Kenneth Sloan wrote:

Guy Macon wrote:

The question of whether computer *should* be prohibited
from using an opening is another question. In my opinion,
the terminology used can beg the question. For example,
in the paragraphs above, the assumption is made that the
storage inside a computer is like a physical book. In many
arguments that I have seen for the opposite position, the
assumption is made that the storage inside a computer is
like a human memory. This is a subject I know a lot about,
having designed hard disks, memory chips, optical storage
systems, and even one system with a bubble memory. I can
say with confidence that the storage in a computer is not
like a human memory *or* a physical book, It isn't like
anything except computer storage.

You have impressive credentials lending weight to your opinions on
computer memory.

But...how much do you know about HUMAN memory?

Quite a lot, actually, if you are talking about the same sort
of knowledge I have about computer memories. Knowing the
exact details of how each transistor (or laser or magnetic
pickup) works together to store bits doesn't tell me much
about how the information is stored and retrieved. That
RAM might contain a C++ heap, a ramdisk with a particular
filesystem, an array of nybbles, or a Forth interpreter.
And I certainly don't know the content of these unknown
data structures. In like manner I know a fair amount about
how a neuron works, how it communicates with the next neuron,
how various chemicals affect it, all without knowing how the
brain organizes memories. Not as much as I know about
computer hardware, but I know about as much about neurons
as any well-read person would know.

This knowledge lets me know certain limited aspects
about the information stored in each case. For example:

A normally operating RAM does not "forget" information.
It stays where it was put and can be accessed at will.
A normally operating brain *does* forget information.
I don't know whether the actual information is lost or
whether the problem is in the retrieval, (the experiments
of Penrose give some clues) but I know that there are
many cases where a normally operating brain can recall
information at one point in time, later cannot, and on
occasion even later can.

Except in the case of special fault tolerant hardware,
a system using RAM is pretty much hosed if something
comes along and destroys a third of the memory array.
Brains often retain a surprising level of functionality
even when severely damaged.

....and that's why I believe that I know enogh to conclude
that they are not the same thing.

Have you designed any of these?

A human brain? That would be that other fellow with a
three-letter name that starts with G...

I have dinked around with neural nets (The old Voice
Direct 364), but never to the point where I shipped
a product with one in it. I would like to try the
RSC-4128 FluentChip that replaced the old VD364.

If not, I suggest you don't have any special standing to judge the
relationship between computer and human memory.

I don't think it really takes special standing to conclude
that the two have different capabilities. Any observer can
see that.

And that is part of the point.

It's easy to dismiss computer memory/processing precisely because we
*can* open the hood and see how it works. This is the "oh - now that I
see how you do it with computers I understand that this task does not
really require Intelligence". That's the entire point of the Turing Test.

You make exactly the same error that almost-IM Innes makes.

Wow. You somehow know what my position is when I have been
careful not to reveal it! I would suspect that perhaps I let
it slip out, except for the fact that you got it wrong.

The reason why I do *not* "dismiss computer memory/processing
because we can open the hood and see how it works" is because
I can do the same thing with the human brain; there really
isn't any mystery about how a neuron works or how it
communicates with the next neuron. Oh, sure, I can also look
at the program running on the computer and not at the "program"
running on the brain, but that's to be expected when comparing
a program written by a human with one written by natural
selection.

As a matter of fact, the ability to "open the hood and see how
it works" is not an innate attribute of computers, but rather
an artifact of the kind of computer we currently use. It is
entirely possible to make a computer simulation that acts just
like a single Neuron (unless, of course, Roger Penrose is right)
and to design a hardware interface so that this artificial
neuron communicates with another just like the real ones do.
In theory, if you hook up a huge number of them in the same
configuration as a human brain and give them the same starting
condition, it would do exactly what a human brain does -- and
I would most definitely *not* be able to "open the hood and
see how it works."

This is the "oh - now that I see how you do it with computers
I understand that this task does not really require Intelligence".

(Switching sides because I know and can argue both sides...)
So you think that the Chinese room and Blockhead are examples
of intelligence?

Quiz question: (fill in the blank) "I will believe that computers are
intelligent when they can ___________________________"

Good thing you didn't write "I will *only* believe that computers
are intelligent when they can..." If you had specified that no
*other* ability would convince me, the following would imply that
I don't think passing the Turing test qualifies -- a question that
I have been careful not to reveal my answer to. That being said;

I will believe that computers are intelligent when they
can independently come up with reasoning that reconciles the
"artificial neuron brain" argument above and the "Chinese
Room/Blockhead" arguments detailed previously. There may or
may not be other things that would convince me, and passing
the Turing test may or may not be one of them.

Back to the original question. I believe it will be fair to have a
computer play chess with it's long-term memory turned off if and when
you find a way to turn off the long-term memory of the computer's opponent.

....and FIDE and USCF introduce rules that forbid anyone who has
a photographic memory and thus can refer to printed books internally
from ever playing a rated game..

Of course, in that case both of the players will have a bit of trouble
remembering how the horsie moves.

As Marie Antoinette said, "Let them play Go!"


