Cheating!!!!

[tama]

I have reason to beleave that some of the top 10 players could be using a computer aid in solving problems. Generaly in Standard mode.

[Richard: I've removed the name of user as nothing is proven yet]

The reason is his record is usualy a mix of getting problems right n wrong, but right now he was a perfect 20 problems correct with almost all of them being over 2000, and he even got a problem correct that no one in the history of the site has ever solved, during his amazing streake. If u look at his standard rating chart he jumps from 1600 to over 2000 in about 2 days!! He solved over 500 problems strugaling to be at 1600 then sudenly he flys to over 2000!? Explain that?

Now im not saing that everyone is cheating because any master or expert can get a rating of 2000 + on this site in standard, but it is a high posibility that there are some users that cheat.

And even if they do cheat its not effecting other users expirienses on ChessTempo.

[tacto]

I actually agree with you on this one perhaps...

If your 2000 + Standard your DEFINITELY 2500 + Blitz here. He also jumped too fast. But who knows maybe he has insane luck for those few problems.

I don't care about cheating though. I use the ratings to track my progress.

I would like it if we had a bellcurve/normal distribution chart to compare ourself within percentiles, but hey, you can't have it all lol.


EDIT: It could also be due to maybe a change in his RD or whatever.

[drahacikfm]

Seems suspicious, yes.  Rating going nowhere for months, then boom!  Up 370 points today.  He did a few problems on May 22 and May 28, and lost rating points both of those days.

Also, time spent on each problem is pretty similar, usually between 80 and 120 seconds.  Enough time to set up the position on the computer and let it think a little.  My times for standard rated problems are all over the place, from 20 seconds to 10 minutes.

[richard]

Last session seems to have been computer assisted.  Too different from previous performance, and got a few problems right that had very computerish best lines that nobody else had gotten correct. I need to add some policy to discourage this and make it clear what happens when such things are detected.  At the moment I don't actually say anywhere, "Don't use a computer to help you solve problems". I'll add an entry to the FAQ for now pending a more detail statement.

I don't care that much about computer assistance as such as the players that are using it simply waste their time (probably even more so than when using a computer in an actual game against an unsuspecting human).  However I do care from the point of view of computer assistance compromising the accuracy of the problem's ratings.

This type of behaviour is always going to be an issue in standard mode, however there are automated tests I can introduce to try and detect it if it becomes too much of a problem (this is only the second case of this type of issue I've seen so far).

If people suspect something strange in the future, it is probably better if they email me privately rather than post on the forum as it is unfair to the person involved if they are in fact just having a lucky run to be publicly accused.

EDIT: I've checked the rest of the (current) top 10 standard users and I am confident they are all legitimate.

Regards,
Richard.

[drahacikfm]

You can be confident I don't use a computer, because you can see that sometimes I miss mate-in-one   

[slacker00]

I've wondered when we'd get  our first "incident" of cheating by computer.  It probably wouldn't  be too tough to do, especially in standard mode. 

But it's just sad.  I mean, what's the point?  It seems like it would just  ruin all of the fun. 

[tama]

You are absolutaly Right! Ive bin having lots of fun increasing my rating but it takes time. !~People that cheat should be Cut Down! Drop KIck Style~!

You no it takes so long to increase when your RD is low, like less then 30. If i start a new acount i would be passed 2000!

[chipschap]

Although it is true that problem ratings are impacted, anyone who cheats here is just cheating himself.  It's so pointless and inane to do that!

I am here to learn and improve.  If someone else is not, that is their loss.

[boybawang]

some chess clients monitor cpu resources to detect computer assistance. But simpler is an automated screen-capture on suspected members display.  Then post here his screen which shows big Fritz!  then Tag as busted!

[richard]

Screenshots are a nice idea (a bit of a privacy issue though , what if the user was also accessing their bank account [or worse :-) ] in another window). Things are a bit more limited on a web application, I do have some anti-cheating ideas but will not be spending time on them until cheating becomes a problem which it hasn't really yet (a couple of isolated cases).

Regards,
Richard.

[buddywinston]

If someone is proved cheating, get rid of him immediately. I used to think they were a mild problem, but  after spending alot  of  time and effort playing the skunks on a correspondence site, losing, and finding out later  they barely knew  the moves of the pieces i have no tolerance.  fortunately, on a site like this they will have little impact.  Normal people  can't see the point, but they're out there.  Once i had  a conversation with one on ICC.  I  said, "Why do you do it?"  He said, "For respect."  I said, "What about  self-respect?"  He said, "I have plenty of  that, i just want other people to know how good i am."  Go figure a mind that works like that.

[richard]

Hi buddywinston,

Yes, cheating is certainly the scourge of online chess sites (online sites with any competitive component at all really). As you point out it doesn't impact other players much on here, but it is still something I like to keep an eye on as it could skew problem ratings if it became a widespread issue.

Regards,
Richard.

[tama]

I don't want to sound envies but there are about 2 users on the best list that take less then 80 seconds to solve there problems in standard. Now i don't know if there Grand Masters or something but it takes me at least 5 min to accurately give a clean calculated solution.

Also there performance in the past inst the best in the world, so perhaps some one could check them out.

[richard]

Hi tama,

I don't see any obvious evidence of cheating in the standard results at the moment. There is one user that has had an unusual rating improvement but if you look at their history (further back than 100 problems so you will not be able to see it) they changed their solving behaviour and started taking MUCH longer to solve problems.

The other thing to keep in mind is that there were a lot more high rated problems before the alternatives were added.  Adding alternatives did two things 1) Made artificially hard (due to good alternatives being marked wrong) problems easier and 2) Better detection of when alternatives occurred meant that some harder problems were removed due to having too many alternatives.  These changes meant that the higher rated users got a spike in their rating as before they were getting some problems around 2200-2300 but after most of those very highly rated problems were gone they got more problems below their current rating (this is why users in the middle of the pack have about a 50% solve rate , but users in the very highest rating range in standard have solve rates over 80% in some cases).

There are no GMs in the standard ratings top 10 (AFAIK),  there was a GM on the top of the blitz ratings last week but they have become inactive now so have dropped out of the top 10 based on inactivity.  However there is often at least 1 IM and 1 FM in the top 10 standard (although the IM has become inactive in the last few days), several of the top 10 standard users have OTB ratings >2000 which makes them pretty good at tactics. CT probably has trickier problems than CTS but for some of these guys they really need to be given harder problems , unfortunately the better I get at removing ambiguity, the harder it is to produce really difficult (and long) problems.  Hopefully if I get the worst of the remaining quality problems resolved in the next update then I can concentrate on producing only new problems for a while, which will help provide a few more hard problems.

It is still possible that there might be some computer assistance going on, but it is not clear to me that anyone is currently  doing this. I do have some ideas on more active ways of detecting cheating in standard mode and I'll start rolling these out if things become too problematic.  One of the nasty things about cheating is that even if it is rare, it makes everyone a little suspicious and sometimes users can get falsely accused.

Regards,
Richard.

[newlook]

I think that all accusations of cheating should be sent via PM to Richard, not posted in the forum. No one else can do anything about it, and false public accusations are bound to create bad feeling.