Chess engines simply play differently than humans. It's fairly easy to detect by pure statistics. For example, chess.com bans about 16 000 players for fair play abuse each month.
Your ban most likely wasn't because of the site detecting A.C.A.S, it was because of your suspicious behaviour patterns. A.C.A.S cannot fix this, it's your responsiblity to play as a human.
Don't want to get banned again? Don't use A.C.A.S against other humans.
Your ban most likely wasn't because of the site detecting A.C.A.S, it was because of your suspicious behaviour patterns. A.C.A.S cannot fix this, it's your responsiblity to play as a human.
Don't want to get banned again? Don't use A.C.A.S against other humans.
@blundererxd said in #32:
> Chess engines simply play differently than humans. It's fairly easy to detect by pure statistics. For example, chess.com bans about 16 000 players for fair play abuse each month.
>
> Your ban most likely wasn't because of the site detecting A.C.A.S, it was because of your suspicious behaviour patterns. A.C.A.S cannot fix this, it's your responsiblity to play as a human.
>
> Don't want to get banned again? Don't use A.C.A.S against other humans.
Just like drugs in fact : what's your problem !? it's so easy to find people using drugs among other people... Don't want to get jailed ? Don't use drugs.
For christ sake, guys, you drive me crazy. Forget about it and keep blind. No problem.
> Chess engines simply play differently than humans. It's fairly easy to detect by pure statistics. For example, chess.com bans about 16 000 players for fair play abuse each month.
>
> Your ban most likely wasn't because of the site detecting A.C.A.S, it was because of your suspicious behaviour patterns. A.C.A.S cannot fix this, it's your responsiblity to play as a human.
>
> Don't want to get banned again? Don't use A.C.A.S against other humans.
Just like drugs in fact : what's your problem !? it's so easy to find people using drugs among other people... Don't want to get jailed ? Don't use drugs.
For christ sake, guys, you drive me crazy. Forget about it and keep blind. No problem.
I think you have two main concerns here that get intertwined and harder to respond to. So I'll split them into 2.
1. Cheating in general. You might feel that at your rating range, cheating is rampant. But if that were the case, then you cannot possibly maintain your rating. If let's say you face cheaters as little as 10% of the time, your rating will drop by around 100 since your expected score for every 10 games would be 4.5. The fact that you're still at your rating range means that fewer than 10% of players you face are cheaters.
2. Technologically undetectable cheating. I'm not sure why this is so bothersome. Technologically undetectable cheating has always existed. The simplest is the cheater I faced before. I beat him quite a few times easy. And then suddenly, he beat me up 5 games straight. Finally, he admitted that he lent his stronger friend his phone to play against me. Those were casual games, I didn't bother to report. There are other technologically undetectable cheating like having a friend watch you and enter the moves in another device with an engine. The possibilities are endless. Trying to solve all of these is a waste of limited resources.
Everyone here is saying that statistical detection is the best method we have. It's not foolproof. It's not 100%, but it works.
TLDR as long as you're maintaining your rating range, you're mathematically not facing a significant number of cheaters. Statistical cheat detection is sound.
1. Cheating in general. You might feel that at your rating range, cheating is rampant. But if that were the case, then you cannot possibly maintain your rating. If let's say you face cheaters as little as 10% of the time, your rating will drop by around 100 since your expected score for every 10 games would be 4.5. The fact that you're still at your rating range means that fewer than 10% of players you face are cheaters.
2. Technologically undetectable cheating. I'm not sure why this is so bothersome. Technologically undetectable cheating has always existed. The simplest is the cheater I faced before. I beat him quite a few times easy. And then suddenly, he beat me up 5 games straight. Finally, he admitted that he lent his stronger friend his phone to play against me. Those were casual games, I didn't bother to report. There are other technologically undetectable cheating like having a friend watch you and enter the moves in another device with an engine. The possibilities are endless. Trying to solve all of these is a waste of limited resources.
Everyone here is saying that statistical detection is the best method we have. It's not foolproof. It's not 100%, but it works.
TLDR as long as you're maintaining your rating range, you're mathematically not facing a significant number of cheaters. Statistical cheat detection is sound.
@GnocchiPup Unfortunately, I cannot agree with your first point. Many believe, their "true strength" is at least 500 pts higher than their rating, and this difference is a consequence of many cheaters.
Guys, read carefully:
-Irwin looks for play characteristic of chess engines: github.com/clarkerubber/irwin
-Kaladin looks for play uncharacteristic of the players: github.com/lichess-org/kaladin
-Irwin looks for play characteristic of chess engines: github.com/clarkerubber/irwin
-Kaladin looks for play uncharacteristic of the players: github.com/lichess-org/kaladin
Hehe, people would love to believe that.
But the math doesn't lie. If cheaters were rampant, the lichess normal curve won't look so normal. I'm thinking there would be a couple of overlapping normal curves if that were true. Imagine 2 normal curves, the non cheaters have our mean at 1500. The cheaters would have their mean maybe at 2500? Then overlap them together, it would look like a mountain with a couple of peaks. The fact of the matter is, we only have one peak, at around 1500, and it descends rapidly on both sides. Except at the 00 marks, where there are little peaks of people parking their rating once they achieve the 00 milestone. Like me.
