Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Internet access to large set of 6-man TBs is now available

Author: Vasik Rajlich

Date: 02:39:28 05/14/04

Go up one level in this thread


On May 13, 2004 at 19:21:05, Omid David Tabibi wrote:

>On May 13, 2004 at 08:18:17, Tony Werten wrote:
>
>>On May 12, 2004 at 22:03:05, Omid David Tabibi wrote:
>>
>>>On May 12, 2004 at 18:03:06, Jorge Pichard wrote:
>>>
>>>>On May 12, 2004 at 17:23:40, Robert Pawlak wrote:
>>>>
>>>>>See my site for more info:
>>>>>
>>>>>www.chessassistance.com
>>>>>
>>>>>Bob
>>>>
>>>>
>>>>As long as there are positions like this one which requires 10 men TBs, Human
>>>>will always be better :-)
>>>>
>>>>Most programs don't even have a clue of what the correct plan is, and for the
>>>>first 30 moves it analyze this position as even, but any IM or Above will
>>>>realize that White is winning from the starting position.
>>>>
>>>>[Event "Shootout (Shredder8, 120'/40+60'/20+30')"]
>>>>[Site "?"]
>>>>[Date "????.??.??"]
>>>>[Round "?"]
>>>>[White "New game"]
>>>>[Black "?"]
>>>>[Result "1-0"]
>>>>[SetUp "1"]
>>>>
>>>>[D]8/8/3k4/3p4/3PpN2/4P1B1/3qBK2/8 w - - 0 1
>>>
>>>You don't have to be an IM to see that white is easily winning here. Three
>>>pieces, especially when having two bishops amongst them, are a deadly
>>>combination against which a lone queen stands no chance.
>>>
>>>Falcon also evaluates the position quite realistically:
>>>
>>>Falcon 1.0.12.8 running on GenuineIntel 733MHz 256MB:
>>>depth     time    nodes   nps  score  variation
>>> 8/19     0.32      60k  189k   1.28  1.f4g6 d6e6 2.g3c7 d2a2 3.g6e5 a2c2
>>>                                      4.c7d8 e6f5 5.d8e7
>>> 9/26     0.59     123k  209k   1.28  1.f4g6 d6e6 2.g3c7 d2a2 3.g6e5 a2c2
>>>                                      4.c7d8 e6d6 5.d8f6 c2c7
>>>10/22     1.03     221k  214k   1.28  1.f4g6 d6e6 2.g3c7 d2a2 3.g6e5 a2b2
>>>                                      4.c7d8 b2c2 5.d8g5 e6f5 6.g5e7
>>>11/24     2.23     504k  225k   1.26  1.f4g6 d6e6 2.g3c7 d2a2 3.c7d8 a2d2
>>>                                      4.g6e5 e6f5 5.d8e7 f5e6 6.e7c5 e6f5
>>>11/24     2.66     605k  227k   1.27  1.f4d3++
>>>12/23     3.82     879k  230k   1.26  1.f4d3 d6e6 2.d3c5 e6f5 3.c5b7 f5e6
>>>                                      4.b7d6 d2c2 5.g3f4 c2b2 6.f4e5 b2c2
>>>                                      7.e5g3 c2b2
>>>12/25     5.41    1242k  229k   1.27  1.f4g6++
>>>13/34     8.60    2004k  232k   1.27  1.f4g6 d6e6 2.g3c7 d2a2 3.c7d8 a2d2
>>>                                      4.g6e5 d2c2 5.d8g5 c2d2 6.g5f4 e6f6
>>>                                      7.e5d7 f6f5 8.d7e5
>>>14/36    20.27    4736k  233k   1.27  1.f4g6 d6e6 2.g3c7 d2b2 3.g6e5 e6e7
>>>                                      4.c7a5 e7e6 5.a5d8 e6f5 6.d8c7
>>>15/33    41.85    9703k  231k   1.28  1.f4g6 d6e6 2.g3c7 d2b2 3.g6e5 e6e7
>>>                                      4.c7a5 e7e6 5.a5d8 e6f5 6.d8h4 f5e6
>>>                                      7.h4g5 b2d2 8.g5d8 d2c2
>>
>>You call that easily winning ?
>>
>>You have a bad_trade score a la Crafty ? 1.2 pawns ? That means your engine is
>>seeing it is easily winning with a tremendous 0.08 score.
>
>I have about 0.25 pawn bad trade score here. But the other 1 pawn advantage is
>from the evaluation.
>

I'm curious - what are you evaluating? Ie. what rules are triggering? (If not
the material imbalance.)

Rybka gives +1.00 from search, exactly as from pure eval. The scaled material
imbalance (bishop pair included, two knights would give a lower score) accounts
for all of it. By scaled I mean that the less material is on the board, the more
the existing material imbalance is weighed.

Vas

>
>>
>>Tony



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.