American Go E-Journal

Michael Redmond 9P on “Calculated Errors”

Tuesday June 24, 2014

“The Chess Life article (Your Move/Readers Write: ‘Catching Chess Cheaters’ 6/23 EJ) says that ‘One interesting statistic is that players make 60 percent to 90 percent more errors when half a pawn ahead or behind compared to when the game is even,’” writes Michael Redmond 9P.

“How would you compare half a pawn in chess to a point advantage in go? I don’t know how big an advantage that is for chess masters, but I think that Regan’s observation that the players’ assessment of a game position — and the assumed emotional value — is affecting their ability to think is also true of go players, but to a lesser extent, depending on how big a half pawn is.”

“The article seems to imply that while the player at a disadvantage might have reason to play a high-risk/high-reward move, the winning player must try to play the correct move always. He uses this reasoning to conclude that the players are actually making errors. I suppose that chess, being a race to kill, does not allow for calculated mistakes, but this seems to be less true of go, and could indicate a difference in the endgame stage of the two games.”

“In go, there can be calculated ‘errors’ by the player with an advantage. As a go game nears its end, the leading player can often calculate a win without playing the optimum moves. My opinion is that top go players will sometimes choose technically incorrect moves when 2.5 points ahead, a calculated choice to simplify the game. Such calculated ’mistakes’ by the winning player are usually minor, and two to three mistakes can add up to a one point loss in actual play when compared to the correct endgame sequence. Anything more than that is probably a ‘real’ mistake.”
photo: Redmond at the 2010 WAGC; photo by John Pinkerton