Anyone following the huge chess cheating controversy?

[rj144]

I have a question for all you chess aficionados.

What was the difference between DEEP BLUE and any other AI in a regular chess game or app like battle chess back on pc ?

Most chess engines aren't AI.

 

[Randlewand]

DId we determine what brand the anal beads are yet?

 

[AristotleAmadopolis]

Most chess engines aren't AI.

They just programs?

 

[SherdogsMVP]

Most chess engines aren't AI.

But what was then the difference than ?

Deep blue was designed to play chess. What separated it from regular AI in single player chess games

 

[Madmick]

But what was then the difference than ?

Deep blue was designed to play chess. What separated it from regular AI in single player chess games

It was a combination of the processing power of the supercomputer IBM built specifically to run it, and the sophistication of the software engine they designed to execute that power. They had teams of computer scientists and grandmasters working together to develop this. This code was a tighly guarded secret, and I've never looked into whether or not the code was opened to the public. It was also designed specifically to defeat one opponent, Garry Kasparov, so there is the possibility that extra care was paid to openings they believed him most likely to play. Apparently they were tweaking the code even during the matches themselves, figuring out where the machine failed, and trying to prevent that failure in future matches.

Kasparov famously believed the IBM team cheated, and used grandmasters to help feed the machine moves at critical positions. There was a documentary that investigated this theory (with a rather heavy pro-Kasparov bias).
https://en.wikipedia.org/wiki/Game_Over:_Kasparov_and_the_Machine
https://www.imdb.com/title/tt0379296/

The competition with chess engines is to write code that is the most efficient at analyzing positions intelligently. The reason is there are far, far more positions than any computer-- even today's supercomputers-- could hope to analyze. We're talking about millions of years. So you can't brute force it. Essentially, what a chess engine is supposed to do is dispense with calculations that are the most likely to waste time by assessing terrible moves. And, in this sense, they're not that much different than the human mind. The tricky part is figuring out how to teach the engine which moves are most likely to be bad, especially since we're learning from the engines, now, that certain moves we wouldn't have considered viable are actually the strongest in a given position-- at least if played by a computer not prone to human mistakes.

You can actually play some of those old chess engines against the best open-source engines today. Of course, today, Stockfish is king. Plug Stockfish into an old game like Chessmaster, and Stockfish will kick the shit out of Chessmaster on its most competitive setting. This is running on the same hardware. The software is vastly superior. This is an ongoing competition in the chess world. The golden standard of these competitions is the one that controls the hardware. In other words, which engine outperforms the others on identical hardware? There is a Twitch channel that specializes in showing nothing but games between chess engines:
https://m.twitch.tv/computerchess/home

 

[SherdogsMVP]

It was a combination of the processing power of the supercomputer IBM built specifically to run it, and the sophistication of the software engine they designed to execute that power. They had teams of computer scientists and grandmasters working together to develop this. This code was a tighly guarded secret, and I've never looked into whether or not the code was opened to the public. It was also designed specifically to defeat one opponent, Garry Kasparov, so there is the possibility that extra care was paid to openings they believed him most likely to play. Apparently they were tweaking the code even during the matches themselves, figuring out where the machine failed, and trying to prevent that failure in future matches.

Kasparov famously believed the IBM team cheated, and used grandmasters to help feed the machine moves at critical positions. There was a documentary that investigated this theory (with a rather heavy pro-Kasparov bias).
https://en.wikipedia.org/wiki/Game_Over:_Kasparov_and_the_Machine
https://www.imdb.com/title/tt0379296/

The competition with chess engines is to write code that is the most efficient at analyzing positions intelligently. The reason is there are far, far more positions than any computer-- even today's supercomputers-- could hope to analyze. We're talking about millions of years. So you can't brute force it. Essentially, what a chess engine is supposed to do is dispense with calculations that are the most likely to waste time by assessing terrible moves. And, in this sense, they're not that much different than the human mind. The tricky part is figuring out how to teach the engine which moves are most likely to be bad, especially since we're learning from the engines, now, that certain moves we wouldn't have considered can be viable-- at least if played by a computer not prone to human mistakes-- are actually the strongest in a given position.

