This is just an observation, but I think it's worth being aware of--
especially by educators.
During the last year of the John von Neumann Supercomputer, I served as
consultant on numerical software. This is not because I was competant to
do so, but because I appreciated numerical analysis.
Would you use the Gaussian elimination program right from Press's and
others' _Numerical Recipes_ to solve a dense linear system of order
6000? People tried it. I found them only when they phoned to
complain angrily that the attempt took all their money.
Those doing things like this were scientists in what has been called the
"softer" sciences--those requiring little math in the curriculum beyond
a first course in linear algebra. These people were competant; yet they
always believed that more was better (more accurate), and didn't understand
why a method they were introduced to (for pedagogical reasons) could fail
or prove hopelessly inefficient for certain problems.
Don't mistake me: I'm referring to many scientists using supercomputers.
Now I was only a User Consultant: were I an Operator, I would probably be
aware of many more from runs being suspended for page block and other reasons.
My opinion is that American universities are forcing students to take too
many specialized courses, in an attempt to bring them to research level by
the time they become graduate students. It is far more likely that a student
with a liberal arts education, who has studied the foundation of his field,
will teach himself these specialized topics, than the overly specialized
student will teach himself a fundamental topic in mathematics--or possibly
Bruce (Gypsy Scholar)
Department of Geological and Geophysical Sciences
Princeton University, Princeton, NJ 08544
bathurst at phoenix.princeton.edubathurst at pucc.bitnet !princeton!phoenix!bathurst