[Date Prev][Date Next][Thread Prev][Thread Next]   [Date Index] [Thread Index] [Author Index

Re: Von Neumann vs. Robert Rosen



Wow, the list is hopping today!
 
Hi Glen,
 
I can't speak for anyone else, but in my view honest questions are never "too pedestrian". (I have a question of my own.... What does "RTFM" mean? I can come up with a number of possibilities that fit the initials, but there's no way to know which, if any, are the words you mean!)
 
You have put your finger on one of the most important aspects of my father's work, in asking what makes Von Neumann's idea for "self-reproducing automata" different from an organism. The reason I say it's one of the most important is because of what it means, for science. The difference doesn't sound all that earth-shattering... until you follow the lines of logic to what the natural consequences are, particularly for science.
 
My paraphrase of my father's argument goes like this: Von Neumann described a complex system as a system that has "more" of something, which he called complexity, than a non-complex system. He described a complexity threshold that a system could (theoretically) cross, which was the demarcation point between complex systems and non-complex systems. In this worldview, complexity is equivalent to a high level of detail and/or intricacy, which we can achieve in a non-complex system by adding more detail and intricacy to it. This definition of complexity differs markedly from the Rosennean one. If Von Neumann's definition were true (meaning, if it commuted with reality), then life is computable, all systems really are just like machines, and we will be able to create a self-reproducing machine which would be a living organism. Von Neumann's idea of a self-reproducing machine was a theoretical idea, based on a particular conceptualization of what complexity might be, and it's never been officially tested (unless we consider the past several decades of science trying like hell to achieve it "a test").
 
Robert Rosen said that this definition cannot be the case: complexity cannot be the same as "complicatedness" because if it were, we wouldn't have the problems we are having in science when dealing with complex systems-- where we can compute planetary trajectories on a galactic scale, and have our computations commute when predictions are compared with behavior of the actual system-- and yet we cannot account for the systemic behaviors of a single-celled organism. If Von Neumann were correct, there would be no such thing as a system being "more than the sum"... etc. Size would matter, in a sense. So, RR followed the problem: If complexity is some property or quality which is innately different from complicatedness, then what is it? His answer ended up being that complexity was an organizational issue. The relations, as specified/created/maintained by the system organization, are the source of the added value in a complex system. The nature of the relations is key, therefore the way some system is organized is what determines complexity or non-complexity (simplicity) of that system. 
 
In this definition, complexity is something which is true of the system organization from the outset. In other words, it isn't something which can be achieved by adding detail to the existing organization of a non-complex system.  No one knows how complex systems spontaneously self-organize in this universe, but, however it happens, complex systems are that way from inception. Therefore, it is not a matter of arithmetic.  The only way to make a simple system into a complex system, in the Rosennean sense, is to reorganize it. Thus, complex systems are innately different from simple systems, and cannot be productively analyzed by reductive means the way simple systems can. All the "added value" disappears as soon as the system is fractionated and we are left with "the sum of the parts". There is no way to figure out what's missing and add it back in. Do you see?
 
My father often used mathematical arguments to illustrate things like this; for example, pointing out that it is not possible to reach infinity by addition. In order for some system to be infinite, it has to include a circularity or what he described as an "impredicativity," in its organization. If you try to straighten the circularity out into linear form, you lose the system. My way of describing this concept is to say that, in a complex system, time itself is incorporated-- as an ingredient-- into the system's organization. With a system like this, there is no such thing as a "state", even in a theoretical sense. (Frankly, it seems to me that there is no such thing as a "state" in this universe, period. It is a condition devoid of time which can only be approximated experimentally and only exists in human theoretical models.) The models based on notions of state, then, are inapplicable to complex systems except in very limited ways. And this is what the difference between Von Neumann and Robert Rosen leads to. Nothing at the foundations of science can remain the same. 
 
There are several "tests" or conditions by which complex organization will reveal itself. One is reduction to syntax: If you can do that to a system and not lose the system in the process, it's a simple system. In order for a system to be computable, it must be reduced to syntax, so we can extrapolate from that: complex systems are not computable. Fractionation is another test: if you can take the system apart and not lose information (which means you can also put it back together again, using the information learned purely from studying the parts), then it's a simple system. If fractionation destroys the system, irrevocably, then... it was complex.
 
