Learning & research

Huxley–Gödel Machine

Inspect hierarchical search and the experiments behind a self-improvement research demo.

Explore this demo ↓Read the full guide ↗Explore the source ↗

OPEN THE COMPLETE COMMAND DECKS

Explore the model in your browser.

Read-only dashboards with energy, governance, stress scenarios and preserved diagrams. No wallet or installation required; all values come from recorded simulations.

Open the HGM research console ↗

A CLOSER LOOK

How should an improvement search be compared with a baseline?

For researchers: inspect hierarchical search and budgeted evaluation through a reproducible simulation and comparison artifacts.

Set the experiment

Inspect budgets, strategies and workload assumptions used by the simulator. The alternative hgm_config.json supports another implementation in this directory.

demo/Huxley-Godel-Machine-v0/config/hgm_demo_config.json ↗

Follow the comparison

Read how each strategy receives work and how aggregate results are computed. Compare like-for-like budgets and seeds rather than names alone.

demo/Huxley-Godel-Machine-v0/simulator/runner.py ↗

Use the import-safe launcher

The compatibility entry point provides explicit config, seed, output directory and UI artifact options for the canonical simulator.

demo/huxley_godel_machine_v0/simulator/__main__.py ↗

REAL REPOSITORY MATERIAL

Inspect. Understand. Reproduce.

Reading the exact source at revision 5b4cebb3.
This browser inspection does not execute the demo.

JSON source1,323 bytesDownload source ↓View on GitHub ↗

demo/Huxley-Godel-Machine-v0/config/hgm_demo_config.json

Select a walkthrough step to explore its source.

Full source text

{
  "seed": 1337,
  "simulation": {
    "total_steps": 120,
    "report_interval": 10,
    "baseline_total_steps": 120
  },
  "economics": {
    "success_value": 125.0,
    "evaluation_cost": 18.5,
    "expansion_cost": 32.0,
    "max_budget": 7500.0,
    "target_roi": 2.5,
    "min_roi": 1.15
  },
  "hgm": {
    "tau": 1.0,
    "alpha": 1.4,
    "epsilon": 0.1,
    "max_agents": 24,
    "max_expansions": 40,
    "max_evaluations": 200,
    "concurrency": {
      "evaluation": 4,
      "expansion": 2
    },
    "quality": {
      "root": 0.58,
      "mutation_std": 0.12,
      "min_quality": 0.05,
      "max_quality": 0.98
    }
  },
  "thermostat": {
    "roi_window": 12,
    "tau_adjustment": 0.12,
    "alpha_adjustment": 0.08,
    "concurrency_step": 1,
    "max_concurrency": 8,
    "min_concurrency": 1,
    "roi_upper_margin": 1.25,
    "roi_lower_margin": 0.15
  },
  "sentinel": {
    "max_failures_per_agent": 12,
    "roi_recovery_steps": 8,
    "hard_budget_ratio": 0.95
  },
  "owner_controls": {
    "pause_all": false,
    "pause_expansions": false,
    "pause_evaluations": false,
    "max_actions": null,
    "note": "Owner can override at runtime via --set owner_controls.pause_all=true"
  },
  "baseline": {
    "mutation_std": 0.09,
    "quality_floor": 0.08,
    "quality_ceiling": 0.96
  }
}

SHA-256 7078d3f008702af067c73f1667403d24fff80be5b2ee7506a489fef1c946a9fc

FROM READING TO A REPRODUCIBLE RUN

Try the selected path.

Python, no demo package install

Run from the repository root with Python 3.12. This selected entry point uses the Python standard library. Keep output in a separate directory so that you can compare runs.

Complete environment setup ↗

SELECTED EXECUTION PATH

python -m demo.huxley_godel_machine_v0.simulator --seed 7 --output-dir /tmp/hgm-demo --ui-artifact /tmp/hgm-comparison.json

What you should observe

Simulation artifacts are written under /tmp/hgm-demo and /tmp/hgm-comparison.json. Inspect per-strategy outcomes and budget use.

The source inspector above reads bundled repository material. Local commands run separately on your computer. Recorded examples may contain historical timestamps, placeholders and simulated metrics.

MAKE IT YOUR OWN

One useful experiment.

Keep the seed fixed and change one budget. Compare performance per unit of evaluation effort, not only the best observed outcome.

THE SYSTEM, MADE VISIBLE

Architecture & relationships

Architecture diagram · source preserved below

View original Mermaid source

flowchart LR
    Operators((Mission Owners)) --> demo_Huxley_Godel_Machine_v0[[Demo → Huxley Godel Machine v0]]
    demo_Huxley_Godel_Machine_v0 --> Core[[AGI Jobs v0 (v2) Core Intelligence]]
    Core --> Observability[[Unified CI / CD & Observability]]
    Core --> Governance[[Owner Control Plane]]

WHEN SOMETHING DOESN’T MATCH

Troubleshooting

Import or dependency error

Confirm the active virtual environment and the selected demo’s requirements. Run python -m pip check; do not install unrelated demo requirements over a working environment.

Unexpected result or missing file

Check the selected entry point, configuration and output argument. Keep the seed and implementation fixed before comparing outcomes.

TRACE THE CHECKS

Verification & next steps

13 tracked test source files are available in this directory. Inspect the tests and their environment before choosing a suite; file counts do not establish test results.

Browse the test sources

For live commissioning, consult the production readiness record.

EVERY VARIANT, PRESERVED

Complete document library

REPRODUCE & INSPECT

Registered commands

Run commands from the repository root after following this demo's guide. Network and owner actions require their documented setup.

npm run demo:hgm:lint

npm run demo:hgm:build

npm run demo:hgm:test

npm run demo:hgm:qa

npm run demo:hgm:worker

Full command catalog and troubleshooting ↗

ORIGINAL DIRECTORYdemo/Huxley-Godel-Machine-v0View on GitHub ↗