>1. If complexity is defined in terms of non-linear interactions (most
>common approach?) and measured in terms of computational complexity (e.g.,
>Algorighmic Information Content), then ecosystems are clearly complex but
>we don't have adequate models to base the measure of complexity on, and
>arguably never can have.
>2. If complexity is synonymous with life (Rosen cites Rutherford that
>"every material system is a simple system"), then it may not apply to
>ecosystems, which, while composed of living organisms, may themselves not
>qualify as a living organism and may be describable as material systems
>with non-material (living) components. "The system" is something we define,
>so "the ecosystem" can be (and is often) defined as the material
>relationships and patterns.
Complexity is definitely a subjective (i.e., observer-dependent)
category. You can estimate algorithmic complexity of anything only
after measurement. Depending on the measuring device, we can get
low complexity or high complexity. The same ecosystem may appear
simple for one species (e.g., bacteria) and very complex for another
species (e.g., a mouse).
Your second point corresponds to the situation when a system
observes itself (semantic closure). This kind of complexity (I
would prefer to call it "organization") cannot be measured
externally. However, we may adjust our measuring devices so that
they will mimic (to some extent) the measuring devices of the
system itself. Then we may understand at least some features
of the organization.
The question if ecosystems (or biosphere) are alive is very
interesting. Some ecosystems may expand in space which is equivalent
to self-reproduction. However, most ecosystems are freely exchanging
their species. Thus, there are only few isolated lineages of
ecosystems. My conclusion would be that ecosystems have a low-level
organization compared to organisms. In particular, ecosystems do
not have specialized "memory components" (something similar to DNA).
However, some dominating species (e.g. mono-species forest) may support
all other components of the ecosystem. The role of the dominant
species would be similar to DNA.
There is much more intregration between specific pairs of species
(symbiosis). These often have high organization and eventually may
become real organisms (symbiotic origin of eucaryotic cells).
>It seems to me that these various definitions of complexity (let alone
>measures) are really "inferential" (in Rosen's terms), not material. Given
>that we can only model nature incompletely, complexity seems to refer more
>to the model than to something natural. Relative (measures of) complexity
>applies, then, only to limited views of nature, which necessarily are
>theories and models. Complexity of a theory is not the same thing as a
>theory about the complexity of nature. Complexity then becomes entirely
>subject to how we define a system, at what scale, etc.
I agree that using different measuring devices we will come to very
different measures of complexity. But if we talk about organization,
then we are not free in selecting measuring devices (e.g., scale).
If we observe some persistence in the system we may hypothesize that
this persistence is due to some kind of semantic closure (self-
production, heredity). It may happen that we find semantic closure
at a lower hierarchical level (e.g., not at the level of ecosystems
but at the level of organisms). Then we would say that observed
patterns at the higher level is the effect of organization at a lower
level.
We need to distinguish optimal systems and systems composed of
optimal components. The latter have no organization at the upper
level.
-Alexei
-------------------------------------------------
Alexei Sharov Research Scientist
Dept. of Entomology, Virginia Tech, Blacksburg, VA 24061
Tel. (540) 231-7316; FAX (540) 231-9131; e-mail sharov@vt.edu
Home page: http://www.gypsymoth.ento.vt.edu/~sharov/alexei.html