New Zappa settings, dead Stockfish!

[Robert Flesher]

Now try these settings in Zappa against Rybka 4. I saw the horrible way it played against Father who showed how ridiculous its playing strength really is. Maybe these new setting is the way to defeat Rybka 4.

No! but they are showing that the top engines can be crushed on occasion with a well thought out attack!

[Robert Flesher]

[Event "Microsoft, Blitz:10'"]
[Site "Microsoft"]
[Date "2010.06.18"]
[Round "1"]
[White "Zappa Mexico II"]
[Black "spark-0.4"]
[Result "1-0"]
[ECO "D47"]
[Annotator "-0.12;0.00"]
[PlyCount "119"]
[TimeControl "600"]

{Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz 0 MHz W=14.0 ply; 3,671kN/s;
BookDF12.ctg B=18.9 ply; 11,940kN/s; BookDF12.ctg} 1. d4 {0} Nf6 {0} 2. c4 {0}
e6 {0} 3. Nf3 {0} d5 {0} 4. Nc3 {0} c6 {0} 5. e3 {0} Nbd7 {0} 6. Bd3 {0} dxc4 {
0} 7. Bxc4 {0} b5 {0} 8. Bd3 {0} Bb7 {0} 9. a3 {0} b4 {0} 10. Ne4 {0} Nxe4 {0}
11. Bxe4 {0} bxa3 {0} 12. O-O {0} Bd6 {0} 13. b3 {0} Nf6 {0} 14. Nd2 {0} Qc7 {0
} 15. Bf3 {0} Bxh2+ {0} 16. Kh1 {0} Bd6 {0} 17. Nc4 {0} Be7 {0} 18. Bxa3 {0}
O-O {0} 19. Bxe7 {0} Qxe7 {0} 20. Ra5 {Both last book move 0} Nd5 {0.00/18 10}
21. Kg1 {(Qa1) -0.12/15 36} f6 {0.02/19 27} 22. Qd2 {(Qc1) -0.09/16 27} Rfd8 {
(e5) 0.00/19 26} 23. Rfa1 {(Rc1) 0.18/17 29} a6 {0.12/20 12} 24. Rc5 {
0.30/17 37} Rab8 {0.13/20 14} 25. Na5 {0.30/17 46} Qd7 {(Ba8) 0.34/21 38} 26.
Qe2 {(Qc1) 0.40/15 18} Rdc8 {0.21/18 9} 27. Rac1 {(Nxb7) 0.40/15 16} h6 {
(Qe7) 0.20/17 8} 28. e4 {0.91/16 16} Ne7 {0.22/19 9} 29. Qc4 {(Bg4) 0.90/17 18}
Kf7 {(Kh7) 0.28/19 42} 30. Bh5+ {(Qd3) 1.28/15 16} g6 {(Kf8) 0.08/19 10} 31.
Bg4 {(Bf3) 1.36/15 17} Ba8 {(Rd8) 0.00/18 9} 32. Rd1 {(b4) 1.40/14 14} h5 {
(Rb5) 0.03/18 15} 33. Be2 {1.61/14 10} Rd8 {(Rh8) 0.34/18 35} 34. Qc1 {
(e5) 1.77/13 7} Kg7 {(Rf8) 0.00/18 8} 35. e5 {1.77/13 7} fxe5 {0.03/18 9} 36.
Nc4 {(dxe5) 1.79/13 8} exd4 {0.00/17 8} 37. Qg5 {(Rxh5) 1.52/13 9} Qe8 {
(Rf8) -0.74/15 5} 38. Qe5+ {(Ne5) 0.51/11 9} Kh7 {-0.79/17 6} 39. Nd2 {
(Nd6) 0.51/11 0} Rd5 {(Qh8) -1.09/19 13} 40. Rxd5 {1.09/8 0} cxd5 {
(exd5) -0.77/19 27} 41. Bd3 {(Nf3) 0.20/13 5} Nc6 {-0.47/20 10} 42. Qf6 {
0.20/14 4} e5 {-0.54/20 7} 43. Nf3 {0.20/15 5} e4 {-0.73/19 4} 44. Ng5+ {
0.20/15 0} Kg8 {(Kh6) -0.70/20 7} 45. Bxa6 {(Nxe4) 0.05/13 4} d3 {
(Ne5) -0.60/19 22} 46. f4 {(f3) 0.20/13 13} h4 {(exf3) -1.35/16 8} 47. Rc1 {
(f5) 0.20/13 7} d2 {-1.12/17 8} 48. Rf1 {0.20/14 4} Nd8 {(Ne7) -0.83/17 18} 49.
Qd6 {(f5) 0.20/13 9} Rxb3 {0.00/16 5} 50. f5 {0.20/13 0} e3 {(h3) 0.00/15 4}
51. fxg6 {1.82/13 7} d1=Q {1.26/16 12} 52. Rxd1 {1.82/12 0} e2 {1.06/16 3} 53.
Bxe2 {1.82/12 0} Re3 {(Qe3+) 1.16/16 3} 54. Nh7 {(Bd3) 1.18/8 0} Qe7 {5.26/16 5
} 55. Nf6+ {5.87/12 5} Kg7 {5.71/18 4} 56. Nh5+ {5.87/12 0} Kh6 {
(Kg8) 6.30/20 13} 57. Qf4+ {5.87/7 0} Kxg6 {7.76/19 12} 58. Rf1 {7.12/13 4}
Rxe2 {8.23/18 5} 59. Qg4+ {7.12/12 0} Kh6 {8.52/18 3} 60. Rf6+ {7.12/12 0} 1-0

