On creating a "companion web site" for talks

By Charles Sutton on December 29, 2016

My last research talk, I tried an experiment during the talk. I created a “companion web site” for the talk, which is an online handout including slides, code, links to papers, and a bullet-point abstract. Here’s my first try at an online handout.

The inspiration was Tufte’s quixotic diatribe against Powerpoint, which recommends written handouts instead of slides. The web site is an attempt to adapt his idea to an all-open all-Internet all-the-time age. Seeing as I have no intention of printing out several hundred paper copies of a handout and carrying them thousands of miles to a research meeting, and that most people have laptops anyway.

Why the handout? Lectures are limited because speech is linear. If we are conversing, when I haven’t understood what you have said, I can ask you to go back. When you lecture, I cannot do that. If I stop a moment to think more deeply about something that you have said, I run the risk of missing your next point.

The handout is designed to help with this. By giving you a written summary, and also my slides, you can “go back” and re-read a point that you may have missed. A second role for the online handout is to help you find more details after the talk, to find exactly which papers, which code, and so on, I was referring to.

If the handout really helps, then it raises an even more interesting question. Why the slides? Tufte argues vehemently against slides as a replacement for written text. Slides have low resolution, which encourages poor statistical graphics, and encourages hiding details which are vital to assessing an argument in science and engineering. Also, most arguments do not fit into bullet points, because bullet points can convey only linear relationships and nesting. The narrative that you are trying to convey in your talk is likely to be more complex.

In research conferences, we already have a handout for each of our talks, the research paper. So maybe we be presenting scientific talks with no slides at all? If not, why not? Sometimes at a conference, I’ll skim the paper during the talk if I am starting to get confused. Sometimes that helps. But papers are information dense, with long sentences full of qualifying statements designed to forestall potential reviewer criticisms. It can be hard to read dense text while also remaining aware about what the speaker is currently saying. Sometimes I think this a blind spot of Tufte’s, that he leans too heavily on information density as a measure of the quality of a display.

We can think of the sequence [ slides, audio of talk, online handout, paper ] as attempting to convey the same information at different levels of density. This allows the reader/viewer to choose the appropriate level. But slides and audio are both low density. Do we really need two low density ways of conveying the information?

Another way of asking the question is: What are slides good for, when they are good? Tufte does seem to accept that good slides are possible. He admits that perhaps the top 10% of presenters aren’t harmed by Powerpoint’s cognitive style, as their own presenting style is strongly enough developed that they impose it on the software, affordances be damned.

In the past couple of years, I’ve been experimenting with different slide designs. I may even starting to understand what principles I am aiming at in my slide designs. But I think that’s a topic for another day.