Re: SUO: The Story So Far - Request for vote
Pat,
The primary point of my suggestion is that the overall efforts
of the various sciences hang together in a way that makes it
possible to relate one to another for useful collaboration.
I certainly admit that the problems are immense, and I don't
want to minimize the difficulties. But I think that we can
usefully look at the history of the sciences as a model for
ontology development. (Before you jump up on this point,
let me explain the details below.)
>>Another approach that has proved to be highly workable in
>>modern science and engineering is to have a limited number
>>(ideally one) very general theory, such as quantum electro-
>>dynamics (QED), of which everything else is a special case.
>>For most applications, an engineer uses whichever special case
>>fits the problem at hand, but every one of them can be derived
>>as an approximation to QED.
>
>John, Im sorry to be blunt, but this is complete nonsense. Almost
>nothing in any field of human endeavor, including most of physics,
>can be derived from quantum electrodynamics, even by approximation.
The word "derived" was perhaps not the best choice, since
historically the development proceeds from the bottom up.
QED itself was developed by merging quantum mechanics with
special relativity, and there still remain many open questions
about how to bring general relativity and other things into
the Grand Unified Theory (GUT).
However, it is possible to take QED and derive from it the
basic equations of special relativity, nonrelativistic QM,
and Maxwell's equations of ordinary electrodynamics.
Then you can derive the basic equations of Newtonian mechanics
as special cases of sp. relativity, and you can also derive
them as special cases of nonrelativistic QM. I certainly agree
that no one who is trying to solve a practical problem actually
does this derivation. That is not necessary, because anyone
who knows the subject has sufficient confidence in people like
Einstein, Feynman, and their ilk that they don't bother. But
the fact that such derivations have been carried out (and well
documented in the literature) is essential.
>.... Most of 'naive physics' cannot be
>derived from classical physics, let alone QED,
As I said above, the equations of Newtonian mechanics have
been derived from QED. I agree that when asked, people will
tell you all kinds of nonsense about how things move or fall,
but those same people can play ball games and race cars at
high speed in which they use their "feelings" for physical
phenomena with far more precise correspondence with the laws
of Newtonian mechanics than they are capable of explaining in
language.
>of human expertise - law, social science, the arts, agriculture,
>psychology, biology, you name it - cannot even be usefully connected
>with physics, let alone derived from it.
The word "derivation", as I said, is probably inappropriate.
But there is enormous amount of contact between physics and
all of those subjects, and it is growing stronger every day.
Just look at the way Gregor Mendel's work on hybrid peas has
been confirmed by studies of DNA, whose structure can actually
be simulated by quantum-mechanical calculations.
And law is extremely difficult because it involves every topic
on earth -- as examples, consider the testimony of DNA experts,
computer experts, etc., in court trials.
>>I disagree with the speed with which Pat
>>dismisses ordinary language.
>
>That 'speed' is the considered endproduct of quite a few years effort
>and experience, if I may be permitted a little chest-beating here.
>(It is also in part based on having spent periods of my life in a
>state of aphasia, which is when I realised that being deprived of
>language has no effect on ones ability to think.)
I didn't realize that you had experienced aphasia, but I am
very stongly convinced of the importance of nonverbal thinking,
and I have also been working in NLP for many years.
But this is another issue that can be discussed endlessly.
Let me just say that I believe you are disagreeing with
something that you think I might say, not with what I would
say if we took the time to get into all the gory details.
>The LANGUAGE of the potter is fine. However, if one uses that kind of
>language as a guide to how to write detailed axioms, they will
>rapidly come to grief. Human natural langauge evolved to support
>communication between intelligent human beings in complex
>social/physical situations, localized in space and time. Ontology
>formalisms must encode all the knowledge that is presupposed in such
>human communication explicitly, in a framework which can support
>rigorously deductive inference and re-use knowledge across a wide
>range of circumstances.
Yes, of course. We agree on this point.
>The two 'niches' are fundamentally different,
>and place sharply divergent pressures on the languages required.
>People can, and often do, say things to one another like: "Know what
>I mean?", with a fair confidence that the appropriate answer is:
>"Yes, or at least well enough to manage for now". No such casualness
>is permissible in an ontology. Using intuitions from one world to
>guide the design of the other is a dangerous business and should not
>be done casually. The current lamentably confused state of the merged
>upper ontology illustrates this, in my view.
I agree with this point. But I also maintain that anything
that can be expressed in any formal language can also be expressed in any natural
language (but not vice-versa).
I would say that the problems of defining a suitable ontology
for NL must include everything needed for formal definitions
in physics, chemistry, biology, mathematics, law, cooking,
football, and anything else you can think of.
John