I have been programming Go for 4 years, and until recently had one of the strongest programs (called SlugGo). Some of the article is correct, but much of it is out of date. The reason the prize "can't" be won is that its time limit expired, and the family that runs the foundation offering the prize decided not to renew the offer.
Many chess programmers are changing to Go because it is still very difficult and computer play is at a level a determined 8 year old child can reach in about a year. The article is correct that computers cannot use brute force enumeration of possibilities to surpass humans in Go because there are far more games of Go than elementary particles in the universe; More's law cannot catch up because there is not enough space to store the information. However, some as yet unknown breakthrough may allow for a huge leap beyond the techniques tried so far. People think and play one way, and computers will undoubtedly use a different "computer optimal" way to play.

David Doshay, Saratoga, CA, USA

I'm not really a Go player but how about this. Ok you're playing Go. How do you play it? Think? Do the first n moves matter - no. Forget them (to a point obviously) and it really reduces the number of combinations. For example. Let's say just give your opponent the least number of choices. Then there's a point where the board wakes up. But what's n eh? And when does the board wake up? Hmm... Someone will work it out. I think it's a two part game, do you?

Steve, stockport, england

Guy Macher: Better grammar checkers are available. They are not widely available because of market forces, not programming incompetence. When Microsoft Word crushed all of its competitors, competition to its crappy grammar checker went away. Microsoft has no incentive to improve theirs, and no one else can get a foot in the door because the demand is not there.

Leo Petr, Toronto, Canada

Although I'm sure it would happen eventually, I don't subscribe to the notion that the first time Kasparov was beaten by the computer. Anyone who wants me to believe that: kindly provide me the analysis of game 2 and tell me shenanigans were not in play.
After game 2 Kasparov's confidence was shattered. You can do that to a human, you can't do that to a machine.

Eventually there will be computers that can out-count humans on any game. It's very easy to say that the computer is stronger if you throw the price of a super computer at it and an army of coders and then crunch numbers. I'm not impressed.

You know when I'll be impressed? When the first computer autonomously designs, and plays, a game that a human can never win. That's when you've got something. Until then you've got an expensive abacus. Computers don't play games. You have to be alive for the word 'playing' to have meaning.

Frances, London, UK

Parkinson explained it a long time ago.

Until computers figure out how to expand the work to fill the amount of time available, they will never replace humans.

Ask yourself, what self-interested IT manager will hire a computer so good that it will replace him and his entire department? Wouldn't he rather hire a computer that always crashes, always gets infected with viruses, always gets hacked, always provides job security, always justifies a bigger budget and more staff for the Emp -- er, Department?

Why, if the computer industry were ever to develop a computer with built-in inefficiency, I bet it would outsell the Mac!

Joe Schembrie, Bellevue, WA

Computers can't beat good poker players even 'most of the time'

Daniel Barnett, Brisbane, qld

While there are two programs that can play Scrabble at the level of a world champion (the proprietary Maven and the open-source Quackle), they are certainly not able to find "the best next move, every time". Maybe 19 times out of 20, but so can the humans, and the imperfection is enough to make man vs. machine tournaments exciting viewing.

John Chew, Toronto, Canada

"Deep Blue won by “brute force” computing power: working out the possible moves and counter moves far ahead of play"

Brute force computation is only a small aspect of chess playing engines. Although there are fewer distinct games of chess than Go, there are still too many to calculate with today's computers. Rather, a certain amount of pattern recognition and alpha-beta pruning of bad moves radically reduce the number of positions that need to be considered. Development of the heuristics in a Go engine will eventually crack the problem.

Charlie, Oxford, UK

Go is well known in the AI world, and has had the "full works" done on it, to no avail. It's as Joseph says, our best computer opponents are for tasks where fast enumeration yields a good result.

The sad result is our other tools are not good at Go either, because they are not much better than fast enumeration, with respect to simulating intellegence, just different.

Adam Luter, Medford, MA, USA

Go is not obscure, and it is played across a wide geography. A look at the list of countries affiliated to the European Go Federation and the wide range of entrants who play in the World Amateur Go championships shows the game is truely a globally played game. If a computer can one day beat a professional Go player I will be sad fo the professionals, they earn a living by undertaking one of the most sustained and mentally challenging activities known to man. For amateurs (like me) we play for the pure joy of the game. I play with slate and shell stones on a wooden Go-ban, and whether a computer would havbe played here or there will not detract from my enjoyment, though it may help me get stronger!

Stephen Bashforth, Leicester, UK

Computers have yet to overcome the top players in Chinese Chess and Shogi primarily because the software has not matched the advances of computer chess, where many programmers are active and highly competitive. However, it will prove only marginally more difficult for a computer to triumph in these two games because the same "brute force" method is used. Go is a very different matter. Those who believe computers will soon rule the world hasten to say that they are on the point of defeating human Go players, but I do not believe it. I think that it will be a long time before a computer program defeats the top players such as Yi Chang-Ho and Yi SeDol.

Derek Jones, Aylesbury, England

I prefer Snakes and Ladders.

jennifer, edinburgh,

A badly researched article - the Ing Prize has expired in 2000.

Michael, Israel,

If a computer ever wins a game using brute force, just say, "Let's have another game, but we have changed the rules a bit". Start the next game immediately.

Ian Semmel, Maleny, Australia

Can a computer create a religion, that no one can beat?

John L. Sillasen, California, USA

The computer couldn't even see the problem with using the word THEN instead of THAN in "moves in Go is more then 250". Perhaps a better grammar checker should be the goal of computer nerds.

Guy Macher, Celebration , Florida

dr venables preller seems to think Go is obscure, but it is now played in over 60 countries worldwide, as the last World Amateur championships testify. I am part of the active Go community in the UK which has hundreds of members and over 20 tournaments each year organised through the British Go Association. In the Far East, where it originated, Go is more popular than chess is in the West.

It does chess a disservice to say it was 'solved' by brute force search through millions of options to find the best move. Chess programs actually use highly sophisiticated search techniques so they can find the best move searching only millions of positions, not trillions. These techniques have been transferred to other problems, and chess research helped the field of artificial intelligence as a whole. Let us hope that the quest for a good Go program brings similar benefits.

Tim Hunt, Milton Keynes, UK

Elementary my dear Watson!
All the games cited take place in a deterministic, closed universe. Even Go.
But just change slightly the rules. Another number of lines or columns, a different number of pieces, a small change in the rules. Do that from time to time.
Man is still on, alive and playing, trying, experimenting, understanding.
The computer is off. Every time the programmers are back to square one.
Man is adapted to an open, undeterministic universe.
The computer is not.
Have a nice open day.

raoul, andorra,

An interesting counter to the Alan Turing test of not being able to distinguish between talking to a human or computer is the Chinese Room theory.

Check out|: http://en.wikipedia.org/wiki/Chinese_room for a better explanation than I can fit in here.

One thing about this article.. its made me really want to play Go!

Stephen, Maidstone,

Perhaps some future quantum computer will make short work of Go's uncomputably large permutations? And perhaps, as people like Roger Penrose and Stuart Hameroff suggest, that will give us some real insight into how human - and animal - intuition works.

Jason Gorman, London,

Computers are excellent at spotting subtle, hidden patterns. This is illustrated by the anecdote (which may be apocryphal but is certainly credible) of the neural network developed by Pentagon contractors to identify aerial photographs in which were hidden camouflaged tanks. Neural networks are "trained" by showing them lots of test data, in this case lots of photographs with and without tanks. The network was trained until it gave a 100% success rate on test data, then came the day of the live demonstration. It got nearly everything wrong. It turned out that the Pentagon had bought a very expensive system for telling whether a photograph was taken on a cloudy day or not. The computer had spotted a pattern in the test data which its human trainers had completely missed.

The problem does not lie in getting computers to recognise patterns; it lies in understanding how we attach significance to them and how to communicate that to the computer.

Ian Kemmish, Biggleswade, UK

I won't be worried until the little plastic devils can play cricket as well :-)

Edward Andrew Green, Upminster, England

Regarding Mr. Whitney's comment: it would be quite a feat if we did fully understand the algorithms we create. The endless bugs in the programs we use every day, however, are ample evidence that we do not.

The behaviors algorithms used in tasks such as game playing, are particularly challenging to predict, since they often incorporate chance (randomness).

Mukesh Agrawal, Pittsburgh, PA, USA

The most sophisticated "algorithms" are not developed by humans, but via evolution within computers, in undoubtedly a similar way to the way nature produced our nervous system's "algorithm". It's stupidity to think there is something magical about "human intelligence".

Joseph Smith, Salt Lake City, Utah

All previous attempts to construct a strong Go playing have failed, the current approaches now favour Monte Carlo simulations to guess the best move! Nobody tries to use an Algorithm to work out what the best move is anymore. Humanity should get the chance of revenge against Deep Blue. Lets see a match between Man and Machine over a Go board!

Ian Davis, Belfast, Northern Ireland

Why are the scientific articles in this paper so much less well researched and written than the arts ones?
There exist computers that can perform fuzzy logic using neural networks. There is no cognitive task that a human can perform that a computer can't (arguably apart from be conscious).

The real question is whether it is worthwhile to design computers like humans or whether it is more efficient to have them specialise in certain tasks. That is, what is the use of making computers imitate humans when they can be more useful as themselves?

