Author: Dann Corbit
Date: 17:14:01 12/09/99
Go up one level in this thread
On December 09, 1999 at 18:45:42, Paulo Soares wrote:
>On December 09, 1999 at 14:14:47, Dann Corbit wrote:
>
>>On December 08, 1999 at 09:34:04, Côme wrote:
>>
>>>Hello everybody,
>>>In this position:
>>>rn1qr1k1/1p2bppp/p3p3/3pP3/P2P1B2/2RB1Q1P/1P3PP1/R5K1 w - -
>>>Does your program find the winning move Bxh7!!
>>
>>Here is computer analysis after 10 full hours on a PIII 500 MHz. I think
>>probably it is NULL MOVE problem, otherwise the correct move would be seen by
>>now.
>>
>>rn1qr1k1/1p2bppp/p3p3/3pP3/P2P1B2/2RB1Q1P/1P3PP1/R5K1 w - - acd 16; acn
>>545972321; acs 36001; bm d3h7; ce 27; id "Complete CCovax A.150 [par=2]"; pv Be3
>>Nc6 Rac1 Bb4 R3c2 Qh4 Qd1 Ba5 Be2 Nb4 g3 Qe7 Rc5 Bb6 R5c3 a5;
>
>I think that are some lines that programs can't see even at
>depth 20.
Absolutely. Any move where the "payoff" is 10 full moves ahead will take 20
plies to see. And if the payoff is farther, the computer may not make the right
choice until the move is seen. Often, however, there is a simple reason why a
choice was not seen. More often than not when the computer analysis is deep
enough to see the benefit drawn in the human analysis, one of the following
problems exists:
{calling the position with a clear benefit the final board position:}
0. The final board position requires an intermediate positional move.
1. The final board position requires a deep sacrifice.
2. There is a zugzwang position between the final board position and the
current position.
3. The stated move is wrong {happens more often than you might think}.
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.