Re: Life without evolution/evolution without life

[John Kineman <***>]

Hi Howard,

Sorry if I gave a wrong impression. Of course VN's view is far from
nonsensical. I only meant that as I see RR's definition it establishes a
primary condition of nature - like saying matter and energy are
convertible. One doesn't then develop a measure of how convertible it is
under various conditions or how that convertibility evolved. I see RR's
definition as saying that function and realization are convertible, but
never precisely. So, its a very different kind of complexity definition
that does not easily lend itself to quantification. On the other hand, the
uncertainty in that relationship probably can be scaled in the sense of how
precisely the relationship between a given function and a given structure
holds; so I have to retract my earlier statement that it can't be scaled at
all - under practical conditions one would want to discuss questions of
"how much" a given system exhibits complex behaviors or simple ones. I'm
thinking that makes more sense from a Rosennean view if the measure is of
simplicity rather than complexity because complexity as a basic principle
is always there, where simplicity is something a complex system can do to
varying degrees. Does that help?

On another track, there are indeed many definitions of complexity that have
value but are not proposing new laws of nature. In landscape ecology, my
adopted field these days, there are simple metrics for landscape complexity
one can calculate based on patch heterogeneity. It is strictly a measure of
structural or pattern heterogeneity and has nothing to do with the
Rosennean definition, and, as you properly say, none of these other
measures or definitions are invalidated.

In the VN case, perhaps it is my ignorance of his definition that led me to
compare the two in this way. If it is clear what metric is being used to
define a computational complexity in VN terms, then, yes, it makes sense to
say at what level of such complexity does a system  run away or below that
level degrade to simplicity. However, I'm not even sure this actually
happens. Why would we suspect there is such a threshold at all? In natural
landscapes the complexity metric can be calculated for many systems, but it
is nothing that metric measures that determines how complex the system will
become; it is other processes that produce the pattern which we then
measure. Is there a feedback from the pattern to the process? Some have
suggested some possibilities, like "ecological memory" whereby a pattern
can be self-maintaining. As far as I know (there's a lot of room in what I
don't know) I havn't seen an analogy in landscapes where a critical
threshold dictates a non-linear shift to one attractor or another, but I'm
sure it must be possible.

Regarding the threshold where systems can begin to evolve, I think the
difficulty I was alluding to is that we must not presume the process
involves going from a simple system to a complex one in the fundamental
sense. And if we are talking about origins, ontologies, it is the
fundamental definition of complexity that is important, not the metric of
structural heterogeneity. In that case systems start out complex and the
interesting questions as I would write them are (actually two): (a) how do
complex interactions produce a simple measurable space-time system that is
stable, and (b) how do some complex systems develop within that space-time
material context, the ability to reproduce themselves and retain and make
further use of natural complexity?

So, you are right -- While I don't believe I am not at all trying to
invalidate VN, I am suggesting his line of investigation may not be the
most fruitful one, if Rosen was right that complexity isn't something that
is constructed. If it is constructed, then it makes more sense to me to
look for the critical threshold where enough of it leads to life, but I
just don't think that is true.




At 08:23 PM 3/22/04 -0500, you wrote:
> John,
>
> I do not understand how I can respond substantively to this line of argument:
>
> [Howard] Von Neumann's fundamental question is: What is the threshold of
> complexity of a system that can evolve increasing complexity rather than
> degrade?
>
> [John K] Rosen's view of complexity, as I read it, would make this a
> nonsensical question.
>
> As I said to Judith: I have no problem with Bob defining complexity
> differently than von Neumann. They both make sense to me. But how can you
> claim that by one definition you can determine that another definition
> becomes nonsensical? Of course I agree that definitions may be
> incompatible, logically inconsistent, or just different, but by your form
> of argumentation you win just "by definition."
>
> Or are you asserting that all of von Neumann's thinking on the question of
> evolvability is actually nonsensical?
>
> Howard