The Glek Variation According to TCEC

In the marathon 64 game match between the two “engines” left standing to battle it out for the TCEC championship, Komodo 1333 and Stockfish 141214, both rated over 3200, the Glek variation of the Four Knights was the opening chosen by humans for the two titans in games 37 & 38. The first game began early enough that I was able to follow it live. I opened the CBDB (http://database.chessbase.com/js/apps/database/) and 365Chess (http://www.365chess.com/opening.php?m=8&n=1004&ms=e4.e5.Nf3.Nc6.Nc3.Nf6.g3&ns=3.5.5.6.47.57.1004) in order to check out which variation would be used. After 4 g3, 365Chess shows the database contains 99 games by GM Igor Glek, the man for whom the variation is named. Surely, I thought, the variation chosen by the TCECers would feature one of the variations promulgated by GM Glek.

The first surprise was 4…d5 since 4…Bc5 is played more often, but the former move is one of the standard moves. It would have been wonderful to see which move the “engine,” left to its own devices, would have played. 4 g3 signals the Glek variation and one would assume the humans would have forced the “engines” to begin the game by answering it with the move the “engine” playing Black considered best. We all know what happens when one makes an assumption…

The next moves through White’s 7th move are all standard, but Black’s 7…Be7 is not standard, as 7…Bc5, and 7…Bd6, have been played far more often, and with better results. GM Glek has faced 7…Bc5 seventeen times, and 7…Bd6 eleven times, while having faced 7…Be7 on only four occasions. Hummmm…

For the final “forced” move, the humans chose 8 0-0, and it has been the most played move by far, but has been outscored, by far, in limited action, by a move near and dear to my heart, Qe2! The last forced move was 8…0-0.

Stockfish 141214 (3218) vs Komodo 1333 (3210)
TCEC Season 7 – Superfinal 37
2014.12.23
C47
Four Knights: Glek, 4…d5

1. e4 e5 2. Nf3 Nf6 3. Nc3 Nc6 4. g3 d5 5. exd5 Nxd5 6. Bg2 Nxc3 7. bxc3 Be7 8. O-O O-O 9. Re1 Bf6 10. d3 Be6 11. Ba3 Re8 12. Nd2 Qd7 (12…Rb8 13. Qc1 Bg5 14. Bxc6 bxc6 15. Rxe5 Bh3 16. Rxe8+ Qxe8 17. Qd1 f5 18. Rb1 Rxb1 19. Nxb1 c5 20. c4 Qc6 21. f3 Qe6 22. Kh1 Qe3 23. Nc3 Qf2 0-1, Benoit Lepelletier 2480 vs David Marciano 2470, 1997 FRA-ch) 13. Ne4 Be7 14. Bxe7 Qxe7 15. Nd2 Qc5 16. c4 Rab8 17. Ne4 Qe7 18. a4 a5 19. c3 h6 20. Qf3 f5 21. Nd2 Rbd8 22. Qe2 Bf7 23. Bxc6 bxc6 24. Nb3 Rb8 25. Qd1 Rb6 26. Nxa5 Reb8 27. d4 e4 28. d5 cxd5 29. cxd5 Rd8 30. Nc4 Rxd5 31. Qe2 Rb7 32. a5 Ra7 33. Ne3 Rdxa5 34. Nxf5 Qf6 35. Rxa5 Rxa5 36. Nh4 Qxc3 37. Qxe4 Ra1 38. Rxa1 Qxa1+ 39. Kg2 Qf6 40. Qa8+ Kh7 41. Qe4+ g6 42. f4 c5 43. Nf3 Qb2+ 44. Kg1 c4 45. Qe7 Qa1+ 46. Kf2 Qa2+ 47. Ke3 Qb3+ 48. Kf2 c3 49. Ne5 Qa2+ 50. Ke3 Qd2+ 51. Ke4 Qe2+ 52. Kd4 Qf2+ 53. Kxc3 Qe3+ 54. Kc2 Qe2+ 55. Kb1 Qd1+ 56. Kb2 Qd2+ 57. Kb1 Qe1+ 58. Kc2 1/2-1/2

Komodo 1333 (3210) vs Stockfish 141214 (3218)
TCEC Season 7 – Superfinal 38
2014.12.23

