Hacking Grad. School Admissions for Comp. Sci.? Go Ph.D.
I recently posted a question to Hacker News, asking, “How do I make myself competitive for a master’s in Computer Science after college?” The colleges in question are particularly the super-competitive and elite that produce legendary talent, such as MIT and Stanford.
The responses are surprising. Most people suggest to shoot for a Ph.D. program, for a variety of reasons:
- Universities prefer to admit Ph.D. applicants over master’s; they’re an investment. Ph.D. candidates publish research that gives the school a return on their investment. It improves their reputation to have more publishing (Ph.D.) students than non-publishing (master’s) students. Hence…
- You are more likely to get funded as a Ph.D. student. Being a master’s student eliminates fellowship opportunities and leaves only the possibility of teaching assistantship. Because Ph.D. students are an investment for the university (see #1), they tend to assign the limited number of TA positions to those students which gives them the highest return.
- The possibility of continuing on to a Ph.D. typically only exists if you initially apply. There are some schools that are exceptions and allow you to opt for the Ph.D. program when your master’s is completed, but most are structured in such a way that they foster Ph.D. applicants from the beginning.
- (For Americans Only) It is cheaper to accept American applicants than international applicants. Lacker, the most up-voted commenter, states that, “a professor at a top-25 CS grad school once told me they accept almost every American citizen who applies to the PhD program because it’s so much cheaper - so worst case, you should be able to go to a well-known school that has merely a good CS program.”
In summary, make the grad school investment as though you are applying to a Ph.D. program. This means that the importance of research and publishing is paramount. Have solid grades and fantastic recommendations from well-published people.
Last, I have heard that graduate admissions councils rank students in this order:
- Undergraduate degrees with publications.
- Graduate degrees with publications.
- Undergraduate degrees without publications.
- Graduate degrees without publications.
Why is it detrimental to have a graduate degree (a master’s) without a publication? Because it shows that you have attended school for a long time without showing a lot of interest in research. If you have done research, then you must not have been very successful.
EDIT: Rearranged the title to save space.
I have no idea how these myths get around. You know who most of the Masters students in CS at MIT and Stanford are? People who got rejected from the PhD. Look at any PhD application form… they let you check a box saying you want to be considered for the Masters if you don’t make it. And even if you do make it but you don’t pass your quals or maintain a minimum GPA, you are forced to leave with just an MS — so clearly, the latter degree is not an aspiration for most PhD students.
And how is it cheaper to admit PhD students when the Masters people are the ones bringing in the money? Yes, PhD students are more valuable because of research output, but every student is a risky investment. With the MS students, at least you are guaranteed a return. Keep in mind many of these MS students are also trying to get research experience for later PhD admissions, meaning they will work their ass off for minimum wage.
Another point: most “masters within PhD” degrees are of little to no use if you want to do anything other than research. The classes involved are tailored to help the students expand their knowledge and skills for research, and are generally very different from those of professional masters programs.
Also, the point about American students being cheaper than internationals makes no sense, especially for private universities where tuition and stipends are the same across the board. (Even for public universities, I think the distinction is more between in-state and out-of-state students.) The ranking of students does sound right, though I think a masters student with good grades is favored over an undergrad with poor grades, assuming both have no publications. This is why many PhD rejects are told to do a masters’ and boost their GPA.
Bottom-line, apply for a PhD if and only if you are interested in research. If you want to get a masters’, that’s what you should apply for. Many programs do have assistantships if you look around. Another good option is to get funded by your company to do a work-study thing.
(Sorry, stumbled across this post and couldn’t help commenting.)
Comment by S on February 2, 2009 at 9:43 am
Thanks for the great reply!
> And how is it cheaper to admit PhD students when the Masters people are the ones bringing in the money?
I never said it was cheaper. Look closely at the original post. Perhaps it could have been worded better: schools like to invest in students that will give them a return on investment. My original question in the YC thread was whether I should apply to a Master’s program to circumvent the competition from PhD admissions. Most folks suggested against this because of the preferential treatment that PhD students are given because of this reason.
> Another point: most “masters within PhD” degrees are of little to no use if you want to do anything other than research.
Isn’t expertise in a subject already being built because of this? And doesn’t that expertise carry more weight in the job market than a broad computer science education?
I could be wrong, but there is a certain depth that a graduate level class can achieve for a specific subject. By pursuing research–either by doing a Master’s thesis or being on the path to a PhD–students can surpass this limitation, and therefore become more competitive. This is assuming that said student has a reasonable idea of what he/she wants to specialize in.
> Also, the point about American students being cheaper than internationals makes no sense, especially for private universities where tuition and stipends are the same across the board.
Isn’t there a large cost associated with H-1B visas that international students require? I believe that was the premise of this claim. Looking forward to your reply.
Adam
Comment by adambossy on February 9, 2009 at 6:32 pm
Schools like to invest in students that will give them a return on investment.
The point is that the returns are a lot less predictable, riskier, and more long-term (which is also why schools are so selective about PhD admissions). Yes, schools do like to invest in research students, but they need ways to fund that investment — NSF grants and the like are only part of it. That’s exactly where (paying) masters’ students come in.
By pursuing research–either by doing a Master’s thesis or being on the path to a PhD–students can surpass this limitation, and therefore become more competitive.
What I meant was that research-based courses don’t train you for most industry jobs. Now whether or not you value the intrinsic benefits of such courses (and can apply what you learn outside research — it certainly can be done) depends on you. But the truth is that the majority of software engineers will find such classes a waste of time. Most aspiring programmers love to rant against how impractical bachelors’ level CS courses are… I can only imagine what they’d say about the PhD ones.
Isn’t there a large cost associated with H-1B visas that international students require? I believe that was the premise of this claim.
But H-1B visas are sponsored by the employer… I don’t think the university has anything to do with it. (If you mean the student visas themselves, they’re F-1, and AFAIK, don’t cost the university anything either.)
Comment by S on February 11, 2009 at 10:13 pm