Author: Drexel,Michael
Date: 12:37:03 02/06/04
Go up one level in this thread
On February 06, 2004 at 15:03:49, Uri Blass wrote:
>On February 06, 2004 at 14:29:30, Thorsten Czub wrote:
>
>>On February 06, 2004 at 12:19:09, Uri Blass wrote:
>>
>>>Here is the line of the game that you posted
>>>
>>>14...Qc7 {-0.30/18} 15. Bxh6 gxh6 {-0.40/15} 16. Qxh6 Rfd8 {-0.54/16}
>>>17.Rd3 Nh7 {0.00/17} 18.f4 Bf8 {0.01/18} 19.Rg3+ Kh8 {0.01/18} 20. Qh5 Rab8
>>>{0.35/17} 21.f5 exf5 {0.85/16} 22.Rf1 Rd7 {0.84/19} 23.Rxf5 f6 {0.84/18
>>>
>>>You can see that shredder evaluated the position as 0.54 pawns for itself at
>>>move 16 and it is more than 0.30 pawns for itself at move 14.
>>
>>right. right.
>>but it is 18 searches deep !
>>
>>in move 20 (that is 12 searches away from Qc7 it feels negative, and in move
>>21 even morenegative.
>>
>>how can the program oversee this, when it sees the things different within
>>12 plies later.
>>
>>>I believe that not pruning could not help shredder to avoid Qc7 and I am also
>>>not sure if it was the losing error(I suggested 17...Ne8 instead of 17...Nh7).
>>>
>>>Uri
>>
>>shredder itself changes the score from + into minus within a few moves.
>>
>>in between there is the fritz-like 0,00.
>>
>>if shredder itself sees now the weakness, why wasn't it able to see the obvious
>>move earlier ? it pruned it away-
>>if the move was wrong, why the score dropping ??
>
>I believe that the following claims are correct:
>
>1)The move Bxh6 was wrong but shredder blundered later(I suggested 17...Ne8
>instead of 17...Nh7)
>
No,Bxh6 is not wrong. White can at least draw.
CS Tal would never accept a draw and lose but that's another story of course.
>2)Shredder searched 15.Bxh6 to reduced depth but this is not the reason for
>playing 14...Qc7(otherwise shredder could see a drop in the score immediatly
>after 15.Bxh6 and it did not happen)
>
>Uri
There is probably not a single program that can find the _only_ defence 17...Ne8
18.Bd3 Bd8 since white can play 19.Nd5.
Michael
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.