Date 2012-02-05 08:59
> sure, you can be an editor, lab assistant, secretary, or investor.
>
> Or a chemist, or a software engineer, or a million other things. Do you think only doctors are designing medical imaging devices? Pharmaceuticals? It doesn't work that way.
>
> but i would definitely trust a medical doctor or medical scientist over a random person.
>
> This is irrelevant. No doubt more than 99.99% of people who receive medical treatments have no clue who came up with them.
>
> also in scientific research you cannot copy other people's experimental values unless you give a proper citation.
>
> Wrong. There is no law requiring attribution in scientific research, although this is required for inclusion in peer reviewed journals.
>
> so it's ok to just take all these evaluation values that fabien took the time and effort to discover?
>
> Rybka doesn't use the same 'evaluation values' as Fruit. This myth has been thoroughly debunked. Feel free to post some examples if you feel otherwise.
1) All this scientific analogies has nothing to do with the rybka issue. Rybka is not part of a scientific project.
2) VR already acknowledge the Fruit influence, so it is moot anyway.
3) If we applied standards generally used in science, the ICGA should be embarrassed by the way they conducted the investigation.
Miguel
Yes, of course you are right. Rybka is not a science project, it is a commercial endeavor, and of course there is no requirement for attribution, although Vas did credit Fruit as a source.
![[us]](/mwf/v12/flags/us.png "United States")
"banned for life" says, "The DT crew almost certainly had better political skills than Vas. I see this whole affair as a political failure more than anything else." so you agree that the lack of a point-by-point rebuttal of the icga report by vasik is not a persuasive stance? how convenient for vasik to say, "i forget", "i don't have the source anymore", and "the pst values may have been from fruit, or it may not have been" (paraphrase).
schroder says, "I think you will find no denial in the VII camp about the pre-Rybka's. What does that tell you?" i think this quote is important. hyatt won his arguments about crafty. this gives more credibility to hyatt. also, was the reverse-engineered code of 1.6.1 accurate, but the reverse-engineered code of 2.3.2a false?
hyatt says "there's a swimming pool of evidence" (paraphrase). i have not seen the full scope of the evidence nor would i understand. i already said that i'm a layman, but my opinion is that hyatt has defended himself, corrected other posters, and given specific coding examples.
banned for life says "I have challenged numerous people to place supposedly copied code side by side here so that we can examine it." apparently there's a disconnect here. banned for life has not seen the reverse-engineered code. banned for life is admitting ignorance. it seems the icga jury had access to this but not the posters here. is the download link to this code available somewhere?
schroder says, "I think you will find no denial in the VII camp about the pre-Rybka's. What does that tell you?" i think this quote is important. hyatt won his arguments about crafty. this gives more credibility to hyatt. also, was the reverse-engineered code of 1.6.1 accurate, but the reverse-engineered code of 2.3.2a false?
hyatt says "there's a swimming pool of evidence" (paraphrase). i have not seen the full scope of the evidence nor would i understand. i already said that i'm a layman, but my opinion is that hyatt has defended himself, corrected other posters, and given specific coding examples.
banned for life says "I have challenged numerous people to place supposedly copied code side by side here so that we can examine it." apparently there's a disconnect here. banned for life has not seen the reverse-engineered code. banned for life is admitting ignorance. it seems the icga jury had access to this but not the posters here. is the download link to this code available somewhere?
> schroder says, "I think you will find no denial in the VII camp about the pre-Rybka's. What does that tell you?" i think this quote is important. hyatt won his arguments about crafty. this gives more credibility to hyatt. also, was the reverse-engineered code of 1.6.1 accurate, but the reverse-engineered code of 2.3.2a false?
This whole thing has nothing to do with winning or losing, but truth finding.
And the truth is there is NO Fruit COPIED CODE in Rybka.
so you agree that the lack of a point-by-point rebuttal of the icga report by vasik is not a persuasive stance? how convenient for vasik to say, "i forget", "i don't have the source anymore", and "the pst values may have been from fruit, or it may not have been" (paraphrase).
No, I don't agree that a point by point rebuttal would have done Vas any good. And for those of us that exchanged posts with him on a regular basis, his not reading the report or losing his source wasn't really a surprise. I still think it might be on the back of a pizza box somewhere...
schroder says, "I think you will find no denial in the VII camp about the pre-Rybka's. What does that tell you?"
There has never been any argument about the pre-release Rybkas. They contain lots of Crafty. But R1B has zero Crafty in it and zero Fruit.
i think this quote is important. hyatt won his arguments about crafty.
No, he certainly didn't. Hyatt has been claiming for years that Vas was using his rotated bitboard code in released versions Rybka. This was false from R1B to the present. There has never been any Crafty in released versions of Rybka.
my opinion is that hyatt has defended himself, corrected other posters, and given specific coding examples.
Ed, Chris and others have shown that there are no coding examples showing copied Fruit in Rybka.
banned for life says "I have challenged numerous people to place supposedly copied code side by side here so that we can examine it." apparently there's a disconnect here. banned for life has not seen the reverse-engineered code. banned for life is admitting ignorance. it seems the icga jury had access to this but not the posters here. is the download link to this code available somewhere?
The ICGA report on Rybka is available from many sources. There is a lot of purported copying of Fruit code alleged in the document, but in each case when you look at the examples carefully, their is no copying to be found.
No, I don't agree that a point by point rebuttal would have done Vas any good. And for those of us that exchanged posts with him on a regular basis, his not reading the report or losing his source wasn't really a surprise. I still think it might be on the back of a pizza box somewhere...
schroder says, "I think you will find no denial in the VII camp about the pre-Rybka's. What does that tell you?"
There has never been any argument about the pre-release Rybkas. They contain lots of Crafty. But R1B has zero Crafty in it and zero Fruit.
i think this quote is important. hyatt won his arguments about crafty.
No, he certainly didn't. Hyatt has been claiming for years that Vas was using his rotated bitboard code in released versions Rybka. This was false from R1B to the present. There has never been any Crafty in released versions of Rybka.
my opinion is that hyatt has defended himself, corrected other posters, and given specific coding examples.
Ed, Chris and others have shown that there are no coding examples showing copied Fruit in Rybka.
banned for life says "I have challenged numerous people to place supposedly copied code side by side here so that we can examine it." apparently there's a disconnect here. banned for life has not seen the reverse-engineered code. banned for life is admitting ignorance. it seems the icga jury had access to this but not the posters here. is the download link to this code available somewhere?
The ICGA report on Rybka is available from many sources. There is a lot of purported copying of Fruit code alleged in the document, but in each case when you look at the examples carefully, their is no copying to be found.
Zach's report is available here:
http://icga.wikispaces.com/file/view/ZW_Rybka_Fruit.pdf
That report has source code from Fruit, and RE'd code from Rybka 1.0 beta side by side for various pieces of the pawn evaluation and such.
Mark's report is available here:
http://icga.wikispaces.com/file/view/RYBKA_FRUIT_Mar11.pdf
His contains additional information, plus the comparison of the fruit and Rybka 2.3.2a evaluation...
People try to discredit both, when it is painfully obvious they have not even read either.
http://icga.wikispaces.com/file/view/ZW_Rybka_Fruit.pdf
That report has source code from Fruit, and RE'd code from Rybka 1.0 beta side by side for various pieces of the pawn evaluation and such.
Mark's report is available here:
http://icga.wikispaces.com/file/view/RYBKA_FRUIT_Mar11.pdf
His contains additional information, plus the comparison of the fruit and Rybka 2.3.2a evaluation...
People try to discredit both, when it is painfully obvious they have not even read either.
Baloney. Both reports have been thoroughly discredited. And the adverse evidence is continuing to pile up. The Dalke analysis is just the latest attack on your sinking ship.
The dalke analysis is really superficial and uses some bad examples. Too much O/S flavor to his comments, where this is a completely different branch of CS...
If chess engines are a branch of CS, they are an incredibly insignificant one. That goes doubly so for engines that are 400 Elo back in the pack.
Chess engines fall under the area of computer science called "artificial intelligence."
Nice to see the insults keep coming. And your credibility keeps dropping...
Nice to see the insults keep coming. And your credibility keeps dropping...
Sorry, chess engines lost their allure to AI people a long, long time ago.
Sorry, but they haven't...
Yeah, right. You are clueless as usual.
Says a computer science AI expert? Read any recent AI books? Taken an AI class? Know any people working in AI / game theory? Didn't think so...
Here is Tim Love's assessment, which is pretty much the norm for non-dinosaurs like you.
Computers and Chess
Early Artificial Intelligence (AI) researchers were interested in Chess - it required calculation but it also had an element of creativity and intuition. Programs were soon developed using ideas introduced by Shannon and many others to more efficiently prune the move trees. As computing power increased, these programs became capable of beating the best human players. Deep Blue (using specialised hardware) played the world champion Kasparov in 1996, winning a game but losing the match. In 1997 it won a rematch, though this may have had more to do with Kasparov's approach than an improvement by Deep Blue. Even without specialised hardware programs like Fritz running on standard PCs can complete for first place in national championships (Holland 2000, for example). So chess is an AI success story - one of the few early dreams which have come true.
In a way, however, chess has been an AI disappointment. The above programs tend to have a fairly simple static analysis routine. They gain their power by number-crunching through as many positions as they can. They don't "plan", or "learn" in the way that the AI pioneers had expected to be necessary, and their development hasn't led to ideas that have been of wider use.
But there is another parallel strand of chess program development. Botvinnik (ex-World Champion and an Electrical Engineer) amongst others devoted time trying to give computers an "understanding" (in the human sense) of chess positions. This approach has fallen into neglect - it hasn't produced powerful chess programs - but now with more powerful computers and programming techniques it might be time for a revival. Benefits include
Improved playing ability - Though using a pure "understanding" approach may not lead to strong chess performance, a hybrid system might work well. As multi-processor facilities increase, more CPU power will become available. Using all this power for number-crunching may lead to diminishing returns. The influence of the "understanding" module may need to vary according to the type of position. Sometimes it might help sort the candidate moves for the number-crunching, sometimes it may determine to move to play by itself. One model might be the way that in some tournaments players are allowed to use computers to augment their play.
Also chess isn't just a matter of playing the "best" move. If the program is losing, or if program notices the opponent being cautious, perhaps it should consider complicating the position or trying to set traps. The ability to define moves as traps and assess their likelihood of success would be useful.
Annotating - Many programs offer an option to annotate a game, but most just evaluate the move played and show the best alternative. Published annotations do much more, pointing out what moves are "forced", when the final, critical error was made, and pointing out when an obvious move isn't good. The ability to define moves as "obvious" would be useful. Also few if any programs are able to explain their behaviour in terms that humans would easily comprehend.
As an exercise it's useful to try to define in computer program terms the criteria for using "!", "!!", "?", "??", "!?" and "?!" in automated annotations. It's not sufficient to say that "!" should be used for a move that leads to a significantly better position than any alternative move would. After all, if someone takes your Queen and you recapture, the recapture is likely to be a significantly better move than the alternatives, but it hardly deserves a "!". Maybe a "!" move is one that in the short term doesn't produce an improved position (indeed, if it's a sacrifice it produces a worse one) but in the long term does improve the position.
When the value of a move changes suddenly as the depth of analysis increases, I think one could say that the move is in some sense "interesting" - oversights, traps and sacrifices fall into this category.
Training - Computers will be more useful as training tools if they play more like humans - some programs are much better at passing the "Turing Test" than others (by the way, Turing was a keen though not very good player and is credited with having written the first computer chess program in 1950). Such programs may also be able diagnose human weaknesses and offer appropriate remedial exercises.
Psychology insights - Chess has been used as a tool by psychologists. The number-crunching approach has provided few insights.
Computers and Chess
Early Artificial Intelligence (AI) researchers were interested in Chess - it required calculation but it also had an element of creativity and intuition. Programs were soon developed using ideas introduced by Shannon and many others to more efficiently prune the move trees. As computing power increased, these programs became capable of beating the best human players. Deep Blue (using specialised hardware) played the world champion Kasparov in 1996, winning a game but losing the match. In 1997 it won a rematch, though this may have had more to do with Kasparov's approach than an improvement by Deep Blue. Even without specialised hardware programs like Fritz running on standard PCs can complete for first place in national championships (Holland 2000, for example). So chess is an AI success story - one of the few early dreams which have come true.
In a way, however, chess has been an AI disappointment. The above programs tend to have a fairly simple static analysis routine. They gain their power by number-crunching through as many positions as they can. They don't "plan", or "learn" in the way that the AI pioneers had expected to be necessary, and their development hasn't led to ideas that have been of wider use.
But there is another parallel strand of chess program development. Botvinnik (ex-World Champion and an Electrical Engineer) amongst others devoted time trying to give computers an "understanding" (in the human sense) of chess positions. This approach has fallen into neglect - it hasn't produced powerful chess programs - but now with more powerful computers and programming techniques it might be time for a revival. Benefits include
Improved playing ability - Though using a pure "understanding" approach may not lead to strong chess performance, a hybrid system might work well. As multi-processor facilities increase, more CPU power will become available. Using all this power for number-crunching may lead to diminishing returns. The influence of the "understanding" module may need to vary according to the type of position. Sometimes it might help sort the candidate moves for the number-crunching, sometimes it may determine to move to play by itself. One model might be the way that in some tournaments players are allowed to use computers to augment their play.
Also chess isn't just a matter of playing the "best" move. If the program is losing, or if program notices the opponent being cautious, perhaps it should consider complicating the position or trying to set traps. The ability to define moves as traps and assess their likelihood of success would be useful.
Annotating - Many programs offer an option to annotate a game, but most just evaluate the move played and show the best alternative. Published annotations do much more, pointing out what moves are "forced", when the final, critical error was made, and pointing out when an obvious move isn't good. The ability to define moves as "obvious" would be useful. Also few if any programs are able to explain their behaviour in terms that humans would easily comprehend.
