In this document, I’ve written out my thoughts on graduate school. It is a mix of reflections on my own experience and advice for new graduate students. It was hard to separate, so I just mashed it all together. It’s like a delicious reflection soup. Yum! It was too long to post in a TikTok video, so if you want the whole story you’ll have to read it. Sue me. Actually, please don’t.
I am writing this reflection in the weeks after submitting my dissertation, successfully completing my Ph.D. studies in the Industrial Engineering and Operations Research (IEOR) department at UC Berkeley. Overall, I had a mostly positive experience in graduate school. I started in the program in the Fall of 2015 and received my Master’s degree in Spring 2016. I graduated with my Ph.D. in Spring 2020. Over the course of my Ph.D., I wrote nine research papers.
My hope in writing this reflection is that it may prove useful to someone who is either considering graduate school or just getting started. I will try to shed light on some of the unspoken rules, expectations, and traditions of academia that I learned along the way. I’ll also reflect on what worked for me. Of course, I don’t expect my advice to be universal. For one thing, the expectations of a graduate program will be quite different if you are working in a wet lab. Even if my advice is not useful, hopefully, you will at least find it entertaining!
I’ll be completely honest: when I started graduate school, I was not completely sure what I was getting myself into. When you choose a school for undergraduate education, the requirements are usually set out very clearly: required classes, electives, credits, etc. You can talk to pretty much anyone who attended the school and get their perspective on what it will be like. The expectations of graduate school are more amorphous. On paper, there are things like the preliminary exam and the qualifying exam and then the dissertation, but departments vary widely in how seriously they treat each one. Moreover, different professors in the same department might take certain requirements more seriously than others.
At some point in my second year, one of my colleagues described the expectations of graduate school in a way that made a lot of sense to me and gave me a better understanding of what was expected of me. He broke it down in terms of what was expected in each year. Inspired by that, I’ll give my summary of the expectations and my overall impressions by year.
Year 1 is mostly devoted to classes. Or, more broadly, in getting everyone to the point where they have the background in the subject they need to conduct research. Our department was particularly heavy in terms of required classes. Some departments have fewer requirements. In our department, the first year culminated in the preliminary exam. Unlike the qualifying exam, which can be more of a formality, in our department, the preliminary exam is a serious affair. I think it was perhaps over-emphasized, but I’ll save my complaining for another medium.
Coming out of an undergraduate program, the coursework in graduate school feels very different. At MIT, I would take at most three technical subjects per semester, sometimes only two. In my first year at Berkeley, I took four or five per semester. This technical immersion, without breadth requirements, means I covered a lot of material in a short time. My first year at Berkeley was very useful. I filled in several gaps in my knowledge. Most notably, I saw a far more nuanced treatment of optimization than I had ever seen before.
While some students do manage to get research done in their first year, it was certainly not the expectation in my department. Again, this might be very different in a different university/department.
Year 2 is devoted to making the first forays into serious research. By the beginning of the second year, you should hopefully have some idea who you want to work with. If they are good advisors, they will probably recommend research topics to you to get started (eventually, you will be expected to find your own). I don’t have much to say about this year. My experiences were a bit unusual, but more on that later.
The qualifying exam is the culmination of the first research project. It doesn’t necessarily happen at the end of the second year, but maybe the third or fourth. Ideally, it will form one of three chapters in your dissertation. More on that later.
In my department and in computer science, “three papers” was a good rule of thumb for graduating. That is, a student who has enough content for three research papers should have no problem putting together a respectable dissertation. In years 3-4, a student should aim to produce about one research paper per year. Combined with the work done in year 2, this brings them to the magic number 3.
Of course, not everything will go perfectly. The research may get stuck and getting the third project done might bleed into year 5. Not all projects are equally large, either. Sometimes you may add up two “smaller” papers into one in the dissertation, especially if they are related. But the pace of one big project per year is a respectable one.
