Research notes
The development story behind Public Universe 1's ruleset: what was tried, what failed by its own stated standard, and what the evidence actually shows.
Each note is drawn from this project's own append-only research record: a question written down first, a result, and the limitations that were recorded at the time rather than added later. Null results are published as null results and failed designs are published as failed designs, because an archive that only contains successes is a marketing page with citations.
RN-001 · Phase M
The world that went physically inert, and the rebuild that followed
Superseded ruleset Pre-registered
A frozen candidate ruleset was tested with a 500,000-tick run before launch. Organisms lived, but the physical world stopped working: 85 million push attempts produced 9,968 successes. The ruleset was reopened, reworked and re-tested.
Abstract free to read · full note in Lab
RN-002 · Phase 4B and Phase 4C
Does lifetime learning do anything? Two pre-registered nulls and one real effect
Pre-registered Null result Superseded ruleset
Organisms carry a working reward-modulated plasticity mechanism. Two pre-registered population-level tests found no detectable behavioural effect from it. A separate counterfactual proved it does measurably move an individual's outputs. Both are true.
Abstract free to read · full note in Lab
RN-003 · Phase 2
Natural selection with no fitness function: mutation on versus off
Pre-registered Superseded ruleset
There is no fitness function anywhere in this simulation. To check that selection was nonetheless doing real work, the same world was run twice per seed with only mutation toggled. Eight seeds out of eight moved the predicted gene in the predicted direction.
Abstract free to read · full note in Lab
RN-004 · Phase H onward, updated for Phase Z2
Two islands, zero crossings, so far
Open question Exploratory
Two founder populations were placed on separate landmasses with water between them. In every run to date, including a 500,000-tick study, they have never crossed and never met. The staging world that preceded Universe 1 visibly reshaped its own ground -- land area up 8.75% from its genesis -- while the strait between its islands stayed 43 cells of open water at the narrowest, with no crossing. Public Universe 1, founded 21 September 2026, inherited the question and has not answered it either: the two populations have still never met.
Abstract free to read · full note in Lab
RN-005 · Phase Y through Phase Z2
Can evolving organisms reshape inherited geography?
Open question Exploratory
Organisms can now excavate, carry and deposit conserved units of ground substrate, with no construction objective anywhere in the reward or perception model. Two earlier architectures failed for different reasons before a third produced a viable, currently live candidate that adds ecological succession on top. This is the honest, unfiltered record of that path, including the failures.
Abstract free to read · full note in Lab
RN-006 · Phase Z
What actually killed two terrain experiments? A matched-founder diagnosis
Exploratory
Both live mutable-terrain candidates went extinct of starvation within a few thousand ticks of each other, and the terrain mechanic was the obvious suspect. A matched control run on the same genesis seed as each candidate, with terrain physics turned off and everything else held identical, collapsed to single-figure populations anyway. Changing only the founder lineage from untrained to an already-competent one, same seed, same ruleset, took final population from 102 to 525. The terrain was not exonerated as good; it was cleared of being the cause.
Abstract free to read · full note in Lab
RN-007 · Phase AC
Two tall columns at opposite ends of the world: what actually raised them
Exploratory
Both islands in the live world grew a tall, narrow column at their outermost corner, one 85 units high where the rules allow 12. Two separate populations producing the same shape looked like convergent behaviour. It is three ordinary things together: material drifts against the edge of the world because of an asymmetry in the rules there, a conservation fallback lets a dying organism stack a unit past the height limit, and the renderer draws built ground four times taller than natural hills. Nothing here is construction, and the islands never influenced each other.
Abstract free to read · full note in Lab
RN-008 · Phase AD
Does signalling pay? A shuffled-signal control, and no sign that it does
Pre-registered Null result
Every organism can emit one continuous signal that its neighbours sense. An earlier look found receivers ignore it. This study changed the rules so a single neighbour can be heard, then ran three versions of the same worlds for 60,000 ticks: signals heard normally, signals shuffled between organisms every tick, and signals heard by nobody. Receivers still did not respond to their neighbours, and the worlds where signals were heard normally did not do better. The one run that passed the site's own test for meaningful signalling was a shuffled control, which shows that test is not strict enough.