As an exercise it's useful to try to define in computer program terms the criteria for using "!", "!!", "?", "??", "!?" and "?!" in automated annotations. It's not sufficient to say that "!" should be used for a move that leads to a significantly better position than any alternative move would. After all, if someone takes your Queen and you recapture, the recapture is likely to be a significantly better move than the alternatives, but it hardly deserves a "!". Maybe a "!" move is one that in the short term doesn't produce an improved position (indeed, if it's a sacrifice it produces a worse one) but in the long term does improve the position.
When the value of a move changes suddenly as the depth of analysis increases, I think one could say that the move is in some sense "interesting" - oversights, traps and sacrifices fall into this category.
Training - Computers will be more useful as training tools if they play more like humans - some programs are much better at passing the "Turing Test" than others (by the way, Turing was a keen though not very good player and is credited with having written the first computer chess program in 1950). Such programs may also be able diagnose human weaknesses and offer appropriate remedial exercises.
Psychology insights - Chess has been used as a tool by psychologists. The number-crunching approach has provided few insights.
![[at]](/mwf/v12/flags/at.png "Austria")
> They don't "plan", or "learn" in the way that the AI pioneers had expected to be necessary, and their development hasn't led to ideas that have been of wider use.
>
Maybe the AI pioneers had wrong expectations. Some of them (Hubert Dreyfus?) believed a computer could never play chess like a GM. According to them, to play at such a high level needs "intuition", "understanding", etc. But these notions are very fuzzy and ill defined, and the number crunching approach has proven that they are not necessary.
Maybe this is the most relevant epistemic value of chess programming for AI: Some (or all) of these "high level" ideas can be successfully simulated by "simple" number crunching.
![[mx]](/mwf/v12/flags/mx.png "Mexico")
I think it's similar to AI in the domain of talking, like in the Turing Test, where the judge talks via text with a human and with a computer and has to guess which one is which. The most successful programs just have a huge database of lines to speak, that are triggered when some pattern is the input, with no code whatsoever in there of true understanding of what is talked about.
The judge can still ask "what is bigger, a cat or the moon?", and while the human understands the question, the computer will just look in its database of contextual answers and say something like "Ugh, I'm allergic to cats, and hate them, let's change the subject." The judge might bite.
I can easily imagine robots from the future advanced enough that can pass as humans, while their coding is no more than table look ups and math algorithms.
The judge can still ask "what is bigger, a cat or the moon?", and while the human understands the question, the computer will just look in its database of contextual answers and say something like "Ugh, I'm allergic to cats, and hate them, let's change the subject." The judge might bite.
I can easily imagine robots from the future advanced enough that can pass as humans, while their coding is no more than table look ups and math algorithms.
The "Turing test" was not just about "natural language discussions." One can define a turing test with a limited communication vocabulary. For example, the person trying to choose which is human and which is computer can only supply chess positions. The computer only supplies the best move. That is a valid Turing test and the challenge is to determine which is which by providing input that somehow differentiates. Say a time limit of 60 seconds for a move. or 1 second per move. And then you analyze the responses to see if one was far more accurate, tactically. Or, in a twisted sense, if one was too good and found a mate in 15 in 1 second... etc...
Worth noting here the only person to win the Loebner prize 2x is our friend Mr. Levy.
http://en.wikipedia.org/wiki/Loebner_Prize
http://en.wikipedia.org/wiki/Loebner_Prize
From your link:
>"...has resulted in some wins that may be due to trickery rather than to plausible intelligence..."
According to them, to play at such a high level needs "intuition", "understanding", etc. But these notions are very fuzzy and ill defined, and the number crunching approach has proven that they are not necessary.
In many other problems, e.g. speaker identification, handwriting analysis, and many others, rule based number crunching hasn't been able to get the job done, and pattern recognition and other techniques much closer to human intuition are necessary. Not coincidentally, these same techniques tend to be useful in solving a broad range of problems.
Even the aspect of chess that should be the most amenable to more general AI techniques, i.e. the evaluation, has retained a simple rule based structure that isn't very similar to that used by top players (with the notable exception of Larry K.).
In many other problems, e.g. speaker identification, handwriting analysis, and many others, rule based number crunching hasn't been able to get the job done, and pattern recognition and other techniques much closer to human intuition are necessary. Not coincidentally, these same techniques tend to be useful in solving a broad range of problems.
Even the aspect of chess that should be the most amenable to more general AI techniques, i.e. the evaluation, has retained a simple rule based structure that isn't very similar to that used by top players (with the notable exception of Larry K.).
And that means? Does he know that Botvinnik's "work" was one giant fraud? :) Hans Berliner debunked "pioneer" in an ICGA journal rebuttal to a paper Botvinnik published in that journal. Chess is, and always has been, "artificial intelligence".
Still mentioned prominently in current AI books.
Still mentioned prominently in current AI books.
Still mentioned prominently in current AI books.
Frequently as an early dead end. No applications outside of chess. The idiot savant of AI...
Frequently as an early dead end. No applications outside of chess. The idiot savant of AI...
Doesn't have to have "an application outside of chess." "chess _IS_ the application. As is Go. Ditto for expert systems. Natural language processing. And the list goes on, except that you have no idea what the field is about...
You're certainly a clueless idiot and a troll. There is a lot of money into AI nowadays. It is not going into chess, it is going into lots of other areas that have general applicability. Chess was a failure. You're just too much of moron to have gotten the message.
It has NEVER gone into chess. Nor other games. What planet are you from? That does not mean it is not a part of the artificial intelligence branch of computer science...
You keep re-defining a term when you get backed into a corner. Funded research in computer chess has never been a big deal. The only such project I know of was the small amount of DARPA money spent by the deep thought guys, and that was not for chess, but for developing an ASIC implementation of Belle to prove the performance benefits of custom ICs.
There IS a moron here. It is not me, however. You need to learn to not venture into deep water when you can't swim. You keep drowning because you don't know what you are talking about...
You keep re-defining a term when you get backed into a corner. Funded research in computer chess has never been a big deal. The only such project I know of was the small amount of DARPA money spent by the deep thought guys, and that was not for chess, but for developing an ASIC implementation of Belle to prove the performance benefits of custom ICs.
There IS a moron here. It is not me, however. You need to learn to not venture into deep water when you can't swim. You keep drowning because you don't know what you are talking about...
You're certainly the stupidest moron I've ever wasted time answering. Here is IBM's take on the matter:
Does Deep Blue use artificial intelligence?
The short answer is "no." Earlier computer designs that tried to mimic human thinking weren't very good at it. No formula exists for intuition. So Deep Blue's designers have gone "back to the future." Deep Blue relies more on computational power and a simpler search and evaluation function.
The long answer is "no." "Artificial Intelligence" is more successful in science fiction than it is here on earth, and you don't have to be Isaac Asimov to know why it's hard to design a machine to mimic a process we don't understand very well to begin with. How we think is a question without an answer. Deep Blue could never be a HAL-2000 if it tried. Nor would it occur to Deep Blue to "try."
"If you go back to HAL in 1968," says Deep Blue development team member Joe Hoane, "2001 came out and a lot of people were introduced to the idea that well, you could have a relationship with a computer. HAL in the movie had a personality and, in 1968, people started to realize that computers are getting interesting, that maybe we've reached another milestone where computers are getting really interesting... solving really interesting problems that we couldn't otherwise solve."
Deep Blue's strengths are the strengths of a machine. It has more chess information to work with than most computers and all but a few chess masters. It never forgets or gets distracted. And its orders of magnitude are better at processing the information at hand than anything yet devised for the purpose.
"There is no psychology at work" in Deep Blue, says IBM research scientist Murray Campbell. Nor does Deep Blue "learn" its opponent as it plays. Instead, it operates much like a turbocharged "expert system," drawing on vast resources of stored information (For example, a database of opening games played by grandmasters over the last 100 years) and then calculating the most appropriate response to an opponent's move. Deep Blue is stunningly effective at solving chess problems, but it is less "intelligent" than the stupidest person. It doesn't think, it reacts. And that's where Garry Kasparov sees his advantage. Speaking of an earlier IBM chess computer, which he defeated in 1989, Kasparov said, "Chess gives us a chance to compare brute force with our abilities."
Deep Blue applies brute force aplenty, but the "intelligence" is the old-fashioned kind. Think about the 100 years of grandmaster games. Kasparov isn't playing a computer, he's playing the ghosts of grandmasters past. That Deep Blue can organize such a storehouse of knowledge -- and apply it on the fly to the ever-changing complexities on the chessboard -- is what makes this particular heap of silicon an arrow pointing to the future.
The worlds of science and enterprise are full of problems with so many variables they can't be solved in real time. A system like Deep Blue that can accelerate solutions by powers of 10 is going to make a difference far beyond the chessboard. (And P.S. - That so much of Deep Blue's innards are "general-purpose" industry-standard hardware is good news to any organization faced with a 7-figure problem on a 6-figure budget.)
The way that the PowerPC chips inside Deep Blue work in parallel to break down and solve a chess-board problem is a pretty good analog for the way many scientists, working independently, advance our total understanding of the universe, or genetics...
Or the way business people confront the complexities of, say, running an airline. Figuring THE best way to schedule 570 planes of 25 different types to 150 destinations for best passenger revenue and most efficient fueling, maintenance, crew deployment, and turnaround servicing is a towering problem. On that scale, the difference between a pretty good solution and the best solution is measured in billions.
The shifting complexities of the chessboard are the airline problem in miniature. For computer scientists, chess is a laboratory benchmark. Back in computing's Jurassic age, in 1950, Claude Shannon, the chief architect of information theory, put it this way: "The chess-playing problem is sharply defined, both in the allowed operations and in the ultimate goal. It is neither so simple as to be trivial, nor too difficult for satisfactory solution."
Satisfactory solutions - to problems far beyond the chessboard - are closer than ever before as a result of the research that has gone into the Deep Blue system. And who knows? As more possibilities open before us, some of those science fiction predictions may come true. But it won't be because of any artificial intelligence. It will be because systems like Deep Blue helped us make better use of the real thing.
Does Deep Blue use artificial intelligence?
The short answer is "no." Earlier computer designs that tried to mimic human thinking weren't very good at it. No formula exists for intuition. So Deep Blue's designers have gone "back to the future." Deep Blue relies more on computational power and a simpler search and evaluation function.
The long answer is "no." "Artificial Intelligence" is more successful in science fiction than it is here on earth, and you don't have to be Isaac Asimov to know why it's hard to design a machine to mimic a process we don't understand very well to begin with. How we think is a question without an answer. Deep Blue could never be a HAL-2000 if it tried. Nor would it occur to Deep Blue to "try."
"If you go back to HAL in 1968," says Deep Blue development team member Joe Hoane, "2001 came out and a lot of people were introduced to the idea that well, you could have a relationship with a computer. HAL in the movie had a personality and, in 1968, people started to realize that computers are getting interesting, that maybe we've reached another milestone where computers are getting really interesting... solving really interesting problems that we couldn't otherwise solve."
Deep Blue's strengths are the strengths of a machine. It has more chess information to work with than most computers and all but a few chess masters. It never forgets or gets distracted. And its orders of magnitude are better at processing the information at hand than anything yet devised for the purpose.
"There is no psychology at work" in Deep Blue, says IBM research scientist Murray Campbell. Nor does Deep Blue "learn" its opponent as it plays. Instead, it operates much like a turbocharged "expert system," drawing on vast resources of stored information (For example, a database of opening games played by grandmasters over the last 100 years) and then calculating the most appropriate response to an opponent's move. Deep Blue is stunningly effective at solving chess problems, but it is less "intelligent" than the stupidest person. It doesn't think, it reacts. And that's where Garry Kasparov sees his advantage. Speaking of an earlier IBM chess computer, which he defeated in 1989, Kasparov said, "Chess gives us a chance to compare brute force with our abilities."
Deep Blue applies brute force aplenty, but the "intelligence" is the old-fashioned kind. Think about the 100 years of grandmaster games. Kasparov isn't playing a computer, he's playing the ghosts of grandmasters past. That Deep Blue can organize such a storehouse of knowledge -- and apply it on the fly to the ever-changing complexities on the chessboard -- is what makes this particular heap of silicon an arrow pointing to the future.
The worlds of science and enterprise are full of problems with so many variables they can't be solved in real time. A system like Deep Blue that can accelerate solutions by powers of 10 is going to make a difference far beyond the chessboard. (And P.S. - That so much of Deep Blue's innards are "general-purpose" industry-standard hardware is good news to any organization faced with a 7-figure problem on a 6-figure budget.)
The way that the PowerPC chips inside Deep Blue work in parallel to break down and solve a chess-board problem is a pretty good analog for the way many scientists, working independently, advance our total understanding of the universe, or genetics...
Or the way business people confront the complexities of, say, running an airline. Figuring THE best way to schedule 570 planes of 25 different types to 150 destinations for best passenger revenue and most efficient fueling, maintenance, crew deployment, and turnaround servicing is a towering problem. On that scale, the difference between a pretty good solution and the best solution is measured in billions.
The shifting complexities of the chessboard are the airline problem in miniature. For computer scientists, chess is a laboratory benchmark. Back in computing's Jurassic age, in 1950, Claude Shannon, the chief architect of information theory, put it this way: "The chess-playing problem is sharply defined, both in the allowed operations and in the ultimate goal. It is neither so simple as to be trivial, nor too difficult for satisfactory solution."
Satisfactory solutions - to problems far beyond the chessboard - are closer than ever before as a result of the research that has gone into the Deep Blue system. And who knows? As more possibilities open before us, some of those science fiction predictions may come true. But it won't be because of any artificial intelligence. It will be because systems like Deep Blue helped us make better use of the real thing.
