Chess in the 21st century is deceptive | So Good News
VWolfgang von Kempelen, an 18th-century inventor and writer, once claimed to have created a chess-playing robot called the Mechanical Turk. How was this possible almost two centuries before computers were invented? Easy. Fraudulently.
The Turkish chess master was a hidden box. At that time, the idea of an automaton that could play chess well was not possible. Even in 1997, when the IBM computer Deep Blue beat then-champion Garry Kasparov at chess, he suspected that one of Deep Blue’s moves during the game was actually human-made.
How times have changed.
Since Deep Blue, the best chess-playing objects have been computer programs, not people. It is very easy to cheat online on platforms like Chess.com where many people and the world’s best play. You simply have the chess program go to the side, input your opponent’s moves as your own, and then the computer moves against your opponent. You are guaranteed to win. Unless, of course, they cheat too.
In September, five-time world chess champion Magnus Carlsen refused to face 19-year-old grandmaster Hans Niemann at the prestigious Sinquefield Cup in St. Louis. Carlsen accused Niemann of cheating. Nieman denied this and admitted that he had cheated on the Internet only twice in his life, at the ages of 12 and 16. According to a Chess.com investigation published on October 4, Niemann may have cheated in more than 100 online games.
There was no hard evidence that Niemann cheated on Carlsen in the Sinquefield Cup match. No device, speaker, computer or cables found. How could it be possible? I remembered that incident Golden finger, where someone cheats at cards with the help of a partner with binoculars and an audio connection. You can put a chess computer in a small silicone-coated box about the size of a small salt shaker. This allows the chess computer to be kept on its person, showing which pieces to move where.
If Niemann had cheated, he could have hired some conspirator to feed Carlsen’s moves into the chess AI, and then communicate the proposed move through some device hidden somewhere within Niemann. Historically, others have cheated by signaling a vibrating device in Morse code from a companion watching the game, or using subtle hand signals from the crowd. This latest mess prompted increased security at chess tournaments, including a 30-minute delay on the video to prevent people from providing real-time help.
Chess.com has discovered that Hans Niemann’s results in many online games are questionable.
Many believe that Niemann somehow cheated. Why? Carlsen’s doubts are significant. Niemann was playing unreal, Carlsen said in a public statement. “During our game in the Sinquefield Cup, I had the impression that he wasn’t tense, or even completely focused on the game in critical positions, while playing in the dark, as I thought only a handful of players could do.”
If Niemann cheated, it might not be at every step. For the best chess players, the next best move often seems relatively obvious. The player must consult the chess AI on certain key points – no less than one or two – in order to defeat the opponent. This is called “selective cheating.” A shrewd fraudster may make cheap moves that they know are suboptimal to throw fraud detectors off the scent.
In a report, Chess.com compared Niemann’s moves to Stockfish, the best chess AI available today. It’s pretty good and will probably beat Carlsen or any other human player every time. He can look 40 steps ahead, which is far more than anyone else. This means that chess AI can make moves that humans never could—moves that seem like bad ideas in the short term to skilled chess players, but in the game work out much better than what human grandmasters can mentally simulate. Human chess players do not make such moves, and if they did, we might suspect them of cheating.
Chess.com found that Niemann’s results in many online games were highly questionable given his skill. They also looked at how long it took Niemann to perform more complex movements — if the movements were performed very quickly, it would be evidence that he was using AI. It’s similar to how “doping” is defined in sports: Drugged competitors perform at inhuman levels of speed and strength.
In September, chess enthusiast Yosha Iglesias analyzed Niemann’s games using a program that evaluates how close each move is to optimality. Iglesias’ findings supported Carlsen’s suspicions: Niemann’s movements were close to optimal. Most chess masters play around 70-75 percent of optimality, while Niemann plays close to 100 percent over the course of several games. Clever Hans may be getting a hint from outside his mind.
Where we used to have machines cheating at chess with humans, now we have humans cheating using machines.
Turk against mechanics.
Jim Davis is Professor of Cognitive Science at Carleton University. He is the host of the award-winning Mind of the Brain podcast. His latest book Becoming Who Your Dog Thinks You Are: The Science of Better You.
Lead image: Humboldt University Library / Wikimedia Commons