Return Match for Komodo

[lkaufman]

Contempt, the type which helps Komodo play odds matches better, was introduced publically in Komodo 9.2. It is not in Komodo 8 which Lyudmil Tsvetkov is using.

Average search depth reached in Neuman match was 31, and just 17 in Tsvetkov game.

The starting position is nearly correct -- b2+f2 is very close. Try again with the actual handicap used, b2+e2.

I can confirm that. The older "Contempt" we even called "Drawscore" since it was what Komodo considered a draw to be worth (so the default -7 just tied to avoid draws). The new Contempt does much more. A positive Contempt discourages accepting early draws, avoids trading material to keep the game complex, and tries to keep a more open pawn structure. These make a huge difference in these games. If Lyudmil privately send us his email I would be glad to get him a more recent copy of Komodo to see how it goes.

Mark

Thanks for the offer, Mark.

If it would not considerably hurt your budget, I will take advantage of the opportunity to thrash newest Komodo in some games.

My mail is ltsvet((at))yahoo((dot))com

Not that it will make significant difference.
I have been following SF and Komodo development closely for the past 3-4 years, and, if some things really change for the better, others simply remain the same.

I also have just a 4-core, so any successful performance migth be attributed to that.

I would also shun posting games on this forum, as the backlash is oversharpish.

Btw., congratulations for the match win, and the new contempt( I know Larry is furious at me for disrespecting contempt, I am doing it out of general considerations, as any contempt simply produces weaker moves, but of course, from a purely psychological point of view, this migth be an excellent option vs humans indeed).

PS. An interesting thing I discovered during trying-out these handicap positions, no matter if the engine plays without a knigth, rook, 2 pawns or just a pawn, is that, with very small specificities for the particular positions, almost always the best 1st move for black is a kingside fianchettoe (g6,Bg7,Nf6), followed by 1...c5, then 1...e5, and 1...d5 comes only in 4th.

I did extensive testing with those positions, and indeed, kingside fianchettoe is always the best line of play, while d5 seems like a low-ranging choice. (Fischer and Kasparov should have known that)

woops, got too long...

How can I be upset with you for not using Contempt, when you don't even have a version that uses Contempt? And how can you comment on the effect of using Contempt, when you don't have such a version? The Drawscore term in older Komodos is just that, something to avoid repetitions, and doesn't affect play very much in general. Contempt is a whole different story. I suggest you use a value of 80 for handicaps of two pawns or similar ones, if you get the new Komodo from Mark. Note that's positive 80, not negative! I don't think you'll get your closed King's Indian positions.

[Lyudmil Tsvetkov]

Contempt, the type which helps Komodo play odds matches better, was introduced publically in Komodo 9.2. It is not in Komodo 8 which Lyudmil Tsvetkov is using.

Average search depth reached in Neuman match was 31, and just 17 in Tsvetkov game.

The starting position is nearly correct -- b2+f2 is very close. Try again with the actual handicap used, b2+e2.

I can confirm that. The older "Contempt" we even called "Drawscore" since it was what Komodo considered a draw to be worth (so the default -7 just tied to avoid draws). The new Contempt does much more. A positive Contempt discourages accepting early draws, avoids trading material to keep the game complex, and tries to keep a more open pawn structure. These make a huge difference in these games. If Lyudmil privately send us his email I would be glad to get him a more recent copy of Komodo to see how it goes.

Mark

Thanks for the offer, Mark.

If it would not considerably hurt your budget, I will take advantage of the opportunity to thrash newest Komodo in some games.

My mail is ltsvet((at))yahoo((dot))com

Not that it will make significant difference.
I have been following SF and Komodo development closely for the past 3-4 years, and, if some things really change for the better, others simply remain the same.

I also have just a 4-core, so any successful performance migth be attributed to that.

I would also shun posting games on this forum, as the backlash is oversharpish.

