Author: Andrew Williams
Date: 07:32:03 05/24/03
Go up one level in this thread
On May 24, 2003 at 10:02:06, Uri Blass wrote: Uri, Thanks for your detailed response. I've interspersed some comments below. >On May 24, 2003 at 09:03:54, Andrew Williams wrote: > >>Hi, >> >>PM lost a game against Deep Sjeng on ICC yesterday. For a long time PM >>thought it was going to win, but it underestimated DS's attack on its >>e-pawn. I think the turning point was here: >> >>[D] 5k2/5p2/1p2p3/p1n5/P1Q5/4RNPP/5PK1/3q3r b - - 0 39 am Nxa4 >> >>PM was very happy with 39... Nxa4?, but this seems to me to "miss the point" >>of the position. PM drops it in 38 seconds if I tell it to search from here. > >Movei need more time but I guess that my hardware is slower. > I should have stated the hardware: TombRaider Dual Xeon 1.7GHz PostModernist Athlon 1.2GHz >It can see the draw after Nxa4 at depth 9 and changes it's mind. > > > >>What do your programs think about this position? Do they ever consider Nxa4? > >Of course. > >Fritz8 and shredder7.04 also like this move at small depth. >I think that only programs with bugs do not like it at small depth. > > >>How do they evaluate the position? > >+1.08 for black at depth 1 and +1.04 for black at depth 2. > > > As always, I'm especially interested in >>amateur programs' analyses. >> >>PM evaluates the position above as +1.37 for Black. Technically speaking, >>it likes the attack on the White King and misunderstands the fact that >>once the e-pawn falls, White will have an overwhelming attack on the Black >>King. > >Why do you think that the evaluation problem is about taking a4. > I think it's a combination of doing that and thinking that Black is ahead. PM would never go for the perp, because it thinks that Black is winning by a lot. >I can see that after Nxa4 Ng5 black has perpetual check by Rg1+ Kh2 Rh1+ > >Even after Qg1+ Kf3 the moves Qd1+ or Nc5 are probably enough for a draw but PM >blundered by 41...Rh2 > >[D]5k2/5p2/1p2p3/p5N1/n1Q5/4RKPP/5P2/6qr b - - 0 41 >bm Qd1,Nc5 > >Uri Thanks. I'll look at this when PM's finished its current game. Andrew
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.