What is Evolution Mode?
Evolution Mode is an artificial-life variation on Conway's Game of Life. Instead of one global rule controlling every cell, living cells carry allele pairs for personal birth and survival rules, vigor, and mutability. Newborns combine neighboring cells' alleles and may mutate, so different rule phenotypes can spread through the board.
It is inspired by inheritance and selection, but it is not a biological model. Aging, inheritance, mutation, food supply, and recovery make it a deliberately stochastic cellular automaton. Replaying the same positions can produce a different future.
🎮 Classic Mode vs Evolution Mode
Classic: Every cell follows the selected global totalistic or isotropic rule. Given the same board and edge setting, every step is deterministic; age only changes the drawing.
Evolution: Expressed personal rules govern births and survival. Inheritance, mutation, and biological age introduce chance. Living ecology adds food and recovery; Open Mutation keeps the original crowding and disease rules.
🧬 The Stored Genotype
Each cell stores five allele pairs. Four pairs currently affect behavior or appearance; the hue pair is retained in the genotype but is not expressed. These are game mechanics, not a claim of biologically complete DNA.
| Gene | Range | What It Does | Dominance |
|---|---|---|---|
| Stored hue | 0-360° | Legacy inherited data; current color ignores it | Not expressed or mutated |
| Vigor | 0-100 | Weights parent selection and can rescue a failed survival check | Higher allele is expressed |
| Mutability | 0-100 | Adds 0-2 percentage points to a newborn's mutation probability | Higher allele is expressed |
| Birth rule | 1 or 2 counts, 1-8 | Helps determine which empty sites can produce offspring | Two-count allele wins; ties use the genotype's first allele |
| Survival rule | 1 or 2 counts, 1-8 | Specifies the neighbor counts under which that cell survives | Two-count allele wins; ties use the genotype's first allele |
🎨 Rules Create the Color
A cell's displayed hue is calculated from its expressed birth and survival counts. There is no color-mutation event. Matching personal rules always produce the same hue, although different rules can land on similar colors.
👨👩👧 How a Newborn Is Made
A birth has two stages. First, the empty site groups living neighbors by their exact expressed birth rule. The most common group is the dominant coalition. If several coalitions tie, their allowed neighbor counts are combined in sorted order, so rotating a setup or changing scan order cannot break the tie. The site's actual living-neighbor count must appear in that combined set.
Once birth is allowed, only neighbors whose own expressed birth rule contains that actual count may pass genes. A lineage cannot authorize a different situation and then reproduce through this newborn. Vigor weighting chooses from this eligible parent pool.
Parent Selection
With two or more authorizing neighbors, two parent slots are sampled with weight equal to expressed vigor + 20. Higher vigor helps, but every eligible neighbor retains a chance.
Allele Inheritance
For each stored trait, the child receives one random allele from each selected parent. With only one parent, its genotype is cloned before mutation.
Phenotype
Vigor and mutability express the higher allele. Rule traits express the allele containing more active neighbor counts; a tie favors the genotype's first stored allele.
Newborn Mutation
Mutation is checked only for births. Probability is the slider rate plus expressed mutability ÷ 5000, so inherited mutability contributes at most 2 percentage points.
🎯 Personal Rules in Action
Seeded cells begin with the familiar B3/S23 phenotype. A rule mutation can replace, add, remove, or shift an allele's allowed neighbor counts. A cell with expressed S24, for example, passes its initial survival test with 2 or 4 neighbors; an expressed B34 rule can help authorize births at empty sites with 3 or 4 neighbors.
📊 No Hidden Global Adaptation
The rule badge shows the most populous expressed personal rule, labeled the Dominant lineage. It is not a global rule applied to the board. “Species” and “lineage” are phenotype buckets: cells in one bucket can still have different vigor, mutability, or recessive alleles. Fitness and Pressure summarize density and recent population behavior; they do not adapt the rules.
A living landscape
Living ecology is the default sandbox environment. Food varies across the board and is consumed by living colonies. Empty ground replenishes its supply. After a cell dies, depleted ground rests for two generations; severely depleted ground rests for seven. Fresh ground stays open to moving patterns. Births require enough food, and single-parent expansion costs more.
Parents with compatible expressed birth and survival strategies mate together, helping colonies retain their identity. Vigor cannot rescue isolated or starving cells, and cells have a maximum lifespan of 60 generations. Local resource pressure replaces population-wide plagues. The habitat map shows actual food, living cells, and recovering ground.
Choose Open Mutation to return to the original rules below. You can switch environments during sandbox playback without replacing the board. Daily challenges and Trials always keep their fixed original rules. Ecology and mixed-environment sandbox runs are excluded from the older Community Runs standings.
Open Mutation: Aging, Vigor, Crowding, and Diversity
Passing a personal B/S check is only the beginning. Evolution Mode applies additional probabilistic checks every generation, which is why a familiar still life or oscillator is not guaranteed to persist.
Biological Age
After age 50, a rule-surviving cell's death chance rises by 4 percentage points per generation, reaching 100% at age 75 before possible rescue effects.
Vigor Rescue
If a cell would die, vigor above 70 grants a rescue chance of (vigor − 70) ÷ 200: at most 15%. Low vigor does not cause a separate random death.
Local Crowding
A surviving cell with 4 neighbors within 30° of its rule-derived hue has a 15% death risk; 5 or more raises it to 30%.
