Vass wrote on 01/09/13 at 08:27:12:


Well to be called a cheater is an extremely serious thing. None of us know this person Ivanov. However if it were you personally who worked hard on their chess, and then had a dream tournament that you would probably never match again that you could brag to friends about and explain how you did it, then it would be very nasty to be disgraced as a cheater or having the label hanging heavily over your head where people would always wonder. Chess accomplishments for an amateur may not mean that much, but they mean something. He has to express himself the way he feels is right, I don't believe there is any correct or incorrect way that a person must respond in such a situation. 

MartinC wrote on 01/09/13 at 12:19:30:

  
Also if you're analyzing a game and look at the first three moves the computer suggests, most of the time, any strong player would tend to pick one of three moves... and Lilov just ignored it when he didn't. The vast majority of the time, there are four or five "candidate" moves. In fact, sometimes the move played wasn't showing at all in the first three moves listed by the computer, Lilov would have to wait and it would change and suddenly Lilov would cheer when it came up.... and occasionally the move he played didn't even come up at all... which he would just write off as being because Houdini wasn't given enough time. 

But most of the time Lilov was just remarking on the "amazing coincidence" that they happened to be one of the computer's top three choices when almost any game of GM strength would show something very similar.    

At times Lilov attempted to make Ivanov's moves to be obviously farcical nonsense based on looking at the position for a few seconds. And I have to admit that it does look very strange to see Ivanov putting a knight on the fifth and allowing the opponent to take it with a cramped knight on the seventh. However it seems a bit too convenient, I mean there are so many times I think GMs do something strange. 

Lilov does seem very convincing at times when he complete ridicules moves, however with a healthy degree of skepticism it would be interesting if people could heavily analyze and see if the move made some sense. At one point Lilov gets excited over the fact that he moves Rec1 instead of Rac1, and was clearly hoping for that to come up as number one in the computer analysis. When it didn't Lilov suggested that maybe he read the move wrong....  
  
Another time Lilov talks about him being a pawn up and losing to a 2000 player after dropping a bishop to a non-obvious endgame tactic, which can hardly be that uncommon especially if someone is running low on time.  
   

MartinC wrote on 01/09/13 at 11:55:26:


I agree with this - computers in shoes and whatnot is great for the paranoid to talk about but in the real world I just cant see how anybody could cheat like that, and stuff in ears/hair etc. surely would be noticeable. tbh the only times I've ever suspected an opponent of cheating is when they are chattering away with other players, and I think a certain amount of casual cheating does go on that way - I wish people would put as much effort into eliminating player communication as they do this paranoid James Bond spy stuff.

Livov actually doesn't seem entirely convincing from that chessbase article:

1) He's doing colleration against the top three choices of the engine. I suppose it'd be possible to randomise what you send when they're close evaluation wise but I dunno.

2) I'm unconvinced by the idea of someone maybe switching to computer cheating like this after round 1 of a tournament. (The earlier one.). It's not a simple thing to set up. Quite a lot of preparation and fore thought.

The camera is a red herring I think, because it'd be orders of magnitude easier to intercept the live broadcast and you'd have far less to conceal on site too.

MartinC wrote on 01/09/13 at 11:16:23:

It is a general misconception that the rating differential is important. You have to look to the number of rating points won in combination with the number of games played or in other words what the probability is to achieve such score against such opponents.

The probability to achieve +400 TPR can be close to 100% but also 0,000001%. A tool to calculate the probability has been developed: http://www.chess-db.com/public/perfprob.jsp

As I said, it's not about the TPR; it's about the chess played. The guy didn't achieve his performance by cashing in on blunders; he tore strong players apart...I wouldn't get hung up on the TPR argument per se.

As for the chess / roulette comparison, what Thorpe et al were doing was inputting physical properties of the spin into their shoe computers, receiving output via same, and then betting while the wheel was still spinning; inputting chess moves would be trivial in comparison...also compare the vigilance of casinos vs chess TDs  I just used the example to show what the state of the art was 50 years ago, so you can begin to imagine what's possible today, especially with a semi-comatose TD playing solitaire on his laptop.

MartinC wrote on 01/09/13 at 11:05:22:


If Ed Thorp could cheat at roulette using a shoe computer in 1961(!), I don't think such a feat is beyond the bounds of human ingenuity, especially when the tournament organisers are asleep at the wheel (unlike casino operators...).

Hummm. Those statistical arguments are basically meaningless. Anyone playing this well in a localised tournament would inevitably play a very strong field.

The grading chart is too because people (even improving juniors) do plateau like that. He gained ~100 elo between Jul 2009 and 2011. Then stable for a couple of years but no reason he couldn't then have improved a bit more to 2250/300 underlying. The gentle uptick in December might suggest that.


Analysis of the games would tell more.

Even there though - has anyone suggested a plausible method by which he might have received entire moves for an entire game? That's very much not an easy problem to solve invisibly.