1. e4 e5 2. Nf3 Nf6 3. Nc3 Nc6 4. g3 d5 5. exd5 Nxd5 6. Bg2 Nxc3 7. bxc3 Be7 8. O-O O-O 9. Re1 Bf6 10. d3 Bg4 11. Rb1 Rb8 12. h3 Be6 13. c4 Re8 14. Bb2 Bf5 15. Nh2 Nd4 16. Bxd4 Qxd4 17. Ng4 Bxg4 18. Qxg4 b6 19. a4 Qc3 20. Qd1 g6 21. h4 h5 22. Bd5 Kg7 23. Re2 Qa3 24. Qe1 a5 25. Bc6 Re6 26. Rb3 Qa2 27. Bd7 Rd6 28. Bb5 Rbd8 29. c5 bxc5 30. Rb1 Rb8 31. Rb3 Rbd8 32. Qxa5 e4 33. Rxe4 Qxc2 34. Bc4 Rd4 35. Qxc7 R8d7 36. Qc6 Rxe4 37. Qxe4 Re7 38. Qf3 Bd4 39. a5 Qd2 40. a6 Bxf2+ 41. Qxf2 Re1+ 42. Kg2 Re2 43. Qxe2 Qxe2+ 44. Kg1 Qe1+ 45. Kg2 Qe2+ 46. Kh3 Qd1 47. Rb2 Qa1 48. Ra2 Qh1+ 1/2-1/2

From the comments left in the “chat” window it was obvious the fans did not care for the choice of opening because some spiced their comments with profanity. How are these eight moves chosen, and who makes the choice? If the Glek variation is chosen, why not stop the forced moves as soon as it becomes a Glek variation when White plays 4 g3? What is the point of forcing the top chess playing things in the universe to play additional moves they may, or may not, play on their on volition?

Here is a recent game played by GM Igor Glek:

Igor Glek, (2438) vs Rustam Kasimdzhanov (2700)
FIDE World Rapid 2014 06/17/2014

1. e4 e5 2. Nf3 Nc6 3. Nc3 Nf6 4. g3 d5 5. exd5 Nxd5 6. Bg2 Nxc3 7. bxc3 Be7 8. O-O O-O 9. Re1 Bf6 10. Rb1 Re8 11. h3 g6 12. Nh2 h5 13. d3 e4 14. d4 Qd5 15. Bf4 Qxa2 16. Nf1 Qd5 17. Nd2 Kg7 18. Nxe4 Bxh3 19. Bxh3 Rxe4 20. Rxe4 Qxe4 21. Bg2 Qf5 22. Rxb7 Ne7 23. Rxc7 Rd8 24. Rxa7 Nd5 25. Bd2 Rc8 26. Ra5 Ne3 27. Qa1 Nxc2 28. Rxf5 Nxa1 29. Rb5 Nc2 30. Rb3 Ra8 31. Bxa8 1-0

Here is a game that began as a Paulson Vienna before transposing, played by one of my favorite female players, Melanie Ohme (OhMy!):

Melanie Ohme (2315) vs Karina Szczepkowska Horowska (2376)
GER-POL w Match 2012 07/21/2012

1. e4 e5 2. Nc3 Nf6 3. g3 d5 4. exd5 Nxd5 5. Bg2 Nxc3 6. bxc3 Be7 7. Nf3 Nc6 8. O-O O-O 9. Re1 Bf6 10. d3 Rb8 11. Nd2 Re8 12. Rb1 Bd7 13. Ne4 Be7 14. Be3 b6 15. d4 Qc8 16. Qd3 h6 17. Rbd1 exd4 18. cxd4 Nb4 19. Qd2 Bf5 20. Bf4 Qd7 21. c4 Rbd8 22. Qb2 Bg4 23. Rd2 Nc6 24. d5 Na5 25. Rc2 f5 26. Nd2 Bf6 27. Qc1 Rxe1+ 28. Qxe1 Re8 29. Qc1 c5 30. h3 Bh5 31. Nb3 Nb7 32. Be3 Nd6 33. Qd2 a5 34. Qd3 Qe7 35. Nc1 g5 36. a4 Kg7 37. Kh2 f4 38. gxf4 Bg6 39. Qd2 Bxc2 40. Qxc2 gxf4 41. Bxf4 Be5 42. Nd3 Bf6 43. Qd1 Nxc4 44. Qg4+ Kh8 45. Qg6 Ne5 46. Qxh6+ Kg8 47. d6 Qg7 48. Qxg7+ Kxg7 49. Nxe5 Rxe5 50. Bxe5 Bxe5+ 51. Kg1 Bxd6 52. Kf1 Kf6 53. Ke2 Bf4 54. Kd3 Ke5 55. Kc4 Bd2 56. Bd5 Be1 57. f3 Bh4 58. Be4 Bd8 59. Bd5 1/2-1/2

