Over at Xykademiqz, a couple of weeks ago, there was a very nice post about the struggle to get students to write. “Very nice” here means that it’s a good description of the problem, not that I’m glad anybody else has to deal with this.
I don’t face quite the same thing– my students generally aren’t writing research papers for journals– but getting senior theses written is often a bit of a chore. And I admit, I’m really bad about pushing students to do this in a systematic way. The problem is that, as I know from personal experience back when I was an undergrad and saber-tooth tigers hunted mastodons in LA, undergrads have a more limited amount of time to sink into thesis research than grad students do, and when writing begins in earnest, work in the lab tends to grind to a halt. So it’s really tempting to just say “Look, the end product will be much better with more data, so don’t worry about draft sections, go work in the lab.
The problem is, of course, the same problem that always trips faculty up in dealing with students, namely that our students aren’t us. That is, the people who go on to become faculty are a small subset of students, and not really a representative sample even within a particular major.
In this specific case, the issue is that writing has always come fairly easily for me, so putting off the writing part in favor of doing more work in the lab didn’t create a big problem. Yeah, it reduced the time I had available to churn out 75-odd pages of thesis, but then, I was probably going to fritter away a lot of that time playing Lunatic Fringe or Spaceward Ho! on the lab computer. Deadlines, like impending death by hanging, focus the mind.
For students who are less facile with the generation of reams of bullshit, though, putting off the writing to work in the lab triggers a kind of nonlinear growth crisis. Time spent getting more data increases the amount of stuff that has to be written up, and also shortens the time available, making the task look much more daunting. Which leads to putting it off more, and then the time gets even shorter, and the problem looks even worse, and then it’s almost the end of the year and there’s a blank page and AAAAAAAHHHHHHHHHHHHHHHHHHH!!!!!!!!!!….
The comment thread to that post is full of good suggestions of ways to kick-start writing, and I really need to be better about systematically implementing them. The one that I threw out that deserves to be expanded on a bit, though, is the comment I made about not starting at the beginning. This doesn’t always work, but for most of my academic papers, it’s ended up being the right approach for me, so I want to put in a plug for it.
The problem is that the very worst part of writing is the bit where you open up a file with a blank page, and need to fill it with words. (Using something like LaTeX to write, where you see fifty lines of header garbage at the top of the page, or beginning with somebody else’s paper and overwriting their text can help a little with this, but only a little.) Compounding that is the fact that one of the most labor-intensive bits to write is the Introduction of a research paper (which may or not be labeled as such, but is an important element of the structure), where you need to put in background information and literature citations and all that kind of thing. It’s really easy to go down the rabbit hole of chasing down citation chains or noodling around endlessly in search of the perfect opening sentence (“Let’s see, ‘Webster’s dictionary defines physics as…’ No, that’s crap. ‘From the very earliest days of human civilization…’ No…”).
On the other hand, if you’re doing the research, you spend the vast majority of your time immersed in the technical details of what you’re doing and how to make it work. Which means that’s stuff you know right away, without needing to look up references, etc. So, that’s actually the easiest place to start– just skip the intro, and type out a description of what you did and how. Describe the procedure, present the data, and come back to the introductory stuff later. The earliest drafts of my technical papers almost always start “Introductory bullshit goes here, blah, blah, blah.” Then a couple of blank lines, and the start of the procedure section.
This is a hard sell for a lot of students, who are used to writing papers by starting at the very beginning and typing until they reach the end. And it makes for some editing work later on, as you need to make the introduction fit with later bits when you get around to it, which sometimes means reordering references and that sort of stuff. But all that is editing, which is fiddly and annoying, but much easier than staring at a blank page waiting for it to contain words. It also tends to provide a better focus when you come back to the introduction– you already know what bits of the incredibly broad field of which your new work is a part will be important later in the paper, so you don’t have to make decisions about what to exclude.
So that’s my advice: don’t worry about the literature survey and putting things in context at the start of the process. Start by writing the bits that you can write without going to the library or a web browser to look up references, and come back to the survey at the end. At that point, you’ll have some momentum, and it will be easier to write the introduction.
This varies a bit from one project to another, of course, but I think the rule of starting with something you don’t need to look stuff up for is good. For my two dog-physics books, the Introduction was very close to the last thing I wrote– I think it was the very last thing for the relativity book, but I could be misremembering. The book-in-progress, on the other hand, started with the introductory section, because that bit is mostly a personal thesis statement, while the main chapters turn on historical anecdotes that needed a good bit of research. And within those chapters, the ones I wrote first were ones based in my own areas of physics, where I could pretty much just sit down and type without needed a lot of reference books and websites.
For some people, outlining can be a good way around this issue, but it’s never worked very well for me. Outlining the Introduction requires the same daunting decisions about what to include, what to leave out, how to frame things, etc., as just writing the damn thing in the first place. So it ends up leading to the same sort of procrastinatory noodling.
Anyway, that’s my tiny contribution to the vast corpus of advice on how to write. Now if I could just remember to start forcing my students to do it earlier…
At least one commenter on the xyacademiqz post suggested starting with the figures and captions, then building the story around them. In my grad student days (when we used viewgraphs, a scarce resource), that was exactly how I planned my conference presentations: with talks, I would make a list of what each viewgraph would be (making sure that the total number was appropriate for the allotted time) before I actually made the first one, while with posters I would sketch the poster layout, specifying the approximate location of each figure and text blurb, before creating any piece of the poster. With papers (and also with PowerPoint, which allows one to add slides at zero apparent cost) I am not always so strict about this rule, but it often helps to know in advance that the key figure will show the result that I want the audience to get, and I therefore need to make sure I build up to that key figure. Scriptwriting for movies or TV often works this way, too (they call it storyboarding), so this method has applications outside of science.
It always makes sense to start with the middle, then work to a conclusion. How else are you going to know what to put in the introduction? When I worked as a consultant, we’d write our report, then the introduction, then finally the executive summary which was more or less an extended one page abstract.
Eric Lund’s comment suggests – through my misreading of “viewgraph” as “graph” – that it might make sense to start with the graphs of the lab results. After all, each graph should tell a story, and you have to make some graphs so you can figure out what is going on. Then, you can just describe how you got the numbers for your graph, what they mean and so on.