[Swaminathan]

Great games, Robert.

You must have done good changes with respect to Attacks/Initiatives/Activity in direction of an attack/Sacrifices.

[LetoAtreides82]

Hello folks, I was bored last night so I thought I'd try to develope a new set of parameters to Zappa Mexico II. I want something that played like Tal on Angel Dust. So after some crazy setting and tweaking I created a super aggressive version of Zappa. It is most likely alot weaker than the default, HOWEVER, if you make a mistake, you die!
Here Stockfish is crushed in 27 moves! Let's call these setting Dissident Aggressor. Another point is that I analyzed the game with Rybka 4.0 and others and they are clueless of the dangers and do not see the coming attack. I may be onto something here !??

[Event "Microsoft, Blitz:10'"]
[Site "Microsoft"]
[Date "2010.06.18"]
[Round "1"]
[White "Zappa Mexico II"]
[Black "Stockfish 1.7.1 JA 64bit"]
[Result "1-0"]
[ECO "B85"]
[Annotator "0.77;0.52"]
[PlyCount "53"]
[TimeControl "600"]

1. e4 {0} c5 {0} 2. Nf3 {0} d6 {0} 3. d4 {0} cxd4 {0} 4. Nxd4 {0} Nf6 {0} 5.
Nc3 {0} a6 {0} 6. a4 {0} Nc6 {0} 7. Be2 {0} e6 {0} 8. Be3 {0} Be7 {0} 9. O-O {0
} O-O {0} 10. f4 {0} Qc7 {0} 11. Kh1 {0} Re8 {0} 12. Bf3 {0} Bf8 {0} 13. Qd2 {0
} Rb8 {0} 14. Rad1 {0} Nd7 {0} 15. Qf2 {Both last book move 0} Nb4 {0.52/19 14}
16. f5 {(Rd2) 0.77/17 43} e5 {0.52/21 13} 17. Nde2 {0.64/16 43} Nc5 {
(b5) 0.48/20 17} 18. Ng3 {1.12/16 87} b6 {(Bd7) 0.48/21 49} 19. Nh5 {
(Qe2) 1.30/15 25} Bd7 {(Kh8) 0.44/22 17} 20. Qh4 {(Rd2) 1.77/14 41} Rec8 {
(Bc6) 0.28/18 25} 21. Be2 {(Qg3) 2.86/12 20} h6 {(Ne6) 3.43/18 23} 22. Nxg7 {
(Nf6+) 5.90/13 23} Qd8 {5.81/17 10} 23. Qg3 {6.74/13 16} Kh8 {6.38/18 14} 24.
f6 {6.74/12 0} Nc6 {(Nxa4) 7.39/18 21} 25. Qh4 {(Bh5) 8.43/13 25} Kg8 {
12.44/18 11} 26. Bxh6 {(Qxh6) 11.35/12 0} Nd4 {(d5) 15.15/17 13} 27. Bc4 {
(Bh5) 12.15/11 0} 1-0

quite fun, although the attack did rather play itself, while Stockfish fiddled and did absolutely nothing.

Strategy-wise the playing strength of Stockfish demonstrated by this game consists of: play out opening according to stored book, then have absolutely no clue what to do and fiddle around. 3000 ELO? Tres drole. However, if you create rating lists out of similar monsters that equally have no idea what to do, then any rating is possible, I suppose.

Maybe what these engines need is opening-book heuristics, so that the engine plays differently depending on which opening it is playing. Human grandmasters understand that each opening has its own long-term objectives, I don't see why it is not possible to teach these objectives to chess engines.