Kasparov on computer chess in the New York Review of Books

[Leto]

I'm not a programmer but I think using current methods there's still a lot of room for improvement for chess engines. Just make a match between Rybka 3 and itself but give one side a time handicap and you'll see how imperfect Rybka 3 is.

However my dream is that someone will someday create an artificial intelligence (I'm thinking more of something like Data from Star Trek) that learns things just the way we do. This artificial intelligence would learn the game of chess just like we do; by learning how the pieces move, learning the strategies of the game, and learning the tactics the same way we would learn them, by reading books. This artificial intelligence will eventually write its own books about its experiences playing against top chess engines and how "unromantic" they are.

A fun thought just came to me. Would the future chess battle be between an intelligent machine and a chess engine powered up by supercomputers? The supercomputers would of course have the computational speed advantage, but would the intelligent machine's creativity prove to be a decisive factor?

That to me would be so awesome, I think about it all the time, I want to see it done so badly. Please someone do it before I die.

How about this?

Yea that's exactly what I'd want. Doesn't even need to look human like in that video.

[jdart]

"Brute force" could cover a wide spectrum of techniques. Current programs that do selective pruning and extension are far from just blindly looking at every node equally. Indeed the old distinction between "Shannon type A" and "Shannon type B" programs is far from clear now. I'm not sure Kasparov, not being a programmer, appreciates this.

It is true that current successful programs do not different radically from each other in their general approach. I'm pretty sure we are going to see future attempts at some quite different algorithm. But the past record of these experiments is not encouraging.

--Jon

[Spacious_Mind]

"Brute force" could cover a wide spectrum of techniques. Current programs that do selective pruning and extension are far from just blindly looking at every node equally. Indeed the old distinction between "Shannon type A" and "Shannon type B" programs is far from clear now. I'm not sure Kasparov, not being a programmer, appreciates this.

It is true that current successful programs do not different radically from each other in their general approach. I'm pretty sure we are going to see future attempts at some quite different algorithm. But the past record of these experiments is not encouraging.

--Jon

I agree, selective search provides the most interesting approach. My thoughts though are that there should be some sort of standard of equality against which chess engines can be measured for their improvement of knowledge and also use this standard of equality as a base for matches against humans. Don't know what that standard should be... but it kind of makes sense to me to keep the interest going between top human players and computers.

regards

Nick

[bob]

http://www.nybooks.com/articles/23592

With the supremacy of the chess machines now apparent and the contest of "Man vs. Machine" a thing of the past, perhaps it is time to return to the goals that made computer chess so attractive to many of the finest minds of the twentieth century. Playing better chess was a problem they wanted to solve, yes, and it has been solved. But there were other goals as well: to develop a program that played chess by thinking like a human, perhaps even by learning the game as a human does. Surely this would be a far more fruitful avenue of investigation than creating, as we are doing, ever-faster algorithms to run on ever-faster hardware.

This is our last chess metaphor, then—a metaphor for how we have discarded innovation and creativity in exchange for a steady supply of marketable products. The dreams of creating an artificial intelligence that would engage in an ancient game symbolic of human thought have been abandoned. Instead, every year we have new chess programs, and new versions of old ones, that are all based on the same basic programming concepts for picking a move by searching through millions of possibilities that were developed in the 1960s and 1970s.

Like so much else in our technology-rich and innovation-poor modern world, chess computing has fallen prey to incrementalism and the demands of the market. Brute-force programs play the best chess, so why bother with anything else? Why waste time and money experimenting with new and innovative ideas when we already know what works? Such thinking should horrify anyone worthy of the name of scientist, but it seems, tragically, to be the norm. Our best minds have gone into financial engineering instead of real engineering, with catastrophic results for both sectors.

Some truth there, I think

Yes, but by the same token, we would apparently be better off giving up on motorcycles, bicycles, and automobiles, and trying to build a method of transportation that uses two leg-like structures, balances like we do, etc?

It is somewhat arrogant to say that humans do this the right way and computers do not. I'm not even convinced that at some basic level, somewhere, someone will never find evidence of "binary programming" in the human being. Won't that be a lark.

