I have yet to watch that particular video, but my understanding is that 99% correlation is never normal for any level of player when these techniques are used to analyze online games.
they analyzed over the board games. The software was not created to catch Niemann, just to show correlations of games and chess enginges. Interestingly, Niemanns game are total outliers. Because they correlate heavily with chess engines. (when he cheated, there are games where he did not of course).
What I mean is: A 2000 player has to think about whether to exchange a bishop in a certain structure, or whether or not to change the pawn structure.
A 2700 player simply knows or at least “feels” the answer to those questions. Plus he has seen dozens of similar positions at home. So there will be few simple mistakes.
Why would he? There isn’t the tiniest piece of hard proof yet. Just lawyer up and play it cool. At this point he is not guilty.
Circumstantial evidence is often enough, even in a court room.
Yes, but all you’re saying is that a 2700 will play better moves than a 2000. A 100% correlation implies that a player has played the best possible move (as understood by an engine) every time, which is a very different thing to not making mistakes. My understanding from how chess.com and others detect cheating online is that not even the best players get anywhere near 100% correlation with the engines. That’s what their anti-cheating algos rely on; because obviously nobody will ever actually prove that someone cheated online (without spyware installed with every chess.com account.)
All of this is fascinating to me. Many years ago, when I was in graduate school, our entire graduate group decided to get into chess. We played a LOT, subscribed to Chess World or whatever the official magazine was that put your official rating on the address label, and we all ended up with huge libraries of chess books. This is early 80s, and when Sargon came out on the Apple II we all bought a copy (it really was terrible in the end game but could play decent for our level in the opening and mid game.) Progress in computer software was happening rapidly but of course we were on Apple IIs with 1 MHz 6502 processors and 64K (I was a freak and had 128K) memory. We played some tournaments - I never realized how nervous I could get playing chess - and watched kids coming in carrying the thick ECO books under their arms. I beat a kid simply by playing out of book early, and he had memorized every opening 20 moves deep but not the “why” and it completely threw him off. Another game in the same tournament we both ran our clock down but I had a minute or two more than him (maybe 15 seconds) and he rushed and I won. (Ended when I had a superior player just tie me in knots, lol!)
I only got up to about 1525 rating before graduating, marriage, kids, etc. So I just recently decided to jump back in, but wow, it appears everyone has computer chess engines and has run every approach and path a million times and I feel like I’d get slaughtered if I didn’t do the same. The same for poker: I played a lot of holdem poker back before it became a TV sport. Thought it might be fun to get back into, but everyone has poker engines where they run every possible situation and it tells you the proper bets and plays, so everyone has optimal play memorized.
Have computer engines ruined chess and poker for the casual player?
No. Most casual players are still terrible like me. You should be fine! Seriously though, the level at which you need to memorize a bunch of engine lines is much higher than most people who play chess will ever get.
At this point, if you read any chess books they are already informed by the engines anyway. Honestly, most players are not good enough to be able to use the engine to improve dramatically—it takes a lot of focus and study.
Having watched the video, I’m convinced that Hans cheated in over-the-board tournaments.
On the other hand, Magnus lost a game against him by playing badly. That game is not one of Hans’ engine-like outlier games. So his withdrawal still looks an awful lot like a fit of pique; informed by nothing more than whatever rumours were going around about Hans at the time. Assuming your opponent cheated is a nice salve for a world champion’s bruised ego.
Heh, man. Vice dropping a brick on Dlugy about cheating in Titled Tuesday:
maybe. However, you always look bad against an engine. Niemann is a 2400/2500 GM + 200 Elo from cheating, way below Carlsen. Hard to believe Niemann could win with black without engine help.
Even if Niemann came out and admitted cheating over the board and how he did it, but would Claim the win against Carlsen was legit, who would believe it?
This is correct.
But this doesn’t really seem founded in reality to me. If nothing else, then the tournament last week has proven that Niemann is strong enough to make it into the play-offs in a world class field. He was under close observation. Cheating was impossible there.
Any strong GM can beat Carlsen when he’s playing as badly as on that day. (*)
Did you notice that no strong GM has called Niemann out as a relatively weak player? I guess we can take this as a hint that everybody accepts that he is strong.
Your post only shows your own prejudice against Niemann. Maybe he has cheated in offline play, but maybe he has not. What sense does it make to talk him weaker than he is or to blame Magnus’s bad day in the office on Niemann?
(*) = To post at least a little bit of chess knowledge here, let me quote the former world championship challenger Dr Tarrasch: “It’s not enough to be a good player. You also have to play well.”
If the only actual evidence we have of cheating is this comparison with engine moves, then we have to be honest about what our evidence is telling us. There are OTB games in Hans’ record that are clearly impossible 100% matches, in which he cheated. Then there are many games - including the Sinquefield Cup - where his correlation to the engines looks no different to any other GM. If we’re deciding based on evidence and logic, we have to accept that he beat Magnus fair and square.
I picked up a potentially important detail about the Let’s Check analysis several YouTubers used to find engine correlation. The source is a very experienced chess programmer.
Let’s Check uses the Let’s Check online DB. It marks a move as an engine hit if at least one engine suggests this move as strongest move.
There are more than a hundred different engines out there. Hundreds of people analyze Niemann’s games. The results stack. If there are only a few people who are interested in finding a high correlation, they can invest time in using old engines like Fritz 5 or Houdini 2. This could inflate the score significantly.
I’m not sure if the info is correct. But if it is, then then engine correlation search with Let’s Check is dead in the water. and other methods are needed.
Wouldn’t the other competitors have similar results?
I don’t think you’re right about this. The thing is, people who are running these “Let’s Check” comparisons are running them on a bunch of historical games from many GMs, and those 100% results are still massive outliers.
e.g. Bobby Fischer’s 20 game winning streak averages 72%
Edit: What Kevin said.
They do not accumulate results. If I run some tests on some games, against a few engines, then they are not linked to other users doing the test with other engines on the same games.
I’m not sure. Maybe it would be possible to define an environment in which the results of his peers should be similar.
Same age, very similar playing style, very similar openings, similar strength of opponents, same kind of tournaments and fast rating gain over a two year period.
Fischer’s 72% is very high. He’s a dinosaur. Not influenced by computers and from a generation which didn’t take certain sacrifices on general grounds. I would have expected his number to be lower.
I’m not sure how exactly Let’s Play works. I’ve never used it. I only wanted to make the people in this thread aware that there are doubts about its usefulness as a tool for cheat detection.
Actually I don’t even like Niemann. More the opposite. But what I like even less are witch hunts, and suspects not getting a fair trial.
If you’re looking for incontrovertible proof that he is cheated, you’re never going to get it unless he confesses - and I don’t see why he would do that.
The problem with Niemann is that he has a very high base capability (no need to doubt that). That makes it much harder to prove whether he is cheating or not (imagine if Magnus Carlsen was cheating… how would anyone know if he was smart about it?) Niemann’s problem is that it is known that he has cheated in the past (which some would say should disqualify him outright), and there is circumstantial evidence that raises suspicions. Some of his games are absolutely astounding - if he is not cheating, we’re talking future WC/#1 player potential (it’s interesting that Regan’s analysis doesn’t trip on those, but a fair amount of chess experts have reacted at this point). Since the pandemic he has gained something like 150 ELO points, which is absolutely insane at this level of the game in such a short time. So … maybe he studied so well during the pandemic that he is now a far superior player than pre-pandemic. Or given that we know he was cheating pre-pandemic (he admitted to cheating when he was 16 - i.e., 2019), he’s just found a more effective way of (almost) getting away with it. There’s just a lot about his career at this point that just doesn’t pass the smell test.
But yes - there is absolutely a possibility that he is innocent of cheating since 2019. Though - TBH - I actually hope not. Regardless of whether he is innocent or not, his career will likely suffer hugely for a long time to come; Carlsen has flat-out said he refuses to play him again, which means that if an organizer wants to hope for Carlsen (undefeated WC, possibly the best player ever) to play in their tournament, they can’t invite Niemann. And there are other top-level GMs who have made supporting noises. Inviting Niemann carries with it the risk of not only a repeat of someone walking out on him, but also that top players simply stay away from your competition. The whole story is tragic, but it is doubly-so if he is in fact innocent.
