Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A test position for programs(was Correspondence Chess and computers)

Author: blass uri

Date: 19:00:47 08/16/98

Go up one level in this thread



On August 16, 1998 at 11:58:09, blass uri wrote:

>
>On August 16, 1998 at 07:29:03, blass uri wrote:
>
>>
>>On August 16, 1998 at 07:02:54, Kai Lübke wrote:
>>
>>>On August 15, 1998 at 18:48:12, Robert Hyatt wrote:
>>>
>>>
>>>>Computers are really not a threat of any kind at correspondence chess.  We had
>>>
>>>[snip]
>>>
>>>>lower levels, computer impact could be felt.  But you aren't going to find a
>>>>2400 correspondence computer for a long while...
>>>
>>>I totally agree. I have a 2080 correspondence chess rating and once lost a game
>>>in 13 moves. In the post mortem I found that almost all of my programs would
>>>have lost just as fast (Junior, Fritz, MChess etc., given the 10-30 min/move
>>>that I typically use in my games).
>>
>>In a correspondence game you can give programs more time then 10-30 min/move
>>
>>The question if they would have lost in 5 hours per move
>>
>>Uri
>I think that the losing mistake is 7.exf7+
>7.Be3 is the move in the opening book of fritz5
>Junior5 after some hours play 7.exf7+ with evaluation 0.89
>it see the line 7.exf7+ Kxf7 8.Be3 Bb4+ 9.Kf2 Re8 10.Ne2 but does not see
>Rxe3
>It is a good test position for computer programs
>I am sure Junior will find not to play exf7+
>The question is how much time does it need (20 hours,200 hours or maybe 2000
>hours)
>I am not going to try more than 50 hours on my pentium200MMX
after more then 13 hours Junior start to smell it is not very good.
The evaluation was only 0.62 pawns advantage for white
with the line 7.exf7+ Kxf7 8.Be3 Bb4+ 9.Kf2 Re8 10.Nc3 Rxe3

when I tried to get out of the internet my computer had a problem and I am not
going to start all the analysis again.

In a correspondence game you can give the computer in average 24 hours per move
and if you see the evaluation goes down it is reasonable to give it more time.
This is the reason I think there is a good chance computers will not fall to
this trap.

Uri
>
>Uri
>>
>>
>>>
>>>Here's the game (I'm White, my opponent has a 2200 rating):
>>>
>>>1. e4 d5 2. exd5 Nf6 3. d4 Bg4 4. f3 Bf5
>>>5. c4 e6 6. dxe6 Nc6 (all theory so far)
>>>7. exf7+ (seconded by Junior, Nimzo, Chessmaster, Rebel, MChess)
>>>7...Kxf7
>>>8. Be3 (seconded by all programs)
>>>8...Bb4+
>>>9. Kf2 (seconded by all except Hiarcs & Nimzo)
>>>9... Re8
>>>10. Ne2 (seconded by all except Rebel)
>>>10... Rxe3!!
>>>11. Kxe3 (seconded by all, trivial)
>>>11...Qe7+
>>>12. Kf2 (seconded by all, trivial)
>>>12...Re8
>>>13. Nbc3 (fairly trivial)
>>>
>>>Now all programs see black up by +2,5 or more and the game is clearly lost (in
>>>the actual game, I carried on until move 23, but it was hopeless...).
>>>
>>>At least Junior, CM and MChess would have played the game exactly as I did and
>>>thus would have lost.
>>>
>>>---
>>>Shep



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.