Monday, November 5, 2012

How to be an Awesome Postgrad Student

Here are a few tips on getting through a PhD or a Master’s, based on my experience doing a PhD in Electrical Engineering at the University of Canterbury. I’ve had a good time so far, and I hope that you can have the best experience possible. This document is based on ideas that I gathered from other written works on how to be an awesome postgrad, my supervisors, friends, parents, wife and various talks that I attended as a postgrad. I hope you can translate any discipline-specific details into your own field. Your PhD experience probably be unlike mine, especially if you are not doing electrical engineering. Get advice from wise scholars in your discipline on how to do be an awesome postgrad student in your context.

1 Before you begin

If you’re about to start a PhD1 then you’re probably motivated by mastery, autonomy and purpose, rather than by survival or money (Pink2011). In a PhD, you’re likely to have plenty of opportunities to develop mastery and autonomy. After all, a PhD is about developing the ability to do independent research. But trouble may come with the ‘purpose’ bit. Before you start, make sure you have a clear idea of why you are about to embark on a PhD. You will encounter discouragement along the way, so I suggest you write down your reasons for starting a PhD. Before I started my PhD, I wrote a letter of encouragement to my future, discouraged self; reading my short letter to myself can put shivers of inspiration down my spine. You will get discouraged, so be proactive while you’re feeling motivated and write something inspiring that will help carry you through the bad times. I’ll discuss the emotional roller-coaster of the PhD experience in more depth in Section 7.

Getting a PhD (or a Master’s) is not for everyone. Don’t do a PhD just because you are smart (and don’t discount the possibility because you didn’t get straight A’s). The chance to do a PhD is a rare opportunity, but that is insufficient justification for doing one. You have to work out if you want to do a PhD.

Have some idea of what you’d like to research, but the exact topic is not all-important. You’re learning how to do research, you are changing – that’s what’s important. You get 6 months to decide specifically on a topic in your proposal, but your topic may change a number of times after your proposal in strange, unpredictable ways. Be aware that you also have the choice of what university to attend. If you don’t like the research that’s happening at one, you can chose another. I’ll give you more detailed suggestions on choosing a topic in Section 3.

Try to pick a good supervisor. The quality of a student-supervisor match is unpredictable, even with extensive gossip. Think of it like gloves and hands. Some supervisors are comfy wool mittens that almost any student could partner with and achieve great success. But you might be the one in a million who is allergic to wool. Other supervisors have a more mixed reputation, but you might get on fine with them. If you’re coming in from afar and you have not had a chance to interact with current and previous students of your prospective supervisor, then at least attempt to do some research on him or her. See Section 4 for more suggestions on relating to your supervisor.

A PhD is not like a job or like undergrad. Good marks do not necessarily imply that you will be great at research. A PhD is not a continuation of an unbroken succession of progressively more advanced schooling programs. It’s something new. In all previous schooling, you were being taught stuff that humanity already knew. Now you’ve got to figure out stuff for yourself that no one else knows.

2 Thesis – this is key!

To get a PhD you must have a thesis: a grain of an idea that your research demonstrates. A short essay by ACM Distinguished Dissertation award winner Olin Shivers explains what a thesis is: www.cs.ucla.edu/~palsberg/shivers.html. So if you’re doing a PhD, your objective is to present written research that supports your thesis. If something brings you closer to a well-supported thesis, then it brings you closer to your goal. If it doesn’t get you closer to a well-supported thesis, then its value to your PhD is questionable.

Your thesis is proven with a written document. You’re becoming a writer, as well as an engineer, scientist, or whatever your discipline calls its practitioners. Therefore, writing is paramount. Learn to write. Read ‘The science of scientific writing’ by Gopen and Swan (1990). If you have to write mathematics, then read ‘How to write mathematics’ by Halmos (1970).

It’s hard to show up at your desk on day one of your PhD and just start writing your thesis. One way to break up a thesis into more manageable chunks is to start by writing a conference or journal paper. Writing a research paper is a much less daunting task than writing a thesis. Once you’ve written a few papers, you use that research material to assemble a thesis.

3 Topic

When you choose a topic, you need to balance your own passion for a subject with other people’s passion for that subject, especially your supervisor’s passion for the subject. Try reading ‘The Art of Doing Science and Engineering: Learning to Learn’ by Hamming (1997); Hamming has a number of ideas on how to do research but he also talks about the value of seeking out important problems. Doing important research gives you a sense of purpose, which helps you stay motivated, as discussed in Section 1. If you strongly believe in the importance of your research, then you will be more able to deal with the negative emotions that come up in the process of doing your PhD.

Is your research topic useful or relevant? I would encourage you not to stress about this question. Every research project is a bet. Most academic papers are not highly cited. But we can’t tell which research papers will be highly cited five years from now. So take a long-term, broad perspective. Your research may be useless and irrelevant, but it’s also possible that if you don’t do it then humanity will miss out on some amazing breakthrough. No one knew what would be useful now. We look back and we think that the great scientists were geniuses (maybe they were, maybe not), but much of their success came from luck.2 Don’t be discouraged, do your research and buy humanity another ticket in the research lottery. We all have to play if someone is going to win.

Remember that your PhD’s most useful contribution to humanity may not be your research but rather you. You’re learning to do research. You may do a lot more research after your PhD and that later research may be more useful than your PhD research, especially since your post-PhD research can be larger in volume. But you have to start somewhere.

In a PhD you must make an original contribution to human knowledge. Given that science has been around for hundreds of years and humanity has been around for eons longer, this sounds like a difficult, impressive thing. But human knowledge has many dimensions. Hamming (1997) tells us that high-dimensional shapes are mostly edge. If you plot a straight course in a random direction in human-knowledge-space, you can probably reach the boundary of human knowledge by reading five books and 100 journal articles. People do it all the time, and they get PhDs. Because a PhD has laser-like focus on a particular topic, being original is not so hard. A child born today, not knowing that he or she has hands, can extend human knowledge in less than thirty years.

Another encouraging exercise is to examine published research and apply the same level of scrutiny to that work as you would to your own. Next time you’re in a seminar, ask yourself ‘Does this speaker ever worry if their research is relevant? Should they?’ Next time you see a paper that uses statistics to show that a particular phenomena occurs, look at the confidence level. One in twenty papers that show something with 95% confidence are rubbish just because of blind chance, to say nothing of experimental error, unnoticed systemic biases or any other source of error. Be bold in the face of your doubts, because other people’s published research is not above question either. This is the scientific process.

4 Supervisor

Your supervisor is there to guide you, but you should not depend on your supervisor to magically volunteer all the help that you need. You need to proactively manage your relationship with your supervisor. Ask your supervisor for help when you need it. Email your supervisor to set meeting times and propose an agenda for each meeting. Send your supervisor written work. If your supervisor can’t understand something you’ve written, then don’t expect a reviewer or thesis examiner to get it!

You need supervision, but you also need autonomy. When you first start you’ll be more dependent on your supervisor. You’ll be new to research and you’ll need more guidance. Then as you enter the adolescence of your research, you’ll know enough stuff that you can create ideas yourself and you might even disagree with your supervisor on some things. Hopefully, you’ll graduate the moment you hit research adulthood.

Most of the time, being a postgrad is like being a teenager. You aren’t fully dependent on your supervisor, but you aren’t fully independent. You aren’t totally wise, but you aren’t totally foolish. You’re a sophomore; a wise (sophos) moron (moros). As the teenage years are a balance between trying things out independently and depending on wise survivors, so are the postgrad years. As the teenage years are full of angst, so are the postgrad years. Some people grow up with a great relationship with their parents through their teenage years, so be hopeful that you can have a smooth relationship with your supervisor. If there’s tension in your relationship with your supervisor, try to talk it over or get help.

5 Workflow

The academic world can be modelled as a huge written conversation. As a new researcher, you’re trying to engage with the academic world by joining this conversation. Most of us were born with two ears and one mouth, so get listening. One key way to listen to the academic conversation is by reading papers. Another way is to attend conferences and listen literally. Reading papers can be exciting, if the material is well-presented and exciting. Papers can be boring, if the material is poorly presented or boring. When you join into the academic conversation by writing an article, you have to demonstrate that you’ve listened well enough that you aren’t repeating someone else’s idea. Therefore, you need to understand and reference the relevant related literature.

I recommend that you keep a working document with summaries of every paper you’ve read. Read a bunch of titles and save references to all the interesting papers (using Endnote or Zotero, discussed in Section 6). Read the abstracts of the interesting papers and print off the ones that you want to read. Then read a paper with a pen and a highlighter in your hand, engaging with the material as you go, making comments in the margins and highlighting key sections. Finally, write a summary of the paper. (Of course, you can interrupt that process at any point if you find that the paper is actually not related to your topic, or if it’s useless to you for some other reason.) That way, when you go to write a paper, half of your related work section and half of your background section are already written! If you don’t do this, you’re more in danger of having to go back and dig up a reference to a paper you thought you read once that said something relevant to your topic.

