A Short History of Sudoku, and Why Ours Has Only Six Rows

Sudoku feels ancient and Japanese. It is neither, quite. The real story runs through Indiana, Tokyo and a retired judge in Hong Kong.

Ask most people where sudoku comes from and they will say Japan, probably a long time ago. The name is Japanese. The puzzle is not especially old, and it was not born there.

Indiana, 1979

The puzzle we now call sudoku first appeared in an American magazine, Dell Pencil Puzzles and Word Games, in 1979. It was called Number Place. The magazine did not credit an author, which was normal for puzzle magazines of the time, but researchers later traced it to Howard Garns, a retired architect from Indiana. He died in 1989, years before his puzzle went round the world, and never knew what he had started.

Garns did not invent the idea from nothing. Grids in which every row and column holds each symbol once are called Latin squares, and the Swiss mathematician Leonhard Euler was writing about them in the eighteenth century. What Garns added was the boxes. That third constraint is the whole trick: it is what lets a handful of given digits pin down every other square.

Tokyo, 1984

Number Place crossed to Japan in 1984, when the puzzle publisher Nikoli began printing it. They gave it a long name that means, roughly, "the digits must stay single". It was soon shortened to the two syllables everyone knows.

Nikoli also set a house style. Their puzzles were made by hand, with the given digits arranged in a symmetrical pattern and a limit on how many a puzzle could start with. In Japan the puzzle became a steady favourite. Elsewhere, for twenty years, hardly anyone noticed.

London, 2004

The person who changed that was Wayne Gould, a retired judge who came across a sudoku book in a Tokyo bookshop. He spent several years writing a computer program that could produce puzzles quickly, then offered them to newspapers. The Times of London printed its first one in November 2004. Within months, rival papers had their own, and by the following summer sudoku was everywhere: on trains, on television, in airport shops.

It is worth noticing what made that possible. A crossword has to be written by a person who knows the language. A sudoku can be generated by a program and solved by anyone who can tell nine symbols apart. It travels without translation.

Why ours is six by six

The standard grid has eighty-one squares. A decent puzzle on it takes most people fifteen to thirty minutes, and on a phone the squares get small.

Sudoku Six uses a grid of thirty-six squares, split into boxes two rows tall and three columns wide. It takes about five minutes. Everything that makes the full-size puzzle satisfying is still there: the scanning, the "it cannot go here, so it must go there", the last few squares falling in a rush.

We also think the small grid is a better teacher. On a nine by nine board a beginner can stare for a long time before finding a first move. On six by six there is nearly always something to do, so you spend your time practising the reasoning instead of hunting for a place to start.

The rule that keeps it fair

Every puzzle on GridPencil is checked by the program that makes it. It will only show you a grid with exactly one solution. That matters, because a sudoku with two valid answers cannot be finished by logic alone. At some point you would have to guess, and a guess is the one thing the puzzle promises you will not need.

If you want more

Once the six by six version feels easy, two neighbours in the notebook use the same muscles. Binary Grid swaps six digits for two and adds a rule about runs of three. Magic Square is far older than sudoku and can be solved by pure reasoning in about four steps.

Of all the HTML5 games people play in a spare five minutes, sudoku may be the one with the oddest career: invented quietly in the American Midwest, named in Tokyo, made famous by a newspaper in London, and now small enough to fit in a browser tab.