JMcD, San Francisco, US

Each rare time I've beaten the computer at chess, I suspect immediately that "it let me win." : )

dick, peterborough, usa/nh

Oh dear, oh dear , oh dear Ben!! Everything you have said about Go was precisely what was said about chess 20 years ago. I can clearly remember reading similar articles. Chess playing was regarded as having something special about it that could not be replicated in a 'dumb' computer. Gary Kasparov (remember him) said in 1992 that he thought a computer would never be able to defeat him at chess, as well he might, as the best machines at the time were no match for him. Well look what happened! Computer power is improving at an exponential rate. It is only a question of time before Go goes the same way as chess, and it will be a far shorter time that most would guess.Incidentally J Bruno your observation that chess can be solved by 'mindless enumeration of possibilities' and that 'it follows that chess is not a good artificial intelligence test after all' is an absolute classic from the naysayers of A.I. That is when AI solves something then it cannot have been intelligent after all!

James Turner, Berwick upon Tweed,

GO may seem to separate humans and computers, but actually, it even separates different types of humans. GO's rules are simple yet learning it to the degree that you one can beat present day computers is no easy task. It seems only some humans the unique intelligence this article discusses.

Sabio Lantz, Pittsburgh, Pennsylvania

Put your money where your wordprocessor is? Why do you think that the electronics of the human brain will never be surpassed? Why do you keep moving the goal posts? If you are confident of your position, make a stand and admit defeat if that position is subsequently lost.

Godfrey Powell, Canterbury, Kent, England

No matter how sophisticated a computer algorithm may be, the human who developed it will understand it perfectly. Can we honestly say that anyone has done more than barely scratch the surface of understanding human intelligence?

Jon Whitney, Murray, Utah USA

The intuitive, wise quality of humans is the thing which sets us apart from clever computers. Why then do we rely so much on a tick-box mentality rather than our judgement? The tick box approach (does the primary school have over 3.8 square meters of space per pupil? 1 point. Did the trainee doctor mention the following key words in his interview? 1 point each, etc, etc) reduces us to the level of rather stupid machines.

We can do so much better. But we do need to display enough courage to make decisions based on experience and wisdom - and sacrifice the false god of complete impartiality (which has never existed) in order to use our talents to their best.

Roddy Campbell, Christchurch, New Zealand

Just an aside. Some AI fellow is working on a word game I invented with considerably more challenge and combinatorial possibilities than Scrabble. Although it could in theory be "brute forced," it would not be easy. To make the solution harder, the computer, I think, should have some limited vocabulary just like humans do. A timely bluff and dealing with uncertainty are part of the game.
My brother and I once tried to learn go many years ago, by playing out an annotated game from the instruction manual. The first move had the comment "Unusual Opening." The second move which placed a bead next to the first one had the comment "Corner Captured!" with that exclamation point. Seeing that, we figured the game was too subtle for us and gave up the exercise.

Peter Roizen, Los Gatos, California

The mathematics of Go are complex but finite. Computer mastery is simply a function of computational power. When computers mastered Western chess it was said that Shogi, or Japanese chess would never be "solved" since it is more complicated. Now 10 years of Moore's law on, the gap has narrowed to such an extent that only world class players can be confident of winning. By 2017 I would guess we will be close to having a "perfect" Go computer. However we are likely over a century away from software mastery of truly intellectual games like table tennis or football. Beating top ping pong and soccer players is far beyond today's computer science.

Veryan Allen, Tendo City of Shogi, Japan

Computers can't play chess. They can only play a different game that happens to have identical rules.

Computer chess was such a hot research topic because it was considered that chess requires intelligent thought. Now it has become clear that the mindless enumeration of possibilities is enough, it follows that chess is not a good artificial intelligence test after all.

Throughout AI the challenge is to find interesting problems that *require* actual manipulation of concepts and are *not* vulnerable to a boring "brute-force" enumeration attack.

Perhaps Go really is one.

Joseph Bruno, London,

The relative obscurity of the game in some outlying country areas might be one reason why the full works of neural networks and fuzzy logic with parallel processing if already unleashed has not, as yet, cracked this one. Likewise some of the other less frequently played diversions such as Diplomacy, Pit and Hindooly.

dr venables preller, Warminster, UK

It has been calculated that there are more distinct games of Go than atoms in the known universe

Who calculated this?

Was it a computer?

Paul Groom, London, England

Wow Ben, a bit afraid there? Maybe you should ask Ray Kurzweil what he thinks, and yes computers will outpace us in "Go" sometime in the next decade. So what? Does it matter? No machine will ever be able to turn a phrase quite like you, right?

Raymond Briant, Princeton, NJ, USA

There are still a few board games that the computer still can´t beat the best human players: Chinese chess, Japanese chess, etc. Although with the current trend in technology, it is assumed that it will be sooner rather than later that the best Chinese chess player will be beaten by a computer.

Kong Kek Kuat, Kuala Lumpur, Malaysia