Post number 7,819,394 talking about how useful or not useful a CS degree is to a programming job.<p>Here are some lessons I learned by going to school that I find incredibly useful:<p>1. How to shut up and get some work done, even if I don't see the point to it.<p>2. How to shut up and get some work done, even if it isn't directly applicable or valuable to a job or something I want in the future.<p>3. How to take feedback and criticism from people who know better than me.<p>4. How to deal with someone I don't like but who has power over me.<p>5. Lots of other fascinating things that have nothing to do with my job but that learning about made me happy and appreciative of life.<p>6. A handful of things that <i>are</i> pertinent to my job, but I see this as added extras because I didn't go to school for the sole purpose of getting a job.<p>7. That a lot of people seem to think that the only things in life worth learning or paying for are those that will be useful at a job. This is a sad one.
I may have been lucky, but I don't recognize this statement when I think of my own CS curriculum:<p><i>The vast majority of what a programmer does, forming good abstractions and avoiding complexity, is almost completely untouched by computer science curriculums.</i><p>That's the <i>majority</i> of what my CS degree covered, in several different programming paradigms. Even the idea that there are "programming paradigms" is something I learned at university, from a great course called Comparative Programming Languages. Granted, it's one thing to study it and another thing to gain years of experience doing it, but it was certainly a big focus. I took a whole course on design and structure of OO programs, which mainly at the time focused on C++, but also covered some alternative OO models and what pros/cons various people have claimed for this style of abstraction. And another one on design and structure of functional programs, which in the case of the one I took was mainly from the ML family perspective. These had some more "theoretical" content as well, especially the functional-programming one had quite a bit on type systems, but the overarching goal of the courses was abstraction and structuring (and even the type-theory stuff was introduced as a means to the end of useful programming abstractions).<p>When I read these kinds of summaries of computer-science degrees, I get the impression people went to some kind of program that was 100% algorithms courses, proving big-O running times and termination properties and that kind of thing. I thought I went to a pretty theoretical one (vs. one with more of a software-engineering bent), but it was <i>maybe</i> 20-25% algorithms. Where are people going where they are taking nothing but Algorithms 101 through Algorithms 1601 for four years?<p>I do agree that the standard tech company interview process seems to mainly test the algorithms part of the CS curriculum for some reason. I think tech interviewers love big-O notation tests more than actual CS academics do.
So I did a 4 year computer science degree about 10 years ago. It's really hard to understand why anyone would bash a degree. What I still find "relevant to practice" from then is:<p>- Any good CS curriculum is basically a calculus, statistics, linear algebra, real analysis, combinatorics, discrete mathematics course with a bunch of CS modules thrown in on the side<p>- All the computational intelligence stuff we did (particle swarm optimizers, neural networks, classification algorithms, data mining, ant colony optimization, genetic algorithms, etc.) these things were all cool but not really mainstream back then. It's all the rage right now. I'm really glad I have a good theoretical background in all of that now, it makes it so much easier to jump back into machine learning and analytics. The math I did makes it even easier still.<p>- I use a lot of "full-stack theoretical understanding" of operating systems, networking, compilers and distributed systems courses to reason about why things could possibly go wrong in production performance problems on large distributed enterprise applications, and how to isolate where the problem is. I do this on a daily basis and am amazed how many professional programmers don't understand what's going on "under the hood". It's almost a Dunning-Kruger effect situation where people don't even know what they don't know.<p>Then this doesn't even begin to mention things like formal methods, software architecture, computer graphics, database theory, etc. Oh, and the philosophy courses, which really went into logic and critical thinking.<p>A CS degree is completely mind-expanding if you get it from a good institution.
People don't understand what computer science is. Don't claim you did CS and it didn't teach you to become a developer, that's not what CS is, and I'm surprised that after a bachelor's and master's the author doesn't acknowledge that.
