Hacker News Reader

Comment by tikhonj | Hacker News Reader

To be fair, they do a pretty bad job teaching CS too. Your average undergrad program is almost entirely learning about CS, not learning CS. A non-trivial of students get through a CS degree never doing anything novel in CS—a real shame because, unlike math, research-level CS problems are eminently accessible to undergrads.

It doesn't even have to be "real" research: I just expect more than classes which have you regurgitate what you learned to pass. And yet that's all you need to get a CS degree. It's a bit like learning to write by filling in blanks.

View on HN

Your average undergrad program... CS degree never doing anything novel in CS

Is this different than any other field? I was under the impression that most BS programs do not require novel research. That doesn't come until post graduate courses.

I think it doesn't come until a second degree, or sometimes a graduate degree. Not post graduate.

Something was lost in translation here. Graduate degree and postgraduate degree mean the same thing, at least in the US.

I don't necessarily mean publishable—that's doable but difficult in CS—but novel in the sense of requiring non-trivial creativity and insight within the field rather than replaying things you were taught directly.

An example might be writing a compiler for a language you design with features not covered in the book. It's probably not publishable, but it is doing CS in a way that just following instructions in a book or from lectures to write a compiler isn't.

So personal projects that extend the material you learn in courses and take 1-2 semesters to complete? I think plenty of schools already do that as part of a BA/BS graduation requirement.

You just must have had a bad CS program, my undergraduate degree involved plenty of those types of projects.

I don't think industry needs every CS graduate to produce something novel. The work of many is just building upon the work of others.

"novel = of value" in this case.

Fill in the blank programs ("Here's the skeleton, make it display a list") and 3 projects with < 200 lines that aren't doing something CS specific are not novel, but all too common and wide spread.

A lot of things of business value go like this, for example: go build this UI in React, etc. You don't need the ability to significantly return something back to academia to do that. You just need good coding mechanics.

Not the point: college is for CS. Not for "building on the work of others." You learn the basics that everyone needs to do the advanced work. If that isn't needed/useful then you need a vocational school instead of a CS program.

Making people do continual simple demos and nonsense programs instead of teaching them actual CS is BS. They can be simple programs, they don't need to build huge things, but what they build should reveal and teach them things beyond how to shift text around the screen.

Are there vocational programs beyond those 20-week'ish coding schools? Because those don't seem to equip people with quite enough skill.

> classes which have you regurgitate what you learned to pass

I think that's what school is, no?

I'm not trying to be glib, I think this is just a reality of how education works in pretty much all subjects (typical US academia, YMMV etc.) Can you name a major that doesn't work this way?

I had to write real proofs (tho I didn't fail completely if it wasn't free of errors) for my U.S. math undergrad degree.

Did those proofs have to be novel, or were the proofs of already known concepts?

I'm sure those proofs at the time were novel to the student but obviously not the professor who assigned the problem. That's how the student learns.

Novel to the student up to a point—math students are taught a variety of proof techniques in each of their subjects, theorems that students are required to prove generally fall to very similar techniques.

They have to be novel in my experience.

I have a BS in Mathematics from a mediocre school (University of Arizona) and exams were always "prove this fact (which you have never heard of before)", never "what is the proof of XYZ theorem (which we studied)"

there is so much room for questions, it's quicker to write the proof than to search an answer. And a readily found answer would, for undergrad, less likely be in proof form, or too distinct for specific topics later on. And the instructor might become suspicious and interogate you, in which case, if the material helped you to understand, that's the point.

Reply on news.ycombinator.com