interesting King's Gambit game

[bob]

Rybka 3 thinks Black is ahead for quite a while, but certainly by 23. Nf5! White is winning. Black should probably have deviated much earlier: maybe with 18. .. d5, since exd5 is met with Nd3+.

[Event "?"]
[Site "chessclub.com"]
[Date "2009.09.15"]
[Round "?"]
[White "Data"]
[Black "Arasan 11.5"]
[Result "1-0"]
[ECO "C34"]
[WhiteElo "2870"]
[BlackElo "2729"]
[TimeControl "240+16"]

1. e4 e5 2. f4 exf4 3. Nf3 d6 4. d4 g5 5. h4 g4 6. Ng1 Qf6 7. Nc3 Ne7
8. Nge2 Bh6 9. Qd2 Nbc6 10. Nb5 Kd8 11. d5 Ne5 12. Qa5 b6 13. Qc3 c6
14. dxc6 N7xc6 15. Qa3 f3 16. Bg5 Bxg5 17. hxg5 Qe7 18. Nf4 Bd7
19. Nxd6 fxg2 20. Bxg2 Qxg5 21. Nd5 Rb8 22. Rd1 Qg7 23. Nf5 Bxf5
24. exf5 Nd7 25. O-O Rb7 26. Ne7 Nxe7 27. f6 Qf8
1-0 {ArasanX resigns}

"Data" has been cooking up this book for years. Each time he loses or draws, he manually tweaks the book. As black, don't play 1. ... e5 in important games and this becomes moot.

My database has e6 as the best move here. What does crafty play?
[d]rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - acd 23; acn 506885582; acs 2813; bm e6; ce -16; pm e6; pv e6 d4 d5 Nc3 Nf6 Bg5 dxe4 Nxe4 Be7 Bxf6 gxf6 Qd3 e5 Ne2 Nc6 O-O-O exd4 c3 f5 N4g3 Qd5 Nxd4 Nxd4 Qxd4 Qxd4 Rxd4 Be6;

The alternative evals record contains:
rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - pm c5;

So I guess that those are pretty well the two best moves for response to e4.

Statistically (from a database of high quality games) the most frequent responses were:
Na6 {3} Nc6 {283} Nf6 {636} a6 {1} b6 {74} c5 {38599} c6 {6652} d5 {720} d6 {1536} e5 {20915} e6 {9312} g5 {4} g6 {890}

Hard to say, but as a human, I will likely play 1. e5 if you play e4. I don't think there is a "best move" at this early stage of the game. Depends on the kind of game you want to play and the kind of positions you like/dislike.

I partialy agree here Bob because playing this,you have to know a hell of opening theory....one example:Ruy Lopez
Dr.D

Ruy lopez. Kings gambit. Guioco Piano. Scotch. Ponzoni. Goring Gambit and others. All depends on the kind of opening you like. I prefer a tactical slugfest when I am playing myself, and generally get it with 1. ... e5. If you play 1. c5 you can get into 2. c4 and such closed systems that are boring (to me).

[bob]

Rybka 3 thinks Black is ahead for quite a while, but certainly by 23. Nf5! White is winning. Black should probably have deviated much earlier: maybe with 18. .. d5, since exd5 is met with Nd3+.

[Event "?"]
[Site "chessclub.com"]
[Date "2009.09.15"]
[Round "?"]
[White "Data"]
[Black "Arasan 11.5"]
[Result "1-0"]
[ECO "C34"]
[WhiteElo "2870"]
[BlackElo "2729"]
[TimeControl "240+16"]

1. e4 e5 2. f4 exf4 3. Nf3 d6 4. d4 g5 5. h4 g4 6. Ng1 Qf6 7. Nc3 Ne7
8. Nge2 Bh6 9. Qd2 Nbc6 10. Nb5 Kd8 11. d5 Ne5 12. Qa5 b6 13. Qc3 c6
14. dxc6 N7xc6 15. Qa3 f3 16. Bg5 Bxg5 17. hxg5 Qe7 18. Nf4 Bd7
19. Nxd6 fxg2 20. Bxg2 Qxg5 21. Nd5 Rb8 22. Rd1 Qg7 23. Nf5 Bxf5
24. exf5 Nd7 25. O-O Rb7 26. Ne7 Nxe7 27. f6 Qf8
1-0 {ArasanX resigns}

"Data" has been cooking up this book for years. Each time he loses or draws, he manually tweaks the book. As black, don't play 1. ... e5 in important games and this becomes moot.

My database has e6 as the best move here. What does crafty play?
[d]rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq - acd 23; acn 506885582; acs 2813; bm e6; ce -16; pm e6; pv e6 d4 d5 Nc3 Nf6 Bg5 dxe4 Nxe4 Be7 Bxf6 gxf6 Qd3 e5 Ne2 Nc6 O-O-O exd4 c3 f5 N4g3 Qd5 Nxd4 Nxd4 Qxd4 Qxd4 Rxd4 Be6;

There are a ton of good choices. From e6 to d6 to d5 to Nf6 to c5 to c6 to you-name-it. e5 is also playable and crafty will play that on occasion although typically not in tournaments unless we have prepared something specific.

What makes e5 different from other possibilities in term of getting caught in a book trap (which is what I think you refer to)?
Fear of gambits?

Miguel

One example is the king's gambit. If white chooses to play f4, which I still think is a not very good opening, you enter a tree where the possibilities become limited, and deep, and book traps suddenly become common. This is what Data is doing. Crafty generally avoids kp openings from either side for this reason, with the main idea being to attempt to avoid the huge set of book traps the book authors have discovered.

[bob]

I think the interesting part of this game was post book. certainly for Arasan. Not sure about Data. It is a real test of king safety because both kings are exposed.

--Jon

Yes, but there is "exposed" and there is EXPOSED. The difference is in how the computer deals with them. As white, you can cook up lines that cause problems for everyone if you make the thing go deep enough. It is a waste of time if you can't be sure your opponent will play that opening, however.

[Dann Corbit]

Rybka 3 thinks Black is ahead for quite a while, but certainly by 23. Nf5! White is winning. Black should probably have deviated much earlier: maybe with 18. .. d5, since exd5 is met with Nd3+.

[Event "?"]
[Site "chessclub.com"]
[Date "2009.09.15"]
[Round "?"]
[White "Data"]
[Black "Arasan 11.5"]
[Result "1-0"]
[ECO "C34"]
[WhiteElo "2870"]
[BlackElo "2729"]
[TimeControl "240+16"]

1. e4 e5 2. f4 exf4 3. Nf3 d6 4. d4 g5 5. h4 g4 6. Ng1 Qf6 7. Nc3 Ne7
8. Nge2 Bh6 9. Qd2 Nbc6 10. Nb5 Kd8 11. d5 Ne5 12. Qa5 b6 13. Qc3 c6
14. dxc6 N7xc6 15. Qa3 f3 16. Bg5 Bxg5 17. hxg5 Qe7 18. Nf4 Bd7
19. Nxd6 fxg2 20. Bxg2 Qxg5 21. Nd5 Rb8 22. Rd1 Qg7 23. Nf5 Bxf5
24. exf5 Nd7 25. O-O Rb7 26. Ne7 Nxe7 27. f6 Qf8
1-0 {ArasanX resigns}

Here are the EPD records, commencing where the last opportunity to save the day was:

