Hacker News new | ask | show | jobs
by jstummbillig 2507 days ago
> It makes a lot of sense. People don't need a CS degree to do programming or even to be really good programmers.

Even more so (and maybe more shockingly) you don't need to be a capable programmer to get a CS degree. I have worked with people who have master's degrees from respectable universities who have not used Git or a command line in their life.

And yes, both are teachable -- but then, everything is. It is hard to quantify exactly how much inexperience in any given discipline hurts the outcome to what degree, but I feel there is a strong bias towards algorithms over practical skills, including having worked on and completed real life projects, that does not correlate well with their importance in common real world situations.

4 comments

This is an ever green topic that we aren't going to solve here, but for what is worth my take is that if you find someone that's self taught that nonetheless knows the asymptotic behavior of the common data structures and can fizzbuzz that's likely to be a better hire than a fresh CS degree applicant. Because you are less likely to get someone that "wants to work on interesting problems" or try to write a new project in Idris on the sly. But that said if you open the door to non-traditional applicants you are going to find yourself with a higher proportion that don't know O(1) from busy beaver and/or can't declare a variable in their supposed language of choice.
Kind of funny this came up. I used fizz-buzz in all my interviews for about two years (at first as a joke). Not a single new CS grad ever solved it in about a dozen interviews. The best I got was pseudo code like "if (i divisible by 5) { ... } else if (etc... "

Turns out modulo is not a well known operation.

Wow. I was using fizzbuzz as stand-in for basic ability to code problems and I guess I don't really care if someone doesn't happen to know modulo, but I am surprised. I certainly would expect modulo to be more well known that left and right shift for example.
> I have worked with people who have master's degrees from respectable universities who have not used Git or a command line in their life.

This is kind of a strange bar to set. There are entire companies that don't use git. In fact, most projects hosted on Google Code (now defunct) were using Mercurial.

While you believe this to be strange, it is just another metric.

If you've not used Git or GitHub or any of the cloud providers for Git, it's a signal. To me it signals that someone uses open-source software or has some passing familiarity with it.

As a person that conducts interviews, the things that people leave public on a GitHub profile have been a pretty invaluable signal of attention to detail or code quality. If I see someone with a dotfiles repository, it likely means that they care about their craft a bit, and it factors into a part of the whole evaluation pie.

I've worked with at least a dozen people with master's degrees who learned just enough git to get by and then went learned helplessness on anything further. I don't understand this behavior.
> I have worked with people who have master's degrees from respectable universities who have not used Git or a command line in their life.

I think you're too young to know that code management systems existed before git.

to be charitable to the GP, I think they are using git as a stand in for any version control system, and it doesn't surprise me at all that many CS grads have not used any VCS