Slashdot Mirror
Detecting Chess Cheats Taxes Computers
First time accepted submitter jeffrlamb writes "Cheating in live chess matches — fueled by powerful computer programs that play better than people do, as well as sophisticated communication technologies — is becoming a big problem for world championship chess. Kenneth W. Regan is attempting to construct a mathematical proof to see if someone cheated; the trouble is that so many variables and outliers must be taken into account. Modeling and factoring human behavior in competition turns out to be very difficult."
I'd just use the CoD system for cheat detection. If they beat me, they cheated. Simple enough.
RTFA, you can use a computer to come up with a better move than you would have on your own.
Check their pockets and make them play in a giant Faraday cage! In a room with only them and an impartial referee. No outside influences, and nobody else to give signals or otherwise interfere.
Chess is obsolete then. Better to pick a game where people can still beat computers.
Give me Classic Slashdot or give me death!
...was hard to read.
None of us know everything. Therefore we're all naïve.
It's cheating in the same sense that using a dictionary in Scrabble is considered cheating if you agreed to no dictionaries before hand - not trying to use qzjkh as a word. In this case, rather than consulting a dictionary, they're consulting a computer to come up with a better solution than they could come up with on their own.
How about assraped by Big Blue. Much more frightening concept. "I'm helium cooled, biatch!"
The world's burning. Moped Jesus spotted on I50. Details at 11.
Technically speaking, chess does have an enormous, yet finite amount of moves. Eventually computers will be able to calculate every possible outcome within a few seconds. Why not just give up on chess and play games that computer's can't play (or at least play like humans)?
Sure, but some moves are characteristically human (very unlikely for a computer).
If you want to read some detailed thoughts on this, see anything written about whole Deep Blue vs. Kasparov thing (especially game 2).
I don't see how detecting chess can cheat the taxes computers. Our detecting chess should be none of the taxes computers business, IMO.
More of the usual bitching I've come to expect from taxes computers. Big complainers, them.
Assume I am a human who, before starting a game, has learnt and will apply the algorithm a computer uses to play chess.
I am then indistinguishable from someone who is cheating by consulting a computer executing this algorithm.
Therefore there is no mathematical proof to determine whether someone has cheated.
This is in the realms of computability 101.
The article doesn't say anything about mathematical proof of whether someone cheated. At the moment he seems to simply be running the decision points (moves) of a game under suspicion against both historic games (to see if the player is playing significantly above their "normal" rating) and against a single computer chess program (to see if the competitor's moves have unusually high correlation with the moves the computer would make). All of this provides evidence of cheating (or lack thereof, as noted in the article where some grandmasters were found to just be playing unusually poorly rather than their competitor cheating), but in no way does it constitute mathematical proof of cheating. Mathematical proof would suggest that there is absolutely no uncertainty, which is clearly not the case.
Already done. "Aided Playing" is called Advanced Chess.
"Unaided Playing" just gets back to the article. The sneaky part is that you don't need to be a moron playing GM moves for an event of Cheating. There was a couple of stories a while back in which GMs only needed one key decision such as "Go for the Win or Keep the Draw" and their own talent was the rest.
My first Journal Entry ever, in 8 years! http://slashdot.org/journal/365947/aphelion-scifi-fantasy-horror-poetry-webzine
I, for one, welcome having an Aided Olympics where sprinters are allowed to use motorcycles and lasers.
Or a football league where the players are allowed steroid use? Sure would make things interesting.
If you play as white, have a rook or a queen on line 7 threatening and keeping the black king locked on line 8. If the AI starts doing crazy shit then discard the whole thing as junk.
I have seen this pattern emerge in every chess-AI I have played. There is this one move where it seems it cannot comprehend anything anymore and starts pushing pieces at you to draw out time.
just use captchas to make sure the content is generated by a person
Are they aiming the computer to detect only if they had outside help from a computer, so that outside help from a human chess player would not be detected?
