Life as a Grad Student

As I finish up my Ph.D. at UCSD, I though I would write down some thoughts and reminiscences and unsolicited advice for new graduate students.

Time management

Learning to manage your time is very challenging and very important. Every day you need to be able to take a step back, scan the horizon for things coming up, evaluate what you should be working on. Sometimes this will make you take on (or drop) a major project...sometimes it will just help you avoid wasting time. There are plenty of time management books out there. The contents are mostly common sense, but the exercise of reading one can help you get committed to making the most out of your days.

Checking email twice a day is better than checking it constantly. Responding to emails in a batch is better than interrupting what you're doing. Turn off IM when you're working; it will drain your time and morale faster than television.

Note that managing your time in grad school is harder than it was in your undergrad days. As undergrads, we spend a lot of time doing problem sets. We know each problem is solvable - and we cry foul if we encounter a "trick question" that isn't, or that requires too much out-of-the-box creativity. Research work is not like that. It's a mixture of predictable tasks (read 5 articles), unpredictable tasks (track down and fix a bug), and open-ended problems (speed up algorithm X). Sometimes your professors will suggest you do things which turn out to be impossible, or unfeasible, or simply a bad idea. This is normal: Most professors have great ideas by having a LOT of ideas and having grad students help filter out the crazy ones.

Don't expect your advisor to micromanage your time for you. And remember that professors aren't always great at time management - they start meetings late and end them later, they turn in grants at the last possible day, they forget about reviews until the last minute. This is part of the academic lifestyle. If you manage your time in a non-academic way, you will have a big advantage over most of your peers.

Put research before classes. If you're in graduate school, you're likely a grade-motivated person, so it may be hard or just feel weird to put something ahead of classes. Bear in mind that grade inflation in grad school is such that if you put in a modicum of effort and don't run over the professor's cat, you'll generally get a B anyway.

Meetings in grad school can become excessive. You'll probably have a lab meeting, which is the one thing you don't want to skip. Journal clubs are invaluable at first for teaching how to present research, what negative controls to include...but not so much for the content (the stuff you learn from journal club is broad but not deep; you'd get more bang for your buck by reading a textbook for an hour). You may end up being expected to attend multiple journal clubs. At one point, I was expected to attend two lab meetings and three different journal clubs each week, and invited to two other lab meetings. There are two good ways to deal with excessive meetings. One way is to simply not show up. This works better than one might think, especially if you put in appearance once every so often. The other way is to go to meetings and quietly use the time to read or review articles, make to-do lists, and catch up on email.

Notes

Keep careful notes about everything you do, when you did it, why you did it. If you work in a wet lab you'll learn all about this, but if you don't...have someone show you what a lab notebook is, and you may be amazed at how methodical and thorough it is. On the computational side: Keep your source code under revision control (e.g. Subversion or CVS). Comment each check-in, and comment the code at the level of intent. Name your files descriptively (bad counter-examples from my hard drive: "DinobotHasSpoken.py", "barf.xls", "TempTempTemp.txt"). Consider keeping "readme.txt" files around to explain large data repositories - I once had to abandon a collaboration because a student hadn't kept track of the correspondence between directories of protein data and the patients they came from. Eight months after graduation, they (understandably) didn't remember or particularly care.

Your advisor, your committee, or reviewers of your papers will ask you questions that can be answered readily if you keep and organize all of your data organized. The same questions will cause huge headaches if you don't. If you try something, and it doesn't work, make a note of it - otherwise later, you might have to go back, and try it AGAIN, and then give numbers justifying the fact that it didn't work.

Professors

Most professors never took formal classes in how to teach - they learned by example, and from practice. Similarly, most haven't had training in their role as in how to manage. I had the advantage that my advisor came from industry, and put real positive effort into mentoring; other professors have a more "sink or swim" approach. Don't feel bad if you don't get as much mentoring and hand-holding as you'd want.

Choosing an advisor is the most important thing. But remember to also cultivate relationships with other professors. You need five people for your committee, and that tends to require some reaching out to come up with the last one or two people. Being a committee member is not a huge committment, but it's also not something you want to be asking a stranger for.

Respect your advisor's time. If you need to blow off steam (e.g. "I can't believe this stupid journal is making me convert all my images to EPS format!"), try to save it for your fellow students. Bring paper to your meetings to take notes. Consider bringing an agenda; advisors would often prefer to sit back and advise and let you drive the meeting.

When you need to meet with your whole committee, DO NOT ask them for a list of times that are convenient. This sounds polite, but it is more bother in the long run, because the lists never overlap. Ask them for a list of days and times that they are absolutely NOT available (e.g. because they'll be in China). Make sure to get your advisor and co-advisor physically in the room...the others technically should all be there, but in practice, they may not.

Sometimes professors will request you do various academic chores - all the little undesirable tasks that don't advance your research, like reviewing papers, organizing meetings, purchasing supplies, etc. You might be "voluntold" to do things (e.g. professor X asks you to do something by cc'ing you on an email to profssor Y saying that you'll do it). Do fair share of the grunt work...while keeping your research as the top priority. Once you're getting assigned as much stuff as you can handle, bring this up with your professor; most will understand when you can't agree to every request. It's fine to make it clear that you put your advisor's requests before those from other sources.

Conferences

The purpose of a conference is to network with other scientists, especially potential collaborators (your first years of grad school) or employers (your last year or so). Why do you think they have three coffee breaks between breakfast and lunch? That's right, to encourage everyone to sit down next to strangers and start up converstations and exchange ideas. At your first couple of conferences, have your advisor (or fellow students) introduce you to as many people as possible. As for the presentations themselves, some will be excellent and get you fired up to do more work. Others will be science you don't understand, presented by people with poor public speaking skills. So it goes.

A poster may get you to a conference of interest (e.g. by convincing the department to fund travel), and it can give you something to talk about when you do your networking. If you do a poster, remember to bring an envelope, and stick it up along the bottom of the poster board, so that people can drop in their business cards; you are expected to send them more information (which can be as simple as a brief thank you and a PowerPoint file). Bear in mind that posters have very little impact, and the time you spend on a poster is time you could have spent on a real paper.

Publishing

Publishing papers is important. They capture the essence of a body of work in a way that's of general interest. They give you a clear goal and positive feedback. They help keep the attention of collaborators. And when you finish grad school and look for a job (especially an academic job), people will look at the list of papers with your name on them.

Start writing the paper early. You can write the Methods section while you are doing experiments, fill in Results as they trickle in, draft the Introduction before you have done any real work. This helps limit the amount of time you'll have to spend digging up forgotten details. And it brings up early the awkward questions of where to publish, what programs to use to generate figures, author ordering, etc.

The primary goal, of course, is to write a brilliant and lucid paper that gets many citations. But be aware that most of the people who evaluate your work won't actually read and digest it. Instead, they'll use their own quick-and-dirty scoring system involving journal names and the ordering of author names. (Lists of impact factors are one guide to journal prestige levels, but different people value journals differently!) At least in biology / bioinformatics, papers with your name as first author are worth the most by far. The other author positions seem to be essentially equivalent, and not really worth fighting over. A convention in many labs is: Grad students are listed first and PIs last, with the grad student who did the most important work as first author, and the main PI as last author.

Start building up a library of papers you can refer back to, and cite. You will likely be citing the same papers over and over again, if you continue work in an area. You don't want to have to look the references up repeatedly, or download PDFs repeatedly...that is tedious and error-prone. The only thing more embarassing than misspelling a cited author is having one of your co-authors email you to say that you spelled his name wrong! Yes, I made both of these mistakes; thankfully they happened BEFORE publication. Also, at one point we lost all greek letters from a manuscript when converting a paper between word processing tools, so that our methods described an elution column 0.5 m in diameter (rather than 0.5 micrometers)...but that's another story

Your database of papers can be as simple as a text file where you store citation information, and PubMed IDs, and your own notes and reactions to papers. EndNote is a nifty-looking program for keeping track of references. If you use Latex (and if so, may God have mercy on you, you poor schmuck), maintain a master .bib file. In the olden days, people kept fat manilla folders with inked-up printouts of papers - even this is much better than no system at all.

