Dewa Kipas vs. Irene Sukandar match recap | GothamChess The original livestream now has 7 million views. Although the 0-3 result speaks for itself, Dewa Kipas did not admit to cheating.
Thanks, that's a really interesting video, even as someone that only plays chess at a very basic level. It seemed clear that he was completely outmatched and wasn't playing with nearly the skill...
Thanks, that's a really interesting video, even as someone that only plays chess at a very basic level. It seemed clear that he was completely outmatched and wasn't playing with nearly the skill you would have seen if he was legitimate.
Seems like a nontroversy. Both parties admit it was a mistake, wholly on Chess.com's part. Levy just assumed this unknown player with strange stats and unusual play at this level was a cheater,...
Seems like a nontroversy. Both parties admit it was a mistake, wholly on Chess.com's part. Levy just assumed this unknown player with strange stats and unusual play at this level was a cheater, something really common in online chess.
Chess.com won't admit it's a mistake, and the jury is out on whether this player is legit. Is this unknown player just better than even a GM from his own country? It's possible, but I think it's reasonable to think this is all just some stunt by a cheater who got caught.
That is a curious take. What is Chess.com's mistake? They have to ban presumed cheaters. Nobody wants to be on a platform with cheaters. Hopefully they didn't punish an innocent man, but it should...
That is a curious take.
What is Chess.com's mistake? They have to ban presumed cheaters. Nobody wants to be on a platform with cheaters.
Hopefully they didn't punish an innocent man, but it should be relatively easy for him to prove.
Also, I don't think Levy admitted a mistake.
“I knew I was in the right, but sometimes, you just take the L. It’s an insane situation.”
How do you even ban cheaters really? They could have a second laptop and copy the game state over and see the best next move. Seems like the best option is to rank players and match players with...
How do you even ban cheaters really? They could have a second laptop and copy the game state over and see the best next move. Seems like the best option is to rank players and match players with longer account histories and consistent stats together.
If you have a brand new account you get matched with cheaters and other new players for a while but eventually you graduate from that and get matched with real players and if you are found to bounce up and down stats too often, you get put in the untrusted pool again. Normal players will win and lose pretty close to an expected formula.
A traditional alpha-beta minimax based chess bot like Stockfish plays in a very distinctive way. Superior to any human play, but very distinctively not how humans play. It would be like if in a...
A traditional alpha-beta minimax based chess bot like Stockfish plays in a very distinctive way. Superior to any human play, but very distinctively not how humans play.
It would be like if in a race, one of the racers curled their limbs into wheels and started driving to the finish line at record speed. Superior to human play, not quite like how humans do it.
That the player had a statistically significant similarity to Stockfish would lead me to lean towards them being a cheater in the end. That's just not possible intentionally. He would have to be tremendously unlucky.
Their prime method seems to be to match moves to those made by chess engines. Even the really good players get a good amount of mismatch, while most cheaters will match relatively closely.
Their prime method seems to be to match moves to those made by chess engines. Even the really good players get a good amount of mismatch, while most cheaters will match relatively closely.
In addition to looking at matching percentage vs engine moves, they also look at factors like time between moves. A human player will take wildly varying amounts of time to make moves - they'll...
In addition to looking at matching percentage vs engine moves, they also look at factors like time between moves. A human player will take wildly varying amounts of time to make moves - they'll think for a while on a position, then make several quicker moves, as long as the opponent plays the line that they're expecting to get played. A player who is depending on an engine doesn't understand that the opponent's move is the expected one - they have to input it to the engine, see the result, then play it on their board, and they have to do this on every move.
There is an interesting article that shows how Chess.com determines how good a player is, by comparing them to the best moves according to all engines (CAPS) and each specific engine (if your play...
There is an interesting article that shows how Chess.com determines how good a player is, by comparing them to the best moves according to all engines (CAPS) and each specific engine (if your play matches one engine closely, chances are you are using that engine.)
Dewa_Kipas’ accuracy, or similarity to how a chess engine would play, during the match was 94 percent; Rozman’s was 76. Over its last 10 games, the account’s accuracy never dipped below 80. It hit over over 99 percent in two of those. Rozman reported him that day.
Chess.com determined that Dewa_Kipas’ moves in chess games matched a chess engine at a rate that is “not reasonably possible for a human player,” higher even than the top-ranked Indonesian chess player, grandmaster Susanto Megaranto: 95.3 to 94.4.
Chess.com has a separate system for people who are losing and then let the game time out, which is closer aligned to what you are thinking, where poor losers are all grouped together and only get to play other poor losers.
That's a great story and the article gets it right by not taking sides or swaying the reader towards unjustifiable conclusions.
Dewa Kipas vs. Irene Sukandar match recap | GothamChess
The original livestream now has 7 million views.
Although the 0-3 result speaks for itself, Dewa Kipas did not admit to cheating.
Thanks, that's a really interesting video, even as someone that only plays chess at a very basic level. It seemed clear that he was completely outmatched and wasn't playing with nearly the skill you would have seen if he was legitimate.
Seems like a nontroversy. Both parties admit it was a mistake, wholly on Chess.com's part. Levy just assumed this unknown player with strange stats and unusual play at this level was a cheater, something really common in online chess.
Chess.com won't admit it's a mistake, and the jury is out on whether this player is legit. Is this unknown player just better than even a GM from his own country? It's possible, but I think it's reasonable to think this is all just some stunt by a cheater who got caught.
That is a curious take.
What is Chess.com's mistake? They have to ban presumed cheaters. Nobody wants to be on a platform with cheaters.
Hopefully they didn't punish an innocent man, but it should be relatively easy for him to prove.
Also, I don't think Levy admitted a mistake.
How do you even ban cheaters really? They could have a second laptop and copy the game state over and see the best next move. Seems like the best option is to rank players and match players with longer account histories and consistent stats together.
If you have a brand new account you get matched with cheaters and other new players for a while but eventually you graduate from that and get matched with real players and if you are found to bounce up and down stats too often, you get put in the untrusted pool again. Normal players will win and lose pretty close to an expected formula.
A traditional alpha-beta minimax based chess bot like Stockfish plays in a very distinctive way. Superior to any human play, but very distinctively not how humans play.
It would be like if in a race, one of the racers curled their limbs into wheels and started driving to the finish line at record speed. Superior to human play, not quite like how humans do it.
That the player had a statistically significant similarity to Stockfish would lead me to lean towards them being a cheater in the end. That's just not possible intentionally. He would have to be tremendously unlucky.
Interesting analogy :P
This analogy is so good that it's popped into my head several times over the past several days since I first read your comment.
Their prime method seems to be to match moves to those made by chess engines. Even the really good players get a good amount of mismatch, while most cheaters will match relatively closely.
In addition to looking at matching percentage vs engine moves, they also look at factors like time between moves. A human player will take wildly varying amounts of time to make moves - they'll think for a while on a position, then make several quicker moves, as long as the opponent plays the line that they're expecting to get played. A player who is depending on an engine doesn't understand that the opponent's move is the expected one - they have to input it to the engine, see the result, then play it on their board, and they have to do this on every move.
There is an interesting article that shows how Chess.com determines how good a player is, by comparing them to the best moves according to all engines (CAPS) and each specific engine (if your play matches one engine closely, chances are you are using that engine.)
https://www.chess.com/article/view/who-was-the-best-world-chess-champion-in-history
Chess.com has a separate system for people who are losing and then let the game time out, which is closer aligned to what you are thinking, where poor losers are all grouped together and only get to play other poor losers.