My undergrad is almost over. These are my thoughts as a graduating senior who has both never been a full time software developer and never been a full time researcher.


December 2015

I need to send an email to the professor I’m doing research with. You would think that after doing it so many times, it wouldn’t be so intimidating, but you would be wrong.

Ph.D. Comics Email

From Ph.D. Comics

Still, there’s a reason this email is harder to write than most. I take a deep breath, and start composing my thoughts.

Subject: Not applying this year

After thinking about things more, I’m not planning to apply to grad school this year. I’m going to try to explain my reasons.


Most friends at Berkeley were surprised I didn’t apply to graduate school. To many of them, it seemed obvious I was going to apply.

So first, I should explain why graduate school looked inevitable. My transcript has more As than Bs. Some of those As are A+s. I’ve taken four technicals every semester, which is more than the average, although I know people who take more. I started taking graduate level courses in junior year, because the remaining upper divisions didn’t interest me enough. Meanwhile, I’ve also done research since junior year, in the Robot Learning Lab at UC Berkeley.

(For the record, I don’t like to brag about my academic accomplishments. It’s both pointless and not actually indicative of your social worth or intelligence.)

Given just that information, it really, really looks like I’m an academic, through and through.

That’s what it looks like.

In reality, I’m not sure I am one.

Getting good grades and taking challenging courses shows you have good work ethic or good talent. They’re signs you have the potential to be a great researcher. They don’t automatically make you one.


September 2014

It’s my first day in the lab, and I already feel like I don’t belong.

I have no idea what I want to research. I only asked for a research position because it’s The Thing Students With High GPA are supposed to do. At least I know why I’m here - I got the top score on the AI final last semester. (When you have no research, no relationships, almost no extracurriculars, a barely functioning social life, and no side projects, you can devote a lot of time to schoolwork. In retrospect, spending 15 hours optimizing a Pacman bot was a waste of time.)

To help give me direction, I meet with a graduate student who will be my main mentor in the upcoming months. We introduce ourselves, and he starts proposing a few potential topics.

At one point, he asks, “Have you heard of neural nets?”

“Nope,” I reply. Internally, I think, Remember to admit what you don’t know. That should be easy, you don’t know anything yet.

“Well, the basic idea is to take your input features, apply a linear transformation represented by a matrix, then apply a nonlinear function to that. It turns out if you repeat this process several times, you can solve a wide range of problems.”

I follow this explanation (hooray!), but I have no idea why this should even work. He admits the theory behind neural nets is a bit shaky, but he points towards an exceptionally cool result: learning to play Atari games purely from visual input.

We talk a bit more, and although I don’t make a decision on what problem to research, I get a better idea of the available options.

I talk with another student to get an account on a lab machine. Yay, I’m on the team! I’m not sure where to start, but I know that when people do research, they write papers, and other people read them. So I start there. For the next few days, Google Scholar is my best friend.

As I read the literature and work through neural net tutorials, I overhear some discussion.

“Hey, do you understand RNNs?”

“Kind of? I implemented one a while back for a NIPS paper.”

“Okay, that’s more than I’ve done. Mind helping me understand LSTMs?”

I don’t understand a word of it, but I try not to let it discourage me. I just started. There’s no way I should know what they’re talking about.

The same song and dance continues for the next few days. Eventually, I decide to research ways to combine Monte Carlo tree search with reinforcement learning. My lectures are separated by 1-2 hour gaps, and I spend them doing research. By which I mean reading the simplest reference I can find, in a desperate bid to find something I can comprehend. I also go to weekly group meetings. Every presentation loses me on literally the first slide, but there’s free lunch, so I can’t complain.

Ph.D. Comics Presentation

I’m basically the guy with the brown afro. Source: Ph.D. Comics

In my first one-on-one meeting, I mention that my problem feels well-solved, and my mentor has to gently remind me that a paper doesn’t make an idea valid. A paper means exactly what it says; the method used works on exactly the experiments mentioned. It may or may not work anywhere else. My faith in scientific studies drops sharply. (It bottoms out a few weeks later, after reading The Control Group Is Out Of Control.)

