I argue that the characterisation of reducability as computability is too weak to be credible. I propose an improved version “intentional computability” and show that it is indeed a stronger criteria, in that there may be specifications for which there exist programs but where there is no systematic way to build such a program from the specification. This undermines the assumption that just because software life would run on a computer that it must be reducible to a computer program as the result of an intentional plan. This, in turn, opens the way to the possiblity that an intentionally irreducible program might arise (e.g. by evolution).
Available in HTML