As an undergrad at a "second tier" university, I want to know what I need to do to compete with CS students at the top schools. Specifically, I'm concerned about admission to grad school. Bonus points for criteria other than GPA.
Seriously, there's nothing university departments like to see in a graduate entrance application than published research. It doesn't have to be related to anything anyone at the department in question is working on (although that wouldn't hurt); simply demonstrating that you are capable of working independently, discovering new things, writing a coherent paper about them, and getting it published will put you ahead of the vast majority of applicants.
Edit: Sole authorships, while not necessary, are ideal -- with co-authorships (particularly involving undergraduate students) there's always a question as whether each author made any significant contribution.
Research could also involve independent study courses and/or an honor's thesis. The latter could take two years to complete, but when you go to apply for grad schools it's something you've completed based on your interests and drive. It's the perfect example of completing something that grad school will require. You will have formed a committee, presented your ideas, executed a plan of work, and defended the results. It's practice for grad school because it's a mini-version of it.
Independent study courses are a great way to explore your interests more intensively especially as you get ready to do your own work. If you're working on research with a professor, some schools will allow you to get credit in this form. Or you could simply do more focused reading with a professor and get credit for it. One of my best experiences as an undergrad was an independent study with two profs (philosophy and psychology) to talk weekly about mind and brain issues. I needed to know them to see if it would fly, and I organized the reading list with their input, but afterwards it functioned as a normal course with me as the only student.
Through a thesis committee and independent study courses, you'll also get to know professors for those all-important letters of recommendation (as Paul recommends). For those, it's better to know 6 profs well so you can then pick the ones that know you and can best speak highly of you. In a class, especially one with more than ten students it's very hard for a prof to see how you stand out. Better to start working more with them directly. Then they'll have much more to write about. In a similar vein, take small seminar classes wherever and as early as possible. That's a straight-forward way to get to know profs (to see who you want to work with or have on a committee) and for them to get to know you.
Also, find ways to go to other schools during the summers to conduct research more in line with your interests and to be surrounded by like minded peers. Schools have programs that will pay you a little bit and put you up in a dorm. If you research those programs quickly, you could find something for as soon as this summer.
P.s. More directly in line with your original question: It's very hard for undergrads to get face time with profs at the better name schools. If you get that opportunity, it will shine through in your letters, but more importantly, in your experiences.
There tends to be a tradeoff between novelty and time when it comes to research -- more novelty requires more time. Where is the sweet spot for publishing undergrad research papers?
Precise advice. Nothing -- absolutely nothing -- beats having a published paper at a reasonably respected conference / journal, at least for admissions in CS-related areas.
It's not what you asked for, but if you haven't read the likes of it before, you should. I wish I could have read it when I was your age. I wonder if I would have believed it.
If you still need advice on graduate admissions, I can tell you that pg and cperciva are on the right track. Undergrad research is particularly great, as it gives you lots of contact with an adviser who will write great recommendation letters for you.
I have a science Ph.D. I have reviewed manuscripts for journals. It is by now almost physically impossible for me to claim that any piece of prose is "totally true". :)
Greenspun says a lot of things I don't agree with. I'm not prepared to endorse everything he says about women in science, for example... but, then, this essay isn't really about that, is it? And he exaggerates a bit for effect, but I believe that the effect is justified.
Yeah, it's essentially true. As Greenspun says: some people love science so much that they can ignore these problems, cheerfully, throughout their entire career. I have quite a few friends like that. I tried to be one of them, and sometimes I wish it had worked. But not very often. Your stomach knows when you are lying to yourself.
Here's what I wrote about grad school back in 1998, when I was still there:
Not that everything about grad school is bad. You can work any 70 hours per week that you want. If you just want to waste time and never graduate, and you find the right adviser, you don't even have to work at all. And the people you meet are generally smart, unusual, and fun. But for me grad school is fun just like playing Tetris all night is fun. In the morning you realize that it was sort of enjoyable, but it didn't get you anywhere and it left you very very tired.
I think it's a bit exaggerated. Certainly the _average_ assistant professor isn't denied tenure: that only happens in about 10-20% of cases as I understand it.
I'd agree that it's a little bit exaggerated. And yet... I'd claim that the real reason why "only" 20% of assistant profs are denied tenure is that the system now weeds them out at the postdoc level. It's now considered normal to require candidates to perform multiple postdocs before awarding them a tenure-track position. The only sure way to escape that treadmill is to write your own grant and get it funded... which, by no coincidence at all, is also the secret to being a successful assistant prof.
You can count the number of first-year grad students per year. You can count the number of graduating Ph.D.s per year. And you can count the number of tenure-track openings per year. And then you know the percentage of people who will eventually leave, or be kicked out, of academia. The rest is detail.
Most of us leave. For a motivated student to be forced out is kind of rare. They're quite happy to have you. You're really, really cheap.
