Carlsen resigns on move 2

[Angus French]

What's it supposed to mean though? 100% of what?

Which was my first question - in the post I linked to.

[JustinHorton]

Yeah, I saw, but I can't see an answer anywhere! Am I missing something?

[Roger de Coverly]

By the way it's not clear to me (I've never used nor seen the feature) what constitutes a "100% game"

One could speculate that it's a game where every move of the player correlates with a move chosen by an engine, There's a "but" in drawing too many conclusions from this. If they run the test to include the second and third best choices and also run multiple engines, there may be half a dozen or more possible moves in a given position that show up the same evaluation. I hope it's clear that if there are that many plausible moves to choose from, chossing one of them doesn't prove much more than that it's a good player doing the chosing. You could sometimes see this in World Championship coverage. There might be three engines running with three choices each, all different and all showing very similar evaluations.

I come back to the need to demonstrate the physical method of consultation in OTB chess. Does Hans have an ever present second or mentor for example? I'd suspect not.

The more basic question is whether a player who has confessed to online cheating or been disqualified in online events should be allowed to take part in OTB play without sanctions. The ECF, 4NCL and other domestic organisers including team selectors seem content to allow it. FIDE too have only taken action in a limited number of detected OTB cases.

[JustinHorton]

One could speculate

(I should add I wanted to know, i.e. see a description either from Chessbase or somebody who understood the tool. I didn't and don't want to guess.)

[Gerard Killoran]

Although Hans' figures are from far more games, I think, having discussed this in the small and insomniac hours on Twitter last night. By the way it's not clear to me (I've never used nor seen the feature) what constitutes a "100% game" and I may not be alone in this.

I linked to the same Boris Avrukh tweet and asked questions about the Let's Check facility upthread. Other of Niemann's 100% games have been found to contain inaccuracies by Niemann (FM Ingvar Johannesson discusses some of the games on his YouTube channel) and I can't help wonder if the results are dependent on what engine is available to Let's Check and on what hardware the facility is run. If different people are producing Let's Check scores for different players - so that the performances of the players can be compared - are the score calculations under the exact same conditions?

I think Ingvar Johannesson does a good job of blowing this 100% nonsense out of the water. His point that Niemann having a below par performance at Reykjavik despite being able to produce a '100% game' makes no sense at all.

[Geoff Chandler]

No way should a player being banned from a chess site be used as an excuse to ban
any player from any OTB event. Whose chess sites banned list do we use to enforce the ban?
Let's say it's chess.com (a topical choice.)

I'm pretty sure chess.com would not like it if became known that organisers started scanning their banned
list when looking at tournament entries. Players would leave chess.com in droves for fear of an accidental ban.

What may happen (or almost certainly will judging by the current height of paranoia ) players may refuse
to play a banned player. Controllers will then have their work cut for them when making parings.
They must keep the Israelis and the Arab nations apart and now the pious players who refuse to play banned players.

"Hi Alex Mac, How is the tournament going?'
"Quite good Geoff, I have a roomful of people all refusing to play each other!"

[Roger de Coverly]

Malcolm Pein at chessbase and in print at Chess

https://en.chessbase.com/post/chess-edi ... rear-ended

He discusses the numerous transpositional possibilities in the opening, so the positions reached weren't especially novel.

(edit) Talking of chessbase, isn't their "Let's check" feature in effect a publication of theory? So if a player reproduces it 100%, isn't that just evidence of a good memory? From that viewpoint, a world champion who likes to write his own theory by following original paths would have a lower match rate than a GM who worked with the achievements of others.

[Matt Bridgeman]

Malcolm Pein at chessbase and in print at Chess

https://en.chessbase.com/post/chess-edi ... rear-ended

He discusses the numerous transpositional possibilities in the opening, so the positions reached weren't especially novel.

(edit) Talking of chessbase, isn't their "Let's check" feature in effect a publication of theory? So if a player reproduces it 100%, isn't that just evidence of a good memory? From that viewpoint, a world champion who likes to write his own theory by following original paths would have a lower match rate than a GM who worked with the achievements of others.

In the Hikaru streaming video he looked at a few games where the system declined to give a result because the game was too theory heavy. I think he joked it wouldn’t touch any Berlins. So in a sense there is some sort of safe guard there for misleading evaluations.

[Geoff Chandler]

100% games with a positive result will be common amongst the top players.
All you have to do is add up the percentages from the top three choices.
Add in four and five top choices if you want to make your finding to fit
your suspicions and give a false reading.

First you cherry pick. I looked for a positive result in the middle game of the knockout section of the Julius Baer Cup.
I chose that section because that is where chess24 landed me when I clicked my tournament of choice.

Arjun Erigaisi - Quang Liem Le,
And following the Ken Regan method I looked at what Quang played from move 10 to move 32 when the game ended.
That is 23 moves. Quang's moves matched the top three choice of the on sites computer 22 times from 23 moves.
I dare say I could find a 100% cherry picking but this will do to show how easily it to massage figures.

The bogey was 17...g5 which was played IMO in a level position to stir things up.

