Presenting a Paper: A Collection of Tips
- Slides should be kept sparse; their role is to guide the
audience (as well as the presenter). The audience should be listening
to you, not reading the slides. If the slides are too dense, the
audience will spend its time reading them instead of listening
to the presenter. Keep your slides simple; no more than about 10 lines
of text.
- A good rule to follow is to keep each bullet to one line only,
even if you really have to work at it. If the bullets seem to require
more than one line, you may be covering more than one idea in a
bullet. In over 7 years of presentations, I've never broken this rule.
- Slides should use large font that is easy to read; even the
smallest font, used in figure captions, must be large enough to be
legible by a bespectacled person from the back of the room. A good
rule here is to have the smallest font be 20pt (Times Roman) for
office rooms (less than 15 people), 24pt for conferences and
workshops.
- Slides should be simple; a great deal of fancy art may be nice to
look at, but is also distracting from the scientific matter being
presented.
- Try to avoid animations, and if you need them at all, make them
simple. I've found that for relatively dense slides, using
bullet-by-bullet appearance (greying out bullets as you move along) is
very effective. But then again, your slides should not be dense, and
so you shouldn't have to use this last-resort measure.
- Never copy the associated paper and show its pages on the
overhead. The information in papers is too dense and impossible to
read when projected. This is considered very poor presentation form.
- Avoid covering parts of your slides with paper; use multiple
slides or overlays for that purpose.
- Avoid turning the lights too low in the presentation room.
- Avoid giving talks during and after meals, if possible. You are
there to get the best possible audience for your work.
- Show up early, and if possible, check that everything works (the
vcr, overhead and/or laptop projector).
- practice, practice, practice. Then practice some more. Practice
giving the talk until you can't hear it anymore, and then practice one
last time. And then another. Giving great talks is like performing at
the Olympics: You can only do it if you've practiced on this talk
until it hurts. And then some.
- Avoid eating too much before your talk.
- Only give demos if you are certain they will work; live demos
rarely work in talks, so they tend to lessen the positive impact of
the talk. It is better to provide a working demo on the Web page, and
give the page in the talk and/or show a video.
- A talk should contain these slides/sections:
1) Title & author(s)
2) Outline (very sparse)
3) Motivation
4) Approach and methods
5) Current results
6) Strengths and weaknesses
7) Related work (if time allows)
8) Summary (take-home message)
- The title slide should be sparse, but contain clear affiliation
and contact information (email address and URL). Make it easy for
people to learn who you are and where you are. Make sure you give
credit to collaborators. If you are giving a talk on behalf of a
group, it could be useful to highlight your name (tastefully!) so that
people recognize which one of the authors is giving the talk.
- The outline slide should be very sparse, and should be presented
briefly, just to give the audience a sense of direction. The outline
is optional in talks lasting < 20 minutes. Never use a generic
outline (i.e., with generic bullets such as "Introduction",
"Background", "Motivation", as above). Instead, use this opportunity
to already give the audience some of the key terms they will hear, and
the take-home message they will see.
- The motivation slide should briefly but clearly situate the work
within the context of the field and present the key challenges being
addressed.
- The approach and methods part of the talk should be very clear;
work out your terminology so that it is consistent with that used in
the field and throughout the talk.
- Make sure your notation is clear and consistent throughout the
talk. Prepare a slide that explains the notation in detail, in case
that is needed or if somebody asks.
- Always label all of your axes on graphs; use short but helpful
captions on figures and tables.
- If you have experimental results, make sure you clearly
present the experimental paradigm you used, and the details of your
methods, including the number of trials, the specific analysis tools
you applied, etc.
- The talk should contain at least a brief discussion of the
limitations and weaknesses of the presented approach or results, in
addition to their strengths. This, however, should be done in an
objective manner -- don't put down your own work.
- If time allows, the results should be compared to the most
related work in the field. You should at least prepare one slide with
a summary of the related work, even if you do not get a chance to
discuss it. This will be helpful if someone asks about it, and will
demonstrate your mastery of the material.
- The summary/conclusion slide should pull the audience out of the
low-level details of the approach, and highlight / reiterate the key
contributions and issues of what you have presented. This slide is
the take-home message of your talk, do it well.
- If you plan to show videos during/after your talk, plan their
timing carefully. Sometimes the process slows the talk down, because
the movies break or don't show properly; this can make your audience lose
interest and focus. It is often better to leave videos for the end,
during a question/answer part of your talk. In any case, test your movies
in the specific room and projector, prior to your session.
- Make your talk clear and memorable. It may help to be
entertaining, but be careful not to do it at the cost of being viewed
as not taking your own research seriously.
- Common mistake: A slide with one top bullet, with many sub-bullets.
Use the top bullet for the title, and make all sub-bullets first-rank.
- Spell check again.
Additional useful tips: