How to Fix Seminars
I’m sick of academic seminars. Even if the presenter has a good paper and speaks well, the audience ruins things. How? Two minutes into any talk, the fruitless interruptions begin. Half the time, they’re premature quality control: “Are you going to deal with ability bias?” “Uh, yes. On the next slide.” The other half the time, they’re bizarre pet peeves. “How does this relate to sequential equilibrium?” “Uh, it doesn’t. It’s an empirical paper.” “Yes, but that reminds me of something Rudi Dornbusch once said. Now I’m going to talk for four minutes.”
Once fruitless interruptions begin, there’s no such thing as winning. There are only different degrees of losing. Even if you limit each question to a single minute of time, your ideas lose their flow. A well-planned talk handles complexities in logical order, allowing relative novices to follow your reasoning. Premature quality control smashes that order to pieces, leaving novices confused and experts bored. And after three pet peeve questions, the audience struggles to remember the topic of the talk – never mind its thesis.
Can’t speakers deflect the pointless interruptions to make the talk train run on time? It’s tough. You can’t just say, “Please, only good questions.” And once you take one silly question, every other squeaky wheel feels entitled to put in his two cents. “Hey, my question’s clearly better than the last guy’s!” As a speaker, then, you have to choose between offending a fifth of the audience or boring the entirety. Professionally speaking, the former is the greater danger.
Still, I come to fix seminars, not to abolish them. My colleague Dan Klein inspired the following alternate rules:
1. Split the talk into two parts. Part 1 is the first two-thirds of the allotted time. Part 2 is the last third of the allotted time.
2. During Part 1, the audience may not ask any questions. No exceptions.
3. However, the speaker retains the option to ask the audience questions during Part 1. If the speaker sees a lot of confused faces, he can query, “Are you familiar with the efficiency case for Pigovian taxation?” and adjust his presentation accordingly.
4. The speaker scrupulously ends Part 1 on time, then turns the rest of the talk over for questions.
Step 4 has two main advantages over the standard method.
First, it’s easier for the speaker to filter out bad questions. Since there is a dedicated question period, multiple people will normally raise their hands. If an audience member asks bad questions, the speaker can call on other people first. If an audience member asks too many questions, the speaker can gently say, “Someone who hasn’t already asked a question…”
Second, the question pool will generally be better. Once the audience understands your point, they can ask questions about the overall thesis rather than Slide 3. Quality control questions now serve their legitimate function: If you missed your opportunity to address a tricky point during Part 1, audience members can correct your oversight in Part 2. The pet peeves still make a spectacle of themselves, but at least the audience walks away knowing your thesis.
A few months ago, I saw Dan Klein use a similar format, though he made Part 1 shorter and Part 2 longer. This Wednesday, I tried it at the Public Choice Seminar. It could easily have been the best academic seminar I’ve ever done. The approach makes so much sense, it makes me wonder if the standard approach is ever better.
My best guess: The standard approach might be better for graduate students. Suppose you’re a public speaking novice. Interruptions are an opportunity to improve your performance in real time. Toastmasters clubs, for example, appoint an “Ah-Counter” to clap whenever the speaker says “ah” or “um.” Alternately, suppose you know less about your topic than several experts in the audience. Their interruptions are an opportunity to prevent you from derailing your own talk with errors and irrelevancies. The background assumption here, though, is that the speaker and the audience both constructively treat the seminar as training. Shredding graduate students helps no one.
Comments on my seminar system are good, but field experiments are better. Have I piqued your interest? Try my approach and tell me how it works.