Komodo Dragon vs GM Wirig knight odds match

[lkaufman]

Here is the link to watch the live broadcast on Wednesday and Thursday:

[Chessqueen]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

[lkaufman]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

[Vinvin]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

[lkaufman]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

OK, now I understand, I wrongly thought that he was giving you knight odds! How many cores and threads was Komodo 12 using in those games? The Dragon is about 150 elo improved from Komodo 12 at Rapid on the FastGM rating list, will have 32 cores, and will use MCTS and Contempt to make the play more challenging for a human, so expect much more interesting games than you saw with Komodo 12. Whether that will be enough to avoid losing most of the games is hard to say.

[Vinvin]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

OK, now I understand, I wrongly thought that he was giving you knight odds! How many cores and threads was Komodo 12 using in those games? The Dragon is about 150 elo improved from Komodo 12 at Rapid on the FastGM rating list, will have 32 cores, and will use MCTS and Contempt to make the play more challenging for a human, so expect much more interesting games than you saw with Komodo 12. Whether that will be enough to avoid losing most of the games is hard to say.

I used my brand new AMD 5950x, 16 threads@4 GHz.
I set the contempt to 150.
The time control for the 3 games was 10m+5s but for the engine, I set the TC to 5m+3s to have time to operate.

[lkaufman]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

OK, now I understand, I wrongly thought that he was giving you knight odds! How many cores and threads was Komodo 12 using in those games? The Dragon is about 150 elo improved from Komodo 12 at Rapid on the FastGM rating list, will have 32 cores, and will use MCTS and Contempt to make the play more challenging for a human, so expect much more interesting games than you saw with Komodo 12. Whether that will be enough to avoid losing most of the games is hard to say.

I used my brand new AMD 5950x, 16 threads@4 GHz.
I set the contempt to 150.
The time control for the 3 games was 10m+5s but for the engine, I set the TC to 5m+3s to have time to operate.

OK, now I'm a bit worried. 32 cores vs. 16 cores is not a big deal, you set Contempt sensibly, and we'll also be setting time internally to roughly half of the allotted time. So the only major differences are the neural net and other improvements in Dragon, and the use of MCTS (I assume you were not using MCTS, it wasn't very good yet with Komodo 12). Each of these does seem to make a big difference in such games, but will they be enough? They were enough to beat Nakamura by 6.5 to 1.5 giving him two "big" pawns each game, but knight odds is nearly double that handicap.

[Chessqueen]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

OK, now I understand, I wrongly thought that he was giving you knight odds! How many cores and threads was Komodo 12 using in those games? The Dragon is about 150 elo improved from Komodo 12 at Rapid on the FastGM rating list, will have 32 cores, and will use MCTS and Contempt to make the play more challenging for a human, so expect much more interesting games than you saw with Komodo 12. Whether that will be enough to avoid losing most of the games is hard to say.

I used my brand new AMD 5950x, 16 threads@4 GHz.
I set the contempt to 150.
The time control for the 3 games was 10m+5s but for the engine, I set the TC to 5m+3s to have time to operate.

OK, now I'm a bit worried. 32 cores vs. 16 cores is not a big deal, you set Contempt sensibly, and we'll also be setting time internally to roughly half of the allotted time. So the only major differences are the neural net and other improvements in Dragon, and the use of MCTS (I assume you were not using MCTS, it wasn't very good yet with Komodo 12). Each of these does seem to make a big difference in such games, but will they be enough? They were enough to beat Nakamura by 6.5 to 1.5 giving him two "big" pawns each game, but knight odds is nearly double that handicap.

Nothing to worry about, I still believe that Komodo MCTS will at least get 3 draws and 2 wins, simply eliminate those bad openings and select the best opening for Komodo MCTS and keep it a little bit aggressive to make him feel the pressure

[Uri Blass]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

OK, now I understand, I wrongly thought that he was giving you knight odds! How many cores and threads was Komodo 12 using in those games? The Dragon is about 150 elo improved from Komodo 12 at Rapid on the FastGM rating list, will have 32 cores, and will use MCTS and Contempt to make the play more challenging for a human, so expect much more interesting games than you saw with Komodo 12. Whether that will be enough to avoid losing most of the games is hard to say.

I used my brand new AMD 5950x, 16 threads@4 GHz.
I set the contempt to 150.
The time control for the 3 games was 10m+5s but for the engine, I set the TC to 5m+3s to have time to operate.

OK, now I'm a bit worried. 32 cores vs. 16 cores is not a big deal, you set Contempt sensibly, and we'll also be setting time internally to roughly half of the allotted time. So the only major differences are the neural net and other improvements in Dragon, and the use of MCTS (I assume you were not using MCTS, it wasn't very good yet with Komodo 12). Each of these does seem to make a big difference in such games, but will they be enough? They were enough to beat Nakamura by 6.5 to 1.5 giving him two "big" pawns each game, but knight odds is nearly double that handicap.

Nothing to worry about, I still believe that Komodo MCTS will at least get 3 draws and 2 wins, simply eliminate those bad openings and select the best opening for Komodo MCTS and keep it a little bit aggressive to make him feel the pressure

I do not believe it and I expect wirig to get at least 5 wins out of 8 games.

[lkaufman]

Here is the link to watch the live broadcast on Wednesday and Thursday:

I see that GM Wirig is beating Human rated around 2000 with Knight Odds, I wonder if he can beat any engine rated around 1950 to 2000 , who will not Blunder and be intimidated by his Blitz rating. He can try to give Knight Odds to Honey14 set up at the Expert level

Note: I offered to play against him with Knight Odds and he did not selected me, he must be scared.

I looked at his show today, but I don't see any games won by White giving knight odds. I see three games at 10' + 5" in which he gave knight odds to (I believe) Vincent Lejeune (a.k.a. vinvin), FIDE 1941, lichess blitz 2225 (vs 2840 for Wirig a.k.a. Bizoo), but Black (vinvin) won all three rather easily. Then he played some 1' + 1" with SF level 8 (which is basically SF 13 without NNUE taking one second per move) at knight odds and again Black won all games I believe. So just based on this, knight odds looks ridiculously large and Dragon should lose every game, but we know from the experience of the Toth match that it is not so easy for Black to win against the dragon as it would appear from these games played by Wirig (both giving and receiving the odds). If you saw him win some games giving knight odds please let me know where you saw them.
By the way, since knight odds is associated with Paul Morphy, who was able to give knight odds successfully to any American player from about 1857 to 1865, it is interesting to point out that computer analysis of his games suggests that he played at roughly the same level as the FIDE rating of Wirig, so we can look at this as an attempt to emulate a match in which Dragon gave knight odds to Morphy!

Hi Larry,
Yesterday, Anthony wanted to train for the match.
The first thing was to get used to the chess.com interface because Antony mostly plays on Lichess.
The second thing was to train with the knight handicap : I propose him to play against Komodo 12 (the strongest Komodo freeware), so I operated the engine manually against Antho.
I can say he improved a lot in a couple of months against engine with handicap. He uses some systems with the help of his very good tactical understanding to play solid, not take risks, forces exchanges. He won 3-0 quite easily. I hope the latest version of Dragon will be stronger than Komodo 12 to see some interesting games !

OK, now I understand, I wrongly thought that he was giving you knight odds! How many cores and threads was Komodo 12 using in those games? The Dragon is about 150 elo improved from Komodo 12 at Rapid on the FastGM rating list, will have 32 cores, and will use MCTS and Contempt to make the play more challenging for a human, so expect much more interesting games than you saw with Komodo 12. Whether that will be enough to avoid losing most of the games is hard to say.

I used my brand new AMD 5950x, 16 threads@4 GHz.
I set the contempt to 150.
The time control for the 3 games was 10m+5s but for the engine, I set the TC to 5m+3s to have time to operate.

OK, now I'm a bit worried. 32 cores vs. 16 cores is not a big deal, you set Contempt sensibly, and we'll also be setting time internally to roughly half of the allotted time. So the only major differences are the neural net and other improvements in Dragon, and the use of MCTS (I assume you were not using MCTS, it wasn't very good yet with Komodo 12). Each of these does seem to make a big difference in such games, but will they be enough? They were enough to beat Nakamura by 6.5 to 1.5 giving him two "big" pawns each game, but knight odds is nearly double that handicap.

Nothing to worry about, I still believe that Komodo MCTS will at least get 3 draws and 2 wins, simply eliminate those bad openings and select the best opening for Komodo MCTS and keep it a little bit aggressive to make him feel the pressure

I do not believe it and I expect wirig to get at least 5 wins out of 8 games.

I agree with Uri. The three games he played with Komodo 12 were really easy wins. Of course Dragon should do better, but better than losing with no chances could still be just losing. I think that his high blitz ratings mean that he won't blunder, he is clearly prepared, and it's very difficult to get an opening where there is something for White to play for if Black chooses his defenses wisely. I just hope the games are somewhat competitive.