4NCL Online

[Richard Bates]

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.

That sounds suspicious - please say when you get a failed denunciation or when you find someone you suspect gets banned before you report them.

Why? Some cheating is really obvious in a way that everything correlates. To illustrate one example: there was a user with a rapid rating around 2400 on lichess, which I think would correspond to roughly 210 - 230 ECF, who was at the same time asking on the forums what was the best response to 1. d4. From context it was clear he had no idea of standard openings, but in games he was playing a variety of response to 1. d4 at IM+ level.

Certainly there are times when a game has seemed a bit suspicious, but not enough to report, and later on I have noticed that the user was banned.

Hopefully though it is legitimate to ask around about the best response to 1.b4? Can’t vouch for responding to it at IM level though

[Roger de Coverly]

However if you indeed used engine assistance, even just once, then your account is unfortunately lost.

But what do they mean by using engine assistance? Particularly when they know their future opponent and their pet lines, players will attempt to find a winning or drawing bust. Engine use before the game is legitimate, or do lichess and chess.com disagree? Unless the platforms have loaded analysis from all known books and articles, players reproducing published theory could also be vulnerable.

[Richard Bates]

As a bit of a diversion, I wonder who is the strongest player in history who would struggle to provide an answer to the question “best response to 1.d4?” (in a negative context, not as a way of choosing between several good option). From an English perspective - Hodgson? Or perhaps aim higher - Speelman? Extensive use of the move 1...d6 throughout a career perhaps a bit of a giveaway.

And in response to 1.e4. Go to the top?

[Joseph Conlon]

I take 'engine assistance even just once' as meaning an open analysis board tab with stockfish running, analysing the very position you are playing in a live game on another tab. Certainly if both tabs were lichess tabs they would have this information.

I don't know how they deal with the question of preparation of specific lines for a specific opponent; for arena play its not an issue, but for specific challenge matches it would be relevant. But I just don't know what, if anything, holds here.

[Roger de Coverly]

I take 'engine assistance even just once' as meaning an open analysis board tab with stockfish running, analysing the very position you are playing in a live game on another tab. Certainly if both tabs were lichess tabs they would have this information.

I would have thought anyone intending to cheat seriously by using external assistance whilst playing would do so on a separate device, probably not connected to the internet.

If their intent is to default anyone task switching whilst playing, they should say so. The inference of this debate is that they are using a witchfinder program, one that checks games for match ups to engine suggestions. That was the separate check that the 4NCL proposed to do.

[Adam Raoof]

However if you indeed used engine assistance, even just once, then your account is unfortunately lost.

But what do they mean by using engine assistance? Particularly when they know their future opponent and their pet lines, players will attempt to find a winning or drawing bust. Engine use before the game is legitimate, or do lichess and chess.com disagree? Unless the platforms have loaded analysis from all known books and articles, players reproducing published theory could also be vulnerable.

I'm a reasonable player. I try to prepare for games, and unusually one season I knew who I would play and what colour I would have. I knew my opponent was stronger than me, so I prepared using games provided by my team mates. I found a line he often repeated, and a pawn sac I knew he would accept that led to an energetic but drawn kingside attack. We played right into the line and drew.

In another season, I learned a new opening, and I had been studying some games that week. Using a computer I found a sideline with a mating attack that I would probably never have found on my own. In the next league match my opponent went down the line all the way to checkmate on the board.

In neither case did I do any original thinking at the board, and most if not all the moves were generated by a computer, guided by me. All I had to do was remember the moves.

On several occasions I have played a long line of theory in the French, and won easily against strong opposition.

In a league match my teammate complained that his opponent got up and left the room once or twice in critical and complicated positions, when you would expect him to be at the board concentrating. He lost the game.

I really don't think these sites have the ability or willingness to differentiate, and they dont give people the benefit of the doubt.

All you can do as a player is play, and treat online chess on these sites as practice. All you can do as an organiser is accept that if you use these mega platforms then major decisions are totally out of your hands and make that clear to your players. Because they are growing so fast they dont have the ability to respond to the changes in demand. They are not that interested in providing tools to organisers, and even lichess manages to implement good features one day and remove others the next without their volunteers realising it. They cant cope with the volume of queries when they are dealing with an exponential increase in members that crashes their servers in the middle of major events.

If you want a different experience, use a different platform.

[DavidWalker]

...
Back in 2014, the ICC had software which could detect whether users under suspicion were switching windows on their PC (in order to check the engine’s evaluation). I don’t know whether lichess has or uses this capability;
...

Lichess does count the number of times that a player's game window loses focus, and this is used for cheat detection - see the lichess privacy notice at https://lichess.org/privacy:

“Blurs” are how many times your game’s window lose focus: i.e.: leaving the game's window once counts as one blur. “Blurs” are only visible to moderators and is one of many factors used in cheat evaluation.

Details of the anti-cheat measures are hard to come by, but lichess founder Thibault Duplessis did outline some of these measures in a 2017 presentation that is available on YouTube: https://www.youtube.com/watch?v=iHP5AdRlRNY. He mentions:

• Checking moves times

• Counting window switches

• Computer analysis comparison

• A neural network

• Detection of analysis in other browser tabs

• Identification of custom code used by known bots

Note that I have no "inside" knowledge about the circumstances surrounding the banning of one of our players by lichess, so what follows is pure speculation on my part. However, imagine a game played at a slow time limit from a rarely used account which is at a level within the top percentile of all lichess players. If the player of this game is spectating on his team mates' games then he may switch browser tabs many times, and he may also be delayed replying with obvious recaptures, forced moves etc. Such a game or games is obviously a huge red flag to any anti-cheat algorithm.

My take-away from this is that to minimise the risk of falling foul of anti-cheat measures, it is safest to play lichess games in a browser with a single open tab and to never switch away from that tab while the game is in progress. I only wish I had shared this advice with my team mates before any 4NCL games were played.

David Walker, (former) Northumbria Vikings team member.

[Kevin Thurlow]

"Note that I have no "inside" knowledge about the circumstances surrounding the banning of one of our players by lichess, so what follows is pure speculation on my part. However, imagine a game played at a slow time limit from a rarely used account which is at a level within the top percentile of all lichess players. If the player of this game is spectating on his team mates' games then he may switch browser tabs many times, and he may also be delayed replying with obvious recaptures, forced moves etc. Such a game or games is obviously a huge red flag to any anti-cheat algorithm."

Or he might make a cup of coffee, or go to the bathroom, or look at emails, whilst waiting for a reply...

[MartinCarpenter]

This could well be it, yes. Obviously most of their chess, and so examples, are based on blitz. So they've very likely calibrated an algorithm or trained that neural network based on quickplay stuff.

Then apply the algorithm to move timings, player behaviour etc in a 45 + increment game and.....

Stupid but a very believable sort of stupid.

[Roger Lancaster]

Some people including myself would differentiate, others including RDC evidently wouldn't. Each, in my view, is a tenable position. But, to avoid conflating what some would regard as different issues, my question addresses only the first.

It's the 4NCL who don't differentiate. They've surrendered control of the issue to lichess.

The 4NCL's problem, in my view, is analogous to that of an honest businessman who suddenly discovers that his business partner is, or probably is, crooked. Possibly he should have undertaken greater due diligence but that's in the past and he's now faced with a situation, not of his choosing, where the best exit route is far from clear.

[Adam Raoof]

This explains a lot. I blur all the time, all through most games!

[Ian Thompson]

Some people including myself would differentiate, others including RDC evidently wouldn't. Each, in my view, is a tenable position. But, to avoid conflating what some would regard as different issues, my question addresses only the first.

It's the 4NCL who don't differentiate. They've surrendered control of the issue to lichess.

Does the 4NCL have a choice? I would have thought not. lichess are unlikely to care whether or not the 4NCL is hosted on their platform, so they're not going to change the way they operate to suit the 4NCL. The only option the 4NCL has is to not use lichess, but where else would they go that wouldn't be similar?

[Roger de Coverly]

Does the 4NCL have a choice? I would have thought not.

I believe it may do. I have the impression that if games are played as "casual" rather than "rated", that examination of the games and presumably conduct during the games is switched off.

At the very least, if lichess will ban accounts for excessive use of task or tab switching during play, participants should be warned against the practice.

[Roger Lancaster]

Does the 4NCL have a choice? I would have thought not.

At the very least, if lichess will ban accounts for excessive use of task or tab switching during play, participants should be warned against the practice.

On that point, I'd agree with RdC. On Tuesday evening I was team captain for three 4NCL teams as well as - unusually, because I don't much like playing on Lichess for all the reasons given in this thread and others - playing. I had to interrupt my game several times to guide our players whose opponents hadn't shown up and to reply to an opposing team captain where we were at fault. Needless to say, this resulted in delays making obvious recaptures and so on but there's very little I, as a team captain, could do about it - and that's the sort of evidence that might, apparently, lead to my being accused of cheating.

[John Swain]

It would be helpful both for online chess platforms like lichess and chess.com but also for organisations seeking to use them like 4NCL and the ECF to publish guidance on how to avoid running into problems, such as unfair accusations of cheating.

It's now clear that switching windows during an online game arouses suspicion, so, for example, you shouldn't check your email during a game. The same is true for checking on the progress of your team-mates. In over-the-board team chess, it's normal to wish to know how the team is progressing and whether a draw will suffice to deliver a team victory or whether you need to play a very risky line if the team is in trouble, possibly "taking one for the team" as a consequence. This is obviously less true for team competitions based on game points rather than match points.

It's possible to view multi-boards (e.g. viewing the four boards of the Olympiad team); this would be the ideal for other online team events.