patrick.net

= 4.5 billion new Masters in Computer Science Graduates in India.

There is nothing in undergraduate or graduate studies in computer science that even remotely is relevant to software development in the real world. The degrees are worthless. All that matters is experience.

I don't get it, Columbia Univ already has online technical degrees at the same costs as on-campus.

http://cvn.columbia.edu

Thus, what's new here, aside from lowering the sticker shock?

In addition, school work is more foundational and background information. Work experience is always more valuable.

robertoaribas says

Dan8267 says

There is nothing in undergraduate or graduate studies in computer science that even remotely is relevant to software development in the real world. The degrees are worthless. All that matters is experience.

I guess that explains why all the people with graduate degrees in computer science are working at starbucks. Oh wait, they aren't. they get good jobs and careers.

The unemployment rate for recent computer science graduates is 8.7% or over 1 in 12 graduates.

But never let facts get in the way of a witty retort.

I think anyone that gets a masters in computer science then goes on to develop software for a company, is missing the point of a good online education.

Computer Science ends at the OS and compiler, it's all Arts and Crafts after that.

It's an alarming trend. I have several colleagues that have gotten their masters degree or doctorates online. What's alarming to me is that most of them don't seem to involve all that much work at all.

It wasn't uncommon to encounter full time grad students at the university that were heavily bogged down. These people getting their degrees online do it on top of a 40 hour work week and don't seem to be the exhausted at all.

I seriously doubt the standards are there. My guess is, the real universities are sick of the online diploma mills taking away from their business and this is their attempt to reclaim it. Regardless, I would always be skeptical of the credentials of someone with an online degree, regardless of where its from.

CaptainShuddup says

Computer Science ends at the OS and compiler, it's all Arts and Crafts after that.

Yes, because this whole Internet thing is just a fad. Nothing worth developing there.

Dan8267 says

There is nothing in undergraduate or graduate studies in computer science that even remotely is relevant to software development in the real world. The degrees are worthless. All that matters is experience.

Nonsense. Any undergraduate CS curriculum worth a damn covers things like data structures, algorithms, and computational complexity (just to name a few), along with a good helping of logic / combinatorics and higher-level math. The lab work includes some quite non-trivial coding (multiprocessor / multithreading, system call implementation, etc), and typically some exposure to various other specialized areas (database design / normalization, networking, AI, etc etc). And that's just undergraduate.

I actually agree with the sentiment that experience is more important than education, but as someone working in software in SV I have seen exactly ZERO of these self-trained wunderkinds out there in the wild. I'm sure they exist, but everyone working at this level has formal training and a CS/EE degree or the like.

CaptainShuddup says

Computer Science ends at the OS and compiler, it's all Arts and Crafts after that.

And the math guys will tell you that computer science ends with von Neumann and Turing. Your OS and compiler are just implementation details. :)

You're being disingenuous. There's quality work being done across the spectrum.

Dan8267 says

There is nothing in undergraduate or graduate studies in computer science that even remotely is relevant to software development in the real world. The degrees are worthless. All that matters is experience.

let me guess: all that matters is YOUR experience, right?

theoakman says

It's an alarming trend. I have several colleagues that have gotten their masters degree or doctorates online. What's alarming to me is that most of them don't seem to involve all that much work at all.

It's not like getting a masters degree gets you a better job either.

Disclosure, I have 2 bachelor's degrees.

JustInTime says

Any undergraduate CS curriculum worth a damn covers things like data structures, algorithms, and computational complexity (just to name a few), along with a good helping of logic / combinatorics and higher-level math

I.e., shit you learn at age 8 on your own.

MershedPerturders says

let me guess: all that matters is YOUR experience, right?

Oh honey, we people in the biz talk to each other just like people in every biz. When I say something, it's not just based on my experience, but also the experience of hundreds of other developers I've talked with personally, the writings of tens of thousands of developers, the personal discussions with management in hundreds of small to Fortune 500 companies, the interactions with hundreds of recruiters in dozens of recruiting agencies. And they all paint the same picture: academics means jack diddly.

Dan8267 says

The unemployment rate for recent computer science graduates is 8.7% or over 1 in 12 graduates.

