Re: emergence (or is that re-emergence)

From: Eric Hawthorne <egh.domain.name.hidden>
Date: Tue, 26 Nov 2002 23:42:13 -0800

Let me first apologize for not yet reading the mentioned references on
the subject,

John Mikes wrote:

>As long as we cannot qualify the steps in a 'process' leading to the
>"emerged" new, we call it emergence, later we call it process.
>Just look back into the cultural past, how many emergence-mystiques
>(miracles included) changed into regular quotidien processes, simply by
>developing "more" information about them.
>I did not say: "the" information. Some.
>
I don't think this is correct.

A fundamental concept when talking about "emergence" ought to be the
pattern, or more precisely, the interesting, coherent, or perhaps useful
pattern; useful
perhaps in the sense of being a good building block for some other pattern.
"Process" is a subset of pattern, in the sense in which I'm using
"pattern". Also,
"system" is a subset of "pattern".

Q:
How do you know when you have "completely" described a pattern?

Two examples, or analogies, for what I mean by this question:

e.g. 1 I used to wonder whether I had "completely" proved something in
math, and
would go into circles trying to figure out how to know when something was
sufficiently proved or needed more reductionism i.e. The old
"Wait a minute: How do we know that 1 + 1 = 2?" problem. The gifted
mathematicians
"teaching" me seemed to have no trouble knowing when they were finished
proving
something. It was "intuitively obvious" <-- load of cods wallop of
course. And I
still wonder to this day if they were simply way smarter than me or
prisoners of
an incredibly limited, rote-learned math worldview. The point is, every
theory;
every description of states-of-affairs and processes or systems
(patterns) using
concepts and relationships, has a limited domain-of-discourse, and mixing
descriptions of patterns in different domains is unnecessary and
obfuscates the
essentials of the pattern under analysis.

e.g. 2 Is the essence of human life in the domain of DNA chemistry, or
in the domain
of sociobiology, psychology, cultural anthropology? Are we likely to
have a future
DNA based theory of psychology or culture? Definitely not. Cellular
processes and
psychology and culture are related, but not in any essential manner.

A:
Let's define a "complete" description of a pattern as a description which
describes the essential properties of the pattern. The essential
properties of the
pattern are those which, taken together, are sufficient to yield the
defining
"interestingness, coherence, or usefulness" of the pattern.

Note that any other properties (of the medium in which the pattern
lives) are
"accidental properties" of the incarnation of the pattern.

Note also that the more detailed mechanisms or sub-patterns which may
have generated
each particular essential property of the main pattern are irrelevant to
the creation
of a "minimal complete description" of the main pattern being described.
As long as
the "property" of the main pattern has whatever nature it has to have as
far as the
pattern is concerned, it simply doesn't matter how the property got that
way, or
what other humps on its back the property also has in the particular
incarnation.

And that "level-independence" or "spurious-detail independence" or simply
"abstractness" of useful patterns is one of the reasons why it makes
sense to talk
about emergence.

 e.g.of "level-independence of a pattern".

1. "Game of Pong"

2a. Visual Basic 2b. Pascal program 2c. Ping-pong table,
      program on PC on a Mac ball,
bats, players

3a. x86 ML program 3b. PowerPC ML program 3c. Newtonian physics of
                                                                        
               everyday objects
4a. voltage patterns in 4b. voltage patterns in
      silicon NAND gates Gallium Arsenide NOR gates (you get the idea)

Key:
-----
1. The "main" pattern being described

2, 3, 4. Lower-level i.e. "implementation-level" or
"building-block-level" patterns whose own
internal details are irrelevant to the emergence of the main pattern,
which emerges
"essentially identical" from all three of very different lower level
"building-block" patterns.

So in summary, an emergent pattern is described as emergent because it
emerges,
somehow, anyhow, doesn't matter how, as an abstract, useful, independently
describable pattern (process, system, state-of-affairs). A theory of the
pattern's essential
form or behaviour need make no mention of the properties of the
substrate in which the
pattern formed, except to confirm that, in some way, some collection of
the substrate
properties could have generated or accidentally manifested each
pattern-essential property.
A theory of form and function of the pattern can be perfectly adequate,
complete, and
predictive (in the pattern-level-appropriate domain of discourse),
without making any
reference to the substrate properties.

This is not to say that any substrate can generate any pattern. There
are constraints,
but they are of many-to-many cardinality from substrate level to pattern
level.

Why are computers, of all forms, so cool?
Because certainly any universal computer substrate can generate
any computable pattern. Are all realizable patterns computable?
Probably. Cool.
Received on Wed Nov 27 2002 - 02:43:51 PST