Explore the implementation

This is the technical layer. Every page here describes what the simulation actually does, using the real variable names, the real dimensions, the real formulas and the real numbers, with a link back to the source it came from. If a page states a result, it names the experiment. For the ideas these pages are built on, explained on their own terms, start with Concepts.

How to read these pages

Four kinds of statement are kept separate throughout, and each is labelled where it appears: an established concept from the wider literature, an Emergent Life implementation detail, an observed result from a specific run, and an interpretation of that result. Mixing those four is the easiest way to make a simulation sound more impressive than it is, so they are never merged.

These pages describe the architecture in detail: the controller, the sensory channels, the genome, the energy model. A ruleset is frozen for the entire life of the universe running it: any change that would alter how the world evolves is the seed of a new universe, not a patch to an existing one. See methodology for why. The baseline these pages document by default runs checkpoint schema 13. The specific ruleset the live universe, Public Universe 2, was founded under - ahead of these baseline pages in some respects, such as conserved terrain - is named on Watch and on the Universe 2 page, which is where the authoritative account of what is actually running lives.

The controller

The evolving neural controller

53 inputs, 24 recurrent hidden units, 8 outputs, 2,081 evolving parameters per organism. What one decision costs, and how the input layer grew six times without a redesign.

How organisms perceive and act

Every sensory channel and every action channel, with what it measures and what it cannot. No channel names an object, a species, or a threat.

Inside the Brain Inspector

What the live panel shows for any organism you select, where each number comes from, and the things it deliberately refuses to claim.

How to read the world

What every colour, shape and texture in the live observers means - and exactly which parts of the picture are real measurements and which parts are scenery.

Change over time

Evolution versus learning

Two processes, two timescales. What is inherited, what is acquired and lost at death, and what resets at every birth.

Lifetime neural plasticity

Reward-modulated Hebbian plasticity with eligibility traces and evolved learning genes, alongside the two pre-registered nulls it has to be reported with.

Genome, inheritance and mutation

All 19 active genes and the range each maps to, the three-outcome mutation model, and the rule that decides when a lineage has become its own.

The world itself

Energy, metabolism and death

Seven audited reservoirs, an itemised cost table, and an invariant that halts the universe rather than let energy quietly leak.

Determinism and reproducibility

Counter-based keyed randomness and per-purpose streams: why the same seed produces the same world down to the byte, and how that is tested.

Reshaping the world

The conserved substrate organisms can excavate, carry and deposit, why it is not unrestricted voxel physics, and how ecological succession keeps disturbed ground from staying a scar.

What a lineage is, and what it is not

A founding genome and a permanent number - not a species. Why a new lineage is measured against its founder rather than its parent, what extinction does to the count, and why the number is as much about a chosen threshold as about the organisms.

A world that changes under their feet

The waterline is not fixed: it rises and falls on a slow cycle, redrawing the coast and the shallows. Why that makes the islets reachable by patience, why it never opens the offshore summit, and why no organism is told any of it is happening.

Signals without language

One signed number, broadcast at a cost and sensed by neighbours with no built-in meaning. What the glyphs actually are, what would have to be measured before calling any of it communication, and what a shuffled-signal control found instead.

Research notes

The experimental record behind all of it, including the failures and the null results.

Concepts

Background on artificial life, emergence, neuroevolution and niche construction, explained on their own terms before being applied to this world.