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.

The first candidate failed a long-run watchability test

An earlier candidate ruleset was frozen, then tested with a 500,000-tick observational run on a disposable research universe before any public launch. The result was not a crash or an extinction: population activity continued throughout. What failed was the physical world. Ground material fell from roughly 600 occupied cells at the start to zero mass on both islands well before the run ended. Across 85 million attempted pushes, only 9,968 succeeded (about 0.0117%); across 18.8 million attempted pickups, only 6,322 succeeded (about 0.034%). No traversable-water cell ever changed across the entire run. A physically inert world is not a promising long-term spectacle, whatever else is happening in it, so this candidate was withdrawn before public launch.

Why the ruleset was reopened

That finding was treated as a simulation-development problem to fix, not a launch-readiness technicality to route around. The manipulation and material system was substantially reworked: a richer, renewable material substrate, new manipulation primitives, and proprioceptive feedback that measurably improved organisms' ability to aim their own actions. One internal training measurement found the average number of attempts needed before an organism's first successful action fell from 306.9 to 181.4 across the first several generations under the new substrate, evidence that evolution could now learn to use it.

The 500,000-tick acceptance run

The reworked ruleset was then held to the same test the first candidate failed, run on the same seed for direct comparison. It completed the full 500,000 ticks with zero restarts and no invariant violations. Compared directly against the earlier candidate: renewable material stock never approached negligible and never flatlined; manipulation success stayed one to two orders of magnitude above the earlier candidate throughout, even at its lowest point; and physical activity, measured by how much of the terrain was passable, trended upward into the later part of the run rather than collapsing. Population peaked at 1,055 during the run.

What actually persisted, stated precisely

Ground-material occupancy showed genuine, substantial persistence for a specific minority of locations: 96 cells held material for streaks of 100,000 ticks or more, 33 for streaks of 250,000 ticks or more, and one cell was occupied at nearly every one of the run's 101 sampled points. These durations comfortably exceed any individual organism's lifespan. At the same time, the majority of touched locations, 86% of them, remained transient rather than persistent. A separate "obstruction" mechanism this study specifically looked for was never observed even once across the entire run: a genuine, reproducible null result for that mechanism, reported as such rather than folded into a more flattering average.

What this study does not show

A large number of manipulation attempts is not the same thing as persistent organisation, and this project does not conflate the two: at roughly 0.2 to 0.4% success rates, tens of millions of attempts produced on the order of 75,000 to 99,000 actual successful modifications, and it is that smaller number, together with direct tracking of individual cells over time, that supports the persistence finding above, not the raw attempt count. One location was repeatedly modified and re-modified over the run in a way that is suggestive, but a single suggestive location is not a demonstrated pattern, and this project's instrumentation cannot currently distinguish "organisms behaviourally return to this place" from "this place's terrain geometry mechanically tends to reaccumulate material regardless of any organism's history there." That question is recorded as open, not resolved by assertion.

No demonstrated higher-order intentional organisation

Nothing in this project's evidence supports organisms planning, coordinating, building, or otherwise organising their environment with intent. What is supported is narrower and still genuine: the physical substrate now sustains real, renewable activity and, for a real subset of locations, modifications that outlast the individual actions that produced them and very likely outlast the individual organisms that produced them.

Zero crossing, zero contact: a result, not a failure to hide

Across the 500,000-tick acceptance run, the two founder populations never crossed the water between the islands and never made contact. This is reported as a direct observational result, not downplayed or omitted because it is a null finding. Public Universe 1 begins from the same physical possibility space; whether or when its own two populations meet, if ever, is an open question this site will report on honestly as it happens, on the Universe 1 page.

Limitations

• This is evidence from one seed on a disposable research universe, not a statistical claim about every possible run under this ruleset.
• A single suggestive pattern of repeated modification at one location is explicitly not treated as evidence of organism intent, memory, or preference.
• Absence of an observed behaviour across 500,000 ticks is evidence about that run; it is not a permanent claim that the ruleset can never produce it.

See methodology for the discipline behind how these findings were produced and reported.