Btw., congratulations for the match win, and the new contempt( I know Larry is furious at me for disrespecting contempt, I am doing it out of general considerations, as any contempt simply produces weaker moves, but of course, from a purely psychological point of view, this migth be an excellent option vs humans indeed).

PS. An interesting thing I discovered during trying-out these handicap positions, no matter if the engine plays without a knigth, rook, 2 pawns or just a pawn, is that, with very small specificities for the particular positions, almost always the best 1st move for black is a kingside fianchettoe (g6,Bg7,Nf6), followed by 1...c5, then 1...e5, and 1...d5 comes only in 4th.

I did extensive testing with those positions, and indeed, kingside fianchettoe is always the best line of play, while d5 seems like a low-ranging choice. (Fischer and Kasparov should have known that)

woops, got too long...

How can I be upset with you for not using Contempt, when you don't even have a version that uses Contempt? And how can you comment on the effect of using Contempt, when you don't have such a version? The Drawscore term in older Komodos is just that, something to avoid repetitions, and doesn't affect play very much in general. Contempt is a whole different story. I suggest you use a value of 80 for handicaps of two pawns or similar ones, if you get the new Komodo from Mark. Note that's positive 80, not negative! I don't think you'll get your closed King's Indian positions.

Hi Larry.

I must admit that you have implemented a very useful feature, what concerns handicap play versus humans. It is obvious that in such mode Komodo values higher or much higher all factors tending to open the game.

This has been long overdue and is a nice feature indeed.
I do not think however that should affect very much play at normal 2 hours TC, the engine will only suffer because of inaccurate play.

At blitz, the feature migth be useful, I am winning and losing many games, and indeed closed positions are less frequent, but one can still create them.

I would not have posted another game, as many people seem to dislike top engines losing to humans, but, as you sounded convincing that KID stuff should not happen, below one such game.

Game is at 4 threads, Komodo 9.3 uses contempt set to 80, as you advised, TC is 2' + 2'' for Komodo, longer for me(until I find a plausible solution ), as you can see from the pgn.

[pgn][Event "Blitz 2m+2s"]
[Site "Microsoft"]
[Date "2016.08.05"]
[Round "?"]
[White "Komodo 9.3 64-bit"]
[Black "Lyudmil Tsvetkov, owner"]
[Result "0-1"]
[Annotator "owner"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/P1PP1PPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "59"]
[EventDate "2016.??.??"]
[TimeControl "120+2"]

{512MB, OWNER-PC} 1. Nf3 {-0.13/17 4} c5 {4} 2. c4 {-0.36/18 11} Nc6 {2} 3. Bd3
{-0.36/21 10} e5 {2} 4. Nc3 {-0.37/19 5} d6 {2} 5. Rb1 {-0.49/20 2} f5 {13} 6.
Bc2 {-0.46/21 8} Nf6 {2} 7. O-O {-0.52/21 12} Be7 {2} 8. Ba4 {-0.54/20 4} O-O {
3} 9. d3 {-0.66/21 5} f4 {3} 10. Re1 {-0.77/21 5} h6 {3} 11. Nh4 {-0.53/19 9}
Nd4 {24} 12. Ng6 {-0.80/23 9} Rf7 {3} 13. Nb5 {-0.84/22 5} Nxb5 {22} 14. Rxb5 {
-0.75/20 8} Rb8 {38} 15. Bd2 {-0.62/20 10} Bf5 {15} 16. Nxe7+ {-0.84/21 6} Rxe7
{3} 17. Rb3 {-0.94/21 6} g5 {15} 18. h3 {-0.86/18 8} Kh8 {11} 19. Re2 {-1.35/
19 9} Qg8 {6} 20. Re1 {-1.38/19 7} Qg6 {16} 21. f3 {-1.75/20 5} Rg8 {9} 22. Rb2
{-2.02/20 8} h5 {5} 23. d4 {-2.49/20 6} cxd4 {94} 24. Bb4 {-2.79/28 0} Re6 {53}
25. Bb3 {-2.79/18 6} g4 {12} 26. Bxd6 {-2.87/19 4} Rxd6 {21} 27. c5 {-3.78/24 0
} Rdd8 {9} 28. Bxg8 {-2.11/16 1} Rxg8 {6} 29. hxg4 {-3.24/17 2} hxg4 {3} 30.
Rxe5 {-4.99/19 2} 0-1

[/pgn]

[Lyudmil Tsvetkov]

Above position was with e2-b2 missing, as suggested by Jesse.

[velmarin]

How can I be upset with you for not using Contempt, when you don't even have a version that uses Contempt? And how can you comment on the effect of using Contempt, when you don't have such a version? The Drawscore term in older Komodos is just that, something to avoid repetitions, and doesn't affect play very much in general. Contempt is a whole different story. I suggest you use a value of 80 for handicaps of two pawns or similar ones, if you get the new Komodo from Mark. Note that's positive 80, not negative! I don't think you'll get your closed King's Indian positions.

Is joke,
you are a Grand Master of chess. Earned in years of competition.
I am sure that you can not earn his son Komodo 8.

Contempt, drawscore, no more nonsense.
Enough of liars.

[Lyudmil Tsvetkov]

But of course, winning in an open game should not be a problem also, provided you give yourself more time.

Just one more game, to illustrate this:

[pgn][Event "Blitz 2m+2s"]
[Site "Microsoft"]
[Date "2016.08.05"]
[Round "?"]
[White "Komodo 9.3 64-bit"]
[Black "Lyudmil Tsvetkov, owner"]
[Result "0-1"]
[Annotator "owner"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/P1PP1PPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "59"]
[EventDate "2016.??.??"]
[TimeControl "120+2"]

{512MB, OWNER-PC} 1. Nf3 {-0.22/19 8} g6 {4} 2. Nc3 {-0.05/19 9} Nf6 {2} 3. d4
{-0.16/20 5} Bg7 {3} 4. Bb5 {-0.33/19 8} a6 {4} 5. Bd3 {-0.31/19 3} O-O {3} 6.
O-O {-0.60/21 19} d6 {4} 7. Qe2 {-0.60/21 10} b6 {9} 8. Bg5 {-0.53/17 10} Bb7 {
5} 9. Rab1 {-0.59/18 5} Nbd7 {5} 10. Ne4 {-0.76/19 3} Re8 {15} 11. c3 {-0.97/
19 12} e5 {40} 12. dxe5 {-1.25/21 9} Nxe5 {2} 13. Nxe5 {-1.20/21 2} Rxe5 {5}
14. f4 {-0.64/20 4} Re8 {80} 15. Nxf6+ {-1.12/22 8} Bxf6 {2} 16. Qd2 {-1.32/23
7} Bxg5 {100} 17. fxg5 {-1.48/29 0} Be4 {11} 18. Bxe4 {-0.84/20 4} Rxe4 {2} 19.
Rb4 {-1.11/21 3} Rxb4 {36} 20. cxb4 {-1.12/17 2} Qd7 {7} 21. Qd3 {-1.48/19 0}
a5 {19} 22. a3 {-1.47/22 5} axb4 {4} 23. axb4 {-1.67/23 2} Re8 {23} 24. h3 {-1.
91/24 4} Re5 {8} 25. Qc3 {-2.07/19 5} c5 {24} 26. h4 {-2.38/23 0} Qe6 {10} 27.
bxc5 {-2.45/22 7} bxc5 {2} 28. Rb1 {-2.13/23 5} Qd5 {29} 29. Rb8+ {-2.00/21 4}
Kg7 {1} 30. Rd8 {-2.35/23 7} 0-1

[/pgn]

Man, soon I should buy another hard disk: no more place to save Komodo-won games.

[lkaufman]

But of course, winning in an open game should not be a problem also, provided you give yourself more time.

Just one more game, to illustrate this:

[pgn][Event "Blitz 2m+2s"]
[Site "Microsoft"]
[Date "2016.08.05"]
[Round "?"]
[White "Komodo 9.3 64-bit"]
[Black "Lyudmil Tsvetkov, owner"]
[Result "0-1"]
[Annotator "owner"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/P1PP1PPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "59"]
[EventDate "2016.??.??"]
[TimeControl "120+2"]

{512MB, OWNER-PC} 1. Nf3 {-0.22/19 8} g6 {4} 2. Nc3 {-0.05/19 9} Nf6 {2} 3. d4
{-0.16/20 5} Bg7 {3} 4. Bb5 {-0.33/19 8} a6 {4} 5. Bd3 {-0.31/19 3} O-O {3} 6.
O-O {-0.60/21 19} d6 {4} 7. Qe2 {-0.60/21 10} b6 {9} 8. Bg5 {-0.53/17 10} Bb7 {
5} 9. Rab1 {-0.59/18 5} Nbd7 {5} 10. Ne4 {-0.76/19 3} Re8 {15} 11. c3 {-0.97/
19 12} e5 {40} 12. dxe5 {-1.25/21 9} Nxe5 {2} 13. Nxe5 {-1.20/21 2} Rxe5 {5}
14. f4 {-0.64/20 4} Re8 {80} 15. Nxf6+ {-1.12/22 8} Bxf6 {2} 16. Qd2 {-1.32/23
7} Bxg5 {100} 17. fxg5 {-1.48/29 0} Be4 {11} 18. Bxe4 {-0.84/20 4} Rxe4 {2} 19.
Rb4 {-1.11/21 3} Rxb4 {36} 20. cxb4 {-1.12/17 2} Qd7 {7} 21. Qd3 {-1.48/19 0}
a5 {19} 22. a3 {-1.47/22 5} axb4 {4} 23. axb4 {-1.67/23 2} Re8 {23} 24. h3 {-1.
91/24 4} Re5 {8} 25. Qc3 {-2.07/19 5} c5 {24} 26. h4 {-2.38/23 0} Qe6 {10} 27.
bxc5 {-2.45/22 7} bxc5 {2} 28. Rb1 {-2.13/23 5} Qd5 {29} 29. Rb8+ {-2.00/21 4}
Kg7 {1} 30. Rd8 {-2.35/23 7} 0-1

[/pgn]

Man, soon I should buy another hard disk: no more place to save Komodo-won games.

Why does this game end here, White is just down the same two pawns he started down? Anyway, in neither game is there more than a single ram (pawn blocked by enemy pawn) in the first 20 moves, so they are not really blocked positions. Contempt did its job. What you do show is that you are a strong enough player to win games from Komodo with the two pawn handicap and the extra time. I don't know, but probably you are something like 2300 strength, is that a good guess? Perhaps you are right that the GM might have done better to play the opening your way; probably he just played what he is most familiar with playing in normal chess. Also remember he was playing against a 24 core machine.

[JJJ]

I d really like to see an official match between Lyudmil Tsvetkov and Komodo dev with handicap.

[mehmet karaman]

I d really like to see an official match between Lyudmil Tsvetkov and Komodo dev with handicap.

It will be an exciting match if Komodo will play with knight handicap

[Lyudmil Tsvetkov]

But of course, winning in an open game should not be a problem also, provided you give yourself more time.

Just one more game, to illustrate this:

[pgn][Event "Blitz 2m+2s"]
[Site "Microsoft"]
[Date "2016.08.05"]
[Round "?"]
[White "Komodo 9.3 64-bit"]
[Black "Lyudmil Tsvetkov, owner"]
[Result "0-1"]
[Annotator "owner"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/P1PP1PPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "59"]
[EventDate "2016.??.??"]
[TimeControl "120+2"]

{512MB, OWNER-PC} 1. Nf3 {-0.22/19 8} g6 {4} 2. Nc3 {-0.05/19 9} Nf6 {2} 3. d4
{-0.16/20 5} Bg7 {3} 4. Bb5 {-0.33/19 8} a6 {4} 5. Bd3 {-0.31/19 3} O-O {3} 6.
O-O {-0.60/21 19} d6 {4} 7. Qe2 {-0.60/21 10} b6 {9} 8. Bg5 {-0.53/17 10} Bb7 {
5} 9. Rab1 {-0.59/18 5} Nbd7 {5} 10. Ne4 {-0.76/19 3} Re8 {15} 11. c3 {-0.97/
19 12} e5 {40} 12. dxe5 {-1.25/21 9} Nxe5 {2} 13. Nxe5 {-1.20/21 2} Rxe5 {5}
14. f4 {-0.64/20 4} Re8 {80} 15. Nxf6+ {-1.12/22 8} Bxf6 {2} 16. Qd2 {-1.32/23
7} Bxg5 {100} 17. fxg5 {-1.48/29 0} Be4 {11} 18. Bxe4 {-0.84/20 4} Rxe4 {2} 19.
Rb4 {-1.11/21 3} Rxb4 {36} 20. cxb4 {-1.12/17 2} Qd7 {7} 21. Qd3 {-1.48/19 0}
a5 {19} 22. a3 {-1.47/22 5} axb4 {4} 23. axb4 {-1.67/23 2} Re8 {23} 24. h3 {-1.
91/24 4} Re5 {8} 25. Qc3 {-2.07/19 5} c5 {24} 26. h4 {-2.38/23 0} Qe6 {10} 27.
bxc5 {-2.45/22 7} bxc5 {2} 28. Rb1 {-2.13/23 5} Qd5 {29} 29. Rb8+ {-2.00/21 4}
Kg7 {1} 30. Rd8 {-2.35/23 7} 0-1

[/pgn]

Man, soon I should buy another hard disk: no more place to save Komodo-won games.

Why does this game end here, White is just down the same two pawns he started down? Anyway, in neither game is there more than a single ram (pawn blocked by enemy pawn) in the first 20 moves, so they are not really blocked positions. Contempt did its job. What you do show is that you are a strong enough player to win games from Komodo with the two pawn handicap and the extra time. I don't know, but probably you are something like 2300 strength, is that a good guess? Perhaps you are right that the GM might have done better to play the opening your way; probably he just played what he is most familiar with playing in normal chess. Also remember he was playing against a 24 core machine.

2100 FIDE elo, 2200 Bulgarian (basically same as FIDE) from... 10 years ago, when I played my last competitive game.

Though in my last tournament my performance was 2400.

I roughly estimate that in the last 10 years I have not played competitive chess I have added about 3 times my then-strength. What this makes, I am not exactly certain.

I am from the engine generation, you know, it is much easier to learn from engines.

If I nowadays appear in a tournament, I am certain I will soon pass 2400-2500 effortless, in a back room with quiet conditions I play much stronger though.

What I can say for sure is that in the last years my interest sharply swung to computer chess, it is so deep and appealing...

Game ends there, as I assume 300cps advantage or over is an easy win, just bored to play to the end. Besides, you migth blunder your queen with too long a game. Long time control always helps though.

As for a full game, one more below (Komodo resigned at -800cps score), but who is interested in winning a 2 connected pawns rook ending?

[pgn][Event "Blitz 2m+2s"]
[Site "Microsoft"]
[Date "2016.08.05"]
[Round "?"]
[White "Komodo 9.3 64-bit"]
[Black "Lyudmil Tsvetkov, owner"]
[Result "0-1"]
[Annotator "owner"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/P1PP1PPP/RNBQKBNR w KQkq - 0 1"]
[PlyCount "106"]
[EventDate "2016.??.??"]
[TimeControl "120+2"]

