The declining value of the MS in Computer Science

I think there are a few issues here. The author points out that there are masters programs springing up whose purpose is to generate revenue. One of my professors calls these mercenary-masters. Often times, these programs segregate their masters students from their undergrads and PhD students - giving them special courses meant for that median mercinary-masters student. That makes it easy to weed out the worst offenders and these programs can be seen as less valuable than a undergraduate degree in CS.

But to put a counterpoint on it, undergraduate education in CS can also be highly variable. My university used to require 20 courses, but because so many other schools required so many fewer, they adjusted down. Heck, most schools don't require students to take operating systems (something that stuck at my school). So, even within undergraduates, I'd say there's a decent difference between someone who took 12 courses, avoiding harder ones and someone who has taken a harder program. In some ways, the same thing that is driving these mercinary-masters programs is driving departments to encourage and keep lower-quality undergraduates within the department.

One of the big problems is that there are a lot of people that want to retrain for software engineering jobs. Yes, they could learn on their own (and many do), but some like the classroom and some like the perceived safety of a piece of paper from a university. We could call these programs post-bac, but that's a lot harder for someone to market on their CV/resumé. So, what is someone that wants to retrain supposed to do? There are undergraduates coming out of top schools with 10-course majors in computer science (including required math courses). As long as the courses are taught at the same level, there's no reason that someone should be able to retrain with those same 10 courses and be seen as having gotten the same education.

So, getting back to the issue at hand, I think there's an issue of whether any CS credential carries the same weight. We only demarcate a few options (BS, MS, PhD), but within a BS, there's huge variation that isn't captured by GPA given that so many schools allow students the option of skipping the hard courses and going with electives like "Web Engineering with Rails" and "Mobile Apps for Android". But part of it is that those course titles offer something that is of value to a lot of companies. For both of those jobs, you don't need to know how a TLB works or why a B-tree is better when doing disk IO. It can be helpful, but there are a lot of jobs that don't need high CS skill.

Maybe the issue is that we don't generally have Software Engineering degrees. There's overlap between software engineering and CS, but I think there's a decent difference worthy of demarcation. Technology is seen as a good path today and I think some of the harder pieces of CS get some people discouraged out of the departments. That usually gives departments two options: water down the major or have fewer achieving the major. Maybe a balance could be struck, maybe a software engineering degree would be the way to go. Maybe universities in America need to start offering MPhil degrees (a research masters) to make the distinction.

But, even if that distinction were made, some schools would be offering software engineering under the CS name and students looking for a fancier piece of paper would send their money that way. We don't have anyone like the AMA dictating standards as they do in medicine. But maybe this is better. Maybe this means that we need to actually look into candidates rather than relying on one piece of paper.

As MS in CS degree holder, I strongly agree with the article re: the idea that coursework-only MS in CS programs are less worthy that programs that include a thesis. I say this as a MS holder who did not complete a thesis and regret it (much more than leaving grad school, which was the right decision for me).

I did (thankfully) participate in research, so I can always talk about the research work even without a publication track record. I have published since in different formats, so that worked out.

But, and this is a crucial issue, a MS in CS of any type helps get you past the HR filter at big companies. When I completed mine, it meant a significant boost in starting salary right out of the gate. If you're looking to join IBM or Microsoft, you probably need to think hard about that signaling credential . . .

Ignoring the disconnect between the academic CS experience and the reality of what most CS programmers work on in the day job, the biggest challenge I see for new practitioners is building a portfolio of software work that a hiring org can understand well enough to make a hiring or interview decision on. How do you communicate that ability and drive to hiring managers outside of your personal network? Do these combined BS/MS programs (e.g., a year of additional coursework to finish off a coursework-only MS) or simply hitting grad school still work better than saying, hey, here is my open-source implementation of path expressions for concurrency control in Common Lisp?

There is a third kind of meaning to an MS-CS degree.

