Physicist, MIT; Author, FAB
I've long considered myself as working at the boundary between physical science and computer science; I now believe that that boundary is a historical accident and does not really exist.
There's a sense in which technological progress has turned it into a tautological statement. It's now possible to store data in atomic nuclei and use electron bonds as logical gates. In such a computer the number of information-bearing degrees of freedom is on the same order as the number of physical ones; it's no longer feasible to account for them independently. This means that computer programs can, and I'd argue must, look more like physical models, including spatial and temporal variables in the density and velocity of information propagation and interaction. That shouldn't be surprising; the canon of computer science emerged a few decades ago to describe the available computing technology, while the canon of physics emerged a few centuries ago to describe the accessible aspects of nature. Computing technology has changed more than nature has; progress in the former is reaching the limits of the latter.
Conversely, it makes less and less sense to define physical theories by the information technology of the last millenium (a pencil and piece of paper); a computational model is every bit as fundamental as one written with calculus. This is seen in frontiers of research in nonlinear dynamics, and quantum field theories, and black hole thermodynamics, that look more and more like massively parallel programming models. However, the organization of research has not yet caught up with this content; many of the pioneers doing the work are in neither Physics nor Computer Science departments, but are scattered around (and off) campus. Rather than trying to distinguish between programming nature and the nature of programming, I think that it makes more sense to recognize not just a technology or theory of information, but a single science.
