Cheaters are Out of Control Today. Is Lichess anti cheating team on vacation today?

@Cedur216 said in #60:
> @delorenflie if you refer to trolls and paranoids, then why did you thumb #1 up and thumb me down?

Because I looked at their games with my own cheat detection tool. And I only thumbed you down because I was pretty sure you didn't. These thumbs up/down does not convey the meaning that words can. Sorry you felt hurt about it.

@Brian-E said in #57:
> It would be of doubtful value because for non-specialists it's impossible to be certain whether someone is cheating or not.
>
> The first category you mention, which you must indeed not post, would require everyone to judge whether it is "obvious cheating". That is something non-specialists simply cannot do.
>
> The second category would tell us nothing whether we are able to judge the play or not. Someone who did not cheat in the game you post could still have cheated in other games.

Nonsense since I'm an expert, so to speak.

Actually they do most of the time ban my reported cheaters. I reported many cheaters. Not only in my own games but also cheaters of the tournament that have not played against me.

I always look at every game played recently by the account and only report cheaters if they played many games in a row nearly perfectly.
Something I probably didn't do once in over 10.000 non-bullet online games...

> Actually they do most of the time ban my reported cheaters

that means it's alright, isn't it?

@delorenflie said in #59:

> lichess.org/forum/general-chess-discussion/analysis-of-lichess-cheating-detection-with-machine-learning-ml-a-mis-use-of-ml--doesnt-work

I would like to pose the following question:

The linked post claims that they use the tosViolation flag as their Yi labels to predict. They go further on to claim that they managed to reverse engineer a 99+% overlap with that flag.

It is probably well known that the tosViolation-flag on lichess is not only set for people who cheat using an engine, but for a fairly large amount of non-engine related violations as well. For example, sandbaggers.

So my question is: How did they achieve this 99% overlap?

I agree that there are far too many cheaters (aka pathetic losers) on Lichess and elsewhere.

Wouldn't it be a good idea to prohibit the creation of such "public" forums and to set up a direct chat of the "emotional support or psychological support" type ?

Lichess could be a little more vocal about anti-cheat. We know it can't be fixed and we don't demand it, but it's always nice to know, that at least something is being done about it. It simply needs to be repeated over and over ;-)

@CSKA_Moscou said in #66:
> Wouldn't it be a good idea to prohibit the creation of such "public" forums and to set up a direct chat of the "emotional support or psychological support" type ?

Yes, topics like this one can be damaging to general morale and misinformation can be spread.

The support would need to be provided by people with good knowledge of the subject, ideally those involved in combating cheating on the site. General information can be given without secrets being revealed. But are there people with time to do that?

Chessdotcom's "Cheating Forum" works quite well. Unfortunately this site is not geared towards groups in quite the way that site is. Something creative and tailored for this site would be needed.

@Brian-E said in #68:
> Yes, topics like this one can be damaging to general morale and misinformation can be spread.
>
> The support would need to be provided by people with good knowledge of the subject, ideally those involved in combating cheating on the site. General information can be given without secrets being revealed. But are there people with time to do that?
>
> Chessdotcom's "Cheating Forum" works quite well. Unfortunately this site is not geared towards groups in quite the way that site is. Something creative and tailored for this site would be needed.

I will be ready to do it, I appreciate contact and helping people. and I think I'm not alone

@anonmod said in #64:
> I would like to pose the following question:
>
> The linked post claims that they use the tosViolation flag as their Yi labels to predict. They go further on to claim that they managed to reverse engineer a 99+% overlap with that flag.
>
> It is probably well known that the tosViolation-flag on lichess is not only set for people who cheat using an engine, but for a fairly large amount of non-engine related violations as well. For example, sandbaggers.
>
> So my question is: How did they achieve this 99% overlap?

That is the issue. This person managed to reproduce most of the tosViolation outcomes, if we assume that the 1% missing is because tosViolation was flagged for something else than cheating, then this person basically managed to reverse engineer the exact parameters that lichess uses. The conclusion, as a fellow ML expert, is not good. But the responses to the post were generally banal, or dismissed the issue, instead of accepting that there is an issue and providing a venue for this researcher, who was doing this work for free, to help improve the system. That, honestly, was quite upsetting to me.