Implementation and indeterminacy

摘要

David Chalmers has defended an account of what it is for a physical system to implement a computation. The account appeals to the idea of a "combinatorial-state automaton" or CSA. It is unclear whether Chalmers intends the CSA to be a computational model in the usual sense, or merely a convenient formalism into which instances of other models can be translated. I argue that the CSA is not a computational model in the usual sense because CSAs do not perspicuously represent algorithms, are too powerful both in that they can perform any computation in a single step and in that without so far unspecified restrictions they can "compute" the uncomputable, and are too loosely related to physical implementations.