Problem is, leaving after a couple of years of a PhD program is the academic equivalent of chewing your own leg off, and most people won't do it -- even if it's the right decision.
You have to be exceptionally mentally strong to leave that culture without grabbing the brass ring, and even then, it can haunt you with feelings of failure and inadequacy for years (people like Jerry, Larry and Sergey are the obvious exceptions.)
The way you get admitted to good grad schools is by impressing specific profs. It's much more personal than undergrad admissions. And profs I know say the thing that impresses them most is good recommendations from other professors they know and trust. So talk to current profs who like you and ask them who they know at grad schools you want to go to who would make good advisors for the kind of work you want to do.
Our CS program here is more like "programming trade school." I doubt recommendations from those profs would be worth much -- what about from the math department?
Get the best recommendation you can, from the best researcher who knows you personally, and who can vouch for your research skills. Classroom performance doesn't count.
If you have to choose, it's better to have a great recommendation from someone unknown, than it is to have a mediocre recommendation from a superstar.
This is a really nice link, I found it really nice to read although I have got into grad school already :-), nice to be reminded why we are here in the first place.
If the gospel according to PG doesn't persuade you, I'm happy to offer some additional anecdotal evidence. Some of the smartest hackers I've ever met are college dropouts or went to a 2nd rate college.
As Twain said, "Never let your schooling interfere with your education."
The order of importance of the many components that would make up your grad school app:
1. Research experience --- The most important factor, though not essential. Not having it is not a handicap. In fact most undergrads admitted for grad school do not have research experience, and their app focuses on no. 2 and 3 below. However if you have done some significant reseach, that fact will dominate the rest of your app, and make 2 and 3 irrelevant to a large degree.
2. Recommendation letters --- there are a couple of factors here. A letter from a better known researcher is more valuable. A letter from a professor who is known to your potential advisor is more valuable. Finally, a letter from a professor who has a history of recommending strong candidates carries more weight (admission committees keep track of such things).
3. Undergrad school and GPA --- A 3.6 from MIT is more valuable than a 4.0 from a second tier school. Your GPA over your junior and senior years carries more weight than the GPA over your first two years. This means you can't afford to let up, but you can make up for the follies of youth (or freshman year).
4. Your statement of purpose (the essay you write explaining why you want to pursue a phd in a particular dept) --- This should be focused
5. GRE scores. These are much less important, and are used mostly as a preliminary filtering mechanism.
Another important issue is that given two applications of equal quality (that is, the two candidates are roughly the same by the criteria above), the more focused one will have an edge. By focus, I mean that between an app saying "I'm interested in theoretical computer science" and one saying "I'm interested in probabilistically checkable proofs and error correcting codes" (these are specific topics of research in theoretical CS), the second one will make the cut every time.
As an applicant from a "second tier" school, you are significantly handicapped. The good news is that this system can be hacked quite well. The best course of action depends on how much time you have left before applying for grad school and the subject you plan to apply in, but I would strongly suggest the following:
a) Increase your gpa as much as possible, between now and when you apply. If your gpa is low right now, showing a substantial improvement just before you apply sends a strong message of maturity.
b) Do research. Find a project in your school to get involved in. Find an informal advisor. If you feel such opportunities don't exist in your school, then find the closest top-tier school in your area. Go visit and find a project there. The goal of this involvement would be to produce a publication and a couple of recommendations. Ideally you should do it in conjunction with point c below.
c) Focus your application. This is the single biggest factor that can boost your chances. Decide the subfield within your major that you want to apply to. Find potential advisors within that area. Read up on their research, contact them about it. Do not spam. If possible go meet these people and get involved. If a professor wants to take you on as a grad student and has funding, then nothing on your application matters and you are in.
Its very possible to get admitted to a top tier program. It takes hard work and knowing how the admission process works. You should also talk to people in your field, as some things vary across disciplines.
You're right about 1 and 2, but you've got the order wrong for 3-5.
Most grad schools view GPA and GRE scores as an initial, binary filter (with the emphasis on GRE scores), and then use the other criteria to rank the remaining applicants in a somewhat flexible way. A 4.0 from MIT is obviously not going to hurt you, but I wouldn't go so far as saying that a 3.6 from MIT is "better" than a 4.0 from a second-tier school. That's just snobbery.
Those numbers were just an example to say that both the gpa and school are factors. That is, an admissions committee would not do a straight comparison between gpas, but would factor in a weight for the school.
This is not snobbery, just the facts of life. I am not saying that a 4.0 student from an unknown school is worse than a 3.6 MIT student. There is no way to objectively know that.
Also gpa matters more than you seem to think. A 4.0 from MIT would do more than not hurt. It would get you admission into a top tier program. The reason graduate programs give importance to gpa is because it takes a sustained effort over 4 years to have a great gpa when you apply. This shows an ability to work hard over a sustained period of time which is a prerequisite for success in grad school and research.
