More Students Prefer Interdisciplinary to CS
prostoalex writes "With increased offshore outsourcing and continuing simplification of such tasks as writing a trivial application, Computer Science degrees are not as attractive for college students anymore, NYT finds. Students prefer interdisciplinary majors, where the programming skills are combined with solid scientific backgrounds in biotech, chemistry or business." From the article: "For students like Ms. Burge, expanding their expertise beyond computer programming is crucial to future job security as advances in the Internet and low-cost computers make it easier to shift some technology jobs to nations with well-educated engineers and lower wages, like India and China."
CS isn't computer programming. CS is computer science.
I'm one of the few. I've done proof of correctness systems, image analysis algorithms, operating system design, game physics algorithms, robotic control algorithms, and network congestion algorithms. I've been lucky enough to be able to do this without having to work in academia. I do have an MSCS from Stanford, which is a great credential, although the education wasn't really that good.
But in most areas of computing, the basic algorithms already exist. (Some of them keep being reinvented; watching the XML fans reinvent LISP is amusing.) Not that many employers really need algorithm development people. I have no idea where you'd go as a computer scientist today. All the old labs (DEC, HP, IBM, PARC) are dead or shadows of their former selves. It's almost down to Microsoft, Google, or academia.
Actually, I'd recommend getting a strong background in numerical analysis and statistics. It's useful to know number-crunching cold. Engineering, financial, database, search, and game work all need number-crunching. It's more useful than, say, combinatorics.
If you're really into theory, you might want to take a new look at proof of correctness. I headed a team to build a proof of correctness system in 1980-82, and it worked, but it was just too slow on a 1 MIPS VAX. 45-minute proof runs for 500 lines of code. Today, that would take one second. It's time to work in that area again. There's some good proof of correctness work going on the hardware area, but not much for software.
(Incidentally, if you think proof of correctness is impossible for undecidability reasons, you're wrong.)
Why do people think it's harder to fire Americans?
Never underestimate the power of stupid people in large groups.