Author: Robert Hyatt
Date: 18:49:49 07/14/04
Go up one level in this thread
On July 13, 2004 at 17:12:39, Peter Berger wrote: >The question raised at CCC by some was actually if the opening book of Crafty >was "en par" with its opponents or if Crafty was "seriously crippled" by it. >Let's take a look. > >Crafty played the following opponents in the tournament: Junior, Shredder, Deep >Sjeng, ParSOS, Fritz, Falcon, Diep, IsiChess, Movei, The Crazy Bishop and >FIBChess. > >An excellent opening for white would leave it out of book with an edge in a >position it it supposed to understand, on the black site equality would be >acceptable given there is some imbalance to work on and it is no dead draw. A >fine opening is basically the same but slightly worse than excellent. I think a point most miss is that it is _impossible_ to get a big edge from the opening book. It just isn't possible unless your opponent makes a mistake. As a result, the basic idea becomes one of "tit for tat". IE give up something to get something. With the idea that what you give up is compensated for by what you gain. IE going for a queenside pawn majority by giving up kingside pawn structure could be a good deal if (a) you defend against the probable kingside attack, and reach an endgame so that (b) the queen-side majority becomes an important feature. In my case, I generally look for some sort of imbalance that favors Crafty, when possible. One thing is for sure, as white I would rather have a score of -.5 being a pawn down, rather than +.5 being a pawn up. Because in the first case, I have 1/2 pawn of positional compensation to fight with, while in the second case, I am a pawn up but in a miserable position. > >This would mean excellent openings in games against Junior, Shredder, Fritz, >Movei, The Crazy Bishop and FIBChess. Fine openings against ParSOS and Deep >Sjeng. > >Openings against IsiChess and Diep did show some minor book problems - but both >games were won by Crafty anyway, so they are partly irrelevant for the >discussion. > >The opening against Falcon was fine too - we wanted to win this game, so Bb4 and >Nd4 were not acceptable as being too drawish. Bd6 looks strange for the >unaccustomed human eye, but it has been successfully introduced into grandmaster >play some years ago. But even if it would be argued that this game led to an >opening Crafty didn't understand too well ( although I wouldn't agree in this >case) ; the game was actually won too. The thing here is that as a human, I'd insist on moving the d or e pawns quickly as black. But that would foil the later ability of the rook going to b6 and then helping on the king-side after the bishop sac. Problem is that Crafty understands none of that... But it can be improved... > >Crafty lost games against Junior, Fritz and ParSOS and drew against Shredder and >Sjeng while winning the other games, which led to a dead tie with Fritz to the >4th place with 7.0/11 and equal Buchholz. No loss of points in the tournament >was due to a weakness in opening book, one might even slightly argue to the >contrary thinking about some of the games. I would agree. I saw two openings that caught my eye. The Falcon (I think) opening with the blocked center pawn was one. The other was the SOS game with an early b3/Bb2 if I recall correctly. Crafty doesn't deal with wing-type openings very well, which will cause it to generally play them pretty poorly since it is trying to directly influence the center more than it should when trying a b3/Bb2 type development. These were not _bad_. And there is nothing that suggests that they were worse than another alternative, in fact. This is _not_ an easy problem. Most important detail, IMHO, is that in _no_ game was it lost right out of book. That is a major detail because it happens frequently enough in general. But it didn't here. In general, if it comes out of book with reasonable chances, I'm happy, whether the first search says -.5 or +.5. I'll try to critique each opening from that perspective soon... > >While I am very much open to criticism, there is simply no data to support it >in this case. > >Of course Crafty's openings didn't always follow the latest hype in grandmaster >chess and some mainlines were avoided. But this was done intentionally, mainly >because Crafty does have some problems with middlegame tactics and king safety >compaired to its main competitors and to emphazise its strength in endgame play. >Another reason was of course the limitted time for preparation for the event. >But I don't buy the argument that Crafty was in any way handicapped in this >tournament. Nor do I. Things went very well, IMHO. It should not have lost the Fritz game. QPPP vs RRN is simply bad for the QPPP in 90% of the games. I'll look at this to see what went wrong. > >Talking about "crippled by book" - it might be worth some time to look at >ParSOS's games ; that's what I would call a crippled book. Dead lost out of book >against Fritz, The Crazy Bishop and Junior; in many other games questionable >positions were reached. In the few games where it accidently reached reasonable >positions ( which unfortunately included the game against Crafty :) ) it was a >very strong contender - it just lost too many points on book. > >And whether taking part in the eternal Najdorf battle of book authors is >objectively really a good idea - looking at the results of the games in the >tournament it looks *much* more like a worse version of Russian roulette to me. >The only program that never seemed to have book problems was actually Junior. > >Credit where credit is due: several opening ideas for Crafty and this event were >burrowed from the recently released book "The Chess Advantage in Black and >White: Opening Moves of the Grandmasters" by IM Larry Kaufman (Random House >2004) after having been checked for usability and tactical oversights; else the >short preparation time would have not been sufficient. It's a very good book in >general IMHO and most probably the first book on openings co-authored by Junior, >Fritz and Hiarcs (which sometimes make it a very funny read :) ). > >Yours sincerely >Peter > >PS: Last message written in Ramat-Gan - will post it tonight at home :)
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.