SUO: RE: Re: Question about Example in KR Book
John:
See comments below.
-----Original Message-----
From: owner-standard-upper-ontology@majordomo.ieee.org
[mailto:owner-standard-upper-ontology@majordomo.ieee.org]On Behalf Of
John F. Sowa
Sent: Wednesday, November 26, 2003 1:04 PM
To: Tom Johnston
Cc: cg@cs.uah.edu; SUO
Subject: SUO: Re: Question about Example in KR Book
Tom,
I think that we agree on a moderate position, except for the
labels. Just a few comments:
> Your definition of "realism" ("Once we have ......" below.) claims
> the middle ground between extreme realism and extreme nominalism,
> which are the versions I was defining. This grabbing of the middle
> ground so trivializes the realism/nominalism issue that Richard Rorty
> (whom we both agree is a nominalist) would be perfectly happy with
> your brand of realism -- as would Quine, another nominalist! In
> claiming that middle ground, by definitional fiat, you have not
> solved an important philosophical problem, but rather covered it up.
I prefer Wittgenstein's attitude: show people how to escape from
the philosophical puzzles and get on with their lives. I admit
that's a cover-up from the point of view of philosophical history,
but the exact positioning of any particular problem within its
historical context goes far beyond the purpose of this mailing list.
[Start TJ]
I think it's a cover-up from the point of view of semantic engineering!
I agree that the focus of this mailing list is semantic engineering,
not general philosophical discussion. Indeed, back last July, for a couple
of months,
I did attempt to explain how nominalistic insights related to semantic
engineering. The philosophical antipodes between which I recommended we
situate our engineering constructs were:
(a) realism vs. nominalism;
(b) analytic sentences vs. synthetic sentences;
(c) observations as reporting what is "out there" vs. observations as
irreducibly theory-laden;
(d) sentences vs. theories as what observations confirm/disconfirm;
(e) Aristotelian definition as listing essential but not accidental
properties
vs. definition based on Wittgensteinian family resemblances; and
(f) the need for our semantic engineering constructions to nimbly respond to
sometimes
very rapidly-changing semantics and to family-resemblance type differences
in definitions
from one business enterprise (the area in which I professionally encounter
semantic engineering
issues) to another. (Different definitions of "customer" was the example I
most frequently used.)
I raised these issues because I thought that doing so was one contribution I
could make to
semantic engineering work. I thought this, not based on any deep familiarity
with the
relevant literature, but not in total ignorance of it, either. The basis for
my hypothesis
that I could do some good for semantic engineering was that first, in
reading the
collection of articles in your Semantic Networks book, I saw no recognition
that such issues
had any import for real engineering work. Your allusion, above, to
Wittgenstein and the fly bottle,
seems to reflect that belief, the belief that all this is interesting
philosophical coffee house
talk, but that all we engineers need to do is see that when we talk that
way, we are just buzzing
around inside a bottle we can and should simply fly out of.
The second basis for my not-well-researched hypothesis is that, in reading
your comments and
those of others in this forum, I again saw the same attitude, or better
perhaps, the same
lack of awareness that if these philosophical issues are treated as
basically irrelevant
to our real world engineering work, we are likely to build semantic bridges
that collapse
under their own weight, semantic cars that crash into telephone poles
despite our best efforts
to keep them on the road.
Hence my attempts to take philosophy very seriously, in this
engineering-focussed discussion group.
And hence this current exchange, in which I felt that your characterization
of a philosophical issue
not only illustrated that cavalier "fly bottle" attitude, but also
misrepresented the realism vs.
nominalism issue enough to seriously mislead others into dismissing the
notion that such issues might
be foundationally relevant to our semantic engineering work.
By now, for those who might be interested, I think I have written enough in
this forum that my
claim for the relevance of philosophy to semantic engineering -- and
specifically my claim that
a robust, "let's get on with creating hypotheses, testing them, revising our
theories -- the whole
scientific methodology as applied to formal semantics" realism is NOT the
right starting point for
an SUO or any other kind of formalized ontology -- is a not unsupported
claim.
