Sea of Troubles

Things I Learnt From Writing a Thesis

You can never research enough This one was unfortunate. Half-way through writing my research proposal, I discovered a book. This book covered most of what I was originally intending to research. Fortunately, I managed to reposition my work. But, if I hadn’t discovered it until later, it would have cost me hours and hours and hours. So, make sure you read every paper even remotely related to your work before you write a single line. Doubly, write down every single paper you read, in a way you can find it again later.

Use LaTeX As far as I know, I’m the only person I’ve spoken to about thesis who hasn’t constantly bitched about Word crashing, being slow, losing revisions, etc. By using LaTeX, I didn’t have to worry about being tied to one application, being unable to diff version properly, or laying out images wrong. Best of all, it seperates rendering text from writing it. This means that while I was writing I couldn’t get distracted by layout or fonts, or anything else.

Use CVS The most valuable thing you can possibly do is implement some kind of version tracking. It helps you roll back if you screw something up, it can help you prove when you wrote or thought of something. Best of all, it helps you make backups (cvs update is your friend).

Use automation (makefiles, graphviz, etc) The more you can automate your task, the easier it gets. In my case, part of my thesis included some text generated from an XML file via XSLT. It’s a real pain to have to go through the generate/copy/paste cycle every time you want to print, or get a page count. Same goes for the Bibliography. But, I cheated, what with laziness being a programmer’s virtue and all. :) My happy little makefile updated my cvs, commit it, ran xalan over the XSLT/XML files, and built the PDF. Every time it built, it picked up the latest images, and included any changes I’d made to the various support files. Try doing that in Word. :)

Use paper Computers are seriously awesome for some things. Searching is absolutely one of them; spotlight in OSX tiger, and DEVONthink before it were incredibly valuable for finding documents using only a word or two. DEVONthink is particularly good at grouping documents, and at finding documents related to each other; this is very good for literature reviews. :) Computers are incredibly awful at other things. You should never, never, never, never think at a computer, especially if you’re doing conceptual work. Among the list of tiny-things-that-saved-my-thesis was the fact that I bought a flip-chart pad (it’s just butcher’s paper). From that point forward, all my big-picture thinking and organising was done on it in sharpie and the results strewn across my walls. Then I switched to index cards to track each paper in my litrev. Suddenly, I could turn around in my chair and see the entirety of what I was talking about. I could shuffle, reorganise and group the papers I was citing. In fact, I can show you the structure of my literature review on the big diagrams I drew to organise it, and show where each card fits on that diagram. The trick here is knowing what tools are good for which jobs. For my purposes, paper was a much better solutions for some problems. Give it a try if you’re not getting done what you need to.

Carry a notebook everywhere One of the scary things about thinking about the one problem/solution so much is that you end up thinking about it in the strangest places. Some of the most useful thoughts I had or decisions I made were in the strangest places. Some of them, I lost; some of those things I really wish I could get back. Once I got back into the habit of carrying a notebook, that problem kinda went away. :)

Start early Yeah, yeah, yeah. Teachers have been telling you forever, and you haven’t failed yet. Let me tell you, this time, they’re right. Your deadlines will sneak up on you, and Barry’s Law applies: “Everything takes longer than you think, even if you take Barry’s Law into account.” This is doubly super-true if you’re doing engineering/”design science” research, becuase you’ll keep tinkering with the system, and putting off the writing-up. This will result you working your ass off in the last week, no matter how hard you work the rest of the time. :)

Be careful with your senses (hearing, sight) This one caught me by surprise. Be aware of what kind of work you’re doing, and what else is appropriate in your environment. If you’re reading papers, you can probably get away with some background distraction; first time mindless scanning you can probably even do in front of the TV. If you’re writing, background music is okay, so long as you keep it low. And if you’re doing heavy thinking, just turn it all off. If you can, stick to playing music on music players (not computers) when you’re not actually using the computer; it’s more distracting than you imagine.

Try to avoid having a line-of-sight to anything moving. Any kind of movement, especially late at night, will trigger your primal “Oh noes! I’m being attacked!” sense and completely throw you off. For our purposes, this means closing any door between you and a hallway, turning off the tv, maximising the window you’re working in, turning off anything likely to pop-up (like IM), etc.

Make time for entertainment and relaxing, or you’ll lose days! Compensate for your hyper-attention and sudden entertainment drought (you /did/ turn off the TV, right?) by making time to relax. When you book out hours (or days, toward the end) to do the work, make sure you block out at least 1/10th (or 1/5th) as much time to relax. If you’re working ten hour days, make sure you have at least an hour where you /don’t/ think about what you’re doing. Once a week, go out of the house and watch a movie or buy a friend coffee.

If you try to work too hard, for too long, you WILL burn out. I lost an entire week this way; I just couldn’t do anything. Far better to take a break preemptively for a few hours, than lose ten times as much through burnout.

Try to keep to your sleep cycle If you let your sleep cycle slip (e.g.: staying up late and sleeping late), you will eventually find yourself coming nearly full cycle. For a while there, I was sleeping at 6am and getting up at 9am, and that hurts.

Turn off the internet The internet is the devil. Turn it off. Yank the blue cable out of your computer, and the wireless access point (if you’re lucky enough to have one). I know you don’t believe me; just trust me.

Check out 43folders :) Riiiiight before you rip out that blue cable, check out for some neat ideas about organising yourself, and the merits of index cards.

Sirius doesn’t work at night. This is pretty UNSW-specific, so I saved it for last. Sirius (the database that gives access to all the e-journals, safari, etc) doesn’t work “at night”. I don’t remember exactly, but it goes off some time after midnight and comes back up in the early hours (5 am?). This is incredibly painful if you need to grab a last minute reference or paper, so plan for it.

That’s it! All done! Sorry it took so long to get out to you. May your experience be more pleasant by learning from my mistakes. :)

[A legacy blog post]