The history of Dots and Boxes

A mathematician's game from 1889

Dots and Boxes has a precise origin, which is rare for a paper game: it was published in 1889 by the French mathematician Édouard Lucas, under the name La Pipopipette. Lucas is better known for the Tower of Hanoi, which he also invented, and for the Lucas numbers.

Lucas presented the game as a diversion, and it spread the way paper games spread — through schools, in the margins of exercise books, with no publisher and no equipment. Within a generation it was being played under local names everywhere: Squares, Paddocks, Dots, Käsekästchen ("little cheese boxes") in German, Timbiriche in Mexico.

Its cultural position is unusual: universally known, almost never taken seriously, and — as it turns out — one of the deepest games ever played on squared paper.

Berlekamp, and the theory nobody expected

The mathematician Elwyn Berlekamp — a founder of combinatorial game theory, and co-author of Winning Ways for Your Mathematical Plays — spent decades analysing Dots and Boxes, and published the results in 2000 as The Dots and Boxes Game: Sophisticated Child's Play.

The book demonstrated something counter-intuitive: a game children play badly for fun has a rigorous theory with named concepts and calculable values. The essentials:

• The endgame decomposes into chains and loops, and the whole game is about their number and length.
• The double-cross — declining the last two boxes of a chain to keep your opponent on the move — is the central technique.
• There is a parity rule connecting the number of long chains, the board size and who should aim to move first.
• Positions can be assigned values using Nimber theory, connecting the game to Nim.

Computers and the modern game

Dots and Boxes is a standard exercise in game programming, and small boards have been solved exhaustively. The state space grows very quickly with board size — every line is either drawn or not, so a 5×5 board has 60 lines and 2⁶⁰ line states before symmetry — which puts the standard tournament board beyond complete solution while leaving the endgame entirely computable.

That is why implementations, including the one on this site, use a hybrid: heuristic play in the opening, and exact analysis from the point where the chains form. It mirrors how strong humans play, because the opening genuinely does not matter much and the endgame is everything.

There is a small competitive scene, mostly online, and the theory is well enough established that serious games are decided by chain counting rather than by tactical oversight.

Common questions