Cheating in chess

[Roger de Coverly]

It looks as if the 4NCL decided to withhold a prize in its Congress to a player who was noted shortly after the completion of the event to have had their lichess account flagged for breaking the lichess terms and conditions.

[JustinHorton]

A posting on a Facebook page of which several forumites are members, by Javier Gil, who's a chess player and coach in Zaragoza, concerning a teammate of his. (Just for declaration of interest purposes - although I haven't played local chess here for a decade, and I don't think I know Javier or any of the players mentioned, I do know and have played at the club. I have made some small alterations for spelling etc.)

LICHESS, You got this one wrong, I promise. Algorithms vs The Human side.

A few days ago, someone I know in my hometown was flagged as a "cheater" on lichess. At present Lichess prefers the euphemistic flagging "This account violated the Lichess Terms of Service".

I'm perfectly aware that there have been many cases of people on line who dispute Lichess decisions, some were honest, some were not, and until recently, I always had my doubts about those cases. I have very little doubts about this one!

Santiago Palomar is a fiftysomething public servant who has a passion in life: his chess. He's been a formidable chess player for over 40 years. Those of us who know him in our hometown know that he'd never do anything to dishoner chess. Never EVER. He'd rather be eaten alive by hungry sharks! Seriously, he would.

The flagging, which in my opinion is completely unfair and unfounded, has caused him great pain and distress because it's so unfair...

He doesn't like computers and until recently he didn't even know how to play online. I teach at a local club and even had to convince him to join lichess!

He appealed liches'ss decision, but this was the answer:

We are continuously improving our cheat detection so that we can aggressively prevent cheating while minimizing the false positives.

After investigating your case, we have determined that our cheat detection algorithms flagged your account with sufficient evidence for the mark to remain.

Lichess has strong detection methods and a very thorough process for reviewing all the evidence and making a decision. The process sometimes involves many moderators and can take a long time. Other than the mark itself, we will not go into details about evidence or the decision making process for individual cases. It is a strict Lichess policy to not discuss the specifics of cheat detection publicly with anyone ever, doing so would make it easier for cheatersto avoid detection and be an invitation to many unproductive debates.

Any attempts to start such a debate will be ignored.

Santiago Palomar's local rating has been higher than 2200 at some point.

His FIDE rating has been as hight as 2130.

His current FIDE rating is around 1955.

You can view it here

His Lichess rating is 1902 (lower than both his FIDE and local ratings!). Check it out here:

So, what are the facts? Out of his recent games, 2 of them had a centipawn loss of less than 20, just about every other one is higher.

This is the first one. (Yes, nice combination at the end giving up the knight, but, frankly, not too difficult. I've seen him do much harder puzzles in our class! )

The following game, which was apparently used by lichess also, was played against Jorge Struch, who happens to be a team mate of his and a very good friend also, and Jorge knows that Santi can still play some wonderful attacking games, so he wasn't at all shocked to lose a quick game against him. This is the game.

So, apparently these were the 2 main games with a centipawn loss of less than 20 (just about every single one of the others has a centipawn loss of +40, most a lot more) which the algorithms used as main proof of his cheating, and one of them played against a great friend of his who would never accuse him of such a thing because he knows how good he is at chess after more than 40 years...and the other, a relatively simple combination in a clearly better position.

And all of that alleged cheating for what - just to gain a few rating points??? hasn't he proven that he doesn't care about rating??? he's played over 700 games on-line for God's sake!! if he wanted to cheat his rating would have gone up ages ago!

I'm sure he has lost many on time because I've seen him play on a phone which he can barely use, and both his FIDE and local ratings are higher, which very few people can say on lichess...

I support lichess, I think it's, by far, the best online chess platform ever, I've been supporting lichess for quite a while now and will continue to do so, but guys, we're humans, the accusation here makes no sense whatsoever.

We are humans, we can and we do nasty things all the time. But Lichess, YOU GOT THIS ONE WRONG. This guy loves chess and would never do anything to harm the integrity of our sport. And as Santi says: "you know, we may not be high rated players, but we can play a brilliant game from time to time.

I can assure you that he can. Check out your database if you don't believe me... look up "Santiago Palomar".

Palomar has left a note on his account:

Estais locos, yo no me ayudo de ordenador para competir. No pienso volver a jugar por aqui mientras no rectifiqueis. Y no voy a cambiarme de cuenta. No tengo nada que esconder. Que os quede muy claro. Ah, y usais mucha informatica pero si no sabeis distinguir las denuncias falsas de las autenticas sois un fiasco de plataforma. Y ahora me borrais vosotros si teneis narices.

