4NCL Online

[J T Melsom]

Roger L,

I can't have total confidence in much at all these days, but anti-cheating mechanisms which are not able to be explained to the accused are a bit too much' black-box' for me. And I suspect strong players are more open to false accusation than weak players who simply copy too many engine moves. If I were a stronger player I'd probably not risk it.

That said I understood 4NCL games were checked by the Regan system? I'm not at all clear what has happened in that regard in this case. Does a clean bill of health from Regan mean the platform's own methods are regarded as equivalent to a false positive? Isn't that where the discretion of event organisers might be exercised?

[Roger Lancaster]

may I ask whether there is anyone on this forum who has total confidence in the anti-cheating mechanisms of Chess.com and Lichess?

Perhaps the two suppliers have already done this but I would like to see them check there engine matching procedure against a regression test bed of games where it was impossible to cheat with engine support eg. games played before 1990. For amusement I checked some of my own historic OTB games about 5 or 6 years ago against deep fritz 12. I got quite a good match for 1970s games, less so for 1960s and 1980s was somewhere in the middle.

Ken Regan remarks, in a video I've mentioned elsewhere, that Capablanca's games circa 1927 show well above-average correlation [compared with other grandmasters up to the Fischer-Spassky era] with modern engines

[NickFaulks]

when I think my opponent is using an engine the lichess software agrees with with me.

How do you know that?

[E Michael White]

Ken Regan remarks, in a video I've mentioned elsewhere, that Capablanca's games circa 1927 show well above-average correlation [compared with other grandmasters up to the Fischer-Spassky era] with modern engines

Yes, think I've heard that or read your earlier comment. That half proves my point as Capablanca couldn't possible have cheated in that way but we need to know how the methodology deals with average Joe in addition to Capa. The big question is how often does Joe match the engine, when he couldn't possible have cheated, and generate false positives.

[Paul Cooksey]

, may I ask whether there is anyone on this forum who has total confidence in the anti-cheating mechanisms of Chess.com and Lichess?

total, no. I think cheat detection is probably better than people give it credit for here, it might well be 99% accurate for blitz with a decent sample size. But 1% is a problem for serious competitions, in itself, and because it gives actual cheats an easy way out.

[Joseph Conlon]

when I think my opponent is using an engine the lichess software agrees with with me.

How do you know that?

As so far 100% of the times (approx 7 or 8 times) I have reported someone for suspected cheating lichess has then banned them for engine use.

[Kevin Thurlow]

"Hi Kevin, whether this was a typo or not it's still an amusing pun ! Perhaps it should be run through K W Regans system to establish whether it is a typo or not."

It was supposed to be an amusing pun, for satirical purposes only, not a serious suggestion that the companies are in anyway corrupt.

Thinking back to the "guilty" party playing better than his rating, that happens all the time. Many years ago, I beat a 225 and then lost to a 125 a couple of days later (both proper games taken seriously, the lower-graded opponent being somewhat inconsistent). A rating or grading is only a rough guide to performance, nothing more. If anyone takes his last 100 games and notes the grading scored for each game, then works out the mean and standard deviation, they will find that ECF gradings are probably "right" within +/- 20 points or so.

I played a Lichess Chess960 event last night, (thanks, Adam!, and I will pay...), first Chess960 for 2 years, first blitz for months, and got some sort of rating (about 1800) over 12ish games. I was not playing well - if I wake up a bit next time, and play properly, will they assume I'm now cheating?

[Paul Cooksey]

if I wake up a bit next time, and play properly, will they assume I'm now cheating?

only if you have high engine correlation, low time use in complex positions, etc It isn't about performance vs rating.

[Joseph Conlon]

Also worth commenting here that lichess is not a company and is not a commercial enterprise - in that it gets money it is via donations from its happy users.

It is open source and is written and supported by volunteers, all the moderation etc is being done by volunteers. So IMO expectations that lichess should be expected to go into great detail about why they ban any one user is unreasonable; its a free platform, all the content is free and I think it works very well.

[David Robertson]

I've now discovered the name of the accused player, and the game that led to the accusation. I've now played through that game, albeit without an engine running, and I'm happy to say from considerable experience of online chess and cheating that I do NOT think the game shows evidence of cheating, not remotely. (NB running an engine was hardly necessary)

No, I will NOT be revealing the name, nor the game. Too bad if this deepens the water. But if any of you were to play through the game, you'd WTF too

[Ian Thompson]

I've now discovered the name of the accused player, and the game that led to the accusation. I've now played through that game, albeit without an engine running, and I'm happy to say from considerable experience of online chess and cheating that I do NOT think the game shows evidence of cheating, not remotely. (NB running an engine was hardly necessary)

No, I will NOT be revealing the name, nor the game. Too bad if this deepens the water. But if any of you were to play through the game, you'd WTF too

I agree -

Now the player involved has been named on this forum, albeit only briefly, their games can be examined. It would be interesting to know what aroused suspicion in these games because it's not obvious to me.

[Jon Tait]

What if someone is only using an engine for part of the game? In other words, mostly making their own moves but referring to an engine at critical moments or for an ongoing blunder check. Even just the latter would increase a human player's strength considerably. That was in fact the main reason I declined to take part in the event. I don't see how such usage can be realistically policed.

The other reason, I confess, was the temptation to access database files of games and analysis. Personally I'd find it very hard to resist doing that if the game entered one of my pet lines and I couldn't remember all the details.

[Adam Raoof]

What if someone is only using an engine for part of the game? In other words, mostly making their own moves but referring to an engine at critical moments or for an ongoing blunder check. Even just the latter would increase a human player's strength considerably. That was in fact the main reason I declined to take part in the event. I don't see how such usage can be realistically policed.

The other reason, I confess, was the temptation to access database files of games and analysis. Personally I'd find it very hard to resist doing that if the game entered one of my pet lines and I couldn't remember all the details.

That is very honest, Jon. However, what makes this different from any normal league or tournament game? I mean, that possibility is present in every game we play, whether online or not. I can think of several players who have been removed from tournaments and leagues for using assistance, but it doesn't mean that I don't play chess any more.

Are we expecting too much? Perhaps all games should be played with the assistance of computers!

"a strong computer will not tell you anything a weak computer does not tell you. A strong computer will simply give you the bad news faster" - Anand

[Roger de Coverly]

However, what makes this different from any normal league or tournament game?

In a normal league or tournament game, players are under constant scrutiny, if only from their opponent. You can hardly sit at the board, tapping away at a laptop, tablet or phone at critical moments in the game. Leaving the board when it's your move isn't really allowed.

If there were webcams monitoring the player and the room in which they were playing that would go some way towards emulating the over the board environment. If players are allowed breaks away from the board, the Rausis method is still possible.

[Roger de Coverly]

It is open source and is written and supported by volunteers, all the moderation etc is being done by volunteers. So IMO expectations that lichess should be expected to go into great detail about why they ban any one user is unreasonable

Evidently the mechanisms used to decide who to ban are not open source, given that no-one knows how they work.

I believe I've seen a comment that if a flag is set from "rated" to "casual" the witchfinding mechanism is switched off. The 4NCL issued instructions that it be switched on.