Conway's Game of Life: Rules and How to Play the Cellular Automaton
Updated
Conway's Game of Life is a single-player cellular automaton invented in 1970 by Cambridge mathematician John Conway. You don't control individual cells during play — you set an initial pattern, then watch the grid evolve generation by generation according to four fixed rules. It's worth learning if you want a hands-on introduction to cellular automata, emergence, or simple computational modeling, and you can run it online or in math software such as Mathematica.
What makes Life a cellular automaton
A cellular automaton is a grid of cells, each in a finite set of states, updated in discrete steps by a rule that looks only at each cell's neighbors. Life fits this definition exactly:
- Grid: an infinite (or bounded, in most implementations) 2D square grid.
- States: each cell is either alive (on) or dead (off).
- Neighborhood: the 8 cells surrounding a cell (Moore neighborhood).
- Update: all cells change state simultaneously, once per generation, based only on the previous generation.
No central controller and no randomness are required. Complex behavior — gliders, oscillators, even logic gates — emerges from these local rules alone.
The four rules
At each step, every cell's next state is decided by how many of its 8 neighbors are alive:
| Current state | Live neighbors | Next state |
|---|---|---|
| Alive | 2 or 3 | Stays alive |
| Alive | 0, 1, or 4–8 | Dies |
| Dead | Exactly 3 | Becomes alive (birth) |
| Dead | Any other count | Stays dead |
In shorthand: survival on 2 or 3, death from underpopulation (fewer than 2) or overpopulation (more than 3), and birth on exactly 3. These rules are applied to the whole grid at once, so a cell's fate never depends on the order you check cells.
Setting up a pattern and stepping through generations
- Choose an initial pattern. Start with a small arrangement of live cells — a single block, a line of three (a "blinker"), or a glider. The starting configuration is called the seed.
- Apply the rules once. Compute each cell's neighbor count from the current grid and write the next generation. In software, this is a single update step.
- Repeat. Feed the new generation back in. Each iteration is one generation, and the pattern may stabilize, cycle, or move.
Expected result: some seeds vanish immediately, some freeze, some pulse, and some travel across the grid. Common pitfall: updating cells in place instead of from a copy of the previous generation — this corrupts the result because later cells would see already-changed neighbors.
Common pattern types
- Still lifes: stable patterns that never change, e.g. the block (2×2 square).
- Oscillators: patterns that return to their starting shape after a fixed number of generations, e.g. the blinker (period 2).
- Spaceships: patterns that translate across the grid, e.g. the glider, which moves diagonally and repeats every 4 generations.
Recognizing these categories helps you predict whether a seed will die out, settle, or keep moving.
Trying it online or in math software
The site Play John Conway's Game of Life (playgameoflife.com) lets you play the Game of Life online as a single-player game, with a Life Lexicon available for reference on named patterns. You can draw a seed on the grid and step through generations to watch it evolve.
For a programmable approach, Mathematica-style environments let you represent the grid as an array and apply the neighbor-count rule as a function, then iterate it. This is useful when you want to test many seeds, measure population over time, or visualize larger patterns than you'd draw by hand.
Either way, the workflow is the same: define a seed, apply the four rules repeatedly, and observe whether the pattern dies, stabilizes, oscillates, or travels.