Academic writing

Writing the paper forces you to do the necessary research and understand what you want to say. Start writing before the research is done. Apply lazy evaluation to the research.

• This forces us to be clear and focused.
• Highlights what we do not understand.
• Opens the dialogue: you can start sharing with other people. You can even consider submitting work in progress to conferences.

Understand the writing not only as the output of research but as the medium.

Make other people interested on your research; the ideas have to be durable and catchy and get people excited about them. Ideas are worthless if you keep them to yourself.

• Write about any idea. You do not need to have a fantastic idea. We all feel that our ideas are insignificant. Not much is loss if that is actually the case, but we can find things while trying to explain these ideas. Get started, do not wait for ideas.
• The idea. The paper needs to have one and only one ping: one clear, sharp idea. It needs to be clear when the paper is finished. If you have many ideas, write many different papers.
  • What is the key thought of any paper we read? We should make it clear to the authors. You can even write a section called: the main idea that makes it very explicit.

Imagine a whiteboard explanation.

• Here is a problem.
• It is interesting.
• However, it is unsolved.
• Here is my idea.
• And it works.
• This is how it compares to other approaches.

A commmon structure for conference papers is the following. You lose readers as the paper advances, so you should carefully plan the first pages.

Describe the problem and state your contributions. All of this is the first page, the one everyone will read.

• How to introduce the problem? Use an example. Four lines and the first example should be there, to illustrate the problem.
• Avoid a rant about how important is your problem. It seems good scholarship, but is demotivating to the reader. Simply put an example: “consider this thing, we will solve it in general”. It is better to describe a small goal, and then actually solve it.
• Contributions! State the list of contributions first: bullet list. This is the specification of what your paper will do. The reader should be excited about these contributions and then satisfied by the paper.

  • Contributions should be refutable, and with forward references(!) to evidence. You should point to where this is done in the paper in parentheses “Exciting claim (Section 3)“ instead of saying “Section 3 introduces…“, which is utterly boring.

    We describe (system)	No
    We give syntax and semantics for (system)	Yes
    We study its properties	No
    We prove soundness and decidability	Yes
    We use it in practice	No
    We build a GUI for Java	Yes

    If something is not referred here, maybe it should not be on the paper.

Related work should not be on the introduction. It distracts your reader from the main idea. The reader knows nothing about the problem. It is better to move this to the end and do it comfortably, where the reader is familiarized with your terms.

• There is a fallacy that says that to make your work look good, you have to make other people’s work look bad. Appreciation is a divisible commodity! do not do that.

  • Warmly acknowledge people who helped you.
  • Be generous to the competition “In his inspiring paper…”, “The lucid work of…”.
  • Express your gratitude to other papers.
  • Acknowledge weaknesses in your research; unawareness of the weaknesses is not helping. This goes at the end of the related work.
  • The related work should not be just a list of citations.

Try not to recapitulate your journey; it was difficult for you to get to this point, but the poor reader does not need to suffer that to understand it was good. The paper seems so simple that the reader feels it is irrelevant? That does not mean you need to scare them. The first approximation should not be scary.

• Exception: explain what you did not do the obvious thing. Why are you not taking the obvious avenue?
• It can be good to have an appendix for the painful details?

Intuition. The reader should be able to skip the details and still have the intuition.

Get your paper read by as many friendly guinea pigs as possible. The ideal would be some expert on the field that wants to read your paper; but any other friendly advice is also useful. Each guinea pig can read your paper only one, so use them wisely.

• You want them to tell you where did they get lost, not only on small orthographic details. Even discuss a bit with them those details.
• A good plan is to send your draft to the competition asking if you are describing their work fairly. They have helpful critique.
• Be grateful for any criticism. This is really, really, really hard; but it is very important. People is putting time into your work. The paper should be such that even Reviewer 2 needs to accept it works. Thank them warmly.
• Mention all guinea pigs on your Acknowledgements.