Re: How we train numerical programmers

[Gerard Jungman <jungman at lanl dot gov>]

Eleftherios Gkioulekas wrote:
> 
> Violence ain't the answer. :> It seems that the real problem is that
> academia is being forced to serve the corporate interest, instead of
> the public interest. Our plight is a side effect of that reality.
> Software is a tool: purchase it, or shove it onto your poor grad students.
> Once upon a time, software WAS research. Do you have any real ideas on
> how to fight this evil?

Hi. Hope you don't mind I am posting this reply to the
gsl list. I've had a few replies to my comments, so
I thought I might as well say something in the open.
It's definitely off the gsl topic, but what the hell,
I've earned the right to stand on the gsl soapbox
for a couple minutes.

I just stuck a short paper on the web. It's nominally about
some project I am working on, but it's really just a thinly
disguised attack at the way things are done in that particular
field. It's at:
  http://t8web.lanl.gov/people/jungman/spf-open.pdf

The executive summary is that, if people realized that they
could use software to communicate their ideas, instead of
creating streams of plots on dead trees, we might all
benefit.

The question is how to support this sort of activity.
Without some kind of incentive, the culture will
never change. So, one possible answer to
your question is that we should try to create
software which encapsulates expert knowledge in
our various problem domains, and which is
flexible enough and extensible enough that
people could imagine adding to it (in the form
of drop in packages, modules, ...) whenever
their research creates something worth sharing.
This would not be perfect or comprehensive, and
it would not solve all the problems, but maybe
it would be a place to start. Maybe it would
give a handful of graduate students a place
to start, instead of forcing them to reinvent
everything in order to get anything done.

As far as academia becoming slaves to the corporate
world, I think that is also true, especially in academic
computer science. But even more disturbing is the
global tendency of academia to try to model itself after
the corporate world, as if the suits had some kind
of secret for guaranteeing progress. This is a sad
sort of perverted social darwinism (some things
never really die). Maybe the corporates do understand
a certain kind of progress; certainly if they
don't make something they can sell then they are cooked.
But there is no guarantee that this sort of progress
is desirable for or applicable to academic research.

We all know the realities of the funding systems
which lead to this kind of "product oriented"
research. But in the end these systems are
supposed to benefit people, not the
other way around.

Anyway, that's even more off topic. So I'll
stop with that.


-- 
G. Jungman