WIRED on Intl Master of Chess beat online by an Unknown Player

[Vanadium 50]

You can't "model your play on an engine", because a human is fundamentally unable to run the engine's algorithms, even if he/she knew what they were.

Whereas, even a strong GM will start to play more conservatively once the win's is the bag.

Can you explain how both of these are true? Why couldn't someone be taught to play more aggressively even when the outcome is not in dispute?

 

[PeroK]

Can you explain how both of these are true? Why couldn't someone be taught to play more aggressively even when the outcome is not in dispute?

It's possible, of course, that a player may actively seek complications in a winning position, but players who reach any reasonable standard (let alone IM and GM) will have the mental scars of won positions thrown away for that very reason. This is another difference between humans and computers: for most humans it hurts to lose a chess game.

This is an area where even GM's lose points in the computer evaluation: the simple winning move that a GM might play without thinking can be seen as a blunder by an engine, which evaluates a much more complicated route to victory as vastly superior. Bobby Fischer, for example, was very principled about this: he always chose the simplest and most efficient way to win. The brilliancies were saved for when he needed them.

The other side of the coin is when the cheats start to run out of time, they fail to see the forced draw. They lose on time in a winning position (because of the time lag using an engine), where a strong player would trade in their winning advantage for a dead draw.

In my view, the chess.com algorithms have to be quite clever to spot cheating, but it's obvious to human eyes.

 

[Infrared]

In my view, the chess.com algorithms have to be quite clever to spot cheating, but it's obvious to human eyes.

I don't think it's always obvious. I agree that weaker players who always blindly copy the computer's recommendations are usually pretty easy to catch, but unfortunately there are more savvy cheaters. If a stronger player plays most of the game on his or her own and only consults the computer for a few critical moments, it can be very difficult to detect. There have been some pretty high profile cheating accusations that have turned out to be false (though this particular case looks open-and-shut).

 

[PeroK]

I don't think it's always obvious. I agree that weaker players who always blindly copy the computer's recommendations are usually pretty easy to catch, but unfortunately there are more savvy cheaters. If a stronger player plays most of the game on his or her own and only consults the computer for a few critical moments, it can be very difficult to detect. There have been some pretty high profile cheating accusations that have turned out to be false (though this particular case looks open-and-shut).

Yes, occasional use of the engine is much more difficult to detect.

 

[onatirec]

I've been following this story since the beginning, and it has never looked good for Dewas (aside from his recent tournament windfall). You can look up his older games on chess.com - they are of very low quality. Even a mediocre player could easily recognize his cheating.

Chess.com has a very advanced (proprietary) detection algorithm. Those GMs which have gotten a chance to peek under the hood practically fawn over it.

 

[jedishrfu]

Yeah I imagine they are very protective of their algorithm because there’s always someone looking to game the system and knowing how that algorithm works goes a long way to doing that.

Not to mention, competitors trying to steal your business as has happened with other websites being reimagined in China.