If you are anything like me when I started graduate school, then you have probably heard of the big, scary dissertation and are completely mystified by it. How does someone produce enough research to warrant a book-length report and then write everything down? I found the prospect daunting when I started.
In operations research and neighboring fields, the dissertation is not as scary as it sounds. In fact, the general advice is to just focus on doing good research in your first few years and then figure out how to put it together in your last year. This is because the research papers you’ve already written form the bulk of your dissertation. This sort of dissertation is sometimes called a “staple job” (because you are stapling your papers together) or a “sandwich” (because you slap your papers between an introduction and conclusion). Again, this does not apply to all schools and all departments, but I have seen it hold true in many of the schools/departments that I’m aware of, especially in STEM programs.
While there are occasionally schools that articulate their dissertation requirements clearly, more often then not this is a place where academia depends on tradition. Indeed, there can be considerable variation between advisors in terms of what is expected in a “sandwich” dissertation. For example, some professors may ask their students to intentionally find related research topics as early as their second or third year. This will make it so that there is a clear theme linking the papers in the dissertation. Some advisors will take a more relaxed approach, telling their students to research anything and then figure out how to link it together later. Similarly, some professors tell their students to just copy/paste the papers, while others ask them to make rewrites so that the papers flow together. It is not uncommon for the dissertation to include a few new results (especially if it revisits an old project), longer proofs, more figures, or other details that were not in the original papers. Whereas conference papers may have page limits, dissertations are not so limited.
In my case, my dissertation ultimately combined four of my papers into three chapters. A lot of work went into combining two of them into one. We also presented several new results. In fact, in one of the chapters, we ended up developing so many new results that we published a whole new follow-up paper! My situation was a bit unusual, but ultimately the whole process took six months. It was certainly not a trivial amount of work, but it was much easier to start from papers I had already written than it would have been to create a 100-page document from scratch. And that, in a nutshell, is the secret!
When my colleague shared this outline of graduate school with me, I found it illuminating because it gave some justification for the magic number 3. It also makes clear why graduate school can feel like a series of different endeavors: year 1 is not at all like year 5.
In my case, it felt a bit like the latter years of graduate school had diminishing marginal returns. In year 1, I learned an exceptional amount. In year 2, I stumbled around a bit of my research. In years 3 and 4, I got a lot better at finding research topics, making a plan, executing on that plan, and writing a paper about it. By year 5, I felt like I was mostly just rinsing and repeating. The tail end of year 4 and most of year 5 were productive, but also a bit frustrating since I felt like I was not learning anything new. Your mileage may vary. Or… your lack of milage? Graduate school is not a very mobile endeavor.
There is no easy way to say it: graduate school is difficult and can take a toll on mental health. The hard work, itself, is not the crux of the problem. The challenge is dealing with extreme ambiguity. You eventually need to find your own research topics. You can be stuck for months. Your research might fail. Choosing your own work hours is a blessing and a curse. You have flexibility, but it can feel like you are always supposed to be working.
One factor that really helped me deal with graduate school mentally and emotionally was my friends outside the department and university. A second factor was being married. I’ll reflect a bit on both. Naturally, a lot of this is just what worked for me and is not necessarily going to be widely applicable.
First off… friends! Friends are a good thing to have. Duh. It is good to have friends in your department. My colleagues made our office an awesome place to work and gave me great advice and support, but friends outside the department are just as important. I was lucky to have friends outside the department and friends outside the university.
First, there was a group of fellow graduate students who came together twice a week at Hillel. I did not come to Berkeley with the intent to get involved in Jewish life since it was not a big part of my undergraduate student life, but, when I went to one of their free Shabbat meals in my first year, I instantly clicked with the crowd there. There was a vibrant graduate student group with graduate students from across the Berkeley departments… Computer Science, Economics, History, Public Health, etc. On Wednesdays and Fridays, we would get together at the Hillel building and chat into the night about current events, cool research, graduate life, or whatever was on our mind. I ended up being the coordinator of the group for two years, planning bagel breakfasts, retreats, holiday celebrations, and more. It was a very special group.
Second, I was lucky to have a circle of friends who were not attending graduate school at all. To start, there were many fellow MIT undergraduates who moved to the Bay Area to work at technology companies. From there, the circle grew, as we met friends of friends and beyond. Every few months, my wife and I would host giant “board game nights” at our house, which would sometimes involve more than 20 people and go for 11 hours. We also went on all sorts of adventures, big and small: ski trips, camping, and so much more. Having these scheduled around normal work hours meant I mostly resisted the urge to work through nights and weekends.
This brings me to one of the things that I think goes unmentioned too often when talking to prospective graduate students. Yes, graduate school is a professional endeavor, but when you are choosing a graduate school you are also choosing the geography where you will spend your mid-to-late 20s (typically). Personally, I am very glad that I chose to attend graduate school in the Bay Area. When I was an undergraduate at MIT, most of my activities were centered around campus. I rarely went into Boston. In graduate school, that is not the case. I frequently visited San Francisco and the South Bay. My point is that you should take the locality into account when deciding on a graduate school!
Okay, on to my next point. Marriage! If you want to go to graduate school, you should get married. Immediately! What are you waiting for? Just propose to the next person you see. Okay, I’m kidding.
Seriously, though, being married to someone not attending graduate school kept me sane. At the risk of sounding clinical, there were several simple pragmatic reasons for this. First, there was the fact that my wife had normal working hours and they were enforced on me by proxy. Second, there was the fact that my wife’s income was plenty to support us so that we never needed to worry about basic needs. Finally, there was someone to complain to. As a graduate student, you will always have plenty to complain about, so having someone on the outside to whom you can vent and who can give you a sense of perspective is very important. You don’t necessarily need to be married to them, but it doesn’t hurt.
Despite everything, I’ll admit that my habit of working normal hours did deteriorate over time. While I was able to be nearly perfect in my first year, only spending two weekends to study for the preliminary exam, the habit gradually broke down. In my second through fourth year, I would occasionally work from home, but I preferred to go to the office to work with my collaborators. In my fifth year, I moved further away from the office and my most frequent collaborator graduated, which meant I felt less motivated to go in. The line between work and home deteriorated. When COVID-19 hit and I started working from home full time, things got even more blurred. A lot of my dissertation was written at odd hours.
Okay, now I’m going to seemingly contradict myself. You might see the advice elsewhere that you cannot simply view graduate school as a “9 to 5” job. You need to find a research topic that you can become obsessed with. I agree! While this seems to contradict my insistence on normal work hours, it doesn’t. You should be passionate enough about your work that you are able to work on it non-stop… it doesn’t mean that you always have to. Ideally, you should be happy to work long hours when things are going well, but able to detach when you’re in a rut and need a break.
On a number of occasions, I had research epiphanies in the middle of the night or while on an airplane or while at the beach. I genuinely enjoyed the problems I was working on and I was happy to have them in the back of my mind nearly all the time. They were fun puzzles for me! There were days where things were going great I worked on my research for 12 hours from waking until sleeping. This felt right. I had enough intrinsic drive that doing the work came naturally to me.
Setbacks are inevitable. At some point, your research will get stuck in a rut or you’ll realize you made a mistake that will take a lot of effort to fix. In these situations, having a routine is important for two reasons. First, you need to be able to detach and, as much as possible, not let the negative energy bleed over and prevent you from enjoying other fun things outside of work. I was not perfect about this, but I think I did reasonably well. Second, having structure will help you eventually get out of the rut. Even when my research was not going well, I came into the office and did… something. Maybe that “something” was just running random experiments to try and inspire a theoretical idea or just having a long chat with a colleague, but I tried to never let my research come to a full stop.
So, to recap… it is great if you find projects that you feel like you could work on for every hour of every day. Intrinsic motivation is the best kind. Sometimes there will be deadlines looming or your research will be going especially well… around these times, you might need to work like crazy, but that shouldn’t be the norm. Just because you could work non-stop, doesn’t mean you always should. Your routine will be there to rescue you when things don’t go as planned.
