Why is Good Science Writing Hard?

by F.

I love good science writing but it seems fairly rare. Why? Some guesses.

The Writers

There seem to be two types here: the scientist-writer and the journalist-writer. Having been reading various scientists’ blogs, I can see the contempt in which “they” (scientists) seem to (often) hold science writers. Here’s an example from Peter Woit, a mathematician at Columbia:

Yesterday I also saw two reviews that I don’t think much of. The first is Gregg Easterbrook’s piece at Slate, The Trouble With String Theory. It’s a very enthusiastic review of Smolin’s book, and when I started reading it my initial reaction was positive, although it did seem a bit over the top. As I read on, besides wondering “Hey, is he going to mention my book too?”, I started to remember who Easterbrook is, and how stupid some of his previous writings on physics were. By the end of it, I was very glad Easterbrook had left me out of it. One sometimes depressing aspect of being on this side of the string theory controversy is seeing who some of one’s allies are.

Easterbrook is best known as a sports writer writing about the NFL, but for some reason various prominent publications feature his writing on other topics. The biggest mystery of all is why places like Slate and the New Republic have him writing about science, a topic he seems to know nothing about, and be actively hostile to.

Representative? I don’t know. But there is something here, as reading lots of science writing will tell you. Journalists need to make stories interesting, and this almost always involves distortion. There’s no other choice. The trick, I think, is to do the least damage to the actually theory under discussion. Even The Economist gets a drubbing now and then by various scientists for its science reporting, and I would say The Economist is better that the vast majority of science writing out there.

Scientists are the best at writing about their work, but many are not very good writers. This seems to be changing, though, as can be seen from the many excellent scientist blogs. Many scientists are fantastic however, such as E.O. Wilson. Journalists tend to oversimplify the science to fit the format of journalism. Often they don’t have the conceptual background to fully compress the research into article-form without huge distortions.

The Subject Matter

Readers want—we all want—stories that fit the person/action schema: “Frank petted the puppy,” “Lolita published a paper on harmonic theory,” and so on. That’s the easiest way to think: about agents (people) doing stuff. Noun + Verb. Any writing book will tell you this, and this is what “eliminating passive voice” is really about. Passive voice covers over the agent, the do-er: “Mistakes were made.” By whom? That’s exactly the point: the author doesn’t want to say.

So you got to get your agents in there. A lot of science stories start with an anecdote: the scientist is doing such-and-such, or the patient did such-and-such, or the three toed sloth did such-and-such. Our Theory of Mind is so powerful, we want to use it wherever we can. The actors in science stories are, then, scientists, animals, patients, historical figures (“Descarte thought X, but he was wrong, as new research shows”). They can also be inanimate things, and this is where it gets tricky.

I was reading this morning an article in The Economist on microbiology. Here’s a paragraph from it, with the agents highlighted:

Under a microscope, bacteria appear to reproduce by dividing exactly down the middle. Instead of growing, giving rise to offspring and fading away to death, the mother cell actually becomes its two daughter cells. The assumption was that these daughters are as close to identical as random variation permits, and that neither is subject to the sort of ageing imposed by the disposable soma theory. But that assumption depended on another: that the components of the mother cell are dealt out equally to the two daughters. Examination of bacterial division using modern techniques has shown that this is not always so. Sometimes one daughter gets preferential treatment, by receiving newly synthesised components instead of old, battered and repaired ones. The more aged a bacterium’s innards are, the more mortal it appears. It is slower to metabolise and likelier to die before it has reproduced.

The bacteria are doing shit. There are a fair number of “agentless” sentences here (“The assumption was”—Who assumed it?) but generally, this is quite easy to read. More

Dr Watve asked himself why this should be, and constructed a theoretical world inside a computer to investigate the matter. In this world, a bacterium can “choose” a mortal strategy, dumping all its old components into one daughter with limited prospects, and betting most of its future—and all of its new bits—on the other. Alternatively, it can divide itself symmetrically. The latter choice brings, if not eternal youth, then an equal pressure to avoid ageing, since neither daughter is more prone to death than the mother had been. But these daughters grow more slowly than the favoured daughters of unequal divisions, because their resources have to be divided between maintenance and reproduction.

Here, the scientist is the actor and the bacteria are acting, too. They are “choosing”—appropriately quoted—and dividing and growing. They are doing stuff. More:

When Dr Watve ran the model, he discovered that the main determinant of whether symmetrical or asymmetrical division was favoured was the amount of food around. In impoverished environments (in the real world, that might include lakes and oceans), the slower-growing daughters of symmetrical divisions had the upper hand because they used what little resources were available more efficiently. That was because the runts tended to die before they could reproduce, thus wasting the food that they had already eaten. In richer places, fewer runts died, and the daughters with shiny, new bits grew and divided very rapidly indeed. And that fits with the finding that bacteria inside the nutrient-rich human gut grow rapidly and in an asymmetrical manner.

And so on.

I’m surprised often at the topics that science writers take up, like evolution. I suppose they feel they can take part in the culture wars that way. In addition, evolution is easy compared to, say, covering physics or math or chemistry. Try to make chemistry exciting. Where are the agents? You don’t even have bacteria! Bacteria are living, “like us,” and we can easily anthropomorphize them (those little cuties). Chemicals? Harder. Astrophysics gets a lot of press, it seems to me, partly because it is so incomprehensible to the layman that the science writers use extended, often mythic metaphors. I hate this, personally. I would rather just accept that, say, I’ll never understand string theory because I (like 99% of us) will never be able to go over the equations and get anything from them.

The last resort to getting an agent into the writing, it seems to be, is dropping into “writer-as-professor” made. Here, you (the writer) are so desperate to get an agent into the story that you include yourself. Not that you use “I,” but you start to say things like, “Now, the thigh bone is associated with the knee bone, which is contiguous to the foozdoozle muscle, and…” Better would be to make the bones agents: “The thigh bone connects to the knee bone.” I think this writer-as-professor made is sort of like a flashback for screenwriters: it’s the first thing you reach for when there is a problem, but it’s the last thing you should consider. It can work, but you have to make sure you’re not just being lazy.

Why Do I Care?

Because I’m compressing some neurology research into an article format and it’s fucking hard! In science, you can be wrong, and mixing up the prefrontal cortex with the amygdala can make one look, to use the scientific nomenclature, like a dumbfuck. My piece starts out well. There’s a case study about a patient. Really compelling. I keep reading my draft. Interesting. Interesting. Interesting. And then huh? I drop the reader off a cliff into the land of “associations between dispositional states of the brain and bodily states.” Ouch. Where are the agents? Associations? Associations are just not interesting—is there causality? What ties the associated entities together? And so on.

Back to work.