Hikaru vs. bots

[lkaufman]

Today chess.com had an event in which participants played against many of their chess bots, which are basically various Komodo (version 13.3 I think) Skill levels modified in various ways. Levels 19 and below use evals from NNs that are designed to make them more "human", but the bots rated over 2400 use the Komode evals with Skill levels 20 thru 22 and a few parameter mods. I watched most of the games by Hikaru Nakamura against the bots. There was no time limit per game, but as prizes were based on how many bots were beaten, there was an incentive to play quickly and to resign bad or drawn positions; I would estimate the rate of play was similar to 3' + 2" games, with the engine playing at hyper-bullet speed. As you would expect, Hikaru beat all the bots below level 20, though in a couple cases he didn't win on the first try, at least at level 19. But he had a tough time with the Skill 20 bots; I think he eventually beat them all, but probably lost more than he won, even though they were only doing seven ply searches and were slightly weakened by other parameter changes. Against the level 21 bots, he could not score against one of them, and didn't win right away against others. Against level 22 bots (9 ply), he beat the "Morphy" bot, but lost all games to his own Hikaru bot! Overall the bots gave him a tougher time than I would have predicted. It seems that his closest match was level 20.
So I decided to determine approximate CCRL blitz ratings for these skill levels, based on long 2' + 1" blitz matches with Arasan 14, which is about 2585 CCRL blitz. Skill level 23 came out 2543 based on this, level 22 was 2382, level 21 was 2196, and level 20 was too weak to test this way but by extrapolation should be about 2000. Given that Hikaru was playing slower than 2' + 1" and that the bots were slightly weakened by parameter changes, it seems that if he were to play for a CCRL rating by playing 2' + 1" games with CCRL rated engines on reference hardware his CCRL blitz rating might be 2000 or even lower! This is really hard to believe, he's nearly 3200 on chess.com blitz, the world's best. Are the scales really 1200 apart? Maybe I'm doing something wrong, but that's what the data suggests! That would imply that the top engines on the CCRL blitz list would be about 4900 on chess.com blitz if all the intermediate engines played there too. Hard to believe.

[Uri Blass]

I understand that draw or loss was the same for hikaru.

It is not the case in case hikaru had to play for a CCRL rating list.

Another advantage hikaru had for the ccrl rating list is that he could think on the opponent time when he cannot do it when the opponents play at hyper bullet speed(I know CCRL use no pondering but still you cannot force the opponent not to ponder in case the opponent is not an engine).

For the time control I understand that the time control in CCRL blitz is
equivalent to 2'+1" on an Intel i7-4770K

I wonder if you use the same hardware or maybe faster hardware.

[lkaufman]

I understand that draw or loss was the same for hikaru.

It is not the case in case hikaru had to play for a CCRL rating list.

Another advantage hikaru had for the ccrl rating list is that he could think on the opponent time when he cannot do it when the opponents play at hyper bullet speed(I know CCRL use no pondering but still you cannot force the opponent not to ponder in case the opponent is not an engine).

For the time control I understand that the time control in CCRL blitz is
equivalent to 2'+1" on an Intel i7-4770K

I wonder if you use the same hardware or maybe faster hardware.

Hardware for this was calibrated based on an old 16 core Xeon, presumably somewhat slower than the CCRL reference machine. So really I overrated the skill levels slightly. If Hikaru (or any human) did play games under CCRL conditions, presumably Ponder would be turned on for the machine to be fair, since it can't be turned off for the human. It's true that he got nothing for draws, so perhaps if we allow for this his fair opponent was between level 20 and 21, still only something like 2050 CCRL blitz.

[mwyoung]

