The History of Backgammon
A family tree running through nearly every civilization that has rolled dice.
Dice Games Older Than Writing
Backgammon's lineage runs back through the Royal Game of Ur and the Roman game Ludus Duodecim Scriptorum ("Game of Twelve Lines"), later simplified into Tabula - a race game for two dice and 15 pieces per side, already recognizably close to backgammon by late antiquity.
Emperor Claudius reportedly loved Tabula so much that he wrote a (now lost) book of strategy for it, and had a board built into his carriage so he could play while travelling.
Tables, Trictrac, and the English Name
Throughout medieval Europe, backgammon-like race games were known collectively as "Tables," with regional variants like French trictrac adding their own scoring quirks. The English name "backgammon" is first recorded in the 17th century, likely from "back" and an old word for "game" - a nod to pieces sometimes having to re-enter and cross the board again.
The doubling cube, which lets either player raise the stakes mid-game, was introduced in New York in the 1920s and transformed backgammon from a pure race into the sharp risk-management game tournament players know today (this app doesn't include the cube - see its rules page for why).
An Early Win for Artificial Intelligence
Backgammon holds a special place in AI history: in 1992, a program called TD-Gammon, built by IBM researcher Gerald Tesauro, used a neural network that learned by playing itself millions of times - one of the earliest successes of what's now called reinforcement learning.
TD-Gammon played at a level competitive with the world's best humans years before similar techniques conquered chess or Go, and its unconventional strategic ideas ended up changing how top human players approached the game.