The relevance of impredicativity

[Steve Johnson <***>]

Amen to that, Jack.

I think there is too much focus on impredicative
modelling in discussing Rosen's legacy to the
exclusion of more immediately useful areas of
exploration. 

I see Rosen's legacy as falling into these broad
categories:  

1) importance of relational modelling
2) insufficiency of state-based Newtonian paradigm, 
3) anticipation, 
4) epistemology of modelling and measurement
5) the argument that life required impredicative
models, closed causal loops etc.

I feel like only item (5) has to do with causal loops,
impredicativities and the line. Items 1-4 have
enormous independent value even if (which I don't
believe) the Church-Turing thesis is ultimately true.

So we can't model impredicativies on a computer. So
what? We can't model real numbers either. We just take
an approximation and in many cases it turns out good
enough.

There is a hell of a lot of relational modelling you
can do in Mathematica (topology, knots, some category
theory). These are all relational subjects they have
nothing to do with the Newtonian paradigm.

I think Rosen's anti-computer ire was mostly directed
against certain media lab types who make silly robots
with a camera attached to a neural net and then
pontificate whether or not it is conscious.

Who is to say that no kind emergence is possible when
highly complex software system reaches certain
threshold? Yes it will not be LIFE but that does not
mean it will not be INTERESTING. Just look at the
Internet. As a human/computer chimera it might well
lead to a meta-system transition of a novel kind.

- Steve










--- Jack Park <***> wrote:

> As Richard Hamming famously said: "The purpose of
> computing is insight, 
> not numbers."
> 
> Cast in that light, just about anything a computer
> can do to illuminate, 
> if only in some small way, candidate pathways to
> enlightnement in the 
> eyes of modelers, problem solvers, then, let it be
> said that a benefit 
> is accrued. If such modelers, problem solvers are
> looking for insight 
> into the nature and solution to the vast array of
> complex and urgent 
> problems facing humanity, then, even better.
> 
> We can sit here and wax poetic or even religious
> that computers can or 
> cannot do this or that. In the end, who really gives
> a shit? Some of us 
> are bucking for a dissertation defense, some of us
> are trying to collect 
> or fulfill a payroll, and some of us really care
> about those complex and 
> urgent problems.  You (Judith) and others can take
> the time to wax 
> thoughtful on the notion of a cookbook, or, you can
> get on with tossing 
> ideas into the ring. I am happy to see a relational
> analysis of the 
> problem space posed just by the notion of a
> cookbook, itself an intended 
> relational model of a perceived discipline. The
> analysis sketched below, 
> itself, is a useful start.
> 
> People, such as Jerry will argue, and correctly so,
> that we cannot share 
> knowledge. Information is the currency of human
> discourse. Gordon Pask, 
> who's entailment meshes are as close to a relational
> space as any I've 
> seen, argued that we each hold, in our minds, models
> of those with whom 
> we engage in social intercourse. Without such
> models, the information we 
> encode in our conversations would likely be just so
> much noise. My model 
> of my son, when he was 3, suggested that an indepth
> dialog on technical 
> analysis of IBM's ticker just wouldn't make much
> sense to him. These 
> days, he might surprise me. So, we can share
> information, and given that 
> our internal models of our listeners allow us to
> perform a kind of "pre 
> interpretation" such that we emit only those ideas
> that we either think 
> to be meaningful--as in, capable of being
> interpreted in a way that 
> meaning was made in the mind of our listener, or
> that we think will 
> provoke a debate (or titilate, when the game is
> humor). Indeed, Marvin 
> Minsky thinks that humor is all about expectation
> failure.
> 
> And, expectation failure is what anticipatory
> systems are all about, and 
> that is what my qualitative reasoner uses as a
> modeling trick to provoke 
> thoughts in the human in the loop. Thus, I would
> argue that even the 
> simplest, purely syntactic, symbolic model holds
> forth the opportunity 
> to illuminate things, ideas, concepts, events which
> the user may not 
> have seen.  In my experience with my program,
> crystalographers who were 
> asked to encode their domain knowledge into the
> process rules of my 
> system all commented that just the process of
> "teaching" my program 
> opened their eyes to things they hadn't thought
> about before. You don't 
> have to solve a problem to discover something; you
> just need a means by 
> which you can hold-forth a dialog within your own
> mind as you formulate 
> some story for some other entity to hear (whatever
> hearing entails). 
> Following a cookbook is just one way to learn
> something. Writing the 
> cookbook is another.
> 
> We, as humans, are extremely fond of telling each
> other what is or is 
> not possible. I submit that that behavior is part of
> the problem. Go 
> model that.
> 
> Jack
> 
> Judith Rosen wrote:
> 
> > *I continue to mull over these ideas...*
> > ** 
> > *What limits me the most is a clear understanding
> of what computer 
> > programming can realistically encode. I use
> computers a lot, and I have 
> > amassed a large cache of mostly intuition-based
> understanding, using 
> > what experts like Jack and others have said as a
> form of "parameter 
> > checking".... but I have grave doubts that this is
> enough to really help 
> > generate a "Rosennean information sorting
> protocol" or something along 
> > those lines... *
> > ** 
> > *So I keep throwing various insights out there,
> hoping something will 
> > prove useful and can be plugged into your much
> larger cache of knowledge 
> > and help you along in your efforts.*
> > ** 
> > *The problems Jack is facing, as I perceive them,
> are:*
> > *1.) Humans are complex systems.*
> > *2.) Human conscious minds are a complex
> system/component of the average 
> > human being.*
> > *3.) Language is a complex system in its own
> right, and is also 
> > a component of the human conscious mind.*
> > *4.) There is a relation between the human mind
> and language which is 
> > also complex.*
> > *5.) /Semantics/ are indispensable _within_ the
> complex system that is 
> > "language"-- as is /Context /(there is a crucial
> relation there, between 
> > those two aspects).*
> > *6.) Semantics and context are also indispensable
> in the complex 
> > relation between human minds and language.*
> > *7.) Computers are finite and are, therefore,
> incapable of fully 
> > modeling any complex system.*
> > ** 
> > *The good news, on the other hand, is:*
> > *1.) This is intended to be an interactive tool
> and therefore its finite 
> > incompleteness is not a fatal limitation: The
> human mind and human 
> > relational ability will come along "for free" in
> any actual use of this 
> > interactive tool. (In other words; it's meant to
> be part of a larger 
> > chimerical system of
> human-and-computer-and-interactive-tool.)*
> > *2.) Reductional models of complex systems can be
> incredibly useful and, 
> > in fact, organisms naturally incorporate the
> formation and functional 
> > use of such reductions within themselves in the
> natural world (i.e.; 
> > Anticipatory Systems) to great effect, all the
> time.*
> > *3.) It should be possible to figure out which
> aspects of the complex 
> > systems being modeled can "safely" (safety being a
> relative term!) be 
> > dispensed with and which cannot, such that the use
> of the models doesn't 
> > generate intolerable "side effects".*
> > *4.) It should also be possible to list which
> potential side effects 
> > would be intolerable (a term which I would
> tentatively define as 
> > meaning: limited/limiting to a negative degree
> /and/ unable to be 
> > compensated for via the interactive human mind
> using the program) and 
> > which potential side effects are clearly tolerable
> (easily compensated 
> > for).*
> > ** 
> > *What I can't answer, however, is whether the
> resulting information will 
> > provide answers which can be translated into an
> information-sorting 
> > protocol via computer programming language rules.*
> > ** 
> > *Judith*
> > ** 
> 


__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com