Joel Spolsky published yet another complaint about what they teach people to get a Computer Science degree. I think he is right in complaining that no university is producing the kind of programmers he wants, but he’s missing one point.

In Argentina there are two different careers related to programming: Computer Science, and Software Engineering. Computer Science, like its counterpart in USA produces scientist. Scientist are people not very much concerned about what’s practical or useful, but by advancing knowledge.

You don’t expect physicists (scientists) to build a bridge. Although they may understand all the forces at play, they don’t have the practical training. You have civil engineers that know how to build a bridge. Civil engineers, on the other hands, don’t play with subatomic particles, the beginning of the universe and black holes. They generally don’t advance knowledge, they build practical things

In the same vein, one should expect nothing else of a Computer Scientist than to use Haskell, push the advance of  type inference, experiment with artificial intelligence, dream of computers with a teracores (that is 1012 cores) and know nothing about deploying servers, Microsoft tools, etc. And you do expect a Software Engineering to know how to use Java, C#, Python or other current languages and never touch Haskell. They should also be able to organize themselves using agile or whatever to produce working practical products.

The Computer Science and Software Engineering careers in Argentina more or less reflect that. It’s not a clear cut but in CS you can find lessons in Artificial Intelligence while on Software Engineering you even find some lessons about laws. Not sure what they are about, but a Software Engineer does require some basic knowledge of licensing.

I thin Joel Spolsky and many others are right about complaining that universities don’t produce software engineers, but I think he is wrong about expecting them out of Computer Science departments. It would be very sad if Computer Science turns into Software Engineering and there’s nobody to dream of type inference and teracores.