On a recent winter afternoon, I sat outside a posh hotel ballroom with Masayuki Mochizuki, the best backgammon player in the world. Mochy, as he’s known, was fresh off a victory in the tournament taking place in the hotel, the skitter of dice still audible as we spoke.
Backgammon is one of the oldest games still played. Egyptologists smashing into ancient burial tombs discovered pristine copies nearly identical to today’s game. Its modern history is more vibrant, booming in the 1970s, acquiring a cocaine-and-disco sheen, and played by Tina Turner, Pink Floyd, Hugh Hefner, and James Bond. The game, an intoxicating mixture of luck and skill, became a popular theater of gambling and panache.
Mochy, 43, picked up backgammon on the ebb of this tide, as a poor university student in Tokyo. “I could make a hundred dollars a day,” he told me. “That was big money for me at the time.” He learned the game as a new wave swelled, as computers started to play strong backgammon. Every night, he analyzed positions that he’d written down during the day, plugging them into his desktop.
Ever since there have been computers, their programmers have made them play games—small, tractable models of the real world. The same games that were played in the ancient cities of lost empires are now played by artificial intelligence. Checkers, chess, go, backgammon, poker, Scrabble—all conquered by superhuman computers. But in perhaps no other game has the presence of machines been as overwhelming as it has been with backgammon.
The digital quest began decades ago. At the world championship in Monte Carlo in 1979, a rudimentary program built by a Carnegie Mellon computer scientist took physical form as a robot called “Gammonoid.” After some lucky rolling, it defeated the human world champion—the first computer to do so at any game. Fellow players assembled in “an indignant and gesticulating mass immediately afterwards and hurled insults at the machine,” the newswires reported.
The pursuit continued in a windowless office at IBM Research, outside New York City, where an erstwhile theoretical physicist unleashed neural networks on the game—those modern artificial-intelligence systems structured to resemble neurons in a human brain. This system so efficiently mastered the game that its ideas were used by NASA for the space shuttle.
This research begat commercial programs with names such as JellyFish, Snowie, and eXtreme Gammon, or XG. The XG software, available for $60, is the demi-god of modern backgammon—consulted, idolized, feared. The software also lives on my phone, where it obsesses, torments, and judges me. (While reporting a chapter of my forthcoming book, Seven Games: A Human History, I developed a severe backgammon addiction.)
My conversation with Mochy turned melancholy as he remembered a time gone by, a time undone by the creep of technology.
“The big downside is it kills action,” Mochy said about the software. “Before bots, everybody believed that they were the best player. ‘I know how to play. I’m the best player. Let’s play for a hundred dollars a point.’ But now the computer tells you, ‘Hey, you suck. You’re burning your money.’” Indeed, in the hotel bar, there were no money games, no $100 bills being passed across the board. There were a dozen players staring at their laptops in silence.
Backgammon pros once made their nut like any other gamester, feeding off rich whales and hopeful fish. Now they make their living teaching those same marks, disseminating the computer’s lessons, more adjunct professor than hustler. Backgammon was once a loamy garden where human delusion flourished. Now it is a sterile laboratory of arithmetic.
These days, it’s not even enough to win a game of backgammon; you have to impress the computer, too. Modern elite tournaments, such as the Ultimate Backgammon Championship, give points not only for winning but for being adjudged by the machine as having played better. However the dice may have rolled, the masters’ plays are run through the algorithmic ringer and assigned a stark number.
But the old, muddy game is still available for us human dabblers. At a recent dinner party, a long-awaited reunion of pandemic-scattered journalist friends, one attendee smuggled in a small travel backgammon set. He and I stole away from the meal, set up the checkers, and spent the evening playing well and playing badly, winning money and losing money, and—heaven forfend—having fun. We didn’t dare ask the machine how we’d done.
Oliver Roeder is a New York–based journalist and author focused on game theory and artificial intelligence. His book Seven Games: A Human History, a group biography of seven enduring and beloved games, will be published by W. W. Norton on January 25
