§01 · Research axis · Life

We grow small lives
in flat worlds,
then we ask them to live in ours.

This programme runs a sequence of two-dimensional petris: voxel worlds with weather, scarcity, and occasionally predators, populated by agents whose policies are not designed but grown. They reproduce, they forget, they invent rituals of their own. The ones that survive long enough are extracted, compiled, and ported into embodied trials.

Active petris12
Generations · 72h12,944
Export trials · MTD03

Petri № 047. Day 83, generation 9,102.

Petri 47 · seed 1729 · entropy budget 0.82 nats/s
World size256 × 256
Tick42 ms
Agents1,284
Species3 stable · 2 fringe

The working hypothesis: real
intelligence does not arrive.
It persists.

Stage i

Grow.

Seed a petri with minimal policies and maximal hostility. Weather, scarcity, and occasional catastrophes. No reward is written by us; the reward is not dying.

Stage ii

Archive.

At each major extinction, lineages are frozen. The archive is a museum of local maxima: not training data, not a curriculum, but a set of strategies that happened to persist.

Stage iii

Port.

Archived lineages are compiled into policies that run on embodied platforms (small drones, tabletop robots, microfluidic petris) and given a different scarcity to confront.

Stage iv

Read what breaks.

Most exports fail. The failures are the interesting part: where the 2D world gave the lineage a habit that only works flat. We study the seams, not the successes.

First embodied trial, in progress.

E·07
Lineage
petri 047 · first lineage
Vehicle
to be confirmed
Field note
Lineage archived, embodiment target under selection.
In progress