Well the suspect selection criteria I use includes top 0.1% highest rated, high win to loss & draw ratio, number of games in progress vs high-quality opposition, move frequency, wins/draws vs previously banned engine users & a ski-slope ratings graph amongst other things.
When you narrow it down to this extent, finding engine users is a bit like shooting fish in a barrel.
I think the point Diophantus was trying to make, (at least I think its worth making), is that someone may really want someone banned and choose a non random sample of games. Eg. I've gotten in a fight with player x so I choose 200 games that fit the requirements and analyze all of them using the method given in that thread. I then pick out the 20 games with the most damning evidence, and send them to the mods. Even though the average of the 200 games might have something like an 80% top three match up, its reasonable that I could find 20 games which had match up rates of 90% or more.
The other thing I'm wondering is what the probability of banning real players with that system is. For instance, would it be preferable if there was a 5% chance of banning a real player, and catching 95% of cheats, or would it be better if the chance of banning a real player was .1% but only 50% of cheats were caught?
Originally posted by clandarkfireClose but not quite. What I was saying is that the method as used leaves the door open to accusations of this sort. I wasn't saying that anyone would do it,
I think the point Diophantus was trying to make, (at least I think its worth making), is that someone may really want someone banned and choose a non random sample of games. Eg. I've gotten in a fight with player x so I choose 200 games that fit the requirements and analyze all of them using the method given in that thread. I then pick out the 20 games ...[text shortened]... t be better if the chance of banning a real player was .1% but only 50% of cheats were caught?
The mod system here makes it unlikely that anyone with an axe to grind (other than a grumpy mod presumably) could get someone banned on the basis of cherry picked games. From the little I know of how the system works the mods do their own data gathering and analysis exercise which they combine with other information to reach a verdict, thus removing the risk of anyone being banned on the basis of accusation alone.
This is very different to some other sites where players can be hounded from the site because certain people shout "CHEAT!" often enough. This occurs no matter what the evidence and often with no evidence at all!
The test is whether the game selection method has any inherent bias. Selecting the x most recently completed games that meet various qualitative criteria does not introduce any bias. It leads only to more relevant results. There have been many instances of players who have played flawed, inaccurate, human chess for many years before succumbing to their silicon habit. Randomly including games from years ago will not help to answer the question of whether they are currently cheating. Including irrelevant data certainly doesn't improve the sample quality.
In practice, though, not only is the analysis of the games and their results automated, but the selection of the games is automated as well. Mods do not ordinarily sift through games or the completed analyses looking for those to include or exclude in their final evaluation. Typically, mods click a few buttons and receive a pgn batch in their email inbox, which then gets loaded into analysis software, which many, many hours later kicks out a detailed, and repeatable result. The opportunities for subjectivity in the statistical process is really quite limited.
Subjectivity in the interpretation of the results, though, is a completely different story, and may include additional evidence and other factors surrounding a case. Any sanction requires consensus among mods, and more importantly, admins, who are responsible for the final outcome.
- Joined
- 16 Feb 07
- Moves
- 27653
Originally posted by DiophantusI may be totally off, but I think the admins actively dislike these sort of public hangings and are even more reluctant to ban folks when there's a lynch mob. Several cases, maybe even Weyerstrauss's case, make me think that the public lynch mob things tends to cause the admins to react even more slowly in an already slow process.
...This is very different to some other sites where players can be hounded from the site because certain people shout "CHEAT!" often enough. This occurs no matter what the evidence and often with no evidence at all!
Originally posted by MarinkatombThere is no single "official rules" for correspondence chess. Each organization has its own. Most of them explicitly ban engine use: see Section 2 of the link below, which lists cc organizations and rules dealing with computer use:
No one seems to have mentioned that in official Correspondence rules, engine use is permitted. If this guy is a correspondence IM, then he probably used an engine to get the title in the first place. Therefore, there is no reason to believe that he'd have achieved a 2400+ rating unassisted on RHP, if that was the case he'd probably be an IM OTB too..
http://www.correspondencechess.com/campbell/articles/a050531.htm
Note that even organizations which permit engine use do not explicitly sanction their use in their rules. Apparently (unless the author of this essay has made a typographic error and the word should be "illegal" ) the ICCF does permit it.
I was about to say that even in cases where engine use is permitted, it is only done with the full foreknowledge and agreement of both players (e.g., specialized tournaments in which all entrants and prospective entrants know that engine use is allowed and are, indeed, expected to use them); but if the ICCF rules "are silent" on the issue that cannot be the case. Such an omission is unconscionable and should be corrected so that players can make an informed decision whether or not to play under the auspices of a particular organization.
Edit: Since all players at RHP are required to explicitly agree NOT to use chess engines, there can be no ambiguity.
Originally posted by Schach AttackICCF now has specifically non-engine tournaments, as does IECG (LSS now that the email part has ceased operating).
There is no single "official rules" for correspondence chess. Each organization has its own. Most of them explicitly ban engine use: see Section 2 of the link below, which lists cc organizations and rules dealing with computer use:
http://www.correspondencechess.com/campbell/articles/a050531.htm
Note that even organizations which permit engine ...[text shortened]... RHP are required to explicitly agree NOT to use chess engines, there can be no ambiguity.
Originally posted by avalanchethecatThe essay linked to above provides one possible answer:
What I can't figure out is why an IM would use an engine?
"In many cases players assume their opponents are using computers, whatever the rules may say."
Someone with a reputation to preserve might find this even more compelling.
What I don't understand is why anyone who loves the game would use an engine and (explicitly or implicitly) call the moves his own.
Edit: If someone believes that engine use is endemic and largely undetectable and intends to "level the playing field" by using an engine himself, he should confine his play to forums/games where engine use has been explicitly sanctioned.