On bad first drafts

From I Can Haz Cheezburger.

My blogging mojo has been channeled almost entirely towards a book project I've undertaken with Julienne Rutherford of UIC and Katie Hinde of UCLA (though shortly to be of Harvard). The book is called Building Babies: Primate Development in Proximate and Ultimate Perspective and it will be published by Springer in 2012. Each co-editor has a chapter in there, and then we have a number of other rather fancy-pants contributors as well.

The first drafts of the chapters were due yesterday. I did not submit my chapter (er, to myself). I'm running about a week late. I thought I would come clean with this, because there are a number of elements of the writing process that I think remain obscure for students and other junior scholars. And after I share a few thoughts about academic writing, I thought I would show you some of the draft I'm working on.

First drafts suck

They really, really do. If you think your first draft is amazing, give it to someone else, and that someone else can't be a pet, spouse or parent. First drafts suck because we write the most obvious things in them, the most vague. First drafts don't have enough context. First drafts are where you use cliches because you haven't figured out how to say what you're saying in a sophisticated way. They are often under-cited. They are out of order. And, they aren't that compelling.

This is why so much student writing is bad -- but it's not their fault. Close together deadlines, ones that align with other projects, and little teaching of time management means most students start writing projects just before they are due. So they essentially submit first drafts of papers, with a little copyediting if you're lucky. Plus, somehow a lot of students have picked up this idea that first drafts are better or more authentic than revisions. This is patently false. They are simply the place our favorite worst stuff goes to die (this is why revision is so often called killing our darlings, to use a term from scio11, though its origin is much older).

But everyone has bad first drafts, so it is absolutely useless to feel bad about them. Give them to your advisor or your colleague if they have said they will read a first draft (otherwise, revise it after consulting with someone else first). They write bad first drafts too. You have to write a first draft in order to get to the revision, and to me, this was a liberating realization. Get it all out now! Don't worry about using the right word! Just get the words on the page, get about the right content in about the right order, and if something is repetitive, just leave it for now. Because after a little breather away from it, or a look from a trusted colleague or advisor, you will hack it up and remake it into something far better.

Revising only sucks sometimes

Revising sucks when you get your first comments back from a colleague, because it is terrifying to share that vulnerable, bad first draft with another person (ever had that moment after you print it out or hit send when you realize your prized metaphor was a trembling nod to your failed attempt as a fiction writer?). It sucks at those moments when you feel at cross-purposes with the thesis of your paper. And it's frustrating, also, that revising is the most important yet under-taught skill in academic writing.

But here's the thing. Revising can be glorious. If you abandon any sense that you own your words, and remember only to own your mind, it allows you to be merciless in cutting out all the badness of that first draft: the cliches, the vague repetition, the jargon. If you return again and again to your outline, or abstract, or data, or whatever materials you keep to help you remember what the paper is about, you will start to see the right shape of the piece. And then you can also build in the context.

The best moments of revision are when you remember why you were writing the piece in the first place. Do you want to produce a fundamental review that will be useful to other practitioners in your field? Do you have an amazing piece of data to share? A well-grounded hypothesis that you want to articulate? What was surprising or compelling about that work when you first set fingers to keyboard?

One last thing I'll say about revising is that owning your mind is not the same as owning your ideas. You need to be willing to let go of being right, and you need to be willing to change if the evidence is against you. Accepting reality and working with it in an interesting way is the mark of a good scientist, and a good revision.

My first drafts suck

The title of my chapter is: "Inflammatory factors that produce variation in ovarian and endometrial functioning" (eventually, I think, I will need to change the title to better reflect the manuscript). I thought this would be an easy piece for me, since I have been doing a lot of work on C-reactive protein, a biomarker for systemic inflammation, and I have been studying the endometrium and ovaries for many years.

I was wrong. Oh, so wrong.

A few quick searches pulled up an embarrassingly large number of citations for chemokines and cytokines, for toll-like receptors, natural killer cells, and other immunological terms I barely remembered from high school and college. So I re-drafted my outline, set aside a lot of time for reading (as in, several days straight), and then finally set to work.

The problem with the literature on this topic is that it is wholly mechanistic. I can now tell you what interleukins are expressed in the periovulatory phase versus the implantation window, or which ones are suppressed or overexpressed for certain pathologies, but I can't tell you what that means in a broader sense, or what produces variation in any of these immunological factors in a systemic way that might impact local inflammation in the female reproductive system.

Here is my section on normal endometrial functioning (alas, given the literature, the section on pathology in the endometrium is far, far longer). First draft ahead! Remember, I am sharing this embarrassingly bad prose for the good of SCIENCE.

The endometrium is composed of the functionalis and basalis layers; the functionalis comprises two thirds of the endometrium and is the part that proliferates and sheds each reproductive cycle. The basalis is adjacent to the myometrium, and is the place from which the endometrium regenerates after menses. The proliferative (also known as follicular) phase is when estradiol promotes proliferation of endometrial tissue, where the secretory (also known as luteal) phase is characterized by progesterone control of decidualization and menstruation. The endometrium typically proliferates with narrow, straight glands and a thin surface epithelium, and angiongenesis continues as ovulation nears (King and Critchley 2010). After ovulation and during the secretory phase, the endometrium differentiates: endometrial glands become increasingly secretory, and by the late secretory phase spiral arterioles form. If implantation does not occur, the corpus luteum degrades, progesterone declines, and this triggers a cascade of events to produce menstruation.

Menstruation is a key inflammatory process of the endometrium. Menstruation is when the functionalis are shed at the end of the human reproductive cycle. The basalis regenerates over the course of the next cycle. The demise of the corpus luteum and the associated withdrawal of progesterone precipitate inflammatory mediators that cause tissue degradation. For instance, progesterone inhibits nuclear factor κ B (NF-κB), which increases the expression of inflammatory cytokines like IL-1 and IL-6 (Maybin et al. 2011). The withdrawal of progesterone is also associated with an increase in endometrial leukocytes and IL-8, which regulate the repair process (Maybin et al. 2011). At this time other inflammatory factors promote MMP production to break down endometrial tissue (Maybin et al. 2011). Further, it is thought that progesterone withdrawal, not an increase in estradiol concentrations, leads to the repair of the endometrium so that it can resume activity for the next cycle (Maybin et al. 2011). Thus, variation in progesterone concentrations may lead to variation in inflammatory activity, degradation, repair and cycling in the endometrium.

First question: why should I care about any of the above? So what if any of this happens? Then, you might not know this, but I do: the only two citations in these two paragraphs are both review papers, and one of the authors overlap between them. Therefore, it's quite under-cited. To be fair, in this section it is less important that I demonstrate the depth of the literature, but a review that only cites two other reviews isn't doing its job.

Do I inspire excitement in my field? No. Do I provide an appropriate context for this material in order to situate the reader? Not so much. Right now, these two paragraphs contain the exact information I wanted them to contain, based on what was in my outline. That is, I've described the basic functioning of the endometrium, and menstruation. It's flat because that's all that I did.

My job in this chapter is to take this vast reproductive immunological literature, pair it with what little we have in anthropology and ecology that helps us understand the way genes and environment might produce this variation, and then describe the necessary context in future work to understand these mechanisms. In some places, a lack of context may help me make my case, because it will demonstrate why anthropologists need to be in the field. But if my whole manuscript looks like the two paragraphs above, it will be an unreadable yawnfest that doesn't contribute a thing to anthropology.

So, I guess I would expand the "kill your darlings" advice. First, accept your darlings. Accept that you have them like everyone else, and that darlings aren't just turns of phrase but entire ideas, hypotheses, fields of thought. Then, once you have accepted that your darlings make you just like every other academic writer out there, from the middle schooler to the full professor, kill them. With fire. Finally, make sure you provide what is left with context or else there is no reason to read what you wrote.

And now, I have been sufficiently inspired to go finish my bad first draft.

References

King, A., & Critchley, H. (2010). Oestrogen and progesterone regulation of inflammatory processes in the human endometrium The Journal of Steroid Biochemistry and Molecular Biology, 120 (2-3), 116-126 DOI: 10.1016/j.jsbmb.2010.01.003

Maybin JA, Critchley HO, & Jabbour HN (2011). Inflammatory pathways in endometrial disorders. Molecular and cellular endocrinology, 335 (1), 42-51 PMID: 20723578