Tips for a Good Conference Talk
Here's some advice I sent to the SIGMOD '05 conference
speakers (I was PC Chair). Some of the points were inspired by a
SIGMOD 1995 note-to-speakers from that year's PC Chair Mike Carey, who
in turn attributes Marianne Winslett and Catriel Beeri.
- Plan very carefully what you can cover in the allotted time. You
have 25 minutes, and a conscientious session chair will ruthlessly
cut you off if you attempt to exceed that limit.
- Design your slides with a large room and audience in mind. Use large
fonts so your slides are visible from far away. As a general
principle, don't put too much on each slide.
- Given the time limit there's simply no way you can present all the
results in your paper, so don't even try. Think of your talk as an
advertisement -- your goal is to entice the audience into wanting to
read your paper. Motivate the problem; describe your overall
approach and your major results. If your work includes experiments,
pick a representative graph or two.
- Don't put everything you want to say on your slides and then read
them during your talk. You will captivate your audience by forcing
them to listen to you, looking at the slides only for cues and
diagrams.
- Keep your examples simple, emphasizing the main points, and give the
audience enough time to digest each example.
- It is strongly recommended that you practice your talk several
times, especially if you haven't given many conference talks in
the past. Time your runs; get your friends to listen and criticize.
- If you are not a native English speaker, make an effort to speak
slowly and clearly enough for a large audience to understand you.
Even if you are a native speaker, you will need to speak loudly and
clearly.
- Position yourself carefully with respect to the screen -- be sure
not to block the audience's view.
- During the question & answer session at the end of your talk, be
sure that your audience knows what question you are answering --
repeat a question if not everyone was able to hear it. It's often a
good idea to repeat questions regardless, to make sure you have them
right and to give yourself a moment to think.
Most of all, have fun, and remember -- your talk serves as an
advertisement for your work and your paper.