Although kinda irrelevant but I seriously liked his lichess videos from before. After signing a contract with chess.com, he has more and more unwatchable. And more and more crazy!!!
Ben Finegold said it himself live on stream. Compare his statement to these prewritten PR statements. Who is telling the truth? I don't know anymore. I'm so confused now. I mean surely a 2 hour presentation means something, right? If only the agreement didn't prevent them writing on the specifics! Things would be so much simpler.... Alas, now we don't know what to believe.
GM Eric Hansen:
The agreement I signed for Chess.com prevents me from writing on the specifics, but I can say that I was more than impressed (and pleasantly surprised) with some of the things Chess.com's fair-play team showed me. They are as capable of keeping the game clean as anything I've ever seen."
IM John Bartholomew:
The cheating-detection methods that Roland and Danny demonstrated to me during a two-hour presentation are the most thorough and sophisticated I have ever seen. The Chess.com team has gone above and beyond to ensure fair play on their site, and this is reflected in the time and seriousness they have dedicated to testing and refining these methods. I have full confidence in Chess.com's ability to accurately identify cheaters.
GM Benjamin Finegold:
"No matter how good chess.com thinks their detection system is, they're probably catching less than half the cheaters"
GM Eric Hansen:
The agreement I signed for Chess.com prevents me from writing on the specifics, but I can say that I was more than impressed (and pleasantly surprised) with some of the things Chess.com's fair-play team showed me. They are as capable of keeping the game clean as anything I've ever seen."
IM John Bartholomew:
The cheating-detection methods that Roland and Danny demonstrated to me during a two-hour presentation are the most thorough and sophisticated I have ever seen. The Chess.com team has gone above and beyond to ensure fair play on their site, and this is reflected in the time and seriousness they have dedicated to testing and refining these methods. I have full confidence in Chess.com's ability to accurately identify cheaters.
GM Benjamin Finegold:
"No matter how good chess.com thinks their detection system is, they're probably catching less than half the cheaters"
Would you buy a used car from this man?
As an aside, in case anyone is wondering and for the record, my patronage has been interrupted by some kind of snafu somewhere which I am working to resolve.
John Bartholomew's remark is particularly silly. No detection method is perfect for the simple reason that cheaters adapt to the methods. If you play classical, there is enough time to check positions with an engine, and cheaters who wish to avoid getting caught can do it just at some difficult or complex situations. How do you catch those? Answer: you can't. It takes only a bit of common sense to see that. Either Bartholomew lacks any, or he's too much of a toady. As Finwgold observed on the original stream, there are also cheaters who set the engine at a lower rating, say 2700. Perhaps eventually those can be caught, but not easily or immediately.
Regarding GM Hansen's remark, a careful reading shows it is not incompatible with Finegold's belief that only half the cheaters at any one time have been caught. You can be impressed with the detection methods while understanding that sophisticated cheaters can still get away with it, at least for a while.
Regarding GM Hansen's remark, a careful reading shows it is not incompatible with Finegold's belief that only half the cheaters at any one time have been caught. You can be impressed with the detection methods while understanding that sophisticated cheaters can still get away with it, at least for a while.
You can’t cheat in the long run successfully. Even if your task is do distribute 0s and 1s randomly an AI will catch you out (human or not).
But agreed: cheating cannot be prevented in advance. In corr. chess it is difficult as well because the time consumption is lacking resp. useless.
Anti-Cheating algorithms are expert systems which can’t be deceived by simple means:
www.reddit.com/r/chess/comments/9s9ml8/how_does_chesscom_and_lichess_deal_with_gms/e8n9uwg
But agreed: cheating cannot be prevented in advance. In corr. chess it is difficult as well because the time consumption is lacking resp. useless.
Anti-Cheating algorithms are expert systems which can’t be deceived by simple means:
www.reddit.com/r/chess/comments/9s9ml8/how_does_chesscom_and_lichess_deal_with_gms/e8n9uwg
"Anti-Cheating algorithms are expert systems which can’t be deceived by simple means"
I don't know what YOU mean by "simple", but it seems simple enough to me to copy a position to the engine and then move as the engine would for two or three moves. How can you "catch" that? It is obvious that you cannot, for the simple reason that you cannot prove that the player didn't happen to make a great move. Even mediocre players occasionally make great moves. If you can get over your quasi-theological belief in the omnipotence of AI, you will realise that a programmer's dilemma is similar to that of the law generally: is it not better to let some guilty people go free than to punish the innocent? (Judge Blackstone famously stated in the 18th century that it was better that 10 guilty go free than to convict one innocent). You can't design an algorithm that will catch all cheaters while never accusing an innocent who happened to come up with a good move.
I don't know what YOU mean by "simple", but it seems simple enough to me to copy a position to the engine and then move as the engine would for two or three moves. How can you "catch" that? It is obvious that you cannot, for the simple reason that you cannot prove that the player didn't happen to make a great move. Even mediocre players occasionally make great moves. If you can get over your quasi-theological belief in the omnipotence of AI, you will realise that a programmer's dilemma is similar to that of the law generally: is it not better to let some guilty people go free than to punish the innocent? (Judge Blackstone famously stated in the 18th century that it was better that 10 guilty go free than to convict one innocent). You can't design an algorithm that will catch all cheaters while never accusing an innocent who happened to come up with a good move.
Cheat a couple of times you won’t get caught. Cheat once more you‘ll will.
(By the way, modern AIs will taught themselves. They „develop“ what‘s relevant, see AlphaZero. It is not „programmed“ in the common sense.)
(By the way, modern AIs will taught themselves. They „develop“ what‘s relevant, see AlphaZero. It is not „programmed“ in the common sense.)
@nayf
You'd be surprised what properly trained machine learning models can predict.
Still, there won't be a 100% accuracy, that's a valid point.
@Sarg0n
Irwin uses a different method than AlphaZero, and it's safe to assume that it is not "perfect" (only "very good"). Both are also limited by the ressources (hardware) they are allowed to use.
You'd be surprised what properly trained machine learning models can predict.
Still, there won't be a 100% accuracy, that's a valid point.
@Sarg0n
Irwin uses a different method than AlphaZero, and it's safe to assume that it is not "perfect" (only "very good"). Both are also limited by the ressources (hardware) they are allowed to use.
@NeverBeenTimid
Stop claiming things you don't know. Ben Finegold clearly said in his stream that he's not being paid a cent by chess.com for playing on their website. They only helped him setting up the stream.
Stop claiming things you don't know. Ben Finegold clearly said in his stream that he's not being paid a cent by chess.com for playing on their website. They only helped him setting up the stream.
This topic has been archived and can no longer be replied to.