Finally, as I originally said, most undergrads do not have any significant research experience when they apply, and so the recommendations and gpa become more important in those cases. There is a strong correlation between recommendations and gpa, with the difference that a gpa is an objective measure. Most competent students manage to find people who will write positive recommendations for them. Its much harder to finesse your gpa this way. Admission committees are well aware of these issues.
I understand what you believe to be true, but I'm not speaking from a hypothetical perspective. I've seen the 3.6 MIT student get rejected, while a 4.0 student from a "lesser" school gained admission. It happens.
As I said, a high GPA from MIT isn't going to hurt you, but it's not a golden ticket, either. If nothing else, grad schools like the GRE because it's a standardized metric. Still, neither GRE score nor GPA are particularly good predictors of success in graduate school, and the committees know this.
I think we possibly have a different understanding of what is a lesser or second tier school. I based my opinion on the original poster's response to pg's comment, where he states that his cs program is like "trade school". I doubt that a 4.0 from such a program would count for much in comparison to an applicant from a top CS department.
As for your anecdotal example, could it not be that the other factors were the deciding ones in this case (like the student from the lesser school having much stronger recommendations or published a paper)? In any case, all I meant by making up this 3.6/4.0 example was that the school matters too.
If I read you correctly, you also seem to be saying that GRE scores are more important than GPA. This is not true at all. In fact GRE scores count for very little compared to your GPA (which counts less than recommendations and research). Remember that we are discussing admission to top-tier schools. MIT does not even consider the GRE in their admission procedure.
Also, what makes you assume that I'm talking from a hypothetical perspective? At my undergrad school, I have personally known about 50 people who've applied and gone to grad school, and GPA correlates very strongly with the ranking of the department they joined. Since then these views have been confirmed by the faculty (including my advisor) that I have talked to at my grad school.
In any case, we have our grad recruiting weekend coming up next week, and I'll do some field research and report back :)
My post was based on my own knowledge of the recruiting process (and the background of many fellow grad students) a top-tier school:
1) GPA and GRE scores are typically the basis for an include/exclude decision to filter a large pool of applicants down to a smaller pool. They don't typically play a primary role in determining the ranking of the candidates after this step. That's a far more holistic (read: subjective) decision.
2) In terms of #1, committees tend to weight both GPA and GRE equally, with a slight preference given to the GRE, because it is the only direct quantitative comparison available between all applicants.
But this debate is mostly navel-gazing, because admissions committees are famously capricious; the exact method used depends on the members of the committee, and their particular biases. The only point I'm trying to make here is that your chances to get admitted to a top-tier school aren't necessarily higher if you went to MIT as an undergrad. That's a myth that is dearly loved -- by MIT undergrads.
(And by the way? Two seconds of googling tells me that you're wrong about MIT and the GRE:
First, I really don't understand what we are arguing about here, since my initial comment already says that research and recommendations are much more important than GPA and GRE. Lets just agree to disagree on the rest.
Second, I never questioned the basis of your opinions, you questioned mine (indirectly).
Finally, maybe you should have spent more than two seconds googling, since MIT CS does not ask for your GRE scores. See the relevant part in this document (which is available on the page you linked):
http://web.mit.edu/admissions/graduate/pdfs/MIT_department_i...
(Edit: Did you go to grad school in CS? Because otherwise the whole discussion loses relevance.)
It's true that you do generally have to take the GRE to get into MIT grad school, but it's not really what you're being admitted based on (although you might be rejected for sucking).
It's funny to even see an MIT-wide admissions page, since it's department based, and especially so for CS students.
MIT has its set of acceptance mistakes. Many (perhaps even the majority) of those make it through its grind and graduate.
At the end of the day, it is a big, big deal to fail a student. Or to put them on a path to take a core class over. Instead, most professors will give them a barely passing grade simply to be done with them.
By the same token, your "cardinal-direction state school" has a lot of folks works really hard and busting their hump for that 4.0. Sure, you could argue it isn't the same as an MIT 5.0 (yes, MIT's scale goes to 5...), but it's damn sight better than a marginal pass.
There's one exception to this: the apocryphal brilliant and distracted MIT student who can't be bothered to get good grades because they're too busy working on cooler stuff.
I've had the honor of working with a pair of those folks, and it wouldn't have mattered where they went to school. Heck, one of them transferred into MIT from a two year junior college.
Seriously, there's nothing university departments like to see in a graduate entrance application than published research. It doesn't have to be related to anything anyone at the department in question is working on (although that wouldn't hurt); simply demonstrating that you are capable of working independently, discovering new things, writing a coherent paper about them, and getting it published will put you ahead of the vast majority of applicants.
Edit: Sole authorships, while not necessary, are ideal -- with co-authorships (particularly involving undergraduate students) there's always a question as whether each author made any significant contribution.