Gödel's Lost Letter writes an interesting post suggesting that complexity classes -- categories of computational problems related by the resources necessary to solve them -- play a similar role in complexity theory as elementary particles in high-energy physics. (Via Chad.) All very fascinating stuff, no doubt. But along the way a much more important issue is raised: when there is a seminar, should we have donuts before, or after?
Back then, Yale computer science used the post-talk-food normal form. That is after the talk donuts were served to the audience and the speaker. Most places then and now use pre-talk-food normal form, but Yale was different. I always wondered why we were different, but it was Yale.
I have to say that Yale is right on this one, and yet almost everyone does it backwards. Some sort of refreshments -- coffee, tea, stale cookies, donuts if you're lucky -- are generally served before a colloquium or seminar, to attract an audience and presumably put people in a good mood. The problem is: we haven't heard the talk yet, so we can't chat about that, and if the audience is big enough we might not even know which person is the speaker. Whereas, if donuts or whatever are served after the talk, not only do you make it more awkward for grad students to scarf some food without sitting through the seminar, but you have offered a very natural topic of conversation -- the substance of the actual talk everyone has just heard. And the resulting conversation will usually be better than the desultory Q&A that follows a typical talk. For one thing, it's just more natural to stand around and chat while sipping coffee or munching a donut than while one person stands at the front of a room and everyone else sits in the crowd (many of whom are restless and ready to scat). For another, students who might be intimidated out of asking a question in front of the whole audience can screw up their courage in a more informal setting. And most importantly, the chances that the actual speaker will get something intellectually useful out of the whole experience are enormously larger if they get to interact with a bunch of people who have just heard their talk. (Not even to mention the abomination of the usual "lunch talk," where the undernourished speaker seminars away in front of a collection of people happily chewing away at their meals.) I'm sure a lot of influential people read this blog. Let's put that power to good use. What do we have to do to change the traditions and make it standard that coffee is served after the talk instead of before?
