Loading
Fetching the latest data…
Fetching the latest data…
Emergence is the word for behaviour that appears in a system without being written into its rules, and it is used loosely enough that it often means nothing at all. A cellular automaton is the cleanest place to pin it down: the entire specification is nine birth bits and nine survival bits, you can read the whole thing in a second, and nothing in it mentions the gliders that some rules produce. This module lets you dial the rule and measure what came out — which turns a vague word into a number you can argue about.
Before you start — None whatsoever. If you can decide whether a cell has two neighbours or three, you have the prerequisites. The entropy readout is the only place any mathematics appears, and the intuitive tier explains it without any.
Every cell looks at its eight neighbours and counts how many are alive. A dead cell comes to life at the counts in the birth list; a living one stays alive at the counts in the survival list. That is the entire rule — eighteen bits, no exceptions, no special cases, and nothing anywhere that mentions a shape.
Do — Open Conway's Life and read the five gauges. Then turn birth 3 off and watch the grid empty: one bit out of eighteen is the difference between the most studied automaton in existence and nothing at all.
The first failure mode is the one that looks like success. A majority rule settles into large stable domains within a few dozen steps: every cell surrounded by its own kind, nothing changing anywhere. It scores at the very top for structure, and it is a dead end.
Do — Open Vote and read structure against activity — 100 and 0. Then compare the two numbers against Life. High structure alone does not tell you a rule is interesting; it tells you the cells are in clumps.
The opposite failure. Under Seeds nothing survives even one step — every live cell dies and new ones are born wherever exactly two neighbours were alive. The population explodes rather than vanishing, so the grid stays full and busy forever, and none of the busyness holds together.
Do — Open Seeds and watch activity go near the top while structure falls below Life. Persistence stays at 100 the whole time, which is the honest reading: something IS alive. It is just never the same something.
Between frozen and boiling is a thin region where structures last long enough to move, collide and interact. It is the only regime in which a cellular automaton has ever been shown to compute anything, and Life sits in it — moderate activity, high structure. No single gauge finds it, which is exactly why there are five.
Do — Land activity in the 5–40 band with structure at 70 or higher, on a rule that is not Life. Start from Day & Night and change one bit at a time; most changes fall out of the band immediately, which tells you how narrow it is.
Novelty is measured against the rule alone: change the seed, the density or the step count and it cannot move. That separation is the point of the module. The rule is a hypothesis about what kind of behaviour is possible; the seed is one trial of it. A result that only appears at one seed is a result about that seed.
Do — Fix a rule you like and run it at three different seeds. The pattern changes completely and the profile barely moves — that stability is what makes the reading a property of the rule rather than of your luck.
Did anything survive? Some rules wipe the grid clean in a handful of steps. This is the share of the second half of the run in which at least one cell was still alive.
Try — Open Seeds (B2/S) at a low density and watch the population explode rather than vanish — persistence answers "is anything there", not "is anything happening".
How much is still moving. A frozen pattern scores 0; a boiling one scores 100. The interesting rules are neither — they are somewhere in the middle, and that band is what the Edge of Chaos challenge asks for.
Try — Compare Vote, Life and Seeds in that order. Activity goes 0, low, very high — and only the middle one has structures in it. No single gauge finds class IV; the profile does.
Are the living cells in clumps, or scattered like static? This is the share of live cells with at least two live neighbours.
Try — Run Vote and read structure against activity: 100 and 0. Perfectly structured and completely dead is a real result, and not an interesting one.
How many different little patterns appear. Look at every 2×2 window in the final grid: if they are all the same, entropy is 0; if all sixteen possibilities show up equally often, it is at maximum.
Try — A checkerboard uses exactly two of the sixteen block types, so it scores 25 — one bit out of four. High entropy is variety, not disorder, and not quality.
A rule is two lists: the neighbour counts that bring a cell to life, and the counts that keep it alive. This scores how far your lists are from every rule that has a name.
Try — Push novelty to 100 while keeping persistence at 100. Novelty alone is trivial — any rule nobody bothered to name is far from the corpus, usually because it does nothing.
Surfacing sources you can verify is a deliberate anti-pseudoscience measure, not a bibliography. Nothing on this page asks to be taken on trust.
Glossary — every term used above, defined once.