I just got back from the Artificial Intelligence '04 conference in Cairns. It was a great conference, met lots of people, went to some interesting talks, saw some fantastic keynote talks, and generally had a good time. I even got to go for a swim and soak up some sun on the last day.
But I am writing this particular entry to capture some points I think are worth recording, based on some of what most people would agree were the less successful talks. If anyone is giving a technical talk, maybe some of the following will be useful...
- Don't stare at the projected slides: a frightening number of presenters will come up to the podium, turn their back to the audience and proceed to stare at the slides on the screen behind them. This alienates the audience, and people will switch off.
- Don't ignore the audience: these people came to listen to you: look at them, scan the room, engage with the people as you talk. Don't stare at your notes, at the screen or at the projector. Read the next bullet point, then talk about it for a while.
- Don't mumble: speak slowly and clearly. If you are nervous, you are more likely to rush. Take your time, as anyone unfamiliar with your topic is unlikely to keep up if you are speeding ahead.
- Don't read the slides: your audience is most likely literate; if you have every word you are going to say on the slide, and proceed to read directly off the screen, you are certain to send people into a deep slumber.
- Don't fill the slides with text: a good slide has no more than about 4 or 5 bullet points. These should highlight key ideas, but not be whole sentences. Do not write whole paragraphs of text. Not only is it too dense, it will be hard to read from the back of the room, and you are more likely to read it rather than talk naturally.
- Don't use distracting animations: animations can very effectively be used to illustrate important concepts, but there is a great danger of being overused. One talk I went to featured an animated logo in the top left of every single page, a spinning 3D logo of the lab: this was annoying and distracting, and did nothing to help the speaker.
- Don't assume too much: your audience is likely not the expert you are in your particular field, so don't jump in and start like you are speaking to your boss; some level of gentle intro is useful to give the audience an idea of what it is you are talking about
- Don't present the paper: some people try to present the entire paper in slide format, which is doomed to failure. Your presentation should be something that will make people want to read the paper. It should feature the highlights, key contributions and main points, but don't bother trying to cover everything; it is just not feasible, and you will lose your audience.
- Don't go for style over substance: some of the templates in the more popular slideshow packages have some horrific choices of fonts an colours. Especially stay away from those with animations; they are nothing but a distraction.
- Don't use a table when a graph will do: graphical information is much easier to comprehend, so always use a graph whenever possible. Don't fill up a slide with columns of numbers and expect people to interpret it - that is your job.
- Don't throw away questions: people asking questions should not direct it only to the speaker; say it loudly and clearly so the rest of the audience can hear you. Speakers should repeat the question (or paraphrase it) before answering, at least to mitigate against the first point.
- Don't miss an opportunity to practice your talk, and practice public speaking in general. Toastmasters have some excellent courses that can improve everyone's speaking technique. Especially for people whose first language is not English, take courses and practice as much as possible. Since most scientific publications are in English, as are most international conferences, this is a critical area you cannot afford to neglect.
I hope the above is of some use, please feel free to leave a comment for any other ideas on what to include in the above list.