Maybe we should be asking whether computer science as we know it is too broad of a field to meaningfully explore in a Bachelor's degree.
We should probably think of Computer Science in the same way. Instead of just vanilla CS, we could split a CS BA/BS into just as many specific concentrates: assembly, architectures, crypto, data related anything (management, safety, security, databases), etc.
But they don't have to learn it all in school. The way for CS to cut itself down to a manageable size is to concentrate on the traditional "hard core". Compilers, operating systems, plus some newer (i.e. from-the-1970s) stuff like networks, distributed systems and concurrency.
As always, people will cry that they never (ha!) do such stuff in their real jobs. But the more day-to-day stuff can be learned on the job. The stuff I listed is where university-style learning can offer a real advantage.
Stats/calc/algorithms weren't hard, I needed practice. I did every question in every textbook and followed up with the prof when I didn't understand how something worked.
Computer science actually pulls together all these different disciplines into one. By that metric, it might be the hardest. But really, nothing is easy or hard, it just takes practice.
As someone who had significant exposure to the world of mathematics for a working professional (4 years in an excellent PhD program) and who codes for a living, I find that what I did in grad school was multiple levels harder than what I do now, or what I get exposed to in the inane CS-focused interviews of some companies. The abstraction level is seldom, if ever, matched.
What about if you have to derive and prove the correctness of the formulas used instead of just memorizing formulas without background?
Funny how both of us are dismissing stuff neither of us did in school :)
I should note that I did study a year in California. There, the computer science was much harder, and the math was much easier (scored straight A's in math courses literally without any effort, and nearly failed the operating systems course that I took).
It's easy to dismiss the difficulties that arise in software as properties of the language, platform, framework, etc. that you're working with, except that we're a few decades into this whole software engineering thing and we're still inventing (and re-investing) tools to solve "simple" problems. Which, of course, is also true in many other engineering disciplines. Sure, the scale may get smaller or larger, but base issues like "this problem would be easy if I could just temporarily disable gravity" or "I've got the perfect mechanism for this if we can allow one solid to pass through another when we assemble it" still crop up with remarkable frequency.
One might reasonably start to think that estimating the difficulty of problems is, in and of itself, harder than it looks.
I picked a pretty good school and we got in depth with assembly, graphics, databases, functional programming, etc. In my time there, I wrote my own programming language and VM, a working raytracer, a neutral net project, and much more. I appreciate having a good solid base in a wide range of computing science concepts.
One of my co-workers are currently doing a degree in a much less prominent school in town, and the classes are much less involved, pretty easy, and sometimes just weird. She doesn't even bother to go classes most of the time. She really isn't learning much that she doesn't know already.
If I had gone to that school, I wouldn't have learned much that I didn't know already either.
The same is true of EE programs. My degree is in EE with CE emphasis but work as a SE now. I've interviewed a number of EE background folks from well known schools and it's very clear they had a completely different experience than I did. Their experience was much less rigorous in terms of the physics/mathematics involved.
Relatively often while trying to help a fresh-from-school hire I'll ask myself silently "when and how did I learn this" and realize it happened before my first day as a full-time software engineer. But it's fairly evenly split between coursework that has proven itself to be better-than-average or through some fortuitous internship experiences.
aside, i interned at a nat'l lab and I wonder if there are any reliable predictors for how good of a training opportunity an internship will be. I think mine was good because I got a lot of attention from two smart people who had more experience and a decent grasp of many best-practices, though they were both relatively fresh from (graduate) school themselves.
As a self-taught programmer, I think you're downplaying how much you can learn in a CS program. I don't pretend to know as much as a CS major. I haven't spent time learning compilers or DB internals, and probably haven't studied CS algorithms enough to pass an interview at Google or Facebook.
Regarding the CS curriculum, an undergrad EE curriculum is similarly shallow: it's a sampling of many different subtopics that often are only loosely related (E&M, DSP, circuits, control theory, devices, etc.). Undergrads won't be experts in any one of those fields.
Since I graduated (grad school) in EE, I've told myself each year that I'll jump from my EE job to a programming job, since I enjoy programming more. But it's getting harder and harder to justify. Will I be leaving a senior EE position for an entry-level software position? Does that make sense at 30? Why am I the oldest person in the office when I interview at a startup? Why is the interviewer fixating on compilers and database internals?
If I were to do it over, I'd get a degree in CS because that's the most direct route to the job, and education continues on the job. Everything else was a diversion. And you're seeing the grass as greener on the other side of the fence regarding engineering/physics curriculum. Learning how to solve physics problems doesn't make you smarter or a better programmer.
I might be downplaying CS programs, I just experienced one of them. In that program, I had no chance to learn compilers, operating systems, electronic circuits, assembly language, or any math. In hindsight, I'm not sure if this was the right choice. I did learn about those topics, but just in my spare time. I was able to pass my CS program while barely going to lectures (which may sound nice, but in practice I did not make any friends; it was a terrible time). I could just as well learn this on the side while studying something that is intrinsically more interesting, like physics or electrical engineering.
At the other hand, I was able to double major, and it's not hard for me to find a job now.
I found a very sweet spot in Embedded Systems programming, which by nature is more or less an intersection of the two disciplines.
I wrote an operating system and a simulated CPU implementing (a subset of) a real-world ISA in my last semester - both comparatively in-depth things. Of course I did have to choose that, and in so doing, I chose a semester of hell.
Similarly, doing an intro-level survey of all of those topics won't leave you with much; there is a concept of depth-over-breadth that I think is lost in programs' attempts to sound rigorous by bundling in everything under the sun. However, none of those things you listed are 'small' topics and could very well be majors on their own.
My school was a liberal arts college, and still required a decent amount of math for the undergrad CS degree.
I've known a few folks who had these kinds of degrees and I'm not sure their schools did them a service graduating them without a proper curriculum. At most schools with C.S. majors, they have alternative majors for people who don't want to do the intensive Maths and Sciences and they're usually called something like "Information Systems" or "Computers for Business" or similar.
My undergrad degree was a BS in Computer Information Systems, and I still had a ton of math. The main change from CS to my degree was replacing the higher end maths like Calc2+ and Physics2+ with business oriented courses like finance and economics.
I graduated from undergrad in 1999 and I think since that time colleges have tried to make CS more 'friendly' to non-geeks. This has unfortunately meant lowering the math requirement.
At the state schools in my state, the curriculum generally consisted of math up through and including three semesters of Calc + Linear Algebra + Discrete + 3 semesters of calculus-based Physics with labs + two semesters of algorithm design and formal methods (which I count as math courses) + Stats + an Operational Research course. You could also swap one of the sciences with a semester on diff-eq if you wished. I think you could get a math minor if you took something like 3 more higher-level math courses and a dual major wasn't that much more after that. It was brutal and I think the second highest number of required credits of any major at the school after EE. It wasn't uncommon to start a semester with 100 students in the class and end with 8-12 at the end.
The non-CS "Info Sys" or "Software Engineering", etc. dropped all the calc, linear algebra, OR, formal methods and required only two semesters of science classes that did not require any calculus (they were called something like "Physics for Business" or something) and no algo design. They actually had fewer classes than the CS major, but the rest of the gap was filled with major specific courses like finance or econ or whatever.
It's way more difficult(if done correctly) than probably almost every other degree except physics, math, and maybe chemical engineering.
I don't know where you got your degree but I felt like I was challenged immensely in my curriculum..with algorithms, upper level maths, cryptography, distributed systems, etc. etc.... but I really worked 80 hours a week to try to get the stuff, even though I could have scraped through.
And taking the degree has helped immensely in industry...it didn't teach me any specific technology, like a bootcamp teaches a MEAN stack, but it made me fluent in the industry...so I am easily able to pick up new tech rapidly..which I credit years of studying computer science in college.
Which is the goal of college...to get you fluent in the world you're majoring in..not necessarily any specific tech, you're supposed to learn that stuff yourself, and it never stops.
As for the industry...that was a huge disappointment. My first job I felt like 'the best office worker'...it didn't have the revolutionary cutting edge feel I wanted.
My current job is way better, but there's a lot of shit jobs in the industry, because you work directly for business people, who typically have no higher motive than money, and you're basically the best paid worker in the 'Rat Race'....
Also, there's way less status than being a doctor, lawyer, or nurse. People dont even remotely understand what we do, therefore there's no connection for them to grant us status.
I would also add that university in general, is more about approaching new and hard topics, and finding a good way to learn and digest them. In computer science you also get to do it facing fascinating topics like lambda calculus or the mathematics behind cryptography.
I actually did Engineering but work in code every day. I noticed much the same as you: it's a million little topics. Control theory, thermo, digital filters, a bit of coding, and so on. But having come at it from another side, I figure the lesson is that you can learn any topic you like, the procedure is the same. Read something, try it yourself, search for solutions. If it's hard look at the prerequisites.
I suppose it's a bit more obvious when you study Engineering that you're not going to use all of it. But after a few years you realise that you use a lot of it. There's this meme going around about "The 20 most important equations". I think I used them all either at work or in uni.
Imperative programming (learned C#, nice course), Computer science project (program something in a group), Logic for computer science, Web design, Computer architecture and networks (terrible: no practicals, just an exam), Databases, Graphics (fun, but only learned how to write shaders in glsl), Functional programming, Data structures, Algorithms, Modelling and system development (we learned working with UML diagrams; the course was way too 'soft' for me), 3d modelling, Software project (again, program something in a group. we did something for a two-man company with a delusional CEO, which went bankrupt while we were working on the project), Security, Concurrency
I filled the rest of the curriculum with math courses. I have to admit the whole thing (not just the CS part) looks so meager when I look at the courses I followed. Can't believe I spent 4 years on it.
To give a counter view, at the university I attended we went deep into the processor, effectively building a simple model from NAND gates all the way up to an optimized CPU pipeline. The class had one of the most interesting grade curves I've every seen, with no B's, and very few C's. Most grades were either A's or D's.
That would allow a deeper theoretical basis but also time to link those to reality through concrete examples. But then, you'd probably lose more of the class...
You know, graduation quotas & all.. :/
That's interesting. The most common complaint in my program from students was that we weren't actually learning how to program. Actually learning how to use a language was considered (for the most part) a "technical detail", and the course was very heavy on algorithm analysis, theory of computation, math, how operating systems function on a theory level, writing a keyboard interrupt handler for a small embedded OS to support backspace (^H) and erase to beginning of line (^U), using threads with locks, etc.
> Computer science is just not very hard, while physics, math, and engineering is.
I think that depends quite a bit on the school you go to. I think the CS program at my school was quite hard compared to the math program. My school was by no means top-tier, and I was very self conscious of that for a while. Then I met a good friend that had a degree from Berkeley's CS program, and over the years I've come to the conclusion that the main difference between his CS program and the one I went through was the quality of the teachers, and I think it was actually a fairly small difference. Much more important seemed to be what you wanted out of the program, and what piqued your interested and caused you to follow it deeper (and good teachers could help quite a bit at this point, as they might have deeper knowledge of specific subjects that could help you).
My point is that I do not believe that CS is necessarily much easier than those programs from what variable experience I have with them, but CS is fairly wide in what it touches, so unless you decide to dive in deeper on a subject, you are likely going to be getting theory plus a survey of what's available. That said, if your school focused more on that actual art of programming than the theory, than might lead to different (but not necessarily worse) outcomes. People go to college for different reasons, and a good many in my class that were complaining about not getting much experience actually programming were looking for college to train them for a programming position immediately after college. That wasn't necessarily what I was there for, but that doesn't mean that a college putting a bit more of that in their program is doing the wrong thing, they're just responding to market realities (i.e. what their students want).
> mathematical theory behind cryptography
In our algorithm analysis class, we had to write a paper on some known algorithm examining its function and complexity. Being security conscious and having an interest in cryptography, I chose Diffie-Hellman key exchange. A lot of the benefit of college is making your own opportunities.
Exactly this. While working for a Fortune 50 engineering conglomerate, a colleague participated in a Career Day at the local university. He interviewed ~50 or so students, a mix of undergrads and graduates. He claimed that of these he found about two of them who might be able to do Fizz-Buzz. That's crazy!
It is entirely possible that my school took this a bit too far (or that might have been the norm at the time). At the same time, some students were complaining that our major wasn't treated the same as others, as our lab time didn't count for the same amount of credits as other majors. For example, I believe a 2-hour of instruction and 3-hours of lab time class for chemistry was 5 credits, but 4 credits for CS, leading to more in-class time required than some other majors. I personally didn't care, and didn't even know about it until my last semester. Depending on how you look at it, it's actually not a detriment, but a bonus of the major.
That said, the electives available to CS did have more practical experience, such as the OO programming one taught using Java (I graduated probably a year or two before they switched from C++ to Java as the main teaching language).
I spent my four of my five school years in France, and one in California. Where did you study ?
I saw the quality difference in the two schools I attended. One had profs teaching by the book and the other had profs that wrote the book.
There is a great deal of difference between different schools, which I think explains much of the varying experiences people have.
[1] http://www.jroller.com/languages/entry/python_writing_a_comp...
I honestly wish I'd picked a more interesting major, like electrical engineering or physics. I feel like I could learn the things I've learned in a few months (which may or may not be true).
Computer science is just not very hard, while physics, math, and engineering is. I think the guys from other fields can be more successful programmers, just because they are smarter (more used to solving hard problems).
In computer science, the only course that required a little bit of creativity was algorithms. It was stuff mathematicians are practically trained to do (be it not in exactly the same setting).