It is currently Thu Mar 28, 2024 6:53 am

All times are UTC - 8 hours [ DST ]




Post new topic Reply to topic  [ 8 posts ] 
Author Message
Offline
 Post subject: Cheating 1.0
Post #1 Posted: Wed Mar 20, 2019 1:01 am 
Gosei

Posts: 1494
Liked others: 111
Was liked: 315
Making use of the excellent gomill, there is a software project to detect cheating
https://github.com/IgorBS/GoCheaterCatcher
I look forward to trying to beat it!

_________________
North Lecale

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #2 Posted: Tue Dec 17, 2019 2:06 am 
Gosei
User avatar

Posts: 1753
Liked others: 177
Was liked: 491
This page doesn't exist anymore. Does anyone know if such a software exists now? Or if there is an ongoing project?

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #3 Posted: Tue Dec 17, 2019 4:41 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
I rather expect that neural networks have made software detection of cheating more difficult than before. In chess, anyway, one thing that distinguished computer play from human play was the deep calculation of variations. Chess engines, and the human cheaters that used them, would make plays that didn't make much sense to human players unless they were based upon deep calculation. The same cannot be said of neural network go bots. In fact, one way to tell a bot is that it misreads a ladder! :lol: This is well known, so a cheater is unlikely to have the same tell. There are other things that bots do that you could program a computer to find. They like to invade on the 3-3 point, they play new joseki and avoid some of the old joseki, they don't pincer very much, they don't extend on the side or make splitting plays on the side very early in the game, etc. The problem with detecting cheaters is that all of these are learnable by humans. And a suspected cheater may be playing better than he or she did before, because of learning those things. It is not easy to detect cheating by looking at game records.

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #4 Posted: Tue Dec 17, 2019 6:53 am 
Judan

Posts: 6087
Liked others: 0
Was liked: 786
Bill Spight wrote:
In chess, anyway, one thing that distinguished computer play from human play was the deep calculation of variations.


How so? Is there no theory in chess that determines correct moves without deep tactical reading? I know little about chess but have heard that there is such theory e.g. for some classes of endgame positions. Are the players punished for knowing and applying excellent theory?

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #5 Posted: Tue Dec 17, 2019 7:01 am 
Gosei

Posts: 1590
Liked others: 886
Was liked: 527
Rank: AGA 3k Fox 3d
GD Posts: 61
KGS: dfan
RobertJasiek wrote:
Bill Spight wrote:
In chess, anyway, one thing that distinguished computer play from human play was the deep calculation of variations.
How so?
In complex positions, computers can calculate deep tactical variations much more efficiently and accurately than humans can. Often a chess engine can find a tactical move in a short amount of time although it would take an hour of reading for a human to verify its soundness. If a human consistently plays such moves quickly, it is a sign of possible cheating.

Quote:
Is there no theory in chess that determines correct moves without deep tactical reading?
There is plenty of opening theory, where excellent moves have been verified "offline". They are basically whole-board joseki. All players have memorized some fraction of this theory to a greater or lesser extent.

Also, as you note below, there is a corpus of endgame theory, where people have found rules for correct play that require a lot less thinking than brute-force calculation.

Quote:
I know little about chess but have heard that there is such theory e.g. for some classes of endgame positions.
Correct.

Quote:
Are the players punished for knowing and applying excellent theory?
No.

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #6 Posted: Tue Dec 17, 2019 7:26 am 
Lives with ko

Posts: 131
Liked others: 1
Was liked: 18
Rank: AGA 4 dan
Universal go server handle: telegraphgo
RobertJasiek wrote:
Bill Spight wrote:
In chess, anyway, one thing that distinguished computer play from human play was the deep calculation of variations.


How so? Is there no theory in chess that determines correct moves without deep tactical reading? I know little about chess but have heard that there is such theory e.g. for some classes of endgame positions. Are the players punished for knowing and applying excellent theory?


