Author: Gian-Carlo Pascutto
Date: 08:12:16 01/21/02
Go up one level in this thread
On January 21, 2002 at 09:34:51, Chessfun wrote: >Sjeng's programmer Gian-Carlo Pascutto had posted at CCC that he needed an >operator for Sjeng at CCT4 as he was unavailable. Hi Sarah! First of all, thanks a lot for operating! I couldn't resist the temptation yesterday to login to ICC and take a few peeks to see how Sjeng was doing. I hope you don't mind if I add some of my own comments inbetween :) >Saturday Jan 19. > >Round 1 >Sjeng v Armageddon 1-0 >Sjeng has white against Armageddon and immediately has a problem being out of >book at move three. Which was entirely my fault due to giving you the wrong book :( Funnily enough, in the position where this happened white had at least three logical moves which were all theory. Sjeng managed to pick another move that was only played by weaker players. Would this have a deeper meaning? :) Armageddon recaptured and brought the queen out...probably a bad idea. Sjeng would be hunting her for the rest of the game. >Round 2. >Ban v Sjeng 1-0 >Sjeng has black against Ban (Deep Junior 7). I made a new book from a 2600.pgn >file found at Mogens Larsen's website. (http://home1.stofanet.dk/moq) By move 15 >Sjeng is out of book and on my other PC with Junior 7 loaded I can see Junior is >still in book. Sjeng manages to calculate and follows the book line till move >20. Sjeng seems to have a good game until move 29. when it plays 29.....Rxc5 >instead of the probably better Bxd6. Sjeng is losing but not out until it plays >37....Qb2, I can see from Sjeng pondering the move and Junior's eval on my >other PC that the move is lost. [D]8/4kr2/p2bppQ1/4q3/Pp6/4P1P1/4BP1P/2R3K1 b - - 0 37 This was a bitter pill. The tactical shot after Qb2? was not that hard to see, and I think most other amateurs will also find it fairly quickly. I will have to work on this...it should not have happened. It also blundered a move earlier: [D]8/4kr2/p2bppQ1/q7/Pp6/4P3/4BPPP/2R3K1 b - - 0 35 Qe5 is not a good move here. >I let the game go on just to get to move 40 :-) >resigning after losing queen for rook. Junior played solid building it's >advantage slowly which wasn't what I had expected, which was for Junior to come >charging as version 7 is apt to do. Junior 7 seems to be very good at complicating the position, and when it does so, it often comes out on top. This is also what happened here. It complicated the position, all the while keeping it's opening advantage. When the tactics started, it came out on top. >Round 3. >Sjeng v Hossa 1-0 >Sjeng stayed in book until move 14. At which point it played Rfc1 instead of >what I saw was in Junior's book Qd3. This Rfc1 allowed black to exchange queens >and with two passed queenside pawns I was a little concerned that allowing the >queens to come off would allow black the chance to move them. I was concerned when Sjeng played a gambit for development, and indeed it did not see adequate compensation. Nevertheless it played correct moves. Allowing the queen trade was strange, but perhaps not unsound. >After the queens >were off Sjeng got its rook onto the seventh where I expected black to >challenge it, however black chooses 18...h6 instead preferring to prevent >Bg5 and allow a later g5. I was a little concerned black would find the move >29....f6 but instead played 29....Ne4 which appeared to me the losing move. >From this position white's lead increased steadily resulting in mate at move >57. What was very strange about this game was Sjengs evaluation. It was showing a huge advantage for white. If I put the position into another program, it'll just say white is somewhat better. I was afraid Sjeng was overevaluating something as that could lead to trouble later on. The end of the game was also very strange. I expected that Hossa would have to give a quality or a piece to stop the white passer, but it stayed on the seventh rank for the rest of the game! Instead, Sjeng wove a mating net for Hossa's king. >Sjeng stayed in book until >move 14 and the position looked pretty balanced. From the moment out of book it >seemed that Gandalf steadily increased it's lead I had the impression that it was pretty balanced until Sjeng made a mistake by playing [D]3rr1k1/pp4pp/n1p2p2/2q1n3/2PN2b1/P1Q1P1B1/2B2PPP/RR4K1 b - - 0 23 b6? This allowed a combination that gave Gandalf total control of the board. It didn't hesitate and went on for the kill. I will have to look into Sjeng's defense. It looked strange, but that could be because it was already lost at that point. >and on move 27. Sjeng played >27...Re7 instead of the likely safer Nb8. After losing the exchange on move 36 >it was only a matter of moves till Gandalf won. I resigned after 48. Bg6+ with >if I recall Gandalf 5 showing mate in 18 on my other PC. > > >Round 5. >Sjeng v Chezzz 1/2 - 1/2 >A strange game, which early on I thought was going to be a clear win, but which >turned into a scrape for a draw. After the opening Sjeng seemed to have a clear >advantage positionally, [D]rn1q1rk1/pb2bppp/1pp1pn2/8/3PPN2/P1N5/1PP1B1PP/R1BQ1RK1 b - - 0 10 Chezzz made a positional blunder by playing a5? It should have played c5! instead. For comparison: Rebel - Sjeng, Dutch Open 2001 [D]r2q1rk1/pbp1bppp/1p2pn2/8/3P1B2/2PB1N2/PP2QPPP/R4RK1 b - - 0 10 Sjeng played a5? and lost. It should have played c5! instead. Your learn best from your own mistakes :) >but choose to play 14. Nh5 instead of the clearly >winning Bd3. >Not to worry as after 14....Bc8 we can still play 15. Bd3 but this >time played IMO the odd looking Qe1. 14.Nh5 was a good move, but indeed the next move should have been 15.Bd3! without question. [D]rnbq1rk1/4bppp/1pp5/p2pP2N/3P4/P7/1PP1B1PP/R1BQ1RK1 w - - 0 14 >After exchanging queens Sjeng played 18. >fxg7 and I was concerned about my opponent playing 18....Re8 however >fortunately he choose to take the pawn on g7 and by move 21. we have a >position that although white holds the keys, seems drawn. I was disappointed when Sjeng decided to trade queens. It has a better chance to outplay another amateur program when kingattacks are involved than when it has to do it in the endgame. (though its endgame has improved quite a lot after the loss to Spiderchess in the Dutch Open, and thanks to some suggestions from Dieter Buerssner) I was chatting with the author of Celes at that point and he remarked that Sjeng had both bishops and a better pawnstructure so it would just outplay Chezzz in the ending. I was not so sure the advatage was that great but I agreed Sjeng would probably outplay Chezzz in the ending. Fate, it seems, is not without a sense of irony. >On my other PC, I have Fritz 7 >running, seeing my opponents evals as they were kibitzed. I was concerned, as >it seemed to me, my opponent was clearly on the right track as far as >understanding the position, while it seemed that Sjeng didn't see that he was >in trouble. Sjeng totally lost track of what was happening. It exchanged it's bishop pair in order to give black 2 isolated pawns on top of the 2 he already had. But after the bishop pair is gone, this not so an important advantage any more! Black just puts the pawns on the other color of the bishop and white has only one piece to attack them. Moreover, the exchange brought an additional weakening of Sjengs b2 pawn. Sjeng has an exponential penalty for isolated pawns, i.e. so that 2 isolated pawns is worse than 2 x 1 isolated pawn. It seems that this penalty is much too large here, and that Sjeng underevaluates the bishop pair. David wrote a few days ago here that his program understands rook activity in the endgame. Sjeng does not understand that, and it clearly showed! Chezzz completely outplayed Sjeng at this. The following position shows how bad it was: [D]1r6/7p/8/p3kp2/2p1b3/P3K1P1/1P2R2P/3B4 b - - 0 36 >Between moves 21-58 it seemed Chezz played almost flawlessly then somehow at >move 60 lost track playing 60.....Rh4+ when either c2 or Kd3 appear to win. >Chezzz having tablebases would also probably have helped it, in the end a lucky >draw for Sjeng. > > >Round 6. >Somnus (Amyan) v Sjeng 1/2 - 1/2 >After a restart caused by pairing issues the opening seemed ok for Sjeng coming >out of book at move 14 and playing 14....b4. Sjeng also looked at Ne5 which >looks clearly the better of the two especially a few ply later. Somehow on move >16...Sjeng played Ng4 with the thought 17. h3 Nh7. However after 17. h3 saw that >Ng7 was no good and choose to put the Knight on h6 instead of the probably >stronger 17....Ne5. I have not been able to analyze this completely yet, but Sjeng may have been caught in an (un?)intentional booktrap. It came out of book with a positive eval, and Shredder agreed, but both programs were quickly dropping in score. Even though Ne5 was a better defense than the simply losing Nh6, I think black is also quite bad in that line. >Meanwhile my opponet is announcing to (64) +1.2, +1.8 >bugging me no end, (no idea why). Although I'm still showing a small minus >score. After my opponent doubled rooks at move 20 it seemed to me immediately >that it would now be possible to hold a draw. After 24. Nc4 I was more than >happy as Sjeng also was. By this time my opponent has stopped posting his + >score as he. I assume like me can see that it means little. I'm happy with Sjengs play there. It did it's best to defend a losing position and succeeded. >My one last concern >was over the move 39....Qb2 when it seemed either Kg7 or Kh7 were relatively >easy draws. Qb2? again :) >But Sjeng seemed to defend very well and handled the position well >and after 48. Qc7 Sjeng can see an easy draw, which my opponent offered on move >51. > > >Without having much experience of using Sjeng my thoughts are that the two >losses against Deep Junior 7 and Gandalf 5 with black are not a surprise. The >two wins against Armageddon and Hossa were both well played games for Sjeng >where it seemed to clearly understand the positions. The two draws were both >lucky. Chezzz impressed me especially it's evals of the position from around >move 35. >Amyan again knew clearly that it was winning and with any luck on it's >part would have won. This is the one thing we disagree on :) Amyan had a good position out of book, did not put enough pressure to break through Sjengs defense and was not able to see the shot after Qb2. I personally wouldn't call that luck :) (Then again, I'm the one who agreed Sjeng was going to outplay Chezzz in the endgame. Hah!) -- GCP
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.