4NCL Online

[Roger de Coverly]

I don't speak on behalf of the ECF or the 4NCL, but I note this in the ECF fair play policy

"Before any sanctions are applied a player’s games will be reviewed using Dr Ken Regan’s
anti-cheating software, and the results will be shared with the player on request. "

Isn't that sanctions by the ECF? They don't control and cannot control what action might be taken by chess.com or lichess. That's a major part of the problem.

[Richard Bates]

Richard,
I mean the Regan method, but it looks like Lichess (and chess.com) very strongly follow the Regan methodology.

Does it? Apologies, I perhaps shouldn’t really be engaging without researching a bit further on how exactly the Regan method works but, well, forum privilege.

I had somehow got the idea that Regan relied on “move matching” eg. High correlation with top engine suggestions. Lichess (the public “centipawn version”) seems to compare strengths of moves played with strength of preferred engine suggestions. Not the same. The former method will potentially be triggered innocently in positions with high levels of forcing or near forcing moves. The latter also allows for quite positions with large numbers of possibilities of close equivalent strength. One of both of these assumptions may be wrong but anyway...

Yesterday I played a blitz game on lichess. My opponent who I have played before with similar outcomes decided I was cheating and told me so (I had centipawn correlation if 0.14 and no inaccuracies). Enough to trigger? I have no idea! This was as part of my aforementioned run of 11 consecutive wins - mostly played with reasonable correlation with computer.

[Richard Bates]

This is what the captain were told at their meeting with a Ken a Regan and does indeed seem to be borne out by the evidence I have seen.

Ok. So if we look at the games of any of the banned players, we should be able to see this, am I right?

Yes and no. In some cases you will be able to see it, but not in others. All the games are available on the 4NCL website so you'd have to investigate for yourself.

Probably worth noting that I think there are some possible errors in the pgns on the website

[Neil Graham]

Perhaps the management of the 4NCL will have to accept that bans for no stated reason are going to trigger such reactions.

But also that failing to ban players strongly suspected of cheating will trigger a similar reaction.

If I was the management of the 4NCL - and I am as glad as everyone else that I am not - I would be praying just to get to the end of this season online, and that we could play OTB next year, probably with a late start.

If a player is sent off in football the rest of the players don't walk off the pitch as a reaction.

(We are know this has happened very very infrequently - so no examples need to be quoted please).

[Roger de Coverly]

I had somehow got the idea that Regan relied on “move matching” eg. High correlation with top engine suggestions.

There's also the extra "you aren't good enough to make those moves legitimately" feature which gives higher tolerance to those of higher rating.

What would be intriguing would be a test of over the board tournaments or 4NCL weekends on the premise that all players had ratings of 1000. How many accusations would this throw up?

[Adam Raoof]

Ken Regans software is based on a wide sample of games, not one single game. It is also more generous towards the potential offender that it might be, and is not mechanically applied like lichess or chessdot. Therefore if his software tells you, as an organiser, that these five players are cheating, then they are cheating. What you do with that information is up to you.

I am afraid most complaints and appeals I have seen from players and captains and parents have either been vexatious or largely based on misunderstanding or based on the principle that their child or club mate could not possibly be cheating. I dont trust the lichess and chessdot method as much as I could if it were public, and I knew what their basis for banning players actually was, but I know they dont use Ken's software.

[Roger de Coverly]

If a player is sent off in football the rest of the players don't walk off the pitch as a reaction.

If a referee arbitrarily sent off a player for no apparent reason, the controversy could run for ages. Would a purely amateur team withdraw from a league in which this seemed a regular occurrence?

[Matthew Turner]

I don't speak on behalf of the ECF or the 4NCL, but I note this in the ECF fair play policy

"Before any sanctions are applied a player’s games will be reviewed using Dr Ken Regan’s
anti-cheating software, and the results will be shared with the player on request. "

Isn't that sanctions by the ECF? They don't control and cannot control what action might be taken by chess.com or lichess. That's a major part of the problem.

Roger,
I am not suggesting it is problem free, but accused players can get information, can see what anti-cheating systems say about their play and can use this in their defence if they choose to do so. If we have lots of players banned in the 4NCL, presumably we'll start to see to see a lot releasing this information to expose flaws in the system.

[Richard Bates]

Roger,
You only get a ban if you play unnaturally well, better than anyone played in the whole of the OTB 4NCL. So the idea that a player could be banned because a parent accessed some analysis from the same IP address is total nonsense.

I think that’s just demonstrably false.

Please demonstrate this.

Note to self - don’t use long words for effect.

The over the board 4ncl games have all been run through lichess software? And nobody played at a standard of those banned in the online 4ncl? Let’s just say I don’t believe that - or at least “standard of play” needs to be defined...

[Adam Raoof]

I applaud the loyalty for teams who have raised protests where one of their players has been accused of cheating.

I have more sympathy for players like Keith Arkell and Danny Gormally, who have been cheated against, where it is accepted to be proven, and where there has been no punishment.

I appreciate that mistakes can be made, and that there can be false accusations (this has happened over the board). But generally speaking I would be confident in the ability of the algorithms to identify cheats, both online but also over the board offline. This is what Ken Regans software was being successfully applied to before lockdown, and that's why it is so good.

[JustinHorton]

This is what the captain were told at their meeting with a Ken a Regan and does indeed seem to be borne out by the evidence I have seen.

Ok. So if we look at the games of any of the banned players, we should be able to see this, am I right?

Yes and no. In some cases you will be able to see it, but not in others.

If we can't see it, what kind of evidence is that?

[Adam Raoof]

I think that’s just demonstrably false.

Please demonstrate this.

Note to self - don’t use long words for effect.

The over the board 4ncl games have all been run through lichess software? And nobody played at a standard of those banned in the online 4ncl? Let’s just say I don’t believe that - or at least “standard of play” needs to be defined...

Please forget lichess and chessdot analysis software totally.

All 4NCL online and over the board events have been run through Ken Regans software which involves several world class engines and supercomputer processing power at his university, over many hours, not just five minutes.

When he picks up an outlier it is because they are performing at ridiculous levels of accuracy over a sample. Even his software is set to be generous to a fault, taking account millions of book theory lines updated regularly, discounting forced moves, and taking account of ratings and junior improvement.

There is no comparison between what he does, and what lichess does, not least because we don't know what lichess does.

[Roger de Coverly]

The over the board 4ncl games have all been run through lichess software?

I believe that to be the case. At the very least that gives a reference point on an event presumed free of in game assistance. It may also flag a player or two for arbiters to keep an eye on in future events.

I hope that OTB chess organisations firmly reject the notion that you take sanctions purely on statistics and that a method of external consultation has to be demonstrated. Cases like the phone in the loo or signalling by third parties are relatively easy. There have been some unproven ones as well which may have involved hidden hardware. The reaction of the accused player has eventually been to retire from play.

Playing in the privacy of your home, in the absence of video evidence or physical presence of an arbiter, how can anyone know for certain one way or another that there's no consultation?

[Roger de Coverly]

There is no comparison between what he does, and what lichess does, not least because we don't know what lichess does.

The 4NCL has got itself into a position where it has to accept lichess bans, That's really what the row is about.

[Matthew Turner]

Justin,
Once we get a sophisticated cheat then a human cannot detect it, that is why we need computer systems. What a human can do is provide evidence that a player hasn't in fact cheated, so before a player is banned by Lichess a human signs it off. Before a player is banned in titled Tuesday (chess.com) then I believe three titled players sign it off.
Where we have a 'sophisticated cheat' then we have to review the statistical evidence very carefully. We know that in the vast majority of cases the player will be guilty (we can tell that by the variance between online and OTB events) but that shouldn't stop us approaching the evidence with skepticism to give the accused the best possible defence.