Write, write, write! Written work can be included in your thesis write-up. Your supervisor can critique it. You might be able to publish it. If you do lab or field work, keep meticulously detailed records and notes. Those notes may be the only lasting products of your, so make sure they contain all the information you’ll need to translate them into well-communicated research. How do you know if an idea of yours is clearly defined or even real? Write it down! Work is not done unless it’s out of your head. Real completed work is computer code that works, scribbles in lab books, chemical solutions made up, etc., but mostly written material for supervisors to see.

Try to accomplish something measurable every day. If you haven’t measurably accomplished something today, then how do you know if you’re a day closer to completing your thesis? Even if you make great measurable progress, you may end up throwing it away. Don’t worry about this. It happens. That’s research. It wouldn’t be research if we knew what we were doing! Make concrete, measurable progress and don’t be paralysed by the fear that your current approach might fail. A failed approach might close off a of range possibilities; you could write about that in your thesis. If you got good grades in your undergraduate years, you might get discouraged by the PhD environment, where positive encouragement is rare and failure is common. It may help to read over your undergraduate transcript and remember your successes. In research, you don’t become a failure by failing over and over again, you become a success by succeeding just once. If success hasn’t come to you yet, keep waiting and keep working.

Here are some examples of measurably accomplished research:

  • Finding and saving an interesting reference.
  • Reading an article, highlighting key passages and summarising it.
  • Reading a related book and writing a summary. You may need to write a summary for each chapter if the book has many relevant ideas.
  • Writing the first draft of a paragraph on a research paper.
  • Filling out an ethical consent application form.
  • Writing a list of materials that you need and giving it to your supervisor or a technician.
  • Doing an experiment and writing a detailed summary of the results in your lab book.
  • Drafting a one-sentence ‘thesis statement.’
  • Printing some of your own written work and handwriting editing corrections. (If you have the right software, get your computer to read your work aloud to you. It can be embarrassing to hear all your mistakes aloud; I used headphones.)
  • Interviewing a research participant with a recorder going.
  • Designing something in computer aided design software.
  • Writing and testing a piece of computer software.
  • Writing one page of your thesis.
  • Incorporating your supervisor’s corrections in your written work.
  • Making written corrections to your journal paper proof.
  • Printing and handing in your thesis!

One way to broaden your horizons and to learn a variety of skills is to embark on small alternative projects. Alternative projects can also be called ‘diversions’ or ‘distractions,’ so your supervisor’s support (or ignorance) of such activities could be pivotal in whether or not you use this technique. Spend an afternoon building something cool. Read and summarise an (apparently) irrelevant paper from an unrelated discipline. Visit a museum. Learn a new computer programming language. Start a blog about your research. I don’t know if doing things like this has helped me complete my degree faster, but I know that they helped me have fun and stay sane while I worked on my thesis. A rare few of my unrelated projects have ended up being useful to the mainstream part of my thesis. In one case, I turned a side project into a conference paper in a research area that is only tangental to my thesis.

Work on improving your research skills. Read ‘How To’ guides (like this one), go to the workshops run by your university and get advice from seasoned academics. See also ‘How to do Research At the MIT AI Lab’ (Chapman1988).

6 Tools

You’re going to need good tools to complete your thesis. When I started my PhD, I deliberately chose software tools that were freely available, ideally ones with permissive software licenses,3 over commercial ones. You could use Microsoft Windows, Endnote to keep track of references, Word to write your thesis and MATLAB to run your simulations. One problem with those tools is that you have to pay to use them. So if you start a business after you finish you’re PhD, you’ll be paying high prices for corporate licenses. Alternatively, you could use Linux, BibTEX to keep track of references, LATEX to write your thesis and Python to run your simulations. No license fees there, and you can freely distribute Python in your commercial software. License fees may not matter in your case, but it’s something I thought about.

Zotero (zotero.org) is a great tool for keeping track of references. Zotero integrates with some web browsers, so you can save references straight out of Google Scholar and many other web pages. When I want to use a reference in LATEX, I export the reference out of Zotero in BibTEX format, then I run a custom made Python script that adds the reference to my master PhD BibTEX file. Qiqqa is another free reference tracker (www.qiqqa.com).

