Why lightning talks work

Last week was the second annual meeting of the Evolutionary Demography Society. It was fabulous. Close to 100 people, over three days, at Stanford. Man there is a lot of good food around there. A friend and I were sitting outside, eating burritos when this dude from a farmers' market booth came over and started giving us samples of gluten-free baked goods. Delicious. Great weather, friendly people. Oh, yes, something else. What was it now? Ah! We talked about a lot of great science. Roughly 70 presentations. Post-reproductive lifespan, evolution of aging, all the topics I like best and every talk on a topic of at least some interest. Fabulous.

Many conferences (but not this one) have what are called concurrent sessions. In one room there might be a series of perhaps 15 minute long talks about matrix models, while in another they are talking about theoretical modelling, and in a third it could be all about field data on hunter gatherers. The benefits of this are allowing more people to give talks in a short period of time, and letting audience members pick and choose which topics they spend their time hearing about. At EvoDemoS, we address these same problems in a very different way. Most presenters give a lightning talk plus poster. The lightning talk is five minutes (plus five for questions), and then after each session there is a break for coffee and posters. But the posters are mostly from the same people who gave the lightning talks, on the same subject. So you get up, give a rapid intro to the work, answer a few questions, and then because there are no concurrent sessions, everyone at the conference knows who you are and what you are working on. If they are interested in it, they come talk to you at your poster. If you didn't bother to print your poster, they already know what you are working on and come talk to you anyway. If they aren't  interested, they don't have to sit through 15 minutes of you talking about it. The frequent and lengthy breaks (made possible by the shortness of the presentations) make it easier to stay alert through the talks, and let us achieve a much higher conversation-to-passive-listening ratio, and it is really the conversations that are the point of the conference for me.

Now the obvious downside is that many speakers are used to having more than five minutes. Some won't come because they can't have more time, use their connections and seniority to push for more time, or simply prepare the same talk they would for a much longer slot and largely ignore the warnings that their time is almost up. One speaker, to remain nameless, was on slide 4 of 26 when the one minute warning came and sped up only slightly. So the moderators need to be a bit firm in some cases. The more senior the speaker, the more likely an overage, in my experience. This is partly a matter of habit, but also that the more senior people often have more work to present. There were a few talks where the theoretical framing got almost completely cut to make time for more methods and results, and that was sometimes problematic. A is consistently greater than B, but what does that tell us? That said, the great majority of the five minute speakers were able to state the question clearly, say a word or two about methods, give a main result (or maybe two) and draw a conclusion or three before inviting us to see the poster. And almost everyone I talked to at the conference, both as speakers and audience, thought it worked well in this context.

For a conference with thousands of people, I'm not sure lightning talks would work. I'd be interested to hear from anyone who has tried it. But for anyone organizing a small conference like ours, I absolutely recommend it.