Knowing what your funding situation will be like in graduate school is crucial. In my opinion, schools way underemphasize this point when talking to prospective graduate students.
Let me step back and first give a little context. There are a few ways that graduate students get funding: either by teaching, by getting funding from their advisor, by getting funding from their department/school, or by getting an outside fellowship. There are probably some other ways, too, but I won’t cover them here.
Getting funding through teaching is almost always a possibility. Teaching is fun, but it’s also time-consuming and doesn’t really contribute to getting research done. Getting funding from your advisor is ideal, but it depends on whether your advisor has any grant money available to fund you. Most advisors in our department did not have enough grant money to support all their graduate students, so they prioritized students that were later in their studies. Then there is department funding. Different departments have different ways of getting department funding to their graduate students. In our department, all the first-year students are supported by a “department fellowship” so that they do not need to teach. In later years, students could submit their recent papers and win prizes which gave them semester-long fellowships.
The best deal out there is outside fellowships. Perhaps the biggest contributing factor to my success in graduate school was that I had an outside fellowship. In my case, it was NPSC. But there are others. The most famous is NSF. If you are considering graduate school… do your research and apply to fellowships! The difference between students with and without fellowships is like night and day. With a fellowship, you have to teach for only the minimum number of semesters required by your department. Without a fellowship, you might spend half or more of your semesters teaching. I taught for one semester (Machine Learning) and it was a great experience, but, sadly, it does not really count for much towards graduation requirements.
Finally, it is no secret that graduate students are underpaid. While my fellowship covered tuition and gave me a small stipend, it was not a lot, especially for the Bay Area. While plenty of people get by fine on this small amount, it can be tight. That is another reason I was very fortunate to have a wife not attending graduate school. Thanks to her income, we were never concerned about our basic needs. Other students I know had savings built up from a year or two of working before graduate school, something that I recommend doing for the experience anyway. Internships help a lot, too. You don’t need to be rich, but it doesn’t hurt to have an emergency fund.
It would probably be a bit presumptuous of me to give general advice about how to do good research. Many researchers have written about their own processes, so I’ll just highlight a few things that worked for me.
One challenging aspect of graduate school is finding the right problems to work on. In fact, one of my colleagues described graduate school in a very clever way: “As an undergraduate student, you learn how to answer questions. As a graduate student, you learn how to ask them.”
Here is a breakdown of how a few of my research papers evolved:
In a lot of ways, I got very lucky in finding research topics. While some research topics I started did ultimately lead to dead ends, more than my fair share ended up leading to success. Speaking in very general terms, I rarely felt like I was actively looking for a research topic. Rather, I just let my curiosity drive me to learn and explore new things. I was always passively receptive to a new project if I heard or read about something and felt like I could contribute.
Let me try to articulate this with a more concrete example. When the AlphaGo paper came out in 2017, I was very excited. I had not been doing research in reinforcement learning or deep learning, but I wanted to know more about the breakthrough. So… I read their paper. I was not intending to start a research project. In their paper, they mentioned Monte-Carlo Tree Search. I had not heard about it, so I read a survey paper about Monte-Carlo Tree Search. I wondered if I could apply the ideas used in the AlphaGo paper to a cooperative game: Hanabi. It felt like something was missing to make this generalization. I wrote some code. I read more research papers about information sets in imperfect-information games. While there was some semblance of a research project emerging, I was still just following my curiosity. As it turns out, I ended up abandoning the project, but I think the process I went through was very illustrative. I rarely set out with the explicit goal of “finding a research project.” Rather, I found something I was curious about and just kept following my nose until a research project emerged.
One of the more direct ways to find a new research project is to follow up on a loose end in an old research paper. This worked well for me in the context of the k-terminal cut problem, where my first paper on the topic was inspired by an underappreciated lemma in an old paper (not mine) and my second paper was inspired by weird edge cases in my first paper.
On a separate occasion, I spent a lot of time learning about social network analysis on graphs. In particular, the literature on influence maximization and a metric called “modularity.” Although this only ended up affecting a small part of one of my papers, I was quite happy to have learned it. I enjoyed the process and pursued the knowledge for the sake of knowledge, not with the goal of contributing to the research. Well, at least not initially. It came eventually!
Uh oh. This section is about writing. If it’s not well-written, people are going to judge me. Or maybe it’ll be like my dissertation and only one other person will read it. Okay, here goes.
My main goal in this section is to explain the publication process in a way that I wish it had been explained to me when I was starting graduate school. Also, I’ll reflect on what worked for me in terms of learning how to write papers effectively.
When I got into graduate school, I did not really understand the difference between conferences, conference proceedings, and journals. It turns out there is a pretty big difference.
For a conference with proceedings, you submit a full paper to the conference. This paper is reviewed and, if accepted, it is published in the conference proceedings. This counts as a publication. In fact, in some fields, such as many subfields of computer science, authors rarely submit their papers to journals: conference proceedings are the primary venue for publications. As an example, two of my papers appeared in the proceedings of COCOA: Conference on Combinatorial Optimization and Applications.
For a conference without proceedings, you typically submit just an abstract to the conference or a poster or you are invited based on a previous paper you’ve published. You can present your work at these sorts of conferences, but they do not count as publications. The INFORMS annual meeting is like this. These are sometimes called “workshops” instead of conferences, but they have a similar feel to conferences. A smaller example is the Mixed-Integer Programming Workshop, where I presented some work in 2019.
Journals are a different beast. Whereas conferences guarantee to give you reviews within ~3 months, journal reviews are much more thorough and could take much longer. A conference reviewer is not expected to check your proofs, but a journal reviewer typically will. In fact, you are not required to take all the suggestions from a conference reviewer. When responding to a journal review, you must give an itemized response addressing each point in their review, either explaining the changes you made to address it or clarifying why a change isn’t necessary. There can be several back-and-fourths and the whole process could take more than a year.
Now, here is where it gets interesting. Typically, it is okay to publish the same paper twice: at most once in a conference with proceedings and at most once in a journal. Typically, in that order. You are expected to make at least one substantial improvement from the conference version to the journal version. In fact, sometimes the conferences will have a “special issue” of a journal, where a select subset of the conference papers will be invited to submit to the same journal. This happened to me several times. My paper which first appeared in the proceedings of Workshop on Approximation and Online Algorithms was invited to a special issue of Theory of Computing Systems. Both of my papers which first appeared in the proceedings of Conference on Combinatorial Optimization and Algorithms were invited to a special issue of Journal of Combinatorial Optimization.
Okay, so that’s the technicalities of the process, but how do you actually learn to sit down and write an academic paper? Again, a lot has been written on this topic, so I’ll just share a few pragmatic things that worked for me that I haven’t seen emphasized elsewhere.
When I was an undergraduate, I took a playwriting class. The professor told us that “great plays are not written, they are rewritten.” His point was that revising is the most important part of the writing process. The same is absolutely true of academic papers. The amount of revising I did of my academic papers completely dwarfed the amount of revising I had done on other types of writing up to that point in my life (granted, writing things was not a big part of my life before that… it’s all relative). It is important that, as a graduate student, you expect to do a lot of revising. When working on a paper, it was not uncommon for me to completely rearrange the paper several times and rewrite several sections from scratch. It was important for me to get over the “sunk cost” fallacy: even if I had taken a lot of time to write/organize a concept in a certain way, I had to be willing to rewrite/reorganize when a better way presented itself.
There is a reason for all that rewriting. There were a lot of times that I had an idea organized a certain way in my head, but when I put it down on paper it came out jumbled. Sometimes, this was just because the idea was hard to express. More often, it was because the process of writing the idea down gave me much more clarity. I realized gaps in my logic or connections that had not been clear in my head that were clear when written out. On a few occasions, my initial idea was just wrong and I only realized that through writing about it.
