Yesterday was Theory Day at U. Maryland (organized by Samir Khuller and Azarakhsh Malekian). I went to the first half and the talks were universally excellent. Though I disagree with Bill's enthusiasm for the idea of the theme. Or rather, I abstractly agree with him, but was not excited about this particular theme. Now if the theme had been Computational Geometry...
Aside from being amused by Mohammad Mahdian from Yahoo! Research's use of a Google search screenshot, the talks prompted me to think about models and our "reasonable" assumptions that we make to simplify the world so that we can reason about it. For example, Mohammad discussed research on sponsor search ads that appear in a column. As part of his model, he assumed that users go through these ads in order from top to bottom. This is, of course, not universally true. It may not even be usually true (he mentioned eye scan research - I'm not sure if any of it was on this topic). And yet it's certainly a reasonable assumption in order to create a workable model. At least, it's reasonable from the theoretical point of view. Are the assumptions we make reasonable from the practical point of view? And how do we determine whether they're reasonable? It was good to be reminded of these issues as I work on some models of my own.
Two drivers hit 10-year-old girl in Seattle, speed away
13 minutes ago