Timofey Galinsky (2424) vs Denis Shilin (2424)
UKR-ch 2000

1. e4 Nc6 2. Nc3 e5 3. Nf3 Nf6 4. g3 d5 5. exd5 Nxd5 6. Bg2 Nxc3 7. bxc3 Be7 8. Qe2 O-O 9. O-O Bf6 10. d3 Re8 11. Nd2 e4 12. d4 Bf5 13. Nc4 Qd7 14. Ne3 Bh3 15. Bxh3 Qxh3 16. Nd5 Bxd4 17. Nf4 Qc8 18. cxd4 Nxd4 19. Qh5 Re5 20. Qh3 Nxc2 21. Qxc8+ Rxc8 22. Bb2 Ra5 23. Rac1 Na3 24. Rfd1 b5 25. Rxc7 Rb8 26. Rdd7 1-0

This is the oldest game found, and it makes me wonder why the variation is not called the “Nimzowitsch variation.” Could it be that there are so many other variations named after Nimzo that it would be too confusing to have another one? Or is it a variation is not named after a player who loses the initial game?

Aaron Nimzowitsch vs Ernst Gruenfeld
Karlsbad 1923

1. e4 e5 2. Nf3 Nc6 3. Nc3 Nf6 4. g3 d5 5. exd5 Nxd5 6. Bg2 Nxc3 7. bxc3 Bd6 8.
O-O O-O 9. d3 Bg4 10. h3 Bd7 11. Rb1 Rb8 12. Re1 Re8 13. Ng5 h6 14. Nxf7 Kxf7
15. Qh5+ Kg8 16. Bxh6 Qf6 17. Bg5 Qf7 18. Qh4 Ne7 19. Rxb7 Rxb7 20. Bxb7 Qxa2
21. Bxe7 Rxe7 22. Be4 Qe6 23. Qh7+ Kf8 24. Qh8+ Qg8 25. Qh5 Be8 26. Qg5 Qe6 27.
Ra1 c6 28. Kg2 Qh6 29. Qg4 Qd2 30. Qh4 Qh6 31. Qg4 Rf7 32. Qe2 Bc5 33. Bf3 Bd7
34. g4 Qf4 35. Ra5 Bb6 36. Rxe5 Bc7 37. Re4 Qh2+ 38. Kf1 Qxh3+ 39. Bg2 Qh6 40.
Qe1 Bg3 41. Re2 Qh4 42. c4 Kg8 43. g5 Qxg5 44. Kg1 Bd6 45. d4 Bh3 0-1

TCEC Rules

The Season 7 Superfinal of the TCEC ahampionship is underway with Komodo 1333 playing Stockfish 141214. I was amazed upon learning game 3 began as a French: Chigorin, 2…c5. In this 64 game match the same opening is being played by both “engines.” This game began during the evening of December 17, 2014 with the same opening played the following game, with colors reversed, and I spent the night watching the games, with both being drawn.
The TCEC people force the “engines” to play 8 moves. When White plays 2 Qe2, after 1 e4 e6, it signifies the Chigorin. Black should be allowed to play any move it computes best in lieu of being forced to play a move it may, or may not, consider best. When Black, on move two, after 1 e4 e5 2 Nf3, plays Nf6 that signifies the Petroff defense. I fail to understand why the TCECers force the programs to play another SIX moves when it is obvious White should choose the third move. Another example is the Najdorf defense. After 1e4 c5 2 Nf3 d6 3 d4 cxd4 4 Nxd4 Nf6 5 Nc3 Black has a variety of choices. Playing 5…a6 signals the Najdorf. At this point White has a plethora of choices. The “engine” should make the choice. After 1 e4 e5 2 f4 White has chosen the King’s Gambit, and Black has many options. It is unfortunate but we will never know what an “engine” would play because humans have made the choice for the machines. Near the end of the line for the game of Checkers, with the advent of the Checker playing program Chinook (http://webdocs.cs.ualberta.ca/~chinook/) the openings would be chosen for the human players in a tournament because so many openings were known to be a forced draw.
One would think that the humans in control of the TCEC tournament would at least choose an opening played numerous times by top GMs, but such is not the case. Take the aforementioned Chigorin game as an example. After 1 e4 e6 2 Qe2 White has, according to the Chessbase Database (http://database.chessbase.com/js/apps/database/), just played the top scoring move against the French, with White scoring 57%. The move Black makes in reply, 2…c5, is the most played move in the variation and can be considered the main line. On its third move the “engine” is forced to play 3 g3 when the most often played move, considered to be the “main line,” is 3 Nf3. According to the CBDB, Stockfish 181114 would play the little played 3 b3. I would rather see SF play this move, since it is the move determined best. Houdini would play another lesser played move, 3 d3. That is the move it should be allowed to play.
Black answers with the most often played move, 3…Nc6, the move considered best by Komodo. Stockfish 5×64 at depth 31 also plays 3…Nc6, but at depth 35 changes to 3…g6. It would be interesting to see a game between these two titans continue from this point.
The next move, 4 Nf3 is the main line, but Houdini 3×64 would play the little played 4 d3. This is followed by 4…Be7, considered best by both SF and Komodo, yet 4…g6 and d6 have been played more often.
5 Bg2 is for choice, as is the reply 5…d5. 6 d3 is the main line, but Houdini 4×64 would play 6 0-0. The next move, 6…Nf6 is almost automatic. The program is forced to play 7 0-0, the main line, but Komodo would play 7 e5, so maybe it is best?! 7…0-0 would seem to be a no-brainer, and it has been throughout chess history, and it is the first choice of both Stockfish and Houdini, but the humans in charge forced the “engines” to play the much lesser played 7…b6. Now White plays 8 e5 and Black retreats his Knight to d7 and the forced moves have ended and the programs can begin to “compute.”
The TCECers could have allowed the “engines” to compute beginning on move two, after White plays 2 Qe2, or they could have chosen the same opening moves chosen previously by the best human players to have played the opening. Instead the knuckleheads at TCEC forced the machines to play an obscure, little played line.
The first move produced by a program, 9 c4, is not only “main line,” but also seen as best by the “engines.” Stockfish played 9…0-0. Komodo would play the most often played move, 9…d4.
Stockfish played 10 Re1 (Komodo would play a TN, 10 a3) and Komodo responded with 10…h6, which is a TN. Stockfish 151214 shows 10…Ndb8 at depth 24, but give it more time to crunch and it produces the move chosen by Komodo.
The move 10…h6 is a deviation from “known theory.” A game was played with the move 10…Bb7 in lieu of 10…h6. Who were the players in this earlier game? Were they well-known GMs who devoted their lives to the Royal game? Not hardly…They were guys of my calibre:

Samuel Minor (2092) vs Matthias Schoene (1846)
Frankfurt-ch 05/15/2006

1. e4 e6 2. d3 d5 3. Qe2 Nf6 4. Nf3 c5 5. g3 Be7 6. Bg2 O-O 7. O-O Nc6 8. e5 Nd7 9. c4 b6 10. Re1 Bb7 11. h4 Re8 12. Nbd2 Nf8 13. Nf1 Qc7 14. N1h2 Nd4 15. Nxd4 cxd4 16. cxd5 Bxd5 17. Bxd5 exd5 18. Qf3 Bb4 19. Re2 Rad8 20. Ng4 Ng6 21. h5 Nxe5 22. Nxe5 Rxe5 23. Bf4 f6 24. Bxe5 fxe5 25. a3 Be7 26. Rae1 Bf6 27. h6 Qd7 28. Rc2 e4 29. dxe4 dxe4 30. Qxe4 d3 31. Rd2 Bg5 32. f4 Bxh6 33. Qe6+ Qxe6 34. Rxe6 Kf7 35. Re3 g5 36. fxg5 Bxg5 37. Rexd3 Bxd2 38. Rxd8 Bc1 39. Rd7+ Kg6 40. Rxa7 Bxb2 41. a4 h5 42. Kg2 Kg5 43. Kh3 Bc3 44. Rd7 Bb4 45. Rd5+ Kg6 46. Rb5 Ba5 47. Kh4 1-0

