Farewell

This is outrageous. I've been following Pepellou for years and I would've never suspected him of cheating. He was always very aware that playing moves that the chat suggested during a rated game is not allowed and would continually warn his viewers about it. Until we're shown any evidence I have to come to the conclusion that this is a terrible mistake done by lichess.

Whatever 'faith' I could've had in the cheat detection and the appeals process on this site is disappearing.

As to the game linked above, there's nothing special about any of Pepellou's moves and it clearly doesn't prove that he cheated.

@KenulL_76 this guy www.twitch.tv/pepellou

He created some high quality content for pychess users as well. For example:
webcache.googleusercontent.com/search?q=cache:d679WfUUogYJ:https://lichess.org/%40/pepellou/blog/practical-chess/MRSJqWxD
webcache.googleusercontent.com/search?q=cache:npFyrJww85sJ:https://lichess.org/%40/pepellou/blog/you-only-need-one-win-per-game/By8vmmHn
webcache.googleusercontent.com/search?q=cache:78tXcxa3vuMJ:https://lichess.org/%40/pepellou/blog/an-instructive-loss/tHnH7SIm

@gbtami said in #42:
> @KenulL_76 this guy www.twitch.tv/pepellou
>
> He created some high quality content for pychess users as well. For example:
> webcache.googleusercontent.com/search?q=cache:d679WfUUogYJ:lichess.org/%40/pepellou/blog/practical-chess/MRSJqWxD
> webcache.googleusercontent.com/search?q=cache:npFyrJww85sJ:lichess.org/%40/pepellou/blog/you-only-need-one-win-per-game/By8vmmHn
> webcache.googleusercontent.com/search?q=cache:78tXcxa3vuMJ:lichess.org/%40/pepellou/blog/an-instructive-loss/tHnH7SIm
thanks

I certainly hope that the game linked above was the cause of his ban. I could have played that game myself. Not one suspicious move.

I can't say anything about lichess decision since it's all private but all I can say is I really hope they did a thorough examination and found near irrefutable proof of cheating or of whatever it is they're accusing him of doing.

I'm not a very good judge of character but I hopped into a few of his streams, seen some of his messages and he sounded like someone who really believed in this website and who has tried his best to share his passion.

We never really know anyone, maybe his ego took over but I'm really tempted to side with the guy who streamed for lichess for so long, who doesn't really hide his identity, who contributed to pychess (thx #42 and probably for other projects I'm not aware of) and who wrote messages like this one: lichess.org/forum/lichess-feedback/cheater-banned-during-the-game-no-points#3

But I reiterate I know nothing. I just hope that any mod reading this is certain of what they're doing.

Pepellou's engine correlation scores in the above mentioned tournament (14 games, 7+5) according to pgn-spy - undecided positions (eval between -2 and +2) :

pgn-spy acl : 13,91 ; blunder rate (100cp+) : 0,92% (2/217)

Magnus Carlsen at the PCL 2018 (6 games, 10+2 time control) :

pgn-spy acl : 12,01 ; blunder rate (100+cp) : 2,37% (4/169)

The acid test would be to compare these scores to Pepellou's OTB games, which I won't do. I'd just say these are remarkable correlation scores for such a quick time control, although I don't know what was the strength of the opposition.

There was a tool developed by Michael Gleason at chess.com called PGNSpy and it was used to run against statistics of players over a number of games.

It was open-source as well as being available in binary form, so lichess may be using it. I do not know what the stats are for the games played in this tournament.

What I recall is that the odds were substantial that Maxim Dlugy was cheating in the Titled Tuesday event on 4 April 2017. Another name that came up as slightly suspicious was HansCoolNiemann, but not nearly as conclusive as that of Dlugy.

The one who beat Hikaru was known to him in person - he had lost a game a long time before when he hung a piece in a drawn endgame. So then Hikaru hung a bishop himself in their Titled Tuesday game and might still have held but played badly after that and lost.

Over on chess.com, Haubi used to play a lot in "daily" chess including a lot of chess960, and has lost to a lot of players. In daily, many of them would have been cheating, of course, as it's rife over there where Haubi would have been on the top board a lot. Meanwhile pepellou is on the staff over there, so I doubt they'd let a cheat on their staff.

Most of the time, we trust the lichess staff and thank them for keeping their site clean, but occasionally, especially if someone so popular is banned, we do want answers.

@hicetnunc said in #46:
> Pepellou's engine correlation scores in the above mentioned tournament (14 games, 7+5) according to pgn-spy - undecided positions (eval between -2 and +2) :
>
> pgn-spy acl : 13,91 ; blunder rate (100cp+) : 0,92% (2/217)
>
> Magnus Carlsen at the PCL 2018 (6 games, 10+2 time control) :
>
> pgn-spy acl : 12,01 ; blunder rate (100+cp) : 2,37% (4/169)
>
> The acid test would be to compare these scores to Pepellou's OTB games, which I won't do. I'd just say these are remarkable correlation scores for such a quick time control, although I don't know what was the strength of the opposition.

The problem with correlations is that they are not sufficient to determine causation. There's still a chance that Pepellou could have played the top engine moves by chance. Also, cheaters will sometimes use the engine in a tough position for a couple of moves and then play the rest of the game out by themselves. I wonder if Lichess is able to detect if you have another analysis tab open in the browser or if it can detect if you have a cheating browser plugin installed. I think that would be better evidence than simply correlation.

The appeal process (lichess.org/page/appeal) states that once your appeal has been denied you can not make any more accounts on Lichess. I would hope that the moderators do a careful analysis before they permanently ban people off the site.

Have a question, where to write in support?
I may not understand here, who knows, please write in private messages on liches

In the event in which Magnus Carlsen achieved those statistics, presumably he was playing against grandmasters, who would have given him a tougher game. And I mean top grandmasters, not Haubi, whom Magnus would probably thrash every game.

You play worse (i.e. blunder more) when you're put under pressure.