GOL-012 · technical companion
Free Cell
The grid becomes a habitat rather than the organism: independent units reproduce, move, stay, or die from local crowding, following authored compass preferences through deterministic conflicts.
At a glance
- geometry
- Square habitat + mobile units
- neighborhood
- Moore · 8 adjacent positions
- states
- Empty position / independent unit + action evidence
- update
- Synchronous birth, fate, and movement proposal rounds
- renderer
- Canvas 2D
What to explore
Independent units should visibly reproduce and travel through a habitat while all simultaneous conflicts remain fair and reproducible. The original dense-field composition was intended to sustain both birth and movement for hundreds of Steps; mechanical correctness that becomes extinct almost immediately does not satisfy the experiment.
Rules, update order and replay
Every Step freezes occupancy and Moore counts. Only occupied positions contain units and evaluate behavior. Birth proposals resolve first through a complete ranked compass order; accepted newborns reserve their targets and do not act until the next Step. Each parent then dies, moves, explicitly survives, or stays according to its count. Movement considers only positions empty in the frozen field and excludes accepted births, so units cannot swap or follow into a same-Step vacancy. Unique claims succeed. Multiply claimed destinations accept no one, and all contenders retry their next preference synchronously. State has no unit IDs and includes occupancy, source outcomes, destination events, accepted birth/movement targets, and generation. Replay requires full state, configuration, topology, profile, and start. Reset/Clear/manual edits clear transient evidence.
Reading the simulation
The renderer distinguishes established units, newborns, movement arrivals, deaths, departures, contested births, and contested moves. Inspection pauses the world, outlines the selected position and frozen-empty candidates, runs the exact field-wide deterministic preview, and reports count, matching actions, ordered candidate counts, accepted targets, prior evidence, and the source's next fate. Every evidence class is named in text and legend as well as color.
Reproduce an opening
Download the example above, open the simulation, expand the controls, then use Share & reproduce → Import JSON. The example uses authored rules and a fixed seeded starting field; it is independent of the site’s editable default composition. Import starts paused at Step zero.
Share links and JSON reproduce the configured starting conditions, including hidden layers rebuilt by Reset. They do not save an evolved board, hand-painted cells, speed or display preferences. Determinism applies to the same rule version and Step sequence; real-time playback rates vary by device.
Larger fields and more involved rules can reduce playback speed. Read the configuration format and automated-access guide for compatibility details.