By this measure, atoms are complex systems. Can you see what the consequences of that realization are?
 
Judith

 
----- Original Message -----
To: ***
Sent: Thursday, June 30, 2005 12:15 PM
Subject: [ROSEN] machines vs. living systems (was Re: J. Craig Venter - frabrication: in stores now!?!)

Hello!  I don't think I've posted here before; but, I've lurked for
quite some time and this is an interesting thread.  But, I have a blind
spot that you guys don't seem to have.  Perhaps you can help me fill it in.

First, to set the context, the role of the mad scientist in sci fi is
all about the lack of full coverage of the madman's teleology (or
perhaps we could say what makes them mad is their myopia with respect to
the causal entailment of their artifacts -- and the processes that
produce the artifacts).  So, it seems to me that this horse has been
fully beaten. ;-)

What I don't understand is this distinction between "machine" and ...
well, "non-machine".  What do you mean by "life is no machine"?
Similarly, what is the real (by which I mean _practically_ useful)
distinction between "mechanistic" or "reductionist" thought or methods
and ... well, whatever else there is?  I ask this, here, because it
seems to be a frequent accusation on this list.

To set a little more context for the above question, I'll add that I
think of something like von Neumann's self-reproducing automaton as both
a machine _and_ an artifact that attempts to tacitly demonstrate a
common theme in Rosen's writings.  (Or, I should say "what I infer as a
theme from my limited understanding of Rosen's writings".)

That automaton (specifically that one, not the many impoverished
reformulations) seems like it directly addresses the issue of a
dynamical process being embedded in a logical ontology, where the
logical ontology is rich enough to support causal entailment beyond the
original prescription present in the automaton, but simple enough to
prevent the automaton from "breaking" or evolving into something wholely
different.  And in so doing, it broadened the definition of "machine" to
include at least one, if not two, of the properties of living systems
that had been part of the popular distinction between machines and
living systems.

My point is that entailment seems to be more about the structure of
explorable possibilities than about any particular system doing the
exploring.  So, by asking my question above about "machines" and the
denigrating way in which that term is used, I'm asking those of you who
_feel_ a tacit difference between a living organism and a machine to
talk about the key elements of the distinction you make.

I honestly can't make the distinction.  A living organism seems very
much like a "robust machine" to me.  By "robust", I mean it is capable
of persisting (in full dynamism) in manifold contexts.  So, the problem
of engineering (not science) is to find ways to increase the robustness
of our artifacts.  And we look to biology for help.

Thanks for your indulgence and I apologize if these points are too
pedestrian.  Feel free to tell me to RTFM. ;-)
-glen

Dan Fiscus wrote:
> This is the fabrication issue that Rosen warned about, upon
> us now. What could we say about this or how events might
> play forward? My immediate and first hunch which I think
> would be compatible with Rosen (not sure) is that whatever
> goal, purpose, function etc. Venter and fellow would-be
> creators of life may have for their creations, these goals will
> not likely be met, achieved, gained or under their control.
>
> I say this since, again following Rosen, life is no machine. If
> Venter's bacterium were a machine - no problem to assume
> one can design, expect, achieve, control the outcome and the
> goal/pupose/function/results/effects of the creation. But if
> the bacterium is alive, then almost by definition it will also
> embody and entail and create its own goals/purposes/
> functions/results/effects/development/growth/evolution,
> etc. There could be no way to expect, determine or control
> that these goals endogenous to the life form would in any
> way match up to the exogenous goals of the creator. A "good",
> realistic, humble creator, perhaps, would not be attached to
> the outcome or results, but would be detached and know
> that the life form would literally "take on a life of its own".
> The surprises may be nightmarish or happy, but they seem
> to me guaranteed to be unknowable and truly surprising.
> And as such, very hard to make any profit off of, unless one
> billed the process as a freak show or adventure, like Russian
> roulette maybe...

--
glen e. p. ropella              =><=                Hail Eris!
H: 503-630-4505                       http://ropella.net/~gepr
M: 503-971-3846                        http://tempusdictum.com