(And if the organizers of this forum do decide to set up linked sub-folders
for each of those
participants who would like them, I would certainly be willing to cut and
paste my stuff into a
more or less coherent narrative, and put it in that folder.)
[End TJ]
> So in the end, I claim it is possible to acknowledge the force of
> the nominalism presented by Kuhn, Quine et al, and still put one's
> trust in the laws of science and technology, to get in one's car
> and drive to the doctor's office where one will be treated with
> a possibly invasive technology that one understands very imperfectly.
What I object to in your statement is the phrase "force of nominalism".
I believe that what makes Kuhn, Quine, et al., significant is their
underlying acceptance of scientific realism. Quine, for example,
recognizes the importance of set theory, which seems to be innocent
enough on the surface. But from the assumption that sets "exist"
in some sense, you can construct all the Platonistic forms and show
that they "exist" in the same sense. Once they accept set theory
and the claim that scientific laws reflect something real, you can
just as easily say that they recognize the "force of realism".
[Start TJ]
In the way you have defined "scientific realism", no reasonable person
could fail to accept it. Kuhn, Quine, and even Rorty and myself, accept it.
The trouble is that your definition covers up the very philosophical issues
that have made realism vs. nominalism debates so enduring, and so engaging
to so many excellent minds. Even more importantly, for us in this forum, it
covers up
what makes those issues relevant to our engineering concerns. It explains
why you
think we can leave philosophizing such as we are doing here behind us, when
we get down
to work as real engineers.
For example, it covers up the issue of what it is that
experience/experiments
confirm or falsify. Is it individual statements? But no individual
statements are
free of theory, and so how is "seeing that they are true" a support for that
theory
through whose conceptual lenses we are doing the seeing?
Is it a theory as a whole (however that is to be demarcated)? If so, any
statement can
be held immune to revision in the face of apparently disconfirming evidence,
just by making
enough adjustments elsewhere in the theory. But then in what concrete,
realist
sense are theories ever falsified? As Kuhn argued, the geocentric theory
collapsed, not under
the weight of conclusive evidentiary refutation, but under the weight of the
adjustments
it had to make to accommodate the flurry of new evidence available through
that new
instrument, the telescope.
We science/engineer types do tend to think of science as the first kind of
thing, as the
confirmation or falsification of individual hypotheses. Quine's comment on
this tendency is:
"The scientist thinks of his experiment as a test specifically of his new
hypothesis, but only
because this was the sentence he was wondering about and is prepared to
reject."
(Pursuit of Truth, p. 14)
It is with science as this second kind of thing that we down-to-earth
engineering types get
nervous, because it seems a small step from this to the conclusion that all
theories are "relative",
all theories are equally supportable, given enough ingenuity. Take this over
to "theories" of ethics
and morality, and now the man in the street gets nervous. And hostile. For
there is real hostility to
nominalism, something well beyond a strongly-held different opinion. I
believe I detect this hostility
in some of the things you, John, have said about Rorty. I think it is
apparent in a couple of the articles
Susan Haack wrote about Rorty. And this is my hunch about where that
hostility comes from. And also about
why nominalists get called "relativists" -- a bad name if ever there was
one! For relativists about science
and knowledge are clearly stupid, while relativists about ethics and
morality are downright dangerous!
And as to deriving a full Platonic set of forms from the concession that we
can quantify
over sets, I'd really need to see the proof. Quine certainly didn't think
that full-blown
Platonic realism followed from his concession that there are sets.
[End TJ]
TJ:
> Conclusion: what I called "nominalism" and you called "realism" is
> nominalism, pure and simple, from the realism/nominalism and truth
> theories perspective....
Now you are trying to force me to accept your label. I would prefer
to say that Quine and Kuhn are closet realists because they accept
science and set theory. Rather than fight over these issues, I would
just prefer to drop all labels, except perhaps in a discussion of
philosophical history.
> ... Scientific methodology is more realist the more stable background
> theories to an experiment are. It is more nominalist the less stable
> those background theories are.
[Start TJ]
On labels. Yes, I was. Sorry about that.
On my quotation about scientific methodology being more or less realist,
more or less nominalist. I don't mind dropping the labels, provided we
don't drop my point along with them. The point being that scientific
methodology is not the simple enterprise I believe you present it as.
Scientific
methodology encompasses both kinds of science I described above, not the
simple
Popperian science in which what is falsified are individual statements, in
which
the force of disconfirming experiments cannot be deflected onto other parts
of the
entire theory than the one hypothesis the experiment was designed to test.
When we think of that Popperian kind of science, scientific methodology does
seem
based on solid, common-sense, rock-kicking realism. But that kind of
observation
and experimentation, and the hypotheses related to it, are only one part of
science.
They are the part closest to the sensory periphery of Quine's conceptual
sphere.
The other kind of science is perhaps best exemplified these days by string
theory.
Most of the predictions of string theory are not yet testable. (A complex
statement
to unpack, meaning something like "there are not yet inferential links
leading from
highly theoretical statements of string theory to statements describing
observed
outcomes of experiments that the community of involved scientists would
assent to.")
Such theoretical statements are around the middle area of Quine's conceptual
sphere,
with basic logic and mathematics being close to the center of the sphere.
When we think of this kind of science, exemplified by such statements as
"Physical reality
exists in eleven dimensions", it isn't quite as clear how rock-kicking
common-sense
realism applies to it, nor how any individual experiment would falsify it.
Now it becomes
possible to understand how a lot of experimental evidence can be treated as
disconfirmatory
by one group of scientists, but perfectly consistent with a favored
hypothesis by another
group. Now "scientific methodology" is not the simple, "let's get on with
it" thing I think
you present it as.
[End TJ]
I would rather just drop the labels here. I certainly agree that
the result of induction alone is nothing more than a summary of past
observations. That is the classic definition of nominalism. But
an inductive hypothesis does not become accepted as a theory unless
and until it has made successful predictions about future observations.
The ability to predict the future is a classic aspect of realism.
Scientific methodology includes both.
Question about Quine: I have read a lot of Quine's work, but I don't
recall that he has ever emphasized or even mentioned that science
makes successful predictions about the future. Do you know of any
place in his opera omnia where he says anything like that?
[Start TJ]
I think Pursuit of Truth contains a few interesting items.
1) "Within this baffling tangle of relations between our sensory stimulation
and our scientific theory of the world, there is a segment that we can
gratefully separate out and clarify ..... It is the part where theory is
tested by prediction." (p.1)
2) "...prediction is not the main purpose of the scientific game. It is
what decides the game, like runs and outs in baseball". (p.20)
The crucial point I wanted to make is that all successful engineers
and car drivers make predictions about the future that they are
willing to stake their lives on. As long as you acknowledge that
point, I don't really care whether you label it moderate nominalism
or moderate realism.
[Start TJ]
What I care about is whether or not we can dismiss these philosophical
issues as
interesting conversational pasttimes, but essentially irrelevant to our work
as
semantic engineers. It seems that you do think we can do that. I do not
think so.
And so I also think that quite strongly philosophical discussions are
relevant to
the engineering concerns of the SUO mailing list. I am willing,
nevertheless, to move
my philosophical discussions over to the ONT mailing list. Since those
discussions
are often prompted by a comment in this SUO list, perhaps the right approach
is to
post a brief reply on the SUO list, referencing a more lengthy reply which I
will then
create on the ONT list.
My fear is that the creation of a separate ONT list reflects and supports
the prejudice
(as I see it) that philosophy is essentially irrelevant to semantic
engineering. My academic
background as a philosopher, and my work as a designer of databases, tells
me that it is not.
So as most of my future stuff goes over into the ONT ghetto, I hope that a
few of you knowledge
engineering guys, at least, won't forget to visit there occasionally.
Thanks to all with whom I've had these interesting exchanges.
Tom
[End TJ]
John