Each of you will be responsible for presenting or co-presenting some number of papers (or equivalent) in the seminar, and leading the discussion on them. The quantitative details will be determined based on our enrollment. Here we provide some guidelines for your presentations.
Selecting your paper(s): there are many exciting results that will be tempting to select for your presentation. However, do take care to gauge whether you and the other seminar attendees will have sufficient background to dive right in, or it will be more fruitful to instead cover some more foundational works that are necessary to appreciate the more recent developments. We are not aiming to come out of this seminar having mastered all of the latest and greatest works on Pseudorandomness and High-Dimensional Expanders (which is not realistic for one semester), but rather to have the preparation to start doing so going forward, for example in the summer and during the Simons semester. It is perfectly reasonable to select extracts from a survey or textbook to cover; you do not need to work directly from the research papers. And hopefully we can have some coordination so that later presentations in the semester can build on earlier ones.
Selecting pre-reading: At least one week before your scheduled presentation, you should share with the course staff a selection of pre-reading that you would like everyone to do before your presentation. This could be relevant background material, a set of key definitions, an overview of the proof strategy, or anything else of your choosing. Make sure that the pre-reading is not more than a couple of hours worth of reading.
Reading: Here is some guidance on how to read a paper.
Getting assistance: Do not hesitate to come to Salil and Jake’s office hours to get help working through your reading and deciding what to present, or to work with other seminar attendees in doing so.
The presentation itself: You may use the board or slides, as you prefer. A general guideline is to try to have a top-down and modular presentation, starting with the main results, then an overview of the strategy used to establish the results (e.g. the statements of key lemmas), and then diving into some of the proofs. This way everything is well-motivated, and if you run out of time, everyone will have still come away with a good takeaway picture. Be selective in what you present; it is better to present a few things clearly than many things in a rushed and hard-to-follow manner. Here is some advice from others:
“How to give a technical presentation” by Michael Ernst.
“How to give a great research talk” by Simon Peyton-Jones.
“How to give a good research talk.” by Stephanie Weirich.
“How to Give Bad Talk” by John Ousterhout, Tom Anderson, Dave Patterson (channeled by Mike Dahlin).