Carlsen withdrawal after loss to Niemann

[M ANSARI]

Looks like a duck, waddles like a duck and quacks like a duck ... well maybe it really is a duck!

[CornfedForever]

Looks like a duck, waddles like a duck and quacks like a duck ... well maybe it really is a duck!

And Kamsky was a 'Super Duck'?

[Pedro]

If that is the case, why wouldn't it apply pretty much equally to Gukesh? Why the huge disparity between Gukesh's declining error rate and Niemann's steady error rate while both were making similar progress in their teen years? But then I don't have a better explanation for the similar rating climb with highly dissimilar error rate histories. It is a puzzle.

1) Niemann crossed the 2300 level in February of 2016, then, with this low aCPL, languished below 2400 until September of 2018, over two and a half years later. Note that he dips below 2300 quite a number of times. Could it be that our budding data scientists filtered those high 2200 games out?
2) Then, with this consistently low aCPL, he crossed into 2500 territory in February of 2021, a year and 5 months later.
3) Again, with this low aCPL, he crosses 2600 in August of 2021.
4) He continues to improve until the present day where he stands at 2699.

One thing worth noting is that we’re talking about SSS and an uneven number of games played.

Also, still no clear evidence that there is a strong correlation between aCPL and rating.

The Brazilian youtuber has just posted a new video in English with news from the investigation. He seems to have made the correlation you referred to in his comment.

Video:

[dkappe]

The Brazilian youtuber has just posted a new video in English with news from the investigation. He seems to have made the correlation you referred to in his comment.

Video:

I made a few further suggestions:

A good second video. A few more suggestions: 1) you’ve made a good start but you need a broader range of players — a few thousand — not just the top players to demonstrate correlation between ratings and aCPL. Also you need to go back a few decades. 2) Did you filter games by rating or date? Niemann crossed 2300 several times over a 2.5 year period before crossing 2400. 3) did you filter for theory? Take out the first 20 ply, for instance and see how much opening knowledge impacts aCPL. 4) eliminate positions with only 1 legal move and also only 1 good move, I.e. the second best move is more than 100 cp worse. You’ll find there’s a dramatic change in accuracy. The same is true of positions with 2-3 good moves, but to a lesser extent.

I’d point out that the number of FIDE rated games is quite small for Niemann, so that’s an issue as well.

Let’s see what the next video brings.

[lkaufman]

I would like to see a similiar analysis about Mikhail Tal's games when he got better.

I know that he made unsound sacrifices.
A strong player may make wrong moves that cause the opponent also to go wrong and
I am not sure if the assumption that always a strong player improve his accuracy when he gets better is correct.

It may in fact be true, but like the validity of fingerprint identification, there’s nothing (NOTHING) in the scientific literature on that topic.

I would point out another player: Gata Kamsky. Gata recently said he gained about 300 rating points in (think he said) about a year. That puts Hans to shame...

My own son Raymond went from about 900 (USCF rating) at age 12 to about 2200 at age 15 in the 1990s, a 1300 elo gain! I know it's not the same reaching 2200 as 2600 or 2700, but my point is that huge rating gains in a couple years as a teenager are quite possible and not suspicious. What would really convince me of cheating is a properly done comparison, using identical engines and time limits, of Niemann's centipawn loss in the short time interval under the most suspicion (OTB, standard time controls only) compared to the same for other teenage players who were similarly rated and active over a similar interval around the same general time. If Niemann's mean CP loss was clearly lower than all the other players being compared, that would be pretty convincing. Has anyone done a study like this yet?

[dkappe]

My own son Raymond went from about 900 (USCF rating) at age 12 to about 2200 at age 15 in the 1990s, a 1300 elo gain! I know it's not the same reaching 2200 as 2600 or 2700, but my point is that huge rating gains in a couple years as a teenager are quite possible and not suspicious. What would really convince me of cheating is a properly done comparison, using identical engines and time limits, of Niemann's centipawn loss in the short time interval under the most suspicion (OTB, standard time controls only) compared to the same for other teenage players who were similarly rated and active over a similar interval around the same general time. If Niemann's mean CP loss was clearly lower than all the other players being compared, that would be pretty convincing. Has anyone done a study like this yet?

The Brazilian above is heading in that direction, but he has too little data as yet. Also not sure his games are all OTB, he’s controlled for single move or single node positions or opening theory. Hopefully he’ll take more of my suggestions.

[swami]

I would like to see a similiar analysis about Mikhail Tal's games when he got better.

I know that he made unsound sacrifices.
A strong player may make wrong moves that cause the opponent also to go wrong and
I am not sure if the assumption that always a strong player improve his accuracy when he gets better is correct.

It may in fact be true, but like the validity of fingerprint identification, there’s nothing (NOTHING) in the scientific literature on that topic.

I would point out another player: Gata Kamsky. Gata recently said he gained about 300 rating points in (think he said) about a year. That puts Hans to shame...

