I was asked by a friend what do computer scientists do. She asked me whether we learn things like creating websites and databases. She also asked if programming is a big part of it. I then wrote quite a big answer, much of it is still unchanged in this essay. I liked so much the insights I had during the writing, that I thought other people might enjoy reading it as well.
We don't really learn how to make websites per se, I had a class about creating databases, though. I think most would disagree with me, but I think programming is almost all there is to computer science.
Usually people watch the nature and then they try to create a mathematical model that will describe and predict its behaviour. It happened -- and still happens -- to physics, to chemestry, to biology and others.
Interestingly, in computer science things started the other way around. The nature was changed in order to describe a mathematical model. The Turing Machine was a mathematical model of computation created by Alan Turing. Later on they created hardware that would simulate that model to some extent (the turing maching is less limited than a computer, even today). That hardware evolved into the computers we have today. They're easier to program than a turing machine and they get faster by the month, but they are not able to do more computations than the turing machine is. In fact, they are able to do less.
So, the computer science began in a very weird way. That's probably one of the reasons that many computer science papers are -- still -- very lacky when it comes to the scientific method . But that's disgressing.
The root of computer science is the study of the ways of doing computation. Besides the Turing Machine there's also lambda-calculus and, in fact, each new language is a new way of doing computation. Though they're all postulated to be equivalent. The programming languages are there to make our job easier at writing algorithms (programming in the Turing Machine is a major pain). There's a bunch we don't know yet about the behaviour of algorithms and such.
Anyhow, after a while people got pretty confortable at writing programs. And they coded a lot. They developed other hardware, new ways of communication information and all that. When that started to happen, computer science started to focus on all the ways we can process the information and such. It also began to focus in the new systems people were creating. I feel that's when computer science started to look like a science.
What a computer scientist do now is to try to understand this new form of nature we created. For example, we try to understand this Internet thing (lots of papers trying to understand how it's evolving, where it's going, how it's going). We've built it, but it's amazing how little we understand our creation. People have created operating systems, but we don't fully understand what we need to do in order to get the most of it. In order to understand all this stuff we've created we need the scientific method, that's why I say that computer science wasn't a science at the very beginning.
Nowadays, a great part of the computer scientist are trying to understand what we have programmed so far and what we have to do to improve it. That's where most of the science lays, we have to understand what's going on those systems we made so we learn to improve them. Much like a physicist has to understand nature in order to change it to his likings.
Of course, I might be underestimating it a bit. But it's very hard to define a label such as "computer science". Take A.I., for instance, it has so much influence of psycology, philosophy and medicine that it's hard to define where computer science ends and the other subject begins. Usually, when there's code to be written (that is, we figured out a bunch of steps to take in order to produce something we want) we call it computer science.
 Walter Tichy, Should Computer Scientists Experiment More?, IEEE Compute May 1998, pages 32-40.