Source note: Local copy of a Summer Into AI 2026 submission originally published on Substack. Screenshots, video, and audio found in the post body are mirrored locally when publicly accessible; profile and avatar images are intentionally not copied. Read the canonical post.
I wanted to build independence as a thing that has to be learned — not declared, not scripted, not voted on a rail I laid down in advance. So I gave thirteen colonies brains and let them figure it out.
Each of the thirteen colonies is a reinforcement-learning agent, hand-written from scratch — tabular Q-learning, no libraries, no LLM, no cloud, no API. Every colony starts knowing nothing. Each round it makes one choice: stay loyal and pay the Crown’s tax, or revolt. Then it gets a payoff and updates its own policy. That’s it. Thirteen little minds, each selfishly trying to do better next round.
The trick is the payoff structure — a coordination game, a stag hunt:
Revolt alone, while your neighbors stay loyal, and the Crown crushes you. Bad move.
Revolt together and you trade as free states, and the Crown’s grip weakens the wider the revolt spreads.
Stay loyal and you’re safe — but taxed.
Nobody tells the colonies any of this. They discover it, each one learning from its own rewards whether independence pays. And out of thirteen independent, self-interested learners, something nobody coded appears: a revolution that cascades down the seaboard — colony tipping colony through the real 1773 Committees-of-Correspondence links (Massachusetts↔Virginia, New York↔South Carolina, Pennsylvania↔Georgia) wired right into the graph. Independence here isn’t a symbol and it isn’t a script. It’s an equilibrium that has to be learned.
And you can flip it. Three sliders — Crown Tax, Crown Retaliation, inter-colony Trade — change the incentive landscape, and the learned outcome changes with them. Drop the tax and the colonies learn that loyalty is comfortable and you get a calm, blue, Crown-ruled coast. Crank the tax, or weaken the Crown’s reach, or richen the free-trade union, and the same thirteen blank-slate minds teach themselves to revolt, and the map goes red. There’s a real phase transition in there, and you’re dragging it around with your finger.
Why this isn’t faked. The easiest way to make a “revolution” demo is to script the dominoes. So I built in the proof that I didn’t. There’s a 🔒 No-Learning button: hit it and the colonies’ brains freeze — they can no longer update their policies. Freeze them and the revolution never comes, no matter how oppressive the Crown: they just flip coins forever, and not one of the thirteen ever forms the conviction to be free. Turn learning back on and, under the right incentives, they find independence again on their own. That’s the line between an emergent result and a puppet show — and you get to test it yourself.
On the theme. Week 2 asked for independence as a mechanic. This is the most literal version I could build: independence as a learned equilibrium in a multi-agent system — the thing thirteen separate, self-interested actors have to coordinate their way into, against a power that punishes whoever moves first and alone. That was the actual problem in 1776, and it’s the actual problem in the math. The revolution isn’t drawn on the screen. It’s computed, by minds that started with nothing.
A nod to Jake Strait . Jake’s been the engine of this week — a dozen builds — and two of them are the perfect mirror for this one. In Drafted he stages the real Declaration vote; in The Convention he sets five Claude instances debating the founding. Both put you inside the founders’ decision. EMERGENT REVOLUTION comes at the same moment from the opposite side: nobody here debates and nobody re-enacts history — thirteen agents that have never heard of 1776 rediscover why you’d revolt, from incentives alone. Jake staged the founding; this one grows it from scratch. Respect, Jake — you set a pace this week that made me want to ship one more.
Watch them learn (give it a few seconds — you’ll see the coast turn):
▶️ Live: https://myles-mf.github.io/summer-into-ai-2026/week-2/emergent-revolution/
💾 Source (one file — read the Q-learning and the payoff game): https://github.com/myles-mf/summer-into-ai-2026/blob/main/week-2/emergent-revolution/index.html
🗂️ Everything I’ve built for the contest: https://myles-mf.github.io/summer-into-ai-2026/
Also written up on my site: https://merlinforge.com/blog/emergent-revolution-week-2/
Six engines of independence this week, and this is the one that argues independence isn't handed down — it's figured out. Thanks for setting the theme, Eric . ★
Original source
Canonical Substack URL: https://merlinforge.substack.com/p/emergent-revolution-13-colonies-that.