FYI.

This story is over 5 years old.

Tech

Big Data's Strategy for Winning the World Chess Championship

When Viswanathan Anand goes aggro, he wins big.
Image: Fisher Photostudio/Shutterstock

Just as Steve Kerr, Jason Kidd, and Derek Fisher have left behind playing basketball to coach it, computers no longer have anything left to gain from playing against people. But they are still being used to outline strategies.

Take, for instance, this year's World Chess Championship. Magnus Carlsen is a handsome, young, reigning World Chess Champion. He rose quickly through the ranks and he might just be the best chess player of all time. But the man Carlsen displaced last year to take the throne can't be counted out just yet.

Advertisement

As he proved with an aggressive and stunning victory in Game 3, which evened up the championship series, Viswanathan Anand is capable of beating Carlsen. A team of computer scientists from Michigan State University have been working out the best way of doing so.

Magnus Carlsen. Image: Frans Peeters/Flickr

In a description of Game 3, the Hindustan Times noted that Carlsen is "known as 'boa constrictor', for wearing down his opponents by slowly putting pressure on them and extracting mistakes." Carlsen's a grinder, who wears you out. As the game goes on—the graph below shows games against Anand, but it's true for Carlsen's career overall—the Norwegian becomes more likely to win.

But Anand controls the meaty part of the graph, where most games typically end.

Image: Randy Olson

I had a chance to chat with Randy Olson, a fourth-year computer science graduate research assistant who has worked on visualizing data on chess games. I asked him why it looked like Carlsen had the advantage in very short games.

"There are only six games in that range, only one of which Carlsen won," Olson said. "It's likely the players reaching an agreement early on in the game that it's not worth playing out and calling a draw."

Olson outlined what Anand can do to win, conclusions he came to after looking at the 77 tournament games that Anand and Carlsen have played against each other, and came to see that game length is potentially the most telling statistic.

"Anand has a considerable win advantage over Carlsen if he can finish the games within the 60-120 ply range, but his chances of victory will slip away if he allows Carlsen to drag the games out too long," Olson said. "Anand is known as a quick-minded, aggressive master of chess who finishes his opponents off quickly, which we also see related in his overall win record. For games that last between 60-120 ply (ply meaning one player's move), Anand wins about 50 percent of his games. Outside of that range, his win record quickly drops to 30 percent and below."

Advertisement

As for Game 3, it seems to confirm what Olson was saying.

"I've watched the games but haven't analyzed them closely yet," Olson said. "What I do know is that Anand played defensively in Game 2 and lost because of it. Anand played aggressively in Game 3 and handily beat Carlsen. With these observations backed by their historical record against each other, it's quite clear that aggressive play is key to Anand's victory. Hopefully he takes the lessons from his Game 3 win to heart so he can win back his title this year."

But, Olson told me, Carlsen is known as an all-around solid chess player, who refuses to specialize in any part of the game. Hence he wins about 40 percent of the games he plays, no matter how long. He also rarely loses, grasping draws from the jaws of defeat.

Of course, that's increasingly true of chessmasters overall, and is certainly true of Anand and Carlsen's history of playing each other—the most frequent result is a draw. It's worth noting, however, that Anand has the advantage historically.

Image: Randy Olson

At the moment, though, they're tied at a game and a half each, with the World Championship on the line.