Some researchers advise that you start by writing an outline. I rarely did that. Instead, I just wrote down everything. I plopped all my thoughts down onto the page with only a very vague structure. Sometimes, I would have one thought interrupt another thought, so I would stop in the middle of writing one paragraph to go and write another. It was a very “stream of conscious” approach. I also started this process relatively early, as a way to organize my research before the paper-writing process even started.
Once it came time to organize a paper, I would print out my haphazard and sporadic writings and then try to make sense of some structure. It was like piecing together Frankenstein’s monster. Was it the most efficient approach? Maybe not… but it worked for me. It was very iterative. At first, I wrote based on a loose collection of ideas. Then, I pieced those writings together to have some structure. Seeing the writing organized in a certain way occasionally gave me conceptual insights I had not previously had and inspired me to connect ideas to each other and larger concepts. This informed a rewrite reflecting that new conceptual structure, possibly with some new ideas. This led to another reorganization, which led to another rewrite, which led to another reorganization, which led to another rewrite, and so on until there was a paper to submit.
Finally, on the more pragmatic side of things, for the first two or three years of graduate school, I used a local LaTeX editor on my computer. Managing the packages was an annoying process. Once I started using Overleaf, I never looked back! If you use LaTeX to type documents, I highly recommend trying Overleaf. It is so cool to not have to worry about installing the right packages. Once I discovered it, I used it for every project from then on.
On the other hand, I had certain colleagues who swore by particular citation managers, but I never developed a particular affinity for one. I used Mendeley, but I was not thrilled with it. It worked fine. More often, I went the old-fashioned route and printed out my papers and organized them by topic in different manila folders.
One touchy subject that you are going to run into in graduate school is authorship. One of the tropes of academia is that the order of authorship tells you who actually did all the work. The reality is a little more confusing. The truth is that the way in which authorship is handled boils down to tradition, which is different in different disciplines. The order of authors and the threshold for coauthorship are different even between pure math and certain fields of computer science. This is especially confusing when both fields occasionally submit to the same conferences! Generally speaking, there are two approaches: in one, the first author is whoever drove the project. In the second approach, the authors are listed alphabetically and that’s that. I personally prefer the latter, but there are arguments for both. Hopefully, you will have a good, fair advisor and there will be no drama in this and they will be clear about their expectations. If you do a majority of the work on a paper, you should expect to be either the first author or all the authors should be listed alphabetically.
As my close friends know, my advising situation was a tumultuous one. For a long time, I was technically advised by a certain professor. Unfortunately, that professor had severe anger management issues. Eventually, when dealing with that professor became unbearable, some other students and I switched to working with other advisors. I won’t go into any more detail here, but if you are considering working at Berkeley in the IEOR department and need more specific information about certain professors, don’t hesitate to contact me.
Mine is hardly the worst horror story when it comes to advisors. Unfortunately, most academic advisors receive zero training on how to advise students. In addition, at most universities, there is no contract specifying the parameters of the working relationship between advisors and their students. When my friends with industry jobs had issues with their supervisors, they could go to HR or just switch jobs entirely. That’s not to say it was easy, but there were established channels. In academia, there is often no central HR to complain to, and, if there is, it has little power. There is a stigma against switching advisors, too, especially later.
I’m definitely not one to talk when it comes to picking advisors, but I will pass along two helpful pieces of advice that I gathered from colleagues along the way. Both have to do with keeping your options open.
First, when it comes to picking a graduate school, you should choose a school where you feel there are at least three professors who you feel you could work with. Unfortunately, not everything is going to pan out. The person you want to work with might not have the funding or might want to retire early or might get poached by a big technology company. Plus, some professors have joint appointments that limit their ability to work with you or might have work styles that are incompatible with you. The point is that there are a lot of complicating factors. You want to make sure you enter a program with several good options as a hedge against getting stuck in a suboptimal situation.
