Two small universes, running live. Each has energy, bodies, and one law: a body that runs out of energy dies. No teacher, no score, no human data. Whatever else appears, the bodies found on their own. Below is what they look like right now, and whether they are passing the tests we wrote before pressing start.
Connecting…
Pale is energy. Coral is where bodies are, deeper coral where they crowd. Dark is empty. Energy arrives by a rule the bodies never see.
Same universe, same rule. Here a body earns a little extra for predicting what its neighbourhood will look like next.
Can a mind ever learn the rules of the world it is inside, from inside?
That is the only question here. Every experiment on this page is one way of asking it. We cannot ask it about our own universe directly, so we build small universes whose rules we know, put minds inside that do not, and watch whether they find out. A yes is evidence that minds like ours can see past their walls. A no would be the deeper finding, and it would be reported exactly the same way.
Each test carries the date it was written. Lines are added, never edited. If a world fails, this page says fail.
Survive or die. Copy yourself or not. No score, no goal, no reward anywhere in the code. The only question is whether something keeps appearing that we did not put in.
Watch Experiment 1The same universe, plus one reward smaller than the cost of staying alive: predict your neighbourhood. Half the first bodies can read the past. Every body also emits a guess at the rule that feeds the world. We know the rule. They never see it.
Watch Experiment 2In today's worlds the rule never changes, so stability may win and evolution may switch itself off. Here the hidden rule drifts slowly. The question becomes whether a mind can track a moving truth, and whether it recovers faster each time.
Bodies are given one new output: a full statement of the rule as they believe it. Nothing rewards the statement. We score it against the truth they cannot see. Three cases run side by side: no reward at all, a reward for prediction, and a reward for prediction plus the ability to read the past. Whichever case reaches the truth, and whether any does, is the answer in its cleanest form.
Each brain gets one extra sense: the state of its own hidden neurons. Nothing tells it what that sense means. If a self-reading channel spreads through the population and raises survival, a model of oneself has value in this universe. If it is ignored, it does not.
Universe A is given a few numbers that come from Universe B. The bodies in A are never told the numbers come from anywhere. If A's bodies learn to predict B better than chance, a mind can detect something beyond its own world from a thin signal.
The same world on a torus and on a sphere, never told which. If behaviour diverges, the bodies have detected the shape of their space without being able to see it.
Bodies evolved in a forward-running world are placed in the same world with energy flowing in reverse. If they cannot survive, the direction of time was written into them. If they can, that is bigger news.
Eight lines, in plain words here. The exact wording of the ledger, and the code itself, are on the method page.