*From:*John Day <jeanj...@comcast.net <mailto:jeanj...@comcast.net>>
*Sent:*Sun Mar 02 21:13:32 EST 2014
*To:*d...@farber.net <mailto:d...@farber.net>,sa...@dsalex.org
<mailto:sa...@dsalex.org>
*Cc:*da...@dslprime.com <mailto:da...@dslprime.com>,d...@bu.edu
<mailto:d...@bu.edu>
*Subject:*Re: [IP] Re Read re Losing a Generation of Scientists
Scott,
You have hit the nail on the head. We are not doing fundamental
research. The sciences are turning into craft. Lee Smolin
first brought this up about physics in the last 5 chapters of
his book, "The Trouble with Physics."
In CS, we have this disease in spades and partly for the reasons
you outlined below, the pursuit of the dollar. I also think to
some degree what I have come to characterize by paraphrasing
Arthur C. Clark, 'Any sufficiently advanced craft is
indistinguishable from science.' We are so dazzled by the
products of Moore's Law that we don't see that what we are doing
is craft.
The trouble with craft is that it stagnates.
The classic example is Chinese "science" prior to Western
contact. See Needham's "Science and Civilization in China." To
some degree, Needham ends up arguing (and most scholars agree)
that 'science' in pre-Qing China was more technique or craft.
There was no theory, no abstraction, no attempt at a theory that
holds it all together. (By their own admission, this problem
still plagues China and India. There are the exceptions, but in
general it is a recognized problem.)
By late Ming (17thC), it had pretty much stagnated and they were
losing knowledge. Needham says that it is because merchants
(capitalists) were at the bottom of the heap. The government
power structure controlled everything. I also believe it is
because there was no Euclid. There was no example of an
axiomatic system. The Holy Grail of a scientist is to do to his
field what Euclid did to geometry. Interestingly Heilbrun
points out in his book on geometry book that the Greeks were the
only ones to develop the concept of proof. Other civilizations
have mathematics, they have recipes, algorithms; but not proof.
Proof is at the root of building theory. Theory gives the ideas
cohesion, shows how they relate in ways you didn't expect, and
shows you where the gaps in your knowledge are. The quest for
theory is more important to avoiding stagnation as the pull of
capitalism.
Needham didn't live to see it. But we now have the example of
how the entrepreneurial drive leads to stagnation. That drive
is fine for exploiting *within* a paradigm, but it won't get you
to the next one. And we have seen the example of that as well.
And we are seeing the same stagnation in CS. One sees the same
the same papers on about a 5 years cycle. The "time constants"
have changed but they are the same papers.
Early CS was much more scientific. We went about things much
more methodically, we were more concerned with methodically
understanding the fundamentals than just building something that
worked. (BTW to your comment: We *did* do a lot of RJE on the
early ARPANET. We had many scientific users submitting jobs on
particle physics, economics, weather simulation, etc. However,
we never saw it as the future. We had much bigger ideas in mind,
for distributed computing (ask Dave). It is really depressing
that 40 years later, things really haven't moved anywhere. The
hardware is 10s of thousands times faster and bigger. You are
right. We have re-labeled RJE, cloud computing, and never
gotten past the 3270/Mainframe days.)
You are right. We do have to get back to this. And there I am
afraid it gets disheartening. We have 30 years of conditioning
the field toward everything else but. I don't see many who even
when they say we need to do it, know how to do it. We have
selected against the ability for decades. I am even finding that
CS students (and professors) have trouble with abstraction. For
a field that one could say was founded on abstraction, this is
really scary.
Take care,
John Day
