Why (Not) Do a PhD in Computer Science?

Several people have been asking me recently about going to grad school. I thought I might take the time to write down my thoughts. So, here's my rundown of reasons why you would (not) want to do a PhD in computer science.

Whatever reason works
for you... (source: QuickMeme)

So, suppose you are a sophomore in college (or, worst-case junior, otherwise it's too late to think of grad school without taking some time off), you have taken a few CS courses and decided that you want to keep hacking like a madman for the rest of your days. First of all, congrats. The job market for CS graduates could hardly be better than its current state. If you want anecdotal proof, just look at the number of recruiters that contacted you through LinkedIn lately.

But I digress. Suppose you are considering grad school as an option after graduating. Maybe you spent a summer in a research lab and enjoyed the experience. Or you heard a professor describe their research near the end of a class, and they suddenly appeared much more ecstatic than in a typical lecture. Or, god forbid, you saw some of those graduate student people arguing fiercely over a whiteboard and thought what they were doing could be fun.

Before you immediately jump into filling an application to graduate program X, let's look at the reasons you would want to do so. Sitting down to think for a while is certainly a good idea -- grad school in CS will take anywhere between 4 and 10 years of your life, and by the mere nature of research, there are bound to be some setbacks along the way. If you don't have a clear enough motivation to push through those, chances are that you will have a miserable time. So let's go over some of the reasons:

• "I'll do it for the money!" Noooo, you have it all wrong. During grad school, you won't be living in poverty (that's just for the philosophy and English PhD students), but you'll also be making way below your potential earnings. A very common misconception says that you'll be earning more with a PhD degree, and that will compensate for the lost earnings during grad school. That is simply not true in the general case. In fact, I've heard times and times again that the time in grad school costs people the equivalent of a house in missed earnings.

• "I'll do it because I like college and don't want it to end!" Sorry to burst that bubble, but college is going to end. And grad school is very, very different. It doesn't have the nice bubble where you are readily taken care of (with housing, food), and it most likely lacks the social environment that makes college the cool experience that it is (you know, chatting with your roommate about life at 3am, or wandering aimlessly around town after trying to crash a ridiculously overcrowded party). Also, if you go into a PhD hoping to just take a few more classes, you are going to have a bad time. A PhD is predominantly about research and classes are mostly a way to get you up to speed on that. For classes, you might want to consider a masters.

• "I'll do it because I want to be a professor." Alright, this is a perfectly legit reason. Virtually any faculty appointment at a research university would list having a PhD as a requirement. Some of the teaching faculty positions don't, but the trend seems to be going away from that direction.

• "I'll do it because my parents want me to." (no, I'm not making this up) Not a lot to say here. Man (or girl) up!

Now, let me tell you about some of my reasons.

• "I do it because I'm intellectually curious (aka I like new stuff)" Whether you call it doing research, playing with new things, or just poking around hoping for interesting answers, this is the main purpose of a PhD. And it is a process that could be both extremely rewarding and frustrating. The first time you get a prototype up and running, doing something no one has done before, you feel like you've just entered god-mode. Very similar to the feeling from that first time you wrote a "Hello world" and you found out you can make computers do things for you. With the added bonus that you are the first person to do that specific "Hello world". Of course, academia is not the only place you can do new stuff (this article comes to mind as an example). But the freedom inherent in academia often allows trying out rather more wild and crazy ideas than in industry.

• "I do it because it teaches me a great skill set" The biggest skill for me is going in an area and figuring out what the interesting unsolved problems are. This is exactly what a PhD is supposed to teach you. You come in having a vague idea of a target area and start refining that idea until you come up with something that is both novel and interesting enough. Of course, by the end you had better made some progress on those unsolved problems. But my big takeaway is that a PhD teaches you how to ask the right questions -- and that is a great skill. As a prerequisite to that, you get to be able to go up to speed in a given field, identifying the current state-of-the-art very fast. That is also invaluable for any domain that requires problem solving, in industry or academia. Of course, in the meantime, you get the typical soft skills of presenting and explaining your work to others, but that is by no means unique to grad school.

Of course it's going to work, this is
exact science... Source: (MemeGenerator)

• "I do it because it allows me to take a broader view" One of the main properties of research is its scope -- problems that are longer-term and higher-risk than what industry tries to solve. If the rewards for solving a problem are immediately clear, chances are industry will solve it faster and better. But in academia we very often get to see the forest, roughly approximating the trees. Especially in a field as practical as CS, the line is often quite fine. Take this argument too far and you find yourself in theoretical math, living in a space of manifolds or the like. Take it too close and a startup in the Bay Area is already eating your lunch, and getting paid four times as much per employee.

• "I do it because I like the people" Finally, a lot of smart people tend to do research and PhDs. Interacting with them can always teach you a lot. One of the most rewarding experiences in grad school for me is forming ad-hoc teams to tackle certain problems. Teams in which everyone is an expert in their respective area, and brings a unique perspective to the problem. This always reminds me of agile development that has been getting increasingly hip in the last few years.

• "I do it because of vanity" Alright, I couldn't leave that one out without lying. By the end of the PhD, you are supposed to be "the world's best expert in X". Now isn't that cool? Even if X turns out to be "The value of stamp collections"...

So, there, these are my reasons to go through all the fun of grad school. What are yours?

Finally, I am by far not the first or the best person in academia to write about this. Actually, I've been refraining to write this post for a while because it's such a cliche to do so in academia. Some excellent posts on the subject come from +Matt Welsh (here and here), Douglas Comer (here), or Nigel Stewart (here).