I have read in a few places, for example here, that a 20-minute presentation (or indeed, any presentation at a conference) is an advertisement for the paper or research program in question. I had never thought of it this way and have tended to worry that I have not been technical enough. I now see that this is probably a mistake.
I want to incorporate this advice, but I'm still faced the problem of determining the best way to translate a full-length paper into a 20 minute presentation. That is to say, I want to get some idea of the process by which a paper is "converted" into an "advertisement" for presentation. I am wondering about which of the following, for example, is advisable:
- Presenting all of the arguments and objections in the paper, but in a more surface way.
- Limiting the paper to a few key points.
- Leaving certain questions open by spending more time motivating the question and then suggesting or gesturing at the answer that the paper gives (this fits with the idea of an advertisement).
I realise that answers will likely take be "all of the above" or "it depends", so really what I'm interested in are additional suggestions, some idea of priorities, and some way of figuring out given the situation how to make the decision.
P.S. My discipline is philosophy.
Actually, all three of those points apply and in fact what is recommended for a conference presentation.
Additional suggestions would be the following.
In case of an original research, focus on the overview of algorithms/procedures and its nuances with respect to the existing ones (if any) or the novelty involved without jumping right down to the algorithmic details.
For a survey paper, note only the highlights of the survey and studies that provide the most impact to the field in each sub-topic. Avoid jumping to rabbit holes and missing out the opportunity to cover the big picture.
Spend a bit more time on how your findings impact your area of research. Discuss the applications and scope and detail the conclusion. I see may participants miss this one out by just elaborating on the technical aspects of their implementation without conveying what is it applied for and how could it be applied.