Today chess.com had an event in which participants played against many of their chess bots, which are basically various Komodo (version 13.3 I think) Skill levels modified in various ways. Levels 19 and below use evals from NNs that are designed to make them more "human", but the bots rated over 2400 use the Komode evals with Skill levels 20 thru 22 and a few parameter mods. I watched most of the games by Hikaru Nakamura against the bots. There was no time limit per game, but as prizes were based on how many bots were beaten, there was an incentive to play quickly and to resign bad or drawn positions; I would estimate the rate of play was similar to 3' + 2" games, with the engine playing at hyper-bullet speed. As you would expect, Hikaru beat all the bots below level 20, though in a couple cases he didn't win on the first try, at least at level 19. But he had a tough time with the Skill 20 bots; I think he eventually beat them all, but probably lost more than he won, even though they were only doing seven ply searches and were slightly weakened by other parameter changes. Against the level 21 bots, he could not score against one of them, and didn't win right away against others. Against level 22 bots (9 ply), he beat the "Morphy" bot, but lost all games to his own Hikaru bot! Overall the bots gave him a tougher time than I would have predicted. It seems that his closest match was level 20.
So I decided to determine approximate CCRL blitz ratings for these skill levels, based on long 2' + 1" blitz matches with Arasan 14, which is about 2585 CCRL blitz. Skill level 23 came out 2543 based on this, level 22 was 2382, level 21 was 2196, and level 20 was too weak to test this way but by extrapolation should be about 2000. Given that Hikaru was playing slower than 2' + 1" and that the bots were slightly weakened by parameter changes, it seems that if he were to play for a CCRL rating by playing 2' + 1" games with CCRL rated engines on reference hardware his CCRL blitz rating might be 2000 or even lower! This is really hard to believe, he's nearly 3200 on chess.com blitz, the world's best. Are the scales really 1200 apart? Maybe I'm doing something wrong, but that's what the data suggests! That would imply that the top engines on the CCRL blitz list would be about 4900 on chess.com blitz if all the intermediate engines played there too. Hard to believe.

I believe. And have come to the same conclusions.
People seem to get taken back when I say all humans play bad chess.

Here is the reasoning. Today's chess engines play at such a level. The best engines see no difference in playing strength between the worst human players and best. As measured by results in match play.

[kasinp]

Today chess.com had an event in which participants played against many of their chess bots, which are basically various Komodo (version 13.3 I think) Skill levels modified in various ways. Levels 19 and below use evals from NNs that are designed to make them more "human", but the bots rated over 2400 use the Komode evals with Skill levels 20 thru 22 and a few parameter mods. I watched most of the games by Hikaru Nakamura against the bots. There was no time limit per game, but as prizes were based on how many bots were beaten, there was an incentive to play quickly and to resign bad or drawn positions; I would estimate the rate of play was similar to 3' + 2" games, with the engine playing at hyper-bullet speed. As you would expect, Hikaru beat all the bots below level 20, though in a couple cases he didn't win on the first try, at least at level 19. But he had a tough time with the Skill 20 bots; I think he eventually beat them all, but probably lost more than he won, even though they were only doing seven ply searches and were slightly weakened by other parameter changes. Against the level 21 bots, he could not score against one of them, and didn't win right away against others. Against level 22 bots (9 ply), he beat the "Morphy" bot, but lost all games to his own Hikaru bot! Overall the bots gave him a tougher time than I would have predicted. It seems that his closest match was level 20.
So I decided to determine approximate CCRL blitz ratings for these skill levels, based on long 2' + 1" blitz matches with Arasan 14, which is about 2585 CCRL blitz. Skill level 23 came out 2543 based on this, level 22 was 2382, level 21 was 2196, and level 20 was too weak to test this way but by extrapolation should be about 2000. Given that Hikaru was playing slower than 2' + 1" and that the bots were slightly weakened by parameter changes, it seems that if he were to play for a CCRL rating by playing 2' + 1" games with CCRL rated engines on reference hardware his CCRL blitz rating might be 2000 or even lower! This is really hard to believe, he's nearly 3200 on chess.com blitz, the world's best. Are the scales really 1200 apart? Maybe I'm doing something wrong, but that's what the data suggests! That would imply that the top engines on the CCRL blitz list would be about 4900 on chess.com blitz if all the intermediate engines played there too. Hard to believe.

I believe. And have come to the same conclusions.
People seem to get taken back when I say all humans play bad chess.

Here is the reasoning. Today's chess engines play at such a level. The best engines see no difference in playing strength between the worst human players and best. As measured by results in match play.

While this sounds like an exaggeration I tend to agree.

In 1994 Kasparov lost to a Genius 3 on P90 in rapid (not even blitz). The big surprise factor was probably removed at that point.
CG3 on P90 is rated 2313 on SSDF. So I would agree that to the best engines the difference in 2'+1" between Magnus and a 2000 player is negligible.

