Advice on giving talks

1 As a graduate student, pounce on every opportunity to give a talk.

  • You get feedback on your work.
  • You get practice giving talks.
  • Almost no one is comfortable giving talks right away. Almost everyone needs lots of “flight-time” in order to settle in. You want to be an experienced talk-giver by the time you give your job talk.

2 Talk content

  • “Clearly lay out a philosophical problem that grabs the reader, and find pure joy in addressing it.” – Delia Graff Fara
  • “Make sure you start your talk with a puzzle or clearly stated problem to get people thinking.” – Benjamin Morison
  • Thought experiment: if you just had 20 minutes to informally explain this paper to a friend while walking around, how would you explain it? That could be the backbone of your talk.
  • “You need to have a clear and pretty straightforward idea at the center of the paper. Keep it simple, and make sure that people don’t forget it. Your audience probably won’t remember complicated technical ideas, so don’t get lost in the details.” – Dan Garber
  • Trust the one main point enough to highlight it, and to let your excitement about it shine through. Convey to the audience what gets you enthusiastic about this project.
  • You can’t get everyone to like your talk or be convinced by it, often due to circumstances out of your control. But it is to a great degree under your control whether they understand what you are trying to accomplish, and understand why you are excited about it. Prioritize influencing what is in your control.
  • “Treat the talk like you’re teaching your work to an audience of smart people that doesn’t already know much about the topic (rather than like you’re defending yourself or like you’re talking about your view to people who’ve already read all the stuff you’ve read).” – Lara Buchak

3 Learn how to “talk through” a presentation rather than reading it.

  • This presentation style is more engaging.
  • Even if you later end up reading a paper out loud, it will improve your presentation to go “off script” several times during the presentation, but you can only do this effectively if you’ve practiced.
  • Think of the talk and Q&A together as an exercise in teaching—your job is to most effectively get your audience to understand your main claim and your argument for it.

4 Use a handout

  • The handout explicitly and prominently displays your main claim and your argument for it.
  • If you can’t give a one sentence statement of your main claim, and an argument for that claim that has no more than about 5 premises, then you need to think more about your paper.
  • The handout can also contain important subsidiary claims, or explications of other arguments you are discussing. But these can never take the place of crisp statements of your claim and your main argument.
  • The talk should follow the structure of the handout. Think of the talk itself as explaining and fleshing out the handout.

5 Day of talk preparation

  • Celebrities have handlers who make all sorts of elaborate arrangements to support the celebrity before a performance. Before giving a talk, make these arrangements for yourself.
  • Remove as much uncertainty as possible:
    • Check in advance that the room is set up how you want it to be
    • Is there a blackboard or whiteboard?
    • Is there good chalk or working markers? (I bring my own.)
    • Will the projector be there on time if you need one?
    • Will it work when you plug the relevant computer in to display your slides? (The answer is always “no”. So test/debug in advance whenever possible.)
    • Show up very early to the talk so that you can set yourself up calmly, and so that you can chat a little with people as they filter in.
    • Talk with some audience members, get a sense of their backgrounds and what might interest them in particular.
    • The people you talk with at the beginning can be natural people to make eye contact with during the talk
  • Nerves
    • Giving lots of talks helps, but frankly I still get a bit nervous before talks
    • Re-frame your own internal feelings of nervousness as positive-valenced excitement. Think to yourself as you notice, eg, elevated heart rate “I’m so excited to be giving this talk”

6 Practice

  • The real secret to giving a smooth talk is practicing it many times before you give it.
  • Practicing your talk means: preparing your handout or slides, going to an empty room that simulates the talk room as closely as possible (best of all is the actual talk room), starting a stopwatch and giving the talk to the empty room.
  • At first, this is painful.
  • If you are like me, it will be terrible in the beginning: you will stumble over words, not be clear, talk yourself into dead ends, and take too long. This is to be expected.
  • Keep practicing, starting over from the beginning or the beginning of the section whenever you get stuck.
  • As you practice the talk, you will find that it gets easier to say what you want, and that the time it takes to get through the materials goes dramatically down (typically by a factor of 2).
  • You can’t tell how long it will take you to talk something through until you’ve done it. If you don’t time your practice you will almost certainly get into a time crunch.

7 Talk structure and timing

  • Your first sentence is a chance (perhaps your only chance) to have the audience’s full attention. Make it count:
    • Just enter your talk and say your first sentence, without any preamble.
    • Make it crisp and designed to draw the audience’s attention.
    • Memorize your first couple of sentences cold (helps overcome nervousness at the beginning).
  • Be able to see the elapsed time as you are giving your talk.
  • If for some reason you seem to be running over:
    • Don’t be flustered
    • No one cares if you present what you intended, they just want a good presentation that does not go over, so:
    • Just calmly remove part of the remaining talk so that you end comfortably on time. If the things removed are on your handout you can say something like “We’ll skip section 6 on the handout for reasons of time, though I’d be happy to talk about it in the question period.”
    • “It can be helpful to make notes in advance [on your copy of your handout] of parts that can be cut if running over time (or digressions that can be added in the much less likely reverse case).” (adapted from advice due to John Burgess)
  • Running overtime (or even letting people worry that you may run overtime) during either the main presentation or the question period is leaving goodwill (and attention) on the table.
    • I think it is ideal to end a few minutes early. It is like giving your audience a surprise gift of a few minutes of time.
  • Have a pre-rehearsed, punchy final sentence, just as you have a pre-rehearsed first sentence. Saying it clearly and then decisively stopping is the equivalent of “sticking a landing” in gymnastics.

8 Breaks

  • Consider taking at least one break in the middle of the talk in order to maximize audience attention.
    • No one can really pay attention for 50 minutes straight.
    • The break wins you a chance to have another “first sentence” that will get the audience’s full attention.
  • Types of breaks:
    • Actual break: e.g., a 3-minute break for stretching.
    • Announce what you are about to do and tell the audience what level of importance it has (and maybe how much brainpower it will need). Examples:
      • “In a moment I’m going to give a high-level summary of the talk so far.”
      • “Next I’m going to give the crucial insight of the talk. This is the most important part—if you are going to give your maximum attention to one thing all day, this should be it.”
      • “Next I’m going to give the idea of how to prove such-and-such theorem. If you don’t need the details, feel free to turn off your brain until the start of the next segment. [proof here]. OK, we’re done with the proof, and welcome back to everyone who turned off your brain before.”
    • Pause. Perhaps you could say after a tricky point—“that was a lot to absorb. Let me pause for a moment to let it sink in.” This would be a good time to have a sip of water.
    • Break character to notice and note something that is happening in reality. Occasionally you will have the opportunity to have a “save the cat” moment by actually helping someone in the room with something.
  • Variation within a talk creates a kind of break. Your talk should have internal variation, for the same reason that a “set” of songs by a band typically contains a variety of tempos or moods. Although most of the talk can be delivered at your baseline, it is helpful to occasionally vary: volume, tone, pace, whether you stand or sit, modality (slides vs blackboard vs just talking).

9 Manner

  • Dress a bit more formally/fashionably than your audience.
  • If you are highly confident, friendly, and outgoing, be yourself. If not, then play the role of a more confident, outgoing, friendly version of yourself.
  • Set your physical boundaries wide in the initial part of your talk
  • Be normal (this takes practice).

10 Q&A general principles

  • A good Q&A is not primarily about you proving that you have lots of knowledge. It is about you using your detailed knowledge to make some good philosophy happen during the Q&A.
  • You are not required to rigidly defend your position. Instead treat the Q&A as a discussion you are leading in which you are helping everyone to fruitfully talk about your ideas and probe what criticisms they might be subject to and how those criticisms might be addressed.
  • Don’t think in terms of convincing the audience members that you are smart. Instead convince them that they are smart.
    • Appreciate the audience’s points, when they make good points.
    • Say a question back and sharpen it up, if it is unclear—or if it is particularly excellent.

Delivering an answer

  • Questions come from particular people. But your answers should be to everyone. Your goal is to draw everyone into the discussion.
  • When giving an answer, give your answer and then stop. It is not your job to keep talking until someone interrupts. Indeed, doing so impedes the flow of the discussion and obscures the point you have just made.

11 If you get stuck while answering a question

  • If you get stuck or confused in answering a question, one good option is to confess. Just say, “I’m stuck here.” Or “I lost my train of thought. Let’s get back on track by …”.
  • If you can’t answer an objection, be honest about it. Say “that’s a good objection, and I don’t have an answer for it right now.” Maybe use the “make things worse” trick—explain why you think the objection is an especially good and tough one (if that is true), and why a promising reply won’t in the end answer the objection. In the worst case, you’ve at least gotten a really good new objection to your paper—such objections are worth their weight in gold!
  • It doesn’t much matter if you don’t have an answer to every objection.
  • What matters more is that you understand what is going on, understand the objections that are raised, and try to get to the bottom of things, rather than throwing up smokescreens in order to avoid a good point.

12 Managing the discussion

  • If you are fielding your own questions, it is your job to manage the discussion in a way that makes it overall most fruitful. This will not be accomplished if you do not constrain the length of each exchange.
  • The normal expectation is that each exchange should be limited to one question and one answer (with the author or commentator being given the last word), but it sometimes turns out that a questioner wishes to ask a follow-up question. It is your job to allow follow-ups when fruitful and time and local norms permit, but also to politely but firmly end such exchanges and move on to the next question when you judge that the time is right.
  • When is the time right? Use your philosophical judgment about when you think the meat of an exchange is over. You aren’t required to give each question the exact same amount of time, and normally no single exchange should go on for more than a few minutes.
  • When in doubt, err on the side of ending question exchanges.
  • You can always add those exchanges to the end of the list, so that they can be returned to at the end of the session if there is time.
  • If someone is rambling, it is generally ok to do a “soft” cut-in: You can gently interrupt and say, “Since there a bunch of other questions, let’s move on now and come back to this if there is time at the end.”

13 Questioners who are difficult to handle:

  • Doesn’t have any idea what is going on, asks a long and irrelevant or incomprehensible question
    • Can be OK to gently nudge toward end of question by semi-interrupting as question drags on (“OK, I think I see what you are asking here. What I want to say in reply is…”), or even in extreme circumstances to cut off the question (“Let me stop you there so that we can make sure to get to everyone in the queue…”).
    • Try to make the question into a good question, or at least a question that leads to something worth talking about.
    • What is difficult about this situation is that the questioner might well be a crank who shows up to all the talks and annoys the whole audience by taking up too much time. Or they might be the chair of the department. Use your best judgment!
  • Super-aggressive questioner who asks a question that amounts to “You are an idiot and your idea is garbage”.
    • Don’t be flustered. This sometimes happens and most of the time the cause is not you but rather something that is going on with the questioner. It is fine to pause and gather yourself before answering.
    • If there is any kind of real question under there, you can simply address the real question (perhaps restating it first by saying “You are objecting to my claim that such-and-such. That is a fair worry. In reply I’d point to …”) and ignore the trash-talk/vitriol.
    • If there is no real question, then you might simply say “Thanks for the comments,” and move on to the next question.
    • Alternatively, you could say “Your comment nicely brings up such-and-such issue…”, where the relevant issue is just something in the vicinity that you wanted to say anyway. Here you are in effect acting like a politician dealing with a heckler at a town-hall meeting.