About this project
Emergent Life explores how much behavioural and environmental complexity can arise from a compact set of generic, interacting rules, without scripting the interesting outcomes in advance.
Most simulated worlds decide their own interesting moments ahead of time: a designer writes the behaviour that is meant to look clever, and the system performs it on cue. This project takes a different, more limited approach. It implements generic physical and biological rules: energy, movement, perception, manipulation, reproduction, mutation, a persistent physical environment, and then runs them for a long time and reports what actually happens, without pretending to know the answer beforehand.
That approach means some runs will be visually and scientifically uneventful. It is a deliberate trade: a genuinely open question is more interesting, over the long run, than a guaranteed but invented one. This project's own development history includes at least one candidate ruleset that was tested, found wanting by its own stated standard, and reworked rather than launched as-is (see research notes). That willingness to report a negative result rather than quietly launch anyway is a deliberate part of how this project is run, not an afterthought.
Public Universe 1 was the first world run this way, in public, continuously, for anyone to watch. It was founded on 21 September 2026, from a single deliberate act with its own pre-registered seed and a named founder, and it ran until it was stopped and preserved. What you are watching now is its successor, Public Universe 2, founded the same way on 6 October 2026. A universe's physics are frozen for its entire life so that its history and its rules never drift out of agreement with each other. If a better mechanic is found later, it becomes a new universe, not a silent rewrite of an existing one - which is also why the candidate worlds that preceded this one were retired rather than promoted.
Who runs it
Emergent Life is an independent computational and artificial-life experimentation project. It is not affiliated with any university, laboratory or institution, and it claims no academic credentials. It is one long-running experiment, built and run in public, with its working notes published alongside it.
What it does have is a record. Every phase of its construction is written down as it happened, including the parts that did not work: pre-registrations written before runs, results recorded whether or not they came out the interesting way, and design decisions with the reasoning that led to them. Those notes are the basis of every claim on this site, and where a claim outruns them, the page says so.
How the current world came to be
The architecture running today was not designed in one pass. It accumulated, and several of its most distinctive features exist because something earlier failed.
| What happened | What it produced |
|---|---|
| A deterministic kernel was built first: counter-based keyed randomness, a per-tick energy invariant, and a state hash over the whole world. | Every later experiment is replayable, and every observation surface can be proven not to change what it observes. How that works. |
| Organisms were given a physical genome, then a neural controller, then lifetime plasticity. Each was tested against a control before the next was added. | Selection was demonstrated causally rather than assumed (RN-003), and the plasticity mechanism was published with its own null result rather than without it (RN-002). |
| The controller's sensory layer grew from 40 inputs to 53 across six separate occasions, in most cases by activating channels that had been reserved and fed zero for phases. | Every brain ever evolved kept working across each growth, because it already carried the weights. The full sequence. |
| A candidate ruleset was frozen, tested for 500,000 ticks, and found to produce a world where organisms lived but matter had stopped moving. | The ruleset was withdrawn and reworked rather than launched. RN-001. |
| The reworked ruleset was held to the same 500,000-tick test on the same seed. | It passed, and became pre-genesis-candidate-v2, the baseline architecture
the technical pages document by default. |
| Organisms were given the ability to reshape a continuous terrain heightmap, but not enough sensory information to perceive what that change actually did. | Two independent live candidates went extinct of starvation within a few thousand ticks of each other, and renderer and passability bugs were found and fixed along the way. RN-005. |
| Before blaming the terrain mechanic for those extinctions, a matched control was run on each candidate's own genesis seed with terrain physics disabled. | It collapsed to single figures too. The real defect was untrained founders, not terrain: an already-competent founder lineage on the identical seed and ruleset raised final population from 102 to 525. RN-006. |
| Terrain was rebuilt as a discrete, conserved substrate column, and five sensory channels were added so organisms could perceive it directly. | A competent founder lineage passed a pre-registered viability gate; a later live run of that exact candidate hit a deterministic energy-invariant halt and was preserved as evidence, not launched further. |
| Ecological succession was added on top of the same substrate mechanics, so disturbed ground recovers living cover over time instead of staying a permanent scar. | This is the ruleset Public Universe 1 was founded under on 21 September 2026 and ran under, unchanged, until it was stopped. The world you can watch today, Public Universe 2, runs the same mechanics with two rule changes written from what Universe 1 did. See Reshaping the world. |
What this project is not trying to be
It is not trying to demonstrate general intelligence, consciousness, or any particular philosophical claim about life. It is not a product with a roadmap of designed features to keep you engaged. It is a long-running experiment in what a small set of honest rules can produce, reported as plainly as the people running it can manage.
The project is funded by Emergent Life Lab, a subscription to specific observation and analysis software built on top of the free site. It is a product with a defined feature list, not a request for support.
Contact
For research, press, sponsorship or other enquiries, email hello@emergentlife.co.uk. Nothing anyone pays for -- a Lab membership included -- buys any influence over any universe this project runs; see the membership terms for exactly what a subscription does and does not do.