Komodo Dragon vs IM Andras Toth knight odds match.

[Chessqueen]

Even Komodo 13.3 using my mere intel i7 3.40 GHz could have won this game without the queen slip versus an engine that is close in strength to IM Andras Toth Dragon 4.6

"mere"...the thing that drives the chess engine should not matter of course...the engine 'should speak for itself'.
So, your results really should be every bit as legit as someone's Threadripper.

I wrote the word mere in comparison to what Mr. Kaufman used against IM Andras which is probably 7 or 12 times faster, but of course I used the entire 15'+10' whereas, Komodo Dragon2 MCTS used half of the time.

Note: What Komodo Dragon2 just accomplished by beating an IM at standard Rapid time control, I thought that was not going to be possible until the next version of Komodo Dragon3, I know that the opening helped a lot and in most of the 1. f4 opening IM Andras was not too comfortable since the engine usually got a very attacking position where the extra Knight of Mr. Andras did not matter too much. I am hoping that the next version of Komodo Dragon3 it could probably give GM Nakamura a Knight Odds in TC of 10'+5" since for his level to expect Komodo Dragon3 to beat GM Nakamura at 15'+ 10" would be too much. Probably Mr. Kaufman can explain how IM Andras managed to beat Komodo Dragon2 today, was it because of a bad opening choice ? I am still trying to go over the first game and find out, but do not fully understand how IM Andras positionally squeeze the win

I think that he won game four primarily because he was familiar with how to play the Closed Sicilian as Black, so he didn't have to think much in the opening. He was probably much less familiar with how to play against 1.f4. In our blitz games with GM Alex Lenderman, Dragon won every game in which it played 1.f4, so that's how I knew it was a good choice against Andras. We had a lot more trouble in the games with g1 knight missing. 1.e4 is then a good move (1.f4 is silly without Nf3 option), but it's too normal for the human to play against.

That is true IM Andras is too familiar with Closed Sicilian since I checked all his previous games on chess.com and he played it a lot all his moves up to move 9. 0-0 0-0 were rock solid which he made instantly and I just let these two engines to finished it, here is the result

[pgn][Event "Live Chess - Chess"]
[Site "Chess.com"]
[Date "2021.06.20"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Dragon 4.6"]
[Result "0-1"]
[BlackElo "2417"]
[WhiteElo "3400"]
[TimeControl "900+10"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "149"]
[WhiteType "program"]
[BlackType "program"]

1. e4 {[%clk 0:15:10]} c5 {[%clk 0:15:08]} 2. Nc3 {[%clk 0:15:10.3]} Nc6
{[%clk 0:15:16.1]} 3. g3 {[%clk 0:15:20.2]} g6 {[%clk 0:15:17.8]} 4. Bg2
{[%clk 0:15:30.1]} Bg7 {[%clk 0:15:27.7]} 5. d3 {[%clk 0:15:24.4]} d6
{[%clk 0:15:35.4]} 6. f4 {[%clk 0:15:15.8]} e6 {[%clk 0:15:42.6]} 7. Qe2
{[%clk 0:15:08.1]} Nge7 {[%clk 0:15:47.6]} 8. Qf2 {[%clk 0:14:56.6]} Nd4
{[%clk 0:15:48.2]} 9. O-O {[%clk 0:14:50.6]} O-O {[%clk 0:15:55.6]} 10. Nd1
Bd7 11. c3 Ndc6 12. Ne3 Qc7 13. Bd2 Rad8 14. Rae1 b5 15. g4 f6 16. Qg3 Qa5
17. a3 Qa4 18. Bf3 Qb3 19. Qg2 Bh6 20. Bd1 Qxb2 21. Bc2 c4 22. a4 cxd3 23.
Bxd3 a6 24. Nc2 e5 25. axb5 axb5 26. Ra1 Ra8 27. Rab1 Qa2 28. g5 fxg5 29.
Ra1 Qb3 30. Rab1 Qe6 31. fxg5 Bg7 32. Rxb5 Rxf1+ 33. Qxf1 Qg4+ 34. Qg2 Qd1+
35. Ne1 Be6 36. Rb1 Qh5 37. Be2 Qh4 38. Nd3 Bh3 39. Qf3 Ra2 40. Qe3 d5 41.
Nf2 Be6 42. Bd1 d4 43. cxd4 exd4 44. Qf4 Qxf4 45. Bxf4 Be5 46. Nd3 Bxf4 47.
Nxf4 Bf7 48. h4 Ra3 49. h5 gxh5 50. Nxh5 Bxh5 51. Bxh5 Re3 52. Rb7 Rxe4 53.
Bf3 Re5 54. Kf2 Rxg5 55. Be4 Re5 56. Bc2 h5 57. Ba4 Ra5 58. Bb3+ Kf8 59.
Rd7 h4 60. Rd6 Rf5+ 61. Ke2 Rf4 62. Kd2 h3 63. Rh6 Rf2+ 64. Ke1 Rf3 65. Be6
Re3+ 66. Kd2 Kg7 67. Rxh3 Rxe6 68. Kd3 Nf5 69. Kd2 Re3 70. Rh5 Ng3 71. Rd5
Ne4+ 72. Kc1 Re2 73. Rd7+ Kf6 74. Rd5 Ke7 75. Rh5 {White resigns} *[/pgn]

[Chessqueen]

It is very easy it, happened to me 19 months ago, and I discontinued using it, but since Mr. Kaufman is a gentleman and he realized what happened, he should start that game from the point before the mouse slipped after this game

[pgn]
[Event "Live Chess - Chess"]
[Site "Chess.com"]
[Date "2021.06.19"]
[Round "?"]
[White "PlayKomodo"]
[Black "ChesscoachAndras"]
[Result "?"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[TimeControl "900+10"]
[WhiteElo "3400"]
[BlackElo "2538"]
[Termination "PlayKomodo won by resignation"]


1. e4 {[%clk 0:15:10]} 1... c6 {[%clk 0:14:41.6]} 2. d4 {[%clk 0:15:19.9]} 2...
d5 {[%clk 0:14:51.5]} 3. e5 {[%clk 0:15:29.8]} 3... Bf5 {[%clk 0:15:00.1]} 4.
Nd2 {[%clk 0:15:39.7]} 4... e6 {[%clk 0:15:03.4]} 5. Nb3 {[%clk 0:15:49.6]} 5...
Nd7 {[%clk 0:15:10.4]} 6. h4 {[%clk 0:15:42.8]} 6... h5 {[%clk 0:15:10.1]} 7.
Be2 {[%clk 0:15:51.2]} 7... Bg6 {[%clk 0:15:15.7]} 8. a4 {[%clk 0:15:39.3]} 8...
Ne7 {[%clk 0:15:14.3]} 9. a5 {[%clk 0:15:27.2]} 9... Nf5 {[%clk 0:14:42.9]} 10.
Bf4 {[%clk 0:15:37.1]} 10... a6 {[%clk 0:14:30.1]} 11. g3 {[%clk 0:15:29.4]}
11... c5 {[%clk 0:14:36.9]} 12. c3 {[%clk 0:15:25.2]} 12... cxd4 {[%clk
0:14:27.4]} 13. Nxd4 {[%clk 0:15:22.2]} 13... Nxd4 {[%clk 0:14:34.1]} 14. cxd4
{[%clk 0:15:23.2]} 14... Rc8 {[%clk 0:14:32.9]} 15. Qb3 {[%clk 0:15:21.3]} 15...
Rc7 {[%clk 0:14:07.3]} 16. O-O {[%clk 0:15:20.7]} 16... Be7 {[%clk 0:13:50.7]}
17. Rfc1 {[%clk 0:15:19.7]} 17... O-O {[%clk 0:13:43.3]} 18. Rxc7 {[%clk
0:15:29.6]} 18... Qxc7 {[%clk 0:13:52.4]} 19. Rc1 {[%clk 0:15:35.2]} 19... Qb8
{[%clk 0:13:49.4]} 20. Qc3 {[%clk 0:15:34]} 20... Bd8 {[%clk 0:12:18.8]} 21. Bd2
{[%clk 0:15:24.9]} 21... f6 {[%clk 0:10:30.4]} 22. f4 {[%clk 0:15:19.5]} 22...
fxe5 {[%clk 0:09:00.1]} 23. dxe5[/pgn]

If it was my own game I could do a takeback easily, but as it is I don't know how to do that, and anyway the game was over before I understood what happened. The third game looked bad for a long time, but finally he overlooked something and Komodo won again. Surprising result today.

In that case offer to play 4 games tomorrow and start with 2 wins for Komodo Dragon cancelling the 2nd game altogether

Playing 4 games plus a possible tiebreak tomorrow is unreasonable for Toth since he's starting at 11 pm his time, but I offered to call Game 2 a draw. This won't help him win the match by Armageddon scoring, but at least he'll earn a few $ and still have a decent chance to draw or even win the match by normal scoring, which would be something.

Komodo was active attacking and aggressive, What Contempt did you use this time? IM Andras will be very lucky to get at least 2 draws tomorrow.

I used 150. I think that the attacking play had more to do with the choice of the first move (1.f4 in games 1 and 3) than with the Contempt setting.

It seems that the choice of opening with 1. f4 surprise IM Andras as Morphy after losing 2 straight games decided to play it and surprised his Opponents well. For Engines Vs human IM and GM the 1. f4 is very good for two reasons most IM and GM usually know the best response for either 1. e4 or 1. d4 and knows those opening by heart.


And the dark square strategy for this game as well ==>

And the King Gambit of course ==>

Yes, 1.f4 was considered to be the main alternative to 1.e4 at (b1) knight odds in the 1800s. I think that 1.e4 was usual only because almost everyone played 1...e5 since that is what they almost all played in standard chess; it is actually not very good due to 1...d5! So 1.f4 may actually be the best move with b1 missing.

Will you write a chess opening book for Knight Odds on how to play Vs engines, since even IM like Andras is having a hard time to play Vs Komodo Dragon 2 ==> https://www.chess.com/article/view/gm-l ... white-book

[Chessqueen]

Even Komodo 13.3 using my mere intel i7 3.40 GHz could have won this game without the queen slip versus an engine that is close in strength to IM Andras Toth Dragon 4.6

"mere"...the thing that drives the chess engine should not matter of course...the engine 'should speak for itself'.
So, your results really should be every bit as legit as someone's Threadripper.

I wrote the word mere in comparison to what Mr. Kaufman used against IM Andras which is probably 7 or 12 times faster, but of course I used the entire 15'+10' whereas, Komodo Dragon2 MCTS used half of the time.

Note: What Komodo Dragon2 just accomplished by beating an IM at standard Rapid time control, I thought that was not going to be possible until the next version of Komodo Dragon3, I know that the opening helped a lot and in most of the 1. f4 opening IM Andras was not too comfortable since the engine usually got a very attacking position where the extra Knight of Mr. Andras did not matter too much. I am hoping that the next version of Komodo Dragon3 it could probably give GM Nakamura a Knight Odds in TC of 10'+5" since for his level to expect Komodo Dragon3 to beat GM Nakamura at 15'+ 10" would be too much. Probably Mr. Kaufman can explain how IM Andras managed to beat Komodo Dragon2 today, was it because of a bad opening choice ? I am still trying to go over the first game and find out, but do not fully understand how IM Andras positionally squeeze the win. I wonder how LCO or Stockfish would have done versus IM Andras

[lkaufman]

Will you write a chess opening book for Knight Odds on how to play Vs engines, since even IM like Andras is having a hard time to play Vs Komodo Dragon 2 ==> https://www.chess.com/article/view/gm-l ... white-book

Right now I'm more interested in figuring out how the engine should play down a piece rather than how the human should play up a piece, which of course partly depends on what openings the human is familiar with in normal chess. Maybe after we've had many Dragon vs. GM knight odds matches it will be possible to determine what works best for the humans.

[Chessqueen]

Will you write a chess opening book for Knight Odds on how to play Vs engines, since even IM like Andras is having a hard time to play Vs Komodo Dragon 2 ==> https://www.chess.com/article/view/gm-l ... white-book

Right now I'm more interested in figuring out how the engine should play down a piece rather than how the human should play up a piece, which of course partly depends on what openings the human is familiar with in normal chess. Maybe after we've had many Dragon vs. GM knight odds matches it will be possible to determine what works best for the humans.

Or these three pawns Odds as White to a Human IM which ended up in a draw too

[d]r1bqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 0 1
These three pawns could be the same as a Knight odds after all,so the misconception that a Knight is worth more than three pawns in the beginning turn out that if you let the engine play the position, it is really only worth 3 pawns; and possibly Komodo Dragon2 could give those three pawns Odds as Black to an IM.

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.06.21"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Komodo-13.3-64bit-bmi2"]
[Result "1/2-1/2"]
[BlackElo "3400"]
[Time "13:36:52"]
[WhiteElo "3400"]
[TimeControl "5400+5"]
[SetUp "1"]
[FEN "rnbqkbnr/p2ppp1p/8/8/8/8/PPPPPPPP/R1BQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "144"]
[WhiteType "program"]
[BlackType "program"]

1. Nf3 Nf6 2. g3 Bg7 3. Bg2 Nc6 4. O-O d5 5. d3 O-O 6. c3 Rb8 7. b3 Re8 8.
Bg5 h6 9. Be3 e5 10. Qc1 Kh7 11. c4 Be6 12. Rd1 Rc8 13. Rb1 Qc7 14. Bc5 Kg8
15. h3 Ne7 16. Bb4 Nc6 17. Bc3 a5 18. Nh4 Bf8 19. Rf1 Bg7 20. Rd1 Qb6 21.
e3 Qa7 22. Ra1 Qd7 23. Kh2 Rb8 24. Qb2 Qd6 25. Rac1 Bd7 26. Qd2 a4 27. bxa4
Rb7 28. a5 Ra7 29. Qe1 Rc8 30. a4 dxc4 31. dxc4 Qc5 32. f4 Be6 33. e4 Re8
34. Rd3 Kh7 35. Nf3 Nd7 36. Bd2 Qe7 37. Qe3 Nf6 38. Nxe5 Nxe5 39. fxe5 Ng8
40. c5 Rb8 41. Rd6 Rb2 42. Qc3 Ra2 43. Ra1 Rxd2 44. Rxd2 Qc7 45. Rd5 Ne7
46. c6 Qxc6 47. Qc5 Qa8 48. Rf1 Ng6 49. Rf5 Rb7 50. Qf2 Kg8 51. Qd4 Qa7 52.
Qxa7 Rxa7 53. Rb5 Bxf5 54. exf5 Nxe5 55. Bb7 Kf8 56. a6 Bf6 57. Kg2 Nc4 58.
Kf3 Nd6 59. Kg4 Bg5 60. Rb6 Ke7 61. h4 Be3 62. Rb3 Bd2 63. Rb2 Bc3 64. Rb3
Ba5 65. Kh5 Nxf5 66. Rb5 Nxg3+ 67. Kxh6 Bd2+ 68. Kg7 Ke6 69. Bc8+ Kd6 70.
Bb7 Ke6 71. Bc8+ Kd6 72. Bb7 Ke6 {3-fold repetition} 1/2-1/2[/pgn]

[Chessqueen]

These three pawns could be the same as a Knight odds after all,so the misconception that a Knight is worth more than three pawns in the beginning turn out that if you let the engine play the position, it is really only worth 3 pawns; and possibly Komodo Dragon2 could give those three pawns Odds as Black to an IM.

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.06.21"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Komodo-13.3-64bit-bmi2"]
[Result "1/2-1/2"]
[BlackElo "3400"]
[Time "17:29:39"]
[WhiteElo "3400"]
[TimeControl "3600+5"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 0 1"]
[Termination "adjudication"]
[PlyCount "178"]
[WhiteType "program"]
[BlackType "program"]

1. Bb2 c6 2. d4 Nf6 3. Nf3 d5 4. Qb3 g6 5. e3 Bg7 6. Nbd2 O-O 7. Bd3 Bh3 8.
Rg1 Rc8 9. Rc1 Rc7 10. Qa3 b6 11. Rg3 Be6 12. Be2 Qc8 13. Rg1 Ne8 14. Ba6
Qb8 15. h4 h5 16. Bd3 Nd6 17. Ba1 a5 18. Nf1 Rfc8 19. Ng5 Bg4 20. f3 Bd7
21. Nxf7 Kxf7 22. Bxg6+ Kf8 23. Qd3 e6 24. Bxh5 Be8 25. Bxe8 Rxe8 26. h5
Qd8 27. Rg4 Bh6 28. Bb2 b5 29. Ke2 Nf5 30. Qa3+ Kf7 31. Qxa5 Qb8 32. Qd2
Rg8 33. Ke1 Qd8 34. Qh2 Rc8 35. Rxg8 Kxg8 36. Rc5 Kh7 37. Bc1 Qf6 38. f4
Bf8 39. Rc2 Be7 40. Kd1 Qf7 41. h6 b4 42. Rb2 c5 43. dxc5 Bxc5 44. Rc2 Qa7
45. Ng3 b3 46. Nxf5 Qa6 47. Ng3 bxc2+ 48. Qxc2+ Kxh6 49. Bb2 d4 50. Qe4 Qc6
51. Bxd4 Qxe4 52. Nxe4 Bxd4 53. exd4 Rc4 54. Ng5 Rxd4+ 55. Kc2 Kg6 56. Nxe6
Rd6 57. Ng5 Kf5 58. Nf3 Kxf4 59. Nd2 Ra6 60. Kb3 Ke3 61. Nc4+ Kd4 62. a4
Rc6 63. Na3 Rb6+ 64. Nb5+ Kc5 65. Kc2 Ra6 66. Kb3 Re6 67. Kc2 Re3 68. Kd2
Rh3 69. Ke2 Kb4 70. Nd4 Kc3 71. Nf5 Kc4 72. Kd2 Kd5 73. a5 Ke5 74. Ne7 Ke4
75. Nc6 Ra3 76. Kc2 Kd5 77. Ne7+ Kc4 78. Nc6 Kc5 79. Ne7 Re3 80. Nc8 Kd5
81. Kd2 Kd4 82. Nd6 Re5 83. a6 Ra5 84. Ke2 Rxa6 85. Nf5+ Ke5 86. Ng3 Kf4
87. Nf1 Ra3 88. Nd2 Re3+ 89. Kf2 Re8 1/2-1/2[/pgn]

[Chessqueen]

[quote=Chessqueen post_id=896578 time=1624322569 user_id=10732]
These three pawns could be the same as a Knight odds after all, so the misconception that a Knight is worth more than three pawns in the beginning turn out that if you let the engine play the position, it is really only worth 3 pawns; and possibly Komodo Dragon2 could give those three pawns Odds as Black or White to any IM. After performing these two tests, I believe that Komodo Dragon2 can give this position or the reverse as Black to any IM with time control of 15'+10" and boy will any human IM be frustrated trying to beat the engine.

[d]rnbqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 1 1

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.06.21"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Komodo-13.3-64bit-bmi2"]
[Result "*"]
[BlackElo "3400"]
[Time "17:29:39"]
[WhiteElo "3400"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "268"]
[WhiteType "program"]
[BlackType "program"]

1. Bb2 c6 2. d4 Nf6 3. Nf3 d5 4. Qb3 g6 5. e3 Bg7 6. Nbd2 O-O 7. Bd3 Bh3 8.
Rg1 Rc8 9. Rc1 Rc7 10. Qa3 b6 11. Rg3 Be6 12. Be2 Qc8 13. Rg1 Ne8 14. Ba6
Qb8 15. h4 h5 16. Bd3 Nd6 17. Ba1 a5 18. Nf1 Rfc8 19. Ng5 Bg4 20. f3 Bd7
21. Nxf7 Kxf7 22. Bxg6+ Kf8 23. Qd3 e6 24. Bxh5 Be8 25. Bxe8 Rxe8 26. h5
Qd8 27. Rg4 Bh6 28. Bb2 b5 29. Ke2 Nf5 30. Qa3+ Kf7 31. Qxa5 Qb8 32. Qd2
Rg8 33. Ke1 Qd8 34. Qh2 Rc8 35. Rxg8 Kxg8 36. Rc5 Kh7 37. Bc1 Qf6 38. f4
Bf8 39. Rc2 Be7 40. Kd1 Qf7 41. h6 b4 42. Rb2 c5 43. dxc5 Bxc5 44. Rc2 Qa7
45. Ng3 b3 46. Nxf5 Qa6 47. Ng3 bxc2+ 48. Qxc2+ Kxh6 49. Bb2 d4 50. Qe4 Qc6
51. Bxd4 Qxe4 52. Nxe4 Bxd4 53. exd4 Rc4 54. Ng5 Rxd4+ 55. Kc2 Kg6 56. Nxe6
Rd6 57. Ng5 Kf5 58. Nf3 Kxf4 59. Nd2 Ra6 60. Kb3 Ke3 61. Nc4+ Kd4 62. a4
Rc6 63. Na3 Rb6+ 64. Nb5+ Kc5 65. Kc2 Ra6 66. Kb3 Re6 67. Kc2 Re3 68. Kd2
Rh3 69. Ke2 Kb4 70. Nd4 Kc3 71. Nf5 Kc4 72. Kd2 Kd5 73. a5 Ke5 74. Ne7 Ke4
75. Nc6 Ra3 76. Kc2 Kd5 77. Ne7+ Kc4 78. Nc6 Kc5 79. Ne7 Re3 80. Nc8 Kd5
81. Kd2 Kd4 82. Nd6 Re5 83. a6 Ra5 84. Ke2 Rxa6 85. Nf5+ Ke5 86. Ng3 Kf4
87. Nf1 Ra3 88. Nd2 Re3+ 89. Kf2 Re8 90. Nc4 Re4 91. Nd2 Re3 92. Nf1 Re8
93. Nd2 Re7 94. Nb3 Ke4 95. Ke2 Re8 96. Kd2 Rc8 97. Na1 Rd8+ 98. Kc3 Rh8
99. Nc2 Rh7 100. Nb4 Rc7+ 101. Kd2 Rc5 102. Nc2 Rc4 103. Ne1 Rc8 104. Nc2
Rd8+ 105. Kc3 Rd3+ 106. Kc4 Rd7 107. Nb4 Rc7+ 108. Kb5 Kd4 109. Nc6+ Kd5
110. Na5 Kd4 111. Nb3+ Kd3 112. Nc5+ Kc3 113. Kb6 Rc8 114. Kb5 Rb8+ 115.
Kc6 Kd4 116. Ne6+ Kd3 117. Kd5 Rg8 118. Nf4+ Kc2 119. Ne6 Kc3 120. Ke5 Ra8
121. Kd5 Ra7 122. Nc5 Ra1 123. Ke5 Ra5 124. Kd6 Ra3 125. Kd5 Kb2 126. Ke5
Rg3 127. Ke4 Rg7 128. Nd3+ Ka3 129. Nc5 Rh7 130. Kd5 Rh4 131. Nd7 Rh6 132.
Ne5 Rh5 133. Ke4 Rh4+ 134. Kd5 Rh5 {50 moves rule} *[/pgn]

[lkaufman]

These three pawns could be the same as a Knight odds after all, so the misconception that a Knight is worth more than three pawns in the beginning turn out that if you let the engine play the position, it is really only worth 3 pawns; and possibly Komodo Dragon2 could give those three pawns Odds as Black or White to any IM. After performing these two tests, I believe that Komodo Dragon2 can give this position or the reverse as Black to any IM with time control of 15'+10" and boy will any human IM be frustrated trying to beat the engine.

[d]rnbqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 1 1

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.06.21"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Komodo-13.3-64bit-bmi2"]
[Result "*"]
[BlackElo "3400"]
[Time "17:29:39"]
[WhiteElo "3400"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "268"]
[WhiteType "program"]
[BlackType "program"]

1. Bb2 c6 2. d4 Nf6 3. Nf3 d5 4. Qb3 g6 5. e3 Bg7 6. Nbd2 O-O 7. Bd3 Bh3 8.
Rg1 Rc8 9. Rc1 Rc7 10. Qa3 b6 11. Rg3 Be6 12. Be2 Qc8 13. Rg1 Ne8 14. Ba6
Qb8 15. h4 h5 16. Bd3 Nd6 17. Ba1 a5 18. Nf1 Rfc8 19. Ng5 Bg4 20. f3 Bd7
21. Nxf7 Kxf7 22. Bxg6+ Kf8 23. Qd3 e6 24. Bxh5 Be8 25. Bxe8 Rxe8 26. h5
Qd8 27. Rg4 Bh6 28. Bb2 b5 29. Ke2 Nf5 30. Qa3+ Kf7 31. Qxa5 Qb8 32. Qd2
Rg8 33. Ke1 Qd8 34. Qh2 Rc8 35. Rxg8 Kxg8 36. Rc5 Kh7 37. Bc1 Qf6 38. f4
Bf8 39. Rc2 Be7 40. Kd1 Qf7 41. h6 b4 42. Rb2 c5 43. dxc5 Bxc5 44. Rc2 Qa7
45. Ng3 b3 46. Nxf5 Qa6 47. Ng3 bxc2+ 48. Qxc2+ Kxh6 49. Bb2 d4 50. Qe4 Qc6
51. Bxd4 Qxe4 52. Nxe4 Bxd4 53. exd4 Rc4 54. Ng5 Rxd4+ 55. Kc2 Kg6 56. Nxe6
Rd6 57. Ng5 Kf5 58. Nf3 Kxf4 59. Nd2 Ra6 60. Kb3 Ke3 61. Nc4+ Kd4 62. a4
Rc6 63. Na3 Rb6+ 64. Nb5+ Kc5 65. Kc2 Ra6 66. Kb3 Re6 67. Kc2 Re3 68. Kd2
Rh3 69. Ke2 Kb4 70. Nd4 Kc3 71. Nf5 Kc4 72. Kd2 Kd5 73. a5 Ke5 74. Ne7 Ke4
75. Nc6 Ra3 76. Kc2 Kd5 77. Ne7+ Kc4 78. Nc6 Kc5 79. Ne7 Re3 80. Nc8 Kd5
81. Kd2 Kd4 82. Nd6 Re5 83. a6 Ra5 84. Ke2 Rxa6 85. Nf5+ Ke5 86. Ng3 Kf4
87. Nf1 Ra3 88. Nd2 Re3+ 89. Kf2 Re8 90. Nc4 Re4 91. Nd2 Re3 92. Nf1 Re8
93. Nd2 Re7 94. Nb3 Ke4 95. Ke2 Re8 96. Kd2 Rc8 97. Na1 Rd8+ 98. Kc3 Rh8
99. Nc2 Rh7 100. Nb4 Rc7+ 101. Kd2 Rc5 102. Nc2 Rc4 103. Ne1 Rc8 104. Nc2
Rd8+ 105. Kc3 Rd3+ 106. Kc4 Rd7 107. Nb4 Rc7+ 108. Kb5 Kd4 109. Nc6+ Kd5
110. Na5 Kd4 111. Nb3+ Kd3 112. Nc5+ Kc3 113. Kb6 Rc8 114. Kb5 Rb8+ 115.
Kc6 Kd4 116. Ne6+ Kd3 117. Kd5 Rg8 118. Nf4+ Kc2 119. Ne6 Kc3 120. Ke5 Ra8
121. Kd5 Ra7 122. Nc5 Ra1 123. Ke5 Ra5 124. Kd6 Ra3 125. Kd5 Kb2 126. Ke5
Rg3 127. Ke4 Rg7 128. Nd3+ Ka3 129. Nc5 Rh7 130. Kd5 Rh4 131. Nd7 Rh6 132.
Ne5 Rh5 133. Ke4 Rh4+ 134. Kd5 Rh5 {50 moves rule} *[/pgn]

I don't doubt that Dragon can give three pawns to an IM, or even a 2500 GM, at Rapid and do well, but it would be much less impressive than knight odds. Maybe the difference between a knight and the three pawns isn't quite a winning difference, but knight odds is certainly the larger handicap, and is much easier for the human as he can play normal openings without having to consider the development advantage afforded to the engine by missing pawns (other than "f"). The reason knight odds is especially easy for the human is that he can play normal chess.

[Chessqueen]

These three pawns could be the same as a Knight odds after all, so the misconception that a Knight is worth more than three pawns in the beginning turn out that if you let the engine play the position, it is really only worth 3 pawns; and possibly Komodo Dragon2 could give those three pawns Odds as Black or White to any IM. After performing these two tests, I believe that Komodo Dragon2 can give this position or the reverse as Black to any IM with time control of 15'+10" and boy will any human IM be frustrated trying to beat the engine.

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.06.21"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Komodo-13.3-64bit-bmi2"]
[Result "*"]
[BlackElo "3400"]
[Time "17:29:39"]
[WhiteElo "3400"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "268"]
[WhiteType "program"]
[BlackType "program"]

1. Bb2 c6 2. d4 Nf6 3. Nf3 d5 4. Qb3 g6 5. e3 Bg7 6. Nbd2 O-O 7. Bd3 Bh3 8.
Rg1 Rc8 9. Rc1 Rc7 10. Qa3 b6 11. Rg3 Be6 12. Be2 Qc8 13. Rg1 Ne8 14. Ba6
Qb8 15. h4 h5 16. Bd3 Nd6 17. Ba1 a5 18. Nf1 Rfc8 19. Ng5 Bg4 20. f3 Bd7
21. Nxf7 Kxf7 22. Bxg6+ Kf8 23. Qd3 e6 24. Bxh5 Be8 25. Bxe8 Rxe8 26. h5
Qd8 27. Rg4 Bh6 28. Bb2 b5 29. Ke2 Nf5 30. Qa3+ Kf7 31. Qxa5 Qb8 32. Qd2
Rg8 33. Ke1 Qd8 34. Qh2 Rc8 35. Rxg8 Kxg8 36. Rc5 Kh7 37. Bc1 Qf6 38. f4
Bf8 39. Rc2 Be7 40. Kd1 Qf7 41. h6 b4 42. Rb2 c5 43. dxc5 Bxc5 44. Rc2 Qa7
45. Ng3 b3 46. Nxf5 Qa6 47. Ng3 bxc2+ 48. Qxc2+ Kxh6 49. Bb2 d4 50. Qe4 Qc6
51. Bxd4 Qxe4 52. Nxe4 Bxd4 53. exd4 Rc4 54. Ng5 Rxd4+ 55. Kc2 Kg6 56. Nxe6
Rd6 57. Ng5 Kf5 58. Nf3 Kxf4 59. Nd2 Ra6 60. Kb3 Ke3 61. Nc4+ Kd4 62. a4
Rc6 63. Na3 Rb6+ 64. Nb5+ Kc5 65. Kc2 Ra6 66. Kb3 Re6 67. Kc2 Re3 68. Kd2
Rh3 69. Ke2 Kb4 70. Nd4 Kc3 71. Nf5 Kc4 72. Kd2 Kd5 73. a5 Ke5 74. Ne7 Ke4
75. Nc6 Ra3 76. Kc2 Kd5 77. Ne7+ Kc4 78. Nc6 Kc5 79. Ne7 Re3 80. Nc8 Kd5
81. Kd2 Kd4 82. Nd6 Re5 83. a6 Ra5 84. Ke2 Rxa6 85. Nf5+ Ke5 86. Ng3 Kf4
87. Nf1 Ra3 88. Nd2 Re3+ 89. Kf2 Re8 90. Nc4 Re4 91. Nd2 Re3 92. Nf1 Re8
93. Nd2 Re7 94. Nb3 Ke4 95. Ke2 Re8 96. Kd2 Rc8 97. Na1 Rd8+ 98. Kc3 Rh8
99. Nc2 Rh7 100. Nb4 Rc7+ 101. Kd2 Rc5 102. Nc2 Rc4 103. Ne1 Rc8 104. Nc2
Rd8+ 105. Kc3 Rd3+ 106. Kc4 Rd7 107. Nb4 Rc7+ 108. Kb5 Kd4 109. Nc6+ Kd5
110. Na5 Kd4 111. Nb3+ Kd3 112. Nc5+ Kc3 113. Kb6 Rc8 114. Kb5 Rb8+ 115.
Kc6 Kd4 116. Ne6+ Kd3 117. Kd5 Rg8 118. Nf4+ Kc2 119. Ne6 Kc3 120. Ke5 Ra8
121. Kd5 Ra7 122. Nc5 Ra1 123. Ke5 Ra5 124. Kd6 Ra3 125. Kd5 Kb2 126. Ke5
Rg3 127. Ke4 Rg7 128. Nd3+ Ka3 129. Nc5 Rh7 130. Kd5 Rh4 131. Nd7 Rh6 132.
Ne5 Rh5 133. Ke4 Rh4+ 134. Kd5 Rh5 {50 moves rule} *[/pgn]

I don't doubt that Dragon can give three pawns to an IM, or even a 2500 GM, at Rapid and do well, but it would be much less impressive than knight odds. Maybe the difference between a knight and the three pawns isn't quite a winning difference, but knight odds is certainly the larger handicap, and is much easier for the human as he can play normal openings without having to consider the development advantage afforded to the engine by missing pawns (other than "f"). The reason knight odds is especially easy for the human is that he can play normal chess.

Yes but if the human has the Black position with all his pieces he can play normal chess, the engine missing three pawn is the one that has to be creative, it is no different than when GM Nakamura played versus Komodo with two pawn Odds and he had all his pieces, but in this case it would be an IM at TC 15'+10"
Here on youtube.com you can see GM Nakamura drawing versus Komodo with two pawns Odds


[d]rnbqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 1 1

Or the human IM playing with White
[d]rnbqkbnr/p2ppp1p/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1

[lkaufman]

These three pawns could be the same as a Knight odds after all, so the misconception that a Knight is worth more than three pawns in the beginning turn out that if you let the engine play the position, it is really only worth 3 pawns; and possibly Komodo Dragon2 could give those three pawns Odds as Black or White to any IM. After performing these two tests, I believe that Komodo Dragon2 can give this position or the reverse as Black to any IM with time control of 15'+10" and boy will any human IM be frustrated trying to beat the engine.

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.06.21"]
[Round "?"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Komodo-13.3-64bit-bmi2"]
[Result "*"]
[BlackElo "3400"]
[Time "17:29:39"]
[WhiteElo "3400"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "r1bqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "268"]
[WhiteType "program"]
[BlackType "program"]

