Stockfish as Artist

Last month Wesley So won the US Championship in St. Louis and apart from his first prize, he also received the brilliancy prize for his game against junior world champion Jeffery Xiong. In that game, So had first sacrificed a pawn, then a knight and finally his queen. What you saw was beautiful and at least as beautiful was what you didn’t see, at least not at first sight: variations that stayed under the surface.

Call it an atavistic instinct, but we tend to value beauty that arises from improvisation at the board higher then when it is created at the player’s study, especially when it is done with the help of computers. Even Garry Kasparov, the first great cyborg of the chess world, was most proud when he had won a beautiful game in which his fabulous opening preparation had played no significant role.

I wondered how big the role of computer-aided preparation had been in the making of So’s brilliancy. I played it over while running the program Stockfish, which is undoubtedly one of the programs used by So with his opening preparations. Stockfish and Komodo are considered the two strongest engines available, but an advantage of Stockfish is that it is free, while the later versions of Komodo are not.

Last year I saw an interview in a Dutch newspaper with a Serbian immigrant who was living on one of the Dutch islands in the Waddenzee, the sea in the north of the country. The Serbian had come to the conclusion that the mindset of the Dutch could be characterized by three words: sale, free and discount. That didn’t mean that he disliked the people of his new country. He also said that the Dutch were reliable and that when you needed help, they stood by. The stinginess of the Dutch is proverbial in the English language and as a true Dutchman I use the free Stockfish.

It turned out that starting with the opening novelty on move 16, all moves played by So were the first choice of my Stockfish. Had he been a lesser player, he might have been searched after the game.

On his 26th move, when he already had a winning position, So spent twenty minutes and then played the very strong 26...e5. This was also the only moment when Stockfish needed more time than usual (at least ten seconds) before it came to the same conclusion as So, namely that 26...e5 is best.I wouldn’t draw the conclusion that So had seen the whole game already on the screen during his preparation. You can’t prepare for everything and Xiong had not played the best defense at every move. But a conclusion that can be drawn is that beauty in modern top chess often stems from the cooperation of man and machine and especially that Stockfish, and undoubtedly other engines as well, is capable of producing a truly beaufiful game on its own.

Click here to view Xiong-So, 2017 US Championship