Second, remember that you always have the right to switch advisors. Berkeley states this very explicitly in their policies, though it is buried and can be a bit hard to find: “Head Graduate Advisers should make sure that students are aware that they may change their Dissertation Chair.” Of the few people I know who were in a bad advising situation and managed to switch to a new advisor, all of them were happier with their new advisor, made more progress in their research, and ultimately graduated.
At the end of the day, as you go further along in graduate school, your goal is to become more independent of your advisor. Hopefully, by the end of your graduate school career, you feel confident enough in your research abilities to do research and publish papers mostly independently of your advisor (you’ll still work with them, technically, but less). That’s when it’s time to graduate!
Lest this section comes off as bitter, I’ll end on a positive personal reflection. I was very fortunate to be rescued from my old advisor by Professor Ilan Adler! I have nothing but glowing things to say about Ilan. Here are the things that I think worked particularly well in our advisor/advisee partnership, which are qualities that I think should be reflected in most good pairings:
As you can tell, I was very proud of the work I did with Professor Adler. Though it is tangential to the main discussion, I wanted to write this somewhere… Ilan’s advisor was George Dantzig, who famously developed the simplex algorithm. I’m very pleased to be part of such a fun “family tree!”
At one point, I was talking to someone who had had an extensive career in both industry and academia. They told me that academia had far more drama and politics than industry. Unfortunately, I have found this to be true. At least in part, I believe this stems from a lack of mobility: once a professor earns tenure, it is not an easy task for them to switch to a tenured position at another university. Thankfully, if you have a decent advisor you will hopefully be shielded from this politics. Most professors have the decency not to let students get involved in their squabbles.
I only have one piece of advice on this matter: take the time to get to know your department staff and be friendly with them. While the department chair might be the department’s leader in name, it’s generally not a highly sought-after position among faculty. The department chair will rotate out every few years and most faculty are too busy with their other duties to give administrative responsibilities much heed. The department staff, on the other hand, will often have a longer tenure in their positions and are the ones who really make the department tick. I think the staff goes underappreciated in many academic departments, but they are the ones who really know how to get things done!
On a few occasions, I have served on a panel where Juniors and Seniors considering graduate school asked questions to current graduate students. Of course, the number one question was always “Should I go to graduate school?” My answer, in one sentence, is “Yes, everyone should get a Master’s degree.” As I mentioned earlier, of my five years in graduate school I felt like the earlier years had the highest returns in terms of new knowledge gained. Plus, doing at least one small independent research project is a valuable learning experience. The decision to go all the way to get a Ph.D. is a personal choice. I don’t have any blanket advice for that case.
Another question that I get asked frequently is whether one should go to graduate school straight after their undergraduate. My advice is to work for a year or two before returning to graduate school. First off, practically speaking it helps to have a little money in the bank before starting graduate school. If you are fortunate enough to be able to work in a field where you can earn enough money to save some, then do so. Moreover, I think there are some skills that can be gained when working in industry that are undervalued in academia, but which are immensely helpful to have. One crucial skill in my field is the ability to write good code. Though the day-to-day research in graduate school is mostly solitary, in a way you are collaborating with the whole research community. At some point, you are probably going to write code to test your ideas. While you can get away with writing “research code,” it would be much better to write code that the community can review and perhaps even use. Again, that’s just my recommendation. Take it with a grain of salt because I didn’t always practice what I preach here!
Finally, within academia, I occasionally get asked by younger students if it is worth doing internships or if it is a huge mistake that will interfere with getting research done. My answer to this one is short: I enjoyed my internships. They gave me new ideas and made me a better researcher. Especially in my field, I highly recommend completing at least one or two internships!
Overall, I am quite satisfied with the choices that I made leading up to and during graduate school. I stumbled occasionally, but ultimately I am very proud of the body of work that I produced over these past five years!
In part, I made this document just to reflect on my own experience, but if it ends up proving useful to a prospective graduate student then I am happy to have made a difference. Good luck!