--
Guy Macon
http://www.guymacon.com/

Guy Macon wrote:
Kenneth Sloan wrote:
Guy Macon wrote:
Not as much as I know about
computer hardware, but I know about as much about neurons
as any well-read person would know.

I'm sorry, but "well read" wasn't the standard I had in mind.

--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/

Henri Arsenault wrote:

I think the difference between tactics and strategy has been
over-emphasized.

For a game like chess, it seems to me that strategy is only a shortcut
for tactics. To take a simple example, the concept of opposition in a
king vs pawn ending is just a way to calculate how to win using fewer
moves - one only has to calculate far enough to ensure that he has the
opposition, and the total win combination is a series of such
sub-calculations put end-to-end.

Isn't this also true for more complex positional concepts? Sure if these
sub-algorithms are programmed into the computer, it could calculate
more deeply, but the distinction between positional knowledge and
tactics is then considerably blurred, and would depend only on how deep
the computer can calculate.

Question; when you play against a computer with a large tablebase,
starting with a position that is in the tablebase, does it seem
to you that it shows what we normally consider to be strategic
thinking? If so, that would support the above theory.

--
Guy Macon
http://www.guymacon.com/

help bot wrote:

A possible solution would be to force computers
to consume pizza and beer, which would ultimately
clog their circuits and lead to a more human-like
imperfect memory of digested book knowledge.
Unfortunately, firms like IBM and Intel have lobbied
against this for years, and they have a lot of clout
(i.e. money).

Years ago I worked on a handhelp field service data
collection decvice for IBM. As part of the acceptance
testing, I had to soak it in gasoline, mustard, vinegar,
boiling water, ice water, bleach, blood, alcohol, etc.
I think IBM would be able to handle the new requirement.

Do you have any idea how hard it is for a six foot tall
burly dude who looks like a biker to procure a bucket
of blood in Los Angeles?

--
Guy Macon
http://www.guymacon.com/

23.08.2007 10:17, Guy Macon:
;

There exist people who disagree with the validity of Turing's test.
They would not agree that "If a program achieves to convince
intelligent people that it is intelligent - it is intelligent."

Agreed. But this is a point well behind this little chit chat here and I
have doubts that Mr. Innes was heading this direction with his criticism.

Greetings,
Ralf

23.08.2007 16:25, Chess One:

Whatever else is introduced to the discussion are side-lines and
sophistries,

.... introduced by yourself ...

Greetings,
Ralf

In article ,
Guy Macon http://www.guymacon.com/ wrote:


Question; when you play against a computer with a large tablebase,
starting with a position that is in the tablebase, does it seem
to you that it shows what we normally consider to be strategic
thinking? If so, that would support the above theory.

Fair question: so let me ask another one...

When Kasparov surprised an opponent with a brilliant strategical move
prepared months in advance, was that strategical thinking (from the
point of view of that single game)?

In both cases, was not the "strategy" calculated in advance? And does
the fact that it was calculated in advance mean that it was not strategy
at all? Is a move strategical only when it is found "over-the-board?

If the answer is yes, then Napoleon's brilliant victory at Austerlitz
(won by creating a weakness in his line that was actually a trap) was
not strategy at all, since he had concocted it in advance.

So if the computer's using advance knowledge is not strategy and the
human's is, please explain the difference?

Henri

In article .com,
help bot wrote:


Really? I am now wondering if his definition of
a "logical system" might have something to do
with this proof. Common sense might tell us that
it is illogical systems which tend to be incomplete
or inconsistent, and that those aspects are some
of the best clues to telling the difference. :D

No, Godel's theorem is a mathematical proof. It is discussed in detail
in the book "the Emperor's New Mind" by mathematician Roger Penrose, and
interesting book whose main purpose was to debunk much of the bullship
about qrtificial intelligence.

Two of the most interesting consequences of Godel's theorem are that
there are non-trivial mathematical theorems that are unprovable, and
that not all of mathematics can be reduced to axioms (arithmetic can and
algebra cannot).

Henri

"Kenneth Sloan" wrote in message
...
Chess One wrote:
"Henri Arsenault" wrote in message
...

(Godel proved in 1934 that any logical system must be either incomplete
or inconsistent).

Hey, what's this? Science!

No - it's Mathematics.

Ah - philosophy as understood in Alabama - 40th in the world in Math, and
47th in the States.

That too, is uh, a datum.

Phil Innes


--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://www.cis.uab.edu/sloan/

"Ralf Callenberg" wrote in message
...
23.08.2007 16:25, Chess One:

Whatever else is introduced to the discussion are side-lines and
sophistries,

... introduced by yourself ...

no sir, introduced by others.

my proposition was very simple, and very much avoided! look at the nonsense
it has spawned! including now this denial by yourself

how strong is the engine; the book?

pfft! hysteria by techno-geeks for daring to raise a question! and that is
entirely normal in the chess computing community, otherwise there could be
claims to science, to data, to objective measures,

and if anyone had any they ever deployed rather than specualted upon, as
'interesting' hypothesies, et cetera.

the question i raise is avoided by others, for interesting psychological
reasons [what else?] - there are no factual answers, after all

phil innes

Greetings,
Ralf