< Earlier Kibitzing · PAGE 2 OF 2 ·
|Feb-09-03|| ||Sneaky: Many years ago the ancient Greeks conducted foot races to determine who was the fastest runner at various distances. Some specialized in sprints while others went for endurance. |
This noble sport endured for thousands of years, until one day the TRAIN was invented... and later, the AUTOMOBILE. But did this destroy track and field events? Of course not!!
I am skeptical that computers will become better overall chessplayers than people any time soon, but even if they do, it won't harm chess a bit.
|Feb-09-03|| ||aulero: There was a man that played like a computer for his entire life, but being a human and not a machine occasionally lost games (very few), got tired (just once) or sick (just once) and eventually died (just once!). But, strangely enough, he was neither the greatest tactician, nor the greatest strategist, nor the greatest technician and absolutely not the greatest thinker. He owned only a very special skill: a natural, intuitive and near perfect evaluation of the chess positions.|
The computer play can be reduced to calculation and evaluation: if they are able to defeat the humans (one, two or else ten exceptions mean nothing) is because they surely calculate better, but also because they evaluate better (than Grand Masters!).
I think that the natural evolution of chess is using computers as drivers use cars: so future champion will not be Kasparov, but the pair Kasparov + Deep Junior (or Deep Junior + Kasparov if you prefer), much like the pair Schumacher + Ferrari (or Ferrari + Schumacher as I prefer).
|Feb-09-03|| ||ughaibu: Did he never get drunk? or fall in love? was he any good at backgammon? |
|Feb-09-03|| ||BLD9802: Argull: I do not own a copy of the world's most powerful chess programs (Fritz, Junior, HIARCS, etc.), but I do own a copy of the Sony Playstation version of Chessmaster II, and I admit that it defeats me in most of the games that I play against it because I do not have the necessary skill to defeat it (yet...ha ha!)
(Incidently, like ughaibu, I find it very difficult to get motivated enough to play a serious game with a computer or even over the internet. To me, chess is most fun as a human vs. human activity, which is why I want to play in a good OTB tournament. After all, nobody is watching me play against the machine and I am not going to win a huge prize for beating it, so why waste all that energy?)|
However, the point of my previous statement was that computers are not yet superior to humans at chess simply because our finest, most serious, and dedicated players can still hold their own when playing them. Chess is an ancient game invented by humans, and we only created computers a few decades ago, therefore the burden of proof is on the computers to demonstrate without a doubt that they are superior to us, not vice versa, and so far they have not done that (they can crush less-skilled players, just as less-skilled players can crush less powerful or "cheap" chess computers, but at the present time their best do not seem able to beat our best).
Also, please understand that our comments weren't meant as personal attacks on you, we just don't agree with the idea that the chess computer is superior to the human chessmaster. :)
|Feb-09-03|| ||mdorothy: At first, I really liked your Ferrari + Schumacher analogy, but the more I think about it, the less I like it. The difference is that it takes almost no chess skill to operate a chess computer. I saw an interview with one of the guys that operated Deep Junior in this match. He said that he isn't much above an average chess player. Thus, you wouldn't always get matches like Kasparov + Deep Junior vs. Krammnik + Deep Fritz. You still get matches like Kasparov + Deep Junior vs. John Doe + Deep Fritz. Maybe even John Doe 1 + Deep Junior vs. John Doe 2 + Deep Fritz. It's called the world computer chess championship. |
|Feb-09-03|| ||ughaibu: Mdorothy: Well said. The idea that GMs use computers as partners to verify the tactical soundness of their strategic plans is completely destructive, it removes all the elements of risk, judgement and intuition that made Tal such a wonderful player. Does anyone seriously want to see such a development just so that they can claim the chess is somehow objectively better? |
|Feb-09-03|| ||Argull: i don't think your comments are against me, but i think chess go to the end. Obviously not coffee chess or chess between friends, but yes for important tournaments. Someone spoke about GM+computer that's the end of chess, risk and interesting are very important in our play.Tahl combinations are not possible with GM+computers. Fischer personality that was very important is not important with computer+GM. If you can't adjourn games because of computers the time for tournament will be lesser, and then chess will be only tactic chess and not interesting strategy chess will be played. Someone told that a program of playstation wins him easily, but Frtiz7 splashed you in a very easy way once and another. but this is not very important, worst is the problem of tournaments. Money is very important and players will use computers to win adjourned games. All of you use computers to see games and no boards why? |
|Feb-09-03|| ||aulero: I prefer much more watching Carl Lewis than Ferrari and Schumaker, and I find Formula 1 boring, nevertheless Formula 1 is not without interesting points. |
I hate the use of computers to extend chess skills, nevertheless they are currently used in this way, only they are not yet officially permitted during games, but ...
