Slashdot Mirror
Chess Improves Machines and Humans Alike
erick99 writes "Chess provides a window into some more arcane philosophical matters. The remainder of this article will focus on two difficult, and interrelated, questions. The first has to do with the nature of reality; the second is about the prospects for human and artificial intelligence in grappling with reality. In both cases, the search for an answer leads through a board game with 32 pieces and 64 squares."
The article has little to do with the game of chess, it is a philosophical piece (it strikes me that invoking religion in a philosophical debate is a bit like invoking Hitler in any other argument...). It's a bit thin too - saying that you can use the same word to describe different things doesn't imply any necessary connection between those things; it could mean we interpret the word based on its context...
I have little time for philosophy: the endless soul-searching and argument over subtle nuance seems pretty meaningless - you can't root an argument in reality when you're debating the existence of reality! Accept that and move on. I happen to agree with Popper about falsifiability, but that's just an opinion...
Perhaps we ought to just accept the universe does exist, then perhaps we can start to do something useful rather than pursue ultimate logical deriveable truths (although I guess the Vulcans got their warp drive first, hmmm)... The greatest breakthroughs in science were made once the ancient Greek philosophic method was turned on its head and transformed into the scientific method we use today. Theory and practice, unified in harmony; either on their own regarded with suspicion - look at cold fusion and string theory...
Simon
Physicists get Hadrons!
A group of chess enthusiasts had checked into a hotel, and were standing in the lobby discussing their recent tournament victories. After about an hour, the manager came out of the office and asked them to disperse. "But why?" they asked, as they moved off.
"Because," he said, "I can't stand chess nuts boasting in an open foyer."
Q. What's the difference between a chess player and a highway construction worker?
A. A chess player moves every now and then.
Which football team has a couple of chess pieces missing?
QPR
Q. What is the difference between a chess player and a couple on a blind date?
A. The chess player mates then chats......
Regards,
(courtesy of Graham Moore)
Q - Which group of women are the best chess players?
A - Feminists. Their opponents begin with King and Queen,
but *they* always start with 2 Queens.
If Chess has these implications, imagine what a good match of GO will do for you! Both man and computer alike! Simple to learn, arcane to master offering a lifetime of fulfillment.
I've read that while computers can offer a credible competition to even a Chessmater, there is no current "go" program that can challenge a true master of that game. Though it's been a while since I've read this, so this may have changed. But this has been a reason why computer logic enthusiasts have been enthralled with this game for many a year.
A little offtopic...but...by the way, while on the topic of Go: did you know that the original selling price of KPT Bryce was determined over a game of go? Eric Wenger (the original developer who based all of the fractal math on the work of Ken Musgrave, originally an aprentice of Dr. Mandelbrot himself) thought that Bryce should be a "Hollywood Tool" and cost over $7000 (back in the early 90's!). But Kai Krause thought it should be a tool to "empower the creativity of the average person" and said the pricepoint should be set at $99.00
So they decided to let a game of Go decide it. Thankfully, Kai won the game!
Quod scripsi, scripsi.
A zen master was once asked, "What is the greatest game ever invented by man?"
He replied, "Chess, of course."
His chela asked, "But, what of Go?"
The master replied, "There was go before there were men."
pandanet.co.jp
"He who would learn astronomy, and other recondite arts, let him go elsewhere. " -- John Calvin, commenting on Genesis 1
I'm sorry... this article feels like an infomercial for pseudoscience. With abstractions like "Does the number 12 exist?" I have to wonder why it made the cut to even appear on slashdot. We could also pretend we're Vulcans and talk about the deflector dish, but it certainly isn't worth slashdot coverage.
"He who would learn astronomy, and other recondite arts, let him go elsewhere. " -- John Calvin, commenting on Genesis 1
Funny, as the current trend in AI research is to eschew abstractions and modeling (referred to as GOFAI - good old fashioned AI) in favor of neural nets and the like. Adherents of embodiment look at chess as exactly the sort of problem stacks the deck in favor of the machines / can't tell us anything interesting about intelligence
Of course, chess is always solvable with sufficient computing power. There's really nothing interesting about it, just an optimized adversarial search tree with some function to evaluate how good board states are, maybe with a table of good endgames tacked on.
