The Chess Thread - Page 111

On September 28 2022 14:55 Magic Powers wrote:


Niemann has played against opponents that are much lower rated than those of Carlsen. If he's overperforming because of his skill, then it's completely expected that he plays more games with higher engine correlation, because his opponents play moves that are more easily refuted by a highly skilled human player. Imagine if Carlsen started beating up 2200 to 2500 ELO rated players, he'd completely destroy them with a much greater than usual accuracy, too.

The ELO rating of Niemann's opponents during those ten selected games was 2558, 2430, 2283, 2427, 2454, 2204, 2376, 2398, 2466 and 2542. That ranges from 2204 to 2558 and gives an average of ~2414.
In those same games Niemann was rated 2478, 2459, 2439, 2465, 2526, 2567, 2571, 2606, 2609 and 2637. That ranges from 2439 to 2637 and gives an average of ~2536. The distance between the total averages is +122. The distance in each individual game is -80, +29, +156, +38, +72, +363, +195, +208, +143 and +95. This fact really sticks out, it gives Niemann a huge advantage in six of those games, and a small advantage in another three. He only had a rating disadvantage once, i.e. 10% of the time over all the selected games.
So the first thing we can see is that Niemann had a very big rating advantage over most of his opponents during these ten games, which means we're comparing apples (Niemann's rise to the top) to oranges (Carlsen's defending of the throne). If we only look at Carlsen's games where his opponents are much lower rated, we should also see him having a very high accuracy, and that's an analysis that I haven't seen anyone do yet for Carlsen or any other top rated player.

https://lichess.org/study/ffYRNE1u

So far I've analyzed three of the ten alleged 100% accuracy games using Stockfish 15 NNUE.

In the first game in the list against Matthieu Cornette, Niemann played only natural moves the entire game. He outplayed his opponent with moves that every GM would consider normal.

In the second game in the list against Christopher Woojin Yoo, Niemann plays 6... d6, which costs ~0.8 evaluation points, but it leads to a commonly known position.
8... Bd7 costs ~0.4.
10... h6 costs ~0.4 (it's also one of the weirdest moves to make. Niemann can eliminate the knight and quickly castle kingside. Instead he blocks the Bg5 idea, which is not as strong compared to the bigger threat of white lining up his rook against the unprotected king. So Niemann played a move that no engine would recommend, and humans would only recommend it if they have an attacking plan, which was probably considered already while playing d6).
His opponent then gives away his advantage with Qc3 and also plays f3, which allows Niemann to go for a strong attacking idea that many GMs should be aware of, busting open the center and creating a mating attack.
Niemann probably set up the idea of castling queenside already when he played 10... h6, leaving him with attacking options down the line. This would also explain the nature of the move.

This is only one of the many examples where Niemann outplayed his opponent, and the engine correlation was certainly not 100%. It's strictly a lie to claim that it was, because all engines would recommend capturing the knight on f5 on move 10 and quickly castling kingside. Maybe Niemann thought there was a Qc6+ idea in that line that could give him trouble, or maybe he wanted to create an attack himself. All engines would also recommend against 6... d6, this opening idea is entirely human and leads to double-edged, imbalanced positions. It's fairly obvious that Niemann had attacking plans the whole game.

In the third game in the list against Miguel Angel Soto, Niemann goes for another imbalanced position, again castling queenside (might be thematic for him?). Both he and his opponent have clearly done their homework and are playing only the top engine moves until move 18 or 19. Wow, look at how Soto is playing a perfect game for so many moves, he must be a cheater. Or maybe they both know their theory, seems like an alternative explanation, right?
Finally Soto makes a small inaccuracy with 19. h3, but it's the second best move so he's probably still cheating. Niemann supports the knight with the very common idea of 19... h5. Every GM knows this idea, in fact most of my opponents would also play this move in 3 minute blitz. Soto has to continue with 20. f3, but he blunders the whole game instantly with 20. Na3 instead. Niemann pounces with the tactical shot of f3. Everyone at 1700+ would see this move. Soto takes the knight on g4, Niemann recaptures, and the rest of the game is easy going because the white king can't escape. Soto resigns soon after.
This game is another example of Niemann outplaying his opponent with obvious moves, there's nothing suspicious about it.

Do you want me to keep going?

On September 28 2022 06:59 Jockmcplop wrote:


Finegold makes great points.

Magnus and all the other super GMs played in a tournament with Hans 10 days before the drama started, and none of them were concerned about him cheating.

As soon as he beats Magnus, Magnus wants to bring up all this old shit about him cheating, even though when Magnus beat him less than two weeks earlier, he was perfectly happy to play him and not say anything.

For the record, Hans won the first game in Miami and then Magnus beat him 3 times. So Hans won a game, but Magnus said nothing about Hans cheating and everyone left the tournament happy.

Magnus loses and suddenly its a problem. Its pathetic.

On September 28 2022 19:40 Magic Powers wrote:


This video features the whole game:
https://www.youtube.com/watch?v=yn4Wd9NjMDk

First thing to note: Abhimanyu Mishra is a player who - at the time - was much lower rated than Niemann. The game itself was for Niemann's final GM norm. Just a bit of relevant information, now on to the game.

That Nh2 move is not suspicious in the slightest, it's a top candidate move that every GM would want to look at and calculate in some depth.
It should also be noted that 16... Nfg4 is a very good move that is about as critical and hard to calculate as Nh2 later.
In the critical position after 17. Qe2, black's knight on e5 is not attacked, so 17... Nh2 becomes a valid option that wouldn't have been available if 17. d4 was played instead. In fact without 17... Nh2 black would have to give white the advantage, so chances are Niemann saw this idea several moves earlier. I'd argue it's not a particularly difficult line to find for GMs, it's much harder to evaluate it correctly (which is a super important skill). In the video it is explained that 17... 0-0-0 would allow 18. d4 (a similar problem as in the other position). So this whole thing is very thematic, not hard to see for GMs. It's also very common to try to restrict the enemy king with a move like 17... Nh2, so it appears to have only upsides and is therefore highly attractive. It'd be surprising if Niemann didn't see this whole idea ahead of time.

Furthermore, Niemann played several incorrect moves in this game.
14... g3 would've been the top engine move giving a greater advantage to black. 14... Bd7 is the second top choice.
22... Nxf4 is the top engine choice with a potentially winning chance for black, and it's also a very human move. Re8 being the second top choice gives white chances to hold, but Niemann apparently turned his engine off at this very moment.
24... Rh6 is the top engine choice, while c6 is a mistake that costs black the whole advantage. The position is now equal.
25... Rh6 is again the top engine choice, while d5 gives white better chances to win.
33... Ke7 (and several other moves) retains black's winning advantage, while b5 gives most of it away. This is strictly a blunder and white has several good responses to it, 34. Bh7 being the best one. But luckily Niemann has the ability to use mind tricks, forcing his opponent to play 34. a4

The rest of the game however Niemann turns on his engine and plays only top engine moves, which is why he wins the game.

On September 29 2022 00:23 Magic Powers wrote:
Selection bias is very strong in the chess community right now. They point out when Niemann looks suspicious (which every super GM does every so often, in particular Carlsen), but not when he looks more than legit.
To give one example of the heavy bias against Niemann, his classical rating improved from ~2300 to ~2500 over a span of 5 years. Want to take a guess how fast Carlsen's journey from 2300 to 2500 was? 1 year. 5x shorter.
But which fact is being pointed out instead of that? Niemann's journey from 2500 to 2700. As if it's not possible that this is just a result of some variance or an unknown bias other than a manipulated performance. The most likely explanation is that Niemann was strongly underrated until more recently.

On September 29 2022 02:33 pmh wrote:



Its funny that you mention this. Because that was another aspect that i found odd after looking at a few things a bit more closely.

His rating progress chart does not make sense. If he is indeed a true 2700 player at age 19 then he is one of the biggest talents in chess. Yet he took stuck 5 years to go from 2300-2500 and was stuck at ~2300 for like 2 years.
This to me does not make any sense,it is not consistent with the typical progress charts of the biggest talents. Where people go from 2300 to 2500 way faster then from 2500 to 2700. This in itself doesnt proof anything obviously but it doess add to the pile.

On September 29 2022 05:16 Magic Powers wrote:
I know there are at least two other players who went from 2500 to 2700 in under two years. Can't remember their names right now. I think it's something that can realistically happen within one year if a player is underrated.

On September 29 2022 05:16 Magic Powers wrote:
I know there are at least two other players who went from 2500 to 2700 in under two years. Can't remember their names right now. I think it's something that can realistically happen within one year if a player is underrated.