How to give a good conference talk.

Oh my. I just finished chairing a session at a conference and decided this would be a good time to share some advice about giving conference talks.

General comments:
When I first started giving talks as a grad student, I was really afraid and nervous. I thought of the audience as a hostile entity that I had to prove myself to. I thought that they all new a lot more about the topic than me and that I had to demonstrate my knowledge and mastery of esoteric points in the research to show that I "belonged" in this group.

This is NOT true. About half of the audience in a conference talk are the other speakers and/or their advisors. The other half are people that have an interest in the area or that know little but would like to move into research in that area. The chances of finding an expert in your talk are low. You might have 1 person who works in an area close enough to yours to offer true criticism. If you give a highly technical talk, you are speaking to that one person to the detriment of the rest of your audience.

Most people in your talk just want to be entertained. They want to understand what you are doing and more importantly why you are doing it. If you start to think of your talks as teaching a class to your peers or giving a group meeting, your talks will improve drastically. And even if there are serious technical questions, your audience will remember you as a good speaker. The key here being they will remember you, which is the reason that we give talks in the first place anyway.

Specific Comments
1. DO NOT run over your allotted time. The most riveting speaker ever will be reviled for running long. Generally, the suggestion is 1 minute per slide. I find that about 75% of this is better. So for a 25 minute talk I would have about 18 slides. This gives you time to really explain the material and not fly through data.

2. Corollary of this is that you do not have to present every last piece of data that you collected between this talk and last year's. It is perfectly reasonable to present data that has never been seen before as "previous data" so that you can focus on one interesting aspect of your results. In fact this is preferred.

3. Get rid of the words. The best way to explain is to show. That's why you're doing a presentation and not simply a speech. Most of my talks now consist of pictures and captions only, except for the conclusions and acknowledgements page. This means that when I do use words they are that much more meaningful. So I might use words to highlight a key experimental question that we are trying to address.

4. As part of this, don't be afraid to deviate from the powerpoint slide template. In fact, I never use them. Every side I make including the title slide starts out as a blank slate. I add each text box and figure without predetermined boxes telling me where they should go.

5. Related to these ideas, be like Feynman. Okay I love Feynman. He was a great scientist and a great teacher and it doesn't get better than that. One of the key points that he emphasized was trying to distill knowledge down to its simplest unit. You don't want to belittle your audience, but make everything as simply as possible. When you finish your slides, go back and ask yourself if there is any better way to explain what you are trying to say. Make it as easy as possible to understand.

6. If you have a short time slot (< 1 hour) do NOT present an outline of what you are going to talk about. This just wastes precious minutes in an already short talk.

7. It is pretty rude to only attend a session for the talk that you are giving. If you are the speaker you should stay for all the talks in that session unless you have a very good reason for not being there (another talk in another session). Of course bathroom breaks are fine and you might bump into someone and have a short conversation, but to gather up your things and walk out right after you finish is inexcusable.

8. Make graphs readable and explain them. Saying I know this slide is busy or I know that there is a lot of data here does not change the initial shock that viewers experience upon seeing a busy slide. Graph axes should be large enough that people in the back row can read them, and even if you are showing a graph that uses a common method that your audience probably knows (like an NMR or FTIR profile) you should explain what each axis is and FULLY interpret the data to your audience. There is nothing worse that a speaker saying, "as you can clearly see..." when we have no idea how to interpret the data.

9. Talk slowly. It can be very hard to follow a speaker who is talking rapidly. This problem is worse if English is not your native language. Please, please, practice talking slowly. A rushed presentation of lots of data impresses no one. Your audience will tune out before you get to the third slide and starting praying that you won't run over.

10. Give context for your work. The most important part of your talk is explaining what you are doing, why you are doing it, and how it relates to the work of others. Cite others liberally and often, you never know when they will be in the audience. Nothing makes me angrier than someone who talks about a particular technique but doesn't cite the very well known source.

Well hope this helps, and the most important advice is to RELAX!!! Remember that most of the audience is probably sitting around being nervous about their talk, or just tyring to pass the day. They are not out to get you.