Author: Robert Hyatt
Date: 07:04:46 06/24/98
Go up one level in this thread
On June 24, 1998 at 03:38:59, Ernst A. Heinz wrote: >On June 23, 1998 at 16:19:32, Bruce Moreland wrote: >> >>I get a fail high a few minutes into this, but the re-search fails low. >> >>I like aggressive play, and I like speculative play that works, but I think > >that programs are seeing something here that doesn't exist. > >Bruce, > >After running for 45 min. "DarkThought" shows the same behaviour as "Ferret" >and switches to 1. Nxe6 in iteration #14. It kept this move over night until >completion of iteration #16 with a slightly positive score of +0.25. > >PV = 1. Nxe6 fxe6 2. Qxe6 a6 3. Bxa6 Bxa6 4. Qxa6 Ra8 5. Qc6 Ra5 ... > >But again this might still be too optimistic (3 pawns for the knight). > This is almost *never* a good trade, unless the pawns are very mobile and can advance quickly enough to recover the material. Otherwise, they only become targets for black's extra piece. I've gone so far as to put some special code in my eval, rather than trying to juggle the piece/pawn values to stop this, as I have lost way too many such games on the chess servers. Crafty won't even consider Ne6 here, it considers N for 3P to be -1.00 right off the bat... now if the three pawns are passed, and are on the 5th rank, it might go for it assuming there's no blockade. >> [...] >> >>I would be happy to be shown a line that has any promise for white. If someone >>thinks that this really does win, or at least offers good counterplay, please >>post a line. >> >>My own analysis shows a nasty check or two, then black is simply up material. > >Starting after 1. Rxe6 fxe6 "DarkThought" suggests 2. Bh6 and feels down with >a score of -0.5 after overnight analysis. > >PV = 2. Bh6 a6 3. Nxe6 axb5 4. Ng7+ Kf8 5. Ne6+ Kf7 6. Nxd8+ Rhxd8 ... > >>As I said before, I like speculative play, and this might be the best way to >>play very strong players, I do not know. >> >>But I don't think this move gets an "!" or "!!" in an absolute sense, I think >>it is "?!", and I don't know if this is properly understood by any of these >>programs, including mine. > >You are absolutely right -- most programs including "DarkThought" seem to >over-estimate the attacking chances of White in this position. > >But anyway, it is such an interesting sacrifice with many chances for Black to >go wrong that I do not mind "DarkThought" to play it. Other games are lost in >much worse fashion than this position (if lost at all) after the sacrifice ... > >=Ernst= it depends on what you want your program to do. Against a GM, Re6 will likely lose *every* game. Against an expert (or even a master) it might well work out ok. But to me, it's the "GM-case" that is interesting, as I can already beat experts and masters without tossing material out the window, and only getting attacked for the trouble.
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.