The point I was trying to make in my previous post was that even though digital
computers dissipate (the microprocessor chip gets hot) that dissipation doesn't
entail the software flow. The flow of the software is independent of whatever
is happening thermodynamically in the hardware. You can accurately trace the
software flow of a program without ever considering the energy state (or
dissipation) of the hardware. This is why "computer science" can be separated
from "electrical engineering". Although a computational device needs something
physical on which to run, one can understand and predict the software flow
without understanding anything about the physical device. (Note, that to make a
computer, you do have to understand the physical device...this sugests that the
software can not understand or make the hardware on which it runs).
Universal computers have the advantage that the same hardware can be made to do
many different things. Its all program dependent. The downside (as Onar points
out) is that universal computers (digital computers) can not perform any self-
organization that is dependent on the thermodynamics of the hardware that
implements the instructions. Therefor, I agree with Onar that a universal
digital computer can not be a dissipative structure (since, for a dissipative
structure, the self-organization is entailed by its own thermodynamics). And if
dissipative structures are important in the definition (or realization) of a
self-producing (or autopoietic system) then we will need to consider things
other than digital computers as models for autopoiesis.
(Don said the same thing in different words in his post).
Jeff Prideaux