Here is Gata kamsky's rating history

https://www.olimpbase.org/Elo/player/Ka ... 0Gata.html

2380 1.1.1988 when he entered fide rating list with 25 games
2345 1.1.1989 got down in the rating based on 16 games
2345 1.7.1989 did not play for fide rating between 1.1.1989 and 1.7.1989
2510 1.1.1990 played 51 games and increased his rating by 165 elo
2650 1.7.1990 played 51 games and increased his rating by 140 elo

Note that Kamsky's rating got down later before getting up again.

Kaspersky warns me not to open the link you posted. But, yes, I think it was from 2345 to 2650 in 1 yr. Pretty darn impressive.

During Gata Kamsky's times, IE in 1989, FIDE didn't have many players hence it was very easy to climb to higher threshold of the rating scale faster.

He just needed a few draws and wins with the likes of Nigel Short, Ivanchuk, Anand, Kramnik and he's all set right up there.

Nowadays the ladder is really long with several hundreds of players

[Uri Blass]

I would like to see a similiar analysis about Mikhail Tal's games when he got better.

I know that he made unsound sacrifices.
A strong player may make wrong moves that cause the opponent also to go wrong and
I am not sure if the assumption that always a strong player improve his accuracy when he gets better is correct.

It may in fact be true, but like the validity of fingerprint identification, there’s nothing (NOTHING) in the scientific literature on that topic.

I would point out another player: Gata Kamsky. Gata recently said he gained about 300 rating points in (think he said) about a year. That puts Hans to shame...

My own son Raymond went from about 900 (USCF rating) at age 12 to about 2200 at age 15 in the 1990s, a 1300 elo gain! I know it's not the same reaching 2200 as 2600 or 2700, but my point is that huge rating gains in a couple years as a teenager are quite possible and not suspicious. What would really convince me of cheating is a properly done comparison, using identical engines and time limits, of Niemann's centipawn loss in the short time interval under the most suspicion (OTB, standard time controls only) compared to the same for other teenage players who were similarly rated and active over a similar interval around the same general time. If Niemann's mean CP loss was clearly lower than all the other players being compared, that would be pretty convincing. Has anyone done a study like this yet?

SOPHIE MORRIS-SUZUKI has more than 1300 elo gain in a shorter time
She went from USCF rating below 700 at age 12 to rating above 2000 in less than 2 years
Of course it is easier when you start lower

http://www.uschess.org/msa/MbrDtlTnmtHst.php?15542082.4

[swami]

Well, part of this thread has been about who we should believe.

I think you mix up something. It is not Carlsen vs. Niemann, Carlsen is not the only one who speaks out suspicions, also other GM's and different AI cheat detection systems are on the trigger. The question whether Hans is cheating or not has nothing to do with Carlsens credibility. so it is completely useless to question his credibility because he said "how" at the wrong time.

Also, there is no question that Carlsen really believes that Niemann is cheating. He may be mistaken about this, but it takes a flat earther to believe that Carlsen is lying.

Just because Carlsen is about 200 rating points higher does NOT mean GM Hans cheated, there are several other upsets where the difference in ratings are much higher and the higher rated player do NOT accuse the lower of cheating. Here are some latest Upsets. These two lower rated indian players were hibernating during COVI-19 and only studying chess the same as GM Hans, Gukesh, GM Pragg, and GM Arjun ==>
https://chessbase.in/news/1st-Bikaner-G ... d-1-report

[Event "1st Bikaner Open International Grandmast"]
[Site "Banquet Hall of Ashirwad Bhawa"]
[Date "2022.10.01"]
[Round "1.14"]
[White "LAXMAN, R.R.."]
[Black "SWAYHAM, P DAS."]
[Result "0-1"]
[ECO "A06"]
[WhiteElo "2378"]
[BlackElo "1902"]
[Annotator "Shahid"]
[PlyCount "106"]
[EventDate "2022.10.01"]
[EventRounds "10"]
[EventCountry "IND"]

1. d4 {1} d5 {9} 2. Nf3 {5} Nf6 {8} 3. c4 {4} c6 {6} 4. e3 {5} e6 {25} 5. b3 {6
} Nbd7 {11} 6. Bd3 {6} Bd6 {54} 7. Bb2 {10} b6 {91} 8. O-O {9} Bb7 {29} 9. Nc3
{49} O-O {17} 10. Qe2 {171} Re8 {83} 11. Rfd1 {37} Rc8 {148} 12. Rac1 {34} Qe7
{63} 13. cxd5 {66} Nxd5 {208} 14. Ne4 {571} Bb8 {189} 15. Ne5 {198} Nf8 {321}
16. Qh5 {315} f6 {155} 17. Nc4 {66} Nb4 {1261} 18. Bb1 {118} Red8 {382} 19. a3
{292} Na6 {171} 20. b4 {24} Rd5 {415} 21. Qe2 {137} Rcd8 {83} 22. Ncd2 {32} Nc7
{104} 23. Nf3 {231} R5d7 {295} 24. Re1 {197} Nd5 {225} 25. Qc2 {80} Qf7 {118}
26. Ng3 {28} Rc8 {92} 27. Ba2 {90} Kh8 {178} 28. Qe2 {399} Qe7 {50} 29. Ne4 {21
} Qf7 {54} 30. Bb1 {151} Qe7 {22} 31. g3 {22} Kg8 {77} 32. Ned2 {8} Qd8 {193}
33. Ba2 {125} Kh8 {15} 34. e4 {7} Ne7 {75} 35. Nf1 {23} Neg6 {69} 36. h4 {9}
Qe8 {109} 37. Ne3 {80} c5 {123} 38. d5 exd5 {136} 39. Nxd5 {41} (39. exd5)
39... Ne5 {75} 40. bxc5 {125} Rxc5 {66} 41. Rxc5 {2} bxc5 {4} 42. Bxe5 {45}
Bxe5 {10} 43. Qc4 {38} Ne6 {371} 44. Nxe5 {92} fxe5 {4} 45. Rb1 {50} h6 {87}
46. a4 {58} Bc6 {163} 47. a5 {408} Rd8 {8} 48. Qa6 {515} Qd7 {46} 49. Ne3 $2 {
116} (49. Ne7 Bxe4 (49... Qxe7 50. Qxc6 Nd4 51. Qb7) 50. Re1 Nd4 51. Rxe4 Qxe7
52. Kg2) 49... Nd4 $19 {50} 50. Bd5 {5} Qh3 {43} 51. Ng2 {39} Bxd5 {49} 52. Qd3
{29} Bc4 {12} 53. Qe3 {5} Ne2+ {4} 0-1


