Author: pete
Date: 12:05:30 02/03/00
Go up one level in this thread
On February 01, 2000 at 11:43:54, Côme wrote: > >In this position : >[D]r2qkb1r/pp1n1ppp/4n3/1B6/Q7/2N2P2/PP3P1P/R1B1K2R w KQkq - 0 1 > >After 8 min on my K6 II 350 Crafty 17.07 don't find Bg5!! > >Junior and Fritz find Bg5!! very quickly. > >Can you test other programs ? > >Best Regards >Alexandre Côme An interesting position ! Bringer gives : 0:00:31.2 ( 8/22) 5658816 0.54 c3-d5 f8-c5 c1-e3 c5xe3 f2xe3 h7-h6 e1-c1 e6-c5 a4-b4 0:01:40.1 ( 9/24) 19149007 0.70 c3-d5 f8-c5 c1-e3 c5xe3 f2xe3 e8-g8 b5xd7 e6-c5 a4-b5 0:02:22.1 (10/24) 27278238 0.50 c3-d5 f8-c5 0:03:31.8 (10/27) 40323858 0.50 c3-d5 f8-c5 0:04:04.7 (10/27) 46908439 0.50 c3-d5 0:10:59.1 (11/28) 131438353 0.56 c3-d5 f8-c5 c1-e3 c5xe3 f2xe3 e8-g8 b5xd7 e6-c5 a4-b5 c5xd7 b5xb7 a8-b8 0:17:56.9 (12/29) 215610054 0.36 c3-d5 f8-c5 0:27:37.6 (12/31) 329175856 0.29 c3-d5 f8-c5 c1-e3 c5xe3 f2xe3 e8-g8 b5xd7 e6-c5 a4-d4 c5xd7 h1-g1 f7-f6 1:29:01.3 (13/33) 1060919861 0.14 c3-d5 f8-c5 c1-e3 c5xe3 f2xe3 e8-g8 b5xd7 e6-c5 a4-d4 c5xd7 1:47:10.9 (13/33) 1290242983 0.15 c1-g5 2:10:40.6 (13/33) 1599978391 0.65 c1-g5 3:19:56.6 (13/34) 2524858403 1.04 c1-g5 d8xg5 b5xd7 e8-e7 a1-d1 e6-c5 7:18:03.2 (14/36) 5666550665 1.09 c1-g5 f8-e7 g5xe7 d8xe7 At first sight this looks rather bad but I think it is easy to misjudge the result. Let's compare : Shredder 4 solves this in less than 1 second evaluating +1.70 : impressive ! But when I looked at the main lines I got a more complicated impression . Shredder evaluates Bg5 with 1.70 right from the start so it's no question of search but of evaluation ; both Shredder and Bringer get their first fail high on Bg5 at ply 13 so this seems to be the depth you have to reach . I think all programs see the main line with Bg5 right from the start so it is simply : some don't like it as much as others . Interesting are the differences of tree size : Shredder only needs 133.373.275 positions until the fail-high at ply 13 where Bringer needs an amazing 1.060.919.861 but this seems to be a direct consequence of Shredder evaluating it so high early on . It seems Bringer also thinks sensibly here ; I have the impression the wrongest judgement is about the doubled pawns at f2 and f3 ( Bringer _hates_ doubled pawns ) ; Blacks main "threat" is Bc5 so this leads to the natural Be3 ( Amyan and Jusupov ) or to : Nd5 ; it might have said to itself : hey this Bc5 would look impressive and has to be stopped ; but I have the nice Nd5 centralizing knight and after Bc5 I have Be3 where I can UNDOUBLE MY PAWNS !!! :-) I personally have the impression that Shredder is the program that most often "guesses" right ; if you take the +1.70 this is not so far away of the 2.46 it has when it finishes ply 15 and the program that evaluates positions like this one ( which I think is not such a tactical one at all ) right most often are simply the better player . So I think this one is not about : " Why can't dumb crafty find this tactical shot fast ? " but more "Why does it evaluate Bg5 so low ?" . IMHO. pete
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.