Posted by
glen ep ropella on
URL: http://friam.383.s1.nabble.com/On-Rosen-On-Donder-On-Blitzen-tp526152p526159.html
Ken Lloyd wrote:
> I think there is another way to prepare for catastrophic change, besides
> agility - complexification. C.f.Reuben McDaniels or Ken Stanley.
>
> This allows for failure in one or more set of features to be ameliorated by
> other sets, and resembles trends in the evolution of species.
I agree. But, it's not that significant of a statement, really. A
complex system will be amenable to the ascription (or attribution) of
sub-systems, many of which will be partially redundant or where one (set
of) sub-system(s) will be satisficing when another (set of)
sub-system(s) fails.
But this is just another way of saying that a robust system will be
embedded in its context.
Going back to Rosen, his definition of "mechanism" and discussion of
"closure to efficient cause" are rhetorical devices to talk explicitly
about how embedded a specified unit is in the whole context. He
identified one type of "closure" that is necessary for his definition of
"complexity", that is the closure of the agency by which an entity is
created and maintained. But this sort of closure would only be robust
to certain types of context change and not other types.
These closures would constitute the actual ("property" vs. "attribute")
boundaries between sub-systems in any system. If you break any closure,
a larger "entailment structure" would provide the safety net and if,
within that larger structure, there was pressure and capability to
restore the previous closure, it would be restored with other materials.
(Under the assummption that efficient cause is independent of material
cause; so the same closure to efficient cause could be restored using
different materials.)
What Marcus and I were talking about when using the word "physics" or
"forcing structures" is that larger entailment structure within which
any closure might be restored. So, a system is "complex" when it has at
least the one type of closure. But a system could be more and more
complex if the closures are embedded inside larger closures.
Embeddedness (and by extension robustness) could then be measured by how
easily it is to realize a closure within an entailment structure. How
close to the larger context is the entity? The closer the entity -- a
particular type or set of closures -- is to the larger entailment
structure, the more easily it is to re-constitute (or maintain) that
entity when the entailment structure changes, assuming the entity is
still _possible_ in the changed context. But, the closer the entity is
to its context, the less significant it is!
So, the trick is: How do we maintain a significant entity (e.g. Intel)
when the context changes drastically? The answer is the strategy of
multi-modeling.
Note that this is all stuff I've inferred from what I've read of Rosen's
work. I make no warrants as to what any Rosenites, including Rosen
himself, might think of my comments here (or anywhere else). So keep
your rock salt handy when you read the above.
--
glen e. p. ropella, 971-219-3846,
http://tempusdictum.comWhat luck for rulers that men do not think. -- Adolf Hitler