Sunday, October 03, 2004

Are we too nice?

Steve Smale talked about his experiences in the Economics Theory Workshop at Chicago, particularly the aggressive questioning. I didn't attend his talk but I did go to a few of the econ theory seminars years ago and it forms an interesting contrast to the CS theory talks which have few usually technical questions followed by polite applause.

The econ theory seminar took place in a medium-size conference room with a long table. Graduate students sat in chairs along the walls. The speaker was at one end of the table and econ professors, usually including multiple Nobel prize winners, around the rest of the table. A copy of the paper was sent with the talk announcement and almost from the first slide the faculty aggressively attacked the speaker with questions about the validity of the model or the importance of the results. (Remember this was the theory seminar, imagine the questions at the political economics seminar). At the end of the seminar time, the talk ended and everyone left the room. No applause.

I don't recommend that we follow this model in theoretical computer science. However we usually go to the other extreme and (outside of crypto) rarely ask negative questions in a seminar. Typically the only negative feedback we get in our field is from anonymous referees and reviewers. If we were forced to defend our research in an interactive setting, we would establish a better understanding of the importance of the models and results of our own research.


  1. The thing I find odd about theoretical computer science is dual applause. First, a round of applause when the speaker finishes, and then after the speaker answers questions. (And sometimes, even if there are no questions, there is still a second round of applause.)

  2. Many STOC/FOCS papers I've seen entail stating a problem which has an obvious solution in O(n^d) time and giving an algorithm with worst-case complexity O(n^(d-\epsilon). Why get nasty over that? It's nice. It's supposed to be nice.

  3. Yes, the "is this research worth it?" type of questions are rarely brought up during seminars and this can leave the non-specialist in the dark as to any deficiencies in the model.

    I wonder why CS theory evolved like that, perhaps math is similar?

    My guess is that the assumptions are fairly upfront and there is little pretense that anything corresponds to "reality." Hence no big deal if the model is unrealistic as long as it produces cool results.

    The situation is obviously different in economics and (I would guess) in more applied areas of computer science.

  4. I gave some talks on information retrieval and natural language processing back in the day, and I can attest that you get far more hostile questions. There's always something to be hostile about if the speaker did an experiment.

  5. I indeed think that theory cs people are too nice and that this explains (partly) why there are so many bad talks in the field. It is very helpful when your
    colleagues explain to you why they don't like your
    talk, results or topic. Much more so than the disingenuous "I liked your talk... looks interesting."

  6. Yes, we're too nice, and it makes giving talks a lot less fun. In fact, I find that CS (theory) people are very likely to discourage audience members from asking any questions except the most mundane (i.e. what does X stand for again?) Everything else is met with "Maybe we can discuss that issue after the talk"--an offer which frequently goes unclaimed. We should be far more discerning, far more vocal, and we should remember that since we're in a theoretical field, our number one responsibility is critical self-evaluation.

    If you enjoy a lively, engaged, and somtimes combatitive audience, consider giving more talks at Hebrew university. That's an exciting seminar.


  7. Reffering also to cs specialization, i was thinking about the following "talk" mechanism:

    25 minutes before the talk the speaker will present her/his first slides which normaly starts the talk. then comes the usual talk.

    so the first part will serve as introduction to the area, and the speaker just needs to repeat her/his say 5 slides - no any other prepartions before the "normal" talk.