Author: Peter Kappler
Date: 23:00:34 11/30/99
Go up one level in this thread
On November 30, 1999 at 20:48:39, Ricardo Gibert wrote: >On November 30, 1999 at 17:43:50, Robert Hyatt wrote: > >>On November 30, 1999 at 03:35:54, Ed Schröder wrote: >> >>>>Posted by Robert Hyatt on November 29, 1999 at 20:17:47: >>> >>>[ snip ] >>> >>>r3q1k1/ppp1rpp1/2n1b2p/8/2P2B2/3B4/PPPQ1RPP/5RK1 w - - bm Bxh6; >>> >>>>>>>00:00:00 1.00 1.02 1.Bxh6 gxh6 2.Qxh6 >>>>>> >>>>>> >>>>>> >>>>>>This looks _incredibly_ dangerous. For every position where Bxh6 works, >>>>>>there are 10 positions where Bxh6 loses. Here a piece for two pawns looks >>>>>>awful if the other pieces can't get over to help out... Speculative play is >>>>>>nice, as in the old days of the Novag gadgets from Kittinger, and in the >>>>>>current >>>>>>play of CSTal... but it can backfire big-time as well... >>>> >>>>>> >>>>>>Crafty doesn't like the sac, unless it sees actual material coming back, >>>>>>because >>>>>>I have spent a lot of time teaching it which types of material imbalance are >>>>>>bad and which types are good. A piece for 2 pawns is always bad unless >>>>there >>>>>>is some tactical conclusion at the end. And given the above PV there >>>>>>obviously >>>>>>isn't anything except a somewhat naked king position for black... >>>>> >>>>>I can not believe you saying this. After 1.Bxh6 gxh6 2.Qxh6 there is a queen, >>>>>2 rooks and a bishop all pointed very dangerous at a naked king. Rebel giving >>>>>+1 is fully justified. >>>>> >>>>>Every 1800 rated chess player will play 1.Bxh6 immediately, no need to >>>>>calculate. >>>>> >>>> >>>>Sorry, but _I_ am an 1800+ player and _I_ wouldn't play it immediately. >>>>Because black can play f5, and it doesn't look nearly so clear after that. The king >>>>has a way out... the black rook now can interpose on the g/h files. The >>>>f file is no longer useful unless white does something else... >>> >>>1.Bxh6 gxh6 2.Qxh6 f5 3.Bxf5 the black king is even more naked. I guess >>>we have a different view on king safety. >> >> >> >>I probably have as big (or bigger) king safety scores as you. However, I >>have a big penalty for trading a piece for 2-3 pawns, as it almost always >>loses unless tactics make it work... >> >>> >>>>It may be good. It may not work. But as a human, I don't play Bxh6 here >>>>without calculating something worthwhile at the end of the sequence. I let >>>>an IM look at this for several minutes... he left unclear whether it was >>>>good or not... >>> >>>Who was the IM? >>> >>>After 1.Bxh6 gxh6 2.Qxh6 f5 3.Bxf5 I am pretty sure all programs soon will >>>display a +2 or +3 score. >> >> >>Note that I didn't say it doesn't work. I said that at the shallow depth >>of 1-2-3-4 plies, it looks _very_ dangerous to do this.. because black has >>some defensive resources that don't look obviously bad until you search >>deeper. > >I think the culprit is crafty does not understand the role the passive rook on >a8 has in assessing the position. The value of a passive major piece in such a >sharp position plummets precipitously. After 3.Bxf5, any evaluation less than >half a pawn in favor of white should be considered a bug. Naturally, the actual >evaluation for white is much greater, but I assume here that the program does >not "see" the win. > >You have said that crafty generally does not play such sacrifices, unless it >"sees" a win, calling programs that do otherwise "speculative". But isn't crafty >guilty of doing the reverse? It is speculating that the sac is unsound despite >positional evidence to the contrary. Crafty trades one form of speculation for >another at the cost of being boring. > Ricardo, I agree with your assessment completely, especially the comment about the rook on a8. Black is effectively playing without that piece for the duration of this attack, which proves fatal. I think it's fantastic that Rebel evaluates the position as +1 after a 1-ply search. If only we could agree in our other discussions... :-) --Peter
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.