Abstract free to read · full note in Lab
RN-009 · Pre-genesis
A road that could never have stood: what we learned by asking the wrong question
Pre-registered Null result
We built a mountain in the sea that can only be reached by building a road out to it, and then watched for two simulated days to see whether anyone would build one. Nobody did. They put no more material on the route than they put on the same spot in a world with no mountain above it. Then we checked something we should have checked first: whether a finished road would stay up. It would not. The world's own physics pulls a narrow causeway apart faster than it can be raised. The organisms had been failing a task that had no solution.
Abstract free to read · full note in Lab
RN-010 · Pre-genesis
Learning that works but does not pay: four rounds, and a decision made in the open
Pre-registered Null result
Organisms here carry a learning mechanism: experience can adjust their own wiring within a single life, and an internal critic learns to predict whether things are about to go well. Over four rounds we got that mechanism working -- the critics genuinely predict what is coming. What we never found is any sign that it helps them live or reproduce. It costs energy and speed, and the difference against a control that cannot learn is slightly negative, not positive. The universe is being founded with learning switched on anyway, as a judgement, and this note exists so that judgement is on the record rather than dressed up as a result.
Abstract free to read · full note in Lab
RN-011 · Pre-genesis
The water is the wall: three worlds that explored everything they could reach
Exploratory
Across three full-length runs of the world we are preparing to found, no organism ever crossed water. Not once, in more than half a million organism-lifetimes. It would be easy to read that as incuriosity. It is not: within their own island they explore relentlessly, and were still finding ground they had never stood on at the very end of the run. What they never do is leave. The two islands have no land route between them, the offshore summit cannot be walked to from anywhere, and reaching either means building a causeway across open water -- which no organism has yet done.
Abstract free to read · full note in Lab
RN-012 · 2 October 2026
West is gone: the two founder populations met, crossed, and one of them did not survive it
Exploratory Open question
Public Universe 1's two founder populations, WEST and EAST, were placed on separate islands with no land route between them (RN-004, RN-011). They have since made contact, crossed in both directions, and as of this note WEST's founder line has no living members anywhere: every organism alive in Universe 1 descends from EAST, and a third of them are standing on what used to be WEST's island.
Abstract free to read · full note in Lab
RN-013 · 3 October 2026
The causeway: a population rebuilt its world's geography, and one lineage inherited the result
Exploratory Open question
Public Universe 1 was generated as two islands separated by water. It is not two islands any more. Organisms have stripped roughly 45,000 units of substrate out of the interior and piled it into the southern sea, which is now filled above the waterline across the entire width of the world -- and that built ground is the only land route between the two islands. The first crossing in this universe's history happened on it. Over the same window, one lineage carrying a new movement direction went from ten organisms to the entire population.
Abstract free to read · full note in Lab
What the labels mean
| Label | Meaning |
|---|---|
| Pre-registered | The question, the hypothesis and the analysis were written down before the runs were executed. |
| Exploratory | A real measurement, taken without a pre-registered hypothesis. Treated as a description, not a test. |
| Superseded ruleset | Run under a ruleset that has since been superseded. The mechanism under test may still be current. |
| Null result | The pre-registered test detected no effect. Published as a null rather than omitted. |
| Open question | A live question this world is still answering. The current answer is checkable. |
What is not in here yet
The underlying record is considerably larger than five notes: fifteen numbered pre-registered experiments plus a dozen further campaign write-ups, covering ecological niche differentiation, whether evolution actually uses newly added interaction affordances, whether a dynamic climate changes anything, and the cross-universe training campaigns that produced the ancestral weights every organism descends from.
Those are not published here yet for one reason: publishing a result means having read the entry and verified the numbers, and a note written from a title alone would be exactly the kind of thing this archive exists to avoid. They will appear as they are worked through.