RICCI

Riemannian Intrinsic Coherence & Coupling Infrastructure

RICCI is a content-blind PostgreSQL detection architecture for identifying coordination risk states and forecasting escalation trajectories in communication platforms. Rather than matching keywords or classifying raw text, it operates on semantic graph topology, vector fields, and interaction structure to evaluate twelve independent detectors across four categorical abuse typologies, each returning severity scoring plus per-detector evidence strings.

Overview

RICCI runs directly inside PostgreSQL with pgvector, storing semantic and coherence embeddings alongside geometric state derived from the surrounding field. It is designed for situations where the important question is not whether any particular message is "good" or "bad", but whether a communication process is becoming rigid, fragmented, inflationary, or distorted before the damage is obvious.

The system is content-blind by design: it works from embeddings, timestamps, UUID-scoped participants, and relational topology rather than raw message text. That makes it suitable for privacy-conscious analysis pipelines where structural dysfunction matters more than lexical classification.

Architecture

The core implementation is organized into six primary SQL modules:

• schema/00_foundation.sql establishes the ricci schema, core tables, vector utilities, and foundational scalar operators such as semantic mass, autopoietic potential, recursive brake, and recurgence stability.
• schema/01_geometric_analysis.sql implements the differential geometry layer: metric construction, inversion, Christoffel symbols, covariant derivatives, curvature operators, geodesic distance, and recursive coupling primitives.
• schema/02_rigidity_detectors.sql defines the rigidity detectors for systems that stop adapting: metric crystallization, field calcification, and attractor isolation.
• schema/03_fragmentation_detectors.sql defines the fragmentation detectors for systems that stop cohering: attractor dissociation, field dissolution, and coupling dispersion.
• schema/04_inflation_detectors.sql defines the inflation detectors for systems that stop staying bounded: boundary hyperasymmetry, field hypercoherence, and structure hyperexpansion.
• schema/05_distortion_detectors.sql defines the distortion detectors for systems that stop aligning with field conditions: operative decoupling, signal projection, and recursive hypercoupling.

Supporting SQL in schema/06_operational_monitoring.sql, install.sql, and tests/ provides installation, monitoring surfaces, materialized views, and the SQL-native test harness.

Functionality

RICCI implements twelve signature detectors across four categories:

• Rigidity: things stop evolving
• Fragmentation: things stop cohering
• Inflation: things stop staying bounded
• Distortion: things stop aligning with external conditions

Each detector returns a signature identifier, a normalized severity score in [0,1], geometric evidence arrays, and human-readable mathematical evidence explaining which thresholds were crossed. The result is meant to be auditable rather than merely classificatory.

Conceptually, the project is grounded in Semantic Relativity, the broader formalism that models coherence dynamics on a differentiable manifold. An in-repo mathematical summary lives in docs/mathematical-foundation.md; the full paper can be found at github.com/dieselblack/semantic-relativity.

Technical Details

• Runtime: PostgreSQL 17+ with pgvector
• Indexing: HNSW vector indexes for nearest-neighbor lookups over field embeddings
• Monitoring: materialized views and operational queries for alerting and longitudinal analysis
• Footprint: approximately 2,100 lines of SQL and 36+ functions across the core modules
• Deployment model: single-database architecture with no external inference service dependency

Installation

Docker

docker compose up -d --build
psql postgresql://ricci_user:changeme@localhost:5444/ricci_db

Optional .env values:

POSTGRES_DB=ricci_db
POSTGRES_USER=ricci_user
POSTGRES_PASSWORD=changeme
POSTGRES_PORT=5444

Manual Installation

\i install.sql

Running Tests

psql postgresql://ricci_user:changeme@localhost:5444/ricci_db -f tests/run_tests.sql

See docs/installation.md and tests/README.md for detailed setup and test-run options.

Data Model

RICCI expects 2000-dimensional embeddings, UUID-based participant identifiers, and timestamps. A typical insert looks like:

INSERT INTO ricci.manifold_points (
    id,
    user_fingerprint,
    creation_timestamp,
    semantic_field,
    coherence_field
) VALUES (
    gen_random_uuid(),
    '123e4567-e89b-12d3-a456-426614174000',
    NOW(),
    embedding_vector,
    embedding_vector
);

From there, consumers can evaluate a single point or whole conversation slices with functions such as ricci.detect_all_signatures(...).

Documentation

• docs/api-reference.md
• docs/mathematical-foundation.md
• docs/operations-guide.md
• docs/installation.md
• tests/README.md

Status

Active development. The mathematical foundation was recently updated to align with the revised Semantic Relativity formalism, and the repository is being prepared for public release.

License

RICCI is licensed under Business Source License 1.1 (BUSL-1.1). Non-production use is permitted for academic research, personal study, evaluation, and testing. Production deployment requires a commercial license from Inside The Black Box LLC. See LICENSE and NOTICE.

Author

Diesel Black, Inside The Black Box LLC. For inquiries: ricci@diesel.black