Once I see papers as promising ideas instead of ground truth, prior work doesn’t intimidate me as much. At the same time, there’s still a vast ocean of prior work, and I can barely keep myself afloat. But hey, I’m doing research! Literally pushing the boundaries of human knowledge. Better to light one candle than to curse the darkness.

It takes me a while to realize reading papers isn’t enough. I need to implement their ideas to understand them better. I write my first line of code two weeks after starting.

It takes me another week to get my code to work.

Then, one more week to understand why it works at all.


No one, not even the most brilliantly arrogant people I know, has ever said research is easy. Professors casually talk about 60 to 80 hour work weeks, and most of that time is spent getting stuck on a problem. Of course, you’re not really stuck, because by failing so many times, you learn when an approach does or does not work. Still, it can be hard to see the upside when you have no positive results.

This makes scientific research a fundamentally different problem from homework. The nicest quote explaining this is from the book “Countdown” by Steve Olson, describing a speech Andrew Wiles gave at the International Math Olympiad.

Solving Olympiad problems is not like doing mathematical research and is not necessarily the best training for research. Working at the mathematical frontiers is more like a marathon than a sprint. Problems can take many years to solve, and you never know for sure whether you’re going to make the finish line. “The transition from a sprint to a marathon requires a new kind of stamina and a profoundly different test of character,” he said. “We admire someone who can win a gold medal in four successive Olympic Games, not so much for the raw talent as for the strength of will and determination to pursue a goal over such a sustained period of time. […] You can forget the idea, if you ever had it, that all you require is a bit of natural genius and that then you can wait for inspiration to strike.”

I read “Countdown” in high school, but the sheer difficulty of research still blindsided me. It led to me not doing as much research as I should have. Given a choice between a two hour homework problem and a two month research problem, it was a lot easier for me to choose the former.


October 2014

An interesting idea pops into my head. I do a literature check, and find my exact idea in a paper from seven years ago.

It’s a bittersweet finding, but I’m still proud of myself. Another researcher had the same idea, and they got a paper out of it! Maybe I’m not a fraud after all. And, I see some interesting extensions that aren’t mentioned in the literature.

For the next week, I make steady progress on implementing my ideas, riding a high of accomplishment. Then, it all falls apart. My classes all ramp up at the same time. I spend my days in lecture and my nights doing homework and projects. Research goes on the back burner. I’m sorry! I don’t have time for you! I need to learn about thread schedulers, about clopen sets and Jacobians, and you need to wait. You have to wait for me. Please.

No amount of pleading can change the facts: by my next one-on-one, I’ve done literally no research. I sheepishly try to cobble something together in time for my meeting, but I can’t do much with no new ideas and no new code. (Besides, researchers are the sharpest bullshit detectors I know. It comes with the territory.)

In my meeting, I hear disappointment, but I can’t tell if he’s actually disappointed or if I’m projecting my own disappointment into his words. I finally have free time that afternoon, but I go home instead. I’m in too bad a mood to do any work.

Instead of meeting up with friends, I lie on my bed and stare at the ceiling, thoughts swirling in my head.

You’re pushing yourself too hard. You need to slow down.

Or maybe, says another voice, maybe you weren’t pushing yourself enough, and now you are. Do you believe in growth through adversity or not?

I do. I really do. That doesn’t mean I can’t push myself too hard.


Graduate school demands you to be excellent.

Although anybody can apply for graduate school, not all applicants will get in. Based on these two sources and conversations with other grad students, here are the implicit requirements for top CS programs.

  • A high GPA. It doesn’t have to be perfect. A 3.5 GPA won’t raise any questions and anything above it is at best a small bonus.
  • Prior research with a professor. If you do not have this, you will 100% not get in. In the best case, your research leads to a publication, but depending on the subfield this is not required. If you’re applying for AI or machine learning, the standards are much higher because so many (almost 50%) of applicants try for these fields. One publication or one paper under review is almost expected.
  • Three letters of recommendation. A common mix is a strong letter from your PI, a decent letter from a course you TAed for, and a weak letter from a course you did well in. Again, stronger applicants will have two research letters, or maybe even three.

