Following is a note by Lotfi Zadeh, who copied a note from Pat Suppes.
I would emphasize the point that the "ideal" of exactness is a goal
that has never been and can never be realized in practice.
Perhaps God has an exact, infallible theory of everything. But the
best that we mortals can hope for is "satisficing" -- to use a term
by Herb Simon. We can always strive for better approximations, but
the assumption that exactness is possible is misguided and misleading.
All our theories are approximations, which need to be tailored for
different applications. A satisfactory approximation for one purpose
may be totally unsatisfactory for another -- and vice-versa.
In fact, *every* complex system of any kind uses different theories
with different approximations for different parts, subsystems, and
modes of operation.
Just imagine all the subsystems of an airplane. Even for airflow,
many different approximations are used for the wings, fuselage, and
interiors of the engines at various speeds, altitudes, and maneuvers.
This point does *not* imply that we must use fuzzy logic and fuzzy
reasoning. What it does imply is that there is no such thing as
an ideal, one-size-fits-all method of representation and reasoning.
John