For international students, it becomes significantly easier to get work authorization and permanent residency in the US if they obtain a higher education degree (basically an MS or a PhD) from an American university. This is one of the main reasons that students from India and China in my university who have a BS and already have very employable skills still go for an MS.

The article isn't so much about the perceived value of an MS/CS in the job market as it is about the author's unhappiness with what an MS/CS means these days- less often as researchy prep for a PhD, more often as a BS/CS++. He also complains about the fact that there are students getting their MS/CS that don't have a BS/CS, therefore sometimes have fewer CS fundamentals than their undergraduate counterparts. I have a feeling that most of the comments are going to be about whether or not someone with a BS/CS should go for an MS, but that's not what the article is about.

EDIT: typo

As a BS & MS in CS degree holder, who did not have a thesis option, I agree wholeheartedly. The article mentions that the coursework programs are basically extensions of undergrad, and that certainly applied to my program. I was a bit non-traditional, as I did my MS later in life and as a working professional (8 yrs after finishing my BS), so I wasn't expecting the full-time deep dive experience anyway. I tried to take advantage of the time I did have though and both refresh my knowledge of CS and extend it in a few areas (mainly, learning functional programming and knowledge of language and type theory) Unfortunately, I didn't get to do any research, but I am trying to make up for that by applying to a post-masters advanced study certificate program.

I also had some folks in my program who were, in my opinion, definitely not graduate-level material (a point made in the article as well). I feel like I have a strong CS background and can take things to a higher level, but because of the weaker folks, some of my classes were (to me) watered down and ineffective and thus not challenging to me.

The post, and the one it comments on [1], bothers me for a few of reasons.

One: As an MS in CS holder I certainly worry that my degree is perceived as worth less, if not outright worthless. I worked very hard for it and am proud of my accomplishment. It represents a great deal of sacrifice. It also represents the 30 credits of undergraduate coursework and six credits of graduate coursework I completed before entering a graduate program. Of course, Ms. Lerner would see the MS on my résumé and throw it in the rubbish pile.

Two: According to Prof. Regehr's second point, training in public speaking and writing is an important aspect of a research-oriented master's program. I already had a B.A. in Journalism so I was pretty well covered on both of those points. And, since I was working full-time, with a child on the way, I was plenty mature. A coursework-oriented degree is what I wanted.

Three: I violently object to the denigration of MSCS holders based on the limited experience of a recruiter. Maybe Ms. Lerner is unable to attract more qualified candidates, and so she's extrapolated too liberally. Her arrogance is what I find most offensive.

[1] http://www.forbes.com/sites/quora/2013/05/09/how-different-i...

I did a coursework-only MS in CS, and I think it was a good move for me. In particular, because:

1) I was able to work full-time while completing the degree, and my employer paid for the tuition.

2) I don't have a BS in CS, but I do have a semi-related BS in engineering, so getting a second BS was not a very attractive option in terms of time or money.

3) I came away with a formal CS credential that will allow me to get past HR screens. I've already had a few recruiters from well-known tech companies contact me since adding the CS degree to my LinkedIn profile that I don't think would have noticed me before.

Does having the MS alone put me on the same level in terms of overall knowledge as the average BS in CS holder? Probably not. I definitely noticed that some of the graduate-level courses I was taking seemed to be easier overall than what I was used to from some of the undergrad courses I had taken in CS. But at the end of the day I did learn a bunch of new stuff, and I know that I can fill in any gaps remaining through self-study or resources like Coursera, so I'm not too concerned.

Getting the degree for free made all the difference though. I wouldn't recommend paying full tuition for one of these programs, which the author agrees with.

At my University in the UK (Warwick) we have three Physics courses:

- BSc: A 3 year taught course with some of the final year devoted to research.

- MPhys[1]: A 4 year taught course that is essentially a longer version of the BSc with good chunk of the final year being a research project (some UK universities have an even larger portion of time in the 4th year for research)

- MSc: A 1 year master's course focussing entirely on research (no exams) designed as a follow-on to a BSc.