You can actually play some of those old chess engines against the best open-source engines today. Of course, today, Stockfish is king. Plug Stockfish into an old game like Chessmaster, and Stockfish will kick the shit out of Chessmaster on its most competitive setting. This is running on the same hardware. The software is vastly superior. So this is an ongoing competition in the chess world. The golden standard of these competitions is the one that controls the hardware. In other words, which engine outperforms the others on identical hardware? There is a Twitch channel that specializes in showing nothing but games between chess engines:
https://m.twitch.tv/computerchess/clip/CoweringFineCrabMrDestructoid

I did read about deep blue team few days ago thanks to this thread as the curiosity got the best of me n heard kasparov thought some tampering was in play.

Has there been anything newer or more advanced than deep blue since ? Have the humans beaten the best AI yet or is AI still king in the chess world ?

 

[rj144]

I did read about deep blue team few days ago thanks to this thread as the curiosity got the best of me n heard kasparov thought some tampering was in play.

Has there been anything newer or more advanced than deep blue since ? Have the humans beaten the best AI yet or is AI still king in the chess world ?

Again, the best engine, Stockfish, is not AI. Stockfish is more advanced than Deep Blue and no player will ever beat it.

 

[Madmick]

Again, the best engine, Stockfish, is not AI.

That's not technically true, anymore.

 

[rj144]

That's not technically true, anymore.

Sure, it's a neural net now, but not pure AI.

 

[Zer]

I have a question for all you chess aficionados.

What was the difference between DEEP BLUE and any other AI in a regular chess game or app like battle chess back on pc ?

I'm definitely not a chess guy but there is an outstanding video on Deep Blue, focused around its development and then its famous matches against Garry Kasparov. It goes into so much detail including the technical aspects of Deep Blue and how it separated itself from other Chess Machines (If you've got a spare 2 hrs)

 

[SherdogsMVP]

I'm definitely not a chess guy but there is an outstanding video on Deep Blue, focused around its development and then its famous matches against Garry Kasparov. It goes into so much detail including the technical aspects of Deep Blue and how it separated itself from other Chess Machines (If you've got a spare 2 hrs)

I'll give it a watch tonight n check it out. I can't believe this recent controversy has ignited a curiosity for so many people in the world of chess

 

[SaiWa]

A body language analysis seems to indicate that Hans Niemann is lying:



The TLDW:

Hans Niemann fakes an accent to sound more intelligent.

Hans Niemann contradicts his explanation for how he acquired his accent and became good at chess.

IMO, I think Hans is a bit of a poser/narcissist, his hair, his accent, his frown pose, he is almost trying to copy Magnus Carlsen and try to humiliate him on the chess world stage.

Generic tabloid level body language analysis. Body language reading isn't reliable at all unless you are extremely familiar with the subject.

He is definitely an arrogant piece of shit though. But I believe it takes that "kind" of attitude (confidence or arrogance) to achieve greatness. Whether you actually achieve that greatness is what differentiate arrogance from confidence...

 

[SaiWa]

You are right I specified cheating in more broader terms regarding matches that are simple impossible regarding accuracy to have happened without digital assistance. Its the cheating without official note. If you dont want to consider that cheating fine but I dont think we are overstepping to consider him someone with an extensive background in cheating. The circumstantial evidence is too strong. Everything fits together from his coach, documented matches ( I have to look up the data gain but it was at least 10 near engine perfect matches ), according to different sources even more than 20 with 100%. That IS cheating. Magnus has 3 btw) to him lying about how he was able to play like this (ok consider it remembering "wrong" but that is much less probable than him lying). Niemann is a cheater who perfected that together with his coach (Maxim Dlugy) as a way to break into the elite bracket. Thats my view on him and the experts that analysed his matches including Carlsen are enough evidence for me.

People focus way too much on the anal beads to discredit and make fun of Carlsens view on Niemann but that was always just one of many possible ways Niemann cheats.^

One perfect match especially a shorter one is nothing suspicious but as often as Niemann only happened once in entire history (correct me if you know more examples) with Sebastien Feller in 2010 with 98% . This one occasion was later proven to be cheating by Feller communicating with two other GMs that used a chess engine and then had developed a very complex scheme to message Feller the correct moves. BUt I get that its very difficult to prove engine accuracy as cheating and there seems to not be one accepted method.

What do you guys think of the Rios match in 2021 with 45 moves and considered near 100% engine correlation?
https://www.chess.com/events/2021-sharjah-masters/02/Niemann_Hans_Moke-Rios_Cristhian_Camilo

This is an excellent summary of the situation:



As this commentator stated the more engines you use the easier it will be to get that perfect 100%. Apparently Hikaru was checking out one of his games and the engine came back with 100% when he streamed live. That part was conveniently left out of the youtube video he released. Kinda fucked up but whatever...