LATEX is great, especially if you like programming. Figuring out the error messages can be hard, but completely changing the formatting of a huge document is easy. You may prefer Word, that’s ok, but I think LATEX is not that hard to learn.

Backup is essential. When you’ve got multiple years of your life’s work stored on a computer, you don’t want to lose it! Make backing up easy. Hard disks are cheap; get a big external disk and software that does automatic backups on regular intervals. Then leave the hard disk connected and the backup software running. Mac OS X does this well with the Time Machine software.

In addition to generic data backup, I recommend that you use a version control system (en.wikipedia.org/wiki/Revision_control). For some time I used Subversion (subversion.tigris.org). Subversion is easy to use and relatively easy to understand, but it keeps all your version history on a remote server. That’s great, especially if you lose all your local data from some reason. But it’s bad if you get disconnected from the remote server. In the September 2010 Canterbury earthquake, and again in the February 2011 Christchurch earthquake, I lost connection to the my department’s Subversion server. I didn’t like that, so I switched to Git (git-scm.com). Git has the advantage that every repository has a copy of the revision history, but I find Git more confusing to use than Subversion. Now I mirror my Git repository on my department’s Subversion server (using git-svn) and also on a flash drive that I carry in my pocket. So my thesis is stored in several places at once: on my computer (the most recent version), on my backup hard disk on my disk (usually about 15 minutes old, sometimes much older), on my department’s server (usually a few days out of date) and on my flash drive (usually weeks or months out of date).

Use scripting to automate those parts of your workflow that are purely mechanical. No sense repeating work that you don’t have to repeat; it’s better to write a script once than to repeat a laborious process many times. I use Python for scripting, but Perl and Bash are good too. Get Matt Might’s “My Ph.D. advisor rewrote himself in bash” scripts: matt.might.net/articles/shell-scripts-for-passive-voice-weasel-words-duplicates. Matt Might also has a number of other good articles on academia and the postgrad experiences, so explore his website a bit.

Amdahl’s law (Amdahl1967) implies that researchers should program with the programming language that enables them to get results fastest, rather than in the programming language that makes the fastest code. Generally, I think this means that researchers should use high-level languages, like MATLAB, Python, or Ruby. I use Python for my research, along with Numpy (numpy.scipy.org), SciPy (www.scipy.org), Matplotlib (matplotlib.sourceforge.net) and PyGnuplot (sourceforge.net/projects/pygnuplot).

You’re going to need a computer. Some university departments provide computers, others do not. Some will try to provide one that you don’t like for some reason. Check on whether or not you’ll have to get your own computer equipment. In my opinion, a minimum is to have a keyboard and mouse that you feel comfortable using for extended time periods. Adding a second screen can be really helpful. Having a laptop is useful if you do research in more than one place.

Perhaps the most important tool that you have for doing research is you. If you start getting wrist pain from using your computer, go see a doctor or a hand physiotherapist. I got some wrist pains and they only started to improve when I got professional help. Eat healthy foods and get exercise. Be socially active; go to parties. Maintain yourself.

7 Emotions

The rhythm of emotional highs and lows is a natural part of the PhD experience. The way you feel about your PhD is normal for people who are doing PhDs, but doing a PhD is unusual. This is why you feel weird and disconnected, and why you have an unusually small amount of support from those you know. A PhD is a unique experience, so others are less able to understand you. So make friends with other postgrads who are going through a similar thing. Personal support is essential. Maintain your friendships.

Doing a PhD is not about being super-smart. It’s about persistence. A PhD will take you through valleys of despair; don’t give up! Hopefully, your supervisor will encourage you in the emotional dimension of your PhD. The first time you have a paper rejected is a unique, devastating experience. Follow all the advice that you get about this experience: don’t take it personally, try to improve your paper with the reviewer’s suggestions, etc. Also realise that it’s a common experience and that it’s normal to feel bad about getting a rejection. My first paper was rejected twice and modified a number of times before I got it published. I’m super-proud now, but I did feel unhappy for a while. I took many of the reviewers’ suggestions and the paper is much better now.