Qe7 was played, Qxg5 was much better:
[d]r1bk3r/p4p1p/1pnp1q2/1N2n1P1/4P1p1/Q4p2/PPP1N1P1/R3KB1R b KQ - acd 18; acn 193464978; acs 1054; bm Qxg5; ce 0; pv Qxg5 Qxd6+ Bd7 Rh5 Qxh5 Qf6+ Kc8 Nd6+ Kc7 Nf4 Nd3+ Bxd3 Qe5 Nd5+ Kb8 Nb4 Qg3+ Kd1 Nxb4 Qxh8+ Kc7 Qxa8 Qxd6 Qxa7+ Kd8 e5 Qd4 Qb8+ Bc8 gxf3 Nxd3;

r1bk3r/p3qp1p/1pnp4/1N2n1P1/4P1p1/Q4p2/PPP1N1P1/R3KB1R w KQ - acd 18; acn 99822718; acs 665; bm Nf4; ce 42; pv Nf4 d5 O-O-O Qxa3 Nxa3 d4 c3 g3 Rd2 fxg2 Bxg2 d3 Nxd3 Ke7 Nf4 Be6 Nd5+ Kf8 Nc7 Rd8 Nxe6+;

Bd7 was played, d5 is better:
[d]r1bk3r/p3qp1p/1pnp4/1N2n1P1/4PNp1/Q4p2/PPP3P1/R3KB1R b KQ - acd 17; acn 155932529; acs 936; bm d5; ce -59; pv d5 Qc3 Qc5 Qxc5 bxc5 Nxd5 Ba6 Nd6 Kd7 Rh6 Nd4 Rd1 Nxc2+ Kf2 Bxf1 Kxf1 Rab8 gxf3 gxf3 b3 Rbg8 Rc1 Rg6 Nb7 Rxh6 gxh6 Nd4 Nxc5+ Kd6 Nb7+ Ke6;

r2k3r/p2bqp1p/1pnp4/1N2n1P1/4PNp1/Q4p2/PPP3P1/R3KB1R w KQ - acd 17; acn 83323285; acs 409; bm Nxd6; ce 81; pv Nxd6 fxg2 Bxg2 Qxg5 Nd5 Ne7 Nxe7 Kxe7 Rd1 Kf8 Nc4+ Ke8 Nxb6 axb6 Qxa8+ Ke7 Qa3+ Kf6 Qg3 Kg7 O-O Re8;

fxg2 was played, Be6 is better:
[d]r2k3r/p2bqp1p/1pnN4/4n1P1/4PNp1/Q4p2/PPP3P1/R3KB1R b KQ - acd 18; acn 294921456; acs 1244; bm Be6; ce -79; pv Be6 O-O-O Kc7 Nb5+ Kb7 Qxe7+ Nxe7 gxf3 gxf3 Nd6+ Kc7 Nxe6+ fxe6;

r2k3r/p2bqp1p/1pnN4/4n1P1/4PNp1/Q7/PPP3p1/R3KB1R w KQ - acd 19; acn 277148189; acs 1426; bm Bxg2; ce 148; pv Bxg2 Qxg5 Qe3 Qg7 O-O-O h5 Nxh5 Qg6 Nf4 Rxh1 Nxg6 Rxd1+ Kxd1 fxg6 Nb5 Ke7 c3 Kf7 Qg3 Kg7 Kc2 a5 Nd4 Re8;

Qxg5 is probably optimal here:
[d]r2k3r/p2bqp1p/1pnN4/4n1P1/4PNp1/Q7/PPP3B1/R3K2R b KQ - acd 19; acn 295420451; acs 1437; bm Qxg5; ce -107; pv Qxg5 Nd5 Ne7 Nxe7 Kxe7 Rd1 Kf8 O-O Nf3+ Kf2 Qc5+ Qxc5 bxc5 Bxf3 gxf3 Nxf7 Kxf7 Rxd7+ Ke6 Rc7 Kd6 Rg7 Rhg8 Rxh7 Raf8 Ke3 Rg2 Rf2 Rxf2 Rh6+;

r2k3r/p2b1p1p/1pnN4/4n1q1/4PNp1/Q7/PPP3B1/R3K2R w KQ - acd 18; acn 206939171; acs 1069; bm Nd5; ce 120; pv Nd5 Ne7 Nxe7 Kxe7 Rd1 Kf8 O-O Nf3+ Kf2 Qc5+ Qxc5 bxc5 Bxf3 gxf3 Nxf7 Kxf7 Rxd7+ Ke6 Rc7 Kd6 Rg7 Ke6 Rh1 h5 Rg6+ Kf7 Rc6 h4 Rh3 Rab8;

