Komodo Dragon vs GM Ben Finegold knight odds match.

[lkaufman]

Ben Finegold, age 52, is FIDE rated 2446, FIDE blitz rated 2501 (no FIDE rapid rating), and chess.com Blitz 2802. His peak FIDE rating was 2563 in 2006, he was among the very best blitz players in the U.S. for many years, and based on his chess.com blitz rating is still a very strong grandmaster in fast games. He is also well known for his entertaining commentary as a streamer.

A VERY large part of a players strength is his knowing what to do in certain types of positions....tabiyas, priyome's, patterns - whatever you want to call them. The stronger the player, the more he can rely on these tools.

So isn't it certain in these 'odds' exhibitions that a human opponents real 'strength' is far less than ones published rating - FIDE, Blitz, etc.?

Well, "strength" is only relative to something, not absolute. If you are saying that a 2500 elo player is not 500 elo stronger than a 2000 elo player in modified circumstances such as odds play, well, I would only agree if the modification is severe, such as playing "crazyhouse chess" for example; we know that in chess960, the superiority of top players is no less than it is in normal chess. If you are saying that all humans play worse as the game becomes less familiar, that is true, and the extent depends on how weird the circumstances are. If the handicap was something bizarre such as that the engine couldn't make "long" moves or more than 3 checks or had to mate in certain ways, then it would diminish the humans strength. If we start with a weird position, that does diminish the humans strength; I would bet a lot of money on this match if it were FRC with knight odds (as we did with Lenderman in 2000). Even the classic handicap of "pawn and two moves" (f7 removed, White plays e4 and a second move of his choice) is a bit tough for humans as they are not familiar with how to exploit the weakened king at the start, which is probably why no human ever beat Komodo at this handicap in a public game. But knowing what to do with an extra piece is the most fundamental piece of knowledge in the arsenal of any strong chess player, and starting a game with all the pieces on their normal squares but with White having "blundered" a knight is the most normal, easily exploitable advantage I can imagine for a human master. He can play any normal opening, he can place his pieces as in a normal game, and he has already won a piece. In fact all tests suggest that human masters/grandmasters are better at winning with knight odds than are engines of similar playing strength.
So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

So if it was FRC Komodo Dragon 3/6/1 could probably give GM Finegold a Rook Odds and Draw. Ask him to play 2 extra games under that condition?

No, rook odds is noticeably more than knight odds, I think that even FRC rook odds is more than standard knight odds. It's surprising how often Dragon can pull off a draw in an endgame with just a pawn or two for the knight, but with a couple pawns for a rook it would lose all endgames.

[Chessqueen]

Ben Finegold, age 52, is FIDE rated 2446, FIDE blitz rated 2501 (no FIDE rapid rating), and chess.com Blitz 2802. His peak FIDE rating was 2563 in 2006, he was among the very best blitz players in the U.S. for many years, and based on his chess.com blitz rating is still a very strong grandmaster in fast games. He is also well known for his entertaining commentary as a streamer.

A VERY large part of a players strength is his knowing what to do in certain types of positions....tabiyas, priyome's, patterns - whatever you want to call them. The stronger the player, the more he can rely on these tools.

So isn't it certain in these 'odds' exhibitions that a human opponents real 'strength' is far less than ones published rating - FIDE, Blitz, etc.?

Well, "strength" is only relative to something, not absolute. If you are saying that a 2500 elo player is not 500 elo stronger than a 2000 elo player in modified circumstances such as odds play, well, I would only agree if the modification is severe, such as playing "crazyhouse chess" for example; we know that in chess960, the superiority of top players is no less than it is in normal chess. If you are saying that all humans play worse as the game becomes less familiar, that is true, and the extent depends on how weird the circumstances are. If the handicap was something bizarre such as that the engine couldn't make "long" moves or more than 3 checks or had to mate in certain ways, then it would diminish the humans strength. If we start with a weird position, that does diminish the humans strength; I would bet a lot of money on this match if it were FRC with knight odds (as we did with Lenderman in 2000). Even the classic handicap of "pawn and two moves" (f7 removed, White plays e4 and a second move of his choice) is a bit tough for humans as they are not familiar with how to exploit the weakened king at the start, which is probably why no human ever beat Komodo at this handicap in a public game. But knowing what to do with an extra piece is the most fundamental piece of knowledge in the arsenal of any strong chess player, and starting a game with all the pieces on their normal squares but with White having "blundered" a knight is the most normal, easily exploitable advantage I can imagine for a human master. He can play any normal opening, he can place his pieces as in a normal game, and he has already won a piece. In fact all tests suggest that human masters/grandmasters are better at winning with knight odds than are engines of similar playing strength.
So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

So if it was FRC Komodo Dragon 3/6/1 could probably give GM Finegold a Rook Odds and Draw. Ask him to play 2 extra games under that condition?

No, rook odds is noticeably more than knight odds, I think that even FRC rook odds is more than standard knight odds. It's surprising how often Dragon can pull off a draw in an endgame with just a pawn or two for the knight, but with a couple pawns for a rook it would lose all endgames.

So a Rook Odds only can be match in FRC against a 2200 Rated player at the most? It would be interesting if the next Match can be against a 2200 players Rook Odds in FRC.

[lkaufman]

Ben Finegold, age 52, is FIDE rated 2446, FIDE blitz rated 2501 (no FIDE rapid rating), and chess.com Blitz 2802. His peak FIDE rating was 2563 in 2006, he was among the very best blitz players in the U.S. for many years, and based on his chess.com blitz rating is still a very strong grandmaster in fast games. He is also well known for his entertaining commentary as a streamer.

A VERY large part of a players strength is his knowing what to do in certain types of positions....tabiyas, priyome's, patterns - whatever you want to call them. The stronger the player, the more he can rely on these tools.

So isn't it certain in these 'odds' exhibitions that a human opponents real 'strength' is far less than ones published rating - FIDE, Blitz, etc.?

Well, "strength" is only relative to something, not absolute. If you are saying that a 2500 elo player is not 500 elo stronger than a 2000 elo player in modified circumstances such as odds play, well, I would only agree if the modification is severe, such as playing "crazyhouse chess" for example; we know that in chess960, the superiority of top players is no less than it is in normal chess. If you are saying that all humans play worse as the game becomes less familiar, that is true, and the extent depends on how weird the circumstances are. If the handicap was something bizarre such as that the engine couldn't make "long" moves or more than 3 checks or had to mate in certain ways, then it would diminish the humans strength. If we start with a weird position, that does diminish the humans strength; I would bet a lot of money on this match if it were FRC with knight odds (as we did with Lenderman in 2000). Even the classic handicap of "pawn and two moves" (f7 removed, White plays e4 and a second move of his choice) is a bit tough for humans as they are not familiar with how to exploit the weakened king at the start, which is probably why no human ever beat Komodo at this handicap in a public game. But knowing what to do with an extra piece is the most fundamental piece of knowledge in the arsenal of any strong chess player, and starting a game with all the pieces on their normal squares but with White having "blundered" a knight is the most normal, easily exploitable advantage I can imagine for a human master. He can play any normal opening, he can place his pieces as in a normal game, and he has already won a piece. In fact all tests suggest that human masters/grandmasters are better at winning with knight odds than are engines of similar playing strength.
So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

So if it was FRC Komodo Dragon 3/6/1 could probably give GM Finegold a Rook Odds and Draw. Ask him to play 2 extra games under that condition?

No, rook odds is noticeably more than knight odds, I think that even FRC rook odds is more than standard knight odds. It's surprising how often Dragon can pull off a draw in an endgame with just a pawn or two for the knight, but with a couple pawns for a rook it would lose all endgames.

So a Rook Odds only can be match in FRC against a 2200 Rated player at the most? It would be interesting if the next Match can be against a 2200 players Rook Odds in FRC.

Well, maybe 2300 FIDE for Rooks odds in FRC Rapid, 2200 for Rook odds in standard chess would be about fair I would say.

