Academic Poll: Paper Torture

If it's the first time you have attempted to write a paper on a particular topic, the introduction can be hard, but once you are writing subsequent papers in a series of papers on that topic you can re-use much of the previous literature search.

The real problem is the results section. You have to make every figure count, especially in a letter length paper, and that means figuring out exactly what to include, how to present it, and how to describe it so someone who is knowledgeable about your specialty but hasn't been looking over your shoulder while you were doing the work. Sometimes this part is easier because you have given a conference presentation on the work and have therefore confronted similar though not identical constraints on what you show. But one way or another, you have to construct a story.

Getting started is the hardest part.

editing.

"Other" explanation: the hard part of writing a paper is (a) dealing with what people aready "know" (or think they know) and (b) the argument.

Because I write in the social sciences, the papers are longer. The set of relevant "data" (evidence) is always large (if not nearly infinite). You want to streamline the argument: these are the important facts x, y, and z, of which y is new and makes the relationship between x and z different than you had thought, and together it now supports conclusion Q.

But instead of being concise and graceful, you have to deal with readers who say "what about fact (?) d, f, g, h, i, j, etc.? What about them? So you have to spend more of the paper dealing with explaining why d, f, etc. are not facts, or are not relevant, before you can get to the good part, the elegant exposition of x, y, z, and Q.

Since I'm writing for a biology audience these days, the hardest part is trying to translate mathematics into English. And not just equations, but the whole mass of basic stuff anyone with a physics or math degree takes utterly for granted. And then I hand it to a biologist, and they say, "This is really abstract and mathematical." I'm about ready to start a biologist reeducation camp.

Explaining in a comment :).

Actually, I would say that the most difficult part is the abstract, BUT only because you can only write it once you have written the entire paper. The figures (aka your data) are the hardest part! My adviser always used to say, once you have the figures, you have the paper. They are the "soul and center" of any (scientific) literature.

Also....you don't already have LaTeX on all your computers...? I'm just a silly little physical chemist, but I TeX'ed up my dissertation, and have WinEdit on my scrawny little labtop (I really, really, really hate Word).

To echo above posters, the first paragraph is easy-- once you have written everything else.

Also....you don't already have LaTeX on all your computers...? I'm just a silly little physical chemist, but I TeX'ed up my dissertation, and have WinEdit on my scrawny little labtop (I really, really, really hate Word).

My desktop machine at work was "upgraded" to Vista earlier this year, meaning that the previous LaTeX installation no longer works. I only use it for technical work (class notes and the book are in Word), so I haven't needed it until now.

I only use it for technical work (class notes and the book are in Word), so I haven't needed it until now.

Yikes. I mean.... I used Word for a very long while...until one of my math friends yelled at me to use LaTeX. I'll assume, however, that the publishers/students wanted a Word format....?

Also, sorry, just to be annoying (sorry!), miktex + winedt should really work on Vista (as well as Windows 7)....www.miktex.org (sorry, getting yelled at by my personal physicist now). But, if you gotta use Word, I empathize.

I really, really, really hate Word

Make room for me on the Word haters bench.

I have it on my desktop machine only because (1) some people, my boss and certain collaborators among them, assume that I can deal with that format; (2) TextEdit (I'm a Mac type) can read straight text .doc files but can't deal with such files if they contain figures, equations, or macros; (3) Open Office was not a viable option at the time I acquired the computer. I avoid creating documents in it because it's amazingly clunky for such an expensive and established piece of software. Equation Editor is a beast. The hyphenation algorithm isn't as good as the one Knuth developed for TeX almost 30 years ago. Worst of all, the format changes from time to time--Office 2007 introduced the infamous DOCX format, which earlier versions (I have Office 2004 and no plans to "upgrade") cannot read and which many scientific journals still will not accept.

As for why Chad typed up his class notes in Word, I can only speculate that his local computer wizards foisted the OS upgrade on him at an inopportune time, and only now has he had the time to reinstall LaTeX. Sometimes the need to get something done right away takes priority over doing it with the right (but nonfunctional at the moment) tool. (It doesn't explain why Chad used Word for the book, but I'm guessing there aren't many equations in the book.)

I have a deep and abiding love of LaTeX. Emacs and TeX are my word processor, as of last semester.

I haven't had to do any scientific writing. It's mostly research papers for me. In that, my greatest obstacle is ensuring consistency -- of various types:

I'm not a fantastic writer, so generally if I don't write a paper in one sitting, inconsistencies in word choice, level of formality, and tone creep in.

