Dragon by Komodo Chess

[BrendanJNorman]

I gotta say, this game is an absolute positional masterpiece by Dragon with MCTS switched ON.

The way Dragon ties up the queenside, freezes the black bishop with 33.e5!! fxe5 (33...Qxe5 34.Qxb6!+-) 34.Be4! is so impressive...then swinging to the kingside once black is tied up.

It's kind of like Vladimir Kramnik on God-mode.

[Event "Dragon vs Gang"] [Site "China"] [Date "2020.11.12"] [Round "5.1"] [White "Dragon by Komodo Chess MCTS"] [Black "Thinker 5.3b Inert"] [Result "1-0"] [ECO "A13"] [PlyCount "90"] [EventDate "2020.??.??"] [TimeControl "180+2"] 1. Nf3 Nf6 2. c4 e6 3. g3 d5 {A13: English, Neo-Catalan} 4. d4 Bb4+ 5. Bd2 Be7 6. Bg2 O-O 7. O-O c6 8. Bf4 dxc4 9. a4 {0.04/14 6.9 725} Nd5 10. a5 {-0.15/13 7.4 716} Nxf4 11. gxf4 {-0.06/13 6.4 661} Bb4 12. Qa4 {-0.22/13 6.7 566} Na6 13. e3 {-0.24/14 4.0 913} f6 14. Nbd2 {-0.28/14 8.6 881} c3 15. Nc4 {-0.28/14 3.8 1127} cxb2 16. Nxb2 {-0.27/15 3.8 1418} Bd6 17. Rfd1 {14.06/0 1713} Nb4 18. Nd2 {-0.11/13 6.5 574} Be7 19. Ne4 {-0.10/13 5.9 609} Nd5 20. Qc4 {0.00/11 6.2 277} Qc7 21. Nc3 {0.00/10 4.1 161} Bd6 22. Ne4 {-0.19/14 6.4 825} Bb4 23. a6 { -0.20/14 3.6 1055} b6 24. Nd3 {0.42/14 6.4 847} b5 25. Qb3 {0.44/14 3.8 943} Bd6 26. Nxd6 {0.57/14 6.4 1037} Qxd6 27. e4 {0.56/15 3.6 1405} Nxf4 28. Nxf4 { 0.62/14 6.3 879} Qxf4 29. d5 {0.68/15 3.6 1266} Re8 30. dxc6 {0.73/14 6.2 920} Rb8 31. Qc3 {0.96/16 6.8 1975} Qc7 32. Qc5 {0.94/16 3.6 1929} Rb6 33. e5 { 0.84/16 3.6 2077} fxe5 34. Be4 {0.88/16 3.6 2354} b4 35. Rd3 {0.76/16 3.7 1851} Rd8 36. Rad1 {2.31/14 6.5 1091} Rxd3 37. Rxd3 {2.33/15 3.7 1405} Rb8 38. Qc1 { 10.88/14 6.3 825} h6 39. Qd1 {18.04/15 6.3 1164} g6 40. Rg3 {57.48/15 6.1 1334} Kf8 41. Rxg6 {250.00/14 6.2 835} Bxa6 42. Qf3+ {250.00/13 6.0 679} Ke7 43. Rg7+ {250.00/13 4.1 635} Kd6 44. Qd1+ {250.00/13 3.3 596} Kc5 45. Rxc7 {250.00/13 3. 3 554} Rg8+ {resign} 1-0

Decent chessplayers will admire the positional genius of this game, while computer nerds will use Stockfish to look for holes...whatever!

I have some very special engines for training games already, but Dragon is definitely going to be a serious part of my game/opening analysis in the future.

Great job Komodo team!

[Jhoravi]

The problem is that if the handicap is high enough to give humans a good chance (like knight odds or even two pawns), the stronger side almost always wins in self play so it can't really learn how to trick the opponent.

If the stronger side always wins then let it prefer shortest victory vs longer ones. Isn't that possible during training?

[Madeleine Birchfield]

The Komodo Settings and Usage page on the Komodo Chess website needs to be updated. It is currently stuck on Komodo 11.01.

https://komodochess.com/store/pages.php?cmsid=14

The Komodo 13 page needs to be updated as well. It says that Komodo 14 is the current engine, when Komodo 14 has gone on sale and replaced with Dragon as the current engine.

https://komodochess.com/Komodo13.htm

[Madeleine Birchfield]

Same goes with the Frequently Asked Questions page, which seems to be stuck on Komodo 9.3.

https://komodochess.com/store/pages.php?cmsid=20

[Marcus9]

Hi Larry, Is there any chance of NNUE specialized for losing positions for use in handicap matches?

We actually did include some such positions in the training which might help a bit in such matches, probably not a lot though. The problem is that if the handicap is high enough to give humans a good chance (like knight odds or even two pawns), the stronger side almost always wins in self play so it can't really learn how to trick the opponent. I suppose we could train it by playing against a crippled version of itself, but that might mess up normal play so we haven't tried that. We are considering possible opponents for such a match. I wonder, which players would readers most like to see facing Dragon in a handicap match (other than Magnus Carlsen who has never shown interest in playing such matches with engines)?

How does Dragon fare at giving knight odd in slow bullet? My dream is to see Hikaru Nakamura against Dragon at knight odds in many games 2 '+ 1 " in a row, like 12 or 24 games.
While perhaps Hikaru would still be the favorite, it could be a periodic event to track Komodo's progress.

[mjlef]

Hi Larry, Is there any chance of NNUE specialized for losing positions for use in handicap matches?

We actually did include some such positions in the training which might help a bit in such matches, probably not a lot though. The problem is that if the handicap is high enough to give humans a good chance (like knight odds or even two pawns), the stronger side almost always wins in self play so it can't really learn how to trick the opponent. I suppose we could train it by playing against a crippled version of itself, but that might mess up normal play so we haven't tried that. We are considering possible opponents for such a match. I wonder, which players would readers most like to see facing Dragon in a handicap match (other than Magnus Carlsen who has never shown interest in playing such matches with engines)?

How does Dragon fare at giving knight odd in slow bullet? My dream is to see Hikaru Nakamura against Dragon at knight odds in many games 2 '+ 1 " in a row, like 12 or 24 games.
While perhaps Hikaru would still be the favorite, it could be a periodic event to track Komodo's progress.

Yes, we definitely need to update these. Note that there is an updated Resume file included in the downloads. For each version.

[AdminX]

Just read this on Twitter.

[Krzysztof Grzelak]

The file shows a ten game match with Lc0 with a score of one win each and eight draws, and the sole loss was by adjudication, not time. Could you please explain what you think is the problem with Komodo or Dragon time management? It's not obvious from the file.

I noticed that, unfortunately, there is a bug with time compared to other engines. The engine does not lose on time but it has problems with it.

[lkaufman]

The file shows a ten game match with Lc0 with a score of one win each and eight draws, and the sole loss was by adjudication, not time. Could you please explain what you think is the problem with Komodo or Dragon time management? It's not obvious from the file.

I noticed that, unfortunately, there is a bug with time compared to other engines. The engine does not lose on time but it has problems with it.

Are you just saying that Dragon and Komodo tend to take more time than opposing engines? If so why would you call that a bug? Do you have evidence that we would do better taking less time?

[carldaman]

Larry,

Curious, have you given any thought of going up against a strong human GM with Dragon set to Armageddon mode as Black and the human as White? The GM would basically get draw odds with White, if he can get the draw, while Dragon would go all out attacking with Black. It would be an very realistic and exciting handicap.