>> Turchin's terminology is one of the barriers for understanding.
>> He uses old cybernetic terms in a very different meaning.
>
>Can you give a specific example?
I meant the notion of control.
>> Traditional
>> cybernetics is interested solely in the function of a system and
>> is not interested in its origin or adaptaion.
>
>Well, depending on what you mean by "traditional", that's not strictly
>true. Certainly von Foerster and Ashby were centrally concerned with
>issues of organizational development and evolution.
I don't have these books on my desk, but I cannot remember
any evolutionary models developed by Ashby or von Foerster.
I agree that cyberneticians often talked about development
and evolution but there was no theory behind it. Thus,
cybernetics, as a theory, is always associated with function
of systems (information and operation theory) rather than
evolution.
>> I don't like the term "control" because historically it was
>> used for a subsystem that leads the system to a target state
>> using all information available. In evolutionary systems
>> there is no fixed target state.
>
>This is an issue, and is addressed in later PCP work beyond POS (see
>at least http://pespmc1.vub.ac.be/control.html).
The first paragraph in this web page defines control in pure
functional terms (as all cyberniticians do). Nothing is said
about evolution or adaptability of control.
In the second paragraph, the approach changes by 180 degrees.
It is written: "We see in the controller an agent which is
responsible for its actions...". I like this much more than the
first paragraph, but the problem is that it does not fit to
traditional cybernetics. Let us take a Watt's controller in
a steam engine as an example. Is it responsible for its actions?
In the figure, the goal is shown as an external input for
the agent. This is exactly what I was talking about in my
previous message: you have a fixed goal (target state
measured by the controller). From the evolutionary point of
view, there is no fixed goal.
>> Also, systems at higher levels
>> in a hierarchy never attempt to gain control over all tiny
>> details of the functioning of systems at lower level. If they
>> try to do this, then the system becomes non-adaptable and
>> sooner or later become extinct.
>
>Of course, we deal with this as well.
My previous critique was targeted at the functional and
non-evolutionary approach in cybernetics, which is 99% of the
old traditional cybernetics. I am sure, you can handle this
problem. Also, it was not a critique of Turchin's ideas
which I think are very valuable. My point is that his theory
of metasystem transition probably was the first attempt to
build evolutionary models in cybernetics. But Jack Horner
(to whom you replied before) did not notice this novelty. And
the reason why he did not notice is apparently the terminology.
I think, there is no reason to extend the meaning of "control"
beyond its limits. Let us leave it as a functional term without
any "agents that are responsible for their actions". But we
definitely need additional terms for evolutionary phenomena.
I think the "semantic closure" suggested by Pattee may be a
good one.
-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