An erudite, intense discussion is taking place in the comments, over at the Computational Complexity blog. At issue: Is complexity theory (a discipline which studies "the power and limitations of efficient computation") a form of math, or physics? You can use it to describe physical processes, but does that make it a scientific discipline? The debates will make your head spin.
Typical comments: "Complexity theory is math. Saying that it is science because we use models of computation that attempt to model the real world is like saying that Riemannian geometry is science because it models the real world (assuming our current theories of relativity)." "Didn't Wittgenstein (for example) make great progress by asking, not what philosophy is, but what do philosophers do? And similarly Turing, by asking not what computing is, but what computers do? And Gauss by asking, not what space is, but what surveyors do? And Mac Lane, by asking not what mathematics is, but what it is that fundamental mathematics does?" Weirdly, the first commenter in the thread forgot to write "FIRST!" in all block caps.
Also fascinating: the stuff about "selective sweeps," in which advantageous genetic characteristics spread throughout a population, and bring with them new characteristics that have no apparent advantage at all. [Computational Complexity]