How To Catch A Chess Cheater: Ken Regan Finds Moves Out Of Mind
“What’s God’s rating?” asks Ken Regan, as he leads me down the stairs to the finished basement of his house in Buffalo, New York. Outside, the cold intrudes on an overcast morning in late May 2013; but in here sunlight pierces through two windows near the ceiling, as if this point on earth enjoys a direct link to heaven. On a nearby shelf, old board game boxes of Monopoly, Parcheesi, and Life pile up, with other nostalgia from the childhoods of Regan’s two teenage children. Next to the shelf sits a table that supports a lone laptop logged into the Department of Computer Science and Engineering’s Unix system at the University at Buffalo, where Regan works as a tenured associate professor. The laptop controls four invocations of his anti-chess-cheating software, which at this moment monitor games from the World Rapid Championships, using an open-source chess engine called Stockfish, one of the strongest chess-playing entities on the planet. Around the clock, in real-time, this laptop helps compile essential reference data for Regan’s algorithms. Regan and I are on our way to his office, where he plans to explain the details of his work. But the laptop has been acting up. First he must check its progress, and Regan taps a few keys. What he’s staring at on the screen reminds him to rephrase his question, but this time he doesn’t wait for my answer. “What’s the rating of perfect play?” he asks. “My model says it’s 3600. These engines at 3200, 3300, they’re knocking at that door.” In Regan’s code, the chess engine needs to play the role of an omniscient artificial intelligence that objectively evaluates and ranks, better than any human, every legal move in a given chess position. In other words, the engine needs to play chess just about as well as God.