Wednesday, May 07, 2008

Any Questions?

A speaker in a seminar talk loves to get questions during the talk for this means that at least one person is trying to follow the talk. A talk with no questions means everyone is either completely following the talk or is completely lost, most likely the latter.

Each question though involves three parties: the questioner, the speaker and the rest of the audience. A good talk has a certain rhythm and questions can disturb that rhythm. So how does the audience feel about the questions? Depends on the question.

  1. Questions that clarify the model or some aspect of the proof. We need these questions to properly follow the talk. When others ask these questions, I learn that I really hadn't understood the model when I had thought I had.
  2. Questions that argue against the model or results. Usually entertaing but can often degenerate into a long argument. The host needs to become a moderator and has to give one of those one-time nerd jokes that have become standard lexicon: "Take this discussion off-line."
  3. Questions that point out mistakes. Usually annoying and serves no purpose unless, of course, it takes down the whole proof.
  4. Questions that prove how smart the questioner is. The most annoying. I cringe whenever I hear a question starting with the word "So".
At the end of the talk the questions usually suggest various extensions to the work that can often go on forever. Most of the audience just wants to escape but is too polite to leave. The host again needs to end the discussion. Having food in another room to continue the discussions in can help immensely.


  1. Why are you annoyed at being corrected? to me that just says that the contents of your talk don't really matter to you. I guess I don't get the point of a talk then. If you don't care about the talk being correct, then you can't want people to come out of it knowing that it is correct. To me that only leaves a few options for the purpose of the talk, to get people to blindly believe your results, to get people interested enough in your results that they actually go read the paper that they wouldn't have otherwise, or to provide a roadmap to understanding the paper that potential readers will need to actually be able to digest it. Unfortunately all of those have kind of negative implications about academia.

    I'll admit I'm a corrector. When I'm being given an argument as to why something is true I like to be able to follow it and when the argument is done being presented I want to know that it was a sound argument. So when I see something that doesn't click I want to make sure that it is a typo and not subtlety of the skipped work that I just did not get.

    And if you feel the same way about corrections in lectures to students, I hope you aren't allowed to teach undergrads because so many of those kids are just blindly copying down what you write on the board, so the corrections are very important to make sure that the blind copiers are getting good notes.

  2. Mistakes in a talk can be very confusing, because they may make you think that the non-mistakes are actually mistakes.

    Whenever I detect something that I believe is a mistake I ask. Whenever I detect something I know is a mistake I don't ask.

    `Mistake' for me is a statement that seems to not fit with the rest of the talk.

  3. Lance, the word 'entertaining' is spelled incorrectly. :-)

  4. Perhaps the author is just being somewhat provocative to solicit comments and increase readership.... I'll bite:

    Because I'm usually slower to understand things, I get tripped up easily by mistakes. So, I find "questions that point out mistakes" quite helpful.

    Questions that "prove how smart the questioner is" can be enlightening for me as well. During such "questions," the asker may bring out interesting connections, alternative perspectives... etc. All of which I find helpful.

    I do get severely annoyed by arrogant showoffs, but there's something to learn from them as well.

  5. I think questions of type 4 come up mostly after talks. This is a good thing. During a talk they can be disruptive and seriously annoy, but afterwards I think they can quite interesting.

    Also, sometimes when a speaker is going fast (for me anyway) I ask questions where my real goal is to rephrase what is going on to make sure I understand, and to get a moment to think and keep up. I'm also more likely to do this with a small audience -- I wouldn't hate to do so in a talk to 8 people, but would never do it with an audience of 80.

  6. It should be noted that all of this depends very much on the type of talk, the type of audience, the size of the audience, and even the size of the room and whether it is a whiteboard talk or a powerpoint talk.

    I am usually much happier giving a talk to a small audience, where the talk can be much more interactive and I enjoy the questions, even those that point out mistakes.

    On the other hand, having someone point out a mistake from the back of a large lecture hall during a job talk to an audience of 50 people is less enjoyable.