I'm only stating that I don't see any other possible solution if not stopping to play chess. Computers like any other machine are extensions of human skills: you can dislike their uses, but not avoid it. After Alekhine's and especially Botvinnik's approach to chess preparation, I think all this a natural and inevitable evolution.
I will always consider Lasker and Capablanca by far over any other player because their approach to the game was ... solving problems on the board.
Ughaibu, I agree with you about Tal's judgement and intuition and the implications behind your post.
|Feb-09-03|| ||Spitecheck: Computers are not better than humans, they are for the betterment of humans. They are simply servants of man, once and if they go beyond those bounds, society has doomed itself. |
Kasparov or for that matter Kramnik or any of their kind could lose 10-0 and it wouldn't make a lick of difference, the computer is still a tool. (albeit useful :))
|Feb-10-03|| ||magerk: the reason computers can beat humans isnt just because of massive calculation but because of mental states.if a computer loses,it doesnt bother the computer, but if a human loses or misses a win,it can seriously affect the way he plays the rest of the match.|
i believe the reason kasparov drew this game wasnt necessarily that he misjudged Bc1! but because he wasnt very confident about Rxc3!. if he was he wouldnt have offered the draw right after he played it.
|Feb-10-03|| ||mdorothy: But, lets all boil this down to what really matters... we can't really call what the computer does 'caclulation'... It's number crunching, yes.. but calculation requires thought. All the computer knows is a fancy formula that was programmed into by many, many humans using their own intelligence and technology. Yes, they don't get tired, and may eventually become overwhelming better than humans in OTB.. but it doesn't matter because that just shows the intelligence of humans to create it. And what happens when you pull the plug?? What do you get from the computer then?? I guess your reply could be 'about as much as you get from a dead GM', but my point is that computers rely totally and completely on humans and their inventions.. (they are our inventions). Anyways, I guess thats the long way of saying I agree with Spitecheck that all computers are just tools. They will never develop real intelligence, and hopefully never even pass the turing test. |
|Feb-11-03|| ||PVS: I think chess computers will be able to pass a Turing Test for chess, that is they will be indistinguishable from a human grandmaster based on the moves they make. As for thinking, that is a matter of definition to some extent, my present intuition is that computers cannot think. |
|Feb-12-03|| ||LemurWarfare: Hello all, my first post here
Let me begin by saying I play near a master level, while that isn't that great, I'm getting a PHD in artificial intellgence and I would like to make a few comments concerning the machine conversation above.
First of all, we have Moore's law, which states that processors double in speed and power and shrink by half every 2 years. Its a linear function thats held for the last 30 years and will make it at least another 20 until the size of a memory bit reaches an atom. At this point, you have quantum effects and classical computation breaks down.
What does this mean? Well the major way computers play chess is calculating positions. Its not inconceivable that 5 years from now a machine will be built that no human will EVER beat at chess, period. Especially since the strongest players are being drawn today. Just recently several other games have also been "solved" using computers. Example, chess could be solved as a win for white, always, although I tend to believe it will be solved as a draw.
Although many advances in heuristics have appeared in the last several years, its the brute force computation thats driving the major increases in chess playing ability.