Interestingly enough, I think the misguided constant push to move CS degrees towards industry relevance more and more makes them more and more worthless -- we have academics going to industry conferences and asking what kind of developers would be preferable to churn out, students whining that they aren't 100% ready to drop in at a company when they graduate. So schools are pushing teaching how to write CRUD apps in java or "a class on the newest hottest mobile app development platform" or "version control" -- all stuff that should be fairly easy to pick up in a day or on the job for someone who knows theory (version control), or has no place in a University (the rest) when they should really focus on the hard stuff, like CS theory (PL, theory of computation, distributed, compilers, algo, concurrency, mathematics), which is MUCH more difficult to learn outside of an academic environment, is universal, and translates to every language and development paradigm whether it's now or 50 years from now.
As someone that had a highly theoretical computer science education, worked in the industry for three years, then returned to academic CS for summer research, I have a few thoughts on this.<p>1. A highly theoretical CS foundation is still very useful. One of the hardest classes I've ever taken in all my studies was an upper-level proof-based algorithms course. Nowadays I can glance at an algorithm and give you it's runtime complexity. This is actually an important skill because it influences how you construct your own methods and such. There were quite a few times at my old employer where I'd be like, "If we just switched this out to a hashmap, we'd have a constant time solution instead of this polynomial one," or something of the sort. Furthermore, that algorithms course changed how I thought about computing (and life) forever. Computational complexity theory is simply wonderous. I'm in med school now, and I will tell you, learning about human biology through the lens of a computer scientist is a truly enriching experience. Computation is everywhere!<p>2. Your everyday programming jobs probably don't require that deep-cut CS coursework BUT there are plenty of specialized jobs that do. I can think of many sectors where you need to be fluent in something like graph theory and ways to traverse graphs, or statistical methods + machine learning. Not all jobs are full-stack dev jobs. Undergraduate CS is supposed to lay a solid foundation that you can build your career on. It's not vocational training. You have no idea where you'll end up! And honestly, I think there's plenty of hindsight bias going on here.<p>3. Man, oh, man, people in academia suck at software engineering (and not to mention, product design). This isn't their fault! They haven't worked on a large codebase before. They don't understand git-flow or continuous integration or test-driven development. Their products need to get A's, not sell to a consumer. I will say, coming from the industry into back an academic environment it has been a bit of culture shock. Tools used, architecture, even ways of speaking <i>about</i> the software are often completely outmoded.<p>All in all, I'd say both are very desirable. I'd recommend self-taught programmers take some classes on the more theoretical/mathy side of computer science. And I'd recommend that the academically-minded CS seniors FOR THE LOVE OF ALL THINGS GOOD take a year off and work in the industry before pursuing grad school.
A degree in literature doesn't teach you how to be a best selling author or a get a job writing tv scripts. On the other hand becoming a good a programmer is much more difficult when you don't understand the fundamental inventions upon which everything else is built. You're limited by what you see in front of you in stack overflow answers and blog posts, with no insight into the why's, only the how's. Self taught programmers that progress beyond simple web pages and crud applications will undoubtedly come across a lot of computer science on their journey. Maybe you don't have to study Milton to be the next Stephen King, but your life will be a lot richer for having done so.
"I hope to see the industry improve in this respect, but in the meantime I'm happy to exploit this imbalance as a competitive advantage."<p>I wonder how much of requiring a computer science degree is simply because the people hiring have computer science degrees. Seems like a self perpetuating insiders club.
<i>"However, I also have to acknowledge that the scale and difficulty of the problems I've worked on is unusual compared to the vast majority of programmers"</i><p>It's likely that the vast majority of programmers <i>wish</i> they were working on these interesting and difficult problems, but instead they are writing another yet another ETL transform for an ERP system (not that there's no value in that, but that <i>was</i> my job, and it <i>was</i> repetitive and dull)<p>So even if a Computer Science degree isn't necessary for the vast majority of programming jobs, without that degree, you can only do one of those vast majority of programming jobs, leaving the really interesting problems to those with advanced degrees (or special domain knowledge)
I did 5 years of heavy CS. Never graduated.<p>The most important things I got were: state machines, appreciation for high level math.<p>Oh and an intuitive understanding of modeling and translating fuzzy things into easily modelable things. That part is super important.<p>And an intuitive understanding of how tiny spec changes turn problems from computationally unsolvable to trivial.<p>But what have the Romans ever done for us?
The value of a computer science degree does not lie in the number of direct applications you will encounter but rather in the school of thought and way of thinking it molds in you.<p>Often, people take for granted how differently you approach problem solving coming from a CS perspective. If anything, I have gained mostly in the ability to break down problems into the smallest pieces possible that work together thus reducing complexity and shaping the problem case. I can isolate core problems and through algorithm design implementation evaluate best solutions. I therefore wholeheartedly disagree with the notion that CS does not teach how to reduce complexity. It does.<p>Above all, I have rid my decisions of emotion and solely focus on the problem at hand. It's brutal, rough and inconsiderate at times, but it produces the desired results.
Can we all agree that college is about getting a knowledge foundation, a nice degree that says you can get things done, and interpersonal relationships, and move on already?
This topic comes up regularly. Yes, a CS education is not enough. You also need to learn programming. You also need to build systems and learn to think about how to build others even more. But for me it has been very useful almost daily and it can make the difference between being on solid ground vs. being out of your depth and sometimes not even knowing it.
>I highly doubt most programmers need to know how to do formal proofs of algorithms, determine asymptotic complexity of algorithms, or even know that much about data structures.<p>Indeed, but I believe that having a solid foundation in theory is important. People have spent a lot of time coming up with ways to reason about common problems, and you don't need to reinvent the wheel.<p>Additionally, while many programmers can get by with skimping on theory, your mathematical knowledge tends to limit problems you can tackle. While you might be able to get away with making some CRUD application without applying any complicated math, you will have a hard time trying to write code to localize your robot with SLAM. Knowing theory only opens more doors to more interesting problems.
I have very mixed feelings about this article. For one thing, there are a couple of different kinds of CS programs. The basic division is, or used to be, whether it was more Math or Engineering heavy. In fact, some schools attach the program to a BS in the Math department and others in the engineering with more exposure to EE and Computer Engineering. Neither of these are necessarily helpful for becoming a programmer in today's market, but they do generate very different experiences.<p>The biggest benefit for me was the complete understanding and model you get of the entire computer. How it works, what theories it is based on, why things are the way they are (in terms of the Turing model, the way microprocessors work, etc).<p>I don't believe this knowledge is helpful from a software engineering point, and I may have been a better programmer if I had spent those years learning more about how to work as a team and how to solve real-world problems. However, I do feel that it has allowed me to switch between different fields of programming without completely being thrown for a loop. It has allowed me to understand the issues that crop up when something goes wrong on a much deeper level, or why problems crop up on one part of "the stack" that I may not have understood had I focused merely on learning how to program. These are personally valuable to me, whether or not they may have been more valuable to my career.
Isn't the tricky bit that a computer science education is in fact an education in the science of computers (no surprises here), yet what most people study it for is an education in their usage?<p>Maybe what we actually need is a separate field, e.g. "practical computing". - which, I accept, would lend itself less well to University education which we have come to expect as a prerequisite for better paying jobs.
As for testing a prospective programmer, what I consider an ideal test would be something like this:<p>1. Pick the three most important lines out of this 100 lines of code.<p>2. Explain why they're important to those who're not familiar with the problem at all.<p>3. Rename its functions and variables to achieve maximum consistency to you.<p>4. Optimize the code for maximum efficiency.<p>5. Optimize the same code for maximum maintainability. (assuming the code will be used for 30 years)
"Whether someone can or cannot solve some cute algorithm problem in a high-pressure situation tells you nothing about that person's ability to write solid, clean, well-structured programs in normal working conditions."<p>It sure tells SOMETHING.Ofcourse there are better ways to evaluate a person as a programmer but I hate it when people claim something without basis.
Annoyingly, there isn't much in the way of change on the horizon for this problem, people seem to like things the way they are.<p>The problem is that there is no good formalized educational standard for software development training. So instead Computer Science gets shoehorned in as a substitute. This leads both to sub-standard Computer Science degree programs from a pure CS perspective (because they are primarily just trade school educations in disguise) which are even worse as regards to training for Software Engineering. CS is very algorithm and data structure focused, which is good in moderation for dev training, but they usually delve into realms that are not useful even in the "well it's good to have a broad knowledge base" sense. Moreover, CS doesn't typically touch at all on the many practical skills, and their fundamentals, needed to be a good dev, ranging from development cycles and styles to QA and testing fundamentals to source control use and management to proper composition and componentization of software to refactoring and dealing with legacy code to deployment strategies and techniques and then on to familiarity with actual operational systems related to all of these things. Each of those topics has sufficient depth to be an entire course or series of courses, yet they are frequently only barely touched on in CS programs, and sometimes they are ignored entirely.<p>I've been in hiring loops numerous times, and one of the quickest things I learned about hiring in tech is that a CS degree on a resume means basically nothing, there is almost no correlation between having studied CS in college and being able to code at even the most basic level. Unfortunately, I don't foresee much change in the future on this problem. CS is still a prestige degree (a 4-year degree in one of the coveted STEM fields) and I have a hard time seeing a legitimate Software Development education fitting in the mold of either an arts or sciences degree program at a liberal arts college.
I think the author has accidentally replaced the word "useful" with "being a good programmer".<p>Sure, most computer scientists become programmers, and CS degrees don't necessarily translate into programming skills. However, getting a "computer science" degree is not the same as getting a "programming" degree and I don't think that its "usefulness" should be determined by the skill someone has as a programmer after getting the degree.<p>The reason getting a CS degree does not make someone a good programmer is simple:
* A large percentage of the courses you will take will be general education (about 20% in my colleges case)
* Another large chunk will be in mathematics (say another 20%)
* One more sizable chunk will be "lost" to CS theory and "row-based" information (remember writing round-robin schedules by hand, or verbally explaining parts of TCP in class)?<p>Leaving us with maybe 40% of classes having some sort of programming lab.
And if your like me you'll specialize in machine learning, will will allow for even less exposure to things like the gang of 4 patterns.<p>However, the flexibility of a CS education is much much greater than if I had just been programming for those 4 years. If I decide I'm more interested in a career in IT security or data science, my courses will have useful. I think that someone who had instead just been programming Java and Rails web apps for 4 years (like I do now) would have a much harder time transitioning into those (potentially higher paying) fields as the market evolves.<p>Also it's important not to discredit general education. Being a strong writer, speaker will always help you. Knowing some history and geography will keep you from sounding like an idiot when you talk to people from other countries.<p>A college education is not the same thing as a trade school and shouldn't be compared to one.
Ater working with Storm and reading Nathan Marz's Master thesis, I got to say, he really is one hella of a developer. He was also very responsive on the questions I asked him on Storm's forum. I take Nathan's advice at its face value cause he is the most well-rounded computer scientist you can ever find.
If any young person is reading this I would caution them not to take such advises blindly. A "good computer science" education beats "no formal education" any day for an average person for the simple reasons.<p>1. Success in tech industry depends as much on your peers as much it would depend on your IQ. Knowing bright, well established people in tech industry is invaluable. Such connections are often built when you go to schools like Stanford, MIT etc. or even a reasonably good college.<p>2. Having a good college name on your resume is going to open many doors for you. You will realize that as a MIT grad when you talk to investors about cutting edge research in a specific field would be far more credible.<p>3. CS degree will not teach you a lot about how to code, build great products. Not more than a passionate coder writing code in his basement. I admit that.
> The vast majority of what a programmer does, forming good abstractions and avoiding complexity, is almost completely untouched by computer science curriculums.<p>I feel like in the course of working on large projects such as a parsing, lexing, compiling program, a student inevitably touches on these topics. And certainly when working with template code (such as in an Intro to CS class), they can see how nicely it was structured for them.<p>However, I agree with the sentiment that it seems like these topics are not directly addressed by most curriculums. At the same time, from my experience it seems like these ideas are applied very differently in different contexts (such as different platforms and languages), so a college class would be more abstract in teaching them. Perhaps it would be effective if they did some examples while teaching these ideas.
