foundation for the Mobius Detection Engine

Integrity Notes¶

• MII impact: Positive — explicit narrative detection improves information integrity assessment
• GI impact: -0.03 (event-driven) — the Iran-Hormuz crisis itself degrades global integrity; the Signal Engine improves Mobius's ability to track that degradation
• Risk: Medium — signal scoring uses heuristic keyword matching; future versions should incorporate multi-source cross-validation and temporal pattern analysis

"We heal as we walk." — Mobius Substrate