|
Who remembers the Bloom County comic strip of the 1980's and 90's?
There was a computer whiz-kid who was always achieving astonishing things by
typing away furiously at his keyboard, much to his father's alarm. The kid
generally managed to fix things when he occasionally goofed, but it was always
spectacular when he goofed and unexpected side-effects from his experiments
began to manifest themselves... One of my favorite episodes had this
kid (whose name I can't recall at the moment) concentrating on his computer,
while Opus (the main character, a lovable and irreverent penguin) was
hanging around, pontificating about something or other. Suddenly the kid said
"Oops." Opus became very alarmed. "What's wrong?!" The kid didn't
answer, typing faster, but finally reassured Opus; "Nothing's wrong..." and
kept on typing urgently. Opus wasn't lulled into a false sense of security.
"Wait a minute!" he said, waving a finger/flipper..."Don't give me that-- I
distinctly heard an "oops"!!!!"
"Oops!" was my reaction, reading Jack's post.....
[Jack, I could have sworn that you posted several things about this
on the list.... but I'm juggling so many damn balls in the air right now that
the green one may have looked like the blue one, or something to that effect.
I'm sorry!]
Ah, life! Every day is a wee adventure! Well, what can I say?
I hope some good comes of this. And, to that end, I have some comments and
questions on Jack's post:
JP: My recent background involves something derived from the artificial intelligence thread called qualitative physics: qualitative process theory. From that, my instincts tell me that one approach to take in modeling this tribe is to ask these questions: who/what are the actors? what are their relationships? what are the states involved? what are the process rules involved? "states" being that n-tuple of observables mentioned in "Organismic Supercategories." I suppose I might rephrase my questions in this way: who/what are the actors? (what are the components) what do those actors entail? I still might ask to enumerate the observable states. So, you want to know how to model the list, organizationally
(including all relations and components, as far as is humanly possible to figure
them out), right? The first step is probably to sort out what is "list" and
what is "ambience" or "list context". This is important but it's
tricky: As with any complex system, the two include aspects
of each other in themselves. On top of that, contextual aspects which have
been internalized into the system's organization are now part of the internal
context and can interact with and change/be changed by one
another. both the internal contexts and the external contexts. Such
interactions cause change in both, and so it goes...
So the first issue is one of context delineation and system
definition.
Incidentally, these ARE reductionist techniques. (I can
already "hear" John M. typing... ) The difference between a reductionist
frame of reference and reductionist techniques is huge. It's the
reductionist frame of reference that does the
harm.
JP: My quest, in some sense, is to determine the mappings necessary to
morph Qualitative Process Theory into RR Theory. My intuitions suggest that this
should be possible.
I've done some reading about process theory and it definitely is an attempt to deal with some of the issues in science caused by the fact that the reality of a continuously interacting system isn't describable using a state-based vocabulary. Process theory is an add-on to the periphery of current scientific thought, rather than a re-evaluation of the foundations. As such, it doesn't have any underlying theoretical explanation for the reason why so much that goes on in the universe seems to be a "process" rather than a "thing", or why such aspects of reality are so hard to deal with, scientifically. Your intuition that Rosennean Complexity Theory (RCT, as Tim calls it) could provide the missing theoretical underlayment is probably valid. It would change certain aspects of Process Theory, though. I'll have to go back over my notes and do a little further research to put my finger on what and where the changes are likely to be. JP: My program adds a different dimension to computation from that which a
"calculator" brings forth; the dimension of discovery. Think of TSC as a
kind of spreadsheet where "what if" questions can be asked. With TSC, the program, itself, has the ability to "play" with the numbers (concepts), to perform mutations, conceptual combinations, on objects stored as symbols (an actor or a relation is a symbol), and to then return to perform modeling computations on the mutated concept. This is called evolutionary programming. The "conjectures" offered by the program may, or may not, stimulate cognitive processes in the user. That's the point. Fascinating-- sort of like jump-starting a lethargic
imagination? This is just like what a fiction writer does-- The
ol' "... if this, then... what?" It's also involved in a process called
"worrying". It's also involved in what is commonly called "day-dreaming"...
These are all ways of following the lines of entailment to see where they would
lead, then going back to the beginning and changing the variables, then
following the entailment again, etc...My brain does all of the above, all
the time, to the point that I have trouble sleeping... If I don't like the way a
movie ended, I "rewrite it" in my head. That's the best way for dealing with a
nightmare, too, by the way. It's a technique I figured out when I was
small: Use your imagination to rewrite the dream so it isn't scary
anymore.
JP: What does it mean to "perform computations" in QPtheory? It means this. Entailments in QP Theory are called "process rules." TSC allows a user to "set a stage" with actors, relations and states (initial conditions). Process rules are then collected to and "fired". That's the equivalent of a "script" being read. Some rules fire, states/relations/actors change, and you form a new stage setting. Which suggests that the entailment chains (process rules) are
going to present the biggest challenge for you. Those have to be right or
it's just a video game. So, what aspects of the discussion list entailment are
you interested in modeling? To give you some navigation equipment in deciding,
think of the types of information you get from applying Aristotle's analysis to
the list. Those are four types of entailment relations.
JP: Having explained my quest in such terms, how would Rosennean thinkers approach the task of modeling this tribe? What questions would they ask? What analytics would they apply to those questions? Before we can gather information, another big question that needs to be answered is "What do you want the model to help you do?" The model is going to represent a piece of the actual system, so what you want to do is create a model (or a group of complementary models) with the capability to generate "predictive information" which is testably accurate. Judith
|