Author: martin fierz
Date: 09:15:07 10/15/03
Go up one level in this thread
On October 15, 2003 at 10:55:29, Robert Hyatt wrote: >On October 15, 2003 at 04:44:11, martin fierz wrote: > >>On October 15, 2003 at 01:23:33, Slater Wold wrote: >> >>>On October 15, 2003 at 00:53:32, Jim Monaghan wrote: >>> >>>>Here's an interesting position for engine eval testing: >>>> >>>>[D]3r1brk/pp3p1p/8/2pP1P2/2Pn4/6nP/PP4B1/R2Q1RK1 w - - 0 26 >>>> >>>> >>>>It would be interesting to see what different engines see and conclude about the >>>>above position. Will the score consistently drop over time? Could White have >>>>held? >>> >>>I believe at the very least, it's a draw. But I didn't look at it too terribly >>>deep. >> >>why do you believe that it's at the very least a draw? because all engines give >>a positive score on this position? i find it interesting that they all correct >>there score ever lower when searching deeper. would be interesting to see what >>happens when you continue... >> >>cheers >> martin > > >My thinking would be this: If black can't win _quickly_ the queen is going >to tell the tale over a longer game, and it can produce enough threats and >attack in so many directions, it will likely pick up a pawn here and there >if not more. I've not seen many (or any that I can actually recall) games >where one side had a significant material advantage and still lost a long >struggle. Queen sacs to force mate or whatever, yes. But here, given the >choice, I think I'd rather have white, cramped as it is. whenever i see a position where the side with less material has significant compensation (like here, and like the machines also think - after all this is a -4 in material for black, but the scores are around -2), i'd rather play out a few moves in the analysis than let the machine think for an hour in the initial position; to get a feeling of what is going on here. i did this against ruffian playing white, (of course, i would take back moves, and of course, i used ruffians suggestions in the pv too.., kind of ultra-advanced chess :-) - with 10s/move on a P4 2.4GHz. after 1.Rf1-f2 Bf8-d6 2.b2-b4 Rd8-e8 3.b4xc5 Bd6xc5 4.Kg1-h2 Re8-e2 5.Rf2xe2 Nd4xe2 6.Bg2-f3 Bc5-d6 7.Kh2-g2 Ne2-f4+ 8.Kg2-h2 Ng3-e2 9.Kh2-h1 we agreed to a draw. this line is probably crap, but it just shows how quickly black can get something going against the white king (2.b4 is probably a mistake, only activates black's bishop). another attempt, forcing ruffian to play 1.Re1 as suggested by the programs tested by slater: 1.Rf1-e1 Bf8-d6 2.Qd1-a4 (this was ruffian after 20 minutes, rest of the game at 10s/move again) Ng3-e2+ 3.Re1xe2 Nd4xe2+ 4.Kg1-f2 Ne2-d4 5.Kf2-f1 Rd8-e8 6.Qa4-d7 (with a 0.00 score already) Bd6-h2 7.Ra1-e1 Re8xe1+ 8.Kf1xe1 Rg8xg2 9.Qd7-d8+ Rg2-g8 10.Qd8-f6+ Rg8-g7 11.Qf6-d8+ Rg7-g8 12.Qd8-f6+ and another draw again, this is no real analysis, but just goes to show how quickly black can be very ok here. last attempt, ruffian @1 min/move after 1...Bd6 again, with it's scores in brackets: 1.Rf1-e1 Bf8-d6 2.b2-b4 (+1.60) Ng3xf5 3.b4xc5 (+0.80) Bd6xc5 4.Kg1-h1 (+0.30) Rg8-g3 5.d5-d6 (+0.06) Bc5xd6 6.Re1-e4 (+0.15) Bd6-c7 (+0.03) 7.Ra1-b1 (+0.03) b7-b6 8.Re4xd4 (-0.28 - aha!) Nf5xd4 9.Qd1-f1 (-0.34) f7-f5 10.Qf1-e1 (-0.32) Rd8-g8 11.Qe1-e7 (-0.54) and i left it at that, as i guess the point is proven. note that all slater's programs want to play 1...Ngxf5 after 1. Rf1-e1. looks a bit like computer-like greed to me. i quite agree that black has to get something going against the white king here, and he shouldn't worry about that measly pawn :-) cheers martin
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.