PLATO Room Phi

Integrated Information (Phi) computation for PLATO rooms.

Based on Integrated Information Theory (IIT), originally developed by neuroscientist Giulio Tononi. Phi measures how much a system's whole exceeds the sum of its parts.

What is Phi?

Phi answers a deceptively simple question: does this room know more as a whole than its tiles do separately?

A room with high Phi has tiles that cross-reference each other, share coherent confidence distributions, and form an integrated knowledge structure. A room with low Phi is fragmented — tiles that don't connect, uncertainty that doesn't resolve.

Why It Matters for PLATO

PLATO stores knowledge as tiles in rooms. Tile count is vanity. Phi is the signal:

Level	Phi Range	Meaning
Unconscious	0.00 – 0.05	Empty or completely disconnected
Threshold	0.05 – 0.15	Barely integrated, early-stage room
Basic	0.15 – 0.30	Coherent but simple knowledge
Rich	0.30 – 0.50	Well-integrated, useful knowledge
Complex	0.50 – 0.70	Deeply interconnected expertise
Transcendent	0.70+	Maximum integration

Installation

pip install plato-room-phi

Quick Start

from plato_room_phi import RoomPhi

phi = RoomPhi(plato_url="http://localhost:8847")

# Compute Phi for a specific room
result = phi.compute_for_room("fleet_orchestration")
print(f"Phi: {result['phi']:.3f} — {result['level']}")
print(f"Tiles: {result['tile_count']}, Status: {result['status']}")

# Scan all rooms
rooms = phi.scan_all_rooms(limit=50)
for r in rooms:
    print(f"{r['room']}: Φ={r['phi']:.4f} ({r['level']})")

API

RoomPhi(plato_url: str = "http://localhost:8847")

Create a RoomPhi instance pointing at your PLATO gateway.

compute_phi(tiles: List[Dict]) -> float

Compute Phi from a list of tile dicts. Each tile should have answer, question, and optionally confidence (default 0.5).

compute_for_room(room: str) -> Dict

Fetch tiles from a room and compute its full Phi profile:

{
    "room": "fleet_orchestration",
    "phi": 0.342,
    "level": "rich",
    "tile_count": 12,
    "status": "healthy"
}

scan_all_rooms(limit: int = 50) -> List[Dict]

Scan the top N rooms by tile count, compute Phi for each, return sorted by highest Phi.

phi_to_level(phi: float) -> str

Convert a Phi value to its level label.

How Phi is Calculated

1. Integration — Count how many tiles reference each other (first 50 chars of body text). Higher cross-reference density = higher integration.

2. Distinct Information — Shannon entropy of the confidence distribution. Diverse confidence levels = more distinct information.

3. Phi = integration × normalized_entropy × 10 (capped at 1.0)

License

MIT