And what does that have to do with whether computer chess is a sub-field of artificial intelligence, which is a branch of computer science? Exactly nothing. DB was always advertised as a "hardware solution to a difficult software problem." Nothing more. But again, I can't find an AI book that does not include a discussion about computer chess, computer go, and even checkers which has been completely solved by the computer...
My statement was:
chess engines lost their allure to AI people a long, long time ago.
Which is fully consistent with statements from Tim Love at Cambridge and Murray Campbell at IBM.
chess engines lost their allure to AI people a long, long time ago.
Which is fully consistent with statements from Tim Love at Cambridge and Murray Campbell at IBM.
Again, what does that have to do with the discussion? Computer chess IS a part of AI. Always has been. Always will be. Game theory is a part of AI. There are many other parts. Not one thing you have provided indicates that computer chess is not AI.
Once again you are showing off your dismal reading comprehension skills.
I have NEVER claimed that chess wasn't once a focus of AI. In fact, my statement chess engines lost their allure to AI people a long, long time ago, clearly indicates the opposite.
I have NEVER claimed that chess wasn't once a focus of AI. In fact, my statement chess engines lost their allure to AI people a long, long time ago, clearly indicates the opposite.
I am "showing off" an understanding of the field of artificial intelligence. Chess is "in it". Always has been. It was more interesting before the computers passed the best humans, but the interest is still there, right along with GO and games including variants of chess...
Wrong. Chess and go are worlds apart.
Generally speaking, artificial intelligence is trying to mimic the processes used by humans when trying to make decisions. Nowadays, computer chess seems to be more "not this" than "this". Humans make chess decisions based on a large amount of pattern analysis combined with some calculations. In computer chess, one tries to come up with terms that are bits and pieces of what we call "pattern analysis", which are also terms that make it into books and articles about normal playing strategy. However, humans also make chess decisions based on a large amount of learning, and the realization that a certain position is "kind of like" a position from a recently analyzed game will redirect the human's thinking, something that no chess programs currently do. Furthermore, I guess that someone who dedicated a program to actually doing this would find that program significantly weaker. Instead, the strategy is search better, prune more, and search unimaginably deep. The extensions idea is somewhat human-based, but we're still primarily dealing with numbers, while real human chess is often, rightly or not, dealing with "feelings" (e.g. Carlsen's recent game against Topalov).
Computers use very different "strategies" than do humans, and it has been shown that the computer strategies are, in general, better for winning chess games. I think that this is less artificial intelligence and more simply an alternative method that works much better than artificial intelligence. I guess it makes an interesting debate, though, as to whether computer chess truly should be considered in the artificial intelligence regime. Whether or not the answer/resolution to this is even significant in a realm any deeper than semantics would also form an interesting debate.
Computers use very different "strategies" than do humans, and it has been shown that the computer strategies are, in general, better for winning chess games. I think that this is less artificial intelligence and more simply an alternative method that works much better than artificial intelligence. I guess it makes an interesting debate, though, as to whether computer chess truly should be considered in the artificial intelligence regime. Whether or not the answer/resolution to this is even significant in a realm any deeper than semantics would also form an interesting debate.
But if computer chess is not part of AI, the same would be true for similar activities.
- Alpha/Beta tree search in general,
- Monte Carlo tree search (as in Go programs),
- pattern recognition (the evaluation function of a chess program has the purpose to make good features to be preferred),
- utility functions and decision theory,
- reinforcement learning (for opening books or positional features),
- precalculation of results and database lookup (as in endgame tablebases), etc.
If one excludes from AI everything that is used or can be used in computer chess, what will remain?
- Alpha/Beta tree search in general,
- Monte Carlo tree search (as in Go programs),
- pattern recognition (the evaluation function of a chess program has the purpose to make good features to be preferred),
- utility functions and decision theory,
- reinforcement learning (for opening books or positional features),
- precalculation of results and database lookup (as in endgame tablebases), etc.
If one excludes from AI everything that is used or can be used in computer chess, what will remain?
> If one excludes from AI everything that is used or can be used in computer chess, what will remain?
You don't exclude everything, you only exclude the domain specific parts. How does knowing that a Rook on seventh rank is good benefits AI?
Each application of AI is implemented in a certain domain.
Can't your argument be made for every single piece of knowledge?
Can't your argument be made for every single piece of knowledge?
Of course not all knowledge is domain-specific.
AI is about emulating 'human intelligence'. Is not "rook on the 7th" something any human chess player has some understanding of? How is doing the same thing a human does "not artificial intelligence".
Rook on the seventh rank is a feature. The problem with chess engines, is that this type of feature must be manually identified and characterized outside the engine, and a specific rule generated to address this particular feature must be generated, once again outside the engine.
