Is Computer Science Really A Science?

[Red]

I got into an argument with a friend the other day about this. I keep telling him it's more of an application of science, math, and logic than a science, like engineering and medicine. He talks about how in a way doctors and engineers can be interpreted as scientists in a way, as doctors use scientific methodology to diagnose patients and stuff like that.

Thoughts?

[miniboes]

The knowledge we can gather about the universe isn't neatly packed into objective categories. We just make these categories whenever we can identify a subset of knowledge which is large enough for a significant number of people to devote a carreer to. So the fact that computer science uses insights from physics, math and logic doesn't really have any bearing on whether it is itself a science or not. What matters is the systematized practice of making predictions and finding testable explanations to build knowledge about the universe. Computer science uses the scientific method to do just that, and is thus a science in my book. I definitely consider medicine a science (as you do not seem to), as well as social sciences like sociology, psychology, economics and political science.

[Jebus]

About as much as political science.

[teo123]

I'd say it's slightly more empirical than mathematics. For instance, many of the things that are commonly accepted in computer science are based on the assumption that P=|=NP, which has never been proven, but it appears to be the case (there are countless problems for which there are known algorithms to quickly check the solution, but there aren't any algorithms to quickly find the solution). In fact, given how many NP-complete problems there are (which can be shown to have a quick solution if and only if P=NP), P=|=NP can even be said to be an example of a falsifiable assertion. That's just theoretical computer science. The praxis appears to be even more empirical. You know, like how reliable an algorithm for speech recognition is...

[Red]

So the fact that computer science uses insights from physics, math and logic doesn't really have any bearing on whether it is itself a science or not.

Why not?

Computer science uses the scientific method to do just that, and is thus a science in my book.

How?

I definitely consider medicine a science (as you do not seem to), as well as social sciences like sociology, psychology, economics and political science.

Well, the process of obtaining medicine is scientific, but the application of it? Not so much.

Sociology, psychology, economics, etc. all fall under the category of 'Social Sciences,' which are more considered soft sciences, while physics, chemistry, biology, etc. all fall under the category of 'Natural Sciences,' which are more considered hard sciences.

[Red]

About as much as political science.

Oh man, you better not say that to like 25% of kids in my grade.

[brimstoneSalad]

It is in part due to methodology, principally because a lot of what goes on in software architecture isn't really comprehensible by human beings.
Once you spend more than a couple weeks on a large program it can just reach levels of complexity you can't follow.

There are a lot of unknowns in terms of how things work or where bugs come from, so programming is a perpetual exercise in hypothesizing about what might be causing the issue and developing tests to try to recreate or correct it and control for variables. Particularly when working with existing unknown code, there's a lot of modeling and experimentation.

There's not as much work with probability like reality has with inaccurate or imprecise measurements... it's kind of like physics if we had perfect instruments to give us the exact values of everything. Not to say that some bugs don't seem to be inconsistent and hard to reproduce. Generating true randomness in a computer is a real challenge, so it's a different kind of science. Somebody only in computer science might have trouble appreciating measurement error and p values for real world science.

[knot]

What criteria are you using? These?

Predictability and testability
Reproducibility

If yes, then I would say it is a hard science.

[miniboes]

Why not?

Because it would be absurd to propose that a science has to be completely autonomous, using no knowledge from other sciences, to be considered a science. By that measure physics wouldn't be a science, because physics is nothing without math.

Well, the process of obtaining medicine is scientific, but the application of it? Not so much.

So is your argument here that if the insights of a science are used in practice, it's therefore not a science? I really don't understand where you're coming from. Do answer knot's question, because I feel like you're just using a weird implicit definition of science.

[brimstoneSalad]

@miniboes I think the distinction he's getting at is the science vs. engineering one.
E.g. how scientists understand the physics and create the formulas that engineers then use to know their buildings won't fall down.
The distinction is a blurry one, though. More of a spectrum between experimentation and just crunching numbers on existing data. Only the least creative/innovative engineer is doing absolutely no science, though. Experiments of a sort can happen on paper or in computer modeling for new large load bearing designs to optimize material use or cost.