The FiO conference is less than a month away - time to start thinking about putting a talk together...By some miracle of science, I already have all of the results for the talk I will be giving in Rochester (in contrast to my usual state of staying up all night the week before the conference desperately trying to get good results). So now I can spend my time making the presentation better.
Most researchers would agree that it's harder to squash months of work into 12 min than it is to give an hour-long seminar on the same topic. Thus, preparing a FiO talk is an exercise in data compression - and it is always a lossy compression scheme. Some strategies I've seen implemented are:
1) downsampling - simply skipping most of the details and giving only a broad overview of the work, but sometimes key pieces end up missing.
2) no compression - 10 point font and 30 equations per slide, such that it would take hours to get through, but the speaker skips over most of it so quickly that no one is able to follow and they run out of time halfway through.
3) no compression - 10 point font and 30 equations per slide, such that it would take hours to get through, but the speaker is famous and no wants to tell them to stop talking. Perhaps the OSA should consider trying the solution developed by the Ig Nobel awards for keeping the conference running on time...
The trick is to find the right basis set of information that can convey the overall picture, without glossing over important details, restating ideas or being redundant redundant. Unfortunately, the optimal transmission of research to the receiver (audience) is both content-dependent and audience-dependent, and optimality criteria are often heuristic. Just looking at the overview of topics at FiO, one can assume the audience will be quite broad, so speakers, please keep it simple so that I can follow!
No comments:
Post a Comment