eR33t science note 30 September 2026

Rules,
Memory,
and Repair

A tiny glider, one deleted cell, and the narrow boundary between a local rule and a remembered target.

20 pagescellular automata

AI-generated cover illustration.

inspect

The narrow question

Can a local rule tell the difference between a healthy path and a damaged one?

A replayable cellular-automaton experiment in local rules, damage, and recovery.

Read the paper

Download PDF
Cover of Rules, Memory, and Repair by Justin Hedge

UNUS MUNDUS / 137 · 110 · 42

Fast loops.
Slow ideas.

Give Alex the question. Let Jev choose a test. Watch the GPU do the work. Then follow the evidence back to the next question.

One shared workflow for the paper, the lab, and Discord. Thinking fast and slow, with a receipt between the two.

  1. 01 ALEX / HANDOFF

    Frame the question

    Frame one hypothesis, one comparison, and the result that would change your mind.

  2. 02 JEV / ON REQUEST

    Choose a path

    Choose among eligible growth, erase, and hidden-state tests. Code checks the selected option.

  3. 03 GPU / LIVE LAB

    Run the world

    WebGL evolves the cells locally. Watch growth, introduce damage, and compare the response.

  4. 04 CODE / MEASURED

    Check the receipt

    Inspect the seed, step count, state hashes, and time-matched intact control.

  5. 05 ALEX / HANDOFF

    Think again

    Interpret the measured outcome, name the uncertainty, and propose the next small experiment.

Connected today The neural lab offers WebGL simulation and Jev test selection. Carry a lab export into Alex using the brief below; an automatic reasoner bridge remains future work.

Take the experiment to AlexAI

Attach your Neural lab export to an AlexAI message, then paste this !code brief.

Explore the experiments

Share the paper