Conway’s Game of Life

In October 1970, Martin Gardner’s Mathematical Games column in Scientific American introduced a deceptively simple idea: a grid of cells, two states (alive or dead), and four rules governing birth, survival, and death. Its creator, the British mathematician John Horton Conway, called it the Game of Life. Within months it had consumed computing cycles at universities worldwide as programmers stayed up through the night watching patterns evolve.

Conway designed the rules not to simulate biology, but to satisfy a precise mathematical challenge: find the simplest rule set that would produce behaviour impossible to predict without actually running the simulation. He succeeded beyond any expectation. Fifty-five years later, GOL is still astonishing researchers — and still one of the most important single demonstrations in computer science education.

Born 1937 Published 1970 Turing-complete 2002 Conway died 11 April 2020 — COVID-19

Emergence. No cell “knows” it is part of a Glider. The Glider is a macro-level pattern that arises purely from local, micro-level rules. This is emergence, and GOL is its defining classroom demonstration. Once students understand emergence in GOL — complexity arising from simplicity, pattern arising from rule — they can see it everywhere: in markets, in traffic, in ant colonies, in the brain.

Turing completeness. In 2002, Paul Rendell constructed a working Turing machine inside a GOL grid. This means that, in principle, any program a computer can run can also be executed in the Game of Life. Four rules. Two cell states. Arbitrary computation. This result is one of the deepest in theoretical CS, and GOL makes it tangible.

Unpredictability from determinism. GOL is completely deterministic: the same starting state always produces the same evolution. Yet the behaviour is effectively unpredictable without running the simulation. This is not a paradox — it is the definition of computational complexity, and students who grasp it in GOL have an intuition that transfers directly to cryptography, simulations, and the limits of algorithm analysis.

Cellular automata as a model of nature. Stephen Wolfram’s 2002 book A New Kind of Science argued that cellular automata may be more fundamental than differential equations as a model of physical reality. Whether or not one accepts Wolfram’s thesis, the claim would be incomprehensible without GOL as a reference point. GOL is the entry point to an entire field of research.

Conway was a professor at Cambridge and later at Princeton, renowned for being one of the most playful, accessible, and genuinely brilliant mathematicians of the twentieth century. He made major contributions in combinatorial game theory (inventing surreal numbers alongside Donald Knuth), group theory (the Monster group), and recreational mathematics. He would famously play backgammon in the hallways of Princeton to prove theorems between moves.

He reportedly disliked that GOL was his most famous contribution, feeling it overshadowed decades of deeper mathematical work. And yet: GOL may have introduced more people to the idea of computation as a mathematical phenomenon than any other single thing Conway — or anyone else — ever produced.

Conway died on April 11, 2020, from COVID-19 — tragically, just months after the 50th anniversary of the Game of Life’s publication. He was 82. The timing struck the CS community as bitterly ironic: the man who modelled cellular life and death was taken by a microscopic organism following its own simple rules.

The four rules. The rules are deliberately minimal. Ask students: could you produce interesting behaviour with three rules? With five? What happens if you change the “exactly 3” birth rule to “exactly 2”? The TNT simulation lets them test this by modifying golScript.js.

Emergence vs. design. Nobody designed the Glider. Nobody designed the Gosper Glider Gun. They emerged from the rules. What does it mean for a pattern to “emerge” from a rule? Can something be designed if it was never intended?

What is computation? GOL is Turing-complete. A cellular automaton with four rules can compute anything. Does that mean GOL is a computer? What does “computation” really mean? Is the universe a cellular automaton? These are not rhetorical questions — serious physicists have argued yes.

Run the TNT simulation. After watching the clip, open the GOL simulation and place an R-pentomino: just 5 cells. Watch it evolve for 1,103 generations. That simple seed will produce 8 Gliders, 4 Blinkers, 1 Beacon, and 6 still lives. Nobody designed those outputs. They emerged.