yet another reason I love being derailed in live talks: don't need to prepare slides which don't get shown
Conversation
Replying to
It's interesting because it takes a different deck structure to support this, but it's really good when it works. Running out of time in a talk that wasn't designed for that is usually pretty awkward.
1
1
I recently gave a talk where we skipped the last third of the slides because we ran out of time due to interesting interruptions. And it was totally fine... but only because that last third was basically independent of the rest.
1
1
Replying to
Yes, interesting point... there has to be some amount of "front loading" for this to work well. Won't work if the main results require a lot of buildup to understand.
2
1
Replying to
Or you have to "build up backwards" the way that a lot of TCS papers tend to (and the way that a lot of other papers tend not to). Start with some baby version of the result early on and then iteratively refine it throughout the talk.
(This convention is probably my single favourite thing about TCS papers and I desperately wish that other fields would adopt it.)
1
1
Replying to
Good point, I hadn't considered this advantage of baby-results (which are good even for non-interactive talks).
1