Also, When doing research papers I often have a working draft going while I am doing the research, in addition to any notes I take. I don't take many notes, though; with the volume of electronic media I have access to I often have the full text of my references on hand.

Because of the above, or because further thought causes me to revise my opinions, the conclusion I take from my research can change greatly and repeatedly.

Finally, laborious proofreading. I find myself poring over my papers, trying to be aware of what I assume, imply, reference and speculate about. Citations are easy; LaTeX/bibtex takes care of that. But sometimes I remove a citation and find that I've alluded to it later, necessitating that I work the citation back in. I often find that a sentence I've skimmed a dozen times actually implies the exact opposite my explicit conclusions. I "dike out" whole paragraphs as unnecessary commentary.

I'm pretty sure all of the above problems stem from either sloppy research or sloppy writing methods. I should probably postpone bashing bits on my first draft and take copious notes instead. Similarly, those notes should help me to carefully check my assumptions.

@Frederick Ross:

I have a special place in my heart for papers I download (or worse, interlibrary loan) only to find out that they have pages devoted to single equations. And by special place in my heart I mean I put my drinks on them at my desk so I don't get those little dirty circles of junk all over the place.

Should be a felony to not mention modeling in the title and have a disclaimer in the abstract :p

Anyhow, for me the hardest part is the discussion / conclusion because that's where reviewers are going to be a pain about it and may really disagree with what we want to say. Aside from one who blew a gasket because we didn't cite the papers of a specific author (presumably his :p) in our intro.

The introduction, hands down.

But for the current paper, the conclusion. That's because I have come to the reluctant conclusion that the work isn't done, yet. And to finish it properly I am going to have to learn a *lot* of mathematics. If only I'd finished it last month, before I discovered this! I could have sent it in with a clear conscience.

William Hyde

The awknowledgement section......who likes to thank others? I want to hog all the credit....so, does anyone want to collaborate with me? :)

IMHO:

(1) the abstract is the EASIEST part to write, so long as you do it LAST, and rewrite repeatedly and with collegial good advice on editing.

(2) the introduction is, thus, the second easiest thing to write, as you do it next to last. You have to resist the impulse to put as much as possible in it.

(3) the description of methods -- this is usually done in too much of a hurry. It should be rewritten many times, and have useful editorial advice as to what you are assuming without having said so.

(4) the description of results -- this should be fairly straightforward if (a) you are honest; (b) you have someone helping you with good graphical presentation.

(5) the conclusion -- harder than the description of results, as you have to connect with what the data actually says, and not what you want it to say. But you are allowed to hint at implications that do not strictly follow logically from the data, but relate to some hypotheses.

(6) some other element -- picking the right problem; building the right team; getting funded; giving dry runs and presentations of partial results to get informal feedback before the referees do their thing; doing a clear-headed and appropriate-depth literature search for the references; selecting the right venue for submission (neither too hard, say Nature, Science, Phys Rev; nor too obscure but safe).

I voted for introduction, because the blank page is *evil*. But I work as a grantwriter (working with scientists to improve/support the grantwriting process), and I write a lot of NSF grants. The hardest part of that are the required "Broader Impacts" and "Intellectual Merit" sections.

The introduction is the most painful, shortly followed by the conclusion. The rest if more or less based on facts only, but these areas are where you set the tone for the paper and to a large degree determine what the reception of the paper is going to be. These are also the areas reviewers always seem to want the most painful, time-consuming edits.

One good trick for the introduction is to write a draft version, then throw away the first paragraph. That paragraph has invariably become a vague, mealy-mouthed, wordy non-entity while your second paragraph turns out to be perfect to start with.

The bit, somewhere in the middle, where I realise I've got so far inside my argument that I've disappeared up my own backside and that I should have gone into science instead of humanities because then I can say "As you can see, the data clearly supports this conclusion" while battling ninjas on unicycles and feeling smug because I wouldn't have to try and interpret Foucault.

I reckon the conclusion, because sometimes what you conclude isn't what you started out trying to prove.

Making graphs. Everything else is easy in comparison.

I voted for the introduction, but I'm trying something new, based on an approach outlined in a neat text, " Write Like a Chemist". Write the experimental section first (easy to do!), then results, discussion, conclusion, introduction and abstract. I'm hoping that this will help avoid the "how do I start" block.

Add me to the non-Word block. And since I use Linux, I can't use Word anyway. LaTeX is what I use for almost everything, except committee meeting agendas.