Thursday, October 25, 2007

Relevant and Revelant Science

Listening to a program on Radio 4 on the way back home (I think it was Material World) one of the items was about work on predicting the support a candidate in an election can get based on his or her appearance. You extract a few features from their face, such as size of nose, plus others that hopefully capture the notion of a good-looking "commanding" appearance, and see how they correlate with election wins. (The feature extraction is probably the main technical challenge here.)

This is the kind of science that gets a lot of media coverage, but is really good science? It seems like a big majority of the scientific research that garners the press coverage is a kind of collective navel-gazing, studies of ourselves and how we behave. Results that purport to forecast an individual's life achievements based on early life events and his appearance --- it's questionable whether you really want to know some of the stuff that comes out. Any study that correlates sexual behaviour with some other feature or activity of our lives is sure to be hot news.

In support of this kind of work, it certainly looks more relevant to people's everyday lives. In contrast to my own research, for example. I wish people were as interested in the stars as they are in their own backyards, but I guess I can't make that happen.

No comments: