From: mig@satlink.com.nospam (mig) Newsgroups: rec.games.chess.misc Date: Sat, 30 Aug 1997 06:04:33 GMT
Well, giving away a not-so-secret, there are very few PC-based machines out there today that successfully solve the Stonewall Attack when they have Black, especially in blitz time controls. I'm not talking a 233 Pentium running HIARCS 6 or Genius 5 at 2hr time controls here, I'm talking just wanting to whomp your PC in a few quick games.
This is basically worthless as far as improving your overall chess skill, I suppose, but it can be fun to exact some revenge. And it does teach you about sacrifice against a good defender.
The trick is in the move order, after some practice you'll be able to get to the desired position almost every time. The key is to arrive to a position, via the Stonewall Attack, in which you can take your time to move all your pieces over to the kingside BEHIND THE LINES, not giving the computer a chance to calculate your threats, as they're all too far away (tech term is "the horizon effect").
The basic desired position of the white pieces:
pawns on c3, d4, e3, f4 bishop on d3 knights on d2 and e5 castle kingside
Then, often using the queen to support them, begin to push the kingside pawns, forcing the black knight on f6 back. Then the rook can lift, the queen can penetrate and IT'S A FIESTA!! Lots of sacrifices are available! As I said, the key to this position is that BLACK HAS VERY LITTLE OBVIOUS ACTIVE COUNTERPLAY. Computers don't really understand the problem with this position, as there are no direct threats. When there are no short-term good moves for them, they start doing silly things. (Give them an hour or more and this can change, but not always.)
The big test for your computer program is whether or not it plays the HORRIBLE ... c4 at some point, hitting the d3 bishop (which just moves back to c2 and maintains its killer diagonal), but LOCKING THE POSITION UP AND REMOVING THE ONLY LEVER AGAINST THE WHITE PAWN STRUCTURE. After ... c4, the only way Black can get play is by playing ... b5, ... a5, ... b4, etc, but again, this is usually too long term for a PC program to see, especially at fast time controls. Simply playing a3 at some point usually removes this plan from the computers list completely.
If Black does the right thing and is smart enough to maintain the tension with the c5 pawn, things are harder for white, but usually still fine. Computers have a hard time figuring out how to generate long-term counterplay, and the d4 and f4 pawns create an effective bind.
Against a strong human player, the Stonewall Attack is just too one-dimensional and there are several positional antidotes to this system. But if you fool with the move orders against your PC, you can arrive to the above-mentioned dream position almost every time. (not allowing ... e5, for example) Doing this will teach you about the type of positions computers don't understand. Even a d-player would start to get concerned when they saw all the white firepower coming over to the kingside, but a computer doesn't see what it can't calculate and doesn't care if you have everything and the kitchen sink around its king if it doens't actually see a winning variation. They also miss "quiet" moves and sacrifices at faster controls.
Neither CM5000 nor Fritz4 can get past move 30 against me with black if I play this system. (P133, 48MB RAM (but I've done this to many programs on friend's machines, including Alexei Shirov's computer!)) I know Bob Hyatt (Crafty's programmer) has gone out of his way to find a solution to this problem, as have most programmers. Crafty with a huge book can usually avoid it, but that's just a huge book working, not the program.
In general it's just get into the dream position, push your kingside pawns a bit, get everybody over to the kingside, slow black up on the queenside if necessary (even if you have to give up a pawn or two just so you gain time to kill the king) and then SAC SAC SAC ON THE KINGSIDE! WHEEEE!
Well, I just played a few games against the CM5000 you have, and here are the results. (Two against Chessmaster, one against "Fischer", all games at 5 0 standard blitz time control) I should note that I'm 2200 FIDE, but I think a B-player could find all the moves I made in the following games!
(Some of Black's last moves look silly as CM realizes it's toast and starts to throw everything away to postpone mate.)
(This is the CM export format; it looks like a few changes need to be made to make it into pgn.
[Title "Mig vs Chessmaster 8-30-1997"] [White "Mig"] [Black "Chessmaster"] [Date "8-30-1997"] 1. d4 d5 2. e3 e6 3. Bd3 Nf6 4. Nd2 c5 5. c3 Nc6 6. f4 Bd7 7. Ngf3 c4 8. Bc2 Bd6 9. O-O O-O 10. Ne5 Rc8 11. g4 h6 12. Qf3 Qc7 13. Qg3 b5 14. a3 Rb8 15. h4 a6 16. g5 hxg5 17. hxg5 Nh5 18. Qh4 Bxe5 19. fxe5 g6 20. Rf6 b4 21. axb4 a5 22. bxa5 Rb5 23. a6 Ra5 24. Rb1 Rxa6 25. Nf1 Nxf6 26. gxf6 Qxe5 27. dxe5 Ne7 28. Qh6 Nf5 29. Bxf5 d4 30. Qg7++ ... ;Game over: Black is checkmated. ... [Title "Mig vs Chessmaster 8-30-1997"] [White "Mig"] [Black "Chessmaster"] [Date "8-30-1997"] 1. d4 d5 2. e3 e6 3. Bd3 Nf6 4. Nd2 c5 5. c3 Nc6 6. f4 Bd7 7. Ngf3 c4 8. Bc2 Bd6 9. O-O O-O 10. Ne5 Rc8 11. g4 h6 12. Qf3 Qc7 13. h4 b6 14. g5 hxg5 15. hxg5 Bxe5 16. fxe5 Nh7 17. Qh5 g6 18. Qg4 Ne7 19. Rf6 Nf5 20. Nf3 Nxf6 21. gxf6 Bc6 22. Bxf5 Ba4 23. Qg5 Qxe5 24. Nxe5 Kh7 25. Ng4 Kg8 26. Qh6 a6 27. Qg7++ ... ;Game over: Black is checkmated. ... [Title "Mig vs FischerR 8-30-1997"] [White "Mig"] [Black "FischerR"] [Date "8-30-1997"] 1. d4 Nf6 2. e3 d5 3. Bd3 e6 4. c3 Be7 5. f4 O-O 6. Nf3 Bd7 7. Nbd2 c5 8. O-O c4 9. Bc2 Nc6 10. Ne5 Rc8 11. g4 Ne8 12. Qf3 Nxe5 13. fxe5 Bg5 14. Qg3 Qb6 15. h4 Be7 16. Nf3 Nc7 17. g5 Bc6 18. Rf2 Qa5 19. Bd2 Rb8 20. Raf1 Qxa2 21. Bc1 Rfc8 22. Qg4 Be8 23. h5 b6 24. Nh4 Qa1 25. e4 Qa5 26. exd5 Qxd5 27. Be4 Qd8 28. Rf6 Qd7 29. g6 hxg6 30. hxg6 Bxf6 31. exf6 fxg6 32. Nxg6 Nd5 33. Ne7+ Qxe7 34. fxe7 Bf7 35. Bh6 g6 36. Bxg6 Kh8 37. Rxf7 Nf6 38. Rxf6 Rg8 39. Qh5 Rxg6+ 40. Qxg6 Rg8 41. Rf8 a6 42. Rxg8++ ... ;Game over: Black is checkmated. ...