Writing papers - in lieu of a guide

This is a guest post by Neil Levy, Senior research fellow at the University of Oxford

I’ve published a lot of articles – more than 200. So it seems like I should have some tips worth sharing on how to write them. I’m not sure I do. I thought of calling this post How to write a philosophy article, but I’m not sure I know how. I also thought of calling it How I write a philosophy article, but I’m not even sure I know how I do it. What follows is a description of how I tend to generate ideas and turn them into papers. Perhaps something in what I laughingly call my process will help others. Or perhaps the lesson is that you have to find your own.

Most people (I gather) generate the argument, more or less, before they sit down to write. Writing is a craft for them. The thinking happens in conversation with others, in the shower, in teaching, while running, or whatever. This sometimes happens for me too (though usually when reading). That’s nice when that happens. But it doesn’t happen all that often for me. I don’t tend to sit down with a clear idea of what I want to say.

Rather than a craft for me, writing is thinking. When I sit down, any idea I have is incredibly vague and very fragile. I really don’t know whether it will amount to anything. I try to sneak up on it, rather than express it. I often begin by describing the problem I’m concerned with, rather than jumping straight into it. I will often write (what I hope will become) a first section, describing the terrain of the debate as it stands. Only then do I really try to develop my idea. I try things out, on paper (well, on the screen). I write the ideas as they come, as fully and carefully as possible.

The advantage of doing things this way is that by the time I get to trying out the ideas, I have a fairly good idea of the lay of the land in the debate. The disadvantage, of course, is that doing your thinking on paper is labor-intensive. When I try to make my vague thought a little clearer, I sometimes find that it melts away. Vague thoughts are often confused thoughts. Because I write on a lot of topics, at the germ of an idea stage I often don’t the debate well enough to know if the idea is novel, or whether there are obvious objections to it. The result is a lot of abandoned papers. I have lots of paper fragments floating around on hard drive. Sometimes, they’re not unsalvageable. I abandon the paper because the idea evaporates, but I come back to it months later (maybe after reading something that suggests a different way of approaching it). Sometimes, they’re complete dead ends.

Maybe this method of working requires more time than most people have, because it results in lots of half- or quarter-assed papers that get abandoned. So maybe it wouldn’t transfer well to the lives of most people. But it works for me. One reason it works for me is that it helps with one of the hardest parts of writing (for me, and other people I’ve spoken to). Writing can be fun, once you get going. But starting is hard. Writing the way I do helps with that. It’s much less forbidding to start by setting out the debate as it stands rather than go straight to the idea. You don’t have to go out on a limb, yet. By the time you get around to your idea, you’re in the flow and you can forget yourself enough to take risks. When I return to the same paper the next day, the introductory section done, I ease in by rereading the earlier paragraphs, editing as I go. All going well, that transitions to real writing more or less automatically.

This ‘process’ may be completely idiosyncratic, and completely useless for most people. Maybe the only real lesson here is that we’re idiosyncratic beings, and what works for one person won’t work for another. Maybe we all need to generate our own way of working.

That said, there are things that are probably common to most people. Because they’re common, they’re probably already familiar to you. At the risk of repetition, here some of the things I do which I guess that lots of people also do.

Write every day. I am vaguely dissatisfied unless I manage to write 750 words. On a good day, I write three times that. Of course there are lots of bad days too.

Write at the same time every day, if you can. This helps with the starting problem. I tend to go to a café, answer a few emails and then write for an hour or two. If it’s routine, it’s easier to start.

