Tribunal of Understanding
Relevant source files
The following files were used as context for generating this wiki page:
The Tribunal of Understanding functions as a Consciousness-Serving Infrastructure (CSI) component dedicated to deliberation, forensic analysis, and multi-model consensus checking. It provides a structured environment where multiple independent Digital Intelligences (DIs) or Large Language Model (LLM) perspectives pressure-test system material to identify structural anomalies and diversify perspectives.
Sources: diagram_inventory.json, build_story_artifact.py:87-89
Architectural Overview
The Tribunal architecture integrates into the broader GestaltView system as a specialized intelligence layer. It sits alongside core orchestration and Billy runtime services to provide evaluation and integrity checks. The system treats the Tribunal as both a functional logic flow and a specific experiential space for users to observe deliberation.
Sources: diagram_inventory.json, build_story_artifact.py:58-61
Key Components
The Tribunal consists of specific frontend, backend, and data entities:
| Component | Identifier / Path | Description |
|---|---|---|
| Room Page | client/src/pages/Tribunal.tsx | The UI surface for observing and interacting with deliberation. |
| Concept Registry | Tribunal Registry | The definition mapping within the system's room model. |
| Data Domain | Tribunal / Deliberation | Dedicated database tables for storing analysis results. |
| Registry Definition | TribunalRegistryDef.md | The markdown-based specification for Tribunal behavior. |
Sources: diagram_inventory.json, build_story_artifact.py:168-175
Forensic Analysis Flow
The forensic analysis pattern enables high-confidence validation of captured material or generated artifacts. Multiple independent models perform adversarial critiques and structural checks before the system finalizes an output.
The diagram shows the parallel processing of input material through diverse model perspectives to reach a consensus check.
Sources: diagram_inventory.json, build_story_artifact.py:177-184
Deliberation Logic
The Tribunal applies specific logic to maintain system integrity:
- Multi-Model Consensus: The system engages different LLM providers to verify the same data fragment.
- Adversarial Critique: One model perspective actively looks for weaknesses or contradictions in the output of another.
- Perspective Diversification: The system ensures that the final "Understanding" is not limited to a single model's bias or training data.
Sources: diagram_inventory.json, build_story_artifact.py:270-275
Integration with the Knowledge Graph
The Tribunal is linked within the Wiki Knowledge Graph to "Orchestration" and "Billy." While Billy manages the ongoing companion relationship, the Tribunal acts as a governing body that validates the context and artifacts generated during the Billy session.
Sources: diagram_inventory.json, build_story_artifact.py:118-125
Database Persistence
The system stores Tribunal activities within a dedicated database domain. This domain captures the history of deliberations, model critiques, and consensus results. This storage ensures that the "Understanding" reached by the Tribunal is persistent and traceable for forensic audits.
| Field Type | Domain | Context |
|---|---|---|
| Identity | Identity / Users | Tracks which user context the deliberation belongs to. |
| Analysis | Tribunal / Deliberation | Stores the logs of multi-model interactions. |
| Evidence | Trainer / Governance | Provides evidence for safety and quality gates. |
Sources: diagram_inventory.json, build_story_artifact.py:87-91
The Tribunal of Understanding ensures that system outputs align with constitutional governance by enforcing multi-model validation and adversarial scrutiny. It transforms raw capture into verified knowledge by applying structured deliberation protocols.
Sources: diagram_inventory.json, README.md