Tuesday 31 July 2012

It Is Done

It still hasn't sunk in. More than two weeks ago, I finished writing my PhD thesis, and a bit more than a week after that (printing, soft binding), it was handed in. Finishing it feels something like this:





My life in the near future isn't changing enormously. I am still going to be working in the same lab as a research assistant and eventually a post-doc (when my thesis has been officially marked, hopefully not more than three months from now). Same work, much higher pay! By staying where I am, I can delay making life-changing decisions for a little while yet. It's not a total cop-out, because I'll hopefully be doing some nice experiments and get some more papers under my belt.

Maybe I will be a scientist when I grow up! I've already done grown-up scientist things like give a talk at my old university. And thinking about most industry jobs makes my insides shrivel up with boredom. Things are looking dodgy at best around the world, so a job offer is nothing to sneeze at.

While I sit around waiting for my files to back up to an external hard drive in preparation for a long-neglected OS upgrade, I'll reflect on my thesis writing process that may or may not be useful to others in the future. Arguably, this is the only part of my PhD that took more or less how long I thought it would. Anyway, here is some of what I did. Keep in mind that this is most relevant to science (physics?) theses, since most people spend all their time doing lab or computer work and only do the write-up at the end.

  • Do your introduction and background theory last. If you've already done some kind of big literature review, then maybe not. In my case, I didn't. By starting with experimental design and results, I knew what I needed to include in the background chapter and didn't include too much irrelevant information on topics that have been discussed to death in review papers and textbooks.
  • This is probably obvious, but write an outline of each chapter and sections, including what sort of figures you want to include.
  • Use LaTeX. Seriously. Even all you biology people. LaTeX + Bibtex (for the bibliography) will make your life much easier and you will avoid having some kind of Word behemoth to deal with. If you can use html tags, you can use LaTeX. Even if you don't, it's very straightforward, and there are many online resources, including PhD templates (your university might even have a template). It is also completely free.
  • As tempting as it may be, don't work all the time. Leave some time for exercise and entertainment every day, even if it's just a short run and a TV show episode. Also, eat healthy and sleep! You'll be more refreshed when you are actually working, and you'll avoid the dreaded Thesis Gut. I planned out what I wanted finished in a given time, and stuck roughly to it. If I finished a section in time, I purposely didn't work in the evening. Everything fell apart in the last 3 weeks and I was working all sorts of hours (and felt like crap), but I kept it up most of the time and it made thesis writing a relatively pleasant experience.
  • Mix up writing and making figures. I found this useful to prevent getting bored of doing just one thing all the time. Also, I spent most of my office time writing and most home time making figures, since it was a bit easier to concentrate like that. In fact, I would usually assign myself a very specific task to work on at home, such as "make a diagram of blah".
  • Back up your thesis to multiple locations. I backed up to dropbox twice or three times a day, and to an external hard drive every couple of days. Nothing ever happened, but better safe than sorry!
  • Try to have all your data when you start writing. I had to go back to the lab for a few things, and it was bothersome and interrupted my writing flow.
  • Have someone not in your group who is knowledgeable about your thesis topic read your thesis in addition to your supervisor/coworkers and local grammar nazi. Their feedback will be particularly useful, because people already familiar with your work will easily understand what's going on, whereas someone unfamiliar with the project may not find things obvious that you and your group members think are obvious.
  • Build up a good figure-making work flow. Sketching out figures on paper before drawing them on the computer is helpful. I got much faster with making figures as time went on. I used Inkscape for making diagrams and Matlab for data plotting. laprint is quite useful for making Matlab plots look nice in LaTeX, but can be bad for figures with subplots. For those (except 3D plots, due to poor performance), I used plot2svg and tweaked fonts and stuff in Inkscape.
That's all I can think of for now. Does anyone else have any thesis writing tips?