I don't think this will end chess, but like John Henry, its the end of an era, even if he did win the piledriver competition. Remember he dies in the legend and the machines take over.
My granddad loves to play checkers, but in the early 60's a checkers program was written that cannot be beaten. I don't see many people my age (28) playing checkers although I gather it was once popular.
However there is hope, the japanese game Go has a branching factor of about 360 at each ply of the game tree ( compared to 36 for chess ). As there are more possible games of Go than atoms in the universe brute force searching has been a total failure. The best computer programs rely on inference rules or pattern recognition, like humans use, but are no better than the average Go player at a local club.
So if your worried about machines, play Go, it will be safe until I get my PHD and start my research group :)
|Feb-12-03|| ||ughaibu: An interesting post. One of the things about go is that pieces dont move and the number of options reduces as the game developes. Japanese compare go with farming and shogi with hunting, they say we change from shogi to go when we pass through middle age. I suspect that the natural way for machines to think about go would be from final positions. I once asked Matthew MacFadian about his go strength, he replied "I have a large vocabulary ["library" I dont remember which] of winning positions". Go is much more a knowledge oriented game than chess or shogi. |
|Feb-12-03|| ||LemurWarfare: In response to ughaibu:
I do believe you can capture pieces in Go, freeing board positions. Although I understand that often this isn't done if the pieces are truly "dead". So its not necessarily true that the branching factor drops. A smart human knowing he/she is playing a brute force machine may very well capture to increase the program's workload.
Also to note, reducing the branching factor from 360 (approx) by 1 isn't very much. Even after 60 moves its still 300. To keep scale in mind, looking only 4 moves ahead from this position involves over 8 trillion positions to evaluate. A daunting task considering that a normal 10 year old could look farther ahead.
As for MacFadian having a vocabulary, its his ability to recognize the patterns of winning positions evolving in the game. Thus pattern recognition is a key element of current approaches. Yet this is only the long term problem, also one must deal with tactics as well. Go is an interesting game, I plan on playing more.
But, I've played chess since I was 6, I'll always love it, even when my robotic minions defeat me.
|Feb-12-03|| ||Fool on the Hill: <LemurWarfare>
A.I.? Until you can get a computer to evolve it's own Machine (genetic) code, it will never be anything but a sophisticated number cruncher. Until the machine is under the stress of Natural Selection it will never evolve. The transistors are not fighting selfishly for their very lives. Things like co-operation, sacrifice, altrusism, as well as genocide, infanticide, greed, cannot be programmed later. They must be part of the original hardware upon which intelligence rests.Language structure, pattern seeking, curiosity are also neccessary building blocks to real intellect. I haven't seen any real progress in dealing with these facets of intelligence.
However Moore's Law moves exponentially, it may very well be that before you have found A.I. computers will beat the Go Masters.
|Feb-12-03|| ||Giuoco Piano Man: The comments on Go were very timely- this week"s on line Go newsletter from the US Go Association mentions Kasparov vs. Junior and that chess may be dying out as an intellectual game- and that computers have a difficult time playing Go at any real level. I am learning Go and it is perhaps more complex than chess- but the games are opposites in many ways. |
|Feb-12-03|| ||LemurWarfare: In response to Fool on the Hill:
Genetic algorithms has been a vital area of research in artificial intellgence for some time. In fact, my dissertation will probably involve genetically evolving neural networks that simulate artificial animals. Ideally one can figure out selection criteria to "force" certain types of things, such as language or altruism to evolve. People have been working on this stuff.
Then you get to write a nasty long paper and become famous.
Actually though, this particular subfield is too new to be getting any respect, professional journals and conferences are less than 10 years old. However I, unlike my compatriots, predict limitless potential...
I will however, dispute your claim that without evolution such aspects of artificial intellgence cannot be developed. In fact, its well known that evolution is the SECOND best way of solving a problem, the best is having a deterministic algorithm. And at this point in my education, I've seen some pretty crazy things. An example, one of the first rule-based AI systems ( 1980~ish ) was developed to diagnosis illnesses. A doctor diagnosis is wrong 1 in 5000 times on average. This program was wrong in 1 in 3 million times. Why don't you see it used everywhere? Cuz the doctors got together and killed it. Call it technophobia, people afraid machines taking over like in terminator. However in my opinion, what do doctors do other than diagnosis and prescribe medicine? Only surgeons really do something. When was the last time a doctor and not a nurse even stuck you with a needle? Rabid self-interest groups defending the status quo, using either the fear of god or the fear of machines to paralyze society to the detriment of everyone but themselves. Get outta the way I say!
But this is all academic ( forgive the pun ), the newscientist.com website has an AI section with many interesting articles, although they only tend to publish the sci fi type stuff. You can get your fill of genetic algorithms and artificial life there.
And before I start pontificating on how many angels can dance on the head of a pin, I'll leave the computer subject alone for awhile. I came here to learn about chess.
|Feb-12-03|| ||ughaibu: LemurWarfare: In go a single stone may be captured but this confers no territorial advantage, it's to do with gaining tempo by the 'ko' rule. Basically the board fills up and there is a finite number of possible full positions, this is why I was suggesting that the computers think from the end rather than the beginning as they do in hunting type games. |
|Feb-12-03|| ||Fool on the Hill: <Lemur>
I thought games were algorithms, but you're right; this is not an AI site.
This thread, and one running thru Kibitzer's Cafe started with the disappointment in Kasporaovs last 3 games with Deep Jr. He seems to be playing for a Draw, rather than attempting to win. He has stepped out of the "Book" a couple of times, and Deep Jr. was not "fazed" (if that word can be applied). So the discussion devolved into debate about Chess software.
More typical commentary can be found in the recent threads over Karpov, (great position player), and Spassky. Even there though, a lot of analysis is being done by computer, so there are references to computer "thought" in a lot of places.
Just type in an opening or a player you want to explore, this site even has games with historical commentary.
I myself enjoy Ruy Lopez and Sicilian games for, among other reasons, the patterns they give the board.
Do you play over the net? Try <GameKnot.com>
|Feb-21-03|| ||mdorothy: <Lemur>
Hey man, you sound very knowlegeable in the computer science field. If you don't mind talking about computers a little more, I want your opinion on an article I just read. It was in Time magazine. (FEB 24, 2003. pg.48) It was talking about physics, quantum physics, and the use of qubits in computers. It said that if this could be utilized further, just a row of 14 atom would be able to calculate more than the biggest current supercomputer. It made referance to a quantum Deep Blue. I am wondering how far you think we are from this possiblity? At this speed, how long would it take computers to completely solve all the possible positions in chess?
|Feb-21-03|| ||refutor: could be years, could be centuries...an advance of being able to use that kind of thing in computing would be a quantum leap, pardon the pun, for calculations, etc. and that technology, i believe would make the solution of chess trivial. in a related note, has anybody thought of the idea of using a massively distributed project like SETI (http://setiathome.ssl.berkeley.edu) to try and make some of the bigger tablebases. if they can calculate and analyze all of that data from space, i'm sure it would make creating a 7 or 8 piece tablebase easy ;) |
|Mar-09-05|| ||sierra: The AI used in chess programs is simply evaluation functions instilled by the GMs, not genetic programming.
After decades of AI research, computers can still only do what programmer tell them to, instruction by instruction. |
|Mar-15-08|| ||whiteshark: Garry Kasparov in an interview right after this 6th (last) game: http://www.youtube.com/watch?v=WWYa...|
|Sep-29-10|| ||RuneFriborg: after leaving the final position in Deep Rybka 3 for 8 hours (reaching depth 25) it says that the move ...f6 results in an even position (-0.11) and that the alternative Kf8 gives a small edge to white (+0.27)|
Also, it seems that Deep Junior was a very strong engine (Rybka agrees in most cases and never dismisses Juniors moves)
< Earlier Kibitzing · PAGE 2 OF 2 ·