Seminar on Distributed Data
CIS 700/1, Spring 2002
Advice on Presenting Papers
Want to read more advice on making good presentations? Try
The most important (and obvious) point is that you need to
understand and be able to explain every important point of the
paper you are presenting, in depth. This will typically involve
reading the paper more than once; it will often require referring to
Of course, deciding which points in a given paper are important enough to
read and research deeply is a judgement call. Signing up to present a
paper does not mean that you have to read the transitive closure of its
references, back to the beginning of history. But it does mean that you
must try pretty hard to understand (the key parts of) what it is saying,
and this will typically involve some background research.
With some papers, even some of the main points may simply be too obscure
or tricky to figure out in a reasonable amount of time. That's life.
But full understanding is always the goal.
- Preparing a good presentation takes serious effort. Allocate at
least ten hours.
- The golden rule of all good presentations is to think about
your audience---who they are, where they are coming from, what they
already know, what they can understand easily, what will take more
effort, what questions they have, etc.
In a reading seminar like this one, presenters must face a somewhat
special problem: everybody listening to the presentation will already
have read the primary paper(s)! So there is no point in simply
regurgitating its contents. Instead, you should aim to...
Spend a few minutes summarizing its key points in your own
words (i.e., bring out what you think is important).
- Briefly review its major parts, to give people an opportunity
to ask questions or introduce their own points for discussion.
- If there are important bits (key definitions, algorithms, or
examples) that were hard to understand or technically tricky, it may
be worth reviewing these in detail.
- Introduce material from supplemental readings that the rest of
the participants have not read.
- Highlight points that seem worth discussing (because they shed
light on broader issues, suggest further work, etc., or because they
seem dubious or involve hidden assumptions)