Some very valid points made in this post. Is CS valuable? I believe it is. I've witnessed the difference first hand in our startup but then again it's "hard tech."<p>The part I question is, the percentage of programmers that will work at a "hard tech" company is so small that, unless the company you're working at is a hard tech company, then there is no relevance to onboarding someone with CS skills. It does not diversify the team. Diversification in a team is of high value. Most engineers want to hire others engineers so that they have something to talk about. But that's a comfort thing.<p>Long story short, soft tech companies should only have one or two hard tech engineers on their team especially in the early days and only test at the skill level required when onboarding.
Another thing that bothered me about the article, although it did make some good points, is that it just assumes that the choice is either four years of employment as a programmer at some level or four years of uni. If you're trying to get a job as a developer as an 18-year-old with a fresh H.S. diploma and maybe some tech school credits, your competition isn't <i>other</i> H.S. diploma carrying developers, it's a 22 year old graduate with a CS degree.<p>If you live somewhere that you can get a job in the field directly out of High School, that's all well and good, but it's alot harder for a variety of reasons than someone who has finished a degree. Which is one reason why so many people go that route.
I didn't see university as a occupational training facility, and I was torn between majoring in CS or history.<p>What I got out of my CS program was... discovering that figuring out how systems and things worked was interesting to me. I also found the prospect of something closer to engineering was a better path to me than academia, teaching, law or some other path.<p>In terms of the relevance of specific coursework, I can't say that I spend a lot of time with advanced math or some of the other topics, but I learned my way around a UNIX system, got my first taste of large scale networking, and a bunch of other things.<p>What I learned most about in school was me.
Doing well in a STEM degree like CS or Math is valuable for two reasons:<p>(1) it usually, but not always, implies that the candidate has at least some coding knowledge; and,<p>(2) [most important] it is a good proxy for logical/mathematical intelligence.<p>As long as IQ tests for employment remain a legal grey area, employers will continue using STEM degrees as a filter.<p>The claim that (2) is the most important is anecdotal, being based on my own experience: I have a degree in mathematics, which doesn't really imply (1) to nearly the same extent as CS, and I have never felt it made it more difficult to get job interviews.
It's always fun being on a project and having the CTO say he's disappointed in you because you cannot 'back-sort a double linked list.'<p>Fun times trying to distract them from their own arrogance.
This analogy is really good:<p>> Whether someone can or cannot solve some cute algorithm problem in a high-pressure situation tells you nothing about that person's ability to write solid, clean, well-structured programs in normal working conditions. This practice is akin to hiring pilots based on how they do on an aeronautical engineering exam. That knowledge is relevant and useful but tells you little about the skill that person has for the job.<p>There's a lot of talk and debate about whiteboard interviews and this, for me, succinctly sums it up
I value my CS degree, but I also agree with the author. I don't think recruiters should ask for a CS degree, or focus on applicants having a formal CS education. There needs to be an simultaneous shift in the industry and in academia that allows for software engineering and computer science to be distinct diciplines with distinct goals: SE gets you into the industry, CS gets you into research. Most other sciences have their envineering partners, and technology should be no different.
<i>The vast majority of what a programmer does, forming good abstractions and avoiding complexity, is almost completely untouched by computer science curriculums.</i><p>It's also almost completely ignored in the typical programming job interview. Which is why when I interview people now I'm most interested in looking at code they've already written, and asking them to break down an application description into a set of data & interface abstractions.
The students you really want working for you are those who went through the rigor of a theoretical CS curriculum while also indulging a more practical skill set. Such a skill set can be honed at hackathons, for example.<p>It mustn't be forgotten that a theoretical curriculum teaches discipline to learn just about anything--if you can learn complexity theory and algorithmic proofs, chances are you can learn to create a CRUD API.
I was just thinking about this today, after I ran out of time in an online interview where I had to write code to correct the formatting of various strings (add spaces, capitalize etc). I was given 20 minutes.<p>I thought, I would have nailed this problem when I was looking for my first developer role, when I knew precisely how python regexps worked. Now of course I just look them up when I need them, which is never.
Seems silly to pay thousands of dollars to read through some books at a slow pace.<p>My employer wants to pay me to get a bachelors in ee or cs (have half of one, but dropped out when it was clear I couldn't afford to finish), followed by a masters/phd, but it's only tangentially about the 'education'. It's all about networking and being able to wave around credentials.
CS education is not useless, but less used. And it is also used-less by the same people who teach it. There are exceptions - but that number is way too small than it should be.<p>If someone wants to learn coding or programming - screw the CS degree, there are better options. I know CS degree holders who can't code more than a few lines or can't debug a large program. And there are better programmers who don't have a CS degree. YouTube will teach better programming than a CS degree class.<p>Apart from programming, if one "also" wants to learn the mathematical aspect of CS and want to work on core-stuff, go for a CS degree. And those skills do not come easy from some tutorials - a academic rigor is probably the best way to earn it. But if you use the O-notation at some jobs - you could become an outcast in the team.<p>When I got into a top college for a CS degree, I had high hopes from it. For the first 1.5 years, I wanted to quit, quit and just quit. It was hopeless - they were teaching few subjects that I was way better at. But for some reason they would not allow me to just give exams and get promoted. I already knew programming, patterns, electronics and even dealt with basic robotics before I joined college. For one example - when they started teaching operating-systems, my expectation was that they will at least give a hands-on on writing a pretty basic OS. No - they were just doing theories. Ultimately, I ended up writing a pretty simple version of a custom OS without any mentoring (and used Google) - because the professors had never even written a boot-loader to start with. I also ended up helping weaker students get upto the mark. In one of the semester assignment evaluations, I was told by a senior professor " you copied programs from the internet - how come they are so well written". I lost trust in them.<p>I did learn some stuff in the later years and came to peace. While I knew programming, I did not have a strong math background - stuff like probabilities, numerical methods, theoretical computer science and bit of advanced algorithms were the things I learnt and explored further. The most important thing was getting introduced to AI which turned out to be a huge area of interest for me.<p>However, all of that bad experience has turned me sour towards academia in general and I do not recommend spending a lot of money and more-important-time learning "used-less" things - unless you are expecting more than programming in CS. And when it comes to CS, the learning never stops.
In my years working in the computing field, I have found that individuals with a Computer Science (or related) degree to be much more adaptable to emerging technologies. Individuals without a degree have proved to be excellent at yesterday's problems, but they have a difficult time moving past their initial expertise.
So basically the guy is saying there IS value, but limited. And in a fairly neutral stance about a rather sensitive topic, he gets to advertise his book as well.I guess with degree from Stanford, he may have written a program to generate that article with a target sentiment and then insert the plug ?
If you are not a DBA and did not get a BSCS, ask yourself this question - when people speak of relational databases - what is a relation? Then look up the answer, which is rather simple (don't reply with it here, unless you rot13 it or something). Maybe ask engineers with a few years experience. Were they right?<p>Where does the concept of regular expressions come from? Can you explain why you can't use regular expressions as an all-purpose HTML parser?<p>I took 1 1/2 semesters on critical sections, race conditions and such. I deal with these problems all the time, and have seen much code where people either never learned about it, or didn't learn it properly.<p>I had the idea once to hash a short list of small numbers as a Goedel number. Looking around I see the idea was not original (in other places I use triangular numbers etc.) I would never have known about those without CS.<p>Yes you have to keep learning more. It is a good foundation to start from though.
Ive heard the value of comp science being exposed to various algorithms but im surprised that knowlege isnt an advantageous starting point? I think in long run it wouldnt matter since if you dont have exposure you will eventually pick up
> and avoiding complexity<p>God it's refreshing to see someone actually put that as one of the primary, principal activities of software development.<p>Figuring out how to do a hard thing in a simple way is not simple, it's actually really hard!