Regardless of a failed start, a bachelor's degree is still needed, even if that person wants to be a paralegal, as HR depts have made paralegal's, a BA/BS-holding *white collar* designation. Thus, it's still better to have that BS in [ CS, EE, Chem Eng, etc ] but then, part-time in a law office, RE appraisal firm, etc, and then, perhaps in the future, re-attempt to move into a programmer analyst/software engineering position.

robertoaribas says

this degree is a masters degree... don't let your reading problems get in your way either!

Cause that makes a whole hell of a lot of difference...

robertoaribas says

that is what somebody who has a chip on their shoulder says.... nothing more and nothing less. You should get over your feelings of inadequacy.

You are making assumptions. That is why you are wrong.

I am happily employed and have no gaps in my resume. Well, that just blows the fuck out of your theory.

You know, it's possible for someone to acknowledge the problems that other people have without having to experience those problems himself. It's called empathy. For example, I'm pro-marriage equality, yet I am not gay, have no gay family members, and have no stake in marriage equality myself. Empathy, try it. It feels good.

Rin says

Regardless of a failed start, a bachelor's degree is still needed

Of course, no one is arguing otherwise. I've said many times on this site that a college degree is a tax for entry into the workforce.

What we are arguing is whether or not that degree, when so many other unemployed people have the same degree, is really going to help you. Having been in IT professionally for twenty years, including paying my own way through college with IT jobs, I can guarantee you that the hiring manager doesn't give two shits about your degree or your grades. There is a huge divorce between academics and the real world, especially in computer science.

Again, I'm not making judgments on whose fault this is. I'm just letting you know how it is.

JustInTime says

You're being disingenuous.

I am and I'm not. 90% of the projects out there, are just using existing libraries, then it's only a matter of semantics and syntax to instruct the compiler to build the end result. But it really has more to do with the rules of Haiku than any complex mathematical or scientific computations.

Much like how we all know or can figure out to run power tools and use various wood turning tools. But how many of us, have the knack and skills to actually build fine cabinet work or a federal style desk?

That is the way I see Software development.

I know and work with plenty of cats that on paper they are software developers, but they are only as effective at it as they instructed to do so. A good software developer who is more of a craftsman than a scientist, can spot bugs and gaps in requirements, long before the first line of code is even written.
We are more sure of what works because we've already fucked up every thing ten different ways to Sunday, we know exactly what will fail, and what will work. Through sheer experience. It's things no text book can teach you.
More knack than science.

Now of course there are projects that are processes in nature, more than a GUI to manage data and content. Those more based in science and math. But certainly an inventory management system has more to do with scrap booking, than rocket science.

Dan8267 says

What we are arguing is whether or not that degree, when so many other unemployed people have the same degree, is really going to help you.

Sure, I got that. In fact, you could say the same for any STEM major. For example, a person could graduated in geology/geochemistry or chemical engineering, but w/o a prior CO-OP at Conoco-Philips, may never find work in the energy sector. The only area where a degree in something leads directly to a job is in a health care area like nursing, PA, physical therapy, etc, where the pipeline is clear and internships are a part of the training.

As for the CS person ... if he's got a friend at a small law office, he could still do paralegal stuff, sit for the Patent Agent exam, and still have a way out of his starting predicament. Without that bachelors degree, that paralegal stint would be closed off & the Patent agent exam would be a pipe dream.

Rin says

Without that bachelors degree, that paralegal stint would be closed off & the Patent agent exam would be a pipe dream.

I know high school drop out ex-ghetto queens, that are paralegals.
That takes no skill at all. It's only a matter of being familiar with the forms for the type of legal issues they work with. They just basically do all the phone and leg work for the actual lawyer who's name is on everything. And they don't make any more money than some car insurance rep.

CaptainShuddup says

I know high school drop out ex-ghetto queens, that are paralegals.

That takes no skill at all. It's only a matter of being familiar with the forms for the type of legal issues they work with.

Up until 10 years ago, paralegals were mainly GED, HS diploma, or AA types. Afterwards, for some mysterious reason, law firms wanted bachelor's holders, regardless of the major. So I imagine those still in the field with a GED, need to part-time it at U of Phoenix (or some other BA diploma mill), to switch offices outside of let's say the ghetto.

Dan8267 says

a college degree is a tax for entry into the workforce.

Pretty much true.

Dan8267 says

JustInTime says

Any undergraduate CS curriculum worth a damn covers things like data structures, algorithms, and computational complexity (just to name a few), along with a good helping of logic / combinatorics and higher-level math

I.e., shit you learn at age 8 on your own.