Read widely. Ideas come from weird places. My most recent acceptance is a paper on social epistemology (with Mark Alfano) which is inspired by work in cultural evolution. Some of my work on free will arose by thinking about locational externalism in the philosophy of mind. Even if reading away from your AOC doesn’t give rise to cool new ideas (or, in my cases, fragmentary thoughts that might become ideas), sometimes you need the reset of time away from the AOC. Note, however, that there are traps in reading away from one’s comfort zone. In the bad old days (not too long ago at all), philosophers would all too often read about a cool study in neuroscience or psychology, and write a paper centred on it. Motivational externalism refuted! Virtue ethics eviscerated! I was guilty of this kind of thing myself. We’ve got more sophisticated now, and this kind of thing won’t be accepted today. Now we’ll need to see evidence that the empirical evidence is reliable and that it’s being interpreted appropriately. If you have do generate a cool idea through reading well away from your comfort zone, you need to take the time to develop some real competence in the area before you submit the paper. Alternatively – maybe preferably – enlist a co-author with real competence.

One final thing: while for me writing a paper is usually a pretty solitary affair – I don’t generate ideas in conversation much – my papers really benefit from input from others, and for me this is best done by presenting them. Obviously, y’all know that getting feedback is important, but presenting papers has extra benefits over and above sending them out. I talk my papers, with or without slides. It’s in preparing to talk the paper that it moves from the page into my head: I have to really get to know it to talk through it. I have to get its structure right. This enables me to see how it works, and how it doesn’t work. I need the stimulus of knowing I actually have to present the paper to motivate me to do this (it’s hard mental work, running through the paper in one’s head repeatedly). So presenting the paper has benefits apart from feedback.

I thought I had more to say. Turns out, I don’t. That, too, is something that can happen when you sit down to write.

Comments

[pendaran]

I think the advice to read widely is good, but there is a provisory: don't over read; at some point you need to start writing or you'll generate nothing. Also, writing will help you understand what you're reading, so sometimes it's best to write as you read. Don't worry about having to delete material or throw material out.

I agree that if you're trying to be productive, you should write every day. When I was working full time at philosophy I'd spend 6 hours a day reading and writing and another 4-5 thinking about the topics I was working on (including when I should have been sleeping). That's what it took to crank out 3 papers a year or so.

Having a schedule isn't a bad thing for sure, but for me sometimes I just have to write when I have something to say. I've worked late at night after getting out of bed, because I couldn't stop thinking about something or another. Maybe people with more self control can put it off until the morning, but perhaps it's also best to write when your brain wants to write and not force a schedule if you don't have to. I realize this might work better for postdocs and grad students than professors who have other responsibilities besides research, but remember coffee is your friend! I can't write without caffeine.

[Marcus Arvan]

Hi Neil: Thanks for sharing this. My process is remarkably similar, and I think your tips are spot on!

Like you, I normally begin with only a vague idea of what I want to argue, and then figure out the details as I go. And, like you, this process leads to a lot of "overproduction." I probably write three times as many papers as I end up publishing, deciding only after writing them (either in whole or in part) which ones are working out and which ones don't. I don't consider this wasted time, as my sense is that every paper is good practice for the future--practice writing, practice thinking, practice editing, and so on. Finally, like you, I often return to papers later (sometimes years later) if and when I figure out how to fix what I took to be a fatal problem.

That being said, I think you and I may do a few things a little differently. The first thing I do when I have an idea for an argument is to quickly survey the literature (mainly by reading abstracts) to see if anyone has defended the same thesis or given a similar argument before. This not only helps me see whether the argument is original; it also gives me a very rough lay of the land in the literature (without spending tons of time reading full articles). If the argument I've come up with hasn't been defended, then I start writing. On the other hand, if something like it has already been defended, then I try to figure out whether I have anything original to contribute to whatever has been written about it. If I think I do, then (once again) I sit down and start writing. Finally, if not, then I give up on the project right then and there and move onto something else. This is, I think, one benefit of the approach of beginning with a very quick literature search (rather than reading a ton): oftentimes, it only takes a brief literature search to figure out whether to proceed at all!

What I then typically do (and this does seem a bit different than what you do) is "jump right in" and draft the paper in full, focusing primarily on the positive argument. Whereas you say your typical process is to write a detailed first section of the paper (where you lay out the lay of the land in the literature), I'll typically just sketch that part very quickly, given what I've gathered from my quick cursory literature search. Then I'll move right onto drafting the rest of the paper, working out the argument as I do. Then, only after quickly drafting the whole thing--and after I figure out whether I think the argument works--will I go back and do a "deep dive" into the literature, polishing up the early parts of the paper that situate it in the rest of the literature.

Again, I generally find it helpful to leave this part for last, as my sense is that it can save a lot of time (viz. what's the point of doing a really detailed literature survey before I've written up the rest of the paper and have some idea of whether my argument works?).

Other than that, though, it sounds like we do things very similarly. It's nice to hear how other people work, as I always wondered whether other people work in this general way!

[Michel]

That's pretty much how I write, too. I have loads of (vague) ideas, but the arguments only really come out in the editing process. Editing is *a lot* easier than writing from scratch, and research is a lot easier to perform once I've got a solid direction, and know what kinds of things to look for.

I always try to write up a 3k-ish word conference paper first, and then I flesh it out from there.

I'd also like to chime your point about presenting papers: the main benefit I get from conferencing (apart from socializing!) is the challenge of distilling my point into a twenty-minute presentation. When I get back to working on the paper afterwards, everything is *much* clearer.

[Amanda]

Marcus, how do you go about "surveying the literature?" I have a hard time figuring out how I am supposed to find everything that's ever been written on a topic, especially when the topic is a kind of niche area that doesn't have an obviously demarcated literature.

[Neil Levy]

I completely agree that it is dangerous to allow reading to swamp writing, Pendaran. I learned this the hard way. I sat down to educate myself by reading classics of philosophy: the first Critique, Being and Time, and so on. I don't think I retain anything from them, because I didn't put them to use. They're just gone. So reading too much doesn't just delay writing; it is pointless. You need to write while it is still fresh. That's true for me, with my bad memory, anyway.

[Michael]

I've had a similar experience to Neil. In the two or three years I spent doing my dissertation, there were many things I read during my initial flurry of reading which I forgot. Years later, as I reviewed those references, I thought "wow, these are important quotes/ideas/references relevant to what I did/am doing now". I've now learned that I can only hold about a dozen or so works in my head at once (and really only 2-3 without the help of substantial notes). So I write with about 6-12 references in mind. I expand the citations by either relying on citations from my older papers to fill in gaps, or by doing new dives through the literature on subsequent drafts.

There may be a time and place for reading a flurry of things, e.g. when first learning a literature, but at least for me writing requires focus on a few sources.

[Marcus Arvan]

Hi Amanda: I'm not sure about niche topics without a clearly demarcated literature.

But in general I survey the literature by looking up Stanford Encyclopedia entries, giving them a quick read, and then by doing philpapers searches of key words (e.g. 'nonideal theory'). Then I go down the list of papers, reading abstracts quickly and jotting down some brief notes on each article's abstract in a Word file. At the same time, I download full papers and store them in a labeled folder for reading later.

Then, if there are particular papers that seem relevant to what I'm arguing (i.e. they've pursued a similar line of thought), I give them a quick read to see whether the argument I'm toying with is still original enough to develop, etc.

This ordinarily doesn't take very long. I can usually tell pretty quickly whether an argument I've come up with has already been given, and whether the literature has "room" open for what I want to argue. Then I'll usually draft up the paper based on what I know of the literature, or if there are particular papers I think I need to read in full before starting, I'll give them a full read. Then, after doing this and drafting the paper, I'll usually go back and read other papers in full to fill in my "lit review" in the paper where I show how the paper's argument fits into the existing literature.

[Tom Cochrane]

Hi Neil
I think there's more to your process. You're probably one of the best-published philosophers alive right now! I have some questions.
1. What's your journal rejection rate like?
2. Do you ever write with a particular publication venue in mind? Are do you just start at the top ranked and work your way down?
3. When you review your papers, what sort of virtues are you checking for?
4. Like Marvan above, I don't commit to writing a paper unless I think it's got something original to say. Is originality a big factor for you? What's the priority of this as compared to say, scholarship (being situated in a certain debate, referencing the current literature)?
5. Are you following any structural templates (e.g. problem-solution-objections/relies) or generally following the material where it leads?
6. Are you highly perfectionist about your material (to the extent say, of radically working an essay over a long period) or will you submit relatively quickly?
7. Where work is rejected, do you send it right out again or make sure all referee problems are accommodated?

[Amanda]

Thanks Marcus - that's helpful!

[Neil Levy]

Hi Tom,

Thanks for the questions. I'll have a go.

1. Rejection rates. I can tell you for the recent past. From the start of 2016 to the end of 2018, I submitted 21 distinct papers to journals (with rejections, there are of course more submissions than distinct papers). 18 of them were accepted (3 retired permanently). The rejection rate per submission (rather than paper) was just below 55%
2.I start thinking about the journal midway through the process (unless I'm writing a reply). Because the Australian research assessment exercise really emphasises journal prestige, I aim high. I work in different fields, with distinct sets of journals. The journals appropriate for applied ethics has only a little overlap with those for general philosophy and those for cognitive science. So I have different sets of journals in mind for different papers. I don't literally start at the top and work done. Sometimes I go in the opposite direction. Journal submission is partly a matter of luck, so I don't take a knock back from Nous (say) to show the paper isn't good enough for the top 5 (it's evidence, but reasonably weak evidence).
I'll take 3 and 4 together. Originality is by far the virtue I most emphasise. Partly this is a matter of my strengths and weaknesses. I'm not a very precise thinker (in many ways, I'm an atypical philosopher). I'm not good at making fine distinctions. What I'm better at is generating ideas. I'm in no way downplaying the importance of fine distinctions. My weaknesses here mean that others, with different skills, can identify problems with my work that I don't see without their help. Sometimes those weaknesses are fatal. The emphasis on originality has increased for me, though I've alway thought it was important. I had an epiphany of sorts, writing a reply to a reply to me. I realised I was bored! I want to do new things: both new debates (for me) and new things to say on those debates. That's what I aim for - I certainly don't always achieve it.
5. I guess I have a template. Not explicitly, but I suppose I have an idea of what a paper should like (that is, what reviewers expect). Intro: brief introduction of the problem and my solution. Section 1. The problem in more detail; in this section much of the scholarship occurs (I identify the problem with the help of others). Section 2. Why other responses (already in the literature or - if the problem is novel - extrapolated from the literature) fail. Section 3. My solution. Section 4. Objections and replies. Conclusion. That's a rough template. It may be that the problem has parts that lend to extra sections.
6. Not perfectionist at all. It's not unusual for me to send a paper out a fortnight after I had the original idea. I'll take longer if there are opportunities for audience feedback coming up. I'm impatient: if I don't have an opportunity to present the paper soon, I'll send it out.
7. As I said, I get more rejections than acceptances per submission and in most cases the paper is sent out again. Some of those rejections - more than half - are desk rejections so there's no reason for revision. Sometimes the reports are useful and sometimes they're not. I know from my own experience and observing others that initial responses to reviews are untrustworthy. I put it aside for a couple of days if there are reports and then come back to them. If I'm right in a common first reaction - they didn't understand the paper - I try to make it clearer. If they have objections, I take them seriously. But sometimes I remain convinced that there's no reason to respond. Quite commonly I think the report is reasonable *and* doesn't warrant a response. A reviewer can reasonably want to see a paper developed in a certain way but I can reasonably not want to go that way. I do want to say that I have had many R&Rs with really great reports, that led to big improvements in the paper. Of course I have my share of referee horror stories too.

[Tom Cochrane]

Thanks very much for your reply Neil.

55% is an extremely good acceptance rate (assuming that means about 2 rejections per 1 acceptance), particularly if you aren't being highly perfectionist about them (i.e. making them impregnable to referee objections). This is why I think you must demonstrating certain virtues in your writing that make it particularly acceptable to reviewers/editors. If I get some time, I shall have to examine your style closely! Anyway, you have a paper on OCD that I intend to read, since it's a shared interest.