Advice / GradSchoolFAQ
Mike Gleicher's Graduate School FAQ
(added summer, 2016):
(added November, 2013):
If you are accepted to our department, then we can talk. I will be back in September 2014.
(added June, 2013):
If you're a student who is considering applying for grad school at Wisconsin and want to work with me, you are probably most interested in the last few questions below. But the summary:
I tend to get a lot of people asking me questions about grad school. I tended to get the same questions over and over, so rather than write short answers each time, I decided to compile my answers so I could say more about each request.
Here is that email, tossed into an HTML page. Its evolved a little bit since its humble beginnings - I have not being trying to keep this that up to date, but a lot of the things are timeless.
Note: Some of this is quite old (circa 2001), but still relevant. While not a lot has been added since, out of date things have been removed.
2009 Update: I am now more actively seeking to grow the group.
2013 Update: We (since the group is more than just me) are actively seekling to continue to grow our group.
Enough people ask me about this that I figure I should write it down... So began this FAQ.
First: All of this is my opinion. The grad school admissions game has no set rules (at least no place I know). My experience includes being on the admissions committee a few times, and doing undergrad advising for a few years.
I used to start at zero, not because it was a CS joke, but because the first question really is more fundamental than all of the later ones
Also, now that I'm using automatic numbering, it's harder to control.
Should I go to graduate school?
Short version: (pick one of these)
This is a tough personal question. My feeling is: you should go to graduate school if, and only if, it is the right thing for you to do. But its such a great thing, that erring on the side of going is probably wise.
This defies the conventional wisdom that says that all smart people should go to graduate school. Many of my colleagues feel this way. Partially from the economic argument (it will pay off), partially from the "what's good for me is good for you" angle.
2001 version: In practice, I don't think grad school is for everyone. Over the past few years, the economic argument hasn't been as strong (although, as the economy slows, ...).
2005 version: The job market today (2005) is quite different, and there are major shifts (such as outsourcing, and the maturing of the field) that make the future hard to predict.
2009 version: The job market is totally wierd nowadays. However, things seem much better for people with CS degrees.
2013 version: The 'CS job market is incredibly hot (note that this is not necessarily true for other fields). If you're good, you can get a good job. Grad school can help make you even more attractive to (some) employers, and there are some kinds of jobs that it's easier to get with a graduate degree.
I've seen lots of smart people for whom grad school was exactly the wrong thing. One in particular was pushed into going to grad school by an advisor (partially for selfish reasons - he wanted the student to work on a project). It was the worst thing that happened to each of them. Fortunately, this particular student dropped out of grad school before they lost their enthusiasm for the field, and has had a successful career in the animation industry.
However: I think grad school is a great thing. It's really the only time where you get to focus your energies on learning about the things that you care about the most, and to be given the time and opportunity to explore things that you might find interesting. It's better than being an undergrad because you can be more focussed, it's better than being a professor because you can devote all of your time to learning and your projects. Maybe being a post-doc is better, but I have no first hand experience.
In fact, the worst thing (IMHO) about grad school is that you don't realize how amazing it is, and you rush through it.
However, if you think you might want to go to grad school...
When should I go?
Immediately after finishing your undergrad degree, unless you have exceptional circumstances.
Basically, "real life" is addictive. You graduate, get a real job, get used to having a real salary, forget how to stay focused and study, ... David DeWitt (who feels much more strongly about this one than I do) told my student some sobering statistics about how students who say they will come back never do.
2013 Update: I keep meeting more people who work for a year or two, realize that the kinds of things you can do with a graduate degree are different from the kinds of things you can do without it, and then come back to grad school.
I have an X GPA, and Y. Where can I realistically think to get into?
First, there are no set rules.
Second, things change. For example, when the economy is good, many smart students want to take jobs, so the applicant pool (especially of domestic students) becomes a little weaker. While I think there are hard thresholds (we won't accept someone we don't think can succeed), I think the total pool quality has an effect: especially at a place like UW where we need to admit enough students to have TAs
Third, there are many factors. Grades, recommendations, experience, research background, ... GRE scores are usually reserved as a "sanity check" (e.g. if they are really bad, we might wonder). I have mixed feelings about the importance of an essay. I think it has a lot more importance at smaller places, or places that admit you because you want to work on something specific. (at Wisconsin, we don't do that - we admit you, and let you decide what you want to work on).
