lichess.org
Donate

What options has Lichess considered to quell the bot players?

Personally I feel as though Lichess has exceptional bot detection. I just watched the tournament where a bot users, FellerDeep, was caught within an hour. Definitely applaud the efforts so far. Cheaters suck, but they are of course a reality of online play today.

But I am curious, what efforts Lichess has considered to further deter potential cheaters from playing tournaments? The BitChess tournament was a paid tournament. A very awesome contender whose name I won't repeat was very clearly frustrated playing against the cheater FellerDeep and quit the tournament as a result.

Whatever your feelings are regarding the well-known GM Wesley So incident, it is undoubtedly a shame that cheaters are so easily able to mix themselves in with the more legitimate parts of our community.

I personally am against anything such as credit card verification or anything that is personally identifiable. Some players enjoy their anonymity. But what would the problem be with a "minimum time on an account/minimum number of games played" before one could compete in a tournament? It seems like a fairly unobtrusive way of maintaining the integrity of the more important function of this site i.e. tournaments, and especially paid ones.

FellerDerp only had an account for less than a few hours. A delay on new accounts so the bot detecting features could have done their job would have easily avoided this.
One thing I personally wish as a tournament organizer:

1) The option to restrict the entry for any player with less then 14 days registration date or at least 100 games played.

2) Since it's our tournament I would appreciate the option to kick players. This option is only necessary if option 1) is not possible.

For now we'll restrict our tournaments on Crazyhouse only and stop with with regular chess. Crazyhouse is *almost* engine free and even the strongest engine couldn't beat someone like JannLee.
A lot fewer players play Crazyhouse compared to regular chess. I cannot play hyperbullet because there is no enough time for me to evaluate moves. These games are all based on premoves and crazy moves designed not letting opponent premove but to do unexpected captures with the king and waste time in the process. I don't know how can someone use computer assistance in these conditions. Some GMs play online only bullet and hyperbullet in order to avoid cheaters. Wesley So played 5+0 games. The conditions are very different from 0.5+0 games.
@BitChess #3

+1 For point #1. But 100 games in bullet/hyper is too small. It can be done in less than 3 hours

I would say something like a minimum of 500 bullet __AND__ a minimum of 48 hours __OR__ 150 blitz __AND__ a minimum of 48 hours, etc.

+1 for the restriction on zh
As someone who has actually looked at the repo, I have a lot less faith than many in the cheat detection. I still like to play here though and the cheating, I guess if I don't notice it then it doesn't bother me.

It's a lot harder not to notice when you're ranked top 10 in the world and some random internet dude is kicking your butt.
When you play random anonymous people, you may encounter someone who is actually very good and easily kicks butt to the #10 in the world. Magnus Carslen is a clear #1. He has a very impressive head to head record against Nakamura! Perhaps Magnus Carlsen learned from Wesley So's mistake and playing here under a pseudonym and never reveals his identity.
I do not have any faith at all in lichess engine detection system. I have seen they banned one russian streamer right on the stream, even though moderator was listening to the stream and wasn't able to understand russian language and reasoning of the steamer. The banned guy was just a strong player, but still got falsely banned.

Also recently I have run into an engine by myself. The opponent had statistics like 12-3. 3 loses were just voluntary leaves on 2-3 move of each lost game, and all won games were simply with 0 mistakes. I have reported all his won games, but he didn't get banned.

And obviously Lance5500 case. You just can't prove he is not cheating, until he discloses his title and real name. But he is still playing and even participates in real money tournaments.
Any strong FAIR player should have solidarity against engines, and this solidarity should be expressed by means of disclosing real names and titles.
Lichess has good options to attract cheaters, like titles and cup rewards and recently money.
A lot of public chat rooms have "verification levels" to protect rooms from unsolicited spam. On Discord, they provide the following restriction options:

https://i.imgur.com/6fT78tD.png

Simple criteria with an aim in mind seems to work well. For instance, this seems to be what they target:

Low - Accounts by people with online presence, thus blocking out automated bots.
Medium - Accounts by users who have signed up prior and is not newly made, thus blocking out offenders who attempt to return, or fishy people in general.
(╯°□°)╯︵ ┻━┻ - You have to try hard just to participate. Anyone who wants to spam must be determined enough to bother.

If it was me, I would introduce the following restriction for tournaments:

Official tournaments (Scheduled by Lichess): To participate, you must have registered with a verified email address.
Important tournaments (Weekly, Monthly): To participate, you must have a non-provisional rating and have signed up on the site for some time (hours / days).
Events (Marathon, Promoted tournaments): To participate, you must have a non-provisional rating and have signed up for the tournament for some time (hours / days).

I think it's important to respect the rights of users who want to remain anonymous, as long as they commit to being a healthy member of the community. Contrary to Wesley So's views, I am of the opinion that we can achieve keeping tournaments free and open while still cutting down the rate of suspicious players.

However, as far as I'm aware, we allocate staff to watch event tournaments and review the top participants as they are going, and it works fairly well. If anything, we would need more manpower and tools. For this reason I support the point delivered in #3 where the organizer should be able to supervise and kick out users who they deem should be disqualified already. Not sure if it warrants the effort, but we'll see.

This topic has been archived and can no longer be replied to.