 

[King Joffery]

So how do they get the message across? by Morse code? long vibration vs short vibration?
That must take a long time to say "Bishop to A4"
The guy must have been smiling on and off intermittently according to the message

Not that long, you don't even need morse code, any move can be represented by 4 coordinates and each coordinate could be represented by 1-8 quick vibrations. Moves to H8 would be the most pleasurable.

 

[Goonerview]

Generic tabloid level body language analysis. Body language reading isn't reliable at all unless you are extremely familiar with the subject.

He is definitely an arrogant piece of shit though. But I believe it takes that "kind" of attitude (confidence or arrogance) to achieve greatness. Whether you actually achieve that greatness is what differentiate arrogance from confidence...

Yeah, he comes across like a real punk.

Apparently, Fide are officially investigating him for cheating now. I hope the truth comes out.

 

[SaiWa]

Yeah, he comes across like a real punk.

Apparently, Fide are officially investigating him for cheating now. I hope the truth comes out.

Meh. He's young and has a chip on his shoulder. Chess (or any sport) needs people like him (not the cheating part but the arrogant personality) to make it interesting. They can't all be quiet and respectful like Caruana. Hell Hikaru Nakamura used to be a punk ass arrogant player as well. He was known as one of the most disrespectful elite chess players on the tour. Asshats that people can root against (or for).

 

[PurpleStorm]

• WSJ NEWS EXCLUSIVE

Chess Investigation Finds That U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times
An internal report reviewed by The Wall Street Journal alleges a previously unknown pattern of likely widespread cheating by Hans Moke Niemann, the player whose September victory over Magnus Carlsen has rocked the chess world.

https://www.wsj.com/articles/chess-cheating-hans-niemann-report-magnus-carlsen-11664911524

 

[Goonerview]

• WSJ NEWS EXCLUSIVE

Chess Investigation Finds That U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times
An internal report reviewed by The Wall Street Journal alleges a previously unknown pattern of likely widespread cheating by Hans Moke Niemann, the player whose September victory over Magnus Carlsen has rocked the chess world.

https://www.wsj.com/articles/chess-cheating-hans-niemann-report-magnus-carlsen-11664911524

Was going to post this. Some Brazilian data scientist has found proof that Niemann’s progression into a top tier chess player was not natural:

https://en.chessbase.com/post/statistical-analysis-of-the-games-of-hans-niemann

It suggests cheating.

What I find most curious about Niemann’s interviews is that he doesn’t seem to explicitly say “I didn’t cheat”, he often tries to obfuscate.

 

[SalvadorAllende]

Again, the best engine, Stockfish, is not AI. Stockfish is more advanced than Deep Blue and no player will ever beat it.

Pretty sure Alphazero kicks Stockfish ass most of the time.

 

[Madmick]

• WSJ NEWS EXCLUSIVE

Chess Investigation Finds That U.S. Grandmaster ‘Likely Cheated’ More Than 100 Times
An internal report reviewed by The Wall Street Journal alleges a previously unknown pattern of likely widespread cheating by Hans Moke Niemann, the player whose September victory over Magnus Carlsen has rocked the chess world.

https://www.wsj.com/articles/chess-cheating-hans-niemann-report-magnus-carlsen-11664911524

Not surprising at all that Rensch had him dead to rights online. Anyone who knows the website knows Rensch doesn't publicly make those allegations unless he has a boatload of evidence. We'll see if FIDE turns anything up about OTB games, but as the article states, Chess.com only descibed his rise as "unusual" and "statistically extraordinary" in that realm, not implausible, or impossible. Now the onus lies with FIDE (which only might prove competent because Hans isn't Russian).

Was going to post this. Some Brazilian data scientist has found proof that Niemann’s progression into a top tier chess player was not natural:

https://en.chessbase.com/post/statistical-analysis-of-the-games-of-hans-niemann

It suggests cheating.

What I find most curious about Niemann’s interviews is that he doesn’t seem to explicitly say “I didn’t cheat”, he often tries to obfuscate.

Cool stuff. I am cynical, though, because even as ancillary probabalistic evidence, I don't see FIDE banning a player on it. They're going to need to find a smoking gun.

After all, FIDE's not going to lead with analysis that shows Hans's average level of play above 2600 in classical has been almost perfectly in concert with what would have been projected, but with his consistency slightly below average for that rating.