The Hans Niemann cheating scandal has rocked the chess world following his stunning defeat of Magnus Carlsen and Niemann's subsequent ban from Chess.com tournaments. IM Ken Regan, Ph.D. - the preeminent expert on identifying cheating in the chess world - joins James to unravel the situation and lay out the evidence.
Thanks...this is quite good. His models for catching cheating - as best he cares to explain them, are quite interesting. More people should watch this from a Computational Complexity Theorist.
For those who do not want to take the time to watch...no cheating detected in the cup or those European tournaments. The post pandemic rating jump...explained.
I wish he had gone into it more but about 1hr 25 min in he talks about something like the benefit of MTCS (as I understand it) his model could be used for prepping opening lines with 'acceptable risks'...in playing non-best engine moves.
lkaufman wrote: ↑Tue Sep 13, 2022 11:47 pm
Differences in motivation and similar factors could account for a modest performance gap, maybe 30 or 40 elo, but not hundreds of elo.
I admit that my statistics is definitely a bit rusty. But I think this argument can't be valid. This should be about assumptions of distribution of data and testing, and I feel that 30 or 40 elo would already really be a lot. Swami gave another one that even meets your personal experiences - could this be another 40-50 elo? You actually gave another one yourself, when you described the reasons ( and in this case casuality makes some sense as you should know) for your own performance at senior events compaired to your peers. I envision ever more strange curves of potential distributions. If there are reasonable alternative explanations for the perceived strangeness of the data, the strangeness just disappears, doesn't it?
It's true that I performed a hundred or more elo better in my three World Senior events, where I prepared for many hours for each game, than in domestic events where I had little or no chance to prepare. But I would look at this as pretty much the upper limit to the value of preparation; I play so many different openings that GM John Fedorowitz told me he stopped trying to prepare for games with me after half a dozen or so because he could never guess what I would play, whereas nearly every Senior opponent was 90% predictable for me in their opening choices. For young players playing other young players in strong tournaments, preparation should roughly cancel out. In the present case I am much more inclined to believe that the identification of which tournaments were or were not broadcast was just not reliable, this seems a more plausible explanation than the idea that he actually played much better in broadcast events. Anyway, in view of IM Ken Regan's analysis of the data, we should give Niemann the benefit of the doubt regarding over-the-board play unless new information suggests otherwise.
If someone gets away with cheating, or stealing or even murder ... and there is no definitive proof beyond a reasonable doubt... then he is innocent. I would say that in this latest tournament that MC exited, there is definitely reasonable doubt. Very suspicious play in his games with Mamadyarov, MC and also very suspicious post game analysis (for me the biggest of all the red flags) especially in the game against Firouzja ... but the reasonable doubt holds. I think things should be investigated in more detail and Chess.com should come out with other information or data it has. Most importantly more scrutiny should be made to stamp out any possible method of cheating before it happens. Remember that just the thought that someone could be cheating is enough of a psychological blow to anyone's psyche and will greatly affect how a top GM plays. I also think online cheating should not be disregarded because it means that same person would also have a higher probability to cheat OTB if he thought he could get away with it. At least keep a period of at least 5 years before putting someone like that in a prestigious tournament. With Hans I think if he was Indian or from some other country with many up and coming talented young players, he would have been crucified and banned for life ... being american certainly helps his case and was probably the main reason he was chosen for the St. Louis tournament and a blind eye turned to his previous online cheating. I highly doubt if Gukesh or Nihal Sarin had been caught cheating that they would get the same treatment.
M ANSARI wrote: ↑Thu Sep 15, 2022 8:18 am
I would say that in this latest tournament that MC exited, there is definitely reasonable doubt. Very suspicious play in his games with Mamadyarov, MC and also very suspicious post game analysis (for me the biggest of all the red flags) especially in the game against Firouzja ... but the reasonable doubt holds.
I’d recommend listening to the podcasts with IM Regan, like this one. Regan’s sophisticated analysis (unlike DrCliche’s naive approach) shows no evidence of OTB cheating by Niemann. It also addresses his play in this most recent tournament.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
M ANSARI wrote: ↑Thu Sep 15, 2022 8:18 am
I would say that in this latest tournament that MC exited, there is definitely reasonable doubt. Very suspicious play in his games with Mamadyarov, MC and also very suspicious post game analysis (for me the biggest of all the red flags) especially in the game against Firouzja ... but the reasonable doubt holds.
I’d recommend listening to the podcasts with IM Regan, like this one. Regan’s sophisticated analysis (unlike DrCliche’s naive approach) shows no evidence of OTB cheating by Niemann. It also addresses his play in this most recent tournament.
I have watched a lot of opinions from many people but I think nobody sees the elephant in the room. I am pretty sure it is impossible to tell if someone is cheating from a single game or even a few games ... if the person has an incredibly powerful computer ... and if the person cheats properly and understands how the cheating algorithms work. Unless it is OTB bullet or blitz ... it is possible to obfuscate the moves enough where it can mask any cheating and still play +200 ELO higher than your rating. I mean imagine for example 7 piece EGTB's ... all the human player would have to do is choose non losing moves even if they are not in the top 5 best moves ... just make humanly logical moves that don't lose and don't go for obscure tactical shots. If you analyze the games of even the greatest top GM's ... there are always a few moves that throw away the advantage immediately or that do not parry an eventual pile up for an attack. The human will eventually get outplayed even in the most dry position. I watched GM Hikaru try and not get adopted (lose 10-0) with pawn odds and you could just see that humans have no chance. Even when he was trying to just draw one game, he would eventually get ouplayed by simple strong moves that were totally human like.
M ANSARI wrote: ↑Thu Sep 15, 2022 8:55 am
I have watched a lot of opinions from many people but I think nobody sees the elephant in the room. I am pretty sure it is impossible to tell if someone is cheating from a single game or even a few games ... if the person has an incredibly powerful computer ... and if the person cheats properly and understands how the cheating algorithms work. Unless it is OTB bullet or blitz ... it is possible to obfuscate the moves enough where it can mask any cheating and still play +200 ELO higher than your rating. I mean imagine for example 7 piece EGTB's ... all the human player would have to do is choose non losing moves even if they are not in the top 5 best moves ... just make humanly logical moves that don't lose and don't go for obscure tactical shots. If you analyze the games of even the greatest top GM's ... there are always a few moves that throw away the advantage immediately or that do not parry an eventual pile up for an attack. The human will eventually get outplayed even in the most dry position. I watched GM Hikaru try and not get adopted (lose 10-0) with pawn odds and you could just see that humans have no chance. Even when he was trying to just draw one game, he would eventually get ouplayed by simple strong moves that were totally human like.
Listen to the podcast. It’s based on quite a bit of data from many years of high level players and all of Niemann’s games since 2020. It’s a good listen.
Fat Titz by Stockfish, the engine with the bodaciously big net. Remember: size matters. If you want to learn more about this engine just google for "Fat Titz".
M ANSARI wrote: ↑Thu Sep 15, 2022 8:18 am
I think things should be investigated in more detail and Chess.com should come out with other information or data it has.
While chess.com's policy is to not comment on fair play violations...it seems that once a player admits publicly (as Hans has) then the player is essentially giving them the 'okay' to go public (as chess.com then did) AND probably feel free to comment further about future/other violations'. They sort of did that by saying his statement did not admit to the extent of his cheating. That sounds fair.
M ANSARI wrote: ↑Thu Sep 15, 2022 8:18 am
I also think online cheating should not be disregarded because it means that same person would also have a higher probability to cheat OTB if he thought he could get away with it. At least keep a period of at least 5 years before putting someone like that in a prestigious tournament.
Possibly. But one could argue that because someone cheated in online card game that said person should not be allowed to play in the WSOP (World Series of Poker). That's...just going to far IMHO.
'Online chess' by individually owned companies who are there to make a profit, confers nothing official or even important upon a player ($$ payouts aside!) and different sites will have different levels of security and rules...and different rating systems. FIDE or National OTB tournaments really need to be kept separate from private company online sites activity. and determinations. To tar and feather someone in the world of official OTB for what they may have done in the wild wild west of online play just seems to me to be somewhere we do not want to go so easily.
M ANSARI wrote: ↑Thu Sep 15, 2022 8:18 am
I would say that in this latest tournament that MC exited, there is definitely reasonable doubt. Very suspicious play in his games with Mamadyarov, MC and also very suspicious post game analysis (for me the biggest of all the red flags) especially in the game against Firouzja ... but the reasonable doubt holds.
I’d recommend listening to the podcasts with IM Regan, like this one. Regan’s sophisticated analysis (unlike DrCliche’s naive approach) shows no evidence of OTB cheating by Niemann. It also addresses his play in this most recent tournament.
