Author: José Carlos
Date: 08:36:27 02/20/02
Go up one level in this thread
On February 20, 2002 at 05:07:41, Janosch Zwerensky wrote:
>Hi,
>
>>(...)
>>Programmers _know_ that any program (not only chess ones) is nothing but a
>>secuence of mathematical calculations.
>> In the very end, some 1's and 0's and the
>>hardware they 'dance' in.
>
>I think this is only one interpretation of what the computer does.
>Apart from it being useful for (low-level)-programming, I also think it is not
>an interpretation that can claim any special status over other interpretations.
>For example, what the computer does could be described on a purely physical
>level. On that level of description, there are most definitely no ones and zeros
>dancing around in the machine.
That's correct. My point was to show the difference between what a programmer
knows about his program and what the user sees. But it's true that you can talk
to the guy who designed or built the computer and he'll give another
description. Or you can talk to a physicist and he'll give a different one.
But all of those descriptions reinforce the point that a computer is
essentially different from a human: a computer is artificial; it can't be
aggressive or passionate.
>On the other hand, one can also describe the machine using broad, high-level
>abstractions - saying that it does these things with lists or those with strings or "tries to find a way to do the worst to its opponent".
You have to be very careful and not confuse concepts. You can abstract some
mathematical calculations into that sentence, but don't compare this to the
'human-applied' equivalent sentence. Metaphores are only a languaje trick.
>If one is writing a C
>program, this point of view isn't all-too-useful, since there is no obvious
>mapping of these abstract concepts to C language features, but if for example I
>wanted to know what the program will respond to my moves, *most of the time* the
>latter description will be more useful to me than the former ones (with a few
>exceptions where a somewhat more low-level understanding of the machine will be
>good for exploiting its weaknesses).
As I said before, you can abstract, but not compare to human behaviour.
>When doing actual computer programming, in my experience at least the answer to
>the question of what mental model of the machine is most useful also *in part*
>depends on the language one is using; for example, I believe that an assembly
>programmer will definitively benefit more from a very-low-level mental image of
>the machine than a C programmer, who in turn might think about his machine in
>lower-level terms than a Lisp programmer, who in turn probably will find
>somewhat lower-level descriptions of the machine more useful than, say, someone
>doing programming in Maple (at the top of this hierarchy, of course, would be
>the DWIM programmer ;)...).
>My point here is that one level of description doesn't invalidate the others,
>and that what we call, for example, emotions in humans are just very, very
>useful high-level descriptions for human states. The fact that it really all can
>be described in terms of quintillions of low-level interactions of neutrons,
>electrons and protons does in my opinion not in any strong sense imply that the
>high-level view of things is illusionary, since high-level descriptions of
>things are useful to the extent that we could not make sense of anything without
>them.
That's another topic. I personally believe in human beings as tremendously
complex machines, and that is in agreement (almost) with your arguments, but I
understand most people don't, and then I argue from the point of view of the
majority.
If I had to answer your arguments with my own ideas (regarding the nature of
human mind), I'd agree mostly with you but, in such a phylosophycal problem ('no
beginning', this is, where do you stop removing abstraction so that you can say
'this is pure reality with no abstraction at all') I accept the human being as a
starting point. That's axiomatic, and thus it doesn't make sense to argue about
it. It's just that we need a starting point. But I accept that if you lower the
abstraction level to particles, and we admit human beings are nothing but
physical molecules behaving according to physical laws, then computer programs
and humans are 'essentially' the same: physical systems without any
'romanmticsm'.
>Another point to consider here is that, if one adopts the point of view that of
>several descriptions of a thing only the lowest-level one is non-illusionary,
>they have to bite the bullet that, for example, light bulbs are an illusion,
>since light bulbs are not elementary particles in any reasonable theory of
>physics. However, it is tremendously useful to think that light bulbs exist,
>because they have some interesting properties that no single part of the light
>bulb has. An abstracted description of something thus does not usually
>contradict a low-level one, but it is merely focused to different aspects of
>that object (and in fact in the case of the light bulb the *correctness* of some
>high-level statements about the light bulb might be deduced by low-level
>physical considerations).
You're reasoning is totally correct. Abstractions are usefull, but that
doesn't mean that abstract concepts 'really exist', whatever that means :)
>> But the users tend to see the program as if it was a person. Tend to used
>>words like 'creativity', 'aggresiveness', 'passiveness', and so on.
>
>In my opinion, whether a program has these characteristics or not depends on how
>one defines them in the first place.
Of course, everything depends on the definition. But I was talking about the
comparison to humans, and hence about humans' definitions. You can't say a
program is creative in the same way Shirov is creative unless you consider
Shirov's chess abilities just a program and thus, consider human beings as
'carbon-based computers'.
>>(...)
>> But as
>>I said, in the end, it's nothing but a mathematical calculation that choses this
>>or that move.
>
>Well, from a purely physicalistic point-of-view the claim that a computer is a
>machine performing calculations might be viewed as metaphorical already.
>
>> Believing that a program can be 'creative' is like believing that
>>it rains because the clouds are sad and cry
>
>The hypothesis that clouds have emotional states won't help us predicting the
>weather.
Not exactly true. Myths were useful in the past, in some sense. For example,
if you believe the clouds hypotesis above, and you notice that clouds usually
get angry when it's hot, then you can predict (when it starts to get hot) that
it's gonna rain. Of course this is not a real case, but just an example to
illustrate the point.
>The belief that a given program is 'creative' is simply too ill-defined on its
>own to be either refuted or confirmed in my opinion.
Not that ill-defined if you use the human definition. But anyway, I believe
too many things are ill-defined in the world, and too many discussions are
pointless due to lack of good definitions (for example 'are computers GM
strength?'; who knows... without a good definition of 'GM strength').
>>(...)
>
>regards,
>Janosch.
Regards,
José C.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.