This is an incredibly demanding funnel. Only the strongest CS students will even meet these qualifications. They apply, the truly excellent ones (or the lucky ones) (or actually, both) get accepted into top Ph.D. programs, they make a decision, and they start their new life. About 14% of Berkeley CS undergrads go to graduate school (counting master’s programs), and I’d say this is actually the top 14% of the major. Maybe top 20%, if you want to be safe.

A quarter of these students will drop out of the program. This is the CMU drop out rate, circa 2014.

You need incredible talent and dedication to even get into a program as good as CMU’s, to make it through the application filter, and a quarter of them won’t finish.

People call academia an ivory tower, but if anything it’s an ivory spire. The top high schoolers are average undergrads. The top undergrads are average graduate students. The top graduate students are average professors. Some get a tenure-track job and some don’t, and the ones who do need to prove they deserve tenure. Meanwhile, everyone competes for grant money to fund the next wave of grad students. It keeps going. There is no top to the spire, because the people who make it high enough have become so enlightened that they recognize the top doesn’t actually exist. It’s an illusion that distracts you from actually doing the research you were hired for.

It takes a very confident person to believe they can make it all the way up the spire. It takes a very dedicated person to actually do so.

These are both awful realizations if you have imposter syndrome like I do.

Off the top of my head, I can name several people smarter than me. (I met them in high school, and they’ve been better than me since.) I can name undergrads with first author publications. (They work in the same lab I do, and they’ve worked longer and harder than I have.)

I’m not saying that’s a bad thing. It’s fantastic I’ve had the opportunity to meet such brilliant people so early in my life, because they taught me to base my self worth on my own standards, not on where I am relative to others. Not everybody can be Elon Musk or Bill Gates. That’s not saying I do this successfully, but I try.

The point is that when I surround myself with people better than me, I lose the ability to rank myself. Am I doing well in the Berkeley CS program? Yes. Objectively, I’m doing very well. Subjectively, I still feel like an idiot who spends way too much time doing who knows what.


January 2015

New semester, same problems. Four more technical classes to juggle, and more deadlock in my research. Four classes is pushing it if I want to do good research, but I love my classes too much to drop any of them. Does learning Kolmogorov complexity help my research? No. Absolutely not. Is proving the existence of incompressible strings super awesome? Hell yes.

I put in hours where I can, but my schedule is busy, and once again I can’t put in as many as I’d like. It’s disappointing, because I actually have some promising preliminary results. With enough training time, I can get a program to self-learn heuristics for playing Connect Four. It’s not amazing, but my AI beats me over half the time, so it’s definitely something.

Unfortunately, my approach requires lots of computation time, and running experiments is becoming infeasible. I can’t make more progress until I find a way to make my experiments run faster, and I don’t hit on any solutions that are easy to implement.

As progress stalls, I find it hard to keep doing research. I have to remind myself why I’m here in the first place. You’re here because AI is awesome. If it was easy to discover something new, it would have been done already. So keep going. It’ll be worth it when it all comes together. Sometimes this works. Other times, it just sounds hollow.

These are the thoughts in the back of my head, while I wait in line for free food. The demand for Berkeley CS majors is so high that companies give free dinner, T-shirts, and stickers at their info sessions. It’s pocket change for them, and every company has to do it to compete. Hungry CS majors are totally okay with this.

I’m waiting with a few friends from the CS honor societies. Some are 100% doing grad school, some are 100% not doing grad school. A few are on the fence, like me.

One of my friends pipes up. “What are we even doing with our lives? Is waiting in line for 20 minutes really worth it for free food?”

Someone else replies. “Well, I’m here to get used to the grad student life. Why are you here, Mister I Already Accepted A Full Time Offer In Industry?”