[Event "1st Bikaner Open International Grandmast"]
[Site "Banquet Hall of Ashirwad Bhawa"]
[Date "2022.10.01"]
[Round "1.15"]
[White "SINGH, PRITAM."]
[Black "NGUYEN, DUC HOA."]
[Result "1-0"]
[ECO "D80"]
[WhiteElo "1899"]
[BlackElo "2360"]
[Annotator "Shahid"]
[PlyCount "43"]
[EventDate "2022.10.01"]
[EventRounds "10"]
[EventCountry "IND"]

1. d4 {1} Nf6 2. c4 g6 3. Nc3 d5 4. Bg5 Ne4 5. Nxe4 dxe4 6. e3 f6 7. Bh4 Nc6 8.
a3 h5 9. h3 e5 10. dxe5 {295} Qxd1+ {32} 11. Rxd1 {48} g5 {9} 12. Bg3 {34} h4 {
15} 13. Bh2 {26} Nxe5 {31} 14. Bxe5 {146} fxe5 {6} 15. Ne2 {178} Rh6 {228} 16.
Nc3 {56} Bf5 {27} 17. Be2 {140} Rb6 $4 {278} (17... Rd8 18. O-O (18. Nd5 c6)
18... Rxd1 19. Rxd1 Rd6) 18. Nd5 $18 {768} Bd6 {1183} (18... Rc6 19. Bh5+ Kd8
20. Nb6+ $18) 19. c5 {247} Bxc5 {18} 20. Nxc7+ {10} Kf8 {30} 21. Nxa8 {75} Rxb2
{5} 22. Bg4 {91} 1-0

As far as I know, no one suspects Gukesh, Pragg, or Arjun of cheating OTB because none of them has been accused of cheating online. While it is true that cheating online does not prove cheating OTB, I think that the converse does apply, namely that NOT cheating online is pretty much proof of NOT cheating OTB, since it is so much easier to cheat undetected online than OTB.

Pragg used to be a National champion in his age group consistently in the league of over 400 players year on year. Since the very young age . So his track record is excellent, in face of extreme competition.

Arjun Erigaisi and Gukesh were not National rank holders at the top until recently (ie after their 13 years of age)

Players grow at different pace, but once they devote whole time practicing chess after getting permission from school to take a whole year off and only appear during final exam, they can really progress faster by just practicing day in and day out. Remember Outliers by Malcolm Gladwell?

[CornfedForever]

I would point out another player: Gata Kamsky. Gata recently said he gained about 300 rating points in (think he said) about a year. That puts Hans to shame...

My own son Raymond went from about 900 (USCF rating) at age 12 to about 2200 at age 15 in the 1990s, a 1300 elo gain! I know it's not the same reaching 2200 as 2600 or 2700, but my point is that huge rating gains in a couple years as a teenager are quite possible and not suspicious. What would really convince me of cheating is a properly done comparison, using identical engines and time limits, of Niemann's centipawn loss in the short time interval under the most suspicion (OTB, standard time controls only) compared to the same for other teenage players who were similarly rated and active over a similar interval around the same general time. If Niemann's mean CP loss was clearly lower than all the other players being compared, that would be pretty convincing. Has anyone done a study like this yet?

Kamsky's 2345 to 2650 rating in 1 yr is actually more 'impressive' than Han's leap over 2 yrs, don't you think? I wonder what the Brazilian would find if he investigated those games of Kamsky's?

I hear chess.com has another announcement coming this week. With the US Championship starting Wednesday, anything they look to say would likely be too late to possibly allow Hans to be 'dis-invited'.

I presume the organizers will delay broadcasts more than 15 minutes (which however it probably enough) and have enhanced scanning (along the lines of the recent Candidates) which I know Fabi seemed to find good.

With all that in place...if Hans is a fraud, I wonder how he might score in the 13 games. 4.5 out of 13? Someone should start a poll.