The many stages of writing a research paper

Timothy Weninger recently submitted a research paper to a computer science conference called World Wide Web. On his way to that, he went through 463 drafts. Bear in mind, this paper has only been submitted, not yet accepted, so there's probably even more edits that are still yet to happen. Welcome to the life of a scientist.

In this video, Weninger created a timelapse showing all the different stages of his writing process, as he added graphs and went through cycles of expanding, contracting, and expanding the text. But mostly expanding. The paper grows from two pages to 10 by the end of the video.

[Video Link]

Via Bill Bell


  1. Man, that’s cool.  It makes the process seem rather orderly.  My process is not orderly… if I had made this video it would be interspersed with clips of me screaming in pain and anguish, and having satori like moments of enlightenment…  then more screaming… 

      1. Your institution’s college (of A&S or whatever) will have requirements. Ask them what to use. Many prefer you to use Word. Horrible, awful, stinking Word.

        1. LaTeX is the best thing ever for long projects (I have written several books with it). Your university probably has a LaTeX style file for dissertations, that will help make it satisfy their format requirements with no effort on your part.

  2.  If your dissertation is in a technical field or has more than one figure, citation, table or other thing that needs to be cross referenced, you should use LaTeX. Period.

    1.  It’s not a technical field (history), but requires extensive citations from a broad number of primary (some archival materials as well as magazines/zines, etc) and secondary sources. 

    2. I have to second, ‘check your submission requirements’. (Also, if you are planning to publish all or part of your dissertation, check the submission guidelines for those publishers.) LaTeX is wonderful, but outside of physics / math / computer science, my colleagues seemed to do pretty well with MS Word + Endnote (a bibliography management and formatting tool). 

      I’d also hesitate to recommend anyone ‘new’ learn latex unless they were entering a field where it is widely used or were using it on private projects only, because being to work well with others (i.e., using commonly understood tools and techniques that facilitate collaboration with your colleagues) is so much more important than using the ‘best’ tool or technique.

  3. I call shenanigans. At no point does this paper exceed 10 pages. In reality, the process of writing a 10-page scientific paper always includes episodes of it being several pages too long and having to be cut down, painfully, word-by-bloody-word. If that doesn’t happen, then clearly you don’t have enough to say to make it worth writing a full conference paper.

    1. Look at the bottom right corner… The content moves out of the frame from time to time. Or how did you conclude it never exceeded 10 pages? Also, there are different “styles” of writing a paper. I usually have most of it in my head, start with the most essential and expand until it is good. 

  4. Neat, but I’ve got to quibble.  From Timothy’s website:

    “While writing this paper I took a snapshot everytime I compiled the latex file. This resulted in 463 individual snapshots.”

    Calling that “463 drafts”, particularly looking at the series of snapshots, is totally misleading.

  5. My dissertation was written for a doctorate in English. It had 5 chapters and an introduction. Each chapter had major versions indicated by letters; a major version would mean something like a new order of parts or significant modifications to the argument. Revisions were indicated by numbers, and typically ran into only single digits. So the third version of Chapter 4 with a three revisions would be 4c3. The hardest chapter to write was Chapter 3, which ended at 3g2, for a total of about 50 revisions over the seven different versions.

    The g version alone had  a total of 23.9 hours of actual editing time, as recorded by MS Word. Typical time to graduation for a doctorate in English is nine years, with only two years being coursework. I was typical, taking seven years to complete my dissertation after my comprehensive exams (while also teaching, generally, twelve credit hours a semester for my teaching assistantship).

    Research is quite demanding, even in a “soft” field like English.

  6. In reply to flagday: Agreed.  I write papers under source control (doesn’t everyone?) so it’d be easy to create such a time-lapse.  But often, checkins can happen even for minor tweaks – noway would those be considered a separate “draft.”  And given GUI LaTeX editors these days (TeXshop, Texmaker) you tend to hit “compile” even after adding a period.

Comments are closed.