Author: Steffen Jakob
Date: 02:21:11 03/22/01
Go up one level in this thread
Hi Eduard! On March 22, 2001 at 02:32:58, Eduard Nemeth wrote: >Why any programs would play Kg3? This is (i think) a very good question for >all program-authors! Here I have no answer. I think, that many programs in this >position would mates and seen "first" 1.Kg3 Kf1 2.Kxf1, oh my bad english sorry, >in german language I need say: "es droht weiterhin zunächst Matt plus Material- >gewinn". Hi Eduard, here is a possible explanation why programs may play Kg3: After 1... Kxf3 white can play 2.e7 and gets two connected passed pawns on the 7th rank. So we have those connected passers vs a single rook. The rook usually cannot stop those passers and therefore a huge positional bonus might be assigned for the pawns. The bonus is maybe big enough that the comp sees an advantage for white. On the other hand the pawns e6 + f7 from the root position dont seem to be as strong as after e7 from the view of the engine. The huge bonus doesnt get assigned here. The consequence is that black tries to delay the move e7 as long as possible. This can be done with Kg3 because this establishes a mate threat. White has to react immediately. As long as the white king moves on the squares e1,f1,g1,h1 black can keep the mate threat. So white goes to d1: 1... Kg3 2.Kf1 Kxf3 3.Ke1 Ke3 4.Kd1. Note that already 6 plies have been played. Now the way to delay e7 is to give some checks with the rook: 4... Rd8+ 5.Kc2 Rc8+ 6.Kb3 Rb8+ 7.Kc4 Rc8+ 8.Kd5 Rd8+ 9.Ke5. Black is out of check now. He can make one more move like 9... Kf8 and then *finally* white is able to play 10.e7. You can see that black can delay the move e7 for 18 plies. BTW: I can tell you all this because it's the reason why Hossa would play Kg3, too :-( It's a good example where humans see very fast what happens but computers need to search deep if they dont have any sophisticated knowledge to get the same result. Greetings, Steffen.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.