The history of backgammon

Backgammon is one of the oldest board games still widely played. Its family tree reaches back to the race games of the ancient Near East, though the game as we know it took shape much later.

Ancient race games

People have played race games, where pieces travel along a track according to throws of dice or other lots, for about five thousand years. Two of the best known come from the ancient Near East.

Senet was played in Egypt from early in its history. Boards appear in tombs, including that of Tutankhamun, and paintings show people playing it. It used a track of thirty squares and throwing sticks rather than cubic dice. The exact rules are not known; modern versions are reconstructions.

The Royal Game of Ur is named after the elaborate boards found by Leonard Woolley in the royal tombs at Ur, in modern Iraq, which date to the middle of the third millennium BCE. The game survived for a very long time. A much later Babylonian clay tablet describes a set of rules, and the British Museum's Irving Finkel used it to reconstruct how the game was played.

These games are relatives of backgammon in the sense that they are race games with chance, and they show how old the idea is. There is no evidence that backgammon descends directly from either of them, and claims that backgammon itself is five thousand years old should be read as claims about this wider family.

Tabula in Rome

The Romans played a game called ludus duodecim scriptorum, the "game of twelve lines", on a board of three rows. Later they played tabula, which used two rows of twelve points, like a modern backgammon board.

The best evidence for what tabula looked like is a short Greek poem by Agathias, from the 6th century. It describes a disastrous turn by the emperor Zeno, who reigned in the late 5th century, in enough detail to reconstruct the position. The board had 24 points, each player had fifteen pieces, and they moved in opposite directions. The poem says three dice were thrown, a difference from the modern game, but the family resemblance is clear.

Nard in Persia

In Persia the game was called nard (also nardshir). A Middle Persian text tells how the game was invented by the sage Wuzurgmihr (Bozorgmehr) as Persia's answer to chess, which had arrived from India. In the story the board stands for the year, its points for days, and the movement of the pieces for the turning heavens. It is a legend rather than history, but it shows that nard was well established and highly regarded in the Sasanian period.

Nard spread with Persian and later Arab culture across the Middle East and Central Asia. Names derived from it, like nardy in Russian, are still used today, and long nardy remains the main form of the game across much of that region (see variants).

Tables in medieval Europe

In medieval Europe the family of games played on this board was called tables (from the Latin tabula). Tables was popular at every level of society, often played for money, and was periodically condemned or banned by church and civic authorities for its association with gambling.

The Libro de los juegos (Book of Games), produced for King Alfonso X of Castile in the 13th century, describes and illustrates many games, including a whole series of tables games with different setups and rules. Some are close to modern backgammon. Other European variants developed too, including French and Italian games and the Greek and Turkish forms that are still played (see variants).

The name backgammon

The word backgammon appears in English in the 17th century, for a version of tables close to the modern game. The usual explanation is that it comes from "back" and the Middle English "gamen", game, perhaps because checkers that are hit must go back to the start. Over time the new name replaced "tables".

In 1743 Edmond Hoyle published A Short Treatise on the Game of Back-Gammon. Hoyle, best known for his writing on whist, set out the rules and gave advice on play and odds. His treatise became a standard reference, and "according to Hoyle" became a phrase meaning "by the rules". The rules of modern backgammon are essentially those Hoyle described, with the doubling cube added later.

The doubling cube

The biggest change to the game in modern times came in the 1920s, when players in New York clubs began using a doubling cube. The idea was simple: at any point, a player could offer to double the stakes, and the opponent had to accept or concede. Who invented it is unknown.

The cube transformed the game. It rewarded judgement as well as checker play, since knowing when to double and when to take became the most important skill. It also shortened hopeless games, since a player in a bad position could simply refuse a double. Today every serious game of backgammon uses the cube; see the doubling cube guide for how it works.

The modern boom

In the 1960s backgammon became fashionable. Prince Alexis Obolensky, often called the father of modern backgammon, organised the first international tournament in the Bahamas in 1964 and helped set up the tournament circuit. Clubs opened, international championships followed, and by the 1970s backgammon was a craze, played in nightclubs and featured in magazines.

The boom also brought serious study. Books such as Paul Magriel's Backgammon (1976) explained strategy systematically, with concepts like priming, timing and the pip count that are still taught today. The World Backgammon Championship moved to Monte Carlo in the late 1970s and has been held there since. Match play with the Crawford rule became the standard tournament format (see match play).

Computers and neural networks

Backgammon has been important in the history of computer game playing. In 1979 Hans Berliner's program BKG 9.8 beat the reigning world champion, Luigi Villa, in a short exhibition match, the first time a program had beaten a world champion at a board game. Berliner himself acknowledged that the dice had favoured the program.

The real breakthrough came with Gerald Tesauro's TD-Gammon, developed at IBM in the early 1990s. It was a neural network that learned by playing against itself millions of times, adjusting its evaluations with a method called temporal difference learning. It came close to the level of the best human players, and some of its choices challenged accepted wisdom. Its opening plays, such as splitting the back checkers with some rolls where humans had preferred to slot, changed how experts play.

Commercial programs followed in the 1990s, notably Jellyfish and later Snowie. The free, open source GNU Backgammon appeared around the turn of the century and is still widely used, and eXtreme Gammon later became the standard tool for serious players. All of them use neural networks and can roll out positions, playing them thousands of times to measure the best play. Their analysis has refined opening theory, cube decisions and match equity tables, and top programs now play better than any human.

Online play grew alongside them, starting with text-based servers in the early 1990s and moving to graphical sites and apps. The computer opponent in Backgammon Lounge follows the same approach: its Hard level uses an opening book and a two-ply search with a neural network trained by self-play. Read about it on play vs computer, or play a game now.

Questions

Keep learning