Yes, there are natural moves in chess - the equivalent of shape moves. The difference is that chess engines quickly and commonly find unnatural moves - bad shape moves - that are are only good if deep tactical variations are working. It's very easy to distinguish even super-Grandmaster play from computer play in relatively short time control.

Chess is in many ways more discrete than Go, in that calculation in Go typically makes conclusions that some moves are slightly more efficient, and calculation in chess typically makes conclusions that some moves are just outright unplayable. So it's probably easier to find out whether a computer did someone's homework in chess than in Go.

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #7 Posted: Tue Dec 17, 2019 7:29 am 
Honinbo

Posts: 10905
Liked others: 3651
Was liked: 3374
dfan wrote:
Robert Jasiek wrote:
Is there no theory in chess that determines correct moves without deep tactical reading?
There is plenty of opening theory, where excellent moves have been verified "offline". They are basically whole-board joseki. All players have memorized some fraction of this theory to a greater or lesser extent.


Going back to at least the 1980s, chess engines made use of opening books rather than try to calculate the best opening moves themselves. But now, IIUC, AlphaZero raised questions about the Queen's Indian Defense that Stockfish used against it, and rekindled interest in the Berlin Defense. And in the past year haven't neural network engines made contributions to chess opening theory?

dfan wrote:
Also, as you note below, there is a corpus of endgame theory, where people have found rules for correct play that require a lot less thinking than brute-force calculation.


Around 25 years ago I met a guy who had written an endgame analysis program that had found an endgame mate in something like 225 moves! :)

Edit: Those are chess moves, so that's like 449 go moves. ;)

_________________
The Adkins Principle:
At some point, doesn't thinking have to go on?
— Winona Adkins

Visualize whirled peas.

Everything with love. Stay safe.

Top
 Profile  
 
Offline
 Post subject: Re: Cheating 1.0
Post #8 Posted: Tue Dec 17, 2019 8:20 am 
Gosei

Posts: 1590
Liked others: 886
Was liked: 527
Rank: AGA 3k Fox 3d
GD Posts: 61
KGS: dfan
Bill Spight wrote:
dfan wrote:
There is plenty of opening theory, where excellent moves have been verified "offline". They are basically whole-board joseki. All players have memorized some fraction of this theory to a greater or lesser extent.
Going back to at least the 1980s, chess engines made use of opening books rather than try to calculate the best opening moves themselves.
Yeah, opening play demands slightly different heuristics from middlegame and endgame play, and of course there had already been a huge amount of human opening analysis done in the past, so it was easiest to hardcode it all in a database than try to make engines rederive it. To clarify, when I said "verified" above, I didn't necessarily mean by computer.

Quote:
But now, IIUC, AlphaZero raised questions about the Queen's Indian Defense that Stockfish used against it, and rekindled interest in the Berlin Defense. And in the past year haven't neural network engines made contributions to chess opening theory?
I'm a little out of touch, but I think people are indeed more interested in having engines play the whole game these days; as you note, the Zero engines are already playing the whole game from scratch. I'm not familiar with the examples you name, but the Berlin Defense regained its popularity in 2000 when Kramnik used it to great effect in taking the world championship crown from Kasparov, and has been pretty hot ever since (it was one of the most popular top-level openings even before AlphaZero).

Quote:
Around 25 years ago I met a guy who had written an endgame analysis program that had found an endgame mate in something like 225 moves! :)

Edit: Those are chess moves, so that's like 449 go moves. ;)
Yeah, they can get pretty ridiculous. Here is a list of long ones. Any mate of this length cannot be executed using general principles so they're pretty much not achievable by humans. (Also, the rules of chess still allow a player to claim a draw if there have been no captures or pawn moves in the last 50 moves (100 ply), so even a computer couldn't get to the end of it in practice.)

Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 8 posts ] 

All times are UTC - 8 hours [ DST ]


Who is online

Users browsing this forum: Bing [Bot] and 1 guest


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group