The should I am shoulding on the article is that it ought to account for the facts on the ground. The question of whether computer science curricula in the abstract ought to include calculus is a different question than should something which purports to be equivalent to a computer science curriculum include it - particularly when four of the five reference degrees include calculus and the sort of person who might pursue this track is subject to Dunning Kruger effects.

Now my personal opinion is inclined toward calculus being part of a computer science curriculum in the absence of compelling evidence that it should not. My reasons are:

+ Calculus is where algebraic skills really get practised to the point of competence because there they are tool rather than an end in themselves.

+ One of the longer edges at the edge of computer science as a science encompasses signal processing and that entails calculus directly or indirectly via statistics.

+ If computer science is a branch of mathematics, constructing an argument is left as an exercise for the reader.

Against calculus the primary argument is vocational. But as with any science curriculum these days, the product is graduates not scientists. We with half of all graduates below average.

> I would appreciate if there was empirical reasoning for why Calculus should be taught in Computer Science rather than the argument be "we need Calc because we're supposed to and every one else has it".

(1) Research in Computer Science. Don't expect to get very far in research in any STEM field without good knowledge of calculus.

Calculus is a pillar of Western Civilization, one of the all time crown jewels, and the root of the most important part of math, 'analysis'. Discard 'analysis' at high risk for the future of your career.

We're talking about the future here, right? That is, you are in education to prepare for the future, right?

E.g., the last paper I published on computer science was heavily based on measure theory, that is, a grown up version of calculus.

At present, commonly some of the best research opportunities are to further 'mathematize' computer science where calculus is just baby talk broadly assumed.

(2) Keeping up in Computer Science. To keep up, you will need to read, and authors will feel free to use some calculus.

E.g., maybe you are interested in models of viral growth on the Internet. Okay, you can quickly come to the differential equation

y' = k y (b - y)

with y(0) known. Just for viral growth, some calculus.

(3) Applications. E.g., maybe you want to do some 'machine learning' using, say, maximum likelihood estimation. There need to solve a problem in non-linear programming. A key part of that work is the gradient, that is, calculus.

In robotics, drones, and other forms of mechanical automation, well need some calculus.

That seems to be setting up an barrier for Calculus that nothing else has to face. Is there empirical evidence correlating an operating systems course to success in the field?

More generally,a lot of the value to the student in a degree is signalling -- not only the bare fact of the degree itself, but also culturally. Even if an online education can't provide the first part yet, it can provide the second one. Dropping a reference to the Church–Turing thesis into the conversation at an interview might get you the job. Such "success" is independent of whether or not knowing the lambda calculus is actually helpful day to day once you're there.

An OS course is directly under CS. Getting asked OS questions during a systems interview is normal. Getting asked calculus questions in most CS interviews (except the specific areas of machine learning and graphics, and even then pretty much never) is unheard of. When's the last time you were asked to differentiate something, integrate something, use Taylor series, double integrals, etc? Having to maximize/minimize some function might come up, but probably in an abstract sense.

I think CS really needs to be split into two paths; software engineering and (for lack of a better term) software development. I use math on a daily basis. Today it's statistics, yesterday it was calculus (motion control tuning), tomorrow it may be trig.

What I do is very different than what a web developer may do. Not better or worse, just different, and it requires a very different skill set. The fact that we not get roughly the same education is ludicrous.

How are you going to understand statistics on a level to be able to build programmable models without an education in Calculus?