I have watched a lot of opinions from many people but I think nobody sees the elephant in the room. I am pretty sure it is impossible to tell if someone is cheating from a single game or even a few games ... if the person has an incredibly powerful computer ... and if the person cheats properly and understands how the cheating algorithms work. Unless it is OTB bullet or blitz ... it is possible to obfuscate the moves enough where it can mask any cheating and still play +200 ELO higher than your rating. I mean imagine for example 7 piece EGTB's ... all the human player would have to do is choose non losing moves even if they are not in the top 5 best moves ... just make humanly logical moves that don't lose and don't go for obscure tactical shots. If you analyze the games of even the greatest top GM's ... there are always a few moves that throw away the advantage immediately or that do not parry an eventual pile up for an attack. The human will eventually get outplayed even in the most dry position. I watched GM Hikaru try and not get adopted (lose 10-0) with pawn odds and you could just see that humans have no chance. Even when he was trying to just draw one game, he would eventually get ouplayed by simple strong moves that were totally human like.
I don't recall that event, could you post a link or at least specifics (when played, what engine, what pawn, time limit)? Komodo beat him 10 to 0 in ten minute games once, but that was with no handicap, and we beat him 6.5 to 1.5 giving him two pawns, but no ten game match giving one pawn.
M ANSARI wrote: ↑Thu Sep 15, 2022 8:55 am
I have watched a lot of opinions from many people but I think nobody sees the elephant in the room. I am pretty sure it is impossible to tell if someone is cheating from a single game or even a few games ... if the person has an incredibly powerful computer ... and if the person cheats properly and understands how the cheating algorithms work. Unless it is OTB bullet or blitz ... it is possible to obfuscate the moves enough where it can mask any cheating and still play +200 ELO higher than your rating. I mean imagine for example 7 piece EGTB's ... all the human player would have to do is choose non losing moves even if they are not in the top 5 best moves ... just make humanly logical moves that don't lose and don't go for obscure tactical shots. If you analyze the games of even the greatest top GM's ... there are always a few moves that throw away the advantage immediately or that do not parry an eventual pile up for an attack. The human will eventually get outplayed even in the most dry position. I watched GM Hikaru try and not get adopted (lose 10-0) with pawn odds and you could just see that humans have no chance. Even when he was trying to just draw one game, he would eventually get ouplayed by simple strong moves that were totally human like.
Listen to the podcast. It’s based on quite a bit of data from many years of high level players and all of Niemann’s games since 2020. It’s a good listen.
This video summarize it better than any other video. Why Magnus Carlsen has NOT spoken since he lost to GM Hans, and where is the evidence, his childish behavior is all due because GM Hans Broke his 53 straight winning games streak.
M ANSARI wrote: ↑Thu Sep 15, 2022 8:18 am
I would say that in this latest tournament that MC exited, there is definitely reasonable doubt. Very suspicious play in his games with Mamadyarov, MC and also very suspicious post game analysis (for me the biggest of all the red flags) especially in the game against Firouzja ... but the reasonable doubt holds.
I’d recommend listening to the podcasts with IM Regan, like this one. Regan’s sophisticated analysis (unlike DrCliche’s naive approach) shows no evidence of OTB cheating by Niemann. It also addresses his play in this most recent tournament.
I have watched a lot of opinions from many people but I think nobody sees the elephant in the room. I am pretty sure it is impossible to tell if someone is cheating from a single game or even a few games ... if the person has an incredibly powerful computer ... and if the person cheats properly and understands how the cheating algorithms work. Unless it is OTB bullet or blitz ... it is possible to obfuscate the moves enough where it can mask any cheating and still play +200 ELO higher than your rating. I mean imagine for example 7 piece EGTB's ... all the human player would have to do is choose non losing moves even if they are not in the top 5 best moves ... just make humanly logical moves that don't lose and don't go for obscure tactical shots. If you analyze the games of even the greatest top GM's ... there are always a few moves that throw away the advantage immediately or that do not parry an eventual pile up for an attack. The human will eventually get outplayed even in the most dry position. I watched GM Hikaru try and not get adopted (lose 10-0) with pawn odds and you could just see that humans have no chance. Even when he was trying to just draw one game, he would eventually get ouplayed by simple strong moves that were totally human like.
Ah yes it was Komodo not SF and maybe the 10 0 was not the one with pawn odds. I remember seeing some of the games on Youtube and was really amazing how easily Hikaru would get outplayed even when he was just trying to draw. That was a while back and my guess is that with NNUE engines are even stronger today as they are getting to within 100 ELO points of 4000 ELO. The best human recorded ELO has not yet reached 2900 ... so a huge gap!
I don't recall that event, could you post a link or at least specifics (when played, what engine, what pawn, time limit)? Komodo beat him 10 to 0 in ten minute games once, but that was with no handicap, and we beat him 6.5 to 1.5 giving him two pawns, but no ten game match giving one pawn.