Of course, this is not much like the way that humans solve the problem
I personally would be much more impressed with a computer that could play baseball.
Those more interested in the aspects of computers and brute-force calculating power vs. human intuition in games like chess might find this article interesting.
The author predicts that while computers will one day defeat even the greatest chess Grand Masters, they will probably never be able to master the Chinese game of "Go".
I mostly agree, but don't forget about geometry/trig, etc...
It was a strange thing back then when philosophers said "Let's not measure things, not even REAL things, instead lets think of the IDEA of spatial relations". The idea of the line, equations, and all of those other fundamentals we all learn today. It was math in the philosophical sense. (a^2 + b^2 = c^2 ) (or the shortest distance between two points is....)
If that had never happened, if they hadn't stepped back from the drawing table to theorize and philosophize, we wouldn't have the solid mathematical foundation we have today.
So, the same may be said of other philosophies. Stepping back from reality, and thinking about things that seem unrelated may eventually turn out to be the exact opposite.
That explains why I never can beat a computer at chess. Whenever I get better, the computer gets even more better!
1f u c4n r34d th1s u r34lly n33d t0 g37 l41d
I find the game to be not only fun but also rife with philosophical implications. It reinforces certain lessons of everyday philosophy, for instance the importance of trying hard (my games vary widely in quality, depending on effort and attention) and maintaining some humility (just when I think I've gotten good, someone comes along and wipes the board with me).
But then he goes on to make a discussion about platonism that could IMHO be made much better (and would be more interesting) in relation to mathematics.
It hapens that I have just (about two hours ago) written a short essay on how to improve in another board game. What I didn't dare saying there is that you cannot seriously improve in go without trying to improve get an overall positive attitude towards life, somehow trying to be on top of it.
I would certainly have loved to see a chess player's take on that topic. Chess is probably still a little more competitive than go (in the Western culture), and they might well know more about it than we go players do.
The article mentions this interesting invocation of chess in philosophy:
Daniel Dennett's evocation of chess computers in his argument for the compatibility of free will and determinism.
I find it far more interesting than the two the article DOES cover, i.e. whether ideal objects exist and whether computers will out-think humans.
If this comment has any particular point, it's that there are many interesting questions that are probably NOT covered by this article, and this might be an interesting forum to bring some of them up.
This article reads like an article on chess that collided with an article on DesCartes' philosophy.
_____
Thank you.
Chess is also an interesting test case for one of Vinge's paths to superhuman intelligence. Namely, the idea that human/machine interfaces may become so intimate that we will in effect fuse with our technology, becoming superhuman in capability.
Kasparov, for example, has been advocating allowing mixed human/computer teams in "Advanced Chess" tournaments. It seems that the human/machine combination, with the right interface, yields far better chess play than either alone.
Some questions that fascinate me:
Frankly I find these more useful questions than the old human vs. computer debate.
Man, that's deep.
Chess isn't nearly as interesting for A.I. as we once thought it was. Essentially it's a closed, well defined formal system. These sorts of things are relatively easy to deal with, compared to problems like "Write a good essay about the history of chess". We have a pretty good idea how to write a really good chess program, but we have no idea how to even begin to algorithmically write a good essay.
Chess is essentially a math problem. "Real world" problems however are a completely different ball game. We need to answer some very interesting and fundamental questions before we can even begin to build any interesting A.I. (A theory of relevance being one, and the frame problem being another).
On the one hand, chess is a very interesting realm for understanding the realms of human and machine intelligence. It is an interesting microworld with enough complexity that it lacks brute force or close-form solutions. Thus it provides a useful test case for understanding rational action. And blitz chess is useful for looking at reasoning under time constraints.
On the other hand, chess is closed - a King will always be limited to moving one square in any direction. With chess, no new moves, new pieces, new board locations can ever appear. Chess is also certain -- there are no ambiguites in the locations of the pieces. With chess the rules and positions are fully known before hand by the exactly two players who adhere to the constraints of the game.