Contrast this with say a handwriting recognition system, which may use a very large feature set along with an algorithm for establishing the weights for each feature that are 'learned' during a training session. This approach provides much more generally useful results that are applicable in a large number of fields (and which can even be used to generate a chess evaluation, albeit one that is much too slow to be useful in a competitive engine).
Contrast this with say a handwriting recognition system, which may use a very large feature set along with an algorithm for establishing the weights for each feature that are 'learned' during a training session. This approach provides much more generally useful results that are applicable in a large number of fields (and which can even be used to generate a chess evaluation, albeit one that is much too slow to be useful in a competitive engine).
I had forgotten to make the point in my post last night that I think that some chess programs are "more AI" than others. For example, I see programs strongly based in knowledge like Komodo and Hiarcs as more AI than programs that are not as much so, like Rybka and Houdini. Are alpha/beta and Monte Carlo techniques really AI? I see the first of these as simply doing something that is effective and that the human brain DOESN'T do very well, and the second of these as hard to say, but which has not yet been very successfully implemented into a chess program itself (instead, when using this with chess, it requires a human to interpret the results).
I see the reinforcement learning and pattern recognition as "biggies", but I don't think that opening books really do this, yet--they eliminate unsuccessful lines, but they don't really do anything about making successful ones that SHOULD be successful, i.e. figuring out why a line failed. Something that I would see as a getting toward "truly AI" would be if the various evaluation parameters in a program were tuned based on the results of games. While this might not necessarily make the program play the best chess, it would certainly be getting more toward artificial intelligence. "Hmmm...maybe the double bishop bonus shouldn't be so high..." Instead, this currently has to be done by hand by people like Larry Kaufman based on thousands upon thousands of games.
I don't see endgame database lookup as AI at all--I think this is one of the least AI parts of computer chess. One way of making it "slightly more" AI would be if a program was able to pick from many drawing lines, the one that is most difficult for the opponent--this would be a very significant advance.
I see the reinforcement learning and pattern recognition as "biggies", but I don't think that opening books really do this, yet--they eliminate unsuccessful lines, but they don't really do anything about making successful ones that SHOULD be successful, i.e. figuring out why a line failed. Something that I would see as a getting toward "truly AI" would be if the various evaluation parameters in a program were tuned based on the results of games. While this might not necessarily make the program play the best chess, it would certainly be getting more toward artificial intelligence. "Hmmm...maybe the double bishop bonus shouldn't be so high..." Instead, this currently has to be done by hand by people like Larry Kaufman based on thousands upon thousands of games.
I don't see endgame database lookup as AI at all--I think this is one of the least AI parts of computer chess. One way of making it "slightly more" AI would be if a program was able to pick from many drawing lines, the one that is most difficult for the opponent--this would be a very significant advance.
> For example, I see programs strongly based in knowledge like Komodo and Hiarcs
What makes you think (apart from hype) that Komodo4 is more knowledge based than R4/H2.0?
Nothing--but I believe the hype. :-)
Larry has claimed that he has found many parameters in top engines to be poorly tuned. Have you noticed any differences in this aspect of Komodo, and if so, can we anticipate a stronger Critter in the not too distant future?
> Larry has claimed that he has found many parameters in top engines to be poorly tuned.
Yes, he is probably right. Just note that better tuned weights does not mean _more_ knowledge...
Agreed. More knowledge should be associated with more evaluated positional features, not more accurate weights. I was just hoping that you planned to take advantage of better weights in future releases.
![[gb]](/mwf/v12/flags/gb.png "United Kingdom")
> artificial intelligence,
Human articifical intelligence as a single strain is an erroneous concept in the first place. I think there should be 2 lines of research into artificial intelligence. One that can reproduce representative male intelligence and another to reproduce the female intelligence. Perhaps we should have a poll as to which team people would rather be on or which one is likely to be solved soonest? All male chauvinist remarks ... should make for humerous reading.
PeterG
> You're certainly the stupidest moron I've ever wasted time answering.
Would your argument be weaker if you didn't include this part?
There isn't much argument there. It was just a matter of picking out one or two of the hundreds of articles that discuss why chess has not been considered to be an AI success story.
As far as insulting Bob, Bob insulted me with his very first reply months back and I'm not very good at turning the other cheek. But I can't say I regret being rude to him. I think his goal was to destroy this forum, and he has pretty much succeeded. As far as I can tell, this is the only thing he's good at.
As far as insulting Bob, Bob insulted me with his very first reply months back and I'm not very good at turning the other cheek. But I can't say I regret being rude to him. I think his goal was to destroy this forum, and he has pretty much succeeded. As far as I can tell, this is the only thing he's good at.
Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill