Thursday, November 12, 2009

Seminar skills

I gave a seminar today, entitled "An introduction to Kolmogorov complexity." The abstract I provided:
This week, I'll cover some introductory Kolmogorov complexity (including how to pronounce it!), definitions and applications to complexity theory,
including the relation of Kolmogorov complexity to the halting problem, and defining resource-bounded computational hierarchies from Kolmogorov
complexity. No background knowledge of Kolmogorov complexity is required; this seminar will be self-contained.
General consensus? It went very well. There was a lot of audience participation. Perhaps too much, since it got a bit derailed with people trying to explain each others' questions and answer them. Afterwards, more senior grad students offered me this advice:
  • Never admit you are wrong. Never erase and edit what you've written on the board. I wrote one thing wrong and then 10 minutes were wasted fiddling with it. Relatedly,
  • Don't answer all the questions. Make sure everyone has a basic understanding. If the question is about details that won't improve a basic understanding, postpone it until after the seminar is over.
  • Proof by assertion. Related to the above two points. If there are too many details, or you don't quite remember how to prove it, or it's too hard, or whatever, then just say, "Obviously, ..." and move on before anyone derails you.
  • Don't let audience members talk amongst themselves.
Obviously some of these should be applied judiciously. I'll have to work on these points, now that I seem to have the basics down.

This post's theme word: expatiate, "to speak or write at length" or "to move about freely."

No comments: