Computers Have Turned Chess Into Poker

• Read original article here

It was as if a bottom seed had knocked out the top team in March Madness: At the Sinquefield Cup chess tournament in St. Louis earlier this month, an upstart American teenager named Hans Niemann snapped the 53-game unbeaten streak of world champion Magnus Carlsen, perhaps the game’s best player of all time. But the real uproar came the following day, when Carlsen posted a cryptic tweet announcing his withdrawal that included a meme video stating, “If I speak I am in big trouble.” The king appeared to have leveled an unspoken accusation of cheating—and the chess world, in turn, exploded.

Some of the biggest names in chess launched attacks on Niemann in the subsequent days, while others rushed to defend him. Niemann, by his own recent admission, has cheated at online chess at least twice before, when he was 12 and 16 years old. These past offenses, combined with what some believed was lackluster chess analysis in his postgame interviews, have heightened suspicions of foul play. On Twitch and Twitter, players and fans theorized that Niemann might have been receiving secret messages encoded in the vibrations of electronic shoe inserts or remote-controlled anal beads. No concrete evidence of cheating has emerged, and the 19-year-old grandmaster vehemently denied accusations of misconduct in St. Louis, vowing to an interviewer that he has never cheated in an over-the-board game and has learned from prior mistakes.

Whatever really happened here, everyone agrees that for Niemann, or anyone else, to cheat at chess in 2022 would be conceptually simple. In the past 15 years, widely available AI software packages, known as “chess engines,” have been developed to the point where they can easily demolish the world’s best chess players—so all a cheater has to do to win is figure out a way to channel a machine’s advice. That’s not the only way that computers have recently reshaped the landscape of a 1,500-year-old sport. Human players, whether novices or grandmasters, now find inspiration in the outputs of these engines, and they train themselves by memorizing computer moves. In other words, chess engines have redefined creativity in chess, leading to a situation where the game’s top players can no longer get away with simply playing the strongest chess they can, but must also engage in subterfuge, misdirection, and other psychological techniques. In that sense, the recent cheating scandal only shows the darker side of what chess slowly has become.

The computer takeover of chess occurred, at least in the popular imagination, 25 years ago, when the IBM supercomputer Deep Blue defeated world champion Garry Kasparov. Newsrooms at the time declared the match a “Greek tragedy,” in which a silicon “hand of God” had squashed humanity. Yet 1997, despite its cultural resonance, was not really an inflection point for chess. Deep Blue, a nearly 3,000-pound, one-of-a-kind supercomputer, could hardly change the game by itself. Its genius seemed reliant on then-unthinkable processing power and the grandmasters who had advised in its creation, to the point where Kasparov, after losing, could accuse IBM of having cheated by supplying the machine with human assistance—a dynamic that today’s accusations of foul play have reversed.

By the mid-2000s, though, upgrades in chess-engine software and commercial hardware made overpowering algorithms more accessible; in 2006, an engine running on a standard desktop computer defeated then–world champion Vladimir Kramnik. Players had already been using engines to evaluate individual tactics. But Kramnik’s loss kicked off the first era of computer-chess superiority, in which even chess elites would rely on software to help evaluate their strategies, Matthew Sadler, a grandmaster who has written multiple books on chess engines, told me.

As engines became widespread, the game shifted. Elite chess has always involved rote learning, but “the amount of stuff you need to prepare, the amount of stuff you need to remember, has just exploded,” Sadler said. Engines can calculate positions far more accurately and rapidly than humans, so there’s more material to be studied than ever before. What once seemed magical became calculable; where one could rely on intuition came to require rigorous memorization and training with a machine. Chess, once poetic and philosophical, was acquiring elements of a spelling bee: a battle of preparation, a measure of hours invested. “The thrill used to be about using your mind creatively and working out unique and difficult solutions to strategical problems,” the grandmaster Wesley So, the fifth-ranked player in the world, told me via email. “Not testing each other to see who has the better memorization plan.”