In the UK, an MPhys is understood to be different to a MSc in the level of research achieved and teaching received.

[1] https://en.wikipedia.org/wiki/Master_of_Physics

I don't see what he's complaining about...

Many people only go for an MS because employers increasingly have a credential fetish.

The quality of the actual education isn't generally that important to students because it typically has little overlap with their future jobs.

MS in CS here. We all know we did it for the diploma: employers value that MS degree, so it gives you a better negotiating position in pretty much every hiring/selling scenario.

As someone with a BS and MS in CS (having written a thesis for the latter), it was amazing to see how weak some of the other grad students were. At my school, many of the foreign grad students didn't have an undergrad CS degree and it showed. This isn't a knock on foreign students, just an observation that they comprised the bulk of the grad students who took the path that omitted the undergraduate work and consequently exhibited shocking failures to understand CS basics.

I'd certainly think very hard about hiring a MS-only candidate, particularly if he/she didn't have publications.

I'm aware that the blog post and the discussion on HN focus on the situation in the USA. Nevertheless, I'd like to add that the situation is quite different in Germany and probably many other European countries. As hinted at by 205guy in this discussion, a BS in CS at a German (or French) university focuses on providing each student with a basic understanding of all important topics in CS. It contains only very few non-CS courses. In order to do an MS in CS, one usually has to have completed a BS in CS. The MS allows students to deepen their knowledge in selected areas - research topics that the professors/assistants work on. Students also work on larger practical and scientific projects and have a lot of freedom to choose their own focus. There is a consensus (at least in Germany) that a BS is not enough for most jobs, therefore a majority of students do an MS. (Generally, universities also require PhD students to have completed a MS in this field.)

For me, getting an MS after working for a couple years has helped open up incredible new opportunities and deepened my knowledge.

I had a much longer comment written, but the only point worth making is this: The research MS is no silver bullet, and I saw a lot of people spend a lot of time working on a rather thin gruel.

Maybe the coursework-only MS is diluting the value of an MS, but there are a lot of research MS holders I wouldn't particularly want to hire, either.

I think these MScs are bullshit and I did one.

However equally bullshit is the idea that you can't write software without a Bachelors in CS. As far as I'm concerned my MSc is a halfway house. It gets you passed HR and let's you talk to people who know these MSc aren't worth anything but only really care if you can make things.

I did an MS/CS (coursework-only) nearly twenty years after completing my BS/CS. After twenty years in industry, I expected to concentrate my coursework on design, software engineering, and "practical" topics like bioinformatics. Instead, I found myself drawn to more theoretical topics, like theory of automata, and especially graph theory, and fell in love with the science of computer science. I don't claim this is causal, but my salary since completing my MS took a 25% bump and has stayed there and above since. YMMV.

I'd argue that for many student from a undergraduate CS background, taking a course-based MSCS program from a top school with a specific direction can be tremendously helpful to the student both professionally and personally. I did not have a specific focus in my undergraduate study in CS and I'm doing HCI+AI for my Master's, which has been min-blowing so far as I'm taking more graduate level courses and exposed to content of much greater depth than my undergraduate study. These courses are usually structured with small research projects/paper readings and I've seen many students come up with research quality work out of the project portion of the classes.

For a student like me, there is also a tradeoff between during research vs doing your own side project vs taking more courses in school. Doing research can be extremely time consuming and potentially affect other course selection density and even performance. I've also seen may CS graduate students who take undergraduate level courses out of MSCS and they max-out the amount of course they can take. These are usually great courses that their previous school does not offer. Doing research in a specific area is probably the most counterintuitive for those who're taking lots of diverse courses. You can argue that these type of students are still figuring out what they want to do and that's what they should have done in their ugrads. But the truth is when you realize you can do much more, the curiosity to discover more is not that easy to stop.Doing research in summer usually is the ideal choice if the students take too many courses in other quarters and they do not want to do an intern.

I agree with the advise to throw away all MS programs that don't waive/pay tuition.

Also I am currently a CS Undergrad in my senior year at University of Illinois Urbana-Champaign and from what it looks like (at least from where I am standing) I would benefit more getting a MS in something such as finance or (what I am doing) chemical or bio molecular engineering. This gives me both the appropriate background to take jobs in CS as well as the ability to work in the field of medicine or both.

Just thought i'd share...

As a non-engineering graduate student at a top-25 CS university, I took 2 classes in the CS dept for fun (my undergraduate degree is CS). The first was a subject on the phD qualification exams and populated by phD students. It was the hardest class I've ever taken, both conceptually and workload-wise. I am prouder of that B+ than any A I've ever received. The second was full of Master's students, there on their employers' dimes. For less than half the effort (including re-writing a teammate's entire final project contribution on the due day because his MBA ass was too busy trying to schmooze the professor into doing it for him), easy A. It was my own personal introduction into the difference between a phD and a Masters - at a fairly decent university.

There are not enough course hours in the undergraduate program for students to become fluent in mathematics, physics, chemistry, biology, economics, the liberal arts (very important), and computer science. A fifth year (that is, a Masters degree) allows for a more complete program. And, somewhere along the line, there needs to be some time for a research project where the student works closely with an expert since research remains an apprenticeship program.

The post suggests that course-only Masters degrees are inadequate and dilutive because, while the survey a body of knowledge, the human contact is missing and that the human contact with a real live researcher is how computer science it best taught. I agree.

I've got an MS from the UW Madison. It has served me well. I got to teach cs240b at Stanford with that MS (cs240b is the papers course in OS). Fun times. I taught it on roller blades because I wasn't good enough to have parking so I had to park at the mall and skate in.

If you are going to a good school where they actually work you (UW worked us well into the wee hours and they were a hacking school when I was there, I watched Joe Moran do a Unix port in 36 hours there - Joe is the guy who implemented the Sun VM system which is something Bill Joy imagined but couldn't do), if you are going to a school like that, go for it.

For me the MS was just a chance to hang out with people who were doing real work. And learn from them.

I was a "coursework" MSCS student at a school which had both. I was earlier on the research track but then decided not to do a thesis after losing interest in research oriented CS (which btw, why all schools should offer both - I'd have had to either drop out or more likely spend a really unhappy year or two doing something I didn't want to do).

Here is where a MSCS ("Coursework") works really great for all parties involved -

1. (And the biggest reason) Provides actual, high skilled nonresidents an easy route to working in USA. MSCS ("Coursework") are doable in 15 months with some schools such as Cornell offering a 9 month program (MEngg not MSCS but same difference). This works great for the student (since CS grads are paid well, they don't have any issues repaying student loans), for the university (they're struggling for cash and this is a welcome inflow of cash) and BigCos such as Microsoft, Amazon, Google etc who recruit most of these and basically pay for these degrees (they're bleeding top talent to the Facebooks, Twitters and startups etc and are always hiring)

2. Students who don't have a BS in CS. Its a great route for them to add some formal education to their resumes and improve the core CS skills. Startups don't really care about your resumes, but it can be hard to get calls from the BigCos without a formal degree in CS or EE. Nothing wrong with that and I can't see why anyone would disparage them.

Finally, I'd disagree with the blog author's two observations. First, his recommendation to the undergrad. If you don't fit into either of the two scenarios mentioned above (nonresident, didn't do a BSCS) there is only one scenario where it makes sense for the undergrad to do a MSCS - if he/she is interested in research and doesn't have a good enough profile to get into a PhD program directly (it'll give you an additional year to try and get some more publications + give you an opportunity to convince a prof at the school you're at to do a PhD).

However, if you're going to reject all the schools which don't give you any financial aid or give only a TA, you'll left with schools which will give you a RA which basically means you are good enough for their PhD program and they accepted you to the MS program only because thats what you applied for.

The second scenario - of Udacity/Gatech's $7000 MSCS eating other schools lunches, unless this $7000 MSCS comes with an CPT, OPT and its own category of H1B visas, I find it really doubtful it'll make much headway with the first group, nonresidents. Although it is definitely quite attractive towards the second group, those who didn't do a BSCS.

If you have a BS in CS already, the only reason I'd recommend taking a coursework-only MS program is if both:

1) Someone else (probably your employer) will pay for it

2) You _know_ it will get you a raise

For example, when I was at Lockheed Martin a long time ago, they would pay for a MS and give you a $N raise when you finished it. 100% worth it.

I know a bunch of people who worked their ass off getting a MS while working a full-time job only to keep working the same job for the same salary after they got their degree, except now they were saddled with $30K in debt.

If you don't have a BS in CS or you're looking to get a masters as a precursor to a PHD, that's probably a different story.

In my experience interviewing candidates, an MS counts for less than a good undergrad. I find that software engineers with foreign undergrad and US MS are often missing some fundamentals.

I wrote a blog awhile back that kinda relates to this. I argued that employers should be focusing on developers with great portfolios of innovative and USEFUL software, rather than focus only on candidates who have earned a degree.

If a hacker decides to get a degree, it shouldnt be because the job he wants requires it. Rather college should be a personal pursuit to accomplish a life goal. Something for them to look in the mirror and say, "Your pretty damn bitchin at this thing called life."

In my experience talking with colleagues (I'm a B.S. CS guy) I've scene both ends of the spectrum. One friend was effectively obligated to earn a MS in CS at Columbia (not sure if that entailed research) but seemed like a stat grab. Then at Bay2Breakers I met two foreign MS CSers also from Columbia who were doing research in semi-supervised machine learning applied to traffic systems. Seemed legit. It was early and I was drunk in the latter experience, but I stand by it!

As an undergrad making the choice of a coursework-based MS and simply graduating with my BS in the next few months, this could not be more timely.

What makes me want to do get the MS is that my undergrad performance was not stellar. I struggled in the middle of my time in college, and while my grades have rebounded and I have been doing well for myself since, my GPA is pretty dismal. I can't see myself getting into any of the good MS programs around the country, and the easy remedy to this is to staying at my university for another year and getting my MS with the goal of having a much better GPA in order to demonstrate my ability.

That said, if the value of such a degree is declining at the moment, maybe it's a lost cause to redeem my academic credentials.

Here's my anecdote about the value of a Masters.

One of my good friends and I started at the same company around the same time. He had just finished his masters and I hadn't. He had 5 or 6 years of real world experience, while I had 3.

I got hired in making considerably more money than him because he was considered a "university" hire.

While this could just be the result of the fact he went through university recruiters and I didn't, it still seems like his Master's didn't help him much.

In my (admittedly short) career, I haven't seen a difference in quality between what people with a BS and an MS produce that would make it worthwhile for me to go back to school. It just wouldn't be worth it for me unless I was really interested in the research.

Google Cached version: http://webcache.googleusercontent.com/search?q=cache:http://...

This hits home as I just finished a very well known part-time course-work only (with a pseudo-thesis option) program.

I actually have a little different opinion - what's killing the value of an MS in CS is the fact that CS curriculum is becoming Software Engineering under a different name. Let me be clear - there is NOTHING wrong with Software Engineering but as someone who is more interested in NP completeness, computability, static analysis, machine learning, etc - much of what passes for CS cirriculum in course-work only programs simply isn't.

This is very timely as I'm currently studying to sit the GRE in August, with the hope of applying MS CS programs at the end of the year (Stanford, CMU, Princeton, etc). Perhaps I should reconsider this?

I'm a non U.S. citizen/resident and I'd like to work in Product Management in the U.S.; I had hoped that an MS CS from a top university would help bridge the gap between my design undergraduate from an unknown international university – helping me land a job at a top tech company as an Associate Product Manager or the like.

