chess.com, Online Cheating

[Tom O'Donnell]

I got this in my messages on chess.com

"Dear Global Chess Community,

Chess.com helps millions of players play millions of games every day. From beginners to World Champions, from casual games to online championships, we are the platform of choice for much of the world’s online chess.

Chess.com feels a deep stewardship for the game. We are driven to provide the most fun and fair experience for our members. And our highest obligation is to keep the game safe from cheating. Our Fair Play Policy is designed to make sure that everyone is playing the same game with the same rules and the same advantages. The chess community is counting on us!

Every single day we close hundreds of accounts for cheating. Among those are the accounts of chess professionals, coaches, and even elite grandmasters. Some of those players have apologized and been given a second chance. Others have threatened us or publicly denied their actions. At times we have been criticized for acting too slowly with top players - but inevitably we come to the right result. We understand that these accusations are heavy and want to be absolutely certain that we are correct enough to defend ourselves in court if needed.

Cheating is wrong. It hurts the game. It gnaws at the conscience of the cheater. It steals from other players - rating points, time, and sometimes money.

Our continued investment in detecting cheating continues to improve our abilities. Our 9 full-time team members dedicated to Fair Play - from data scientists to detectives - have made recent breakthroughs in our algorithm and capabilities, allowing us to detect cheating with much greater accuracy (even a few moves).

This month we have closed more titled player accounts than ever before, and many more closures are coming. Those caught will be permanently banned from Chess.com, the largest online chess platform with more players, more games, and more cash prizes than any other. Being banned from Chess.com means never being able to participate in the largest online chess community, and the world’s best online events.

On Tuesday, April 28th at 12:00PM Pacific Time, I addressed the topic of Fair Play and Cheating in a State of Chess.com address. I invite everyone to watch the full replay of that broadcast, and the items concerning Fair Play specifically here.

I am making a personal plea to everyone who plays on Chess.com to sincerely consider their actions and the consequences that will inevitably follow when you are caught. We have decided to send this letter now to allow any and all players of all levels to come forward privately before the end of the day on Friday, May 1st, and ask for amnesty from the Chess.com Team. Your communication with us will be 100% confidential. This is your last opportunity to save face and move forward without forcing us to close your account. This offer extends to all titled and non-titled players who read this letter and wish to come clean. We understand the unique nature of what this letter is asking people to do. It’s a difficult thing to admit to. We are conservative in our approach to closing titled player accounts, but when we do, we do it with conviction and are absolutely prepared to defend our conclusions.

We are continuing to invest in our team and capabilities. And we are about to close more cheaters on our site than we ever have before. Please hear my plea to never even consider using outside assistance of any kind. It’s a vicious addiction, and one that will catch up to you with unpleasant consequences. If you ever have given in to this temptation, or are currently breaking the rules, please consider doing what's right and contacting our team now for amnesty. You can reach us at fairplay@chess.com. And again, everything you say will be held in the strictest of confidence.

Thank you to everyone who gave this letter their consideration, and I'll see you on Chess.com!

Danny Rensch
Chief Chess Officer, Chess.com
"


There was a link provided which discusses this in more detail:

https://www.youtube.com/watch?v=oRih...outu.be&t=3163

Relevant part starts at about 56:00

- They allegedly caught a bunch of titled players cheating, including three 2600+ players. He doesn't specifically say so, but I assume he means 2600+ FIDE.
- They have adopted a number of new mechanisms to catch cheaters.
- There is an amnesty period until May 1st for people to turn themselves in.
- They have allegedly received over 300(!) confessions from titled players who admit to cheating as a result of some system-wide cheating hunt in March.
- A pile of titled players allegedly will get the boot May 2nd(?). I guess these are people who don't admit to cheating by May 1st.


This is obviously a very expensive endeavor. Rensch notes nine people working full-time on this, plus he implies that there are some part-time people.

Based on the sheer number of people who appear to be cheating I think the CFC should seriously consider completely separating out online ratings from otb ones.

[Vlad Drkulec]

We have been discussing this on the CFC forum and also at the executive level and specifically what seem to be unfounded accusations of cheating by chess.com. I know of at least three instances of players who are not cheaters who have been accused of cheating. The key parallel in all of the cases is that the person changed their focus to another window.

Chess.com's evaluation function is not accurate. I played a line of the exchange Queen's Gambit Declined which the chess.com engine indicated that both I and my opponent played with 98.4% accuracy, which I lost. Komodo put the accuracy numbers much lower at 85% and 79% which seemed more realistic about that particular game.I have seen a similar pattern with one of the CFC members who was accused of cheating. A number of his games showed an accuracy rating of 99.3% or so but if you actually play through the games which were short, they played theoretical lines where the opponent made some blunders which he pounced on. In one instance, there were many opportunities for improvements on a game that indicated a 99% accuracy.

I have been accused of cheating right after I blundered a piece in a highly theoretical Marshall Attack though it was FICS and not chess.com. This happened in 2007 when I was returning to chess after more than a decade of absence from the chess world. I stopped playing online for some time after this.

When I do online lessons many of my students want to play a game as part of the lesson and we often use chess.com as the platform. I usually leave chessbase running in the background so I hope I don't get one of those notices on May 1st, though the games are always unrated. The only time I change my focus is when chess.com disconnects which happens a little too often.

I am suspicious of the claim that they can detect cheating of even a few moves. What does that even mean?

That said, you are probably going to hell if you cheat at chess online so don't do it.

[Serge Archambault]

0123456789

[Aris Marghetis]

Over the years, thousands of non-cheaters were banned from Chess.com . There are entire websites discussing this. In 2018, Danny and Gerard (one of their tech guys) made a Youtube video about their "sophisticated" system against cheaters and viewers were quite critical of them (https://www.youtube.com/watch?v=lYR09lIBsek). One of the problems often talked about is the way they handle those players. They close their accounts first without any warning or discussion. Then, no explanation is given to what exactly triggered this. Then they want non-cheaters to admit they cheated in order to reactivate their account. Over the years, many players admitted having multiple web pages opened while playing on chess.com (very often, browsing through non-chess related web pages) but many also claim the only "problem" they were having is that chess.com site went down as they were playing. Only to come back to a closed account. At the end of the day, most of those having their accounts closed don't even know why it happened. Not given any explanation. They claim there is some kind of appeal system to this but a lot of people have mentionned to just have been ignored after appealing the initial "decision". Also, giving a "second chance" to an admitted cheater makes no sense.

It's tremendously frustrating. One of my students is in that boat now, and it's really grinding us.

[Sam Sharpe]

It's REALLY tough to not be banned on chess.com if you are a correspondence player, as your DB (allowed) is mostly built on the result of engine analysis.

[David Ottosen]

I have spent a lot of time in my professional career working on detection of players trying to cheat online in games, including for using bots. While not true in every single case, I do believe (especially in the case of a chess site where there is zero benefit to the site to wrongly accuse or incorrectly ban a player) that by and large the decisions are correct, especially given the methods used to detect. Trust me, it's significantly more detailed than "let's see how often this player is at 99% accuracy", and the policy is almost always "innocent until proven guilty".

Frankly, I believe that this statement "most of those having their accounts closed don't even know why it happened" is not true - they just don't want to admit that they cheated.

[Kerry Liles]

I have spent a lot of time in my professional career working on detection of players trying to cheat online in games, including for using bots. While not true in every single case, I do believe (especially in the case of a chess site where there is zero benefit to the site to wrongly accuse or incorrectly ban a player) that by and large the decisions are correct, especially given the methods used to detect. Trust me, it's significantly more detailed than "let's see how often this player is at 99% accuracy", and the policy is almost always "innocent until proven guilty".

Frankly, I believe that this statement "most of those having their accounts closed don't even know why it happened" is not true - they just don't want to admit that they cheated.

Perhaps you or someone can answer the question: "if a player's account is locked [ie: for cheating] are they informed of that fact separately and does that include information about what they are accused of doing?"

For example... <our software detected you had multiple browser windows open during the game> or something along those lines?

Several people seem to have experienced their account locked out and ONLY CAN PRESUME they have been accused and apparently convicted of cheating. There also seem to be a large number of reports of NO form of investigation or redress... that surely is not a good thing.
[aside: it would be amazing if anyone could post the details of a situation where they were locked out due to cheating and were reinstated with an explanation of the failure of the detection software]

I have largely avoided online play because I cannot be bothered with this sort of stuff.

PS: what happens if you run online chess in a virtual machine? Does the online software have any ability to detect that? Is that for some reason forbidden?

[Aris Marghetis]

Perhaps you or someone can answer the question: "if a player's account is locked [ie: for cheating] are they informed of that fact separately and does that include information about what they are accused of doing?"

For example... <our software detected you had multiple browser windows open during the game> or something along those lines?

Several people seem to have experienced their account locked out and ONLY CAN PRESUME they have been accused and apparently convicted of cheating. There also seem to be a large number of reports of NO form of investigation or redress... that surely is not a good thing.
[aside: it would be amazing if anyone could post the details of a situation where they were locked out due to cheating and were reinstated with an explanation of the failure of the detection software]

I have largely avoided online play because I cannot be bothered with this sort of stuff.

PS: what happens if you run online chess in a virtual machine? Does the online software have any ability to detect that? Is that for some reason forbidden?

The situation I am dealing with now is quite frustrating because we still don't know exactly WHY the account was blocked. So how to defend the case? Their argument so far is that telling us why the account was blocked could compromise their detection methods. I'm trying to escalate within chess.com support, to at least figure out what non-illegal-but-false-positive behaviours to avoid.

Thanks so far to people like Vlad & Serge who have posted their suspicions about it.

[David Ottosen]

Perhaps you or someone can answer the question: "if a player's account is locked [ie: for cheating] are they informed of that fact separately and does that include information about what they are accused of doing?"

I don't know how they inform players, but I can definitely say it's very unlikely that they'll be given information about how they were caught. Let me give you an example.

Let's say that chess.com somehow determines that a very popular cheating program always takes exactly 0.135 seconds to play it's first move with black. They run their cheating detection algorithms and determine player X always takes 0.135 seconds to play his first move with black. They bar the player for using illegal cheating program.

They would obviously never tell the player "We know you are using cheating program X because you always take 0.135 seconds to play your first move with black", because the cheater would then tweak his program to not do this and go back to cheating on a new account. They would just tell the player "Your business is no longer welcome here".

Can there be false positives? Sure, and it's possible chess.com doesn't have a great appeal process for those players. However, you do have to remember - every time they say "your business is no longer welcome here", it costs them money for very questionable benefit (slight reputational value, but most players aren't even aware of chess.com banning cheaters at all). I'd say they would generally bend over backwards to be certain before barring a player, and it doesn't surprise me that if they do, they are unlikely to provide much in the way of appeals - they likely are close to 100% certain before they take action, and nothing you can say in your appeal is likely to change that.

Further, I can tell you from many years of experience, here is how those appeal conversations go:

Me: You are banned from our site.
Cheater: Why am I banned from the site.
Me: You used cheating program X.
Cheater: That's outrageous, I did no such thing, How can you prove it.
Me: I can't divulge our detection as..
Cheater: It's my right to know!!
Me: It's actually not
Cheater: Do it or else
Me (sighs with regret): You take 0.135 seconds for every first move with black.
Cheater: That proves nothing. Maybe I'm just a consistent player.
Me: rolls eyes
Cheater: You are taking away my rights, I would like your legal contact information and I will be suing you directly for slander and libel.
Me: rolls eyes so hard my hairline recedes
Cheater: you haven't heard the last of this
Me: yes I have

[Lucas Davies]

ugh cheating? come on

[Alex Starks]

I hope chess.com improved their cheating detection algorithm. About two years ago my son was playing 3-day move game against player who had about 500 points lower rating than him when game started. My son was expecting fairly easy win but ended up being very frustrated and disappointed to lose his game. His opponent was making moves that even 10 year old found to be suspiciously good. By the time their game was over his opponent already had more than 100 games winning streak and his rating was way over 2000 at that time.
I decided to run their game in the chess engine and his opponent moves were 100% accurate after the first 3-4 moves.
Even if chess.com algorithm could not detect that cheating was involved I'm wondering how they could miss the fact that someone has over 100 games winning streak and go from beginner level to GM in just a couple of months. I reported my findings to chess.com and they told me that they are going to check it out using their algorithm. There was no any other feedback from them. I checked account couple of days latter and it was closed.

[Kerry Liles]

Ruslan Ponomariov tweeted earlier today a much better "solution" to online cheating:

here: https://twitter.com/Ponomariov/statu...082644995?s=20

For those who don't twitter here is what he recommended:

But there is much better solution: don't let the cheaters know that they were caught. Just make a different pool of pairings that cheaters could only play against each other and have fun to see how they will struggle against each other. Lots of fun and quality games guaranteed!

[Peter McKillop]

Ruslan Ponomariov tweeted earlier today a much better "solution" ....

I wonder if anyone has completed a psychological study of what makes chess cheaters tick? It's hard to imagine someone putting their integrity on the line for such a small return.

[Vlad Drkulec]

I hope chess.com improved their cheating detection algorithm. About two years ago my son was playing 3-day move game against player who had about 500 points lower rating than him when game started. My son was expecting fairly easy win but ended up being very frustrated and disappointed to lose his game. His opponent was making moves that even 10 year old found to be suspiciously good. By the time their game was over his opponent already had more than 100 games winning streak and his rating was way over 2000 at that time.
I decided to run their game in the chess engine and his opponent moves were 100% accurate after the first 3-4 moves.
Even if chess.com algorithm could not detect that cheating was involved I'm wondering how they could miss the fact that someone has over 100 games winning streak and go from beginner level to GM in just a couple of months. I reported my findings to chess.com and they told me that they are going to check it out using their algorithm. There was no any other feedback from them. I checked account couple of days latter and it was closed.

I would expect most players in lessons with a coach would be 100% accurate after the first 3-4 moves unless they are absolute beginners. Most would probably be accurate for the first 10 to 15 moves depending on the level. Most people that read one book would be accurate for at least three or four moves.

Winning 100 games might be suspicious but maybe not if an IM started playing with the 1200 rating that is assigned on your first game on chess.com. If you are a 2500 OTB player playing your first turn based game there are going to be many disappointed opponents along the way.

[Fred Harvey]

I wonder if anyone has completed a psychological study of what makes chess cheaters tick? It's hard to imagine someone putting their integrity on the line for such a small return.

Fundamentally we live in a simple world, where no-body likes to lose, and playing chess on-line isn't really that important.