EDCM Analyzer Platform

A React + TypeScript full-stack application implementing the Energy–Dissonance Circuit Model (EDCM) for analyzing interaction dynamics across text and audio inputs.

🚀 Features

1. Dating Mode (/dating)

• Goal: Help users take clean risks and communicate without escalation.
• Projections: Pace, Power Balance, Intent Clarity.
• Safety: Non-alarmist pattern detection (coercion, boundary erosion).
• Tools: Draft analysis, rewrite suggestions, clean exit generator.
• Audio Support: Voice note and call recording analysis with structural feature extraction.

2. Political Mode (/politics)

• Goal: Analyze rhetoric mechanics without partisan bias.
• Projections: Pressure, Semantic Clarity, Responsibility Locus.
• Tools: Trajectory forecast, definition prompts.
• Audio Support: Speech and debate transcript analysis with prosodic features.

3. Consciousness Lab (/lab)

• Goal: Scientific visualization of "consciousness fields."
• Features: Raw metric access, time-series charts, vector topology.
• Audio Support: Structural paralinguistic feature display in methodology view.

🔒 Privacy & Storage Architecture

Local-First by Default

• No Account Required: All analysis runs client-side using the EDCM engine stub.
• IndexedDB Storage: Sessions saved locally in browser storage automatically.
• Zero Server Dependency: Full functionality without creating an account.

Optional Cloud Sync (Subscriber Feature)

• Hybrid Storage Model: When a user subscribes and logs in:
  • Cloud sync is enabled by default for automatic backup
  • All new and edited sessions sync to PostgreSQL backend
  • Can be toggled off in Settings (opt-out)
• Merge Strategy: Cloud and local sessions merge by lastModified timestamp
• Graceful Degradation: If cloud sync fails, data remains safe locally

Audio Privacy

• Transcripts: Auto-sync when subscribed (default ON)
• Raw Audio Files: Local-only by default; requires explicit opt-in for cloud storage
• User Control: "Delete audio, keep transcript" button available

🛠 Tech Stack

Frontend

• React 18 + TypeScript
• Vite
• Tailwind CSS
• Zustand (State Management)
• IndexedDB via idb (Local Persistence)
• Recharts (Visualization)
• Web Audio API (Recording)

Backend

• Node.js + Express
• PostgreSQL (Neon)
• Drizzle ORM
• Zod (Validation)

🧬 EDCM Engine

The engine (client/src/edcm/engine.ts) is a deterministic stub that computes 9 core metrics:

• C (Constraint Strain)
• R (Refusal)
• D (Deflection)
• N (Noise)
• L (Coherence Loss)
• O (Overconfidence)
• F (Fixation)
• E (Escalation)
• I (Integration Failure)

Audio Integration

When audio features are provided, they modify the engine as non-diagnostic structural signals:

• Fast speech rate (>150 wpm) → increases Urgency and Escalation
• Low pause density (<0.2) → increases Pressure
• High volume variance → signals instability (maps to Noise/Refusal)

Critical: The system does NOT detect emotions. It analyzes pacing, turn-taking, and prosodic acceleration only.

📊 Projection Governance

Raw metrics are NEVER shown to users by default. Instead:

Dating Projections

Pace = categorize(weighted(E, urgency, dE, R, F))
Balance = categorize(weighted(L, D, R))
Clarity = categorize(weighted(O, D, N, C))

Political Projections

Pressure = categorize(weighted(E, urgency, R, dE))
Clarity = categorize(weighted(O, D, N, C))
Responsibility = categorize(weighted(L, R, D))

Thresholds and weights are defined in client/src/edcm/projections.ts.

🗄️ Database Schema

-- Users
CREATE TABLE users (
  id VARCHAR PRIMARY KEY DEFAULT gen_random_uuid(),
  username TEXT UNIQUE NOT NULL,
  password TEXT NOT NULL,
  is_subscribed INTEGER DEFAULT 0,
  created_at TIMESTAMP DEFAULT NOW()
);

-- Sessions (Cloud Sync)
CREATE TABLE sessions (
  id VARCHAR PRIMARY KEY DEFAULT gen_random_uuid(),
  user_id VARCHAR REFERENCES users(id) ON DELETE CASCADE,
  mode TEXT NOT NULL, -- 'dating' | 'politics' | 'lab'
  content TEXT NOT NULL,
  tags TEXT[],
  audio_transcript TEXT,
  audio_features JSONB,
  timestamp TIMESTAMP DEFAULT NOW(),
  last_modified TIMESTAMP DEFAULT NOW()
);

🔌 API Endpoints

Authentication (Stub Implementation)

POST /api/auth/register
POST /api/auth/login
POST /api/auth/subscribe/:userId

Session Sync (Subscribers Only)

POST   /api/sessions              - Save new session
GET    /api/sessions/:userId      - Get all user sessions
GET    /api/sessions/:userId/:id  - Get specific session
PATCH  /api/sessions/:userId/:id  - Update session
DELETE /api/sessions/:userId/:id  - Delete session

⚠️ Disclaimer

This tool observes patterns in expressed signals. It does NOT:

• Diagnose people or mental states
• Infer intent or detect lies
• Adjudicate factual truth
• Identify "dangerous people" or "predators"
• Detect emotions from audio

It DOES:

• Analyze interaction dynamics under constraint
• Surface behavioral patterns (pace, clarity, balance)
• Provide structural analysis of speech (rate, pauses, turn-taking)

🚧 Development

Setup

npm install
npm run db:push  # Initialize database schema
npm run dev      # Start full-stack app on port 5000

Swap Real EDCM Engine

Replace the stub in client/src/edcm/engine.ts with your production engine:

import { analyzeText } from './your-real-engine';
export { analyzeText };

Ensure it adheres to the same interface:

function analyzeText(text: string, options: AnalysisOptions): EDCMResult;

📝 License & Ethics

This is a research tool for understanding interaction mechanics. Users must:

• Obtain consent before recording conversations
• Use analysis results responsibly
• Not rely on this tool for safety-critical decisions

Meaning is co-created by the observer. Signals are not determinations.