Ask HN: Why do many CS graduates lack foundational knowledge?

Your question already hints at the answer.

The "foundation" depends on the context. Your process may be different but many companies will consider D&A as foundational knowledge as it is the closest thing we have so far to a standardized measure.

[dead]

maybe you can describe specific example? topics hinted here are broad.

concurrent programming for example also not taught same always. myself in school learned about concurrency like threads and mutex theory but “really” learned first time on job.

* Database Systems (relational algebra, SQL) * Concurrent Programming * Network Programming

These are not required courses in the CS curriculum of most schools in the US. Elective yes, but at some smaller schools, these are not even options.

Have you checked out the degree map of some of the schools? (Even the most well known ones) And you will see it.

For DB systems, I had to specifically choose this as a deep course and dived into creating a database from scratch.

I would argue those are specialized areas, not base knowledge. Moreover, what kind of questions are you asking? It is more likely that you have a misaligned assessment of an undergraduates knowledge.

The answer is simple: Those are typically optional topics in a CS curriculum. You might as well ask why an undergrad doesn't know compression algorithms, compilers, robotics, software reverse-engineering, cryptography, computer vision... All of the above are optional and will only be taken by those who are interested in them.

The more important question is: Are you explicitly mentioning databases, concurrency and networks in your posting? If not, then it explains why candidates are not filtering themselves out.

  Exposure to computer architecture and organization, information management, networking and communication, operating systems, and parallel and distributed computing.

  It seems most are exposed to them partially through project work but without the base knowledge.

I wouldn’t consider network programming to be foundational to all computer science, tbh. Are these interns expected to be writing their own sockets? Similarly, are the intern projects involving writing out the operational trees of your sql? I would consider these to be great if the intern knew of these things but I wouldn’t expect an intern to be functional in any of it. (I would put kernel writing, compiler programming, security, mobile development, etc. in the same category. If they know some and it applies to the intern project cool, but I brought an intern on expecting them to be worse than useless lol. The goal is to get them towards useful for when the company hires them.)

I think you're forgetting how computer science programs operate. You're listing a bunch of specialty skills that are likely covered once in a topic class and never again. You said "I don't mean data structures and algorithms," which excludes the actual foundational skill that they had to build on in every class, in addition to raw programming chops.

Why don't they know database systems? They might have taken a database course for 4 months 3 years ago and never touched a database again because it's not a trade school. School just validates that you can learn a series of related skills over a few months when necessary.

What's the last thing you started and dropped after a few months just before the pandemic? How comfortable would you be if you interviewed for a job exclusively on that skill?

You hit the nail on the head in your post—these are things _you_ consider foundational knowledge.

I graduated with my Bachelors in CS in 2016, and those classes were optional senior electives. You were required to take a certain number as well as some required ones (i.e., Computer Architecture, Analysis of Algorithms). I chose to take Database Systems, Data Mining & Machine Learning, Robotics, Computer Vision, etc. as electives but not Concurrent Programming or Network Programming because I already felt comfortable with those topics. Others chose classes in topics like mobile application programming or programming language theory.

Those topics may be foundational for you, but not for others.

Databases and networking were optional topics at my undergraduate institution, Cal Poly San Luis Obispo, where many of its CS grads went on to have successful software engineering careers in Silicon Valley. As for concurrency, we had a few required courses (computer architecture, systems programming, operating systems) where we learned about multi-threading and IPC. In fact, one of my assignments for my systems programming course (CSC 357 in Fall 2007) was to write two HTTP servers: one using POSIX threads and another using fork and IPC. This assignment actually helped me land an internship with Lockheed Martin, who was looking for someone with experience with writing multi-threaded code.

In general, however, fresh CS grads from even good universities need to be onboarded on software tools and certain types of systems when they have their first software engineering jobs. For example, many college students don’t know git. Basic sysadmin and more advanced Unix command line skills are also generally not taught.

for some, 'compiler construction' would count as foundational skill. But realistically nowadays it is a skill that is neither necessary, nor all that helpful in a typical day as SWE. There was a very narrow sliver of time (maybe 95-05) where this was a skill you'd be using outside of your CS class. While I do agree with your basic sentiment of 'CS doesn't teach the same things as 20 years ago', I don't think that's necessarily a bad thing. Those who like to dig to that level will do so on their own (just like not everybody likes to study analytical topology).

I agree that those are foundational topics and having graduated from a top 10 CS program can tell you that they were all optional. There were many grads who didn't have knowledge of concurrent programming at all which always struck me as a bit odd,

4 years isn't long enough for all these things + everything else you didn't list. It sounds like a long time but it just isn't enough.

Look for competency, not knowledge

Because it takes much more than an undergraduate degree to learn the field properly. If you find a graduate who is clueful that's probably because they've been hobby programming since age 12, or working in the summers somewhere they could get good experience.

Databases, network programming and concurrency (beyond basic constructs like mutexes) are not fundamental. SQL is certainly not fundamental to the major, that's a specialty only needed by a small percent of programmers.

I would suggest looking at the CS departments page from whatever university you mainly recruit from to get an idea what the core of their program looks like.

CS programs are designed with broad input from a wide assortment of interest groups -- specific industries or companies or whatever. Different universities tend to emphasize the "foundations" of the industries physically nearby. Sometimes this is done intentionally to create a pipeline of labor to large companies that need it -- more often it's subconscious osmosis.

There's a bell curve to aptitude and drive to learn. The curriculum at any university reflects the intersection of that median in the student body with a social narrative.

Those aren’t foundational computer science knowledge.

As an undergrad, my priority was learning what was most frequently on interviews and what most jobs cared about. None of this fit the bill.

SQL? Yes. Database theory? Have never discussed it beyond "what is an index"? So I never looked at that topic again.

Concurrent programming? Never dealt with it outside of courses and jobs that care about it mention it, so I self selected out. So I never looked at that topic again after the course.

Network programming? Took a course in it, but outside of a few devops use cases, I have never had any reason to recall that knowledge. I just memorized 5 versions of that test and went in to it with that.

My advice to my undergrad self would be to basically abandon anything that is not fun projects (so you get familiar with the languages themselves), hackathons (so you have culture fit), and leetcode.

I can't imagine the average ROI on learning these things is great.

A degree is not at all any basis of skills. It just is a ballpark estimate of what a person might know.

I've been doing interview of freshers since long.

Sometimes they'll know a lot of things, sometimes they can't tell a computer apart from a file cabinet.

One can't predict what a person knows based on their degree, yes it does give you a general idea but that's not the case each time.

Moreover, there are various other factors that you should consider while doing the interview, if they are freshers, like they might be nervous, they might have travelled long and came to the place, many many factors.

The typical core CS curriculum that all CS students are required to take stops around introductory computer architecture and data structures/algorithms. All of the more advanced courses are generally elective courses that students can choose to take (or maybe a requirement that you take 2 or 3 of N courses as a 'breadth' requirement).

None of the courses you listed I would expect of all CS students in an undergraduate degree, and quite frankly, databases is something I would explicitly expect few CS students to have taken (the only branch I'd expect to be less popular to take would be specialization into numerical modelling, although that's more because I expect the people taking such courses to be science majors and not CS majors).

when you say like concourrent/network programming do you expect them to be able to use async functions and make http requests, or do you expect them to have read all the works of leslie lamport XD (second is probably more like masters or specialization level i would say)

A lot of the comments opine that these are not fundamentals and they are right.

My impression is that the philosophy of CS curriculum at most places is to keep the barrier of entry low. That's why bootcamps are a thing, and my fellow CS grads are excited about ChatGPT being able to write code. On the other hand, there is a growing shortage of electrical engineers because the level of gatekeeping is too high.

I am an electrical engineer myself and I have working knowledge of all the things you mentioned.