« Making virtual worlds grow up | Main | Art or fake? »

March 08, 2007

Computer science vs. Computer programming

Suresh blogs about a favorite topic (and hobby horse) of mine: the image problem that Computer Science has over computer programming. Programming is really just a generalization of calculation, so to think that Computer Science is primarily about programming is like thinking Mathematics is primarily about punching numbers into an equation, or rearranging symbols on the page until you get a particular form. Of course, that's probably what most people in the world think math is about. It wasn't until college that I realized that Mathematics was more than calculation, or that Computer Science was more than just programming. Naturally, computer scientists are annoyed by this situation, and, if they were more inclined to self-ridicule, they might call it the CS vs CP problem: Computer Science vs. Computer Programming [1].

On a related point, recently, I've been interacting with a hard-working and talented high school student who isn't sure what she wants to study in college. Much to my delight, I frequently find myself connecting problems she's interested in with topics in theoretical Computer Science. For instance, just today, we talked about random numbers. She was amazed to learn that the P vs. NP question connects deeply with the question of whether strong pseudo-random number generators exist [2,3], and also that a sender and receiver can use a common pseudo-random number generator to make their conversation difficult to follow [4].


[1] Sadly, the US Government also seems to think that CS = CP. When they say that demand for degrees in computer science is forecast to grow dramatically in the next 20 years, they don't mean that we'll need more algorithms people; they mean more IT people.

[2] The P vs NP question is one of the most important questions in all of mathematics, and the Clay Institute offers a cool $1 million for solving it, as one of its Millennium Questions. (My advisor likes to say that if you could show that P = NP, then you'd make a lot more than $1 million off the patents, and playing the stock market, among other things. The startling thing is that even this description underestimates this questions significance.)

[3] There are several great introductions to the P vs NP question, and many of them touch on the connection with random numbers. This one (starting on page 34, but see pages 37-38 in particular) is one that I just found via Google.

[4] Most scientists think of the pseudo-randomness of random number generators as either being a nuisance, or only being useful to replicate old results. Engineers - such a clever lot - use it to make many modern communication protocols more robust by hopping pseudo-randomly among a set of allowed frequencies.

posted March 8, 2007 08:37 PM in Computer Science | permalink


if they were more inclined to self-ridicule, they might call it the CS vs CP problem:

I like that :)

Posted by: Suresh at March 8, 2007 11:43 PM