Last week I was in Germany for a Dagstuhl Seminar on Methods for Modelling Software Systems. It was both quite fun and quite productive. I had the chance to talk to researchers whose work I’ve been studying since I started my Ph.D., and to get feedback from them regarding my own interests.
I guess part of the success of Dagstuhl seminars is that the environment is much more informal and flexible than that of a conference. There’s no official program prior to the start of the seminar. Talks and discussions can be added along the way. At lunch and at dinner, you’re assigned to tables randomly, so you get to hear perspectives from several people instead of joining the same clique all the time. And since Dagstuhl is basically in the middle of nowhere, there’s not much to do at the end of the day other than to visit one of the castle’s rooms and chat with other participants over cheese and wine.
In comparison, the conference format feels somewhat broken: There’s very little time for people to present their stuff, there’s too much going on at once, and the emphasis is on defending and selling your work rather than on exploring it and poking around to improve it. On the other hand, however, a big conference can hold a few thousands of participants, whereas a Dagstuhl seminar, I guess, would break down with over 50 people.