Giving a Talk

The main way how science is communicated in CS is through publications (papers) and talks (presentations) at conferences or during visits of other research groups.

Papers document your scientific work, explain your methodology, and report on your findings. They ideally contain all the details that are necessary for someone else to reproduce your work.

To attract people to engage with your work, you submit the papers to conference and, when accepted, you can give a talk about your work.

The goal of that talk is not to convey every single technical detail and to "show off" you intellectual aptitude or the unparalleled depth of your work but to raise the interest of people in your work and attract them to engage with it. Like anywhere else in our society, attention is a scarce resource. Ideally, a good presentation leads to curiosity, interesting discussions, new contacts, scientific collaborations, more scientific work, and so on.

Here are some suggestions on how to give a good talk:

• The most important thing is that you are understood. People need to understand what you are talking about. Nobody will follow a presentation in which they are lost.
• Hence, make sure that you assess the level of knowledge of your audience correctly. One typically tends to assume that the audience has more knowledge than they actually do because after you have engaged with your work for so long and so deeply you tend to think that some of it is actually very easy to understand and you tend to assume that everyone else will pick it up quickly. This is not the case.
• As a rule of thumb, I recommend putting yourself into the situation of explaining your work to a fellow student who attended the same "introduction to X" course with you where X is your research area. In doubt, don't assume that anything is clear.
• It is challenging to understand and decide what to leave out. You cannot talk about every detail and you need to decide what you won't say. It is much better if people understand only the essential aspects of your work than if they get lost in the details.
• This is hard because you and your audience are at different spots: They are at the entrance of your thought process, you are deep down. Put yourself in their position and pick them up appropriately.
• Every researcher looks for the answers of the following questions:
  • What is the problem and why is it an important one?
  • What is your contribution and how does it improve on the state of the art? (what can you do what others couldn't?)
  • How does your technique work? on a high level
  • What are the results? How well did you perform? How successful were you? Make sure you answer these questions well. This allows people to decide if they want to engage with your work further.
• Use as little text as possible and as many visuals as possible
• Don't use the slides as your notes (i.e. don't do PowerPoint Karaoke). They are a visual guide to what you are saying.
• Use examples instead of abstract treatises.
• Make sure you are on time: You will use more than 1' for a slide so make sure you don't have too many slides.
• End on the conclusion/future work, do not use heading slides (they do not convey information and hinder the flow)
• Practice the talk with your fellows and keep track of the time

References

• How to Give a Great Research Talk, Simon Peyton Jones