2001 version: As a rough rule of thumb, a 3.5 GPA (from a good place) and a "generally good" package (letters, ...) should get you into most of the second tier places (bottom half of the top ten, like UW). Students with a 3.0 GPA and some amazing experience (research publications, ...) might get in. Doing well in hard CS courses counts more than overall GPA, and positive trends always look good (we all screw up our freshman years, if you screw up your upper level courses, that can be a bad sign).
2003 version: After 2-3 years of a bad CS job market, grad school admissions has gotten crazy. All good grad schools are getting hundreds (if not thousands) of applicants. The number of spaces in the programs is not going up (if anything, its going down, since a tighted economy means less funding). Over the past 2 years, I've seen good students with good grades who have trouble getting in anywhere in the top 20. These are people who 3-4 years ago would have no problem getting in here.
In this kind of admissions market, the problem changes. You need to stand out from the crowd, and you need to make it through mechanical cutting processes.
2013 version: The good CS job market (over the long term) has changed the admissions game, again. We get huge numbers of applications from Asia - so getting in if you're from China, India, or really any Asian country is really really hard. However, if you're from the US, the story is quite different. The admissions committee seems to be much more willing to take chances on people.
I am considering working for a year or two after my BS. Is this a good idea?
From an admissions point of view, I don't think it matters much. The biggest issues will be you references: your professors will forget you, but the people you work with might be able to write letters.
From a personal standpoint, this can either be a great thing, or a bad thing. Depending on your personal situation, you might need some time to experince the real world, and figure out "what you want to do with your life." Students who come back from the real world are always there for a reason: they didn't just come to grad school "because they couldn't think of what else to do."
The downside is that you might get used to the real world - the real salary, the "normal" hours, the paid vacations, ...
What can I do to improve my chances?
Well, I am assuming raising your GPA is out of the question, so...
(revised for 2013)
The biggest thing is to show some initiative. Do something out of the ordinary. A student who takes normal classes and gets good grades is common - a student who has an interesting internship, or has done a research project, or even did an interesting project just for fun, is more likely to stand out.
Make sure that some people know you. If you just "do well in your classes" you'll get things that we call "DWIC" letters (did well in class). Having a professor say "Student was in my class and got an A" tells us little, having a professor or supervisor who worked directly with you and can give specific examples of your abilities is much more compelling. (the reference letters are important!)
For example: get some research experience. Show that you can do the stuff, and build a relationship with a potential letter writer. A research supervisor gets to know you a lot better than someone who teaches a class.
Where should I go?
Where is good for graphics?
First, read the last thing. When I went to grad school I had no idea I wanted to do graphics, and kindof stumbled into it. I know plenty of people who change their mind.
Also, the places that are "good" for graphics, aren't necessarily good places to be a graphics grad student. Sometimes these places attract lots of students, and its easy to get lost in the shuffle.
On the other hand, there is something to be said for being from a place with a strong group. You will learn at least as much from your peers as from anything else.
(2005 Update: I used to have a list of places with graphics groups, but it had gotten woefully out of date.)
Will they pay me?
If you are in a professional masters program, probably not.
Your best bet is to not have them pay you. If you go to grad school apply for all of the "prestige" fellowships (like NSF). These are given to new grad students, are very selective, but are nice since they pay well and are very prestigeous.
At most good places, most/all students are supported. Here are Wisconsin, we used to have enough TA jobs that we could pay all students (that's changed recently). Generally, if you are working on a PhD, you will be funded as an RA (Research Assistant) which means that your advisor will be paying you to do research. Generally, most masters students are supported someway.
Should I get a Masters degree or a PhD.
This is a tough personal question. If you don't know the answer already, you should assume that you just want a masters degree but be open minded to the fact that you will change your mind.
In fact, if you think you know the answer, you should be open minded that something will change your mind.
Generally, this means you should consider whether a program will allow you to do one or the other.
What area of CS should I work in?
Graphics, of course :-)
Seriously, you should work in an area that you enjoy working in. Especially if you are getting a PhD.
Push comes to shove, your thesis topic may not have a huge impact on your life. (there are plenty of people who change areas post-PhD, and some who continue to work on the topic for the rest of their careers).
Often, other factors will influence you. For example, the personalities of the faculty and research groups. For example, if you come to Wisconsin and don't want to work in a lab environment, the graphics group might seem less attractive. Or if you come here and find that you don't get along with Eftychios and I.
Finally, it is not a crime to not know what you are interested in. When I applied to grad school I did know: I was definitely going to have a career in VLSI CAD tools. By the time I showed up to grad school 8 months later, all I was sure of was that I did not want to work on this stuff. I ended up starting out in graphics/physical simulation - fields I pretty much didn't know existed before I got to grad school.
I want to work with you. Can I?
Let me break that into two parts: (1) if you need to be admitted here, and (2) you're already a Wisconsin student.
Aug 2009: I am actively seeking to grow my group, so I am eager to find students that are a good fit.
I want to come to Wisconsin and work with you. Can I?
If you are not a student at Wisconsin, you first need to become one.
The admissions committee is pretty much area blind: a departmental committee evaluates your application. Just because I think you'd be a good graphics student has almost no effect. Even those times that I am on the admissions committee, I am generally given responsibility for only a portion of the applicant pool, and have little influence over most graphics applicants.
Getting admited to our department is difficult. We only take people with really good records, and we don't even take all of the people who are "good enough". So, if you think you might be good enough, apply. If you're good enough, you might get in. But there are so many applications, many good people get rejected.
And if you're good enough (and lucky enought) to get in, and serious enough about graphics, that will lead to the next answer...
I'm already a Wisconsin grad student. Can I work with you?
If you're interested in what I do, and fit in well with the culture of our group, I will probably want to work with you. But there's a question below on how I pick students to work with.
Back in the old days, I used to say "I have a hard time saying no to good students who fit in with our group." Things have changed a bit. Taking on too many students isn't good for anyone (me, my existing students, or the new student).
If you are already a grad student here in our department (CS) and are interested in the kinds of things I do, please come talk to me. I am still on the lookout for people who are good matches for our group. Even if I am unable to take on another student, I may have some other ideas for you. Its important to me that good graphics students in the CS department get to work on the things that they want to - even if we need to find non-traditional ways to do it. (see the last question below)
Update 2010: We have hired another faculty member, Eftychios Sifakis, with interests in graphics. He'll be starting in January 2011.
Update 2013: if you're a really good fit, and/or have background in a domain area we're working in, anything is possible.
If you are already a grad student here in another department (like ECE or ME), we should probably talk. Be warned that it is difficult for me to support a non-CS student, and unlikely that I would support a non-CS student that I haven't already had a working relationship with. You should probably start by taking CS559 as a way to demonstrate your abilities. Also, be warned that switching to CS from some other department is very hard.
How do you pick students to support as an RA? (added 2013)
Note: the question of how I choose RAs is not exactly the same as choosing who I work with.
Generally, I prefer to work (on research) with students that I have already gotten to know well enough that I am fairly confident that we are a good fit. This is as much for them (the students) as for me: as an advisor, I have many students; as a student you have one advisor (so you better have the right one).
Most students I work with take a class with me first. Also, I like to do small projects with students (usually as a directed study).
Can I do Graphics at Wisconsin?
This is a slightly complex question.
First, there is the issue of being accepted in our CS program. At present, admission is "area blind" - that is, we don't care what you say you're interested in, just that you're good. When you get here, you can figure out what you want to work on.
Once you're here, there's the issue that the graphics group is
2013: I am often looking for new students to work with.
However, the situation is not totally bleak:
So, if you're a Wisconsin CS student and interested in graphics, talk to me. Visit the lab and get to know the graphics students. Take the classes.
I am applying, is there anything I can do to help my chances?
By the time you apply, there is little you can do to change your record, and that's what really matters.
Mentioning specifics about your research interests (in your statement) is good. It is good to show that you've done your homework and identified people (like me) and projects in the department that you find interesting. If you're interested in graphics, definitely express that as one of your area choices.
Make sure that you choose good letter writers, people who simply write "this student did well in my class" (we call it a DWIC letter) aren't as useful as people who have really worked with you.
These next two questions were added in 2010, since they are literally what people are asking. You should have figured out the answers already, but just to make it totally clear.
Do you have openings on your research team / in your research group?
In case you haven't figured it out: I don't "have openings." When I find someone I want to work with (and they want to work with me), we find a way to do it - after they have been admitted by the admissions committee.
Are you recruiting students for your group for next year?
Usually, students are asking the same as the "openings" question. But I am always encouraging good students to apply (and to come, if the admissions committee accepts them).
Will you accept Ph. D. students for next year?
No. I do not accept students. The admissions committee does that.