The most surprising result to me was that a slight idealization of Google's PageRank algorithm satisfies and is in fact uniquely characterized (at least in terms of the partial order of PageRanks output at the very end) by a nice set of five pretty reasonable axioms. The rest of the talk was some nice incompatibility results, more mundane classification results, and some stuff on personalized ranking systems where each person gets their own perspective of the "social choice" function arising from everyone's votes. Anyway it was a pretty good talk, and reminded me of Kleinberg's class that he taught while he was here. Lots of neat results, lots of open questions.
Later on I went to the generalists' meeting, where I was very saddened to find out that there are legitimate students of linguistics (me being only an armchair one, really) that think deFrancis's book "Visible Speech: The Diverse Oneness of Writing Systems is a good book. Supposedly it is much less stupid about the Chinese writing system in particular than much previous scholarship, and this I can believe, but I disagree from the very core of my intellectual being with some of the other claims it makes about langugage generally, especially about the relationship between written and spoken languages. I struggle to imagine how anyone can confront the linguistic situation and possibilities of deaf culture, or the very real dialect-formation processes in written communication on the internet, and believe that the profound depth of human experience, communication, and linguistic behavior is somehow divinely ordained to be essentially oral and aural, and only secondarily anything else.