The Short Talk
Charles Van Loan
Department of Computer Science
The short talk (<= 20 minutes) is a fixture at most scientific conferences. Assuming that you are using slides, this note focuses on how to give a good short talk.
I think that if you can give good short talk then you can probably give a good 50-minute presentation because the additional time permits a certain flexibility. For example, in a 50-minute talk I think that the speaker can risk losing the majority of listeners for 10 minutes while a technical aspect is embellished for the "experts". In contrast, the short talk requires a more sustained level of clarity if it is to be successful.
It is more profitable to dwell on the positive aspects of the short talk framework than to stew over why you were not given a longer time to speak. Here are some perspectives:
A short talk is a long abstract. If a one-hour talk corresponds to a paper, then a short talk corresponds to a long abstract. It should communicate without distracting detail.
A short talk is a captivating lead paragraph. For a reporter, the quality of that lead paragraph will determine who reads on. A successful short talk will encourage the listener to ask questions, or to buttonhole you outside the lecture hall, or to pursue your literature pointers.
A short talk is a commercial. It is a great occasion to advertise your research accomplishments and/or expository skills. But unlike a prime time spot on network TV, it should give an accurate picture of the product!
The right perspective depends on the audience. Are you speaking to undergraduates, a small group of research friends, visitors from a granting agency, or drop-ins at a large multi-session conference?
Content and Structure
Amount of Material. Resist the temptation of cramming too much into your talk. This is perhaps the greatest pitfall of the short talker and the underlying psychology is clear:
"What I have to say is too important to be edited down to my allotted time."
"If I do not overstock my talk, then the audience will think that I do not have much to say."
The crammed short talk is a negative statement. It indicates that the speaker is incapable of editing for importance. Of course, crammed talks are often the consequence of personal enthusiasm for the material. But recognize that it is possible to edit what you say without chilling your enthusiasm.
Audience Make-Up. If you have a heterogeneous audience, then you may have to make hard decisions with respect to the assumed level of expertise. No one will be happy if you appease the unprepared with ten minutes of background notation and then placate the expert with ten minutes of rigorous minutia. The best course of action is to peg your talk at a level of expertise that suits your goals. State those goals early on and apologize (once) to the potentially offended parties.
Organization. Like any presentation, a short talk should have a beginning, a middle, and an end. Start with an outline of the problem and why it is important. Then delineate the central contribution with suitable pointers to related work. Do not end with references but with a conclusion. Repeated slides are a good way to emphasize important points. For example, at the end of your talk you may want to display an early slide that itemizes the main points of your talk.
References and Acknowledgements. Be gracious, especially if people in the audience are involved. Vague collaborators can be listed on your opening slide.Specific collaborations and references (by name) should appear "in-line". Stepping through a brief bibliography is one way of outlining related work. There is not much point in giving complete bibliographic information on a slide, but be prepared to provide reference details should there be a question.
Algorithms and Proofs. The presentation of algorithms and proofs requires great care. If a low dimension example captures the essence of an idea, then use it. If you feel that a step-by-step approach is instructive, then be informal enough so that the whole thing fits on a single slide. If the enforced brevity is too disruptive, expand selected details on subsequent slides.
Benchmarks and Data. With respect to benchmarks and other statistical data, remember that facts do not speak for themselves. A few well chosen results can enliven a talk and clarify difficult concepts. Too many statistical results again point to a speaker who is incapable of editing.
Anticipating Questions. After you have finished writing your talk, think of the three most likely questions that might be asked. If any of these are particularly "nasty", then you should revise your talk with an eye towards heading off potential embarrassment. The impression that people have of your work will be more positive if its shortcomings are delineated by you and not by a member of the audience.
How well your short talk is received is as much a function of delivery as content. Here are a few pointers.
You do not have to say everything that is on a slide. For example, the audience can read an equation faster than you can speak it.
Concentrate on the audience and not the projection screen. If you must reference an item on your slide, use a laser pen.
Do not fixate on a V.I.P. who happens to be in attendance. It will bother the rest of the audience and you will appear too eager for approval.
Avoid the peek-a-boo style of gradually uncovering a slide to force the audience to read along with you. The slide is a different medium than the blackboard and it is better to gracefully accept its disadvantages.
Give the audience 4-5 seconds to scan a slide before you begin talking about it.
If you have a dramatic one-liner, then have a slide with just that one-liner.
Do not hardwire jokes into your talk by placing them on a slide. It is better to let the humor level be established dynamically as your talk unfolds.
Never apologize by saying "there is not much I can do in such a short talk." If you feel that way, then you have no business giving a short talk.
Write clear slides. Avoid cutesy icons. It is difficult to imagine a nice slide that has more than 7-8 lines of material. If there is a lot you want to say with respect to certain slides, talk slowly. The slides should not be a verbatim transcript.
It is fine to piece together slides from different talks. But make sure the styles mesh so as not to telegraph the fact that you are cobbling together material from prior presentations.
Never use a slide that is a photocopy of a journal article or report. The printing will be too small and it will be a tell-tale sign that you have not sufficiently edited your talk.
Practice your talk to anticipate time of delivery. It is helpful to identify "dispensable slides" which can be jettisoned without damage should you run out time. Keep a subtle eye on the clock or have a friend in the audience give you an agreed upon signal when it is time to wrap things up.
Of course, you should end with the usual "are there any questions?" Unfortunately, if no one raises their hand then (right or wrong) negative conclusions are reached. The best guard against this is to set an appropriate pace. If you arrive breathless at the end of the talk, then the mood is not conducive to questions. Another trick is to relate your talk to the work of somebody in the audience. That person will often feel obliged to comment.
In the end it takes practice. Expect to blow away a few audiences before you deliver your first twenty-minute masterpiece!