*“Equations are just the boring part of mathematics. I attempt to see things in terms of geometry.”*

-- Stephen Hawking (b. 1942)

I have collected some of the best academic-style talks that have videos available online. Many of them are from the following websites:

- Google Video
- TED talks
- Microsoft Research Channel [note: only works on Windows Machines]
- Youtube

- Randy Pausch's Last Lecture and on Time Management.
- Luis von Ahn on Human Computation.
- Hans Rosling on Global Statistics.
- Jeff Han on Multi-touch Interface.
- Simon Peyton Jones on How to write a research paper.
- Simon Peyton Jones on How to give an academic talk.

My criteria of good talks (some of them also apply to classroom teaching):

- Visualize
**everything**that is not obvious.

I think visualization is by far the most important aspect of good talks. It not only helps your audience to understand your points, but also allows you to cover much more stuff in a short time. Some topics might be harder to visualize than others, but that's exactly where creativity comes in. For example, in my own teaching experience, I found the Ford-Fulkerson algorithm (for max-flow) quite difficult to teach. My students had trouble with the conventional style of drawing flows as numbers on edges (which is perfect for other graph problems like shortest-paths or spanning trees). But if you can draw edges with various**line widths**which correspond exactly to the amount of flow, that would be much easier to understand. - Intuition, intuition, and intuition!
- Less than three (3) equations! Better no equations at all.

Even for theoretical talks you need to visualize your equations into diagrams...

However, this is the major difference between talks and teaching: in the latter you do need to cover important technical details. - Don't let details overshadow your main theme.

A negative example here is Al Gore's award-winning*An Inconvenient Truth*, which is a great movie but he simply presents way too many individual details without organizing them in a meaningful way. I first watched it in June 2006, right after the conference of NAACL 2006, where I just saw many good talks and more bad ones. So I rank Gore's presentation just like a bad talk (with good contents) in our conferences. There is no doubt that his visualizations are awesome, but even that did not prevent me from falling asleep in the middle. In fact, many people in academia can do a better job of presenting such (scientific) materials. - The three jokes rule: one at the beginning, one in the middle (right before people feel sleepy), and one at the end. This rule is of special importance in job talks.

See also Writing Tips.

Liang Huang Last modified: Wed Apr 11 23:07:31 EDT 2007