Some papers have "multiple first authors" - that is, there's a footnote in the full paper declaring that the second author contributed equally. If you're in a situation where authorship is closely mixed, try NOT to end up doing this. People don't read these footnotes - they'll rate your publication record by scanning through a bibliography. If you're not first, you're better off just agreeing to be a second author and getting something real in exchange (like "you'll be first author on this other paper" or even just "we'll buy you lunch"). Similarly, it's always nice to be named in an acknowledgments section, but it's not necessarily worth asking for it - it's better to save up that bit of leverage than to be repaid by text that nobody reads. (On the flip side: If you are first author, don't hesitate to hand out these "consolation prizes")

Free Time

A friend of mine told me that her advisor told her to be sure to have one day in graduate school that would be free time, with no need to think about research at all. And I told her, "Save that day up - don't blow it in your first or second year!" And there was a pause as she got the joke. And then she laughed. Then she looked a little sad.

But seriously - we all need to have some down-time. People have a finite amount of energy for research, and when the energy is used up, additional time grinding away in lab is likely to be spent chatting or browsing the web. Take pride first in how much you can get done, not in how much time you spend living in the lab. And go for quality first in your free time, too; find things that can recharge you without sucking up a day. I knew one student who took up kayaking, which he loved...but which probably set his graduation back a year.

The Home Stretch

At some point, you'll start to think it's time to graduate. You'll notice that your list of publications no longer fits on one screen, and you'll think about how nice it would sound to hear "Doctor" before your last name, and you'll wonder just what you have to do to graduate. Well, ultimately, what you have to do is persuade your committee! Talk with them, starting with your advisor - and bring it up early on, to manage expectations. For me, simply deciding that I had one Ph.D. worth of research was an important first step.

A common source of friction is that advisors want to keep their (most valuable) senior students around for just a little longer. (It's very rare for advisors to be pushing students out the door - if that happens, the student is probably long overdue!) In the last year or two, it's normal to feel a little burned out, jaded, discouraged, disillusioned, or cranky. I've heard of students who even tried ultimatums (ultimata?) like "I'm leaving in May, with a Ph.D. or without it!". Of course, that doesn't seem like the best strategy (even if their bluff wasn't called, what kind of a recommendation letter would they get?) Two ideas that may help sweeten and speed up the exit process: If you can commit to some post-graduation work (like continuing to help junior students for a few months), your advisor will feel less alarmed about letting you go. And once you have a job or post-doc lined up, your committee will feel some positive pressure to let you go.

Writing a dissertation is time-consuming. Be prepared to spend days just following the formatting directions, setting up margins, fixing fonts, merging reference lists. Take the time to figure out how to get your word processor (Word, Latex, whatever) to do things like figure numbering and tables-of-contents for you. The formatting requirements are strict, and not always logical. "Why are figure captions below figures, but table captions above tables? Why, in the twenty-first century, do I have to care about tables wrapping across printed pages?" These are questions to keep to yourself, or to vent to fellow students who are at (or past) the dissertation-writing stage, since junior students may not be so sympathetic about the nuisances afflicting someone who is graduating!

Spend the most time on your first chapter. This is where you explain why your research matters, and the broader context it fits into; chapter 1 must answer the "So what?" question. This is also where you get to give your view of the field, and where you get to state open problems. Start chapter 1 first, but finish it last.

You should definitely re-use published papers (especially ones where you're first author) as chapters in your dissertation. It was very gratifying for me to see my dissertation jump from around 30 to over 100 pages after inserting some publications. There's no formal page requirement (this isn't High School English, after all), but 150 pages is generally a reasonable target.

Go see a dissertation defense before your own. I learned a lot by doing that: I added some big-picture slides, and I re-scheduled for a much larger room!

Your defense itself should be anticlimactic. (In fact, the whole process of writing a dissertation can be anticlimactic if you have a great publication record) To help ensure that things go smoothly, meet with everyone on your committee after emailing out the dissertation and before the defense. If they have concerns, this is where you bend over backwards to address them. And if they're already content...then it is a perfect chance to do some networking, and get more contacts for a post-doc position or job in the Real World. This may be your last one-on-one meeting with them, make the most of it.