Women's World Championship, Game 2

Tan wins a "drawn" endgame.

The first game of the Women’s World Championship match between Ju Wenjun and her challenger (and predecessors) Tan Zhongyi was a peaceful draw, and game 2 seemed headed in that direction as well. The champion didn’t have too much trouble on the Black side of a Four Knights English with 4.g3, and while Tan enjoyed a mild initiative throughout the single-rook ending that resulted after 31.Rxa7 seemed a sure draw.

Instructively, it wasn’t. Drawn, yes, but as years of Magnus Carlsen victories from “drawn” endgames have shown us (or should have, if we were paying attention), we’re all far more capable of losing allegedly equal positions than we’d like to admit. One culprit in all of this is the engine, of course. Used properly it’s a great teaching tool, but one of its limitations at the moment is that it will evaluate a wide range of positions with the infamous “0.00” score. The problem is that not all zeroes are relevantly equivalent. Broadly speaking, there are three kinds or types of moves that can result in a 0.00 eval: (1) There are moves that force the draw or at least bring the draw closer and make it easier to achieve. (2) Moves that keep the game drawn but don’t make progress, generally making it somewhat harder to achieve. (3) Moves that don’t “officially” spoil the draw, but force one to play like a superhero to squeeze it out by the skin of one’s teeth. The naive engine-user keeps seeing all those triple-zeroes and thinks the loser blundered when the zeroes are finally gone, but much more often the problem was a series of Type 2 moves, often with a Type 3 move at the end, followed by the inevitable failure to find all the perfect moves at the last stage.

That is what happened to Ju. Had she played 31…b4! (a Type 1 move) the game would have ended speedily and peacefully. Instead, her 31…c5 was a Type 2 move, and her 33…Kf8 may have been another. 38…Rf5 was probably a Type 3 error, and the fatal error was 40…Ke8. Tan’s technique was perfect after that (and it was great before that, too, starting from at least move 32 after Ju’s initial error), and she won deservedly to take a 1.5-0.5 error.

Here’s the game, with my comments.