Who is "you"? Are you talking about yourself? Surely you're not talking about the average 8-year-old? If so, you're either delusional or just being disingenuous.

Like I said, I'm sure there are 8-year-olds who can code circles around me and everyone I know. I have yet to meet such a person in my professional career.

Dan8267 says

There is nothing in undergraduate or graduate studies in computer science that even remotely is relevant to software development in the real world.

All of the computer science classes I took as an undergraduate except Artificial Intelligence have proved relevant. At some point even that might pay off - the Barrelfish guys found declarative programming in Prolog useful for dealing with configuration issues. I've yet to use my three semesters of calculus, three semesters of physics, two semesters of English literature, etc. although discrete math and statistics are useful.

It's impossible to do anything interesting without understanding computational complexity, data structures, and finite automata which are taught in Data Structures and Algorithms classes.

Programming Language survey courses expose people to different native programming paradigms (functional languages don't have shared state which mutates, declarative programming languages produce results by rules not recipes) and concurrency models (like the actor model where actors can send messages to each other, receive messages, or create new actors). When people have seen those things they can use them where appropriate even when they're not native (I've used the actor model for orders of magnitude better performance than I got with shared threads and state in commercial C/C++ products). Where they don't know such things exist or haven't applied them people flounder.

Computer Architecture lets people understand where real world concerns like memory locality (on NUMA systems), cache locality, and TLB locality have huge performance impacts that sometimes swamp computational complexity. That knowledge is necessary when performance matters especially in the post-web world where you might have 100,000 machines each with a $1000 annual cost of ownership and every 1% improvement means a million dollar difference to the bottom line. High performance techniques can also be exploited to get more diagnostics into production code - I know from experience what you can do with trace level logging in production systems running at 300,000 operations a second ( 300,000,000 log entries per second at 100% means 10 log entries each is a 1% performance penalty and around the measurement noise floor).

Compiler Construction teaches parsing structured data and should involve building a compiler as a semester duration project. While not of the same magnitude as projects in industry a semester is better than a week or two.

Operating Systems teaches concurrency, file systems which are data structures in secondary storage, and scheduling which are all useful for interesting real-world problems.

If I'd taken a distributed systems class that'd have been relevant too. Lamport's happens-before relationship is essential to understanding distributed systems and I used Paxos in a commercial object storage system.

There are people doing paint-by-number software that's more about gluing existing libraries together, translating business process into code, or matching operating system to device interfaces. That sort of thing is well-defined and doesn't benefit as much from first-hand or institutional experience (for example, I worked with a guy whose time at the Digital Systems Research Center overlapped with some very distinguished people and learned a lot second hand). The same attributes make outsourcing to lower-cost areas viable like Chandigarh, India where $1365 a month covers a 4-bedroom vila, food, and live-in help and make that sort of work less relevant to people living in the first world wishing to earn a living doing software.

The degrees are worthless.

While a degree only proves that the holder knew the subject matter well enough at the time of mid-terms and finals to earn passing grades, the semi-skilled labor working as contingency recruiters uses degrees to prune the plentiful resumes from less experienced people down to a manageable number.

Once past the "experienced" threshold a degree is not needed for some one familiar with the relevant parts of its curriculum and isn't going to help some one who can't put that into practice.

JustInTime says

Surely you're not talking about the average 8-year-old?

It has been my observation that the people who are great programmers started between 5 and 10 and learned everything on their own. Sure, the typical 8-year-old can't do that, but the typical 8-year-old that becomes a professional software developer can. The smart-dumb gap is every bit as wide as the rich-poor gap. The smart people today are very smart.

Dan8267 says

It has been my observation that the people who are great programmers started between 5 and 10 and learned everything on their own. Sure, the typical 8-year-old can't do that, but the typical 8-year-old that becomes a professional software developer can

So you're saying that it's like becoming an artist or athlete, where it's better to start young and use the plasticity of youth to one's advantage.

Rin says

Dan8267 says

It has been my observation that the people who are great programmers started between 5 and 10 and learned everything on their own. Sure, the typical 8-year-old can't do that, but the typical 8-year-old that becomes a professional software developer can

So you're saying that it's like becoming an artist or athlete, where it's better to start young and use the plasticity of youth to one's advantage.

Mozart, Gauss, Olympic contestants, STEM, it's always better to start young, especially when a field is extremely competitive.

I never understood how anyone could expect to succeed in software development when the first programming they did was in a college 101 course.

I didn't get my first computer until I was 28.
I was almost a geezer a geek years before I even got my first programing job.
There is math in computer programing but it's more in the syntax of the code, more so than it is knowing a formula or a mathematical proof.

I've done everything from complex GUI programing using only the GDI to creating file formats, saving to various formats, sending and receiving bits of information in buffered increments, telephony programs, sapi, tapi, I've even built a custom SMTP service to canonicalize and munge the email headers for sending email encrypted with domain keys.

And I've never once felt like I was doing any math.
But then again, I don't get the whole math and music correlation either. Though I'm quite more than capable of playing a Minuet or a fugue.

Dan8267 says

Mozart, Gauss, Olympic contestants, STEM, it's always better to start young, especially when a field is extremely competitive.

Typically, pure math, as in Olympiad stuff, is what's encouraged by the school systems for the young. As for the other stuff, like the physical sciences, that's relegated for HS, whereas programming isn't even considered core curriculum.

Dan8267 says

I never understood how anyone could expect to succeed in software development when the first programming they did was in a college 101 course.

I've hired engineers who did that in their 30s and worked out great.

I've also hired people that were programming from grade school through age 25 who needed a lot of hand-holding.

Years of experience does not imply years of different experience and growth. Lots of self-taught programmers are content with a succession of small projects with limited lifetimes which don't teach the skills needed to be productive on larger commercial products that evolve over years.

Rediscoveries also don't get people as far as when they start standing on giants' shoulders.

CaptainShuddup says

There is math in computer programing but it's more in the syntax of the code, more so than it is knowing a formula or a mathematical proof.

Certainly memorizing a formula or a proof might be comparable to remembering syntax, or knowing how to look it up, but that's not the main way that programming is like mathematics.

(edited - not the main way in previous sentence)

If you can actually derive a formula from other more basic things you know. Or if you can figure out how to prove something, on your own (not memorizing some proof), these skills are a lot like what programming sometimes requires.

People who enjoy a good puzzle are likely to have an aptitude for Math and programming because of this. So it's that underlying trait.

But also, about proofs, even just being able to dig in and comprehend fairly advanced Mathematics proof. That's comparable in difficulty to figuring out someone elses perl code that's hardly commented at all and that does something somewhat complex.

marcus says

That's comparable in difficulty to figuring out someone elses perl code that's hardly commented at all and that does something somewhat complex.

When I inherit someones code the first thing I do is delete 90% of the useless comments. Most are either redundant, obvious, misleading or reminders to not forget to fix a bug that was not fixed.

It gets in my way of reading the code, I want to follow the code and understand what the code is trying to do. Comments just get in the way and makes it harder to read. Besides just because the last developer commented what the code is trying to do, doesn't mean it's the best way or effective.

The only comments I'll leave if they are calling external compiled libraries that I don't have the code for, provided what is being passed in, and returned is not so obvious. Also I'll leave comments in a service like for a semaphore and locks, as what is going on isn't always as easy to distinguish by that one line of code, unless it is commented on what the lock and unlock is achieving during that process.

marcus says

but that's the main way that programming is like mathematics.

this was supposed to read but that's NOT the main way that programming is like mathematics

CaptainShuddup says

When I inherit someones code the first thing I do is delete 90% of the useless comments

Well you're more of a programmer than I ever was, based on what you say (even if it's very difficult to reconcile that with the logic or lack of logic you demonstrate so often. Not to mention sloppy syntax and grammar relative to the usual 14 year olds skill with writing in english).

You may not have done much with perl though. I have, although like my other programming experience - mostly in computer science courses. Maybe Patrick will weigh in on this. Perl is very terse. Some tiny comments here and there are invaluable.

It has been my observation that the people who are great programmers started between 5 and 10 and learned everything on their own.

I started at about that age punching in BASIC code from magazines, and using the TRS-80 Model I in the Radio Shack store to learn a few things. Later got my own Atari 400 and wrote my own BBS software, one of the few in North Georgia at the time.

Doesn't mean I didn't benefit from going to Georgia Tech later for a Computer Science degree, learned from some people there too.

Today the vast majority of programmers I meet, couldn't give a shit about their code or their jobs, because they work for managers who are worse than Dilbert PHB.