Dear John,
There are some interesting and major
differences in our world views, which show up in this latest difference in our
convictions. Comments below,
-Rich
Sincerely,
Rich Cooper
EnglishLogicKernel.com
Rich AT EnglishLogicKernel DOT com
9 4 9 \ 5 2 5 - 5 7 1 2
-----Original Message-----
From: ontolog-forum-bounces@xxxxxxxxxxxxxxxx [mailto:ontolog-forum-bounces@xxxxxxxxxxxxxxxx]
On Behalf Of John F. Sowa
Sent: Friday, August 05, 2011 11:52 AM
To: ontolog-forum@xxxxxxxxxxxxxxxx
Subject: Re: [ontolog-forum] intangibles (was RE: Why most classifications
INare fuzzy)
JFS:> On 8/5/2011 12:04 PM, Rich Cooper
wrote:
> An engineering model is in no way even remotely
> similar to a possible world.
The difference between your two
definitions is not only insignificant,
it's almost nonexistent:
1. "A possible world is,
for the most part, a fantasy of what
might
possibly happen, but can't be experienced in this
world."
2. "a model (for us
engineers) is an abstraction of reality
subject to
validation."
The word 'fantasy' is a value
judgment. A more neutral term
is 'specification'. The term
'abstraction of reality' is
another term for 'specification'.
RGC:> Here we disagree on
nomenclature based on world views. An abstraction of reality is not
usually a specification, though a specification is often an abstraction of
reality.
A model is based on past history –
the reality comes first, not last. So an IDEF0 process model, for
example, is based on experiences in the history of similar projects analyzed
statistically, compared and merged to create a set of constraints that serve to
constrain the fantasy to something that might someday be real.
A requirements
specification (as used in engineering) is a document describing a fantasy which
may someday be realized – that specification comes first, and the
realization may or may not ever be realized. Many projects end with the
requirements specification once the estimates of cost and schedule to complete are
filled in. Other projects are only constructible using unobtainium –
like fusion for example. Richard Feynman’s statement that
antimatter is matter moving backward in time is a good example of an untestable
hypothesis with current technology. So constructing a time machine is, as
I say, pure fantasy at this time until we get some unobtainium. Likewise,
effective fusion seems to be a fantasy until we find a way to get huge amounts
of tritium from the moon, where it may not even actually be found as expected
in scientists’ fantasies.
A formal
specification of a standard (like
SQL-92) is (in nearly all cases) a description that encourages producers to
develop products and services that meet the said specification. In math,
you use the word “specification” in a formal way such that a
possible world is anything that can be forward chained from a set of axioms and
rules. In engineering a specification is an elaborate document with
equations and NLP statements that codifies the nature of a fantasy in vivid
language, yet to be realized, and will ultimately be only an approximation
thereof, not a rigorous mathematical construal of a set of axioms and rules
whose closure is adequate for implementing the desired fantasy.
My sentence above “a model …
is an abstraction of reality” is only the first part of the
definition. The “subject to validation” phrase is more cogent
to the definition than the “abstraction of reality” phrase for
those reasons above. I understand that you are used to a certain format
for mathematical language which can make a complete definition in a single
sentence or equation cast as a definition, but it takes much more than that to
complete an engineering definition. There are many more “is-a”
terms conjoining the specification part.
The only question is whether there is
any detectable difference
between the two qualifying phrases.
As I discussed this above, I consider the
differences to be huge, not just detectable, but different in quality and
structure.
For the first, the term 'possible'
means that the specification
is sufficiently similar to our real
world that it could be
realized by some modification.
Before that modification has
been made, it couldn't be
experienced. But afterwards, it could.
I prefer “by some realization”
to “by some modification” since modifications may not often be
necessary, but the specification is only partially complete by definition in
engineering, and not adequate for constructing even a plausible design
document, which is much more detailed and elaborate. Specification
requires about 8% of a system development effort, while design, if properly
performed and validated, takes about 30% and testing takes about 50%. The
other 12% is chewed up in document production for the various developmental
stages.
For the second, the term 'subject to
validation' means that one
could verify that the specification
is consistent with the laws
of the world (i.e., sufficiently
similar) that it could be
"validated" (i.e., realized
by some modification). Before that
modification has been made, the model
couldn't be experienced.
But afterwards, it could.
Engineering specifications for most
purposes aren’t expected to conflict with laws of nature in any way, but
often they do in subtle ways not evident in the specification until realization
brings out the errors and inadequacies of the specification. Even large
projects that are supposedly new (think the first space vehicles) are based on
well known laws, but the closure of the specifications contain many
requirements for unobtainium and inconsistencies that are not known in the specifications.
Those two qualifying phrases were
stated in different ways,
but their implications are either
identical or sufficiently
similar that they have a very high
overlap.
Again, if you use the term “engineering
model” I disagree. If you are discussing math, then you may have a
valid point.
I'll admit that some modifications
necessary for some possible
worlds could be difficult or
impossible to achieve by modifying
our current world. An example
would be a trip back in time to
meet Beethoven, Julius Caesar, or
Moses. But it could be solved
by an engineering model of a time
machine, which would be equally
difficult to validate.
There is no engineering model of a time
machine; it is a fantasy specification not an engineering model.
So the only difference between the
two is a matter of degree.
If we allow mental models and/or
virtual reality as well as
ordinary engineering models, that
difference vanishes.
If we allow virtual reality, we are
indulging fantasies. Only real reality is the end product of
engineering. It could be an engineering task to develop a virtually reality
simulation, but then the reality is the construction of the virtual reality
simulation, but the virtual reality is not itself the engineering part.
> The world is much larger than FOL and ontology
can
> possibly enfold.
I'll admit that the world may be more
complex than we can
imagine. But any specification
that we can imagine and
put into words could be translated to
FOL.
John
Someone famous (you probably remember
who) said something like “reality is more complex than we can even
imagine”. Do you recall the exact phrase, and perhaps who it was
that famously said that? Engineering is concerned, in practice, with what
is not known about a specification’s applicability to reality, and which
must be planned, validated and tested to become functional.
I think a lot of our disagreements
are based on these differences in world view, which acknowledges the
subjectivity of our separate world views. Ontology, to be useful in
engineering must address these subjective issues or it will fail to be useful
in engineering applications.
HTH,
-Rich