Once computers were reliably beating grandmasters, cheating-by-computer became a serious threat, Emil Sutovsky, the director general of the International Chess Federation, told me. The federation implemented its first anti-cheating measures in 2008.

That’s not to say chess was “solved” (in the sense that a perfect set of moves has been devised for every position), as checkers is; there are more possible chess games than atoms in the observable universe. Sadler believes “human frailty”—that we aren’t machines—kept chess exciting: People would still forget their pregame analysis, fail to predict their opponent’s strategy, and end up in positions they hadn’t prepared for. The computers in this first era of chess engines were very good on defense, but they still had weaknesses, Sutovsky said, such as struggling to determine the value of sacrificing a piece for long-term benefit.

But that all changed on December 5, 2017, when AI researchers at Alphabet announced a new algorithm, AlphaZero, which had surpassed the best existing chess engine simply by playing games against itself—over the course of just four hours. AlphaZero used a neural network, an approach to artificial intelligence that mimics the human brain and, in a sense, allows a machine to learn. Other chess engines quickly incorporated the new technology, heralding the modern era of total computer domination.

In the first era, humans would devise attack strategies, then refine them in games against machines. AlphaZero crushed these earlier engines by “playing extremely aggressive chess,” Sadler said. The modern, neural-net engines are eager to sacrifice; and they exhibit a strong grasp of openings, positional structure, and long-term strategy. “It started to look a bit more [like] a human way to play,” Sutovsky told me, in describing this transformation. Or even superhuman, he said: The new chess engines seemed to have insight into “the tactical skirmish, but also could plan for some long-lasting compensation for material loss.”

To understand just how superior machines have become, consider chess’s “Elo” rating system, which compares players’ relative strength and was devised by a Hungarian American physicist. The highest-ever human rating, achieved by Carlsen twice over the past decade, was 2882. DeepBlue’s Elo rating was 2853. A chess engine called Rybka was the first to reach 3000 points, in 2007; and today’s most powerful program, Stockfish, currently has more than 3500 Elo points by conservative estimates. That means Stockfish has about a 98 percent probability of beating Carlsen in a match and, per one estimate, a 2 percent chance of drawing. (An outright victory for Carlsen would be almost impossible.)

Where chess engines once evaluated human strategies, the new, upgraded versions—which are freely available online, including Stockfish—now generate surprising ideas and define the ideal way to play the game, to the point that human performance is measured in terms of “centipawn” (hundredths of a pawn) loss relative to what a computer would play. While training, a player might ask the software to suggest a set of moves to fit a given situation, and then decide to use the computer’s sixth-ranked option, rather than the first, in the hopes of confusing a human competitor who trained with similar algorithms. Or they might choose a move tailored to the weaknesses of a particular opponent. Many chess experts have adopted the new engines’ more aggressive style, and the algorithms have popularized numerous tactics that human players had previously underestimated.

The advent of neural-net engines thrills many chess players and coaches, including Sutovsky and Sadler. Carlsen said he was “inspired” the first time he saw AlphaZero play. Engines have made it easier for amateurs to improve, while unlocking new dimensions of the game for experts. In this view, chess engines have not eliminated creativity but instead redefined what it means to be creative.

Yet if computers set the gold standard of play, and top players can only try to mimic them, then it’s not clear what, exactly, humans are creating. “Due to the predominance of engine use today,” the grandmaster So explained, “we are being encouraged to halt all creative thought and play like mechanical bots. It’s so boring. So beneath us.” And if elite players stand no chance against machines, instead settling for outsmarting their human opponents by playing subtle, unexpected, or suboptimal moves that weaponize “human frailty,” then modern-era chess looks more and more like a game of psychological warfare: not so much a spelling bee as a round of poker.

In that context, cheating scandals may be nothing less than a natural step in chess’s evolution. Poker, after all, has been rocked by allegations of foul play for years, including cases where players are accused of getting help from artificial intelligence. When the highest form of creativity is outfoxing your opponent—as has always been true of poker—breaking rules seems only natural.