Snake: The 1997 Nokia Game That Taught the World to Play on a Phone

Snake is remembered as a Nokia game. It is not, quite: Nokia shipped it in 1997, and by then the idea was already two decades old. What Nokia did was put it in the pocket of a very large fraction of the human species.

1976: Blockade

The ancestor is Blockade, an arcade cabinet released by Gremlin in 1976. Two players each drove a dot that left a solid trail behind it, and the goal was to survive longer than your opponent by forcing them into a wall, or into a trail, without crashing into one yourself.

It is the same core idea in its purest form: a growing obstacle that you yourself are creating. Blockade was widely cloned almost immediately: Bigfoot Bonkers, Surround on the Atari 2600, and later, most famously, the light cycles in Tron, which are the same game with better marketing.

The single-player variant with food to collect appeared on home computers through the late 1970s and 1980s under a long list of names: Worm, Nibbler, Nibbles. By the time QBasic shipped with Nibbles as a sample program in 1991, the game was effectively folk knowledge: no single owner, endlessly reimplemented, familiar to everyone who had ever typed on a computer.

1997: the Nokia 6110

Taneli Armanto, a design engineer at Nokia, adapted the game for the 6110 handset. The constraints were severe by any standard: a monochrome display of 84×48 pixels, or 1,344 usable cells at best, and a numeric keypad as the only input device.

Those limits turned out to suit the game exactly. Snake needs a grid, not resolution. It needs four directions, not a controller. It needs no sound, no colour and no story. There is possibly no other game in existence that would have survived that hardware unchanged, and Snake did not merely survive it. The version on that screen is the version most people picture when they hear the word.

Why it never got old

Most games from 1997 are unplayable now, or playable only through nostalgia. Snake is not, and the reason is structural: the difficulty comes from the player's own success, not from a designer's escalation.

There are no levels, no enemies and no scripted increase in speed in the original design. The board gets harder purely because your own body is filling it. That produces a difficulty curve perfectly matched to whoever is playing. A beginner faces an easy board because their snake is short, and an expert faces a nearly impossible one for exactly the same reason.

It also means the game has real strategy hiding inside what looks like a reflex test: the difference between a score of 40 and a score of 150 is almost entirely about how you shape the free space. We wrote that up in Snake strategy.

The mathematics underneath

Snake turns out to be a genuinely interesting computational problem, which is unusual for a game this small. Playing perfectly means finding a route that visits every cell exactly once, a Hamiltonian cycle, and finding those efficiently is a well-known hard problem in graph theory.

That is why Snake keeps reappearing as a benchmark for pathfinding and reinforcement learning research. A game that fits in a few hundred bytes contains a problem that is still awkward for machines. On a standard 20×20 grid the theoretical maximum is a snake of length 400, which fills the board entirely; solving that reliably is a research result, not a high score.

Multiplayer, which is where it started

There is a nice symmetry in the modern multiplayer arena versions: they are, functionally, a return to Blockade. Several players, one shared field, everybody's trail a wall for everyone else. The single-player version with food was the variant, and the last twenty years have quietly brought the game back to what it was in 1976, just with more than two players and no cabinet.