Thoughts from the savvy HN crowd?

Has anyone done the UW Professional Master's Program? It's a coursework-only MSCS, two years, designed for working engineers. I've been looking at it for a little while, and would love to get some feedback from someone who's done it. I'd only apply if I could get my work to pay for the majority of the tuition.

My BS is in Physics, not CS, I have five years experience in industry, and my goal would be to cement my foundational knowledge and complement the self-study I've done over the years.

What would you suggest to a non-CS undergrad doing a Masters in one such cash-cow program where there are hundreds of similar students? How does one get the maximum out of it?

I earned a MS in computer science in 2006, so things have obviously changed since then. However, I found it was hugely beneficial for me because my undergrad (also in CS) was very, very weak. As a result, I had the opportunity to realize my deficiencies in an environment where I could learn from them.

I spent four years in the MS program and, looking back, I could have done more with the time I had there. Even so, it was well worth it and opened a lot of doors for me.

Companies should be intelligent enough to hire for what they really need. Hiring an MS just for the sake of the MS is not always the best strategy.

In a research-driven job, someone with a strong research background in CS makes sense.

For a job where you want a top craftsmen and producer of code, it would seem to be much more tangential.

For the latter situation, companies that hire purely based on academic credentials will likely be making suboptimal, if not outright negative EV hiring decisions.

An interesting question is what are the chances of landing programming jobs which nurture a coder's talent more than studying for a year or two more to get an MSc.

We hired so many people with MS degree, we treat them no differently than people with no MS degree.

Actually, they are paid less as they are considered to have less experience...

As a current CS undergrad, is it worth it to pursue an MS after I finish my BS? I understand it that in many ways it is a highly personal choice yet, I don't want to end up wasting time I could be spending doing something more productive/beneficial/interesting. I've really enjoyed most of the courses I have had to take so far (OS,DS&A,Lang. Design,...) but I still wonder how much an MS would bring to the table.

I first realised that an MS in CS was a red flag back in 2000.

I was doing a whole ton of interviews and the candidates who had cross-trained to CS from some unrelated discipline by getting an MS were almost universally unable to cut it in interviews.

So, hardly "declining" value.

Of course, I'd be happy to accept the distinction between the two types of MS that the article discusses.

In Australia, the Masters of Computer Science is generally coursework. The reason is that we have Honours years in Australia, which is the usual gateway to PhD studies. And we don't hand out consolation Masters degrees for people who drop out.

It is definitely true that universities are using MS degrees as a way to fund the programs. It is a shame though because it seems to deteriorate the quality of the program when they load up as many MS students as possible.

The non-research/terminal MS may also be useful for those wishing to teach computer science at the junior/community-college level.

At Google a BS in CS will start out the same as a MS in CS. If you have a PhD you'll start out one level higher.

Indian Bachelors -> US Masters CS is a well worn path.

I have read extensively on this topic over the last few months as I'm about to start an MS CS program. Without fail, everything I've read has a subtle undercurrent of disdain for people who don't have a BS CS who want to get an MS CS.

While I understand, respect and even agree with the reason why - the fact remains that I studied Communication, Beer and Skirts for my undergrad degree, didn't start programming until two years ago when I was 30 and now more than anything in the World want to be a software engineer.

So for somebody like me - the absolute lowest form of life in software engineering - is spending two years doing a coursework only MS CS (first year is a fast paced BS CS) a waste of my time and money?

I was married when I started my computer science degree. My wife let me get through my first year and then told me I had to get a job. So I did all of my studies on Lynda, focussing only on practical subjects like PHP and AJAX. Even though I started in 2010 and dropped out at the beginning of 2011, I just got offered my first $100k job. Tell me, if I was able to accomplish all of that at home, then why would I pay tens of thousands of dollars to attend a University where I am considered a Junior when I leave and join the industry? It just doesn't make sense.