6 minute read

These notes are a work in progress!

Writing is not something that has come easily to me. Through middle and high school I routinely would not complete essays I needed to write for class.

I noticed a big change after I started doing research. After a couple of years of reading and writing papers and absorbing advice on how to do so, I noticed that both my research writing and my essay writing dramatically improved. I found it much easier to write.

I still struggle with writing and finishing pieces, but it’s much better than it used to be.

In this post I’ll provide some scaffolding for many of the resources I have used when writing papers thus far. I hope to provide a structured collection of resources that both provide a coherent philosophy on the role of papers in (computer science) academia and actionable tips for improving the effectiveness of your own work.

I am acutely aware that I am not in a real position of authority to be offering advice to anyone on writing. Luckily, you don’t have to take my word for it. I have provided copius links to other people with decades more experience than me.

Community

What is research about? This lecture from UChicago helped establish my point of view.

Research is about community. See this lecture: The Craft of Writing Effectively

Fundamentally, research is about a community of people. From this framing, a research paper is an argument aimed at changing the minds of that community. The rest of this post will be framed with this perspective in mind. It justifies many other points to come.

Know Your Audience

Since a paper is designed to convince people of something, it’s important to understand who those people are.

Why are they reading your paper?

  • reading group
  • literature review
  • related work
  • ideas for their own work
  • just interesting

How are they reading your paper?

  • How to Read a Paper
  • How to (seriously) read a scientific paper

  • What does your audience already know or believe?
  • What does your audience not already know or believe?
  • What forms of evidence might convince your audience?

  • why do people read papers?
  • how do they read them?
  • who are you writing for, specifically, and why?
  • what things do they know or believe?
  • what things do they not know or believe?
  • what things might convince them?
  • speaking their language

Write papers that are easy to read!

Story

  • example-driven
  • beginning, middle, end
  • one key message
  • conflict-resolution

narrative arc

  • exposition, rising action, climax, falling action, resolution
  • rising action = problem
  • climax = key idea
  • falling action = consequences

The Research Paper Story Archetype

Let’s look at how the narrative arc typically plays out in academic papers.

Two related story archetypes:

  • Question-Answer: Construct a research question and find an answer.
  • Problem-Solution: Identify a problem and construct a solution.

Seems to me that problem-solution is possible as “research” in CS, because systems embody ideas. Systems are formal representations of ideas. TODO: honestly this probably does exist in other areas, e.g. development of biomedical techniques.

This dichotomy is overly simplistic. Many papers lie somewhere in between these two extremes; however, research communities often lean to one end or the other. TODO: this claim is a bit weak. Really need to flesh out this part!

everything else revolves around this central point

  • summary: title, abstract, intro
  • auxiliary artifacts: twitter, blog, talk, etc.
  • context: background, related work
  • broader impact: discussion, future work

it’s non-fiction! you don’t have to stick to prose style. use structure to your advantage!

Some notes on how to write research papers in CS (skewed towards PL):

Some great writers/presenters (typically correlated) who you should probably check out (mostly in PL) (alphabetical by last name):

  • Emery Berger
  • Derek Dreyer
  • Jana Dunfield
  • Ron Garcia
  • Dan Grossman
  • Philip Guo
  • SPJ
  • Shriram Krishnamurthi
  • Conor McBride

Paper Summaries In Situ

  • Title
  • Abstract
  • Intro

These are (i) affordances to readers to help them understand your paper and (ii) in situ sales pitches for your paper. Use them effectively!

Title

Abstract

The most important part of the abstract is the conflict-resolution pair. Get as quickly as possible to the conflict, as quickly as possible to the solution, and as quickly as possible to the consequences.

Intro

The introduction should provide a short, but complete summary of the entire paper. Its structure should mirror that of the paper such that you can put inline forward references to later parts of the paper in the proper order. You can go into more detail about the consequences of and evidence for your resolution here than in the abstract.

A Spectrum of Evidence

There are far too many types of evidence to mention in one place here, but I will nonetheless point you at some links to get started.

The important point for this post is that methods vary not only by the topic and idea you are writing about, but also by community. Different communities value different types of evidence. For example, POPL is generally interested in formal proofs (where applicable), whereas CHI is interested in user studies (again, where applicable). There are notable exceptions to these, and they should not be treated as hard and fast rules. Just remember: evidence must convince a community. That means they may be looking for statistically rigorous methodology or a description of your research process, etc.

Selling Your Paper!

If a paper falls in a forest…

A research paper does little to impact a community if no one in that community reads it! Thus we must “sell” our papers. We’ve already seen that portions of the paper can help sell it. But there are auxiliary artifacts as well. This can be done by

  • tweeting about it
  • giving a talk
  • writing an article
  • writing a blog post

Some resources on how to give talks (note: significant overlap with paper writing techniques and vice versa, so use those, too!)

How to Write: Nuts and Bolts

  • two styles of writing: full text beginning-to-end vs. outline first

Draft then edit! Write first, edit second! It’s important to separate them into separate phases. This is a technique that can be difficult to internalize but is tremendously useful not only for writing, but also for most (if not all) creative activities.

For general tips on writing nonfiction, consider On Writing Well by William Zinsser. This is my dad’s go-to book for writing tips.

For a more systematic approach to writing clearly, try Style: Lessons in Clarity and Grace by Joseph M. Williams and Joseph Bizup. This is my undergrad advisor’s go-to book for writing tips.

For your really low-level style points, try The Chicago Manual of Style. My parents, who are editors, swear by this book.

LaTeX

Just learn it. “Learn LaTeX in 30 Minutes” - Overleaf

Wrapping Up

  • community
  • the paper is a medium for convincing people in the community, everything else falls out of that principle

See Philip Guo’s “How to write a good HCI research paper (tips from senior colleagues)”