You will probably suffer from impostor syndrome (Harvey and Katz1985Laursen2008). You’ll forget something from your undergraduate degree and suddenly you will think that you’re a fake, that you aren’t that smart and that sooner or later, someone’s gonna find out that you’re a fraud. But that’s not true! Many academics get the same feeling. Read about impostor syndrome and try to remember that the more you know, the more you know you don’t know. Your increased knowledge of a field implies even greater awareness of the gaps in your knowledge. Don’t fret about it. Keep reading and try to laugh when you forget something that seems basic. You are becoming an expert. A non-expert would be blissfully unaware of that basic thing that you learned, forgot and quickly re-learned.

8 Before you quit

Because a PhD is a roller-coaster of emotions, it’s a bad idea to quit at the first sign of trouble. However, your PhD may go so badly that it would be much better for your to leave and get a job. It’s ok to quit. University is expensive self-knowledge. You might have gotten unlucky with your selection of supervisor or project.

Beware of the sunk cost fallacy. If you put a year or two of your life into a degree that you later detest, you may feel that you have to complete in order to justify your past. But the best thing to do is to look only at the present and the future: is it worth spending the remaining X years and Y months to get Z degree? It doesn’t matter that you’ve already spent A years B months already. It’s easy to do the math, but emotions have great power over us.

Try to separate the facts from your immediate feelings. You may feel that your PhD is pointless, but what is the fact? You may feel that you really want to be called ‘Doctor,’ but what objective benefits will that have for your life? If you quit, one fact is that you won’t have the degree. If you continue, one fact is you may continue to feel miserable. If you quit, one fact is that you may have to search for a job. If you continue, one fact is you may be glad that you persevered.

Feelings are important. I’m not suggesting that you make some Spock-inspired emotionless decision. Beware of feelings; your emotions can oscillate. Try to objectively examine the facts about reality and the facts about your feelings. You might decide to quit because you believe that you will feel horrible for a long time. Or you might decide to continue because you are willing to endure a bit of an unpleasant experience to achieve your goal.

I recommend that you try to make changes before you quit. Get lots of advice from people you trust, from your supervisor and from an objective academic (i.e. a student support person or another academic who is not your supervisor). Ultimately, the choice to complete your degree (or start one) is yours alone. A PhD is a highly personal endeavour and your decision will mostly impact you. If you do quit, then remember that the world outside of the university environment is full of wonderful things! If you chose to continue, then focus on the wonderful dimensions of the academic world; you get a chance to do Science!

9 A final encouragement

I’ve tried to be realistic in my assessment of the PhD experience and I’ve dwelt on some of the bad aspects. A PhD is a hard journey, but it can be great fun and you can advance the state of human knowledge. Whether you decide to start a PhD, to continue one, or to discontinue one, take an optimism outlook. Count your blessings: get a piece of paper and write a numbered list of the good things in your life. Even if you feel unhappy, smile at someone today.

References

    Amdahl, G. M. (1967). Validity of the single processor approach to achieving large scale computing capabilities. pages 483–485.

    Chapman, D. (1988). How to do research at the MIT AI lab. MIT Artificial Intelligence Laboratory.

    Gopen, G. D. and Swan, J. A. (1990). The science of scientific writing. American Scientist, 78(6):550.

    Halmos, P. (1970). How to write mathematics. Enseign. Math, 16:123–152.

    Hamming, R. R. (1997). The Art of Doing Science and Engineering: Learning to Learn. CRC Press, 1st edition.

    Harvey, J. C. and Katz, C. (1985). If I’m So Successful Why Do I Feel Like a Fake: The Impostor Phenomenon. St Martins Pr, 1st edition.

    Laursen, L. (2008). No, you’re not an impostor. Science Careers.

    Pink, D. H. (2011). Drive: The Surprising Truth About What Motivates Us. Riverhead Trade, reprint edition.

    Taleb, N. N. (2007). The Black Swan: The Impact of the Highly Improbable. Random House, 1st edition.

Notes

1If you’re doing a Master’s degree, then you’ll have to mentally replace ‘PhD’ with ‘Master’s’ throughout this document. I think the same things apply, but then, I never did a Master’s degree.

2Great research results come from unpredictable places. Read ‘The Black Swan’ by Taleb (2007) to learn more about unpredictable, high-impact events. Pay special attention to Chapter 7, where Taleb discusses the psychology of searching for Black Swans.

3In a nutshell, open source software comes in two varieties: software that you can include in commercial products without being required to release your source code (permissive, like the BSD or MIT licenses) and software that requires you to release any modified source code (copyleft, like the GPL).

This post is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License.

No comments: