Many folks have noted this advice from well-known novelist Cormac McCarthy on writing scientific papers, usually with a sense of disbelief that he has anything to do with scientific papers at all. I felt the same way, but he’s been associated with the Santa Fe Institute (physics and math) for some years now, and has been providing editing help. (Edit: fixed the first line of this post on writing because it went out making no sense whatsoever. Argh.)
It’s sound stuff. His general rules (which include: remove what you can, pick a theme or message and make sure that everything is devoted to getting it across, one message per paragraph, use concrete language and images, and more) are all worthwhile, and similar to writing advice from other well-known authors (from Mark Twain’s “eschew obfuscation” to Elmore Leonard’s “If it sounds like writing, I rewrite it”) McCarthy’s last suggestion is a key one: read your work aloud. I did that, for example, with every entry in The Chemistry Book, and found any number of phrases that needed to be reworked.
I wonder, though, if the read-it-aloud advice works best if you’re already a fairly smooth speaker (or at least have an ear that can tell when you’re listening to one!) Reading your own prose out haltingly or uncertainly probably won’t work; you have to stress-test it by going right through it at realistic speed. That’s when the potholes will reveal themselves. In the same way that it can be hard to write fluent prose, it can be hard to read it out fluently, and often the same people who are having trouble with the former are having trouble with the latter, too. That’s not always the case, though. There are actors and politicians who can put someone else’s speech over with great effect, but could never have written anything like it themselves. And at the other end of the spectrum, you have people like Nabokov, who said “I think like a genius, I write like a prize-winning author, but I speak like a small child“.
Applying general writing advice to a scientific paper is somewhat tricky, though, since it is a rather specialized form. I find writing a scientific manuscript much slower going than putting almost anything else to paper; it’s like high-stepping through a swamp. But that probably makes such advice even more valuable, because it’s too easy otherwise to make your manuscript so dense as to be unreadably unpleasant (and thus incapable of getting its points across clearly). Turning research results into clear prose can be hard, even for a very good writer, which is even more reason to put the effort into doing it.
I once worked at a company where we had to write up detailed multipage research reports every six months. I would just sit down and bang the thing out; doing the structures and schemes probably took longer than writing about the work. But I had a new colleague whose first report got sent back for reworking because it was just too long to read. Not that anyone had necessarily read it – I was never convinced that anyone really did that. Had I only known a few years later that my last report for that company was actually my last, I would have done what I’d always wanted to do and worked in the line “. . .and I will pay $5 to the first person who tells me that they have read this sentence“. Sadly, I missed my chance.
My colleague came to me for advice on how to shorten the thing; they had tried and gotten nowhere. If anything, I think the report had gotten a bit longer on revision. I had a look, and I swear it started out something like “My lab has been working on structural variations of the lead series involving numerous stereoelectronic diverse substitutions in order to alter the pharmacokinetic behavior in various assays so as to modify the half-life of the scaffold which will lead to a more favorable dosing profile by addressing different metabolic liabilities through blocking sites of oxidation with a variety of substituents with electron-withdrawing effects and/or steric hindrance which are expected to slow the rate of the various enzymes responsible for the clearance of the lead molecule, with is believed to take place through bioconjugation of the. . .” and it went on from there. On and on, down the page, an ever-flowing river wandering through a trackless landscape.
I read enough of this to be sure that it went on in the same human-wave-attack manner (it did), and asked “So what are you trying to say? What have you guys been doing?” “Oh, we’ve been putting groups on the left-hand ring next to the phenol to slow down the clearance” “OK,” I said, “Say that instead of your entire first page”. I believe the response to this was something like “Well, I can’t just say that!”, to which I asked “Why not?” Somehow it just didn’t sound right to them – that wasn’t formal enough or detailed enough, and it didn’t show that every single box had been checked and every single possibility had been gone over. But when you did that, of course, you ended up with the doorstopper, the embolism of prose that had been sent back as too much even for middle-managers to work through.
In the end, the report did get hacked down to an acceptable length somehow. But I doubt if reading Cormac McCarthy’s rules would have kept it from being produced in the first place!