Top choice - 17
2nd choice - 4
3rd choice 1
21 moves and the 17th move = 23.

The game is 100% sound, nothing suspicious at all and few forced moves, which should be ignored
or not if was on the prowl to insult and insinuate Quang. (I'm not, I'm fudging figures.)
All this from one sound game that we all can see is totally clean.

I'd leave all the hunting and who is really cheating up to the experts who have much more
tools to use, times per move, which I ignored, previous games, algorithms and 'other things.'
(Have you never wondered about those cookies that hit your machine when you log on to play online.)
Play mean but stay clean, you never know who is watching and how they are doing it.

And they spend a lot more time doing it than a dickhead like did with me with 5 minutes to spare.

The game. (which I say again, is clean I've used it to show how easy it is to give lopsided cherry picked data)
https://chess24.com/en/watch/live-tourn ... kout/2/2/4

Edit: forgot to mention, I did not give the on sites computer time to change it's mind. I took it's first 5 second
evo, which is another reason why the above data is flawed. Someone else doing it and the computer may
change it's choice and knock out a few of those top 3 moves.

[Roland Kensdale]

Yeah, I saw, but I can't see an answer anywhere! Am I missing something?

https://m.youtube.com/watch?v=jfPzUgzrOcQ (link given in an earlier post) at 2.13 shows a chessbase help page which attempts an explanation. Not convenient for me to look this up at present, but suspect this is not an adequate explanation.

[Keith Arkell]

Edit: Oops, I'm happy to admit I got that one wrong. I read a stat from 12 September by someone on Boris Avrukh's Twitter claiming an average perf of 2850 by Hans Niemann. I just did the maths and it wasn't over a long period, as I thought, but just for 2 tournaments in a row: the Capablanca Memorial and the TePe Sigeman.

[JustinHorton]

I know that some want to challenge the stat produced by Yosha - Echecs that Nieman has played 10 perfect games during a period of a couple or so years (100% accuracy) while Carlsen in his entire life has only played 2 perfect games and Bobby Fischer none.

Just on it, as it happens, but I didn't know somebody had gone through all of Fischer's games, is that available anywhere?

[JustinHorton]

Ok. The Iglesias video: I've not really had the chance to see it before now. I've made some notes when watching (which in itself is perhaps not the best way to follow what it's saying, but what can you do) and I have a few observations and questions.

1. Quite early on, Iglesias says that she never came across the Let's Check function until the previous day. OK, it's not a disqualifcation offence, but ideally you'd want statistical analysis to be undertaken by people familiar with the tools involved! That said, she has received help from somebody who is familiar with the function (somebody I am aware of, incidentally, and would vouch for) but even so, it's not the best.

2. It's a bit of a problem looking at the explanation at 2:13 because you can't see all of it! OK she does read it out but it's no real substitute. Fortunately the text can be found here, to which link I'll return.

3. That said, the information "this value shows the relation between the moves made in the game and those suggested by the engines" is of limited assistance unless we know what's meant by "the engines" - which engines? to score, do you need to match the top choice of any one of a number of them? - and what ply we're talking about.

4. As the Reddit piece I linked to says, the Let's Check function specifically says "this correlation [the percentage figure people are getting excited about - ejh] isn’t a sign of computer cheating, because strong players can reach high values in tactically simple games" and as Iglesias then proceeds to look at a number of individual games on the basis that a 100% score makes them suspicious, she seems to be ignoring that advice.

5. I didn't follow everything she said (not least because making notes makes you lose your concentration) but I don't think - correct me if I am wrong - that she said anything about the different strength of opposition that Hans faces, compared to Magnus. I mention this partly in connection with the figures produced here, which purport to show that Hans outscores Magnus in 100% games by five to every one of Carlsen's, and it's suggested that it is a great deal easier to score a high percentage when you are playing weaker opponents. (But I don't know this. I imagine a properly rigorous statistical analysis would prove or disprove it.)

Also, as an aside, it's been suggested that the figures in that tweet accidentally miss out Carlsen's 2019 games. Is this so? I have not checked.

6. The bit that really interested me in the video was the observation that Hans scored a high overall percentage, i.e. measuring over a tournament rather than a single game, in a number of consecutive tournaments. There's some calculation of the probability of that, which I don't know whether it's statistically valid: but even ignoring that, it does catch the eye. Then again, players do have streaks (hey, I was 39 games unbeaten on lichess until yesterday).

I hope that makes sense. Apologies for not coming to any conclusions and also if I've missed or misinterpreted anything. I was out of my depth for some of it, people with more experience in the field may do better.

[JustinHorton]

Meanwhile Vice have raised the temperature further with this piece which should enhance the reputation for integrity of neither Max Dlugy nor Danny Rensch.

[NickFaulks]

However, it is even harder to argue with the easily calculable statistic that to go from 2508 to 2700 during 1 year and 5 months required of Niemann a performance level of 2850. This is indisputable and very easy to calculate.

For the benefit of those of us whose maths isn't too sharp, could you explain why it is indisputable and very easy to calculate? I can't get anywhere near it.