models that bite back
Locked
 
|
A sidebar conversation regarding the "reality" of models 'The story that I have to tell is marked all the way through by a persistent tension between those who assert that the best decisions are based on quantification and numbers, determined by the patterns of the past, and those who base their decisions on more subjective degrees of belief about the uncertain future. This is a controversy that has never been resolved.' — FROM THE INTRODUCTION TO ''AGAINST THE GODS: THE REMARKABLE STORY OF RISK,'' BY PETER L. BERNSTEINSee http://www.nytimes.com/2009/01/04/magazine/04risk-t.html?_r=1&ref=magazine -tj -- ========================================== J. T. Johnson Institute for Analytic Journalism -- Santa Fe, NM USA www.analyticjournalism.com 505.577.6482(c) 505.473.9646(h) http://www.jtjohnson.com [hidden email] "You never change things by fighting the existing reality. To change something, build a new model that makes the existing model obsolete." -- Buckminster Fuller ========================================== ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
Tom,
Some of us look to both the patterns of the past and a subjective belief about the uncertain future when making decisions. And sometimes the way we interpret past patterns is as subjective as our anticipation of the future. Why set up a non existent conflict? O Tom Johnson wrote:
============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
I agree with Orlando that there is no need for a conflict here. The Bayesian paradigm provides a unified framework for decision making that integrates a subjective interpretation of the past record and views of the future. Further it is a paradigm that in a principled way modifies current beliefs according to incoming data--Bayesian learning. In an important sense the Bayesian paradigm does resolve the controversy.
George
On Thu, Jan 8, 2009 at 10:22 PM, Orlando Leibovitz <[hidden email]> wrote:
-- George T. Duncan Professor of Statistics, Emeritus Heinz College Carnegie Mellon University (505) 983-6895 Life must be understood backwards; but... it must be lived forward. Soren Kierkegaard ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
All well and good, …unless something in the environment develops
a continuity of divergence Phil Henshaw NY NY www.synapse9.com From:
[hidden email] [mailto:[hidden email]] On Behalf Of George
Duncan I agree with Orlando that there is no need for a conflict
here. The Bayesian paradigm provides a unified framework for decision making
that integrates a subjective interpretation of the past record and views of the
future. Further it is a paradigm that in a principled way modifies current
beliefs according to incoming data--Bayesian learning. In an important sense
the Bayesian paradigm does resolve the controversy. George On Thu, Jan 8, 2009 at 10:22 PM, Orlando Leibovitz <[hidden email]>
wrote: Tom, A sidebar conversation regarding the "reality" of models 'The story that I have to tell is marked all the way through by a persistent
tension between those who assert that the best decisions are based on
quantification and numbers, determined by the patterns of the past, and those
who base their decisions on more subjective degrees of belief about the
uncertain future. This is a controversy that has never been resolved.' — FROM THE INTRODUCTION TO ''AGAINST THE GODS: THE
REMARKABLE STORY OF RISK,'' BY PETER L. BERNSTEIN ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
Phil Henshaw wrote:
> > All well and good, …unless something in the environment develops a > continuity of divergence > A model can be built around whatever hunch and evaluated in a Bayesian framework. At some point, if the divergence really exists, the model will reflect that in its likelihood. It's all well and good. Marcus ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
That's only in you model, and leaves out the rest of the world. My "hunch"
is it's good to watch the rest of the world for diverging continuities too... Phil Henshaw NY NY www.synapse9.com > -----Original Message----- > From: [hidden email] [mailto:[hidden email]] On > Behalf Of Marcus G. Daniels > Sent: Friday, January 09, 2009 12:25 PM > To: The Friday Morning Applied Complexity Coffee Group > Subject: Re: [FRIAM] models that bite back > > Phil Henshaw wrote: > > > > All well and good, unless something in the environment develops a > > continuity of divergence > > > A model can be built around whatever hunch and evaluated in a Bayesian > framework. At some point, if the divergence really exists, the model > will reflect that in its likelihood. It's all well and good. > > Marcus > > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > lectures, archives, unsubscribe, maps at http://www.friam.org ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
Phil Henshaw wrote:
> That's only in you model, and leaves out the rest of the world. My "hunch" > is it's good to watch the rest of the world for diverging continuities > too... > Nothing prevents a person from explicitly representing and revising beliefs about the world in a model, especially in an ABM. ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
Well Marcus, isn't that is entirely the point, and why models are unreliable
and need help? A model invariably represents only a person's belief's about the world. The physical subject being represented is both fabulously more complex than any belief system can be, and full of things that are differently organized and requires it's own language of description. It's why one needs a different mode of description for each way of describing what a person is. It's why science, being one language of description, is incomplete. In some cases, a common language seems adequate for many subjects, but only when you are careful to ask the same kind of question of each subject, consistent with that common language. To use common terminology for different things you do need to ignore the discrepancies as insignificant, though. As when your economic system collapses because they were not actually insignificant, that turns out to be an error. It ends up being much safer to think of the physical world as complexly changing place needing many languages of description and close attention, and for science, to watch the fit of your model to see if discrepancies are developing. In order to pick up significant errors due to emerging discrepancies, you need to become aware of what's happening. One way is to watch closely for them. You can also rely on hearsay. The world is full of independently evolving systems, each changing it's organization in response to its own place in the world, in its own way, and developing emergent behaviors as it does. Lots of systems we share the environment with seem sort of diffuse and passive, and others rather distinctly individual with strong independent individual reactions to being interfered with. There's no 'book' you just have to watch. If you're not watching and only wait till you loose your job to know that you should have been watching, (like a lot of us are at the moment) you're out of a job. It's like we were imagining an open road and were driving along in our car and didn't see the water coming because it wasn't on the map. The water coming was real obvious to the people looking out the window who were repeating saying in increasingly urgent tones "hey there's water coming". Am I wrong to be stunned at how difficult it is to get an acknowledgement here that living in a physical world means that theory is not enough? Phil Henshaw NY NY www.synapse9.com > -----Original Message----- > From: [hidden email] [mailto:[hidden email]] On > Behalf Of Marcus G. Daniels > Sent: Friday, January 09, 2009 2:12 PM > To: The Friday Morning Applied Complexity Coffee Group > Subject: Re: [FRIAM] models that bite back > > Phil Henshaw wrote: > > That's only in you model, and leaves out the rest of the world. My > "hunch" > > is it's good to watch the rest of the world for diverging > continuities > > too... > > > Nothing prevents a person from explicitly representing and revising > beliefs about the world in a model, especially in an ABM. > > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > lectures, archives, unsubscribe, maps at http://www.friam.org ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
Phil Henshaw writes:
> A model invariably represents only a person's belief's about the world. > Consider surveys of undecided voters where during a debate the surveyed turn their individual dials to indicate approval or disapproval. > The physical subject being represented is both fabulously more complex than > any belief system can be, A library is fabulously more complex than most any individual's belief system as well. > and full of things that are differently organized > and requires it's own language of description. Military simulations, for example, are often large federated systems, where each part is designed by a different domain expert. Some parts of the models could even be delegated to human decision makers, as in the voting example. Complexity, subjectivity, and quantification are different issues. Marcus ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
And,... how does a poll, or a military analysis tell you what emotions are
going through people's minds? That kind of clairvoyance is what you're claiming, you know. It seems to me a yes/no vote has insufficient variety in comparison to thought, and a potential kill ration won't tell you if going ahead will unusually piss people off. Phil Henshaw NY NY www.synapse9.com > -----Original Message----- > From: [hidden email] [mailto:[hidden email]] On > Behalf Of Marcus G. Daniels > Sent: Friday, January 09, 2009 5:35 PM > To: The Friday Morning Applied Complexity Coffee Group > Subject: Re: [FRIAM] models that bite back > > Phil Henshaw writes: > > A model invariably represents only a person's belief's about the > world. > > > Consider surveys of undecided voters where during a debate the surveyed > turn their individual dials to indicate approval or disapproval. > > The physical subject being represented is both fabulously more > complex than > > any belief system can be, > A library is fabulously more complex than most any individual's belief > system as well. > > and full of things that are differently organized > > and requires it's own language of description. > Military simulations, for example, are often large federated systems, > where each part is designed by a different domain expert. Some parts > of > the models could even be delegated to human decision makers, as in the > voting example. > > Complexity, subjectivity, and quantification are different issues. > > Marcus > > ============================================================ > FRIAM Applied Complexity Group listserv > Meets Fridays 9a-11:30 at cafe at St. John's College > lectures, archives, unsubscribe, maps at http://www.friam.org ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
Phil Henshaw wrote:
> And,... how does a poll, or a military analysis tell you what emotions are > going through people's minds? Given a hunch or actual evidence that a class of emotions have relevance to an interesting mass behavior, a poll could be open-ended, where those polled would describe their feelings about some stimulus, and then a knowledge engineer would listen carefully and formalize what they heard. A model of psychological or sociological phenomena doesn't need to be fixed. There could even be a feedback process where the polled individuals would review the formalism to see that they agreed that the interpretation was a accurate and if not, refine it. Alternatively, one could in-principle accuse a person, or something more extreme, and take careful notes on reactions based on available context and infer a sub-model from those observations. Such stories, for many individuals, could be converted into computer programs that describe how each agent changes from state `Happy' to `Sad' to `Mad' (or whatever states described) on the basis of different kinds of stimulation. Of course, the polled individuals could lie or be delusional, or be easily led by careless interviewer. But for the moment, the mind is still something of a black box. Marcus ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
There is much further to fall, and I think it's likely the Obama plan will
aggravate the failure of the system and push it over the next edge. It will certainly not relieve it of strain and allow it to heal. The Obama plan is designed by the same theory that caused the collapse, and intended to pump up the process of harvesting multiplying returns from our diminishing, dangerously unstable, and increasingly unresponsive set of physical resources. If we pull out all stops to continue on that path as intended it will probably push the system to a significantly greater failure that may be relatively permanent. The only thing that will work is for the people who have financial claims to be paid more than the physical system was able to produce to rescind those claims, i.e. come to a realization that having taken too much money out of the system enough debts need to be forgiven or enough money put back as needed to relieve the system of unachievable obligations to them. Phil Henshaw NY NY www.synapse9.com ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
|
|
In reply to this post by Marcus G. Daniels
My e-mail address has been changed. The earthlink address is now a time-
limited address. Please remove my name from that address: [hidden email] ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org |
«
Return to Friam
|
1 view|%1 views
| Free forum by Nabble | Edit this page |