Monday, March 16, 2009

Interactive Computation: Computing without Boundaries


As a bit of light relief from the entrenched world views and ill-tempered politics of the evolution/ID debate I read this paper after being tipped off by my son. It’s called “The Interactive Nature of Computing: Refuting the Strong Church-Turing Thesis” by Dina Goldin and Peter Wegner. Its thesis is that in the contemporary environment of ubiquitous interactive computing Turing's mathematical function model of computation, namely “Input=>Process=>Output”, makes heavy weather of today’s realities and needs upgrading. They propose a three tape machine model that supplements the input/output tape(s) of standard Turing computation with another tape that represents that the results of the machines output interacting with the surrounding world.

My first thoughts here were that it all depends on where one is drawing the boundaries. The closed systems of deterministic physics conform to the non-interactive Turing model in that the triplet “Input=>process=>Output” maps to “initial-conditions=>process=>outcome”. Even within a deterministic system one can draw boundaries in that system in such a way that subsystems become, in effect, computing machines that interact with their surroundings after the manner of Goldin and Wegner’s modified model. Hence, in one sense the old functional model of computing seems, at first sight, to include the new model.

However, Goldin and Wegner’s model really comes into its own when one realizes that it is not humanly possible to draw boundaries round total systems: the real systems we study are in the final analysis open systems and thus Goldin and Wegner’s model of computation is the more natural one and a better approximation to reality; that’s my current opinion anyway. All in all I think this is an area research to keep an eye on.

No comments: