Komodo Dragon Vs. GM Perelshteyn Where is the Link ?

[lkaufman]

I am almost positive that Komodo dragon can give Carlsen, MVL or Caruana the f2 pawn Odds at T/C 15'+10"

This is a weird comment. Don't you remember that Dragon already defeated Nakamura (who is probably better than two of the three named players in Rapid) in four Rapid (15' + 10") games at f2 + b2 or c2 (and another four at g2 + b2 or c2) odds? If Dragon can win with f2 plus another non-edge pawn in a match, why even talk about f2 handicap?

[Chessqueen]

I am almost positive that Komodo dragon can give Carlsen, MVL or Caruana the f2 pawn Odds at T/C 15'+10"

This is a weird comment. Don't you remember that Dragon already defeated Nakamura (who is probably better than two of the three named players in Rapid) in four Rapid (15' + 10") games at f2 + b2 or c2 (and another four at g2 + b2 or c2) odds? If Dragon can win with f2 plus another non-edge pawn in a match, why even talk about f2 handicap?

I do NOT believe that Nakamura is better than Carlsen at Rapid, but that is my Opinion, but Carlsen has not played versus Nakamura at 15'+10" enough games to verify it.

[lkaufman]

I am almost positive that Komodo dragon can give Carlsen, MVL or Caruana the f2 pawn Odds at T/C 15'+10"

This is a weird comment. Don't you remember that Dragon already defeated Nakamura (who is probably better than two of the three named players in Rapid) in four Rapid (15' + 10") games at f2 + b2 or c2 (and another four at g2 + b2 or c2) odds? If Dragon can win with f2 plus another non-edge pawn in a match, why even talk about f2 handicap?

I do NOT believe that Nakamura is better than Carlsen at Rapid, but that is mu Opinion, but might be close to Carlsen, but C arlsen has not played versus Nakamura at 15'+10" to verify it.

Of course they have played many games at that time control. I don't know the exact score, I think it's close with a slight Carlsen lead. But you named three players, and Naka is pretty clearly stronger than two of those three at that time control, and clearly close to Carlsen.

[Chessqueen]

I am almost positive that Komodo dragon can give Carlsen, MVL or Caruana the f2 pawn Odds at T/C 15'+10"

This is a weird comment. Don't you remember that Dragon already defeated Nakamura (who is probably better than two of the three named players in Rapid) in four Rapid (15' + 10") games at f2 + b2 or c2 (and another four at g2 + b2 or c2) odds? If Dragon can win with f2 plus another non-edge pawn in a match, why even talk about f2 handicap?

I do NOT believe that Nakamura is better than Carlsen at Rapid, but that is mu Opinion, but might be close to Carlsen, but C arlsen has not played versus Nakamura at 15'+10" to verify it.

Of course they have played many games at that time control. I don't know the exact score, I think it's close with a slight Carlsen lead. But you named three players, and Naka is pretty clearly stronger than two of those three at that time control, and clearly close to Carlsen.

Therefore, Velvet which is rated 2863 by CCRL is better than Nakamura, but of course we need at least 1000 games

105 MinkoChess 1.3 64-bit 4CPU 2884 +21 −21 46.0% +24.8 43.7% 725
87.5%
106 Velvet 2.0.0 64-bit 2863 +28 −28 47.9% +13.9 43.7% 403
62

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.11.19"]
[Round "?"]
[White "Dragon-2.5-64bit-avx2"]
[Black "Velvet-v2.0.0-x86_64-avx2"]
[Result "1/2-1/2"]
[BlackElo "2860"]
[Time "17:33:24"]
[WhiteElo "2860"]
[TimeControl "180+3"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PP1PP1PP/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "45"]
[WhiteType "program"]
[BlackType "program"]

1. d4 e6 2. Nf3 Nf6 3. Nc3 d5 4. Bg5 h6 5. Bh4 Be7 6. e3 O-O 7. Bd3 c5 8.
O-O Nc6 9. Qe1 cxd4 10. exd4 Qb6 11. Kh1 Nxd4 12. Nxd4 Qxd4 13. Rd1 Kh8 14.
Bb1 Qb4 15. a3 Qb6 16. Bxf6 Bxf6 17. Rxf6 gxf6 18. Na4 Qd6 19. Qh4 Kg7 20.
Qg4+ Kh8 21. Qh4 Kg7 22. Qg4+ Kh8 23. Qh4 {3-fold repetition} 1/2-1/2[/pgn]

[lkaufman]

[quote=lkaufman post_id=912390 time=<a href="tel:1637363853">1637363853</a> user_id=4773]
[quote=Chessqueen post_id=912389 time=<a href="tel:1637363590">1637363590</a> user_id=10732]
[quote=lkaufman post_id=912376 time=<a href="tel:1637355483">1637355483</a> user_id=4773]
[quote=Chessqueen post_id=912355 time=<a href="tel:1637347201">1637347201</a> user_id=10732]
I am almost positive that Komodo dragon can give Carlsen, MVL or Caruana the f2 pawn Odds at T/C 15'+10"