My translation:

This is madness: I don't play with computer assistance. I absolutely won't play here again until you put this right. And I'm not going to change my story. I have nothing to hide, you be very sure of that. No matter how much expertise you have, if you can't tell the difference between what's real cheating and what isn't then you're a joke of an organisation. Now delete this to show how big you are.

EDIT: following the posting it seems that sense prevailed and Sr Palomar's account has been restored to him.

[Mick Norris]

EDIT: following the posting it seems that sense prevailed and Sr Palomar's account has been restored to him.

That's good to hear

[NickFaulks]

EDIT: following the posting it seems that sense prevailed and Sr Palomar's account has been restored to him.

That's good to hear

Yes, but not everyone has the assurance to bring such pressure on Lichess. It must be assumed that every one of their convictions is suspect.

[John Swain]

EDIT: following the posting it seems that sense prevailed and Sr Palomar's account has been restored to him.

That's good to hear

Yes, but not everyone has the assurance to bring such pressure on Lichess. It must be assumed that every one of their convictions is suspect.

It would be helpful to know why, after receiving Sr Palomar's appeal and rejecting it in the rather haughty manner Justin Horton quoted, Lichess in the end caved in and restored his account. Is there a further stage beyond an appeal? It seems that some pieces of the jig-saw are missing.

It is good news that platforms like Lichess, in the case of Sr Palomar, and Chess.com, in the case of Justin Horton, do occasionally admit the error of their ways. Their anti-cheating methods are not water-tight.

Organisations which piggy-back on these platforms, taking money from players but washing their hands of problems, requiring their customers to refer an appeal over a ban from the platform to the platform and not to them, should take a good hard look at themselves.

[Roger de Coverly]

It must be assumed that every one of their convictions is suspect.

In the absence of physical evidence such as a witness or an audio-visual recording, the evidence is always circumstantial. Also circumstantial is someone playing much better or even much worse than their supposed standard without a previous history of either variability or past success. It is however an indication. Actually in online play someone playing lots of games with undetected engine assistance is presumably going to disappear into rating levels of GM and beyond.

The "even much worse" point is that if someone had developed an undetectable means over the board of consulting external assistance and used it to claim to be a strong player, they could be found out if they ever had to play in circumstances where their consultation method couldn't be employed.

[Matthew Turner]

That's good to hear

Yes, but not everyone has the assurance to bring such pressure on Lichess. It must be assumed that every one of their convictions is suspect.

It would be helpful to know why, after receiving Sr Palomar's appeal and rejecting it in the rather haughty manner Justin Horton quoted, Lichess in the end caved in and restored his account. Is there a further stage beyond an appeal? It seems that some pieces of the jig-saw are missing.

It is good news that platforms like Lichess, in the case of Sr Palomar, and Chess.com, in the case of Justin Horton, do occasionally admit the error of their ways. Their anti-cheating methods are not water-tight.

Organisations which piggy-back on these platforms, taking money from players but washing their hands of problems, requiring their customers to refer an appeal over a ban from the platform to the platform and not to them, should take a good hard look at themselves.

John,
The case of Sr. Palomar is indeed interesting and it does show that Lichess does on occasion reverse flagging. I am not clear in this case whether
1. Lichess got it wrong
2. Sr. Palmar made a big fuss and reinstating the account was the path of least resistance
3. Sr. Palmar provided evidence that made Lichess change their minds

I note in English chess we are not seeing players making a big fuss (publicly) about being flagged

In the case of the 4NCL, people are learning all the time, but with reference to your last paragraph I think it is worth bearing in mind
1. The first season of the league was free to enter
2. There was a robust appeals process in place at the last congress which was independent of Lichess

[JustinHorton]

The case of Sr. Palomar is indeed interesting and it does show that Lichess does on occasion reverse flagging. I am not clear in this case whether
1. Lichess got it wrong
2. Sr. Palmar made a big fuss and reinstating the account was the path of least resistance
3. Sr. Palmar provided evidence that made Lichess change their minds

Well we know that Sr Palomar did appeal and the appeal was in the first instance rejected, which kind of supports 1 rather than 2.

[Roger de Coverly]

2. There was a robust appeals process in place at the last congress which was independent of Lichess

Is it the case that if lichess flag an account, there's nothing the 4NCL can do about it, apart from removing from display the cross references between lichess handle and real name?

There seems to be reason to believe that lichess take action on a dangerously small number of games. It's possible that a player having played 1000 legitimate games may have sought assistance in one or two of them. What lichess is detecting is exceptional performances and a 1 in a 1000 superb game is well within the realm of plausibility without invoking an explanation of external assistance.

[Matthew Turner]

Roger,
The 4NCL have an ongoing dialogue with Lichess. If a player was flagged for games during one of their events and the player successfully appealed to the 4NCL, then this would be discussed with Lichess. Obviously, the 4NCL cannot force Lichess to do anything but they would certainly be listened to and their evidence taken account of.

[Javier Gil]

It would be helpful to know why, after receiving Sr Palomar's appeal and rejecting it in the rather haughty manner Justin Horton quoted, Lichess in the end caved in and restored his account. Is there a further stage beyond an appeal? It seems that some pieces of the jig-saw are missing.

Yes, good point.

When I wrote that post, I didn't think I'd manage to covince lichess that they'd made an error, I just thought it was the right thing to do.

As to why they reversed their original decision, I think a few things might have contributed:

- The fact that Santiago was Ok with providing his full name and details.
- The fact that he'd already played 700 games on-line, and there was nothing suspicious about his results (if anything, it was surprising that his on-line rating was actually lower than his otb rating).
- He didn't have more than 1 online account. (it's illegal, but Lichess does allow it under certain circumstances).
- The fact that he's been FIDE rated for more than 15 years (and again, his FIDE rating has always been higher than his best online rating), and he has a history of defeating some pretty good players, including GM Forcen (rated over 2550) in January this year, IMs Kovacevic and Glavina, etc. in OTB tournaments.
- Only those 2 games had a centipawn loss lower than 20.
- No meaningful gain would be derived from cheating in those 2 games.
- Other reasons that I ignore.

I'm just glad that lichess was open minded about considering all of these and not just the algorithms...

[John Swain]

It would be helpful to know why, after receiving Sr Palomar's appeal and rejecting it in the rather haughty manner Justin Horton quoted, Lichess in the end caved in and restored his account. Is there a further stage beyond an appeal? It seems that some pieces of the jig-saw are missing.

Yes, good point.

When I wrote that post, I didn't think I'd manage to covince lichess that they'd made an error, I just thought it was the right thing to do.

As to why they reversed their original decision, I think a few things might have contributed:

- The fact that Santiago was Ok with providing his full name and details.
- The fact that he'd already played 700 games on-line, and there was nothing suspicious about his results (if anything, it was surprising that his on-line rating was actually lower than his otb rating).
- He didn't have more than 1 online account. (it's illegal, but Lichess does allow it under certain circumstances).
- The fact that he's been FIDE rated for more than 15 years (and again, his FIDE rating has always been higher than his best online rating), and he has a history of defeating some pretty good players, including GM Forcen (rated over 2550) in January this year, IMs Kovacevic and Glavina, etc. in OTB tournaments.
- Only those 2 games had a centipawn loss lower than 20.
- No meaningful gain would be derived from cheating in those 2 games.
- Other reasons that I ignore.

I'm just glad that lichess was open minded about considering all of these and not just the algorithms...

Many thanks for providing this very useful and helpful extra information!

[Roger de Coverly]

- Only those 2 games had a centipawn loss lower than 20.

Are they really issuing bans on the evidence of just two games? The theory behind the mathematical statistics used to infer outside assistance from quality of play requires a reasonable amount of evidence to eliminate the alternative hypothesis that the engine matching is down to luck, skill or both.

[Javier Gil]

Are they really issuing bans on the evidence of just two games? The theory behind the mathematical statistics used to infer outside assistance from quality of play requires a reasonable amount of evidence to eliminate the alternative hypothesis that the engine matching is down to luck, skill or both.

Since that information is unlikely to be made public, we'll probably never know. But those 2 are the only recent games (not sure if any of the other 700 would be in this category also) with a centipawn loss lower than 20. All the rest are above 40, many of them much higher. All of these were usually 5 minutes + 3 seconds increments.

[John Hodgson]

- Only those 2 games had a centipawn loss lower than 20.

Are they really issuing bans on the evidence of just two games? The theory behind the mathematical statistics used to infer outside assistance from quality of play requires a reasonable amount of evidence to eliminate the alternative hypothesis that the engine matching is down to luck, skill or both.

If true, this is disturbing.

Last week I played a 'perfect' game at 60+15 on lichess - 14 moves of theory, a semi-TN suggested by StockFish as part of my preparation before the game - and then 8 rather obvious moves, 7 of which were SF's first choice and only the last SF's second choice (trapping a queen for +10 when there was something stronger).

When I analysed the game afterwards and found the SF correlation the thought that this could be used as evidence of cheating rather spoilt the pleasure one would normally gain from playing a reasonable game.

Does anyone know whether lichess and chess.com use their own rating systems to determine whether someone has played above their grade, or even if they take rating into consideration at all? Nearly all of my opponents at longer time controls have been graded between 180 and 210 ECF but their online ratings have been 1500-1800, which means there will be many seriously under-rated players when these ratings stop being provisional. Does this increase the risk of being called out for cheating?