Cheat Detection - False Positives

Submitted by BorgQueen on Mon, 06/15/2009 at 9:24pm.

I used to serve a large site called gamereplays.org as a global administrator in charge of moderation.  I am called StarTrekker there.  Retired from service there now, but in the process of working for that site, I oversaw many cheat accusation topics where cheats were detected using all the methods at our disposal.  We caught and banned many cheats, costing them their account and their right to play in clanwars.cc.  We did our best to ensure that we had 100% accurate verdicts before pronouncing guilt.  However, every so often there WERE false positives.  Innocent people banned for cheating.

I wonder what the false positive rate is for cheat detection here at chess.com?  I wonder this since I strive for perfection.  Yes, that's a bit of a play on my Borg persona there, but still.  I'll try to explain:

After every game I play is completed, I analyse it with my Deep Fritz 11 engine.  Whether it be OTB or Online Chess.   I judge my performance based on that analysis and I do my best to adapt and learn so I can play better next time. 

Every move is analysed.  I have the top ten engine moves showing during the analysis.  I compare the game moves to the top ten and I note every one of my move's position in the top ten.  Obviously, I aim to get lots of 1's - where Fritz would play what I did.  I often get lots of 1s in a game.  I often get 2s and 3s too.  And yes, I often manage at least one ">10 move" per game lol!  I hate it when I get streaks of those!

When I get moves which are not deemed "best", I let it analyse for a while longer and I study out the position.  I've done this ever since I've owned a chess engine, starting with my Novag Constellation Forte unit.  The aim is to improve by analysing those things that I miss during my game.

Now, in this otb game, when analysed like this, there were lots of 1's.  No, I am not trying to "blow my own trumpet" or brag.  I am just showing that I am capable of getting these 1's, even OTB.  And I am doing that for a reason.  When I play online chess, I am more accurate than OTB, since I have references, opening books, databases, the analysis board, my own physical analysis boards and oodles of time.  Therefore I get many more 1's than OTB.  Every now and then I play a near perfect game (compared to the analysis).  Haven't quite managed that perfect one yet, but I am trying!

So what's my point?  Simply that I'm worried.  I value my chess.com membership very highly and I would sure hate to fall victim to a false positive just because I've played well.

I sure hope the methods used to detect cheats here are accurate! :)

I wonder if any other chess.com members are worried about playing too well?!

Thanks for reading :)

» posted in BorgBlog
 

Comments:

by BorgQueen - 4 months ago
Adelaide Australia
Member Since: Aug 2008
Member Points: 4526

Update:

Someone I know, who shall remain nameless, has identical concerns to me and is seriously contemplating deliberately playing worse to avoid a possible wrong conviction.

This person I mention has been contacted by chess.com to mention that he has been under suspicion of cheating - not a "you are cheating" type message but a "we think you might be cheating and so we're sounding you out" type message.  Now this is a huge concern to me and others.  Why?  Well they obviously DON'T have terribly accurate means of detecting cheats; if they did, then there would be no need for such a message.

I am also very seriously considering playing well below my best on chess.com as I value my account much higher than my rating or record.  I would rather play sub-par and be sure my account remains intact, rather than play my best all the time and get a "we think you're cheating" message such as the person I mentioned... or worse, a wrong conviction. 

And no, I will not mention the name of the person as it would bring unwarranted negative attention.  Cheats deserve all the shame they get, and a punch in the nose imo, but suspected cheats deserve nothing short of the benefit of doubt.

by BorgQueen - 5 months ago
Adelaide Australia
Member Since: Aug 2008
Member Points: 4526

You have to purchase it and then install it like any other software.

I'd do what I can to mentor anyone, but my time is certainly limited.   Join the Borg Collective and we can discuss games and any aspects of chess all you like in our forums.  You just have to put up with all us sci-fi nuts ;)

by gjsgregskyplayer - 5 months ago
Okeechobee United States
Member Since: Apr 2008
Member Points: 166

Borgqueen, How do I get Fritz 11; and make it simple so I can do it. I am computer "challenged" so to say.I also need a mentor, if you are or have the time...Greg

by BorgQueen - 5 months ago
Adelaide Australia
Member Since: Aug 2008
Member Points: 4526

computers still fail strategically...and this is where decent players will catch the computer cheats.

Will they?  I fail strategically too a lot of the time ^^

by Cret1n - 5 months ago
Sydney Australia
Member Since: May 2008
Member Points: 536

I agree. I think that banning is ridiculous since there must be even a small % of doubt in cases where people are banned. Perhaps they should be unable to play for a month if suspected...or perhaps only allowed unrated games?

I think a fair indication of a cheat occurs when a person plays very strong chess and yet plays very fast! I have come up against a few guys like this...no sooner have I spent hours of analysis on my own board when my opponent has a response ready within minutes [every move] Sure it is happening...but computers still fail strategically...and this is where decent players will catch the computer cheats.

by BorgQueen - 5 months ago
Adelaide Australia
Member Since: Aug 2008
Member Points: 4526

No, I haven't tried the computer analysis feature yet.  Thanks!  Good idea, I'll have to try it out with some of my games that after analysis, I have pulled lots of 1's. 

Whether my engine is perfect or not, I have always judged my performance on it.  I probably shouldn't because sometimes, when I analyse normally, my move might get a 8/10... but then I get distracted by something, come back and find that after 5 more minutes of analysis, my move is now like 2/10! Makes me wonder how many of my "1's" are actually not that great a move lol!

Nonetheless, I still strive for the "perfection" of the engine.

Thanks for posting :)

by GeeBH - 5 months ago
Dudley England
Member Since: Dec 2008
Member Points: 1288

I have similar concerns at the moment (just posted some in chess nuts notes).  Have you tried the computer analysis feature on any of your games? It might be interesting to compare chess.com's analyses with yours (via Fritz). My guess would be that the site admins would use this sort of approach, at least in part, to determine how close a player is to the "perfection" (so-called) of an engine.

 

Add your comment:

Join Chess.com for free to add your comment! Already a member? Then login now to comment.