Independent Computer Chess Discussion Forum
https://open-chess.org/
Is there a real statistical analyses anywhere? I'm clicking around and I see 1. examples, 2. observations and statements, but no 3. stats other than stuff like "99% the same moves as Houdini". The last is above the level of many engines' self similarity, so something else must be meant (or it is made-up). But where is the analyses, taking all of his games, calculate similarity game by game, and do the same with a control group?Jeremy Bernstein wrote:Here's a rebuttal from Andrew Martin: http://www.youtube.com/watch?feature=pl ... zbPytRKY#!
Very tricky situation. I disagree with Martin's stance, but I understand his reluctance to damn Ivanov on the basis of Lilov's videos. The statistical evidence is pretty tough to refute, though. See also: Jens Kotainy, about whom a very lengthy (and very ironic) article in last month's "Schach" was written (with rebuttal by Jens, as well).
jb
You can do a correlation yourself. But I believe this guy:marcelk wrote:Is there a real statistical analyses anywhere? I'm clicking around and I see 1. examples, 2. observations and statements, but no 3. stats other than stuff like "99% the same moves as Houdini". The last is above the level of many engines' self similarity, so something else must be meant (or it is made-up). But where is the analyses, taking all of his games, calculate similarity game by game, and do the same with a control group?Jeremy Bernstein wrote:Here's a rebuttal from Andrew Martin: http://www.youtube.com/watch?feature=pl ... zbPytRKY#!
Very tricky situation. I disagree with Martin's stance, but I understand his reluctance to damn Ivanov on the basis of Lilov's videos. The statistical evidence is pretty tough to refute, though. See also: Jens Kotainy, about whom a very lengthy (and very ironic) article in last month's "Schach" was written (with rebuttal by Jens, as well).
jb
from here, referring to this document. (correction: the document is referring to the Zadar Open, not the Old Capital Open).In a separate interview for ChessBase, Alex Karaivanov contacted Dr. Kenneth Regan, a renowned statistics professor and International Master from the University of Buffalo who have studied extensively the correlation between computer and human moves in tournament games.
Question: Let’s imagine that an effective anti-cheating FIDE commission was already in place at the time of Borislav Ivanov’s participation at the Bulgarian “Old Capital” Open tournament. Would the current sum of statistical evidence, e.g. the unprecedentedly high correlation with the first choice of the chess engine Houdini 3.0, have been sufficient for such a commission to ban Borislav Ivanov from attending FIDE rated tournaments in the future?
Dr. Kenneth Regan: That is exactly the philosophical question posed in my cover letter to the Ivanov report in January. I would like the chess world to determine this, in consultation with experts in other fraud-detection areas. I believe that a commission should be empowered to recommend procedures that would operate for judgments within the chess world. The procedures and judgments must be consistent with applicable laws, but should not require the use of general laws to implement sanctions against players.
Also as my report says, I cannot imagine the statistical evidence in any case being stronger than for Zadar. As it comes to Ivanov’s performance at the “Old Capital” Open in Veliko Tarnovo, even when the games from the first two rounds are included, I get significant deviations in both the MM and AE tests, and an IPR of 3149 with 97.5% confidence above 2940.
A bit of information in FM Valeri Lilov's new video that I regard as most surprising and important is that Ivanov took about ten seconds for most of his moves, regardless of position. GM Kiril Georgiev had stated this for his game against Ivanov in the rapid chess tournament in Kustendil (Bulgaria), where this would be less surprising. If that is true, then it meets the standards of observational evidence of cheating that I had in mind when composing my "Parable of the Golfers" policy page. I regard (outside-evidence + 1,000-1 odds) as having higher-precedence than (statistical-only 1,000,000-1 odds) in human terms, and perhaps this will inform your own deliberations about how a commission should regard the evidence. If this observation is true, then I will give a straight Yes answer to your final question, in my own opinion of guidelines that the chess world still needs to establish for itself. As it comes to my results with strong deviations, they are only based on the Ivanov’s games from Veliko Tarnovo and not Kustendil, for which I have only the game versus GM Georgiev. Sincerely, Ken Regan.
Thanks, that is a good resource (in contrast to those Lilov videos).Jeremy Bernstein wrote:this document. (correction: the document is referring to the Zadar Open, not the Old Capital Open).