Carlsen withdrawal after loss to Niemann

[M ANSARI]

Well ... the games databases will be scrutinized now by some very smart people in chess and out of chess. Chess is probably not a good game to try and cheat and hope to get away with. Hans has said he has only cheated online but made it seem it was only twice and never in important games and only when he was a kid and was never against the top players. Now we have one Titled Tuesday where he cheated every game apparently. Also now it seems like he also cheated in some OTB games ... something he adamantly says he did not do. The information is in the databases of games that have already been played. Personally I have a feeling that this is just the tip of the iceberg. The good thing that will come out of is that more effort will be placed on anti cheating measures. I think you can get away with cheating in a few games, but once you cheat in many games, there will be a signature or information in the collective data. Of course you can try to confuse or obscure the data by playing suboptimal moves or even non winning or losing moves in some positions, but most likely if enough games are played this dissonance can be erased. Also the date the game is played needs to be correlated with the relevant engines of the time ... there might be a preference to certain engines even though they are not the strongest. LC0 style is probably the most human style and cheaters might opt to go for that engine more than others. LC0 even at extremely shallow depth can guide even the strongest GM towards a winning position. If you have LC0 combined with any modern SF you could get something that resembles human chess with the tactical safety net of a good bean counter. Of course today's SF has NNUE so you don't really need any other engine and can simply use multi PV and choose the most human looking move. On critical very tactical situations where "only" moves are required ... you can then become a tactical wizard. Remember we are not talking about a weak chessplayer, but about a chessplayer who could be 2300 to 2400 ELO. The difference between a 2400 ELO and 2800 ELO is just a few critical move choices.

[DrCliche]

Take a look back at the original tweet thread and you’ll find that the OP didn’t think to generate a control group and, incredibly, declines to provide one when challenged.
The OP is obviously not fit in a scientific/educated/skilled sense to be taken seriously, there’s not much point in wasting time challenging his data collection and selection exercise for cherry picking and manipulation until it could be wrapped into some confirmation bias conclusion.
Moral of the story - never accept data tables and conclusions at face value, always check first before repeating.

You don't know what a control group is. You mean to say that the original tweeter didn't control for confounding variables. Of course, outside of obscure techniques that you've never heard of (like synthetic control methods), controls aren't "generated". And it's hardly relevant that the original tweeter didn't control for confounding variables, because I did.

(That is, I "checked first before repeating", which you would know if you'd bothered to follow your own advice in this thread. Alas, you appear to glide through life blissfully unburdened by a sense of irony.)

I gathered extra data in order to account for what I thought were the three most likely confounders, namely age, fatigue, and opponent strength. None of them were anywhere close to being statistically significant in my analysis. Andrew Grant suggested another (absurdly unlikely) confounder, that Niemann's strength increases (very) disproportionately compared to that of his opponents when players know their pairings ahead of time, and that advance knowledge of pairings is strongly and positively correlated with a tournament's broadcast status. But as of yet he's offered no plausible mechanism for such an effect, much less demonstrated that any data exists to support it.

If the broadcast data is accurate (still a big if, but I've seen no indications otherwise), the extent to which the original tweeter could have meaningfully cherry picked the data is very limited. The only thing he could have done is choose a particular interval of Niemann's classical games. But that's hardly relevant, as it would obviously be concerning if any player ever had a two year span in which their classical tournament performances were so strongly correlated with broadcast status, and to such a crazy degree of statistical significance.

Given a p-value of 0.0008, such a phenomenon is quite unlikely to occur due to random chance, even if you have the option of picking the most damning two year window in a given player's career. And if Niemann has other spans that don't follow the same pattern, it doesn't mean he never cheated, it just means he probably wasn't aggressively cheating then.

[Dariusz]

Well ... the games databases will be scrutinized now by some very smart people in chess and out of chess. Chess is probably not a good game to try and cheat and hope to get away with. Hans has said he has only cheated online but made it seem it was only twice and never in important games and only when he was a kid and was never against the top players. Now we have one Titled Tuesday where he cheated every game apparently. Also now it seems like he also cheated in some OTB games ... something he adamantly says he did not do. The information is in the databases of games that have already been played. Personally I have a feeling that this is just the tip of the iceberg. The good thing that will come out of is that more effort will be placed on anti cheating measures. I think you can get away with cheating in a few games, but once you cheat in many games, there will be a signature or information in the collective data. Of course you can try to confuse or obscure the data by playing suboptimal moves or even non winning or losing moves in some positions, but most likely if enough games are played this dissonance can be erased. Also the date the game is played needs to be correlated with the relevant engines of the time ... there might be a preference to certain engines even though they are not the strongest. LC0 style is probably the most human style and cheaters might opt to go for that engine more than others. LC0 even at extremely shallow depth can guide even the strongest GM towards a winning position. If you have LC0 combined with any modern SF you could get something that resembles human chess with the tactical safety net of a good bean counter. Of course today's SF has NNUE so you don't really need any other engine and can simply use multi PV and choose the most human looking move. On critical very tactical situations where "only" moves are required ... you can then become a tactical wizard. Remember we are not talking about a weak chessplayer, but about a chessplayer who could be 2300 to 2400 ELO. The difference between a 2400 ELO and 2800 ELO is just a few critical move choices.

