Live data from Hacker News

Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

wsj.com

91–100 of 339 posts

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#91
This is the most interesting part for me:

     Computers have “nearly infallible tactical calculation,” the report says, and are capable of beating even the best human every single time. The report says dozens of grandmasters have been caught cheating on the website, including four of the top-100 players in the world who confessed. 
I can't really comment it, but I leave it here if you haven't read the article.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#92
post #19

The next 'move' for cheaters is to use chess computers in a way that passes 'Chess Turing Test' and makes cheating indistinguishable from normal human play under analysis. When there is money in the game, there is incentive to cheat. > The report says dozens of grandmasters have been caught cheating on the website, including four of the top-100 players in the world who confessed. There are probably smart cheaters alr…

Cheat detection isn't done by only by move analysis, but by an extensive profiling of a person and games based on many factors beyond even just the moves. For instance one of the easiest ways to catch a weak player cheating is move times. Such a player will have no idea whether a move is trivial or works only due to an exceptionally precise and lengthy series of counter-intuitive calculations that no human could do w…

This is exactly how I get clued in on how someone is cheating in a shooting game I play.

You can tell how experienced someone is based off the gun they use (some are stronger than others), whether they use cover or just run out into open spaces and shoot, how they move, whether they use 'gadgets' like grenades, and so on. A lot of novice players don't even use the sprint function to run.

When someone who literally just walks around the map but can laser everyone with headshots (which have a significant damage multiplier)? They're cheating.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#93

The fact that an athlete in a competitive sport was allowed to partake in an event even after admitting to cheating not only once, but twice, is outrageous in itself.

This is, I think, the single biggest factor in the proliferation of cheating. Most cheaters are smart enough to do the math on how they should win at the sport, or is smart enough to hire someone who can do the math. In other words, cheaters are rational actors. Allowing people back into the sport swings the EV math heavily in favor of cheating if the penalty isn't massive given that the chance of getting caught is s…

The alternative - an environment where innocent people were sometimes getting expelled - would be worse when it's just a game.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#94
post #28

Just to make sure I read this right: he most likely cheated in 11 online tournaments from 2015 - 2020. Why not analyze his recent and over-the-board games?

Kenneth Regan did that and came to a conclusion that there were no cheating.

I'd say he could not come to a conclusion that there were cheating.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#95
post #86

Earlier quoted context omitted.

Which shows ... I don't know ... naivety, or cockiness, or something. If I'm cheating at chess (or poker or whatever) online and I know that there are likely to be some form of anti-cheat scanning happening... why wouldn't you run your chess client on a computer beside you - then everything surrounding that is undetectable (processes, focus, CPU utilization, etc.)

People are still rather short-sighted when it comes to what a computer can do. They can really only think in front of them. If I put information into this program, I get information out. They don't think that applications can monitor their own meta-state. Or the state of other applications. So I'm not terribly surprised that he thought running the engine in another browser window would have been sufficient. He might…

Most people don't realize that the online chat functions for a lot of customer service sites show every character you type, not when you hit 'enter' or 'send'.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#96
post #19

The next 'move' for cheaters is to use chess computers in a way that passes 'Chess Turing Test' and makes cheating indistinguishable from normal human play under analysis. When there is money in the game, there is incentive to cheat. > The report says dozens of grandmasters have been caught cheating on the website, including four of the top-100 players in the world who confessed. There are probably smart cheaters alr…

You could train a neural network to filter through likely human moves from the engine's top recommendations and never get caught.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#97

The fact that an athlete in a competitive sport was allowed to partake in an event even after admitting to cheating not only once, but twice, is outrageous in itself.

Are we calling chess a sport?

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#98

Earlier quoted context omitted.

Cheat detection isn't done by only by move analysis, but by an extensive profiling of a person and games based on many factors beyond even just the moves. For instance one of the easiest ways to catch a weak player cheating is move times. Such a player will have no idea whether a move is trivial or works only due to an exceptionally precise and lengthy series of counter-intuitive calculations that no human could do w…

What happens when it’s a strong chess player who is cheating? Even a strong player can benefit from consulting a computer. Chess games can win fail based on a few moves. A strong player would only need to consult the computer on a few moves to get a considerable advantage.

The strongest players would be the "best" cheaters but also the least likely to cheat. Top n players for small n would only need one "this move is important" hint per game to significantly improve their rating. Very hard to detect of course. But when you hear the actual top players talk about chess they genuinely seem interested in playing the game and not so much driven by chasing some accomplishment of winning. It's quite hard to get to that level without having the genuine interest.

But also their games are subject to the most scrutiny and study and they themselves will spend a lot of time publicly talking about and analyzing their own games, those "cheat" moves would stand out as ones which were hard to see and had bad explanations after a while.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#99

Why don't cheaters just use two machines or even just their phone and a laptop? The evidence is often around other processes or browser tabs running on their device (and in this case also focus loss) but an immediate thought must be to simply use multiple devices.

Traditionally chess engine moves were identifiable easily enough by regular players, and chess.com even has algorithms in place to detect this.

Thats all going to change now though, and its totally possible to cheat using a second computer with an engine that will be undetectable.

Re: Chess Investigation Finds U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times

#100

Earlier quoted context omitted.

The admissions were from when he was 12 and 16; many so societies generally believe in redemption from childhood transgressions.

Report says over 100, as late as 2020.

He already admitted to cheating a lot. He didn’t say he cheated in two games, he said cheated at two times in his life. He even detailed how his second time was pumping up his rating, you can’t pump a rating with only one game. Sounds like if it went to 2020 then he would have been 17 at least, not 16 but that’s pretty close at least.
Post reply on HN