[markboylan]

there should be some sort of standard of equality against which chess engines can be measured for their improvement of knowledge

Maybe they could use chess.

[Stephan Vermeire (Brutus)]

Hi there,

To start with: I am both a chess-programmer and a biologist. Here is my perspective.

My belief is that the future lies within exploring neural networks and parallel computation. Related to chess it might probably be combined with the currect brute force approach.

Neural networks can achieve great performances. That's the result of evolution and learning. The problem is: We have hardly a clue what is going on in a brain! Evolution leads to the best solution, not the most understandable or reducable one. Programmers want to be in control of what they do. If you look at a neural network, you just know that is is working without having a clue why! Nevertheless, neural networks have great potential. Imagine the power of an electronic neural network composed of metals instead carbons and fluids that has the size of a human brain. Human current transduction is slow! It would outperform a human being by all means.

Parallel computation has a future. A very complex future for programmers I am affraid. I believe that one of the main challenges will be to make neural networks controllabe so they can be made part of a design. Of course hardware developments will be one of the keyfactors in speeding op the performance also. My belief is that exploration of the potential will be in the field of software initially. If it demonstrates to be worthwhile, hardware will follow.

I don't say neural networks are better than a negascout algorithm. The current searchalgorithms are at least a very effective on serial computers. Also consider the following: A human chessplayer will use some sort of negascout algorithm while investigating the best line of play during a match. in addition to that, humans play intuitively while discovering patterns. Perhaps the ultimate chess program is a combination of both conventional search algorithms and a creative neural network. Time will tell.


It is an interesting era we are living in!

Stephan

[michiguel]

http://www.nybooks.com/articles/23592

With the supremacy of the chess machines now apparent and the contest of "Man vs. Machine" a thing of the past, perhaps it is time to return to the goals that made computer chess so attractive to many of the finest minds of the twentieth century. Playing better chess was a problem they wanted to solve, yes, and it has been solved. But there were other goals as well: to develop a program that played chess by thinking like a human, perhaps even by learning the game as a human does. Surely this would be a far more fruitful avenue of investigation than creating, as we are doing, ever-faster algorithms to run on ever-faster hardware.

This is our last chess metaphor, then—a metaphor for how we have discarded innovation and creativity in exchange for a steady supply of marketable products. The dreams of creating an artificial intelligence that would engage in an ancient game symbolic of human thought have been abandoned. Instead, every year we have new chess programs, and new versions of old ones, that are all based on the same basic programming concepts for picking a move by searching through millions of possibilities that were developed in the 1960s and 1970s.

Like so much else in our technology-rich and innovation-poor modern world, chess computing has fallen prey to incrementalism and the demands of the market. Brute-force programs play the best chess, so why bother with anything else? Why waste time and money experimenting with new and innovative ideas when we already know what works? Such thinking should horrify anyone worthy of the name of scientist, but it seems, tragically, to be the norm. Our best minds have gone into financial engineering instead of real engineering, with catastrophic results for both sectors.

Some truth there, I think

He hit it on the nail. There is less and less Artificial intelligence and more and more domain specific "how can I get two more elo points". In fact, many positions that reflect the fundamental problems of today's engines are many times dismissed as "not worthy of attention because they are rare". Engines continue to suck at endgame planning, but... who cares.

Miguel

[Spacious_Mind]

there should be some sort of standard of equality against which chess engines can be measured for their improvement of knowledge

Maybe they could use chess.

Unfortunately not. You are comparing maximum search depth with selective human search and that's not equality.

back to apples and oranges regards

Nick

[markboylan]

Unfortunately not. You are comparing maximum search depth with selective human search and that's not equality.

Oh. Winning isn't the object of the game anymore?

[Spacious_Mind]

Unfortunately not. You are comparing maximum search depth with selective human search and that's not equality.

Oh. Winning isn't the object of the game anymore?

Of course but under similar conditions would be a better achievment right?

You have already won, with aid of maximum ziga core computers if that makes you feel better

hopping on to my pogo stick off to play basketball regards