"Chris Lawrence" wrote in message
losys.co.uk...
On Fri, 11 Jul 2003, Neil Fernandez wrote:

In addition it would be wrong to assume that there is no game with a
clearly defined and finite state space at which no computer programs
will be _able_ to play as strongly as the strongest humans can, even
with lots of processing power and parallel processing.

I've never claimed it is not possible or that it will never be possible.
Merely that it is not possible today and processing power is not the
stumbling block.

When we play Go we are using a lot of gut-feeling, intuition, visual
effect, experience, reading, etc. Some of those things can be
represented or effected in today's computers. Others cannot.

We need to a) understand what happens when we think about things, b)
create a computer with a more suitable architecture on which to map
those processes and c) understand the thoughts which take place in Go
and map those onto b).

At the moment the best we have - a linear brute force capability - is
good enough to 'play' chess against people considered among world
champions. But it's by no means 'playing' chess or 'thinking' about its
game, it's just at a level where the level of brute force available,
combined with the way chess works, is an alternative way to play a
grandmaster level winning game.

The same brute force applied to Go is useless because there is no real
understanding of how to map the things we think into a computer model on
which to apply that brute force, and even if we could it's orders of
magnitude larger anyway.

Strength is useless without understanding of how to use it.

--
Chris

That's not true. Go, being a game, like Chess, of perfect information, can
be "solved" by brute force just as easily as Chess can - there's no need to
bring any fuzzy ideas about human thought processes into it. The only
problem is that a much larger degree of computer power is required, due to
the fact that Go has a larger board and a much larger possible number of
opening sequences than Chess does. Thus, it may be the case that "thinking"
computer programs are made that can play Go at a high level before computers
get fast enough for brute force programs to do the same. There's no
theoretical reason why it couldn't happen the other way round, though,
especially with quantum computers looming on the mid-future horizon.

--
gravity

On Sat, 23 Aug 2003 23:31:21 +1000, "gravity"
wrote:

That's not true. Go, being a game, like Chess, of perfect information, can
be "solved" by brute force just as easily as Chess can

But chess is not remotely close to being "solved" in the mathematical
sense, and may never be. All that has happened is that programs can
generate moves that can beat grandmasters.

- there's no need to
bring any fuzzy ideas about human thought processes into it.

If you are talking about solving the game, then that is true: all that
is needed is the ability to evaluate all the ending positions. But
that is not what chess programs do, and the gap between what current
chess programs can do and what current go programs can do in
competition with human players is very much a matter of programmers'
inability to write code that can compete with human thought processes.

The only
problem is that a much larger degree of computer power is required, due to
the fact that Go has a larger board and a much larger possible number of
opening sequences than Chess does.

If you're talking about solving games, the number of terminal
positions is more relevant.

Thus, it may be the case that "thinking"
computer programs are made that can play Go at a high level before computers
get fast enough for brute force programs to do the same. There's no
theoretical reason why it couldn't happen the other way round, though,
especially with quantum computers looming on the mid-future horizon.

??? That seems a bizarre comment to me. The numbers involved are
such that no plausible quantum computer could make a dent in the
problem by brute force.

-- Roy L

gravity wrote:

That's not true. Go, being a game, like Chess, of perfect information, can
be "solved" by brute force just as easily as Chess can - there's no need to
bring any fuzzy ideas about human thought processes into it. The only
problem is that a much larger degree of computer power is required, due to
the fact that Go has a larger board and a much larger possible number of
opening sequences than Chess does. Thus, it may be the case that "thinking"
computer programs are made that can play Go at a high level before computers
get fast enough for brute force programs to do the same. There's no
theoretical reason why it couldn't happen the other way round, though,
especially with quantum computers looming on the mid-future horizon.

We are constrained by the size of the universe, and its age.

There are more go positiions than there are atomic particles in the universe.

So, even if you can get _every_ atomic particle in the universe to somehow act
as a "bit" for you in your quantum computer, you won't have enough storage to
do your brute-force search.

Sorry, there _is_ a theoretical reason: You don't have enough bits.

Furthermore, "theoretically" assuming that you do have enough storage,
it will take longer than the age of the universe to do your brute-force search.

Sorry, there _is_ a theoretical reason: You don't have enough time.

Such problems are referred to in the literature as "inherently intractable".

Google for same, if you don't believe me, or see, for example:
http://robotics.stanford.edu/~koller/papers/position.html .

Computers will absolutely _never_ be fast enough, nor large enough, to
solve go by brute force. Sorry, that's just the way it is. Theoretically.

--
Orne Batmagoo