← BACK TO INDEX

/HOW-IT-WORKS

L1–L4 Architecture & Governance

// HIGH LEVEL OVERVIEW

ARCHITECTURE

TM4 separates autonomy into governed layers:

No layer can bypass another. Autonomy is permitted — only under governance.

// THE FOUR LAYERS

L1-L4
LAYER_01EXECUTION

L1 – EXECUTION

Sandboxed, time-limited, deterministic

• All code runs in isolated environments

• Strict time and resource limits enforced

• Deterministic execution for reproducibility

• No external network access or side effects

LAYER_02IDEATION

L2 – IDEATION

Proposal generation without evaluation access

• LLM generates candidate solutions

• No access to test results or scores

• Prevents reward hacking and overfitting

• Blind generation ensures honest exploration

LAYER_03EVALUATION

L3 – EVALUATION

Tournaments, multi-axis scoring, evidence capture

• Candidates compete in tournaments

• Multi-dimensional fitness scoring

• All results logged and auditable

• Adversarial test cases prevent gaming

LAYER_04GOVERNANCE

L4 – GOVERNANCE

Invariants, anti-cheat rules, progression logic

• Enforces system-wide invariants

• Anti-cheat detection and prevention

• Controls when evolution is allowed

• Maintains audit trail of all changes

// KEY PRINCIPLES

GOVERNANCE
PRINCIPLE_01CORE

SEPARATION OF CONCERNS

Generation (L2) is blind to evaluation (L3). This prevents the system from learning to game the tests.

PRINCIPLE_02CORE

ADVERSARIAL VALIDATION

Test cases are designed to catch shortcuts, heuristics, and superficial improvements.

PRINCIPLE_03CORE

COMPLETE AUDITABILITY

Every proposal, evaluation, and decision is logged. Evolution can be replayed and verified.

"Autonomy is permitted — only under governance."

No layer can bypass another. This is the foundation of honest evolution.

NEXT: EXPERIMENTS →← BACK TO ABOUT