< Earlier Kibitzing · PAGE 2 OF 2 ·
|Jan-06-13|| ||senojes: Having read the article, "Bulgarian player strip searched after suspection [sic] of cheating," by Peter Doggers, Chessvibes.com, December 30, 2012, I copy-and-pasted the score of some of Ivanov's 9 games in the Zadar Open `A' chess tournament held 16-22 December 2012, given in the article, into my Houdini 3 chess engine running under Fritz 13 on an Intel i7-3770 quad core processor. |
I then compared Ivanov's moves with Houdini 3's first recommendation. Allowing for the fact that the order of Houdini's choices can sometimes change the longer it is calculating a move, and that I am not infallible, my tentative findings are:
Round 1. Ivanov, B. (2277) v Schachinger, M. (2422), 1-0 (this game above): From Ivanov's move 9.Bg5 to his 35.Nf6+, the last move in the game, EVERY ONE OF IVANOV'S 27 MOVES WERE IDENTICAL TO HOUDINI 3'S TOP RECOMMENDATION.
The probability that Ivanov, a 2722 rated player would by chance find exactly the same 27 moves in a row as a chess engine like Houdini, which has an ELO rating of 3334 (Wikipedia), or indeed any other chessplayer, would literally be ASTRONOMICAL.
And as we shall see, this is not the only game in that tournament where a long series of Ivanov's moves were identical to Houdini 3's first recommendation.
It will be interesting to see if Ivanov is charged with fraud over this, as he may be since this is a major challenge to the integrity of FIDE and other chess tournaments and presumably the prize-money was significant and he tied for second place, what his defence (if any) would be.
The `just coincidence' argument wouldn't work. I have a friend who is a retired forensic scientist who often had to appear as an expert witness in criminal courts and he told me that the entire `DNA fingerprint' argument is based on the billions to one improbability that two unrelated individuals could share the same satellite DNA structure. As a last resort defence lawyers sometimes argue it is `just coincidence' that their client's DNA matches that found at a crime scene. But juries almost always reject that argument as not being a reasonable doubt.
But the improbability that 27 moves in a chess game (except perhaps for some long opening theory variations) could by sheer coincidence be identical would be many orders of magnitude greater than two unrelated individuals sharing the same satellite DNA structure. And as we shall see in my comments on Ivanov other games in this tournament, he made similar long series of moves in those games which were identical to the first recommendations of Houdini 3.
Stephen E. Jones
|Jan-06-13|| ||csmath: The game is clearly chess engine game, by Houdini. Ivanov lost the last round after transmissions of the games were stopped. |
He definitely had outside help through some type of communication device that was not found.
This is similar to Singh's case. At the time Singh was cheating with Sharma there were also people here talking how he was achieving that by himself. Lo and behold since then he was never able to repeat any result he did during cheating.
Watch for Ivanov, he is definitely cheater in my view, I analized couple of "his" winning games including this one and they are Houdini engine games barring coincidence which is not something I believe in. Not even Carlsen can play tactically so precise as Ivanov "did."
|Jan-06-13|| ||csmath: By the way, GM Klaric is of the same opinion, that Ivanov concealed communication device they were not able to find.|
|Jan-06-13|| ||senojes: To get some idea of the improbability of two chess games containing a series of 27 identical moves in a row by chance:|
Assume for the sake of illustration that there are 4 candidate moves (e.g. as listed by a chess engine) in each of 27 positions in a chess game. The probability that two players (or one player and a chess engine) would play a series of 27 identical moves in a row by chance in a game would then be of the order of 1 in 4^27, i.e. 1 in ~1.8x10^16 or 18 followed by 15 zeros. To be more exact, using Windows 7's calculator, it would be 1 in 18,014,398,509,481,984!
If someone wants to quibble about the number of candidate moves then they can plug their own estimate in. I used 4 candidate moves per move to be conservative. The improbability will still be ASTRONOMICAL that Ivanov (rating 2277) by chance played exactly the same move as Houdini 3 (rating 3344) for 27 moves in a row in a game of chess.
And as we shall see, that also happened in at least one other Ivanov game (I haven't checked then all out yet) in the same tournament.
Stephen E. Jones
|Jan-08-13|| ||Maatalkko: <senojes> I have to quibble a little with your math. In a wide-open position, sure there might be an average of four candidate moves that a GM would consider. However, there are also plenty of positions with forced recaptures, obvious plans, etc. where there are either one or two possibilities. For example, if there are 27 moves but seven are forced, then 4^20 is "only" one in ten trillion. |
Obviously, your point remains valid. Try to play a consultation game with a friend of similar strength to you, and you will see how quickly two equal players will choose totally divergent paths. The odds of ANY two players choosing 27 identical moves are basically zero, let alone two players 1000 points apart in strength.
|Jan-08-13|| ||csmath: There are at least two games by Ivanov that match move-by-move Houdini engine.
The best chess players in the world cannot do this, they deviate and sometimes not bad at all but they deviate. This guy was clearly using chess engine and not just any engine. He was using Houdini version 2.|
|Jan-08-13|| ||csmath: When Topalov was accussed of cheating in the aftermath of San Luis and in Corus the evidence about usage of chess engine was not there. In the Feller's case the statistical evidence is vague although he was obviously cheating. Similarly in Singh's case some time ago. In this case I think there is a strong statistical evidence that Ivanov has been cheating in at least two of "his" best games.|
|Jan-08-13|| ||ikipemiko: <csmath> who is the other cheater from Bulgaria?|
|Jan-08-13|| ||Maatalkko: <csmath> The difference is that Feller was cheating on key moves and executing simple moves himself, whereas Ivanov was cheating move-by-move. Topalov, if he was cheating (I am not implying anything either way), would also have been a key move cheater. |
Key move cheating can't be caught using pure statistics because some of the moves are human. It just appears that the cheater makes brilliant decisions in key situations, like 48. Na5! and 50. Nc6!! in Bindrich - Tregubov. Key move cheating is the great potential scourge of GM competition.
What Ivanov did was much more crude: he took the first choice move of the first choice program. He must have really wanted the 2,200 euros for some reason, and he probably would have gotten them if not for the broadcast shut down in round 8.
Ivanov is not strong enough to be a key move cheater. His independent moves against Jovanic were blunders, and he failed to generate a coherent plan as White against Predojevic. Hopefully, nobody will make the mistake of broadcasting Ivanov's games in the future and 15 minute delay will be mandated soon.
|Jan-09-13|| ||Mendrys: I'm not so sure. Perhaps time will tell but at this point Ivanov has not been sanctioned nor has any "real" evidence of him cheating been brought forward. It's always easy to say "well, perhaps he hid the device he cheated with so well they couldn't find it after strip searching him."|
|Jan-09-13|| ||perfidious: <Mendrys> Ivanov is most fortunate that he is not facing some of the posters here, as he would already be drawn and quartered.|
While circumstantial evidence may be strong, that is not enough in the free world. Maybe proof will turn up down the road.
|Jan-10-13|| ||Maatalkko: Well, there may not be enough evidence to convict him beyond a reasonable doubt, but I think there is a preponderance of evidence. It's like the OJ case; though he wasn't convicted beyond a reasonable doubt, he was successfully sued in civil court and nobody pretends he is innocent. |
<perfidious>, have you seen Ivanov's game against Predojevic? He is hopelessly outclassed against a 2600 when the broadcast is down and then plays identical to Houdini (thus better than Carlsen) against similar players when the broadcast is live. I think chess is in trouble if we are willing to call such results plausible because cheating technology will be easier and easier to access over time.
|Jan-10-13|| ||Jim Bartle: In his excellent video Lilov posted at chessbase suggests tournaments install equipment which jams all electronic communication. |
I think it will be interesting to see when Ivanov plays again, and how well. If he plays soon and has good results against good players we ought to figure he's playing honestly. Otherwise, probably not.Ee
|Jan-10-13|| ||perfidious: <Maatalkko> Ivanov-Predojevic was a nasty business, featuring boatloads of weak play by White, and resembles what GMs have done to me from time to time.|
You're quite right; it will be progressively more difficult to catch cheaters out as time passes.
Solutions to fight the obvious tendency to gain access to such programmes do not readily admit of themselves-perhaps full body armour and sensors capable of detecting anything remotely resembling software, as one is in the tournament hall. If anything is detected, blood on the rafters!
|Jan-11-13|| ||Shams: <Jim Bartle><I think it will be interesting to see when Ivanov plays again, and how well. If he plays soon and has good results against good players we ought to figure he's playing honestly. >|
There was an egregious case of cheating a year or so ago at an American tournament, which was proved when the player in question failed a rudimentary chess exam after his last round game. (As I recall the exam consisted of a skeleton position with an open e-file and a back-rank mate and the player in question did not find 1.Qe8+ Ra1xe8 2.Rxe8#. Such a demonstration should satisfy anyone that the player was cheating when he racked up win after win in the tournament.
Which leads me to wonder, why not have a committee of GMs quiz a suspected cheat when the game finishes? If Ivanov found each of his winning moves OTB he should be able to justify them in a post-mortem. Give his intuition the benefit of the doubt, even-- he needn't prove a win in every line, just a winning idea. But if he badly flails and is unable to explain his play, that would be further damning evidence. Of course, this approach is more viable when the suspected cheater is an FM and not a GM.
|Jan-17-13|| ||dpak: dpak: <senojes> "The probability that two players (or one player and a chess engine) would play a series of 27 identical moves in a row by chance in a game would then be of the order of 1 in 4^27, i.e. 1 in ~1.8x10^16 or 18 followed by 15 zeros"|
Only if choice of moves would be expected to be random. But players form plans and have styles, so the probability of a random match is meaningless.
<csmath> "This guy was clearly using chess engine and not just any engine. He was using Houdini version 2."
Another source compared his games to Stockfish.
The question is can anyone explain - or better still PROVE - HOW the moves were conveyed to him, given that a search failed to detect any means? To use senojes' forensic analogy: if a suspect's fingerprints are found on the murder weapon, that is highly incriminating. But if a medical examination establishes that he has lost the use of his hands (in an accident prior to the murder) then that would tend to rather undermine the prosecution's case.
|Jan-18-13|| ||goldenbear: These moves are so logical. This is the kind of position where it is possible to find computer moves. Notice Ivanov plays the "human" 24.R6d2, instead of 24.Rd7!.|
|Jan-21-13|| ||senojes: If Ivanov has been reported correctly in Chessbase News, "Cheating scandal – Borislav Ivanov speaks out," 17.01.2013:|
How did you manage to beat the Croatian grandmasters?
I dont think there was anything special in my games at all. The Croatian GMs made some horrible mistakes during our games. Of course I practiced a lot with the computer, and after beating Rybka and Houdini by 10-0 each, i was absolutelly sure that no-one was gonna stop me winning.
then either he is the strongest chess player who has ever lived, far stronger than Rybka (rating 3136) and Houdini (rating 3253), or he is lying.
Because even Magnus Carlsen (rating 2861) or World Champion Anand (rating 2772), could not beat Rybka and Houdini "by 10-0 each". Indeed, as far as I am aware, neither Carlsen nor Anand could be expected to beat Rybka or Houdini in even 1 game out of 10.
So how could mere 2342 rated, non FIDE titled, Ivanov do it, not just once but "by 10-0 each"!
|Jan-21-13|| ||perfidious: <senojes> Those engines are lucky they don't have to face me; if Ivanov can beat them with clean scores, they'd have no chance in a match against me, either!|
They would probably take a game or two, though.
|Feb-04-13|| ||goldenbear: What is White's idea if 12.Nxe4? I must be missing something...|
|Feb-04-13|| ||Shams: <goldenbear> Per Shredder, White has the tiniest of edges on 12...Nxe4 13.Nxe4 Bxe5 14.Nf6+ Bxf6 15.Bxf6. I suppose Black didn't fancy selling out the dark squares like that though.|
|Feb-04-13|| ||goldenbear: <Shams> Thanks. I'm suprised an IM wouldn't play that in a heartbeat. Why give up a pawn? Good to know I wasn't being blind.|
|May-13-13|| ||Jigsaw42: Ivanov is definitely a cheater.|
|Jun-07-13|| ||Catholic Bishop: <There was an egregious case of cheating a year or so ago at an American tournament, which was proved when the player in question failed a rudimentary chess exam after his last round game. (As I recall the exam consisted of a skeleton position with an open e-file and a back-rank mate and the player in question did not find 1.Qe8+ Ra1xe8 2.Rxe8#. Such a demonstration should satisfy anyone that the player was cheating when he racked up win after win in the tournament.>|
Yes, then the current women's world champion would be convicted a cheater after miserably failing a BN checkmate
|Jun-07-13|| ||Shams: <Catholic Bishop> Embarrassing for her no doubt but not the same as missing 800 level tactics.|
< Earlier Kibitzing · PAGE 2 OF 2 ·