Peter

[jdart]

Nakamura is an awesome blitz player, but it is known for a long time that even GM games are frequently full of suboptimal moves or even blunders, especially blitz games. The engines never lose attention or make shallow-depth blunders, and won't lose on time. So it is not too surprising to me that they are hard to beat, even with somewhat weakened search.

Arasan (latest version) on an 8-core was recenty matched by GM Jakhongir Vakhidov, FIDE 2585, on ICC. He's played 17 games at 5 0 and lost all of them (several on time).

[lkaufman]

Nakamura is an awesome blitz player, but it is known for a long time that even GM games are frequently full of suboptimal moves or even blunders, especially blitz games. The engines never lose attention or make shallow-depth blunders, and won't lose on time. So it is not too surprising to me that they are hard to beat, even with somewhat weakened search.

Arasan (latest version) on an 8-core was recenty matched by GM Jakhongir Vakhidov, FIDE 2585, on ICC. He's played 17 games at 5 0 and lost all of them (several on time).

Latest Arasan is 3290 CCRL blitz, presumably low 3300s on 8 cores. The bots Naka played are not "somewhat weakened", they are searching 7 to 9 plies instead of mid twenties, with the estimated CCRL blitz rating down from 3600 to 2000-2400 depending on the bot. So a thousand or more elo below your Arasan. At depths 7 thru 9 bots make plenty of blunders too, just not the most obvious ones. I fully expect a strong engine like Arasan to win every blitz game from a GM, but imagine it limited to 7 to 9 plies?

[mhull]

Today chess.com had an event in which participants played against many of their chess bots, which are basically various Komodo (version 13.3 I think) Skill levels modified in various ways. Levels 19 and below use evals from NNs that are designed to make them more "human", but the bots rated over 2400 use the Komode evals with Skill levels 20 thru 22 and a few parameter mods. I watched most of the games by Hikaru Nakamura against the bots. There was no time limit per game, but as prizes were based on how many bots were beaten, there was an incentive to play quickly and to resign bad or drawn positions; I would estimate the rate of play was similar to 3' + 2" games, with the engine playing at hyper-bullet speed. As you would expect, Hikaru beat all the bots below level 20, though in a couple cases he didn't win on the first try, at least at level 19. But he had a tough time with the Skill 20 bots; I think he eventually beat them all, but probably lost more than he won, even though they were only doing seven ply searches and were slightly weakened by other parameter changes. Against the level 21 bots, he could not score against one of them, and didn't win right away against others. Against level 22 bots (9 ply), he beat the "Morphy" bot, but lost all games to his own Hikaru bot! Overall the bots gave him a tougher time than I would have predicted. It seems that his closest match was level 20.
So I decided to determine approximate CCRL blitz ratings for these skill levels, based on long 2' + 1" blitz matches with Arasan 14, which is about 2585 CCRL blitz. Skill level 23 came out 2543 based on this, level 22 was 2382, level 21 was 2196, and level 20 was too weak to test this way but by extrapolation should be about 2000. Given that Hikaru was playing slower than 2' + 1" and that the bots were slightly weakened by parameter changes, it seems that if he were to play for a CCRL rating by playing 2' + 1" games with CCRL rated engines on reference hardware his CCRL blitz rating might be 2000 or even lower! This is really hard to believe, he's nearly 3200 on chess.com blitz, the world's best. Are the scales really 1200 apart? Maybe I'm doing something wrong, but that's what the data suggests! That would imply that the top engines on the CCRL blitz list would be about 4900 on chess.com blitz if all the intermediate engines played there too. Hard to believe.

I recall that in the 1980's, strong titled players were losing blitz games to, as Michael Stean put it when he lost to Chess 4.6 in 1977, "bloody iron monsters", which in his case was a mainframe. The ratings of those period machines was nothing comparatively.

So maybe you're not wrong here.

[Uri Blass]

I understand that draw or loss was the same for hikaru.

It is not the case in case hikaru had to play for a CCRL rating list.

Another advantage hikaru had for the ccrl rating list is that he could think on the opponent time when he cannot do it when the opponents play at hyper bullet speed(I know CCRL use no pondering but still you cannot force the opponent not to ponder in case the opponent is not an engine).

For the time control I understand that the time control in CCRL blitz is
equivalent to 2'+1" on an Intel i7-4770K

I wonder if you use the same hardware or maybe faster hardware.

Hardware for this was calibrated based on an old 16 core Xeon, presumably somewhat slower than the CCRL reference machine. So really I overrated the skill levels slightly. If Hikaru (or any human) did play games under CCRL conditions, presumably Ponder would be turned on for the machine to be fair, since it can't be turned off for the human. It's true that he got nothing for draws, so perhaps if we allow for this his fair opponent was between level 20 and 21, still only something like 2050 CCRL blitz.

I tried to play against Nero6.1 that has ccrl rating of 1447 at 2+1 and won with white and I believe that even at this time control I am not weaker than Nero and probably better than it.
I am sure nakamura is more than 600 elo better than me at 2+1 so he has to get more than 2050 CCRL rating because I guess I can get more than 1450 in ccrl 2+1 inspite of not being a strong blitz chess player.

Here is the game
I did not look at Nero's pv
Nero blundered by 34...b5

I blundered by 40.b7 but still won after Nero blundered by 50...Rd8

[pgn][Event "Computer chess game"]
[Site "DESKTOP-7QE6S12"]
[Date "2020.12.05"]
[Round "?"]
[White "àåøé"]
[Black "Nero_61"]
[Result "1-0"]
[BlackElo "2000"]
[ECO "A85"]
[Opening "Dutch"]
[Time "20:08:51"]
[Variation "2.c4 Nf6 3.Nc3 e6 4.e3 Bb4"]
[WhiteElo "2400"]
[TimeControl "120+1"]
[Termination "normal"]
[PlyCount "139"]
[WhiteType "human"]
[BlackType "program"]

1. d4 f5 2. c4 Nf6 3. Nc3 e6 {(e7e6 g1f3 f8b4) 0.00/7 4} 4. e3 Bb4 {(f8b4
f1d3 b8c6) +0.10/7 4} 5. Ne2 Nc6 {(b8c6 a2a3 b4c3) +0.12/7 4} 6. a3 Bxc3+
{(b4c3 e2c3 g7g6) -0.12/7 4} 7. Nxc3 g6 {(g7g6 b2b4 e6e5) +0.06/7 4} 8. Be2
d6 {(d7d6 b2b4 a7a6) +0.02/7 4} 9. O-O e5 {(e6e5 b2b4 a7a6) +0.02/7 4} 10.
dxe5 dxe5 {(d6e5 d1d8 c6d8) -0.02/6 3} 11. Qxd8+ Nxd8 {(c6d8 c3d5 f6d5)
-0.35/7 3} 12. Rd1 Ne6 {(d8e6 a3a4 e5e4) +0.07/7 3} 13. Nd5 Nxd5 {(f6d5
d1d5 e5e4) +0.07/7 3} 14. Rxd5 e4 {(e5e4 b2b4 e8g8) +0.04/8 3} 15. b4 c6
{(c7c6 d5e5 b7b6) +0.10/7 3} 16. Rd1 a5 {(a7a5 b4b5 c8d7) +0.19/7 3} 17.
Bb2 O-O {(e8g8 b4b5 a5a4) +0.13/7 3} 18. c5 axb4 {(a5b4 a3b4 a8a1) +0.13/7
3} 19. axb4 Rxa1 {(a8a1 d1a1 h7h5) -0.08/7 3} 20. Bxa1 g5 {(g6g5 e2c4 h7h5)
-0.12/7 3} 21. Bc4 h5 {(h7h5 d1d6 g8f7) -0.07/7 3} 22. Rd6 Kf7 {(g8f7 h2h3
h5h4) -0.02/8 2} 23. Be5 h4 {(h5h4 e5f4 g5f4) -0.92/6 2} 24. h3 Ke7 {(f7e7
d6e6 c8e6) -0.86/6 2} 25. Rd2 Re8 {(f8e8 e5d6 e7f6) -0.02/6 2} 26. Bd6+ Kf6
{(e7f6 d2a2 e8d8) +0.09/7 2} 27. f3 exf3 {(e4f3 g2f3 f5f4) +0.17/6 2} 28.
gxf3 f4 {(f5f4 e3f4 e6f4) +0.23/7 2} 29. e4 Ng7 {(e6g7 g1h2 g7e6) +0.20/6
2} 30. Kg2 Be6 {(c8e6 c4e2 e6c8) +0.19/6 2} 31. Bxe6 Nxe6 {(g7e6 d2a2 e8d8)
+0.21/6 2} 32. Ra2 Rd8 {(e8d8 a2a7 d8d7) +0.26/6 2} 33. Ra7 Rd7 {(d8d7 a7a8
b7b5) +0.39/7 2} 34. Ra8 b5 {(b7b5 g2f2 e6d4) +0.36/6 2} 35. cxb6 Kg6
{(f6g6 d6e5 d7d2) -0.77/6 2} 36. Bc5 Kf6 {(g6f6 c5f2 f6e5) -0.85/6 2} 37.
Ra7 Rd2+ {(d7d2 g2f1 f6e5) -0.87/6 2} 38. Kf1 Nd4 {(e6d4 c5e7 f6g6) -0.95/6
2} 39. Bxd4+ Rxd4 {(d2d4 a7c7 f6e5) -0.86/7 2} 40. b7 Rxb4 {(d4b4 f1e2
b4b3) +0.28/7 2} 41. Ke2 c5 {(c6c5 a7a6 f6e5) +0.24/5 2} 42. Kd2 Ke5 {(f6e5
d2c3 b4b1) +0.16/8 2} 43. Kc3 Ke6 {(e5e6 a7a6 e6f7) +0.04/8 1} 44. Ra6+ Kf7
{(e6f7 a6a7 f7f6) +0.09/5 1} 45. Rc6 Rxb7 {(b4b7 c6c5 f7g6) +0.42/7 1} 46.
Rxc5 Kg6 {(f7g6 c5c6 g6h5) +0.35/7 1} 47. Rc6+ Kh5 {(g6h5 c6c8 b7b1)
+0.37/7 1} 48. Kd4 Rd7+ {(b7d7 d4c4 d7d2) +1.36/8 1} 49. Ke5 Rd3 {(d7d3
c6c8 d3f3) +1.19/8 1} 50. Kf5 Rd8 {(d3d8 c6c7 h5h6) +0.27/7 1} 51. e5 Rd7
{(d8d7 e5e6 d7b7) +0.40/7 1} 52. e6 Ra7 {(d7a7 f5f6 a7a3) +0.37/5 1} 53.
Kf6 Rh7 {(a7h7 e6e7 h7h8) +0.32/5 1} 54. e7 Rxe7 {(h7e7 f6e7 g5g4) -M4/6 0}
55. Kxe7 g4 {(g5g4 h3g4) -5.09/3} 56. fxg4+ Kg5 {(h5g5 c6e6) -5.00/3} 57.
Rc5+ Kg6 {(g5g6 c5f5 f4f3) -7.23/9 1} 58. Ke6 f3 {(f4f3 c5f5 g6g7) -7.15/7
1} 59. Rf5 f2 {(f3f2 f5f2 g6g7) -M5/7 1} 60. Rxf2 Kg7 {(g6g7 f2f6 g7h7)
-M4/6 0} 61. g5 Kg6 {(g7g6 f2f6 g6g5) -M4/6 0} 62. Rh2 Kg7 {(g6g7 g5g6
g7h6) -M6/8 1} 63. Rg2 Kh7 {(g7h7 e6f6 h7h8) -M6/8 0} 64. Kf7 Kh8 {(h7h8
f7e8) -6.12/3} 65. Kf6 Kg8 {(h8g8 g2g4 g8h8) -M3/5} 66. g6 Kh8 {(g8h8 g2g4
h8g8) -M2/3} 67. g7+ Kg8 {(h8g8 g2g6 g8h7) -M1/3} 68. Ke7 Kh7 {(g8h7 g7g8)
-14.05/3} 69. g8=Q+ Kh6 {(h7h6) -M0/1} 70. Qg6# 1-0[/pgn]

[lkaufman]

I understand that draw or loss was the same for hikaru.

It is not the case in case hikaru had to play for a CCRL rating list.

Another advantage hikaru had for the ccrl rating list is that he could think on the opponent time when he cannot do it when the opponents play at hyper bullet speed(I know CCRL use no pondering but still you cannot force the opponent not to ponder in case the opponent is not an engine).

For the time control I understand that the time control in CCRL blitz is
equivalent to 2'+1" on an Intel i7-4770K

I wonder if you use the same hardware or maybe faster hardware.

Hardware for this was calibrated based on an old 16 core Xeon, presumably somewhat slower than the CCRL reference machine. So really I overrated the skill levels slightly. If Hikaru (or any human) did play games under CCRL conditions, presumably Ponder would be turned on for the machine to be fair, since it can't be turned off for the human. It's true that he got nothing for draws, so perhaps if we allow for this his fair opponent was between level 20 and 21, still only something like 2050 CCRL blitz.

I tried to play against Nero6.1 that has ccrl rating of 1447 at 2+1 and won with white and I believe that even at this time control I am not weaker than Nero and probably better than it.
I am sure nakamura is more than 600 elo better than me at 2+1 so he has to get more than 2050 CCRL rating because I guess I can get more than 1450 in ccrl 2+1 inspite of not being a strong blitz chess player.

Here is the game
I did not look at Nero's pv
Nero blundered by 34...b5

I blundered by 40.b7 but still won after Nero blundered by 50...Rd8

[pgn][Event "Computer chess game"]
[Site "DESKTOP-7QE6S12"]
[Date "2020.12.05"]
[Round "?"]
[White "àåøé"]
[Black "Nero_61"]
[Result "1-0"]
[BlackElo "2000"]
[ECO "A85"]
[Opening "Dutch"]
[Time "20:08:51"]
[Variation "2.c4 Nf6 3.Nc3 e6 4.e3 Bb4"]
[WhiteElo "2400"]
[TimeControl "120+1"]
[Termination "normal"]
[PlyCount "139"]
[WhiteType "human"]
[BlackType "program"]

1. d4 f5 2. c4 Nf6 3. Nc3 e6 {(e7e6 g1f3 f8b4) 0.00/7 4} 4. e3 Bb4 {(f8b4
f1d3 b8c6) +0.10/7 4} 5. Ne2 Nc6 {(b8c6 a2a3 b4c3) +0.12/7 4} 6. a3 Bxc3+
{(b4c3 e2c3 g7g6) -0.12/7 4} 7. Nxc3 g6 {(g7g6 b2b4 e6e5) +0.06/7 4} 8. Be2
d6 {(d7d6 b2b4 a7a6) +0.02/7 4} 9. O-O e5 {(e6e5 b2b4 a7a6) +0.02/7 4} 10.
dxe5 dxe5 {(d6e5 d1d8 c6d8) -0.02/6 3} 11. Qxd8+ Nxd8 {(c6d8 c3d5 f6d5)
-0.35/7 3} 12. Rd1 Ne6 {(d8e6 a3a4 e5e4) +0.07/7 3} 13. Nd5 Nxd5 {(f6d5
d1d5 e5e4) +0.07/7 3} 14. Rxd5 e4 {(e5e4 b2b4 e8g8) +0.04/8 3} 15. b4 c6
{(c7c6 d5e5 b7b6) +0.10/7 3} 16. Rd1 a5 {(a7a5 b4b5 c8d7) +0.19/7 3} 17.
Bb2 O-O {(e8g8 b4b5 a5a4) +0.13/7 3} 18. c5 axb4 {(a5b4 a3b4 a8a1) +0.13/7
3} 19. axb4 Rxa1 {(a8a1 d1a1 h7h5) -0.08/7 3} 20. Bxa1 g5 {(g6g5 e2c4 h7h5)
-0.12/7 3} 21. Bc4 h5 {(h7h5 d1d6 g8f7) -0.07/7 3} 22. Rd6 Kf7 {(g8f7 h2h3
h5h4) -0.02/8 2} 23. Be5 h4 {(h5h4 e5f4 g5f4) -0.92/6 2} 24. h3 Ke7 {(f7e7
d6e6 c8e6) -0.86/6 2} 25. Rd2 Re8 {(f8e8 e5d6 e7f6) -0.02/6 2} 26. Bd6+ Kf6
{(e7f6 d2a2 e8d8) +0.09/7 2} 27. f3 exf3 {(e4f3 g2f3 f5f4) +0.17/6 2} 28.
gxf3 f4 {(f5f4 e3f4 e6f4) +0.23/7 2} 29. e4 Ng7 {(e6g7 g1h2 g7e6) +0.20/6
2} 30. Kg2 Be6 {(c8e6 c4e2 e6c8) +0.19/6 2} 31. Bxe6 Nxe6 {(g7e6 d2a2 e8d8)
+0.21/6 2} 32. Ra2 Rd8 {(e8d8 a2a7 d8d7) +0.26/6 2} 33. Ra7 Rd7 {(d8d7 a7a8
b7b5) +0.39/7 2} 34. Ra8 b5 {(b7b5 g2f2 e6d4) +0.36/6 2} 35. cxb6 Kg6
{(f6g6 d6e5 d7d2) -0.77/6 2} 36. Bc5 Kf6 {(g6f6 c5f2 f6e5) -0.85/6 2} 37.
Ra7 Rd2+ {(d7d2 g2f1 f6e5) -0.87/6 2} 38. Kf1 Nd4 {(e6d4 c5e7 f6g6) -0.95/6
2} 39. Bxd4+ Rxd4 {(d2d4 a7c7 f6e5) -0.86/7 2} 40. b7 Rxb4 {(d4b4 f1e2
b4b3) +0.28/7 2} 41. Ke2 c5 {(c6c5 a7a6 f6e5) +0.24/5 2} 42. Kd2 Ke5 {(f6e5
d2c3 b4b1) +0.16/8 2} 43. Kc3 Ke6 {(e5e6 a7a6 e6f7) +0.04/8 1} 44. Ra6+ Kf7
{(e6f7 a6a7 f7f6) +0.09/5 1} 45. Rc6 Rxb7 {(b4b7 c6c5 f7g6) +0.42/7 1} 46.
Rxc5 Kg6 {(f7g6 c5c6 g6h5) +0.35/7 1} 47. Rc6+ Kh5 {(g6h5 c6c8 b7b1)
+0.37/7 1} 48. Kd4 Rd7+ {(b7d7 d4c4 d7d2) +1.36/8 1} 49. Ke5 Rd3 {(d7d3
c6c8 d3f3) +1.19/8 1} 50. Kf5 Rd8 {(d3d8 c6c7 h5h6) +0.27/7 1} 51. e5 Rd7
{(d8d7 e5e6 d7b7) +0.40/7 1} 52. e6 Ra7 {(d7a7 f5f6 a7a3) +0.37/5 1} 53.
Kf6 Rh7 {(a7h7 e6e7 h7h8) +0.32/5 1} 54. e7 Rxe7 {(h7e7 f6e7 g5g4) -M4/6 0}
55. Kxe7 g4 {(g5g4 h3g4) -5.09/3} 56. fxg4+ Kg5 {(h5g5 c6e6) -5.00/3} 57.
Rc5+ Kg6 {(g5g6 c5f5 f4f3) -7.23/9 1} 58. Ke6 f3 {(f4f3 c5f5 g6g7) -7.15/7
1} 59. Rf5 f2 {(f3f2 f5f2 g6g7) -M5/7 1} 60. Rxf2 Kg7 {(g6g7 f2f6 g7h7)
-M4/6 0} 61. g5 Kg6 {(g7g6 f2f6 g6g5) -M4/6 0} 62. Rh2 Kg7 {(g6g7 g5g6
g7h6) -M6/8 1} 63. Rg2 Kh7 {(g7h7 e6f6 h7h8) -M6/8 0} 64. Kf7 Kh8 {(h7h8
f7e8) -6.12/3} 65. Kf6 Kg8 {(h8g8 g2g4 g8h8) -M3/5} 66. g6 Kh8 {(g8h8 g2g4
h8g8) -M2/3} 67. g7+ Kg8 {(h8g8 g2g6 g8h7) -M1/3} 68. Ke7 Kh7 {(g8h7 g7g8)
-14.05/3} 69. g8=Q+ Kh6 {(h7h6) -M0/1} 70. Qg6# 1-0[/pgn]

What actual ratings do you have, either FIDE, national, or just online? I vaguely recall you estimating a number in the 2000-2200 range but I don't really remember what you said long ago.