2 cents (and 13 points) on good academic talks

10 Aug 2017

One of the perks of being in academia is having the opportunity to listen to smarter people than myself talk about the cool things they have done. I find it a little frustrating that good research is often let down by bad presentation. As a regular attendee of many talks across departments, it is still shockingly easy how a presenter can rise above 95% of the talks I’ve seen - I will just jot down a few, with the caveat that I do not always practice what I preach.

caveat emptor: As with everything academia-related, Matt Might already has great tips for how to make a presentation better. I also found this document pretty useful. And of course, many of these tips are due to mentors and colleagues.

Big picture

  • Start with a question and tell a story to answer it. Tell the audience the purpose of the talk.

  • Prepare (!!!). Flip through the slides a few times. Even the best slides do not present themselves.

Presentation style

  • Do not talk to your slide. Do not stand in front of your slide.

  • On the flip side, presentation is not performance art. Being understood is better than being impressive.

  • Therefore, slow down.

  • Shift your gaze across the room regularly.

  • Modulate your voice. If someone is dropping off, speak louder and slower while looking in their direction.

What to present

  • KISS: Keep it short and simple. No-one (AFAIK) has ever complained about a talk being shorter than planned.

  • Present only the simplest non-obvious example.

  • No more than ~6 lines and one set of equations per slide (if you want anyone to listen to what you’re saying). I like sentence fragments, not whole sentences.

  • Have nothing on your slides you do not want to talk about.

  • Use title slides to summarize take-away insights. At the end, repeat important insight instead of a chronological recap.

Questions

  • Take time to digest questions. Avoid “waiting-to-speak” syndrome. Repeat questions so that you and the audience are on the same page.


comments powered by Disqus