Saturday, October 02, 2004

Selecting a Research Area

The Computer Science department at the University of Illinois, Urbana-Champaign requires all new Ph.D. students to attend a seminar in their first semester. As part of the seminar, various faculty members from different research groups describe the work they are currently engaged in. This is a fantastic idea, as it gives students the opportunity to learn about all the research being conducted in the department and helps us choose which area we would like to work in.

There's one thing I dislike, though; so far, some of the presenters (and almost all the students who I've spoken to) appear to believe that one of the primary reasons to select a research area is the potential impact that your work can have. Don't get me wrong; I completely understand that this could be a motivating factor for many people. Should it be the most important, though? What about research because you enjoy it? Is creating (or discovering) knowledge for its own sake not the most powerful motivation for a researcher? Sure, it must be incredibly gratifying to see your work in daily use. But if that's your reason for research, you might have done better to join a software company and work on products which are used by thousands of people. You could have even joined Microsoft!

Note: I'm not denigrating people who choose such careers; where would the world be if we all spent every waking minute wondering about the precise relationship between NP and BQP? Sure, there are people who would love to work at Microsoft; I'm not one of them. (What about Google? - ed. Ok, you got me!) What irks me is that there isn't anyone presenting their research area and telling us that we should choose it because it's beautiful; because we might love the field, its techniques and open problems. Now that would motivate me. I get a little tired of "Join us because we're relevant!" and "Join us because we're hot today/this week/year!". That's not what I hoped for; it's not what one would expect from people as deeply in love with their field of study as many of the faculty here are.

To be fair, though, even if that's what motivates the faculty, it might be difficult to convey it to students. I wouldn't be able to explain why I love algorithms and not databases to someone who had never taken a theory course. So perhaps this isn't a good way to attract new students to your field. Still, it would be nice to hear it once in a while. And the theory group is presenting their research next week; there's hope yet!

One final comment; what good research does not have an impact? For years, number theory was considered the "purest of pure" mathematics with no practical applications. Today, computational number theory is the basis for cryptography; its impact cannot be denied.


Anonymous said...

One major issue would obviously be funding. When you work on something which eventually enters the industry, you end up creating wealth and jobs, which in turn, (is the only way besides state sponsorship) helps you mobilise funds and resources for your research. So naturally you should expect that the source of these funds would determine the direction of future research to some extent.

I couldn't agree with you more, actually. Unfortunately, industry demand has a huge impact on research, just as it does on students' choice of majors. I remember when I was in my XII, I came across a few books by Russian authors (I think Mir publishers) including "General Methods for Solving Physics Problems" (B.S. Belikov, if I remember correctly). It made me fall in love with physics and I was sure I could spend my whole life reading and appreciating the subject. Something similar happened after I read (parts of) calculus books by I.A. Maron and Piskunov. They were all theory, with very few problems, but you could appreciate the sheer beauty of mathematics.

But then, you know, how many would opt to do a BSc. in Physics and Maths. ? I certainly don't see it happening in India, where there are few prospects for such scientists, besides a handful of CSIR labs and Govt. funded research institutes(DRDO, ISRO, etc).

Of course there are exceptions, and we know a few of them were there in Pilani, but then such people are hard to come by. Hats off to them, anyway.


Davin said...

Hey, asking that "what good research does not have an impact?" question really got me thinking. I'm guessing the slow move to Neural Networks and somesort of Nano tech is the direction to go.

Nitish said...

Yeah, that's a popular and interesting area. An odd coincidence; I very recently re-read this classic talk by Richard Feynman: There's Plenty of Room at the Bottom

Nitish said...

Faraz, you're absolutely right. Unfortunately, we live in the real world where funding is a requirement for research (though theoretical Computer Scientists are luckier than most in not needing huge equipment budgets). On the other hand, this could be a good thing because it discourages pointless abstract work. To be funded, your work must either provide short-term benefits or be fundamental enough to generate new insights which will be useful in the long term. Of course, far more industries and organisations are willing to fund the former kind of research than the latter. Sad, but true.

Another thing I have to agree with is that as long as there aren't good jobs available in India for students of math, physics and other sciences, talented people won't enter these fields. I hope that the situation is rectified soon.

While we're on the subject, here's something else I was considering quoting: John von Neumann wrote, in the first paper of his collected works, "As a mathematical discipline travels far from its empirical source, or still more, if it is a second and third generation only indirectly inspired by ideas coming from reality', it is beset with very grave dangers. It becomes more and more purely aestheticising, more and more purely l'art pour l'art. This need not be bad, if the field is surrounded by correlated subjects, which still have closer empirical connections, or if the discipline is under the influence of men with an exceptionally well-developed taste. But there is grave danger that the subject will develop along the line of least resistance, that the stream, so far from its source, will separate into a multitude of insignificant branches, and that the discipline will become a disorganised mass of details and complexities. In other words, at a great distance from its empirical source, or after much `abstract' inbreeding, a mathematical subject is in danger of degeneration."