For the record here are the two games played by the computer programs:

Stockfish 141214 (3218) vs Komodo 1333 (3210)
TCEC Season 7 – Superfinal game 3
C00 French: Chigorin, 2…c5

1. e4 e6 2. Qe2 c5 3. g3 Nc6 4. Nf3 Be7 5. Bg2 d5 6. d3 Nf6 7. O-O b6 8. e5 Nd7 9. c4 O-O 10. Re1 h6 11. Bf4 d4 12. h4 Bb7 13. Nh2 Rb8 14. Ng4 h5 15. Nh2 g6 16. Nd2 a6 17. Nhf3 b5 18. Rab1 Qc7 19. cxb5 axb5 20. a3 Rbc8 21. Rbd1 Ba8 22. Ne4 Rfe8 23. b3 Qa7 24. Rb1 Qc7 25. b4 cxb4 26. axb4 Bxb4 27. Rf1 Qb8 28. Nf6+ Nxf6 29. exf6 e5 30. Bh6 Bc3 31. Bh3 e4 32. dxe4 Rcd8 33. Qxb5 Qxb5 34. Rxb5 Na5 35. Rc1 Bxe4 36. Nxd4 Bxd4 37. Rxa5 Bb2 38. Rf1 Bxf6 39. Bg2 Bf5 40. Bf3 Bc3 41. Ra2 Rd3 42. Be2 Rd4 43. Rc1 Rde4 1/2-1/2

Komodo 1333 (3210) vs Stockfish 141214 (3218)
TCEC Season 7 – Superfinal game 4
C00 French: Chigorin, 2…c5

1. e4 e6 2. Qe2 c5 3. g3 Nc6 4. Nf3 Be7 5. Bg2 d5 6. d3 Nf6 7. O-O b6 8. e5 Nd7 9. c4 O-O 10. Re1 h6 11. Bf4 d4 12. h4 Bb7 13. Nh2 Re8 14. Qg4 Bf8 15. Bxc6 Bxc6 16. Bxh6 f5 17. Qg6 Qe7 18. Bf4 Qf7 19. Qxf7+ Kxf7 20. Nd2 b5 21. Nhf3 Be7 22. cxb5 Bxb5 23. Nc4 Bc6 24. Ng5+ Bxg5 25. Bxg5 Reb8 26. Na5 Bb5 27. Rad1 Rb6 28. b3 Rab8 29. Re2 Ra6 30. Nc4 Bxc4 31. bxc4 Rab6 32. f3 Rb2 33. Rde1 Rxe2 34. Rxe2 Rb1+ 35. Kf2 Nb8 36. Re1 Rb2+ 37. Re2 Rb1 38. g4 fxg4 39. fxg4 Nc6 40. h5 a5 41. Kf3 Rd1 42. Bd2 Ra1 43. Be1 a4 44. g5 a3 45. Kf4 Rb1 46. g6+ Ke8 47. Ke4 Rb2 48. Kf3 Nb4 49. Bxb4 cxb4 50. h6 gxh6 51. c5 b3 52. c6 Rxe2 53. Kxe2 bxa2 54. c7 Kd7 55. c8=Q+ Kxc8 56. g7 a1=Q 57. g8=Q+ Kb7 58. Qf7+ Ka6 59. Qxe6+ Ka5 60. Qd5+ Ka4 61. Qc6+ Kb4 62. Qc4+ Ka5 63. Qc5+ Ka4 64. Qc4+ 1/2-1/2

If any of this is logical to you, please leave a comment and elucidate the AW because none of this makes any sense to me whatsoever and I am certain Mr. Spock would question the logic behind the moves.