This is a weird comment. Don't you remember that Dragon already defeated Nakamura (who is probably better than two of the three named players in Rapid) in four Rapid (15' + 10") games at f2 + b2 or c2 (and another four at g2 + b2 or c2) odds? If Dragon can win with f2 plus another non-edge pawn in a match, why even talk about f2 handicap?
[/quote]

I do NOT believe that Nakamura is better than Carlsen at Rapid, but that is mu Opinion, but might be close to Carlsen, but C arlsen has not played versus Nakamura at 15'+10" to verify it.
[/quote]

Of course they have played many games at that time control. I don't know the exact score, I think it's close with a slight Carlsen lead. But you named three players, and Naka is pretty clearly stronger than two of those three at that time control, and clearly close to Carlsen.
[/quote]

Therefore, Velvet which is rated 2863 by CCRL is better than Nakamura, but of course we need at least 1000 games

105 MinkoChess 1.3 64-bit 4CPU 2884 +21 −21 46.0% +24.8 43.7% 725
87.5%
106 Velvet 2.0.0 64-bit 2863 +28 −28 47.9% +13.9 43.7% 403
62

[pgn][Event "Computer chess game"]
[Site "MININT-UB2PIMJ"]
[Date "2021.11.19"]
[Round "?"]
[White "Dragon-2.5-64bit-avx2"]
[Black "Velvet-v2.0.0-x86_64-avx2"]
[Result "1/2-1/2"]
[BlackElo "2860"]
[Time "17:33:24"]
[WhiteElo "2860"]
[TimeControl "180+3"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PP1PP1PP/RNBQKBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "45"]
[WhiteType "program"]
[BlackType "program"]

1. d4 e6 2. Nf3 Nf6 3. Nc3 d5 4. Bg5 h6 5. Bh4 Be7 6. e3 O-O 7. Bd3 c5 8.
O-O Nc6 9. Qe1 cxd4 10. exd4 Qb6 11. Kh1 Nxd4 12. Nxd4 Qxd4 13. Rd1 Kh8 14.
Bb1 Qb4 15. a3 Qb6 16. Bxf6 Bxf6 17. Rxf6 gxf6 18. Na4 Qd6 19. Qh4 Kg7 20.
Qg4+ Kh8 21. Qh4 Kg7 22. Qg4+ Kh8 23. Qh4 {3-fold repetition} 1/2-1/2
[/quote]

I am almost positive that Komodo dragon can give Carlsen, MVL or Caruana the f2 pawn Odds at T/C 15'+10"

This is a weird comment. Don't you remember that Dragon already defeated Nakamura (who is probably better than two of the three named players in Rapid) in four Rapid (15' + 10") games at f2 + b2 or c2 (and another four at g2 + b2 or c2) odds? If Dragon can win with f2 plus another non-edge pawn in a match, why even talk about f2 handicap?

I do NOT believe that Nakamura is better than Carlsen at Rapid, but that is my Opinion, but Carlsen has not played versus Nakamura at 15'+10" enough games to verify it.

Probably velvet 2 is indeed stronger than Naka or even Carlsen in rapid; ccrl ratings were roughly in line with humans in Rapid on the old (circa 2008) hardware at top human level but not on a modern i7[/pgn]

[Chessqueen]

Probably velvet 2 is indeed stronger than Naka or even Carlsen in rapid; ccrl ratings were roughly in line with humans in Rapid on the old (circa 2008) hardware at top human level but not on a modern i7
[/quote]

So you are saying that the Rapid ratings needs to be adjusted by CCRL based on the Modern i7, I can see why velvet after 140 games so far is performing so good, I set the contempt of Komodo Dragon to 45 with 2 pawns odds, probably I need a different contempt based on 2 pawns odds

[lkaufman]

Probably velvet 2 is indeed stronger than Naka or even Carlsen in rapid; ccrl ratings were roughly in line with humans in Rapid on the old (circa 2008) hardware at top human level but not on a modern i7

So you are saying that the Rapid ratings needs to be adjusted by CCRL based on the Modern i7, I can see why velvet after 140 games so far is performing so good, I set the contempt of Komodo Dragon to 45 with 2 pawns odds, probably I need a different contempt based on 2 pawns odds
[/quote]

Yes; CCRL Rapid list used to be 40/40 on ancient hardware. Presumably the ratings were supposed to be about right in human terms, at least at the top human level (say 2800). Then they changed it to 40/15 on modern i7 hardware because the modern hardware was roughly 40/15 times faster. So the engines play at the same strength now at 40/15 as they did on the old hardware at 40/40. But the ratings implied that the human also had 40/40 (assuming they were accurate in human terms at 2800). But a human plays much worse at 40/15 (roughly same as 15' + 10" inc) than at 40/40, probably more than a hundred elo worse. So if the same engine is now playing the same human at 40/15 on an i7 instead of at 40/40 on the old hardware, the results should swing in the engine's favor by more than a hundred elo. So even allowing for the deflation of human FIDE ratings in recent years, I would say that CCRL should have raised their ratings by 100 elo when they switched the reference hardware, if the goal was parity with human GM play. But I don't think this was a priority for CCRL. There is also the issue that CCRL ratings need to be contracted by some percentage (i now estimate 25%) for predicting human rating equivalence, so this means that lower down the scale, the ratings need to be raised by much more than 100 for estimating the human equivalent.