Rb8 was played. Ne7 is better:
[d]r2k3r/p2b1p1p/1pnN4/3Nn1q1/4P1p1/Q7/PPP3B1/R3K2R b KQ - acd 18; acn 247317299; acs 1220; bm Ne7; ce -109; pv Ne7 Nxe7 Kxe7 Rd1 Kf8 O-O Nf3+ Kf2 Qc5+ Qxc5 bxc5 Bxf3 gxf3 Nxf7 Kxf7 Rxd7+ Ke6 Rc7 Kd6 Rg7 Rhg8 Rfg1 Rxg7 Rxg7 h5 Rg6+ Ke5 Kxf3 Rf8+;

White has you in his power by now...
[d]1r1k3r/p2b1p1p/1pnN4/3Nn1q1/4P1p1/Q7/PPP3B1/R3K2R w KQ - acd 17; acn 234823630; acs 1302; bm Rd1; ce 211; pv Rd1 f6 Nb5 Qg7 Nxa7 Na5 Qd6 Rb7 b3 h5 Nb5 Nac6 O-O h4 Rxf6 Rg8 Nxb6 Qe7 Qxe7+ Kxe7 Nd5+ Kd8 Nd6 Ra7 a4 Ne7 Nf5 Nxf5 exf5 h3 Be4;

[yanquis1972]

i really like the idea of making a serious attempt at resurrecting the king's gambit myself -- it sounds like, from what youre suggesting & despite yr claim that you still dont think its a very good opening, he may be doing that? wonder if he has any ammo that would be of use to top level human players.

[michiguel]

i really like the idea of making a serious attempt at resurrecting the king's gambit myself -- it sounds like, from what youre suggesting & despite yr claim that you still dont think its a very good opening, he may be doing that? wonder if he has any ammo that would be of use to top level human players.

I doubt it. Gaviota fell into a trap with Data that no human would. Data played a suboptimal move, but the best reply required to be found either 1) an amazing calculation beyond the horizon for blitz or 2) the move is in book. Rather than that, gaviota went to catch some material and lost miserably. Perhaps a stronger engine would have performed better, but objectively speaking, Data's move in the opening was not good. Now that Bob mentions this, the whole strategy makes sense. I was surprised when I saw it.

BTW, King's gambit is a sound opening, not a crazy one, but it does not have the teeth and the constant initiative of the Ruy Lopez at the highest level. Spassky and Larsen among others played it once in a while. If you play it often, I feel that black being well prepared will have good chances to equalize with no problem.

Among computers is a different story, IMHO.

Miguel

[Eelco de Groot]

I don't think 18... d5 would have impressed Data very much because of something like 19. Dxe7+ Kxe7 20.Pxd5+ or 19...Pxe7 20.exd5 both of which seem winning, but this is just my own analysis, not checked. Ancalagon thinks that Black would have been okay however if he plays 18... Ng6:

This is what Ancalagon thinks after 18... d5, three best moves:


[d]r1bk3r/p3qp1p/1pn5/1N1pn1P1/4PNp1/Q4p2/PPP3P1/R3KB1R w KQ -

Engine: Ancalagon 1.3 Weak Squares 180 Board Control middlegame 50 endgame 50
Build 298 (Athlon 2009 MHz, 256 MB) by Romstad, Costalba, Kiiski, de Groot

5 3:46 +0.35 19.Qxe7+ Nxe7 20.Nxd5 Nxd5 21.O-O-O Bb7
22.exd5 a6 23.Nc3 Kc7 24.Rd4 Rag8
25.gxf3 gxf3 26.Ne4 Rd8 27.Nf6 b5
28.Kd2 (110.104.365) 486

5 3:46 +0.33 19.Qc3 Qb4 20.O-O-O Qxc3 21.Rxd5+ Bd7
22.Nxc3 fxg2 23.Bxg2 a6 (110.104.365) 486

5 3:46 +0.29 19.Nxd5 Qxa3 20.Nxa3 fxg2 21.Bxg2 Bd7
22.Nb5 Ne7 23.Nd4 g3 24.Kd2 Nc4+
25.Kc3 (110.104.365) 486
______________________________________________________________

6 6:29 +0.39 19.Nxd5 Qxa3 20.Nxa3 Bd7 21.O-O-O g3
22.Nb5 f2 23.Be2 Ne7 (187.575.832) 481

6 6:29 +0.27 19.Qxe7+ Nxe7 20.Nxd5 fxg2 21.Bxg2 Bd7
22.a4 Rc8 23.Ne3 Rc5 24.b4 Rc8
25.Rd1 a6 26.Nd6 Rc7 27.Ndf5 N7g6 (187.575.832) 481

6 6:29 0.00 19.Qc3 Qb4 20.O-O-O Qxc3 21.Rxd5+ Bd7
22.Nxc3 fxg2 23.Nxg2 a6 24.Rd6 Kc7
25.Rdh6 g3 26.Nd5+ Kb7 27.Rxh7 (187.575.832) 481
______________________________________________________________

7 14:14 +0.62 19.Nxd5 Qxa3 20.Nxa3 Bd7 21.O-O-O g3
22.Nb5 f2 23.Be2 Rb8 24.Nd6 Be6
25.Nf5 Bxd5 26.Rxd5+ Kc7 27.Nxg3 (384.692.673) 450

7 14:14 +0.43 19.Qxe7+ Nxe7 20.Nxd5 fxg2 21.Bxg2 Bd7
22.Nd4 Rc8 23.O-O-O Nxd5 24.exd5 (384.692.673) 450

7 14:14 0.00 19.Qc3 Qb4 20.O-O-O Qxc3 21.Rxd5+ Bd7
22.Nxc3 fxg2 23.Nxg2 a6 24.Ne3 Kc7
25.Rd2 Ne7 26.Rh6 (384.692.673) 450
______________________________________________________________

8 52:17 +0.60 19.Nxd5 Qxa3 20.Nxa3 Bd7 21.O-O-O fxg2
22.Bxg2 Rb8 23.Nb5 g3 24.Nf6 Ke7
25.Nd5+ Kd8 (1.386.199.007) 441

8 52:17 +0.58 19.Qxe7+ Nxe7 20.Nxd5 fxg2 21.Bxg2 Bd7
22.Nd4 Rc8 23.O-O-O Nxd5 24.exd5 (1.386.199.007) 441

8 52:17 +0.19 19.Qc3 Qb4 20.O-O-O Qxc3 21.Rxd5+ Ke7
22.Nxc3 Ng6 23.Nxg6+ fxg6 24.gxf3 gxf3
25.Bb5 Bb7 26.Kd2 (1.386.199.007) 441
______________________________________________________________

9 170:49 +0.52 19.Qxe7+ Nxe7 (4.288.676.940) 418

9 170:49 +0.52 19.Nxd5 Qxa3 20.Nxa3 fxg2 21.Bxg2 Ba6
22.Kf2 Kc8 23.b4 Nd4 24.Rad1 Ne6
25.b5 Bb7 26.Rh5 Bxd5 27.exd5 Nf4
28.Rh6 (4.288.676.940) 418

9 170:49 +0.25 19.Qc3 Qb4 20.Qxb4 Nxb4 21.O-O-O fxg2
22.Bxg2 Bd7 23.a3 Bxb5 24.axb4 d4
25.Rxd4+ Bd7 26.Rd5 Re8 27.Rxh7 Kc7 (4.288.676.940) 418

best move: Qa3-c3 time: 194:15.688 min n/s: 415.890 nodes: 4.847.490.217

To Ancalagon 18... Ng6 seems a bit better and maybe just holding, Stockfish preferred 18... d5 and I'm not so sure d5 is enough to hold the position for Black, because the score is dropping further when you follow the main line.

r1bk3r/p3qp1p/1pnp4/1N2n1P1/4PNp1/Q4p2/PPP3P1/R3KB1R b KQ -

Engine: Stockfish 1.4 JA (64 MB)
by Tord Romstad, Marco Costalba, Joona Kiiski

16.01 3:29 -0.52 18...d5 19.Nxd5 Qxa3 20.Nxa3 g3
21.O-O-O fxg2 22.Bxg2 Bg4 23.Nxb6+ Bxd1
24.Rxd1+ Ke7 25.Nxa8 Rxa8 26.Rh1 Rg8
27.Rxh7 Rxg5 28.Kd2 Ke6 (116.167.817) 555

17.01 10:27 -0.56 18...d5 19.Nxd5 Qxa3 20.Nxa3 fxg2
21.Bxg2 Rb8 22.Nb5 g3 23.O-O-O Bg4
24.Rd2 Ke8 25.Nd6+ Kf8 26.Nf6 Rd8
27.Nf5 Rxd2 28.Kxd2 Nc4+ 29.Ke1 (350.879.231) 559

18.01 25:18 -0.58 18...d5 19.Nxd5 Qxa3 20.Nxa3 fxg2
21.Bxg2 Rb8 22.O-O-O Ke8 23.Nb5 Kf8
24.Rh6 Bd7 25.Nd6 Kg7 26.Nf6 Be6
27.Nf5+ Bxf5 28.exf5 Rbc8 29.Nh5+ Kg8 (847.445.827) 557

19.01 62:39 -0.66 18...d5 19.Nxd5 Qxa3 20.Nxa3 fxg2
21.Bxg2 Rb8 22.O-O-O Ke8 23.Nb5 Kf8
24.Rh6 Bd7 25.Nf6 Be6 26.a3 g3
27.Nc7 Ke7 28.Nxe6 fxe6 29.Nxh7 (2.095.526.451) 557

20.01 176:51 -0.68 18...d5 19.Nxd5 Qxa3 20.Nxa3 fxg2
21.Bxg2 Rb8 22.O-O-O Ke8 23.Nb5 Kf8
24.Rh6 Bd7 25.Nf6 Be6 26.Nc7 Rc8
27.Ncd5 Rd8 28.Rdh1 Kg7 29.Rxh7+ Rxh7
30.Rxh7+ Kg6 (5.971.798.442) 562

21.01 425:47 -0.58 18...d5 19.Nxd5 Qxa3 20.Nxa3 fxg2
21.Bxg2 Rb8 22.O-O-O Ke8 23.Nb5 Kf8
24.Rh6 Bd7 25.a3 Rc8 26.Ne3 Ne7
27.Nd6 Rd8 28.Ndf5 Nxf5 29.exf5 Kg7
30.f6+ Kg8 (14.503.758.092) 567


After 18...d5 19.Nxd5 Qxa3 20.Nxa3 *


[d]r1bk3r/p4p1p/1pn5/3Nn1P1/4P1p1/N4p2/PPP3P1/R3KB1R b KQ -

Engine: Stockfish 1.4 JA (64 MB)
by Tord Romstad, Marco Costalba, Joona Kiiski

16.01 1:02 -0.80 20...Bd7 21.O-O-O (39.567.162) 631

16.02 1:32 -0.60 20...Be6 21.Kf2 fxg2 22.Bxg2 Rb8
23.Nb5 Rb7 24.Rh6 a6 25.Nd6 Rd7
26.Nf5 Bxf5 27.exf5 Ne7 28.Rd1 Nxf5
29.Rxb6 (58.489.871) 631

17.01 3:39 -0.80 20...Be6 21.Kf2 (137.978.678) 628

17.03 5:08 -0.58 20...fxg2 21.Bxg2 Be6 22.Kf2 Rc8
23.Rad1 Ke8 24.Kg3 Kf8 25.Nb5 Rd8
26.c3 Kg7 27.Nd4 Nxd4 28.cxd4 Nc4
29.Rc1 Nxb2 (194.377.529) 630

