21 April 2010
During the last fifteen years or so, I've given a lot of keynote talks. I've always found these kinds of talks rather awkward. If you give a talk during a session at a conference, you pick one subject to talk about. You know that there's multiple tracks, so if people come to your talk that implies a certain amount of interest in your topic. But with keynotes you're speaking to the whole conference, so I feel I can't make my talk too tightly focused. I may like talking about the intricacies of modeling temporal events, but I feel that that's too narrow a topic for a broad ranged audience
So for the last decade or so, I've gone in the direction of ExtemporarySpeaking, choosing a subject late on that I think can appeal to a broad range of an audience. Extemporary speaking also allowed me to ditch the horrors of slide decks, which I felt resulted in a more fluent talk. It didn't please everyone, from time to time people complained of lack of preparation, but my sense is that most people found it worked.
I blame my colleague Neal Ford for jolting me into a different way of doing things. Neal does use slides, but he works hard to make them work well to support what he's saying. He designs a visual channel to support his spoken audio channel, rather than just a progression of bullet points or cheesy stock photos. (Stock photos are the bullet points of the twenty-first century.) I liked his approach, so started to figure out how to do what I do with any good idea - steal it. As I explored using slides as a visual channel, I quite enjoyed the way it jolted my thinking and pushed me into trying some things that I thought would be useful.
However going back to using slides has a downside. Extemporary speaking gives me a lot of flexibility to choose my talks at the last moment. My strategy for keynotes is to give a vague title and then decide what to talk about depending on what I think the audience would find most useful to hear, and what I have most energy about saying. With slides you are more fixed to a pre-defined agenda.
So starting last autumn with a talk in Stockholm, I tried a slightly different twist. Rather than give a single one hour talk, I gave a suite of three twenty-minute talklets. This way I could build a bunch of twenty-minute slide-backed talklets, but leave it late before I decided which combination to use on a given day. Furthermore I could mix slide-backed talklets with extemporary talklets. Not just does this allow me to keep some flexibility, it also allows me to hit a narrow subject with a talklet, since even if some people find one topic uninteresting, they may like one of the others.
Part of the reason I focused on twenty minutes is since that length seems to be a common length for TED talks. Watching them, it struck me that this was enough time to get an idea out without sagging in the way that an hour-long often does. Shorter lightening talks are very popular these days, but I think those are too short for one speaker to fill an hour with. Keeping to twenty minutes is certainly a challenge for someone as loquacious as me, but it strikes me as a worthwhile discipline.
I've done this style about a dozen times now, and I've made it my default style of speaking. I'm gradually building up my stock of slide-backed talklets, so I have a wide choice of things to talk about. If you see me give a keynote-style talk in the next year or so, this is likely to be what you'll get.