By contrast, the field of human affairs evolves continuously to create new scenarios, new possible movements, new roles, and new players. Everyday slashdot has articles about the novel activities of people (from scammers using TTY relays to new chipsets to new laws). I would argue that decision making under conditions that are uncertain, open-ended, massively multiplayer, and subject to changes in the rules are a bit different.
They say one must learn to crawl before learning to walk. In some ways, learning about the intelligence required to play chess is like learning to crawl. That even the decision making underpinnings of playing chess is so hard to understand says something about how hard it will be to understand true intelligence in open-ended situations the poeple deal with every day.
Two wrongs don't make a right, but three lefts do.
Today's philosophers of mind are asking the questions that direct AI researchers toward identifying and solving the interesting research problems.
Those of use who have studied and performed research in AI know that "android epistomology" (the study of the space of possible thoughts in an android mind) is a very vibrant and important topic that is widely debated. The term "android epistemology" was first coined by Clark Glymour in a sourcebook on this topic.
Rudolf Carnap was the first to combine propositional logic with natural language to come up with a general philosophy of high-level thought. His ideas were rigorous enough to be considered computer programs, and yet he came up with them in 1928!
Recently, we heard about the Robotic Race, a 150-mile race of autonomous vehicles, where the winner only made it 7 miles. Want to know why the winner didn't get farther? It got a tire stuck in sand, and wasn't "smart" enough to realize that flooring the accelerator wasn't doing any good, so it burned the tire off, right down to the rim. Had it included in its space of possible mental states the idea it could disengage an axle, it could have gotten out of its hole and kept going. It didn't have the "mental capacity" to step back, reflect, and consider an alternative idea.
The question of how we, as humans, are able to adjust our "space of mental thoughts" to external conditions is hardly even addressed in the modern AI literature, and yet it's precisely this kind of question that philosophers identify as an important problem and ask first!
So, we owe philosophy a debt for often framing the correct questions for other to later answer.
It seems rather clear to me that abstracts exist. Obviously not in the physical sense, but they must exist. If they did not, we would have very little basis for calling two similar (but obviously different) objects the same. For example, if I saw two animals and had no abstract of what an animal was, how could I say it was an animal? If one was a dog and the other a cat, how could I differentiate unless I had some preconceived notion of what a dog was? Furthermore, if I saw a species of dog which I had never seen before and had no idea existed, how would I still know it was a dog without some abstract conception of what a dog is? Arguably, an individual thing, such as a particular dog, has potentiality (the potential to exist in reality) while an abstract always exists in reality, on the basis that it needs to physical status to exist. This could be applied to the question of AI and chess as well. Since it would seem practically impossible for any person or machine to hold all the possible (or abstract) variations on a chess game, there must be some way we arrive at 'new' undiscovered ways of playing. I would assume this to be something that chess programs tend not to use, behaving randomly. When faced with a decision, a human will often choose randomly or emotionally, possibly choosing what would seem a poorer choice. A chess program, especially one that is playing a particularly talented human opponent, would likely not suspect such acts, instead 'thinking' the opponent would behave in the most logical way possible. How we could teach computers intuition is anybody's guess.
Every windows user is a sadomasochist.
Even better, GoZen, the game that can only be played by ethereal brain masses on the highest philosophical plane of existance.
Slashdot sucks
An interesting twist on chess is taking a position and attempting to deduce something about what must have occurred in the game previously. For example, has a promotion occurred or not? What must have been white's last move? I don't know whether there exist computer algorithms for solving these sort of problems - a brute force approach would probably be useless. It's possible to construct quite interesting and non-trivial puzzles of this sort. The logician Raymond Smullyan's delightful book The Chess Mysteries of Sherlock Holmes starts with some easy examples and builds up to some really mind boggling examples.
The world is everything that is the case
Chess has proved the value of a brute force approach--even without a lot of AI routines, simply searching the game tree and adding up the value of the men left on the board is a workable algorithm. Good chess programs improve on that significantly with rules to prune the tree search, and further rules to score a board position. That doesn't work so well in Go: There are 361 points on a Go board, with a typical game lasting some 200 moves--an unimaginably large number of game combinations. Worse, there's no easy way to assign a value to a board position once you've brute forced your way through the combinations. The combination of these two factors is one reason why there are no really good Go playing programs, as there are in Chess.
Go is a great game to play on the Internet. You can order all the books you need to get you started, and then you can play on the 'net. There's not bad Go implementations at Yahoo Games, etc., but eventually you will move up to the real go servers like Kiseido or Panda, both located in Japan.
1. This is high-school grade philosophy
Actually that's pretty good by slashdot standards.
I want to know why if I program in some function that determines how good a board is, and the computer goes and tries all possibilities to a certain depth of moves to determine the best move using either a minimax algorithm or something like that, why is this considered A.I.? The computer isn't doing anything I didn't specifically tell it to do.
Wouldn't real AI be writing a program that plays a whole lot of chess and "learns" what makes a board/move good and that's how it decides how to play?
I just don't get why a computer playing exactly how it's programmed is considered AI and not learning anything on it's own (on its own is loose here, if it was specifically programmed to learn I'd still consider it learning on it's own).
For instance, we wrote a Kalah player in a CS class I was in. You know how my team decided how to rank boards, we wrote a program that cycled through thousands of possibilities for the different weightings of each pit and then compared the results when using those weightings. In my head, that's A.I., the computer just decided for itself what the best evaluation function was (albeit we told it how to decide) as opposed to simply using one we hard-coded in and having it search really deeply (which in my mind is not AI at all, just a computer playing a game).
I've always find it a bit amusing that Go enthusiats always have to get their say about how much better their game is (true or false is besides the point) whenever the subject of Chess is brought up. While, Chess enthusiats never have to do the same thing when Go was brought up. I play loads of chess, but I find Go interesting to talk about and won't say things like "Screw GO. Play chess."
:).
A bit like Linux advocates saying "Screw Windows. Run Linux." everything Windows was brought up.
I let the reader draw their conclusions about this statement
(P.S. I run linux both in the office and at home. And I am completely at loss in Windows).
Mode (3) smart-aleck mode. Press * to return to main menu.
Don't want to flame, but the article does seem like a bunch of "pseudo-intellectual" (forgive me for using that phrase) 14 year olds sitting around, playing chess and thinking their minds are advanced. Half way through the article I thought they would break out matrix-like statements saying "there is no pawn."
Seriously, it just sounds like a half-assed book on Hume or such that somehow had pages of "Chess for Dummies" inserted randomly.
Did the Ruy Lopez exist before its 16th-century namesake started playing it? A Platonist might say it did, as part of an abstract set of all possible chess openings. But chess itself has a finite history. The game originated around the seventh century A.D., and its modern rules became standard in the 15th century, not long before Ruy Lopez de Segura was playing. Platonic ideals are normally defined as timeless, yet in this case they seem also to be historically grounded. The world of abstractions seems to depend on our world.
Does that mean that the number Pi didn't exist before it was discovered? It did, Platonism as he refers to in this article at least, is just stating that fact that that number although not defined (hence taken a particular meaning for us humans) has always existed.
Saying that Pi didn't exist before we noticed it is equivalent to saying that the outter most particles in the universe, the ones propelled by the big bang, don't exist since there's no way for us to reach them (they are moving at the speed of light outward).
Perhaps in some sense, all chess moves, positions and games are "out there," but they have a rather limited existence if nobody plays them. Interestingly, it appears physically impossible for any computer or other material entity ever to store complete information about the game. By some estimates, the number of possible chess games exceeds the number of particles in the universe.
Here's one, the number of different pathways a neural signal can take through the brain is WAY higher than the number of particles in the universe... does that mean we can't form some of these because nobody would be able to count them?
Both of these paragraphs don't add anything to the text, IMHO.
Anyone care to tell me otherwise in a logical manner?