[Chessqueen]

It is interesting what you mention, about 2 pawns versus a Knight but it all depend on the position and who play first for instance, this is a draw weather Black play first or White play first.

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2022.01.08"]
[Round "?"]
[White "Dragon-2.6-64bit-avx2"]
[Black "Dragon-2.6-64bit-avx2"]
[Result "*"]
[BlackElo "2000"]
[Time "19:40:33"]
[WhiteElo "2000"]
[TimeControl "180+2"]
[SetUp "1"]
[FEN "8/8/8/p6K/4N1p1/8/k7/8 b - - 0 1"]
[Termination "unterminated"]
[PlyCount "6"]
[WhiteType "program"]
[BlackType "program"]

1. ... a4 {(a5-a4 Kh5xg4 a4-a3) +5.69/46 21} 2. Kxg4 {(Kh5xg4 a4-a3 Ne4-d2
Ka2-a1 Kg4-g3 a3-a2 Nd2-c4) 0.00/64 7} a3 {(a4-a3 Ne4-d2 Ka2-a1 Kg4-g3
a3-a2 Nd2-c4) 0.00/75 10} 3. Nd2 {(Ne4-d2 Ka2-a1 Kg4-g3 a3-a2 Nd2-c4)
0.00/76 4} Ka1 {(Ka2-a1 Nd2-c4 a3-a2 Nc4-a3) 0.00/78 5} 4. Nc4 {(Nd2-c4
a3-a2 Nc4-a3) 0.00/80 27} *[/pgn]

[Chessqueen]

Here is if White play first, so it all depends on the position.

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2022.01.08"]
[Round "?"]
[White "Dragon-2.6-64bit-avx2"]
[Black "Dragon-2.6-64bit-avx2"]
[Result "*"]
[BlackElo "2000"]
[Time "19:44:48"]
[WhiteElo "2000"]
[TimeControl "180+2"]
[SetUp "1"]
[FEN "8/8/8/p6K/4N1p1/8/k7/8 w - - 0 1"]
[Termination "unterminated"]
[PlyCount "14"]
[WhiteType "program"]
[BlackType "program"]

1. Kxg4 {(Kh5xg4 Ka2-a3 Kg4-h5 Ka3-b4 Ne4-d2 a5-a4 Nd2-b1 Kb4-a5 Kh5-g6
Ka5-b4 Kg6-f5 Kb4-b3 Kf5-g5 Kb3-c2 Nb1-a3+ Kc2-b2 Na3-b5 Kb2-b3 Kg5-f5
Kb3-b4 Nb5-d4 Kb4-c5 Nd4-c2 Kc5-c4 Kf5-e4 Kc4-c3 Nc2-d4 Kc3-c4 Nd4-c2)
0.00/47 3} Ka3 {(Ka2-a3 Kg4-h5 Ka3-b4 Ne4-d2 a5-a4 Nd2-b1 Kb4-a5 Kh5-g6
Ka5-b4 Kg6-f5 Kb4-b3 Kf5-g5 Kb3-c2 Nb1-a3+ Kc2-b2 Na3-b5 Kb2-b3 Kg5-f5
Kb3-b4 Nb5-d4 Kb4-c5 Nd4-c2 Kc5-c4 Kf5-e4 Kc4-c3 Nc2-d4 Kc3-c4 Nd4-c2)
0.00/52 4} 2. Kh5 {(Kg4-h5 Ka3-b4 Ne4-d2 a5-a4 Nd2-b1 Kb4-a5 Kh5-g6 Ka5-b4
Kg6-f5 Kb4-b3 Kf5-g5 Kb3-c2 Nb1-a3+ Kc2-b2 Na3-b5 Kb2-b3 Kg5-h6 Kb3-b4
Nb5-d4 Kb4-c4 Nd4-c2 Kc4-c3 Nc2-a3 Kc3-b3 Na3-b5) 0.00/56 4} Kb4 {(Ka3-b4
Ne4-d2 a5-a4 Nd2-b1 Kb4-a5 Kh5-g6 Ka5-b4 Kg6-f5 Kb4-b3 Kf5-g5 Kb3-c2
Nb1-a3+ Kc2-b2 Na3-b5 Kb2-b3 Kg5-h6 Kb3-b4 Nb5-d4 Kb4-c4 Nd4-c2 Kc4-c3
Nc2-a3 Kc3-b3 Na3-b5) 0.00/63 4} 3. Nd2 {(Ne4-d2 a5-a4 Nd2-b1 Kb4-a5 Kh5-g6
Ka5-b4 Kg6-f5 Kb4-b3 Nb1-d2+ Kb3-c3 Nd2-b1+ Kc3-b2 Nb1-d2 Kb2-c3) 0.00/69
4} a4 {(a5-a4 Nd2-b1 Kb4-a5 Nb1-a3 Ka5-b4 Na3-b1) 0.00/70 3} 4. Nb1
{(Nd2-b1 Kb4-a5 Nb1-a3 Ka5-b4 Na3-b1) 0.00/74 6} Ka5 {(Kb4-a5 Nb1-a3 Ka5-b4
Na3-b1) 0.00/74 8} 5. Na3 {(Nb1-a3 Ka5-b4 Na3-b1 Kb4-a5) 0.00/76 6} Kb4
{(Ka5-b4 Na3-c2+ Kb4-a5 Nc2-a3) 0.00/65 5} 6. Nc2+ {(Na3-c2+ Kb4-a5 Kh5-g5
Ka5-b6 Kg5-g4 Kb6-a6 Kg4-f5 Ka6-a7 Kf5-g5 Ka7-b7 Kg5-f4 Kb7-a6 Kf4-f5)
0.00/70 3} Ka5 {(Kb4-a5 Nc2-a3 Ka5-b4 Na3-c2+) 0.00/74 3} 7. Na3 {(Nc2-a3
Ka5-b4 Na3-c2+ Kb4-a5) 0.00/78 15} Kb4 {(Ka5-b4 Na3-c2+ Kb4-a5 Nc2-a3)
0.00/77 3} *[/pgn]

[Cornfed]

So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

Yes, it makes sense what you say about simple knight odds in relation to patterns etc. Some of the edge would go away but most of it would probably remain. It is really the other variants where the better players would be neutered somewhat by not having them at their disposal.

I can't prove this, but I would think the difference in strength (whatever your measure, elo, wins, etc) would shrink considerably between myself and Carlsen in variants compared to what it is in 'normal chess.

[Chessqueen]

So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

Yes, it makes sense what you say about simple knight odds in relation to patterns etc. Some of the edge would go away but most of it would probably remain. It is really the other variants where the better players would be neutered somewhat by not having them at their disposal.

I can't prove this, but I would think the difference in strength (whatever your measure, elo, wins, etc) would shrink considerably between myself and Carlsen in variants compared to what it is in 'normal chess.

Probably a player of your caliber if Chess Opening are given to you up to move 20 against Carlsen you can probably last another extra 15 to 20 moves before being mated

Note: Try against Komodo Dragon 2.6 set the UCI Elo = 2850, but play an opening up to move 20 and see how long you will last. Or let Komodo Dragon play itself the first 20 moves.

[Cornfed]

So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

Yes, it makes sense what you say about simple knight odds in relation to patterns etc. Some of the edge would go away but most of it would probably remain. It is really the other variants where the better players would be neutered somewhat by not having them at their disposal.

I can't prove this, but I would think the difference in strength (whatever your measure, elo, wins, etc) would shrink considerably between myself and Carlsen in variants compared to what it is in 'normal chess.

Probably a player of your caliber if Chess Opening are given to you up to move 20 against Carlsen you can probably last another extra 15 to 20 moves before being mated

Note: Try against Komodo Dragon 2.6 set the UCI Elo = 2850, but play an opening up to move 20 and see how long you will last. Or let Komodo Dragon play itself the first 20 moves.

I don't have Dragon 2.6, so no...

Actually, if deGroots investigations are anywhere near right, when it comes to variants where pieces are not on their natural squares (as in good ol' regular chess) many of the things that go toward making a top flight chess player (Magnus, for example...and he has a great memory) go out the window with the randomization. I certainly think I would 'last' more than another 15-20 moves...but again, I can't prove it.

[Chessqueen]

Ben Finegold, age 52, is FIDE rated 2446, FIDE blitz rated 2501 (no FIDE rapid rating), and chess.com Blitz 2802. His peak FIDE rating was 2563 in 2006, he was among the very best blitz players in the U.S. for many years, and based on his chess.com blitz rating is still a very strong grandmaster in fast games. He is also well known for his entertaining commentary as a streamer.

A VERY large part of a players strength is his knowing what to do in certain types of positions....tabiyas, priyome's, patterns - whatever you want to call them. The stronger the player, the more he can rely on these tools.

So isn't it certain in these 'odds' exhibitions that a human opponents real 'strength' is far less than ones published rating - FIDE, Blitz, etc.?

Well, "strength" is only relative to something, not absolute. If you are saying that a 2500 elo player is not 500 elo stronger than a 2000 elo player in modified circumstances such as odds play, well, I would only agree if the modification is severe, such as playing "crazyhouse chess" for example; we know that in chess960, the superiority of top players is no less than it is in normal chess. If you are saying that all humans play worse as the game becomes less familiar, that is true, and the extent depends on how weird the circumstances are. If the handicap was something bizarre such as that the engine couldn't make "long" moves or more than 3 checks or had to mate in certain ways, then it would diminish the humans strength. If we start with a weird position, that does diminish the humans strength; I would bet a lot of money on this match if it were FRC with knight odds (as we did with Lenderman in 2000). Even the classic handicap of "pawn and two moves" (f7 removed, White plays e4 and a second move of his choice) is a bit tough for humans as they are not familiar with how to exploit the weakened king at the start, which is probably why no human ever beat Komodo at this handicap in a public game. But knowing what to do with an extra piece is the most fundamental piece of knowledge in the arsenal of any strong chess player, and starting a game with all the pieces on their normal squares but with White having "blundered" a knight is the most normal, easily exploitable advantage I can imagine for a human master. He can play any normal opening, he can place his pieces as in a normal game, and he has already won a piece. In fact all tests suggest that human masters/grandmasters are better at winning with knight odds than are engines of similar playing strength.
So I guess I would say that I agree with your statement in general, but not in the specific case at hand, classic knight odds.

So if it was FRC Komodo Dragon 3/6/1 could probably give GM Finegold a Rook Odds and Draw. Ask him to play 2 extra games under that condition?

No, rook odds is noticeably more than knight odds, I think that even FRC rook odds is more than standard knight odds. It's surprising how often Dragon can pull off a draw in an endgame with just a pawn or two for the knight, but with a couple pawns for a rook it would lose all endgames.

So a Rook Odds only can be match in FRC against a 2200 Rated player at the most? It would be interesting if the next Match can be against a 2200 players Rook Odds in FRC.

Well, maybe 2300 FIDE for Rooks odds in FRC Rapid, 2200 for Rook odds in standard chess would be about fair I would say.

I know that most people only are interested in seeing Odds versus GMs, but it would be nice to see how Komodo Dragon 2.6.1 does in a future event
maybe 2300 FIDE for Rooks odds in FRC Rapid, since we have seeing many many Knight Odds against GMs already.

Note: I predict Komodo Dragon 2.6.1 vs GM Finegold to draw with 2 wins 2 loses and 2 draws next Saturday.

[Cornfed]

I know that most people only are interested in seeing Odds versus GMs, but it would be nice to see how Komodo Dragon 2.6.1 does in a future event
maybe 2300 FIDE for Rooks odds in FRC Rapid, since we have seeing many many Knight Odds against GMs already.

GM's...gravitate towards 2300 as they get older and less active in serious competitive competition.