Friday, April 27, 2012

Computing Conceptions

From Georgia Tech, many of us have been closely and concernedly watching our southern neighbors at the University of Florida, where budget cuts and a seemingly hostile dean conspired in an attempt to decimate their well-respected computer science department.  Dismantling the department, at a time when computer science research is so critical, would be a laughably bad decision on the part of the university.

I don't claim to fully understand the rationale behind this plan.  However, I have a feeling that part of the reason such an idea was even being seriously considered has to do with a couple misconceptions of what computer science research is, and I fear that such misconceptions extend beyond the state of Florida.  And I don't just mean that the average person doesn't understand computer science; that's to be expected. I mean that many academics, even scientists, don't understand the basics of what computer science is about and therefore tend to devalue it, especially as an academic discipline.

First, many people seem to assume that computer science is just programming or fixing computers.  As a graduate student at Yale, where computer science is a small department, I was often asked by other Ph.D. students why computer science even has a Ph.D. program.  They didn't view it as an academic pursuit, but more as a trade skill.  I fear that many scientists view computer science as limited to programming or getting computers to work, probably because that's the way most, say, physicists use computers.  They have little understanding of the beautiful, deep results and insights that computer science has brought the world.  Viewing it as an instrumental non-academic field, people think it would be okay to kill-off computer science research and leave the professors to teach programming (which, admittedly, is an important part of a computer science department's job).

(clip art from here)
something computer scientists do not normally do

The other, very related, misconception, one that was clearly in play at the University of Florida, is that the computer science department was somewhat redundant because the electrical and computer engineering department already has the word "computer" in it.  Their reasoning sounded more sophisticated than that, but only superficially.  But computer science and electrical engineering are very far in their central concerns.  Computer science, for the most part, is divorced from concerns about electricity, physical media, or anything of that sort.  Whether you work on operating systems, machine learning, or the theory of computation, you mostly don't really care about the underlying hardware, whereas electrical engineers do.  Greg Kuperberg, writing on Scott Aaronson's great blog post on this issue, puts it better than I could:
"Apparently from (Florida engineering dean) Abernathy’s Stanford interview, and from her actions, she simply takes computer science to be a special case of electrical engineering. Ultimately, it’s a rejection of the fundamental concept of Turing universality. In this world view, there is no such thing as an abstract computer, or at best who really cares if there is one; all that really exists is electronic devices.
[...] Yes, in practice modern computers are electronic. However, if someone does research in compilers, much less CS theory, then really nothing at all is said about electricity. To most people in computer science, it’s completely peripheral that computers are electronic. Nor is this just a matter of theoretical vs applied computer science. CS theory may be theoretical, but compiler research isn’t, much less other topics such as user interfaces or digital libraries. Abernathy herself works in materials engineering and has a PhD from Stanford. I’m left wondering at what point she failed to understand, or began to misunderstand or dismiss, the abstract concept of a computer."
(image from here)
something usually not of research interest to computer scientists 

It looks like a disaster in Florida has so far been avoided. And with each passing year, more scientists will have, at the very least, taken some basic computer science in college -- it is part of our job to teach the important concepts in our introductory courses.  I'm hoping this will improve the perceptions of our field.  But in the meanwhile, it has become apparent that we have much more PR to do to.

(image from here)
now we're talking!