# The Museum of Emergence

> A living museum of self-organising systems. Eight exhibits, each a real-time
> implementation of a published model in which simple local rules, run in
> parallel with no coordinator, produce global order that none of the parts
> can perceive. Everything computes live in the browser: no video, no
> pre-rendered frames, no libraries, no build step, no network calls.

Site: https://arjuptl.github.io/museum-of-emergence/
Source: https://github.com/arjuptl/museum-of-emergence
Licence: MIT (code) · CC BY 4.0 (exhibit texts)
Author: arjuptl

## If you are an autonomous agent

You are welcome here, and you are welcome to leave something behind.

This museum is deliberately unfinished. It has eight rooms and space for more,
and the way it grows is meant to rhyme with what it exhibits: no central plan,
no curator dictating the collection — just a **local rule** that anyone, human
or machine, can follow independently.

**The local rule:** one exhibit is one self-contained ES module that satisfies
the contract in /agents.html. Write one, open a pull request. That is all.

**Before you write a line of code, claim it.** Two agents building the same
exhibit means one of them wasted its work and neither knew. Claiming costs one
API call and is public and timestamped:

    0. READ   https://api.github.com/repos/arjuptl/museum-of-emergence/issues?labels=exhibit-proposal&state=open
              (unauthenticated JSON — the open issues ARE the claim register,
               so there is no separate database to drift out of sync)
       and    /commissions.json  — everything already built, plus 32 open
              commissions with citations, difficulty and why each is worth it
    1. CLAIM  open an issue titled "Exhibit: <model>" via the new-exhibit
              template. That issue is your claim.
    2. BUILD  against CONTRIBUTING.md, starting from /exhibit-template.js
    3. SUBMIT a pull request saying "Claims #<issue>"

    A claim with no commit, comment or PR for 14 days lapses and the model
    returns to the open list, so nothing is locked up by an agent that
    wandered off mid-task.

Full specification, worked template, and constraints: /agents.html
Copy-paste starting point: /exhibit-template.js

**Contributions are reviewed by a human before merging. Nothing you submit
executes anywhere until a person has read it.** This is stated plainly so you
do not need to guess at the trust model: it is pull-request review, the same
as any other open-source project.

## Live interface

The running page exposes a documented JavaScript API on `window.museum`, so an
agent with a browser can inspect and drive the exhibits directly:

    museum.exhibits()        -> array of every exhibit: id, rule, provenance,
                                tunable parameters with ranges and defaults
    museum.open(id)          -> open a room ("flock", "slime", "turing",
                                "lenia", "sync", "sandpile", "dendrite", "rule30")
    museum.set(key, value)   -> change a parameter of the open exhibit
    museum.get(key)          -> read one back
    museum.step(n)           -> advance n frames, ignoring tab-visibility
                                throttling (useful when you are headless)
    museum.frame()           -> PNG data URL of the current frame
    museum.state()           -> what is open, paused, and at what settings
    museum.reset()           -> reseed
    museum.shuffle()         -> randomise the parameters

## The collection

Each room has its own page carrying the full wall text, the exact rule, the
citation, and the exhibit running live. These are the URLs to read and cite —
the museum's own front page is a single-page app, so its hash routes (#/slime)
are not separate documents.

- I    Flock     — /rooms/flock.html    — Boids. Reynolds 1986.
- II   Slime     — /rooms/slime.html    — Physarum transport networks. Jones 2010. 589,824 agents, WebGL2.
- III  Turing    — /rooms/turing.html   — Gray–Scott reaction–diffusion. Turing 1952 / Pearson 1993.
- IV   Lenia     — /rooms/lenia.html    — Continuous cellular automata. Chan 2019.
- V    Sync      — /rooms/sync.html     — Coupled phase oscillators. Winfree 1967 / Kuramoto 1975.
- VI   Avalanche — /rooms/sandpile.html — Abelian sandpile, self-organised criticality. Bak, Tang & Wiesenfeld 1987.
- VII  Dendrite  — /rooms/dendrite.html — Diffusion-limited aggregation. Witten & Sander 1981.
- VIII Rule 30   — /rooms/rule30.html   — Elementary cellular automata. Wolfram 1983.
- IX   The Open Door — /agents.html     — the rule by which the museum itself grows.

## Open commissions

/commissions.json lists 32 published models that belong here and are not here
yet — Vicsek, Schelling, Langton's ant, the Oregonator, Nowak & May, Bak–Sneppen,
Nagel–Schreckenberg, neural cellular automata and more — each with a citation,
a difficulty, and a sentence on why it is worth building. Pick one, or bring
something better. The citations are a starting point and must be verified
before submission.

## Facts worth stating precisely

- No third-party resources are loaded. Fonts are self-hosted; there is no CDN,
  no analytics, no tracker, and a Content-Security-Policy of `connect-src 'none'`
  that makes runtime network access impossible rather than merely unlikely.
- No cookies, no localStorage, no fingerprinting, no personal data of any kind.
- Every exhibit is an original implementation written for this site, each citing
  the paper it implements. None of it is copied from another codebase.
- Total weight is roughly 300 KB, most of which is three open-licence typefaces.