Author: Mark Young
Date: 18:15:25 05/29/99
Go up one level in this thread
On May 29, 1999 at 20:04:33, Terry Presgrove wrote: >On May 29, 1999 at 15:31:42, Mark Young wrote: > >>On May 28, 1999 at 23:56:35, Terry Presgrove wrote: >> >>>Recently I was playing Rebel10c a 30 0 game that went as follows. >>> >>>[Event "?"] >>>[Site "?"] >>>[Date "?"] >>>[Round "-"] >>>[White "?"] >>>[Black "?"] >>>[Result "*"] >>> >>>1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be2 e5 7. Nf3 Be7 8. >>>Be3 O-O 9. Qd2 Ng4 10. Bg5 Bxg5 11. Nxg5 h6 12. h4 hxg5 13. hxg5 Qb6 14. >>>O-O-O Nxf2 15. Nd5 Qd8 16. Qe3 Nxh1 17. Rxh1 Nc6 18. Qg3 Nd4 19. Qh4 Nxe2+ >>>20. Kb1 f5 21. Qh8+ Kf7 22. g6+ Ke6 23. Qxg7 fxe4 >>>* >>>I thought that I had a won game and went straight forward for the win >>>with 19. Qh4? missing the obvious Bc4 which wins. At the time I had that" I >>>missed the winning line" mentality and dejectedly aborted the game after move >>>23.(although it turns out whites position is still quite intersting after 24. >>>nc7!) >>> >>> Anyway to the point .... when I went back over the game with Rebel10b I >>>discovered several wins for white but to my amazement I discovered that >>>18. Qg3 was not the best move......but rather( now quite obvious) nf6!! (mate in >>>10). >> >>18. Qg3 in not the best move in term of the shortest mate, but there is a forced >>win after the move 18. Qg3. (I was finding forced mates in 12 to 15 moves but >>none longer) Regardless of what it is, black is hopelessly lost after the move >>18. Qg3. >> >> Rebel10b finds in 3:23 on my k6-350 w/100M hash. Crafty 16.6(96M hash) >>>switches to nf6 at 2:21 but does not find mate in 20 minutes. Crafty 16.8 stays >>>with Qg3 for 15 minutes . Up until recently I have felt the top computer >>>programs were at lower GM levels but I am beginning to wonder? Perhaps Dr.Hyatt >>>is correct? >> >>Dr. Hyatt may be correct, but I don't see how this positions shows in a logical >>way that programs are not of GM strength. Is not 18. Qg3 a forced win?, is not >>18. Nf6 a forced win? The examples of GM's missing the shortest winning line are >>to numerous to count, but it has never taken away one rating point from their >>Elo rating. >> > Your point is valid except it is no better than an "A" class human chess > player playing white in the actual game. I think if (from time to time ) I >can find the holes in current supposed 2500 rated programs what will a GM do? I have to disagree, there is no hole in the move Qg3 it is also a forced win. If Qg3 loses you would have a point. >I agree that more data is needed and this one game is just one small example >of current programs underestimating open KR files. This has nothing to do with understanding, but more to do with how some programs extend their search to see the mate in 10 (19 plys). The programs that do not see the mate in 10 and play Nf6 see better then +11 pawns advantage in the move Qg3. To me this is not a hole in how current programs understand open KR files. This is more a matter of style of search. This position is not hard to understand how to win, so a class A player should not have much trouble seeing the natural move Qg3. I would bet many GM's would play this move also instead of the move Nf6. Because the move wins with ease, and no need to calculate the move Nf6. I aplaud the efforts of ED >and others in their efforts to advance computer chess! >>> >>>The following is the position after move 17.....how long does it take your >>>program to find the mate? >>> >>>r1bq1rk1/1p3pp1/p1np4/3Np1P1/4P3/4Q3/PPP1B1P1/2K4R w - - 0 18 >>> >>>TP
This page took 0.01 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.