Diversity Bonus
A cell still marked to die, with at least 2 total neighbors but no more than 1 similar-color neighbor, gets a 25% rescue chance.
Infant Mortality
An allowed birth has 10% mortality at 4 neighbors or 25% at 5+. Five or more neighbors near the parents' average hue add another 20% check.
Periodic Cull
Every 50 generations, boards at least 12% full face a density-scaled disease cull. Base risk ranges from 5-35%; the most populous color bucket takes up to 1.6× risk and the runner-up up to 1.3×. Sparse experiments are not culled.
🔬 Mutation Types
After inheritance, a successful mutation check targets one allele. The newborn flashes white for that generation. A mutation can land on a recessive allele—or an add operation can pick a duplicate count—so the event does not always cause an immediate rule or color change.
| Mutation | Share of mutations | Change Amount |
|---|---|---|
| Vigor | 20% | Shift one allele by up to ±20, clamped to 0-100 |
| Mutability | 15% | Shift one allele by up to ±15, clamped to 0-100 |
| Birth Rule | 30% | Add/remove a count or shift a count by 1 |
| Survive Rule | 35% | Add/remove a count or shift a count by 1 |
👁️ Reading the Interface
The visuals expose the expressed phenotype, while the lineage panel counts exact personal rules:
| Visual | Meaning |
|---|---|
| Cell hue | Calculated from expressed B/S counts, not from a color gene |
| Brightness / saturation | Cells dim and desaturate with age; higher vigor raises base saturation |
| White flash | A newborn passed through the mutation branch this generation |
| Species by personal rule | Counts and population share for each exact expressed B/S phenotype |
| Dominant lineage badge | The currently most numerous expressed B/S phenotype—not a board-wide rule |
🎲 Stability Is Different Here
Matching cell positions do not prove a repeat because ages and hidden alleles can differ and random checks remain ahead. The open Sandbox therefore treats only extinction as an automatic terminal state; finite Trials also stop at their explicit success or failure condition.
⭐ Sandbox and Evolution Trials
The Run Mode selector is available in the desktop Evolution panel and the mobile Evolution sheet. Choose an experiment seed, starting habitat, and explicit opening policy before launch. Ranked Balanced and Sparse habitats use the same deterministic 50×50 torus on every device. My Board preserves a nonempty custom setup as unranked Practice. The canvas HUD reports the objective, horizon, live metric, and policy checkpoints while the Trial runs.
| Mode | Objective | Finish |
|---|---|---|
| Sandbox | Open-ended inherited-rule experiment | No Trial horizon; use Play, Pause, Step, and Clear normally |
| Endurance | After generation 25, stay at or above 300 cells; finish generation 250 at or above 500 and earn at least 600 protocol points | Choose policy at 80 and 160; Bronze 600, Silver 740, Gold 940 |
| Diversity | Finish 300 with a qualifying streak of 30: effective diversity ≥3 and dominant phenotype ≤65% | Choose policy at 100 and 200; Silver/Gold require longer balance, more qualifying generations, and higher mean diversity |
| Adaptation | Survive 300 and average at least 20 fitness across the final 50 generations | Choose policy at 100 and 200; Bronze 20, Silver 25, Gold 30 final-window mean |
🏁 Results, Retry, and Medals
A Trial commits one terminal result, stops the simulation, and shows its objective, score, measured ecology, and deterministic protocol receipt. Retry Same Seed restores the exact prepared setup; New Seed reruns the strategy under new randomness. Ranked guided runs keep one idempotent personal-best medal per Trial. My Board is clearly marked Practice and cannot change progression. Trial results are excluded from Community Runs.
🔒 A Run Keeps One Configuration
The first Play or Step captures the run's mode, board size, edge wrapping, active rule, mutation rate, and leaderboard eligibility. Those setup controls, pattern placement, randomization, and rule import remain locked until Clear. Speed and visual aids can still change because they do not alter the automaton's result.
For Classic experiments, Export Board (RLE) writes the trimmed live-cell bitmap and the exact active totalistic or Hensel notation. Importing that RLE restores partial Hensel neighborhood classes exactly rather than expanding a rule such as a partial B3 class into all B3 neighborhoods. The imported bitmap is then selected for placement. Evolution genotypes are not part of standard RLE and are not claimed to round-trip.
🕹️ Strategy: Choose the Experiment
For Classic Mode
Use known patterns and precise spacing, choose wrapping deliberately, and tune the global B/S or Hensel rule. This is the mode for repeatable construction, exact periods, and deterministic comparisons.
For Evolution Mode
Seed active patterns or a soup with room to spread. Mutations occur only at births, so still lifes create no new variation and still face biological aging; only sufficiently dense boards face the periodic cull.
Tune Mutation Pace
A lower slider preserves successful combinations longer. A higher slider explores personal rules faster but can disrupt an established lineage. Hidden mutability adds 0-2 percentage points either way.
Leave Breathing Room
Dense same-color clusters face local crowding, infant mortality, and harsher culls. Expanding fronts and mixed-color boundaries give rare phenotypes more opportunities.
🧬 Ready to Evolve?
Toggle Evolution Mode, seed an active population, then choose the open Sandbox or a finite Trial. Run it twice: chance, inheritance, and ecology can tell a different story each time.
▶ Start Evolving