Learn to love Alaska
If that's the case, just have the computer play Computer vs Computer.
If they created another league for computer-aided play, they would still be trying to eliminate cheating from the non-computer-aided play league. So that doesn't solve anything.
And, who exactly is shunning technological enhancements? How much press have human vs computer matches gotten? And how frequently do people play chess against computers?
Computers in chess is not really a new thing, and it certainly isn't shunned from chess in general. But, this is a league that is meant to test one player against another player, not one player's team and tech vs another player's team and tech. This league, and any other leagues that decide they don't want technology aiding players, have no reason to change that position if they don't want to. It's their league, they are playing a game, there is no right or wrong, there is simply what the people involved prefer. If a majority of players in the league wanted the technology, they would probably allow it.
1 (short ton / firkin) = 89.1432354 slugs / keg
In smallish events (say, under 16 players) at the most elite levels, cheating detection is not difficult. Danialov's Toiletgate accusations aside (translation: Grandstanding,) on a small scale it is easy to detect cheating:
1) All players are completely segregated and in a sterile environment during play.
2) All players can be metal-detected before play - absolutely no electronics not provided by the tournament allowed.
3) Any 'private' areas not under observation (toilet area,) hook up broadcast frequency counter(s) in the room, such that any elecronics usage in the room trips it.
4) Any live transmissions are broadcast delayed and spectators not allowed in the room. (Can't give advice if you can't see the board.)
Simple.
Now, something like the Olympiad, different story to police (or set up similar conditions.) But, then again, that's not "World Championship" chess - just elite.
(Or we can keep doing the best job we can and that is actually pretty good.)
So check the players for technology, and block outside communications. No algorithm needed.
Or the same...
There are two types of people in the world: Those who crave closure
Football would be far more entertaining if they mined the endzones...
"He's running! 30, 20, 10 KABOOOM!" Ohhh too bad, 2nd down.
Do not look at laser with remaining good eye.
Not necessarily, but it does make it profitable!
Is it just my observation, or are there way too many stupid people in the world?
Pshhh.... qzjkh is a perfectly cromulent word.
Although I am not a Grand Master of Chess. I can still hold my own. Playing online these days is not like the real situation. Visualizing, anticipating, seeing the expressions on the opponents face, watching his eyes, and facial quirks. All these play a part in the strategy. ONLINE CHESS GAMES are a farce. Though fun, and challenging at times. It's nothing like the real situation.So quit griping about online chess games. MAKE an appointment and do it LIVE on a green and white map with ivory pieces.
Mining the endzone is silly, by that time it's too late. Much better to mine the last few yards instead.
Whether it's 'allowed' now or not, it happens.
Need a Python, C++, Unix, Linux develop
put the players in an environment where outside input is impossible.
Each player goes into a separate black box faraday cage. They get a touch screen display of the board the opponent is looking at. There are 3 video cameras, wathing the board, player and reverse views. The only signals going through the faraday cage are the video feeds and the board control interface.
The video feeds are out only so no problem there. The control to update the opponents move is the only feed in and the only hole in the system that needs hard monitoring.
Computer vs. human is totally different from computer + human. One is adversarial and ends up with people being made to feel small as computers brute force their way into a win using techniques that would never ever work for a person. The other allows playing that's better than either a human or a machine could manage on their own.
Need a Python, C++, Unix, Linux develop
That would bring a new meaning to the "red zone."
The classic remark by Dreyfus, "No computer can play a decent game of chess", has been inverted. Today's commercial chess programs, running on ordinary desktop machines, or even laptops, can beat any human. No grandmaster has won a tournament against a chess program since 2005. Pocket Fritz 4 on a phone now plays at the grandmaster level.
Hence the cheating. About once a year, a major chess player is caught cheating.
It turns out that, even at the grandmaster level, about 1 human move in 10 is clearly suboptimal. So, one computers got close to the grandmaster level, they could beat humans just by not making mistakes.