18.01 8:31 -0.78 20...fxg2 21.Bxg2 (320.692.385) 627

19.01 19:05 -0.82 20...fxg2 21.Bxg2 Rb8 22.Kf2 Be6
23.Nb5 Rb7 24.Kg3 Rd7 25.Rh6 Kc8
26.Rah1 Rdd8 27.Rxh7 Rhg8 28.Nbc7 Bxd5
29.Nxd5 Kb7 30.R7h5 (719.505.607) 628

20.01 39:17 -0.70 20...fxg2 21.Bxg2 Rb8 22.Kf2 Be6
23.Nb5 Rb7 24.Kg3 Rd7 25.Rh6 Kc8
26.Rah1 Kb7 27.Rxh7 Rxh7 28.Rxh7 a6
29.Nbc3 Nd4 30.Nf4 Bc4 31.b3 (1.466.238.716) 621

21.01 81:37 -0.90 20...fxg2 21.Bxg2 Rb8 22.Kf2 Be6
23.Nb5 Rb7 24.Kg3 Rd7 25.Rh6 Kc8
26.Rah1 Kb7 27.Rxh7 Rxh7 28.Rxh7 a6
29.Nbc3 Nd4 30.Nf4 Rc7 31.Rh8 Bc4 (3.072.754.787) 627

22.01 179:11 -0.86 20...fxg2 21.Bxg2 Rb8 22.Kf2 Be6
23.Nb5 a6 24.Nbc7 Bxd5 25.Nxd5 Kc8
26.Kg3 Kb7 27.Nf6 Nd4 28.Nxg4 Nxg4
29.Kxg4 Nxc2 30.e5+ Ka7 31.Rac1 Ne3+
32.Kg3 Nxg2 33.Kxg2 Kb7 (6.779.027.832) 630

best move: f3xg2 time: 221:06.672 min n/s: 630.142 nodes: 8.359.890.186

[jdart]

That's more or less in line with my own analysis. But this game and others show you can't trust engine evals very far in the King's Gambit unless they're based on a really deep search.

Of course, also, from an opening perspective, Black doesn't even have to play 6... Qf6 but can choose Bh6 or f5.

--Jon

[BubbaTough]

I think Data is a great opponent for testing out learning algorithms. LearningLemming has had many great battles, and there are not that many other opponents out there to learn about the kings gambit from. My dream was that someday LearningLemming would gain enough respect for the gambit to venture it itself, but that is not looking like it will happen anytime soon.

-Sam

[bob]

i really like the idea of making a serious attempt at resurrecting the king's gambit myself -- it sounds like, from what youre suggesting & despite yr claim that you still dont think its a very good opening, he may be doing that? wonder if he has any ammo that would be of use to top level human players.

I doubt it. Gaviota fell into a trap with Data that no human would. Data played a suboptimal move, but the best reply required to be found either 1) an amazing calculation beyond the horizon for blitz or 2) the move is in book. Rather than that, gaviota went to catch some material and lost miserably. Perhaps a stronger engine would have performed better, but objectively speaking, Data's move in the opening was not good. Now that Bob mentions this, the whole strategy makes sense. I was surprised when I saw it.

BTW, King's gambit is a sound opening, not a crazy one, but it does not have the teeth and the constant initiative of the Ruy Lopez at the highest level. Spassky and Larsen among others played it once in a while. If you play it often, I feel that black being well prepared will have good chances to equalize with no problem.

Among computers is a different story, IMHO.

Miguel

That's the problem with this approach. The moves played by Data are not the best moves from a pure chess perspective, they are the moves that lead to positions where a program might be too greedy and eat itself into difficulty. You can watch the games here and there and see this as a common theme. Retreat this piece to defend this key square and all is well. But try to save that pawn, or grab that pawn, and trouble follows...