“You know, I don’t really know. Force of habit, I guess.”

We laugh, and no one acknowledges the financial commentary. We all already know it’s true, and it doesn’t need repeating.

We get a plate of pad thai and eggplant. It’s good, and getting it for free makes it even sweeter.


Let’s talk about the elephant in the room: the money.

The software industry pays stupidly well. Based on Berkeley Career Center surveys, in 2014 EECS graduates averaged a starting salary of $108,000. Once you spend enough time in Silicon Valley, this makes sense, but if that large a salary for new grads doesn’t seem slightly insane, you should recalibrate your perspective.

That’s not even getting into the perks. Companies like Google, Facebook, and LinkedIn sell themselves by offering insane perks. Free lunch, free dinner, we have a ball pit on campus, we have a masseuse on campus. We built a slide from the 2nd floor of the building to the ground floor, because it helps express our inner child. You’ll never go hungry, you’ll never go thirsty, and by the way we have some interesting problems that need solving, so why don’t you join us?

Alternatively, new grads can join a startup. Come here, and you can change the world! You can exploit a temporary inefficiency of the market. The first exploiter (or more commonly, the best exploiter) can create tons of wealth. We’re changing how people connect with each other. We’re changing transportation. Join now, and if we make it big you can live like a king.

There’s only one way a Ph.D. program can compete with the money in industry. A CS Ph.D. program sells itself with research. You will tackle the hardest problems, the ones where no one knows the answers. You will devote the next few years towards learning as much as you want in your quest to solve one of those problems. It will be hard, difficult, and poorly paid, but in the end you will learn what it feels like to truly know something.

If you’re looking for money, grad school isn’t for you.

If you want to coast through life with a reasonable living salary, grad school isn’t for you.

The only valid reason to go to CS grad school is because you want to do research.

Money isn’t my driving motivation. I care more about self-satisfaction and impact. Still, ignoring the possibility of six-figure salaries straight out of college would be insane. I suspect that’s why grad students worry about selling out on their principles.

Ph.D. Comics Sell Out

Source: Ph.D. Comics

In many respects, the mindset of Ph.D. candidates matches the mindset of startup founders. Both demand unending dedication. Both can take over your life. It takes around 5 years to get a Ph.D., and it can take 5 years for people to even learn about your startup. Both could easily get a lucrative job in industry, but the whole reason they chose this path was to avoid being just another employee. You shoulder a lot of stress, but you’re not supposed to tell anybody about it because you need to look like you know what you’re doing.

If a startup is a bet against the world that you can build something people will pay for, then a Ph.D. is a bet against the world that you can discover something no one has ever seen before.

That’s not to say they’re the same. Paul Graham says everybody should try founding a startup. No one says everybody should try getting a Ph.D.


September 2015

I chose to do an internship instead of research the summer before senior year. I was disillusioned with research after running into so many road blocks, and didn’t feel like I was self-motivated enough to try for a Ph.D. program.

It was a good experience, and overall I’m very glad I decided to take the offer. At the same time, when the semester started I found myself enjoying research a lot more. Maybe it was the light course load in the first few weeks, or maybe it was the realization that I actually knew what I was doing. After one year of research, I could actually hold a conversation with my mentor. At least, for a while.

Riding my newfound wave of motivation, I set up meetings with my PI to find a way to get a reasonable application in time for deadlines. I start a new project, one that could lead to a paper faster than my old one. (As an aside, later this semester I found a paper where the authors independently worked along the same lines I did, this time from 2014 instead of 2007. They ran into the same computational barrier as me, and got over it by throwing more computing power at the problem instead of developing algorithmic improvements. I feel cheated, but at least I was on the right track.)

There’s an undercurrent of satisfaction to my life, but it’s buried beneath the weariness of living a life between problem sets. I get more research done than last semester, but it still doesn’t feel like enough. On especially bad weeks, I get one hour of leisure time a day, tops. Soon, it’s hard to tell if I’m satisfied at all.

