Author: Stephen A. Boak
Date: 17:45:07 08/05/01
[D]r3kb1r/pp1bnppp/1qn1p3/3pP3/3p2Q1/3B1N2/PPP2PPP/RNB1R1K1 w kq - 0 1 In an OTB game yesterday, in a very interesting non-book position (see FEN diagram) arising a couple moves after the novelty 6...Qb6!? (the known Qc7 is likely better) in the French Advance Var, Black played 8...Bd7 and missed an opportunity for Nb4!, winning the Bishop pair and removing the important kingside attacker that covers h7-g6-f4. I had thought of the Nb4 idea, but discarded it in favor of Bd7, wanting to continue with development (Rc8 when possible, etc). White replied 9. a3, permanently preventing Nb4. Here's the early game score including the diagrammed position and a few following moves: 1.e4 e6 2.d4 d5 3.e5 c5 4.Qg4 cxd4 5.Nf3 Nc6 6.Bd3 Qb6 7.0–0 Nge7 8.Re1 Bd7 9.a3 a5 10.Nbd2 Ng6 11.h4 Qc7 12.Qg3 Line This was my inaugural French Defense (I like to change my openings every now and then, for variety and to learn something new), so I don't have much experience or understanding of normal French lines. I'm getting that now, heh heh, from the game lesson and post mortem analysis. Fritz 6 analysis (18 ply) indicates that Black is slightly better after 8...Bd7. Black lost the actual game via subsequent play that included one major error. You could rightfully say my understanding of the French Defense Advance Variation has taken a big leap based on this single game! :) After full game analysis (Athlon 1.33GHz DDR, 266FSB, 512MB PC2100) with HT 256MB and Blunder Check set at 20 min/move, Fritz 6 scored 8...Bd7, the actual move, as 0.00 and gave as best move: 8...Nb4 -0.25/16 The below ply by ply analysis of Fritz 6 is after stepping forward manually through the game to the critical position above, with Fritz 6 set to show the top 5 moves. After 15 ply, Fritz 6 did not even include Nb4 on its list of top 5 moves! After 16 ply, Nb4 was 5th on the list. After 17 ply, Nb4 was 2nd on the list. After 18 ply, Nb4 was 1st on the list. The computer is completing ply 19 at this time. Analysis by Fritz 6 (19 ply, through line 3): 1. ³ (-0.34): 8...Nb4 9.Qxd4 Qxd4 10.Nxd4 Nxd3 11.cxd3 Nc6 12.Nxc6 bxc6 13.Be3 2. = (-0.25): 8...Bd7 9.Nbd2 Nb4 10.Nxd4 Nec6 11.N2f3 Nxd3 12.cxd3 h5 13.Qf4 3. = (-0.19): 8...Ng6 9.Nbd2 Be7 10.Nb3 Bd7 11.Nfxd4 h5 12.Qg3 Bh4 13.Qh3 4. = (-0.16): 8...h5 9.Qf4 Bd7 10.b3 Nb4 11.Nxd4 Nxd3 12.cxd3 Ng6 13.Qe3 5. = (-0.13): 8...g6 Analysis by Fritz 6 (18 ply): 1. ³ (-0.34): 8...Nb4 9.Qxd4 Qxd4 10.Nxd4 Nec6 11.Nxc6 bxc6 12.Be3 c5 13.Na3 2. = (-0.25): 8...Bd7 9.Nbd2 Nb4 10.Nxd4 Nec6 11.N2b3 Nxd3 12.cxd3 h5 13.Qf4 3. = (-0.19): 8...Ng6 9.Nbd2 Be7 10.Nb3 a5 11.Rd1 Qc7 12.Nbxd4 Ncxe5 13.Nxe5 4. = (-0.16): 8...h5 9.Qf4 Bd7 10.b3 Nb4 11.Nxd4 Nxd3 12.cxd3 Ng6 13.Qe3 5. = (-0.13): 8...g6 Analysis by Fritz 6 (17 ply): 1. ³ (-0.31): 8...Bd7 9.Nbd2 Nb4 10.Nxd4 Nec6 11.N2b3 Nxd3 12.cxd3 h5 13.Qf4 2. ³ (-0.31): 8...Nb4 9.Qxd4 Qxd4 10.Nxd4 Nxd3 11.cxd3 Nc6 12.Nxc6 bxc6 13.Be3 3. = (-0.16): 8...Ng6 9.Nbd2 Qc7 10.Bb5 Be7 11.Bxc6+ bxc6 12.Qxd4 f6 13.exf6 4. = (-0.16): 8...Nf5 9.a3 Bd7 10.Nbd2 5. = (-0.16): 8...h5 Analysis by Fritz 6 (16 ply): 1. = (-0.25): 8...Ng6 9.Nbd2 a5 10.a4 Bb4 11.Re2 Bd7 12.Nb3 Rc8 13.Bxg6 2. = (-0.19): 8...a5 9.Na3 Bd7 10.Nb5 Nxe5 11.Nxe5 Bxb5 12.Qf3 f6 13.Qh5+ 3. = (-0.19): 8...Bd7 9.Nbd2 Ng6 10.Nb3 Rc8 11.h4 Nb4 12.Nbxd4 Nxd3 13.cxd3 4. = (-0.19): 8...Nf5 9.a3 Bd7 10.Nbd2 g6 11.Qh3 Be7 12.Nb3 Rc8 13.Rd1 5. = (-0.19): 8...Nb4 9.Qxd4 Qxd4 10.Nxd4 Nxd3 11.cxd3 Nc6 12.Nb5 Rb8 13.a3 The score difference due to playing 8...Nb4 instead of Bd7 is not clearly in favor of Nb4 according to Fritz 6, until it has looked relatively deeply into the position. I suspect the power of the d3-bishop to support a White attack is latent, and not readily apparent until deeper lines are analyzed. Then Fritz 6 realizes the path to the best score for Black involves an immediate Nb4. I believe a program's evaluation factors (knowledge & weighting) have a lot to do with a suggested best move and the behavior of pruned alpha-beta trees. Perhaps other programs want to immediately eliminate the d3-bishop. ANALYSIS QUESTIONS-- 1. Is 8....Nb4 the best move in the position? [I think so, based on my analysis and 'understanding' so far, although I realize other moves may be quite playable, and a stronger player may have improved judgement] 2. How long does it take other programs, and at what ply level, to score 8...Nb4 as best move? 3. What score is given, instead, for the text move 8...Bd7? 4. Why did Fritz 6 score 8...Bd7 as 0.00, and 8...Nb4 as -0.25/16 (i.e. 16 ply), after Blunder Check full game analysis at 20 min per move, when the manual method of stepping Fritz 6 through the moves, showing top 5 moves per position, didn't score 8...Nb4 as top move until ply 18? A. Is this due to differing HT entries, since Fritz 6 conducts Bunder Check in a backward moving fashion? B. Is this due to alpha-beta cutoffs & resulting pruning differences--from forcing Fritz 6 to calculate/display the top 5 choices in continuous analysis mode, versus Blunder Check mode which only calculates a single, best move? C. Any other possible reasons? After Fritz 6 completes 19 ply I will post the results, and then use Chess Tiger 14.0 to analyze the same critical position. --Steve
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.