Sanketh said...

Nitish agree on all the issues you had with those seminars. I get the feeling though that you maybe making too much out of what they say. I think the professors are justified in quoting the validity of their work because most fields are not necessarily "pure" research. Somehow research has taken a whole new meaning now-a-days and I guess being at an American University gives you a better picture of that. Funding has a lot to do with it. There is a professor of mine who doesn't go around looking for money because he says almost all the donors expect him the compromise on his research. He gave us a simple statement:


If you have the money you must be doing research. Why else would you have the money? And more the money you have better the research.
I am sure there exist a few professors who do it for the love of the subject and nothing else. This professor at NC State is a good example.

I think the thing is the people giving those seminars didn't put in half as much thought into what they were saying as you did. I guess they didn't know Nitish John Korula was there :). Maybe I should let them know.

Ranjith said...

"By and large it is uniformly true that in mathematics there is a time lapse between a mathematical discovery and the moment it becomes useful; and that this lapse can be anything from 30 to 100 years, in some cases even more; and that the whole system seems to function without any direction, without any reference to usefulness, and without any desire to do things which are useful." - von Neumann

Anonymous said...

Sometimes beauty and "impact" coincide. One of the reasons I like theoretical cryptography is that it addresses fundamental practical questions (e.g. "what is a secure cryptosystem, really?") with the tools of complexity theory. The answers it comes up with are non-obvious and actually beautiful in their audacity and simplicity. Then there's the unexpected applications, like secure multiparty computation, which sound absolutely impossible when you first hear of them. So the two aren't mutually exclusive.

As to why the faculty are using the "impact" and "relevance" to sell their research areas -- maybe they believe that's what it takes to get students. I remember talking to a theorist a long time ago and mentioning I was interested in the theory of cryptography. "That's funny," he said, "most Americans aren't interested in theory." I bet if you go talk to some of these people independently, they'll give you the other side of the story.

Anonymous said...

forgot to sign last comment... -David Molnar

Jeff Erickson said...

So... um... how was the theory group's presentation?

Since none of the other faculty want to say it, I will: If you're going to study theoretical computer science, you should study theory because it's beautiful, but only if YOU think it's beautiful, only if you don't think some other more marketable area of computer science is more beautiful, and only if you are motivated more by beauty than by fashion, relevance, impact, practicality, fame, fortune, or the ability to pay the rent.

And to answer your final question: MOST good research has NO impact, at least in the long run (say, more than ten years in the future). None. Zero. Nada. Zip. Zilch.

Nitish said...

Since you bring it up, the presentation didn't happen. For whatever reason, the faculty member who was meant to give the presentation couldn't make it. Prof. Harandi (our graduate programs director for off-campus readers) said that he would try to arrange another theory presenation, but it was very unlikely. Sad, but it can't be helped. I didn't particularly want to mention it, so I was hoping no-one would ask.

As a personal statement, I think theoretical CS is beautiful, which is why I would love to spend my life studying it. A lack of fashion, relevance, impact, practicality, fame, fortune, and/or the ability to pay the rent doesn't alter that, though I'm constantly being criticised for impracticality when I say things like this. God bless my parents, though, who've never once said anything reproachful on the subject.

Perhaps I'm being unrealistic, but I wish more people were motivated by similar reasons. I guess that's a little presumptuous of me, though; who am I to comment on other people's motivations?

I'm intrigued by your last statement, Prof. Erickson. Do you mean that most good research has only a short term impact, if any? It's always possible that in a few years, someone will design a better algorithm or an improved system, and relegate the original work to musty archives. Still, perhaps the original work inspired others to take up the problem, or suggested the new approach the improved version took. It was of value, then, if only for a while. Or am I misunderstanding that comment?


Nitish said...

David, I don't know how I forgot to respond to your comment. You're right about an exciting confluence of beauty and impact in areas like theoretical cryptography. One of my friends, Sambit Nayak, (currently at the Indian Institute of Science, Bangalore) is interested in this area because he loves complexity theory and number theory. (Incidentally, I think he reads aporetic fairly often.)

I had heard that Americans weren't particularly interested in theory, too, though I don't know how true this is. The UIUC theory students that I've met are all American. Anecdotal evidence, but I wonder...

Anonymous said...

And the Arun-posting-comments deluge continues...

I think the professors shouldn't be trying to "attract" students to their field so hard...if the desitre to work in that field comes to the person naturally, it's worth a *lot* more. But some amount of "nudeging" does help every once in a while.

And've been nominated for the Be-Pestered-By-Arun-For-Admission-Related-Info second time in 2 years. This time for the Applications-To-The-US category. :D