One Friday night, I get dinner by myself, then head straight to the lab. My work needs to be done now. In fact, it needed to be done two days ago.

Could have been playing board games with friends. Could have been relaxing at home. Nope, lab. Lab. Lab lab lab. Research research research.

It’s quiet that night, as most people have already left. No one wants to work on Friday. I’m not here by choice, and neither are they.

I start working at 8 PM. Five hours later, it’s done. Not well, not prettily, but good enough, and it’s not worth staying later.

As I walk out, I think, Wow, done by 1 AM. That’s actually not too bad. It’s a bit depressing that I’m serious about this.

Out of curiosity, I look around, and one other person is still doing work.

Godspeed, and good luck.


By my first semester of senior year, I knew what research was like. I had full-time job offers waiting for me, and all my major requirements were done. If I wanted to relax and do nothing, I could have.

So of course, I signed up for three grad classes, TAed for the first time, and did research on top of it. At some point in high school I gained a deep unease with myself that always pushed me to learn more and do more. Appeasing that part of my brain while making sure I don’t overload has been one of the harder challenges of my life.

I’ll be the first to admit I have a huge self-deprecation complex. On one hand, it’s probably a net positive for success, especially success in research. Without natural curiosity, there’s no way I could deal with academia, because the only reason anyone would choose the academic life is because they wanted it. (There are exceptions, like international students who need to keep student status for their visas, but the ones I’ve met like research as well, because otherwise they wouldn’t have made it to Berkeley.)

On the other hand, intense self criticism is the first step on the road to depression. To say there’s angst in academia is an understatement. You don’t need to look very hard to find the black humor and horror stories. I’ve already shared lots of Ph.D. Comics above. All the things you see in those comics? They’re real. A bit exaggerated, and any one student will only run into a fraction of the issues depicted, but it’s all true.

As proof, there’s The Ph.D. Grind by Philip Guo. It’s a detailed account of the author’s experience getting a Ph.D. at Stanford, and I consider it required reading if you’re considering grad school. More than anything else, it conveys how soul-crushing academia can be. (That’s not to say his experience was common. In retrospect, he says he was a 30th percentile in happiness at Stanford, but also said it could have been even harder if he was at a weaker school.)

Stories about the malaise in academia are all over the place. There is sampling bias, where only the more extreme opinions get told, but it’s a bit worrying when doom and gloom stories are easy to find and sunshine and rainbow stories aren’t.

A 2015 study at the University of California Berkeley found that 47% of graduate students suffer from depression, following a previous 2005 study that showed 10% had contemplated suicide. A 2003 Australian study found that that the rate of mental illness in academic staff was three to four times higher than in the general population, according to a New Scientist article.


Look, I am in fact a career academic. I know exactly what’s attractive about it, I’ve made considerable financial and personal sacrifices to get myself to a position where I can work in a university environment and spend my time doing groundbreaking research. And yet. The gateway into this life is a Ph.D., and the Ph.D. system is deeply, deeply fucked up when it isn’t actively abusive. Doing a Ph.D. will break you. It’s pretty much designed to break you. Yes, even you, you who are brilliant (that almost goes without saying; it’s because you’re brilliant that you’re contemplating doing a Ph.D. in the first place). You who are resilient and have survived several kinds of shit that life has thrown at you just to get to the point where you’re about to graduate with a brilliant degree. You who have the unconditional support of your family and friends and partners. If you have every admirable personal quality you can think of, if you have every advantage in life, still, getting through a Ph.D. will grind you down, will come terrifyingly close to killing your soul and might well succeed. It will do horrible things to your mental and physical health and test to breaking point every significant relationship in your life.


“You make time for hobbies,” she told me. “This isn’t undergrad anymore. This is the rest of your life.”

Source (need Quora account)

Ph.D. Comics Cancer


So, if it’s so bad, why is choosing not to go to grad school such a hard decision?