I agree with what you wrote.
There are many indications that Hans was indeed assisted by the chess engine.
The analysis of my esteemed predecessors in this forum thread, or, for example, the analysis of games using the ChessBase Centipawn module by FM Andrii Punin, indicates that Hans chose the strongest line of engine playing at the level of at least 2900 or probably much more. Of course - it didn't have to be Stockfish at all.
I wonder, if Hans actually used a chess engine, why he did it in such a "reckless" way ? In your post above you wrote how hypothetically one can cheat in a way that would certainly draw less attention than the way Hans allegedly did it. Did he already feel so confident and unpunished that he thought he wouldn't get away with it ? I'm afraid that's exactly what happened.

I'm counting on the fact that this incident ... will go well for chess. I look forward to even more effective methods of detecting cheating during face-to-face games and on online servers. I know people (I wrote this on some of the threads of this forum) who unfortunately experienced false positive signals and bans, although they certainly did not assist the engines while playing online (I witnessed it). And unfortunately, according to my experience and my friends' experience - Lichess is leading the way. That is, we have in this case the opposite situation, when honest players are punished. This is an imperfection/weakness of today's systems for detecting cheating while playing chess online. According to signals from my friends and chessengeria.com readers, the best at detecting cheaters is chess.com and ICC. Next is chess24.com then PlayChess, where the fewest people play but PlayChess probably has the best security of its kind. Of course, this is only the opinion of a more small percentage of users, so this should be taken more as a signal / curiosity. The Centipawn analyze module is an interesting verifier insofar as anyone who owns ChessBase16 can check, for example, the play of their opponents for chess engine support.

Certainly a very interesting and timely topic. At the urging of my blog readers, I will probably write and publish an article on chessengeria.com on this topic in the future, when more is known (I am waiting for a statement from chess.com and Magnus). Thank you very much for the interesting discussion.

[DrCliche]

I only wonder how Magnus was able to conclude that there was OTB cheating in his game, the game itself wasn't too suspicious if you assume that the 20 moves of prep was legitimate.

Someone posted an old Norwegian interview earlier in the thread where Magnus talked about cheating in a way that seemed almost prophetic. One of the things he said was that playing at the top level requires trust in your peers, because if your opponents worry that you're being guided by an engine at crucial moments, then you have an enormous psychological edge, never mind the actual edge you'd get from being able to play perfectly whenever you deemed it necessary. (He also said something to the effect that he'd be completely unstoppable if he got to consult an engine just twice per game.)

Of course I have no special insight into Carlsen, the Sinquefield bigwigs, and the decisions any of them have made, but it seems at least possible that he quit simply because he was unable to banish the constant worry that Niemann was cheating, and felt that it destroyed his ability to play to his customary level, as further evidence by his loss. If he wasn't comfortable playing out the tournament while Niemann remained, but there was no definitive proof of foul play, so the tournament wouldn't give Niemann the boot ... at this point in his career Carlsen may very well be so practical that if he knows he has no realistic chance of playing his best, he won't play at all.

[Chessqueen]

I only wonder how Magnus was able to conclude that there was OTB cheating in his game, the game itself wasn't too suspicious if you assume that the 20 moves of prep was legitimate.

Someone posted an old Norwegian interview earlier in the thread where Magnus talked about cheating in a way that seemed almost prophetic. One of the things he said was that playing at the top level requires trust in your peers, because if your opponents worry that you're being guided by an engine at crucial moments, then you have an enormous psychological edge, never mind the actual edge you'd get from being able to play perfectly whenever you deemed it necessary. (He also said something to the effect that he'd be completely unstoppable if he got to consult an engine just twice per game.)