{512MB, OWNER-PC} 1. Nf3 {-0.12/18 5} g6 {3} 2. Nc3 {-0.03/20 8} Bg7 {2} 3. Rb1
{-0.26/20 11} Nf6 {2} 4. d4 {-0.35/21 6} O-O {3} 5. Bb5 {-0.35/21 12} a6 {3} 6.
Bd3 {-0.52/19 5} d6 {2} 7. O-O {-0.59/20 5} b6 {3} 8. Re1 {-0.33/19 8} Bb7 {3}
9. Ne4 {-0.48/19 8} Nbd7 {3} 10. Bg5 {-0.57/19 8} Re8 {6} 11. Nxf6+ {-0.65/18 8
} Nxf6 {2} 12. Qd2 {-0.63/18 6} Qd7 {19} 13. a3 {-0.89/20 0} e6 {29} 14. Rbc1 {
-0.76/18 7} c5 {34} 15. c3 {-0.91/18 4} Bxf3 {25} 16. gxf3 {-1.17/23 0} d5 {7}
17. Bf1 {-1.09/21 9} cxd4 {19} 18. cxd4 {-1.03/18 4} b5 {9} 19. a4 {-1.01/18 6}
bxa4 {11} 20. Ra1 {-1.39/19 1} h6 {65} 21. Bxh6 {-0.64/19 6} Bxh6 {2} 22. Qxh6
{-0.83/17 4} Nh5 {11} 23. Be2 {-1.02/18 8} Qd6 {25} 24. Rxa4 {-1.63/29 2} Qf4 {
101} 25. Qxf4 {-2.02/32 0} Nxf4 {2} 26. Bxa6 {-2.01/28 3} Reb8 {23} 27. h4 {-2.
11/29 0} Rb6 {23} 28. Rea1 {-2.17/30 0} Kg7 {21} 29. Kh2 {-1.32/21 2} Kf6 {12}
30. Kg3 {-1.80/25 14} Nh5+ {9} 31. Kg4 {-1.92/26 1} Ng7 {5} 32. R1a3 {-1.95/27
11} Rb2 {43} 33. Kf4 {-2.35/29 0} Rxf2 {12} 34. Ke3 {-2.53/29 0} Rh2 {8} 35.
Bd3 {-2.70/30 5} Rxa4 {3} 36. Rxa4 {-2.76/28 3} Rxh4 {3} 37. Ra2 {-2.78/29 1}
Nf5+ {9} 38. Bxf5 {-2.30/26 1} Kxf5 {3} 39. Ra7 {-2.03/24 2} f6 {6} 40. Re7 {
-2.77/28 7} g5 {6} 41. Rf7 {-3.70/26 4} Rh2 {14} 42. Kd3 {-4.06/29 0} Rf2 {8}
43. Ke3 {-2.72/23 2} Rb2 {2} 44. Rf8 {-3.12/27 2} Rb3+ {3} 45. Ke2 {-3.25/27 1}
Ra3 {15} 46. Rf7 {-3.51/28 4} e5 {47} 47. dxe5 {-4.59/34 0} Kxe5 {1} 48. Kf2 {
-4.58/29 2} f5 {12} 49. Rd7 {-5.06/25 3} Kf4 {10} 50. Rxd5 {-6.16/29 1} Rxf3+ {
2} 51. Kg2 {-6.94/26 2} Kg4 {10} 52. Ra5 {-8.65/22 3} Rb3 {4} 53. Kf2 {-14.21/
24 1} Rb2+ {6} 0-1

[/pgn]

[Uri Blass]

<snipped>

2100 FIDE elo, 2200 Bulgarian (basically same as FIDE) from... 10 years ago, when I played my last competitive game.

Though in my last tournament my performance was 2400.

Cannot find your name in the fide list and they certainly should have also non active players who played 10 years ago unless your name is written
Ljudmil Tzvetkov and in this case you have a fide rating of 2095.

https://ratings.fide.com/card.phtml?event=2905850

If this is your name then
in your last tournament in 2006 your score was 1 out of 4 and you lost 16 rating points.