Some people care deeply about research. They care about research enough to accept the hardships that comes with it. I can see where they’re coming from. There’s something profoundly beautiful in the pursuit of knowledge for its own sake, in searching for answers only because we know we don’t know everything. I don’t have the words to express it properly.

There’s an old saying: if you could be happy anywhere else, don’t go into academia. Some people can’t be happy anywhere else. It is what it is.

(Philip Guo calls this a unicorn job. It’s a job where you put up with lots of stress for that 10% or 20% of the job you actually want to do, because you enjoy it that much.)

I got it, but I couldn’t live up to the ideal. My motivation for research fell right on the boundary. Interested enough to keep doing it semester after semester, but not enough to put in the hours I needed to create work of my own. A lot of sections from The Ph.D. Grind resonated with me, but my favorite was this one.

I discovered over the past 5 years that I love being a spectator of research, but the burden of being a continual producer of new research is just too great for me.

When I don’t do research for a while, I miss the feeling that I’m in touch with the cutting-edge and playing a hand in shaping the future. In AI too! Holy shit, I get to experiment with algorithms for computer learning at a top-tier university! How awesome is that opportunity?

When I do research for longer, I realize how much of a novice I am and how unlikely it is any of my contributions will be important. There are so many subfields, so many subsubfields, papers and papers and papers and it feels naive to believe one more paper can stand out from the rest.

I know it’s unrealistic to reach that far, that reality is complicated and ideals are dreams for a reason, but it always makes me feel a bit guilty to see others make it closer to scientific purity.

One day, Andrew Ng, one of the biggest names in machine learning, came to give a talk on the deep learning work he did at Baidu. Every talk on deep learning gets a packed auditorium. A talk on deep learning by Andrew Ng? It was packed 15 minutes before it started.

He gave a broad overview of the current research trend towards bigger models and bigger datasets, then talked about his work developing smart assistants like the Baidu Eye. The way he talked about it, you could tell this wasn’t just something he worked on, it was something he believed in. This was his vision of the future, and he was there to play a part in making it real.

Another day, a visiting postdoc came to our lab for group meeting. After the meeting, I overheard a conversation between that postdoc and a graduate student. They started talking about Ph.D. Comics, and the postdoc said she could never relate to it. She was always a happy grad student.

A third day, a professor said, “Looks like I’ll be done by 11 PM. Get to leave early for once.” The way he said it left no doubt. He was tired, but he loved computer vision and he loved his work. He was here by choice.

Passion and faith in your work are what carries you through a Ph.D. program. Nothing else will.

It was a high bar, and I wasn’t sure I crossed it. As the months passed by, I realized I had made my decision a long time ago. I was just too afraid to say it out loud.


December 2015

I need to send an email to the professor I’m doing research with. You would think that after doing it so many times, it wouldn’t be so intimidating, but you would be wrong.

I know what I want to say. I just need to say it.

I take a deep breath, and start composing my thoughts.

After thinking about things more, I’m not planning to apply to grad school this year. I’m going to try to explain my reasons.

  • I haven’t been researching faculty at any schools I might be applying to. Given that some of the deadlines are Dec 8th, that means I won’t have a very polished application. But more importantly, I haven’t had the motivation to work on applications. Of course, very few people look forward to writing applications, but people interested in grad school will usually have the motivation to trudge through it because they see it as necessary to get to where they want to go.

It’s true, I haven’t. It has to mean something.

  • I haven’t notified people writing my letters of recommendation. Again, this isn’t a deal breaker by itself, but it’s certainly somewhat bad etiquette, and it goes back towards the idea of motivation. If I believed I wanted to apply this year I certainly would have notified faculty sooner.

  • I talked with some other seniors/grad students, and the general consensus was that you should only go to grad school if you are very sure you want to go. If you’re only somewhat interested, you should internally downgrade interest by a lot.

Maybe grad school isn’t for me.

