Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: WM test bugs

Author: Vincent Diepeveen

Date: 10:30:34 06/28/02

Go up one level in this thread


On June 27, 2002 at 13:36:19, Peter Berger wrote:

>On June 27, 2002 at 04:04:07, Vincent Diepeveen wrote:
>
>>Hello,
>>
>>Several positions in WM test the solutions given aren't the
>>objective best moves.
>>
>>For example i get impression that at position 16:
>>
>>
>>r1k4r/1b1q1pp1/p6p/3n4/1B3Pn1/1P1B4/P1PQ3P/2KRR3 w - - bm Ba5; id "CSS WM-Test
>>16) Lb4-a5 (K)";
>>
>>Undoubtfully Ba5 is a good move, but completely material
>>winning is c2-c4 here. You get directly a rook at the 7th rank,
>>swallow f7 + g7, and black is completely immobilized the rest of
>>the game. Soon also Rd1-e1-e7 happens and it's a mate in XXX soon
>>then.
>>
>>I can imagine why a human plays Ba5 here and doesn't take a risk (note
>>the chessprograms want this all initially too), but let's look objective
>>to the position. Not finding c4 soon means a program is a tactical joke.
>>
>>As a human i wouldn't find c4 so soon either. I would go for Ba5, because
>>i know how much i suck in tactics.
>>
>>Best regards,
>>Vincent
>
>This is what I get with Crafty 17.13:
>
>14->  28:00   0.00   1. c4 Nxb4 2. Qxb4 Nf6 3. Re7 Qc6 4.
>                                    Bf5+ Kb8 5. Rd6 Qh1+ 6. Rd1 Qc6
>               15    36:53   0.00   1. c4 Nxb4 2. Qxb4 Nf6 3. Re7 Qc6 4.
>                                    Bf5+ Kb8 5. Rd6 Qh1+ 6. Rd1 Qc6
>               15   125:37   0.20   1. Ba5 Ngf6 2. c4 Nc7 3. Qb4 Re8 4.
>                                    Rxe8+ Nfxe8 5. Qc5 g6 6. Bxg6 fxg6
>                                    7. Rxd7 Kxd7 8. Qd4+ Kc8 9. Kd2 Bf3
>               15-> 125:37   0.20   1. Ba5 Ngf6 2. c4 Nc7 3. Qb4 Re8 4.
>                                    Rxe8+ Nfxe8 5. Qc5 g6 6. Bxg6 fxg6
>                                    7. Rxd7 Kxd7 8. Qd4+ Kc8 9. Kd2 Bf3
>
>Crafty 17.13 gets 1. c4 very fast with negative score and then drawscore soon.
>It needs a long time to find 1. Ba5 with a slight positive score. I expect it to
>fail-high on Ba5 in ply 16.
>
>If 1.c4 is completely material winning it must be very deep and tough for
>computers. If you let Diep think for a long time, will it keep with 1. c4 or
>will it change its mind to 1. Ba5 eventually ?
>
>Regards,
>Peter

DIEP wasn't on pc beginners hardware when running the position.





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.