1. Bb2 c6 2. d4 Nf6 3. Nf3 d5 4. Qb3 g6 5. e3 Bg7 6. Nbd2 O-O 7. Bd3 Bh3 8.
Rg1 Rc8 9. Rc1 Rc7 10. Qa3 b6 11. Rg3 Be6 12. Be2 Qc8 13. Rg1 Ne8 14. Ba6
Qb8 15. h4 h5 16. Bd3 Nd6 17. Ba1 a5 18. Nf1 Rfc8 19. Ng5 Bg4 20. f3 Bd7
21. Nxf7 Kxf7 22. Bxg6+ Kf8 23. Qd3 e6 24. Bxh5 Be8 25. Bxe8 Rxe8 26. h5
Qd8 27. Rg4 Bh6 28. Bb2 b5 29. Ke2 Nf5 30. Qa3+ Kf7 31. Qxa5 Qb8 32. Qd2
Rg8 33. Ke1 Qd8 34. Qh2 Rc8 35. Rxg8 Kxg8 36. Rc5 Kh7 37. Bc1 Qf6 38. f4
Bf8 39. Rc2 Be7 40. Kd1 Qf7 41. h6 b4 42. Rb2 c5 43. dxc5 Bxc5 44. Rc2 Qa7
45. Ng3 b3 46. Nxf5 Qa6 47. Ng3 bxc2+ 48. Qxc2+ Kxh6 49. Bb2 d4 50. Qe4 Qc6
51. Bxd4 Qxe4 52. Nxe4 Bxd4 53. exd4 Rc4 54. Ng5 Rxd4+ 55. Kc2 Kg6 56. Nxe6
Rd6 57. Ng5 Kf5 58. Nf3 Kxf4 59. Nd2 Ra6 60. Kb3 Ke3 61. Nc4+ Kd4 62. a4
Rc6 63. Na3 Rb6+ 64. Nb5+ Kc5 65. Kc2 Ra6 66. Kb3 Re6 67. Kc2 Re3 68. Kd2
Rh3 69. Ke2 Kb4 70. Nd4 Kc3 71. Nf5 Kc4 72. Kd2 Kd5 73. a5 Ke5 74. Ne7 Ke4
75. Nc6 Ra3 76. Kc2 Kd5 77. Ne7+ Kc4 78. Nc6 Kc5 79. Ne7 Re3 80. Nc8 Kd5
81. Kd2 Kd4 82. Nd6 Re5 83. a6 Ra5 84. Ke2 Rxa6 85. Nf5+ Ke5 86. Ng3 Kf4
87. Nf1 Ra3 88. Nd2 Re3+ 89. Kf2 Re8 90. Nc4 Re4 91. Nd2 Re3 92. Nf1 Re8
93. Nd2 Re7 94. Nb3 Ke4 95. Ke2 Re8 96. Kd2 Rc8 97. Na1 Rd8+ 98. Kc3 Rh8
99. Nc2 Rh7 100. Nb4 Rc7+ 101. Kd2 Rc5 102. Nc2 Rc4 103. Ne1 Rc8 104. Nc2
Rd8+ 105. Kc3 Rd3+ 106. Kc4 Rd7 107. Nb4 Rc7+ 108. Kb5 Kd4 109. Nc6+ Kd5
110. Na5 Kd4 111. Nb3+ Kd3 112. Nc5+ Kc3 113. Kb6 Rc8 114. Kb5 Rb8+ 115.
Kc6 Kd4 116. Ne6+ Kd3 117. Kd5 Rg8 118. Nf4+ Kc2 119. Ne6 Kc3 120. Ke5 Ra8
121. Kd5 Ra7 122. Nc5 Ra1 123. Ke5 Ra5 124. Kd6 Ra3 125. Kd5 Kb2 126. Ke5
Rg3 127. Ke4 Rg7 128. Nd3+ Ka3 129. Nc5 Rh7 130. Kd5 Rh4 131. Nd7 Rh6 132.
Ne5 Rh5 133. Ke4 Rh4+ 134. Kd5 Rh5 {50 moves rule} *[/pgn]

I don't doubt that Dragon can give three pawns to an IM, or even a 2500 GM, at Rapid and do well, but it would be much less impressive than knight odds. Maybe the difference between a knight and the three pawns isn't quite a winning difference, but knight odds is certainly the larger handicap, and is much easier for the human as he can play normal openings without having to consider the development advantage afforded to the engine by missing pawns (other than "f"). The reason knight odds is especially easy for the human is that he can play normal chess.

Yes but if the human has the Black position with all his pieces he can play normal chess, the engine missing three pawn is the one that has to be creative, it is no different than when GM Nakamura played versus Komodo with two pawn Odds and he had all his pieces, but in this case it would be an IM at TC 15'+10"
Here on youtube.com you can see GM Nakamura drawing versus Komodo with two pawns Odds


[d]rnbqkbnr/pppppppp/8/8/8/8/P2PPP1P/RNBQKBNR w KQkq - 1 1

Or the human IM playing with White
[d]rnbqkbnr/p2ppp1p/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1

The point is that the human has to consider the consequences of early moves by the queen or either bishop of the opponent. For example after 1.e4 Bb7 he already can't play a normal opening. He has to think for himself from the first move. At knight odds he can just pretend it's a normal game and be glad the opponent is missing a knight. I'm sure that every top GM would much rather have knight odds than those three pawns (given the same color). In Morphy's time the only games played with pawn odds were with the f pawn gone, since it gave only disadvantages to the odds-giver. Even two pawn odds didn't exist until Kasparov introduced it as a compromise between f7 and knight odds.