But the math doesn't lie. If cheaters were rampant, the lichess normal curve won't look so normal. I'm thinking there would be a couple of overlapping normal curves if that were true. Imagine 2 normal curves, the non cheaters have our mean at 1500. The cheaters would have their mean maybe at 2500? Then overlap them together, it would look like a mountain with a couple of peaks. The fact of the matter is, we only have one peak, at around 1500, and it descends rapidly on both sides. Except at the 00 marks, where there are little peaks of people parking their rating once they achieve the 00 milestone. Like me.
@GnocchiPup said in #34:
> I think you have two main concerns here that get intertwined and harder to respond to. So I'll split them into 2.
>
> 1. Cheating in general. You might feel that at your rating range, cheating is rampant. But if that were the case, then you cannot possibly maintain your rating. If let's say you face cheaters as little as 10% of the time, your rating will drop by around 100 since your expected score for every 10 games would be 4.5. The fact that you're still at your rating range means that fewer than 10% of players you face are cheaters.
>
> 2. Technologically undetectable cheating. I'm not sure why this is so bothersome. Technologically undetectable cheating has always existed. The simplest is the cheater I faced before. I beat him quite a few times easy. And then suddenly, he beat me up 5 games straight. Finally, he admitted that he lent his stronger friend his phone to play against me. Those were casual games, I didn't bother to report. There are other technologically undetectable cheating like having a friend watch you and enter the moves in another device with an engine. The possibilities are endless. Trying to solve all of these is a waste of limited resources.
>
> Everyone here is saying that statistical detection is the best method we have. It's not foolproof. It's not 100%, but it works.
>
> TLDR as long as you're maintaining your rating range, you're mathematically not facing a significant number of cheaters. Statistical cheat detection is sound.
Thank you for your answer. At least it comes with a sound explanation, even if i'm not sure i understand your conclusion since I'm currently 100 rating points below what should "theoretically" be my rating range. For now, it seems like there is a step at 1600 i can't manage to pass and i feel like there is more "helped" players in one range than in an other : for example it's easier for me to win 1600-1700 players than 1500-1600 and it doesn't feel right. So, what's your conclusion if i'm not maintaining my rating range ? I've reached my max ?
> I think you have two main concerns here that get intertwined and harder to respond to. So I'll split them into 2.
>
> 1. Cheating in general. You might feel that at your rating range, cheating is rampant. But if that were the case, then you cannot possibly maintain your rating. If let's say you face cheaters as little as 10% of the time, your rating will drop by around 100 since your expected score for every 10 games would be 4.5. The fact that you're still at your rating range means that fewer than 10% of players you face are cheaters.
>
> 2. Technologically undetectable cheating. I'm not sure why this is so bothersome. Technologically undetectable cheating has always existed. The simplest is the cheater I faced before. I beat him quite a few times easy. And then suddenly, he beat me up 5 games straight. Finally, he admitted that he lent his stronger friend his phone to play against me. Those were casual games, I didn't bother to report. There are other technologically undetectable cheating like having a friend watch you and enter the moves in another device with an engine. The possibilities are endless. Trying to solve all of these is a waste of limited resources.
>
> Everyone here is saying that statistical detection is the best method we have. It's not foolproof. It's not 100%, but it works.
>
> TLDR as long as you're maintaining your rating range, you're mathematically not facing a significant number of cheaters. Statistical cheat detection is sound.
Thank you for your answer. At least it comes with a sound explanation, even if i'm not sure i understand your conclusion since I'm currently 100 rating points below what should "theoretically" be my rating range. For now, it seems like there is a step at 1600 i can't manage to pass and i feel like there is more "helped" players in one range than in an other : for example it's easier for me to win 1600-1700 players than 1500-1600 and it doesn't feel right. So, what's your conclusion if i'm not maintaining my rating range ? I've reached my max ?
@kinsana said in #38:
> for example it's easier for me to win 1600-1700 players than 1500-1600 and it doesn't feel right.
A single blunder may be equal to 300+ rating points in term of winning chances. Weaker players may take advantage from any your mistake if you allow them.
> for example it's easier for me to win 1600-1700 players than 1500-1600 and it doesn't feel right.
A single blunder may be equal to 300+ rating points in term of winning chances. Weaker players may take advantage from any your mistake if you allow them.
Cheat detection on lichess is a joke. Lichess does a good job and has probably a very good detection system, in my honest view. But in my view, and like @kinsana 's view, it is the cheaters that have got their hands on too smart cheat programs and bots. Also, several players are smart enough to cheat/use their programs in only parts of their games. Most cheaters wil not choose the engine moves that is clearly too strange and the most advanced, i.e. like the nextbestmove-calculator. That is what I mean by stating that the cheat detection system "is a joke".
This topic has been archived and can no longer be replied to.