  • Assuming I magically finished my application and got into a top graduate program, I’d still lean towards not going. After a Ph.D., I don’t think I’d want to stay in academia, and would likely try to get into an industry lab. At that point, the Ph.D. essentially becomes a 5-6 year long process of improving research skills enough to apply for those positions, which doesn’t sound worth it to me.

It hurts a bit to lay it all out. I’m not sure why, but I know that on some level it doesn’t matter if I get why or not. I don’t have to go for a Ph.D. if I don’t want to, and right now I don’t want to. That’s fine.

To me, there are many reasons to go for a Ph.D., but I’m not completely sold on any of them:

  • To prove to yourself / the world that you can get through a Ph.D. program, because only a small percentage of people can do so. I don’t [have] a strong need to prove myself to be better than other people, and care more about my personal standards for how much I can push myself. (I suppose that could also be seen as an argument for why I should do a Ph.D., which is all about pushing boundaries.)

  • A very strong belief that what you’re doing your research in is important. Strong enough to deal with the stresses of academia, and solid enough to keep going even after realizing just how big your field is. I believe that AI research is very important and impactful, and enjoy reading about new advances, but the costs of creating new research are too high for me.

Thanks Philip, for putting it more succinctly than I can.

  • A personal fulfillment achieved by understanding something very deeply. For theory-leaning people (which I’d classify myself as), this may also be the joy from finding an especially elegant proof or neat connection between different topics. This is definitely hard to find outside academia, but aspects of this still show up in industry, and being an expert on X is common advice for standing out among your peers.

  • A love for teaching with the plan to go for a lecturer position after a Ph.D. I’ve liked TAing this semester and approve of education in general, but again this isn’t a strong enough interest to make me want to teach full time.

I’m not a grad student. And if I have doubts down the line,

Maybe this is already clear, but I’m not saying any of these beliefs are set in stone. It is entirely possible that I’ll change my mind about many of these things, However, my expectation is that my beliefs aren’t going to change [enough] by next year to make applying worth it. I still plan to do more research this week + finals week to set up opportunities for myself to gain motivation for research, but if they change enough to make me want to go to grad school I’ll just have to wait a year on applications.

I’ll deal with them then, because I don’t need them now.


I read and reread, wanting to get it just right. Then, I click send.


I get a reply two days later. My professor was very understanding.

A while after making my choice, I found myself idly considering whether I could put together an application, and had to mentally beat myself with a newspaper. You can’t change your mind! Application deadlines were 3 days ago! It’s done! Live your life with the choice you made.

(Okay, I didn’t use that much melodrama, but I did actually think these things.)

Even now, I’m worried I chose wrong.

I’m worried I didn’t actually make a choice. I only sent out that email when it was implausible I could have even cobbled together an application.

I’m worried that I’m doing a disservice to the world by not going for a Ph.D., because I think I could actually handle it if I found the right idea to keep me going. In a group meeting near the end of this semester, I realized I followed 90% of it, and knew exactly where to look to learn the remaining 10%.

I shouldn’t be worried. It’s not like I’ve completely shut the door on getting a Ph.D. Those people I consider smarter than myself? I know some who also didn’t go for grad school, and they’re doing fine. It’s not like my choice is unique or special. It’s not like every smart student has to go to grad school.

I know this rationally, but the rest of my brain disagrees.

Still, for now I’m okay with this. I’ve thought about this topic for a while, and I could do with worrying about other things. My social life’s a mess. My romantic life’s a mess. Don’t even get me started on whether I think my mental state is okay or not. There are more important things, and I’ll get there one step at a time. Or maybe, I’ll just relax and play a video game or two. (Or three.)

After all, what’s the hurry? It’s only graduate school.


Thanks to Allan Peng, Margaret Sy, Ronald Kwan, and Ziv Scully for reviewing drafts of this post, as well as others who preferred to stay anonymous.

If you’re interested in more about this topic, I recommend these notes on applying to Computer Science Ph.D. programs, the Ph.D. Comics archive, and the articles on Philip Guo’s site.. Odds are that if you’re considering grad school, you’ve visited many of these links.