skip to content W3C SVG

How to Write a Great Research Paper

2014-03-01
  • Simon Peyton Jones

    • Don’t wait; write

      • Most people think: Get idea, Do research, Write paper
      • Instead: Get idea, Write most of the Paper, the paper develops the idea, use paper as forcing function to Do research

    • Identify your key idea

      • Your idea should be a re-usable insight that can be conveyed to others
      • Your goal with the paper is to spread the idea as if it were a contagion

    • Tell a Story

      • For an approachable paper, structure the narrative as if you were delivering it from the whiteboard
      • Overall structure suggestion for a conference paper (with page counts and expected readership)
        • Title (1000 readers)
        • Abstract (4 sentences; 100 readers)
        • Introduction (1 page; 100 readers)
        • The problem (1 page; 10 readers)
        • My idea (2 pages; 10 readers)
        • The details (5 pages; 3 readers)
        • Related work (1-2 pages; 10 readers)
        • Conclusions and further work (0.5 pages)

    • Nail your contributions right there in the intro

      • State the problem briefly (give a particular example and then generalize outward from there)
      • State your contributions
        • Make them crunchy (refutable claims are interesting and will draw readers forward)
        • Instead of giving a bulleted list of what each section is, add forward references in to your introduction (e.g., evidence in section 4, related work in section 5, etc.)

    • Save Related Work for later

      • Put it just before your conclusions, not just after the intro
        • Related work tends to get over-compressed in order to not bulk up the paper, making it incomprehensible for non-experts, and redundant for experts. It creates a barrier.
      • Throughout the paper, add forward references to related work
      • Use related work to make critical comparisons between your idea and previous work

    • Put your readers first

      • Take the most direct route to the idea (don’t take them down the same bload-soaked path, with blind alleys, that you took), except when necessary to dismiss obvious paths that are follies

    • Listen to your readers

      • Find guinea pigs to read your paper (experts and non-experts)
      • Maybe two at a time, because a guinea pig can only read the paper for the first time once, and you want to save some guinea pigs for the later drafts.
      • Be clear with them about what kinds of reactions you want
      • If a guinea pig got lost somewhere, have a dialogue and try a whiteboard explanation. If the whiteboard explanation works, that should probably be in your paper.
      • Getting expert help: maybe send draft to a ‘competitor’ saying “could you help me ensure that I describe your work fairly?”
        • They’re likely to be your referees anyway