Tilbage til Skak

 




PC-SCHACH has quite sadly come with its last issue.
I will miss the many deep articles etc.

The last issue had a letter from F.Wiesenekker from Wels.
He had seen something strange with Genius2 and Genius3.

Kg2,Bf7,g3/Ke4,Be7,a3,h5


White easily draws with the king on h1 and giving the bishop for
blacks a-pawn. Black has the wrong bishop and cannot promote his pawn.

But genius plays the loosing g3-g4?? after some minutes.
After a longer search it finds a better move.

Hobby Computer Centrale gave this explanation:
Genius sometimes want its king to have more space and may eventually
give up a pawn for that......

I have examined the position.
g3-g4?? occurs in the search to a depth of 10 ply.
I have tried to set the black bishop on most squares and done a 10-ply search.
The result is, that genius plays g3-g4 if black can play Bg5.
Now why that?
Black can play his king to b2 and win the white bishop:
1.Bg8,Kd4 2.Kf3,Kc3 3.Kf4,Kb2 4.Kf5,a2 5.Bxa2,Kxa2
But white can defend with:
1.Bg8,Kd4 2.Kf3,Kc3 3.g4,h4 4.g5!,Bxg5 5.Kg2,Kb2 6.Kh1,a2 7.Be6,a2
8.Bxa2,Kxa2
The point is, that WITHOUT the g-pawn the program recognisis this as a
drawn KBP/K position with the wrong bishop.
But with 1.Bg8,Bg5! black keeps the g-pawn on the board.
Call it computerreasoning or artificial intelligence...

White plays 1.g4?? and pushes the loss of the bishop out of the horizon.
But with a longer search it sees it doesn't help or that white plays his king
to g6 and draws.