I interviewed for a position at Buffalo, and I had dinner with Ken Regan there. Fascinating guy, with a lot of varied interests and a lot of depth. He had some interesting stories to tell about alledged cheating at chess.
Does that mean i cant play because of my cerebal implants?
Then read the summary. It's clear enough to suggest humans cheating by having a computer make their move.
RTFS, they intend to "construct a mathematical proof" to show that a given move, or number of moves, indicates cheating.
This is impossible to prove because it's always possible that the human made those moves on his own. By the same logic that you can assume a human player can only go so deep in the search tree, you can't assume a human player to arrive at a move solely by use of an optimal or deterministic process. A meatbag can see any valid move and decide to play it for any reason. You can't mathematically prove cheating unless you see them cheating. For all you know the player is just lucky,.
So check the players for technology, and block outside communications. No algorithm needed.
What if the "outside communication" is scratching your left ear when you want to know whether to pincer or castle? And an observer responds by touching their tie after they've run the simulation? It's harder to detect than simply checking for electronics on the body.
Parse error: Detecting Chess Cheats Taxes Computers
Parse error: (Detecting) Chess Cheats Taxes Computers
Parse error: (Detecting Chess) Cheats Taxes Computers
Parse error: Detecting (Chess Cheats) Taxes Computers
Parse error: Detecting Chess (Cheats Taxes) Computers
Parse possible: (Detecting Chess Cheats) Taxes Computers
Parse possible: (Detecting Chess Cheats) Taxes Democrats
Parse SUCCESSFUL: (Detecting Chess Cheats) [consumes] Computer [resources]
There are tournaments for that. The best chess programs are at a much higher rating than the best human players. They rate chess programs by playing against eachother.
What if the "outside communication" is scratching your left ear when you want to know whether to pincer or castle?
That doesn't matter if the venue itself is sealed against outside communication, which would prevent anyone in the audience communicating with a computer system outside and relaying moves to the contestant. Deep Blue and its descendants aren't exactly something a spectator could hide in their coat.
If cheating occurs in a sealed room, the judges can be sure that it's collusion between two people and I'm sure they have a great deal of experience with that.
If God forks the Universe every time you roll a die, he'd better have a damned good memory.
not trying to use qzjkh as a word
You appear to have spelled jozxyqk incorrectly.
It's easy to do, since finding it in the dictionary is usually such a pain.
If God forks the Universe every time you roll a die, he'd better have a damned good memory.
When I was young... (yes keep off the lawn please!)
I participated a school-chess evening, and drove my contenders mad... just searching for reasonable valid options, not having a strategy, end-play ect., they were thinking "why the f**k does he do that? he has to have some meaning for that stupid move".
Eventually I got 7th in the pool. (of 7) so my strategy did not work. but had a good evening.
Okay, that's a fair attempt!
I think it will also pose a challenge to computers since they can't just raw search it like chess, but to me that's that limitation on the programming side, heuristics.
My first Journal Entry ever, in 8 years! http://slashdot.org/journal/365947/aphelion-scifi-fantasy-horror-poetry-webzine
well the thing is... if you want to play chess against a computer, you can. Why have a league where humans just act as move proxies for computers? If all I did was input my oponennets moves and make the moves I was told to make, then, am I even playing the game? Or am I just taking credit for the work of the people who created the chess program?
If you suggested an AI vs non-AI league, where competitors could pit their AIs against eachother, then yes, absolutely. Or even a league where people pit their AI players against real ones... but... to have a person stand as a proxy for an AI and then try to claim credit as a player....that seems absurd to me.
"I opened my eyes, and everything went dark again"
Indeed. Klingon's allowed in Scrabble, right?
Random Thoughts From A Diseased Mind (Not For Dummies)
He tried to find articles on the subject, but turned up nothing. “It is one of those situations that it is hard to believe that this hasn’t already been covered in the literature,” he said.
