cheating detection

@Munich said in #29:

Others might share their experience.
OK, this is my experience: so far I played three games where I know my opponent's account was later marked for ToS violation. I lost all three but in none of them I suspected my opponent might have been cheating. Even now, knowing those users were (probably) marked for cheating, if you ask me if I think they were cheating in those games, my honest answer would be: "I have no idea." And if you ask me why I lost those three games, my answer would be: "Because I made mistakes I shouldn't have made." (Meaning not some mistakes beyond my comprehension but simple mistakes I should have realized at the moment that they were mistakes.)

To be honest, I don't understand how so many players - including many who are much weaker than me - can be so sure they can reliably recognize a cheater from a single game. Neither can I understand how can they be so confident about this ability that when they report someone and the user is not marked soon, their primary hypothesis is that something went wrong with the process, not that they might have been wrong.

On multiple occasions I have seen an opinion that the omnipresent paranoia about cheating harms online chess much more than actual cheating. And I'm afraid I have to agree.

@Rider_of_Theli said in #30:

Have you considered that some of your reports are probably wrong ?
yes! read the thread!
I do not expect all my reported games to be positives. But when I started this threat NONE where positive, and from past experience, every 2nd or 3rd report I did was positive. So I wondered.
But shortly after I started this thread, I received a "someone you reported was banned" message, could be coincidence.

@mkubecek thinks in #28:

Highly unlikely.
[that this thread triggered an admin to look at it]

@mkubecek said in #31:

an be so sure they can reliably recognize a cheater from a single game.

May I send you a game from todays banned cheater that I reported?
then tell me what you think

@Bazigar_64squqres said in #12:

You also could've used better language.
Its a quote "Only a fool worries about what he can't control." -Kvothe, Son of Arliden
(google accredits multiple people for it)

I'm surprised how even a passing mention of cheating is met with hostility nowadays. Munich is merely pointing out that the percentage of reports he makes that result in a ban are much lower then previously (as recently as a couple months). He wants to see if others also noticed this, which if true, would mean lichess either relaxed their ban system or cheaters have evolved better ways of getting around it. Either way, good cheat detection is something that should be welcomed by everyone, and is in fact the best way to combat cheating paranoia in an era where cheaters are far more sophisticated.
As far as I'm aware @Munich you're not allowed to cheat in casual games, but I have no idea if they run cheat detection, or do it as rigorously.
Also important to remind some people in the forum that cheaters can:

1. Not have a high rating
2. Not play the engine move on every turn
3. Use a bot to move for them very quickly when then need to, and take their time on other moves
4. Use a niche or weaker engine to lower their accuracy.
5. Throw games to manipulate rating or stay under the radar.

As far as relaxing the ban system, if that is indeed what happened, we really don't want to become like chess.com where cheating runs rampant and players take weeks to get banned. Even when they play the engine move every move of every rated game for 20 games in a row.

@Destroyer942 said in #35:

Munich is merely pointing out that the percentage of reports he makes that result in a ban are much lower then previously (as recently as a couple months).
I would put this down to the clustering illusion. https://en.wikipedia.org/wiki/Clustering_illusion

Someone seeing that many of their reports are resulting in an account being closed for cheating, then seeing that many of their reports do not result in that, will naturally result in them thinking: "something has changed". But that isn't so. The new run of reports not resulting in an account closure is merely the natural behaviour of random processes which the human mind has difficulty accepting as natural.

@mkubecek said in #31:

OK, this is my experience: so far I played three games where I know my opponent's account was later marked for ToS violation. I lost all three but in none of them I suspected my opponent might have been cheating. Even now, knowing those users were (probably) marked for cheating, if you ask me if I think they were cheating in those games, my honest answer would be: "I have no idea."

So, u cant spot cheaters while they're in ur face and one the other hand u call others paranoid....
Now think about the unreported cases, how high is the number in reality?
Ur conclussion is embarrising.

@Destroyer942 said in #35:

I'm surprised how even a passing mention of cheating is met with hostility nowadays

You can thank the aggressive and hypocritical paranoiacs for that. (I'm not saying OP is one of them)

Cheat detection is good, regardless of a loud minority of sore losers. No matter how good your cheat detection is, you won't get rid of these (especially when you have a bullshit metric like ACCURACY)

Here's the problem in this thread:

But meanwhile none of my reports result in a ban

This is most likely just biased perception. I'm not surprised if a mod comes in and says "you've hardly reported anyone in the recent months" (I've seen enough of these dudes)

even though I am very sure it was a cheat (very obvious, PM me for an example).

@Munich I don't think you're in any place to say "very obvious". This is a plain subjective and self-serving judgement. It holds no relevance for me. As a regular user not only you lack the tools to analyze your opponent's behavior (Irwin / Kaladin / browser telemetry...) but you also lack the knowledge to draw correct conclusions from said tools.

@EvilPyrokar said in #37:

So, u cant spot cheaters while they're in ur face and one the other hand u call others paranoid....
Or I'm just honest enough to admit that it's not as trivial as some people think. And also that the actual reason I lost those games were my own mistakes so that it didn't really matter if my opponents cheated or not.

@Munich said in #14:

you are right, if cheaters continue to cheat, their rating should continue to climb.
However, even though this is not a metric on its own, I noticed that cheaters often have fresh accounts. I started to abort games when I notice that they are like "2000?" rated with only 30 games played. Though, if that happens too often, I risk myself a penalty time out for aborting too often games.

Many cheaters have not made it even into the 2000s, previously. What I experience at the (around) 2000 rapid rating level are maybe players that had previously been caught more early, and thus they never reach the 2000+ ratings. But as of late, they seem to have done so.

I also think that lichess check routinely any player above 2500 rating (as not too many have that), maybe by first calculating their performance over their last 20 games., and if the performance is then a 3000+ performance, maybe then an automated quick check is done. How exactly cheaters are caught - I dont know. Lichess keeps that secret, so cheaters do not know how to circumfence cheat-detection.

But ye, cheaters hardly reach 3000+, at some point they get caught before that.

Well munich, if you agree that all cheaters should reach 3000+ at one point, have you not seen your opponent rapidly progress following the game with you? and if they do achieve their rating goal or get banned before that, why haven't the mods banned most your reports that you inform us have been replied not to, yet? You should see a massive rating climb on their rating graph penultimately if they are cheating, no? Even if they are purposefully losing here and there, they should amass a massive rating or get banned doing it because that is the point of cheating.

and if you think they are losing enough games, to mimic the progression of the average player, what is the material difference between matching them and another random player strictly relating your outcome from the game?