Carlsen is just a poor loser, I like how GM Caruana and other top GMs talk about GM Hans, instead of blaming him all chess players of all different ratings sometimes play games that are at a higher level than their actual ratings, probably because their brains automatically recognized a pattern and trust his intuition and even is he was unable to explain it after he played versus Firouzja does NOT mean that his brain was not able to pick it for him from past pattern recognition , but to blame GM Hans of cheating is wrong

[AdminX]

I only wonder how Magnus was able to conclude that there was OTB cheating in his game, the game itself wasn't too suspicious if you assume that the 20 moves of prep was legitimate.

Someone posted an old Norwegian interview earlier in the thread where Magnus talked about cheating in a way that seemed almost prophetic. One of the things he said was that playing at the top level requires trust in your peers, because if your opponents worry that you're being guided by an engine at crucial moments, then you have an enormous psychological edge, never mind the actual edge you'd get from being able to play perfectly whenever you deemed it necessary. (He also said something to the effect that he'd be completely unstoppable if he got to consult an engine just twice per game.)

Carlsen is just a poor loser, I like how GM Caruana and other top GMs talk about GM Hans, instead of blaming him all chess players of all different ratings sometimes play games that are at a higher level than their actual ratings, probably because their brains automatically recognized a pattern that it has stored in the beautiful and most complicated computer ever created ( the Brain ) and even is he was unable to explain it after he played versus Firouzja does NOT mean that his brain was not able to pick it for him from past pattern recognition , but to blame GM Hans of cheating is wrong

He may or may not be a poor loser as you say, but that is rich coming from you. Didn't LiChess ban your account for cheating? I know you said it was because your account got hacked ... right that must be it.

[emadsen]

Here's some more fuel for the fire, but the tl;dr is that Niemann is overwhelmingly likely to be a chronic and habitual OTB cheater... If you run a regression on the above dataset, you will find that whether or not a tournament was broadcast live explains 67% (!!!!!) of the variation in Niemann's performance over that time period. (The rest of the variation is likely random, or at least isn't explained by age, number of rounds, or strength of opponents.)

Discussion of divergent performances at tournaments in New York and Charlotte, where Niemann got his second and third GM norms: https://www.youtube.com/watch?v=AG9XeSPflrU.

Fascinating statistical analysis and interesting video clip, DrCliche. Thanks for taking time to research and explain your findings.

Well ... the games databases will be scrutinized now by some very smart people in chess and out of chess. Chess is probably not a good game to try and cheat and hope to get away with... The information is in the databases of games that have already been played. Personally I have a feeling that this is just the tip of the iceberg... I think you can get away with cheating in a few games, but once you cheat in many games, there will be a signature or information in the collective data.

I agree. To quote AC/DC, "It's a long way to the top if you wanna rock and roll." There will be a record of games along the path to the top that will be scrutinized by strong GMs and strong mathematicians.

I wonder, if Hans actually used a chess engine, why he did it in such a "reckless" way ?

Perhaps for the reason I suggested:

2) Hans realizing there's no audience for the bad-boy, trash-talking villain in the 2400s; that no one cares unless the villain rises to the top and knocks the kings and princes off their thrones; so Hans decided to leverage computer assistance to get himself on the stage.

In other words, Hans wanted to get on the big stage, and fast.

[CornfedForever]

...but it seems at least possible that he quit simply because he was unable to banish the constant worry that Niemann was cheating, and felt that it destroyed his ability to play to his customary level, as further evidence by his loss...

Let me point out that Carlsen's "customary level"...is bound up in his fairly customary choice to play demonstrably suspect openings against people he has some contempt for (...at least let us say 'weaker', much weaker) and while that can be of use to him, it also makes it more likely that he is going to find himself in a 'poor' position and get outplayed in the bargain.

[CornfedForever]

I agree. To quote AC/DC, "It's a long way to the top if you wanna rock and roll." There will be a record of games along the path to the top that will be scrutinized by strong GMs and strong mathematicians.

Yes, but it would be helpful to have a 'strong mathematician' who is also a strong chess player. This because he might be able to see thru mere numbers...which are only going to point towards what 'might be the truth'.

'Where have you gone, John Nunn, our game turns it's lonely eyes to you'...

[dkappe]

Did nobody consider that this whole affair isn’t a set up? Chess.com and Magnus are now a corporate entity with massively increased profile and a valuation in proportion to public interest in chess in general. Venture capital backers of chess.com must be rubbing their hands with glee. Breaking into mainstream with a Magnus-Hans televised grudge match posing as World Chess Championship by-passing FIDE, anyone?

The investors in chess.com look to be private equity, not venture capital.