Author: John Stanback
Date: 08:58:39 09/26/99
Go up one level in this thread
On September 25, 1999 at 18:44:17, Peter McKenzie wrote: >In a game on ICC Beadle (LambChop) reached the following interesting position: > >2kr1b1r/ppp4p/1q4p1/3p1p1b/3N4/2PQPPP1/PP1N2P1/R4RK1 w - - 0 17 > >Tactics: > >Here LambChop blundered with Nxf5, it needs an 8 ply search to avoid Nxf5. The >key variation being Nxf5 gxf5 Qxf5+ Kb8 Qxh5 Qxe3+ Rf2 Bc5 Rf1 Qxd2. This is a >10 ply variation, but we have a couple of checks in there so the check extension >sallows it to be seen in 8 ply. > >My q-srch prunes out captures that look bad, so Qxd2 is pruned as the simple >algorithm I use thinks d2 is protected by the R on f2 :-( I guess if my pruning >algorithm in q-srch was smarter then I'd see this line one ply earlier. I'd >also see it earlier if I just looked at all captures in q-srch... > >How many ply do other programs need to avoid Nxf5 ? I know that crafty needs 8 >ply too... > > >Positional: > >I was slightly surprised when Bob Hyatt said he thought white was doing badly >anyway (crafty agreed). I personally think that white is doing fine, I actually >prefer white's position. > >Here is how I see the position: >White has 2 knights vs two bishops, so white should try to keep the position >somewhat closed. Currently the knights are quite well posted, while the Bh5 is >quite bad although the Bf8 is potentially very good but currently not doing >much. > >White's pawn structure is weakened, but black isn't currently in a position to >take advantage of this. White will probably play f4 eventually, leaving holes >on e4 and g4 but this doesn't look too bad for white as black doesn't have a >knight to sink in there. Also, the white knights cover those squares quite >well, and after f4 the Bh5 is really locked out. > >White has a clear and simple plan of advancing on the queenside. I think black >should try to open the position and infiltrate the dark squares, but its not >clear to me how to achieve this. > >I think white should play 1.b4 which is in harmony with the plan of advancing on >the queenside, it also dissuades c5 (thereby keeping the strong knight on d4) >and has the added bonus of possibly threatening Nxf5. I wouldn't worry about >Nxf5 too much though, as even if it wins a pawn the variations look very open >and dangerous for white. > >After 1.b4, ...c5 looks poor. Simply 2.bxc5 followed by 3.Rab1 with good >initiative for white. > >Another line is 1.b4 Bd6 2.f4 Rhe8 3.a4 threatening a5 Qa6 b5 trapping the black >queen :-) > >Maybe something like Qf6 is best for black? The idea being to play g5 >eventually. Pretty slow though. > >So there you have it, the opinion of someone used to be 2300 Fide but these days >struggles to maintain a 2000 ICC bullet rating :-) > >LambChop gives white as +0.4 after 9ply, and wants to play a4. > >So what do other people/programs think? > >cheers, >Peter Zarkov on a K6-333 takes 1.4 secs to fail low in the 6th iteration. It then chooses a4 and sticks with it for a while with a score of -0.63 pawns for white. My pruning of captures in qsearch is not very aggressive which might help here. 00:01.0 28800 6> -52 Nxf5 Qxb2 g4 Bxg4 fxg4 gxf5 gxf5 00:01.4 44928 6 -108 Nxf5 Qxb2 g4 Bxg4 fxg4 gxf5 gxf5 00:01.9 72960 6 -88 b4 Bg7 a4 Rhf8 a5 Qd6 f4 00:03.9 176000 6 -78 a4 Bh6 b4 Rhe8 Nxf5 Bxe3+ Nxe3 Rxe3 00:05.1 239590 7> -78 a4 Bh6 b4 Rhe8 Nxf5 Bxe3+ Nxe3 Rxe3 00:06.6 306560 7 -79 a4 Bh6 b4 Rhe8 f4 Qf6 c4 00:18.1 860037 8> -79 a4 Bh6 b4 Rhe8 f4 Qf6 c4 00:26.9 1358720 8 -63 a4 Bh6 f4 Qf6 a5 a6 N2f3 Rhe8 00:34.7 1801488 9> -63 a4 Bh6 f4 Qf6 a5 a6 N2f3 Rhe8 John
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.