I'm not criticizing Kenneth W. Regan for the way in which his work was reported in the popular press, but Matej Guid and Ivan Bratko have had a couple of articles published in which they evaluate past world champion chess players with computer programs. Their ICGA Journal articles aren't free to read, but ChessBase.com has articles (which I haven't read) based on those journal articles here and here.
My work isn't closely enough related to that of Regan and Guid/Bratko that I know the politics involved: sworn enemies? friends? never heard of each other? But if it's "never heard of each other", they should talk.
Oblig SNL reference: the All-Drug Olympics. http://www.hulu.com/watch/124975/saturday-night-live-update-all-drug-olympics
I could see using the AI as an assist rather than a replacement for reasoning.
For example, pulling up a database of similar positions, the moves made and the results, and overlaying that on the board by color coding certain moves in certain ways. Or the part that computes whether a particular branch of the tree is more likely to lead to a win overlaying it's information on the board in a similar fashion. Or the player putting in a couple of test moves he or she thinks are particularly interesting and seeing what the possible outcomes are.
A player may have knowledge of his opponents algorithm or psychology that the computer lacks. The computer may point out that in a couple of moves there is likely to be huge pressure in a certain area that the human missed, but the human might figure out the move that breaks that up when the computer couldn't.
I want interfaces that augment human decision making, not supplant it. Computers are very good at certain kinds of reasoning. People are very good at other kinds of reasoning. Lets try to combine the two to make a greater whole instead of having one or the other.
Need a Python, C++, Unix, Linux develop
Pocket Fritz 4 achieved a higher Elo rating than any human, and that was on a PocketPC in 2009. And mobile hardware has evolved at an amazing pace since then.
Pshhh.... qzjkh is a perfectly cromulent word.
I think you mean "kwyjibo."
Jesus was all right but his disciples were thick and ordinary. -John Lennon
Or the same...
Hardly. Right now it's a very delicate balance of taking enough performance enhancing drugs to give you the edge, but not enough that it's too obvious and trying to choose the right drugs to not get detected. If they removed all obstacles then things would be much more interesting!
I get that reference. Let us congratulate each other on our taste!
You try pulling something like that at the olympic games and tell me how that works out for you...
I want interfaces that augment human decision making, not supplant it. Computers are very good at certain kinds of reasoning. People are very good at other kinds of reasoning. Lets try to combine the two to make a greater whole instead of having one or the other.
Hmmm. And I want strong AI. Not sure that the distinction you are making between computers and humans is valid. I would hope that there is only one kind of reasoning in the universe, dude.
Computers handle chess by projecting out the possibilities far into the future. Humans do this too, but there is a lot of evidence to show that humans also pattern match against previously seen positions in a very sophisticated way that computers don't do yet.
Currently, a lot of how Google is 'intelligent' is from slurping enormous quantities of data and doing vast statistical analysis of these sets and predicting likely outcomes.
Computers tend to do things this way. People tend to do things by a more sophisticated and detailed analysis of smaller sets of data as well as the ability to correlate ideas and concepts that seem totally unrelated and would never make it into the same dataset in a computer-run analysis.
So yes, I think there are two different approaches to reasoning. It may be that computers can be made just as good as what we do today. But they are not currently, so it makes a great deal of sense to find ways to intelligently combine the two kinds of intelligence. Have the computers provide us the results of there analysis in a way we can easily integrate into how we think about problems.
Need a Python, C++, Unix, Linux develop
You've obviously never gone up against David Copperfield.
Here's my problem of Kasparov losing to IBM: the computer was totally programmed specifically with his games. Now, granted, it was programmed with lots of other games also. In a grandmaster tournament like the U.S. Open, the players (pretty much all Grand Masters and International Masters) are in a round-robin tournament. I want to see IBM's software playing a different GM every day from a field of 20 grandmasters and see how it scores.
When you sympathize with stupidity, you start thinking like an idiot.