This blog is about how to give a physics talk, designed as a resource for my students and postdocs (typically 10-20 minute talks), though I will also add some comments from my experiences as a professor (for the 1 hour talks).
What I say here is not written in stone, of course, just a set of rules that works well, according to me. If you have any suggestions please do share them with me. Read on for the rules, as well as to find out why a 'microcentury' is a relevant unit of time in this case.
i) Slide format: Use pdfs for your slides. I avoid PowerPoint slides as they do not always display well on different machines. Fonts and formatting do not always come through, with disastrous results. Avoid layering. Avoid animations, no matter how cool they are.
ii) Number of slides: About half (for 10 or 20 minute talks) to one-third (1 hour talks) the number of minutes available for the talk.
iii) Slide layout: Font should not be too small (especially for members of the audience above 50!). Slides should not be too cluttered with pictures or equations. I have seen slides where every sentence was underlined three times, or where seven colors were used on every slide.
iv) Sections of the talk: I like to arrange it into the following parts
a) Cover slide - the title should be catchy but technically correct and not too long. It should list the name of the presenter prominently (and institution), as well as collaborators, and also acknowledge funding.
b) Talk Outline - not too many sections should be listed (you don't want to imply it is going to a tedious talk with serious potential for running over time, see below for more).
c) Main section - the meat of your talk, where you build your model and share it's predictions. If you are giving an invited talk, do not forget to mention the contributions of your host (department) to the topic. They'll invite you back if only because you praised their work to their colleagues.
d) Conclusion - should be brief, should also refer to possibilities for future work, and thank funding agencies again (better to bow too many times than too few).
v) Overall plan: Do not forget to thank the person who introduced you (who is also likely your host), acknowledge what a pleasure it is to visit/be included, and compliment the other speakers in the session/at the conference.
For the actual talk: First tell them what you are going to tell them, then tell them, then tell them what you told them. When I initially begin assembling my talk, I find I am saying too much and rushing through it. When I am finally ready to deliver it, I use far fewer words (what you leave out is as important as what you put in - e.g. the audience will not question what you do not say), and adopt a relaxed pace, with meaningful pauses thrown in. Give the audience a take home message.
vi) Level: When I give 10 and 20 minute talks (typically contributed at a conference) I focus on the technical aspects as the session typically consists of experts in my field. When I give 1 hour talks (whether seminars or colloquiua), my aim is to speak at a level which can carry along as many people in the room as possible, rather than try to impress the two experts in my specialized field sitting in the first row.
vii) Interaction: In the 10 and 20 minute talks, questions are typically fielded at the end of the talk. In the 1 hour talk at the very beginning I request the audience to interrupt me with questions whenever they feel like. The best talks I have given involve a steady, but not overwhelming, exchange with the audience. Such an exchange is a very good indication of how well the audience is following me.
viii) Timing: I always aim to finish early. An audience likes nothing more than to learn than that it has completed its task - that of listening - ahead of time. It is also good to have some margin in hand in case there are problems with the previous speaker (running over!) or the audio-visual system.
For the 10 minute talk, usually it is 8 minutes for the talk and 2 minutes for questions. For the 20 minute talk, often the last 5 minutes are left for questions.
For the 1 hour talk, I aim to talk for 52 minutes, which approximates a "microcentury", as calculated by John von Neumann, and reported by the great mathematician Gian Carlo Rota at MIT (see his Ten Lessons I Wish I Had Been Taught, available on the internet). I have seen colloquiua go over by 30 minutes. Going over is the easiest way to upset people and not get invited back (applies to teaching classes).
ix) Answering questions: Answer questions with honesty and transparency. If you don't know the answer, say so. This is actually your 'get out of jail free' card. If the answer is complicated say you are willing to discuss it afterwards.
Comments