home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!gatech!paladin.american.edu!auvm!VAXF.COLORADO.EDU!POWERS_W
- X-Envelope-to: CSG-L@vmd.cso.uiuc.edu
- X-VMS-To: @CSG
- MIME-version: 1.0
- Content-transfer-encoding: 7BIT
- Message-ID: <01GOL2WPWUQA007LS3@VAXF.COLORADO.EDU>
- Newsgroups: bit.listserv.csg-l
- Date: Wed, 9 Sep 1992 15:19:49 -0600
- Sender: "Control Systems Group Network (CSGnet)" <CSG-L@UIUCVMD.BITNET>
- From: "William T. Powers" <POWERS_W%FLC@VAXF.COLORADO.EDU>
- Subject: Nits; Stella etc.; Gary's paper; raising organisms
- Lines: 149
-
- [From bill Powers (920909.1430)]
- Chuck Tucker (92-09-09)
-
- Every now and then you decide to put all the arguments into some kind
- of order. Very nice. If you keep doing this, we may yet figure out
- what we're talking about.
- ----------------------------------------------------------------------
- -
- Greg Williams (920909!) --
-
- >Bill, I think we've reached an important consensus about my claims.
- >Now, we can go back and nit-pick some more, or ...
-
- I want to nitpick some more.
-
- >>I agree with you that lots of this kind of deception goes on.
-
- >>YOU GOT IT!!! But there's still more to get....
-
- I think that we agree that deception is a form of manipulation. We
- also seem to agree that rubber-banding is a form of manipulation (if
- anyone doesn't understand what rubber-banding means, ask). But one of
- these forms is control, while the other is not.
-
- When you manipulate action by applying a disturbance to a known
- controlled variable, the action will correlate with the disturbance
- somewhere close to -1.00, as long as the other person continues to
- control that variable relative to the same reference level. Accurate
- predictions can be made of individual actions over considerable
- periods
- if the reference level stays the same for a while, as it will usually
- do if you request it, or if the level is high enough.
-
- When you manipulate action by deception, you are betting that your
- deception will create a perception in the other person that will lead
- to the action you want. This may work and it may not. Furthermore,
- it's basically an open-loop process, in that deceptions typically
- follow a formula which is applied in the expectation that the
- predicted action will occur. Deceptions clearly do not work on
- everyone; those who employ them rely on population statistics to find
- people who do in fact behave as the manipulator wants. Advertizers pay
- to modify the perceptions of tens of millions of people, but they're
- satisfied if perhaps one person out of a hundred is deceived, and one
- out of ten of those produces the intended action as a result. In an
- audience of 10,000,000 that's 100,000 sales.
-
- So manipulation by deception resembles an S-R phenomenon more than a
- PCT phenomenon. A theory of manipulation by deception can't predict
- actions of an individual. Its predictions would be statistical, and
- therefore would apply only to a population. And proving that
- manipulation by deception actually works would be extremely difficult,
- because you have to take into account the people who would perform the
- wanted act (buy the car, drink the beer, hand over the money) even
- without the deception. A theory of manipulation by deception exists on
- the same plane as theories of diet and psychoanalysis.
-
- Manipulation, therefore, seems to mean at least two things that are so
- different from each other than they don't intersect anywhere. The
- rubber-banding kind is reliable, precise, repeatable, and predictable
- from basic principles. The deception kind is statistical, unreliable,
- and dependent on peculiarities of individual experience. It is
- possible to predict the outcome of the rubber-banding kind of
- manipulation for an individual. It is not possible to predict (in the
- same sense) the outcome of a deceptive manipulation of an individual,
- either from experience or from basic principles.
-
- Are there any forms of manipulation that have the qualities of a
- control-type manipulation? Or are the others all statistical and
- relevant only to populations?
- ----------------------------------------------------------------------
- --
- Eric Harnden (920909) --
-
- The Dynamo approach get too close to programming for most people to
- learn it easily. Anyway, it's expensive, too, isn't it?
-
- I think we should start leaning on Pat and Greg Williams to get their
- simulation toolkit running. They write in C, so someone with a Mac
- could adapt it for the Mac world. And maybe there are some mainframe
- types out there who would do the same for other machines. Since this
- will be a language for talking about PCT, it won't need all the
- industrial whistles and bells.
-
- > ... all of which is to say that i think we've got workable
- >alternatives to ascii graphics for the effective communication of
- model >structures.
-
- That's what we need, all right. But cost and skill at programming are
- still barriers for lots of us.
- ----------------------------------------------------------------------
- -
- Gary Cziko (920909) --
-
- The paper would be easier to read if you would do a search on "and"
- and convert most of them to periods and a new sentence.
-
- I hope it works -- sounds like you mashed down pretty hard on some
- toes, there. Your appeal to those who AREN'T satisfied is probably the
- most realistic way to go. The majority of people who publish in any
- field are going to defend what they're doing as perfectly adequate.
- Proving that it's not will not deter them for a moment. They don't
- even bother to understand your arguments; they're too busy thinking up
- how to demolish your conclusions.
-
- The real test will come with publication.
- ----------------------------------------------------------------------
- --
- Thomas Baines (920909) --
-
- >I'm not very far along with PCT so you'll have to wait 'til I catch
- up. >Try this, however...
-
- I can't even understand it. How can I try it?
-
- >This reception/emission sets up patterns which reenforce one
- >another in some places & cancel one annother elsewhere.
-
- It's much too soon to start offering alternatives to PCT. Most people
- take about 2 years to get comfortable with control theory, so they can
- reproduce it from scratch from basic principles. I have a feeling it
- would take me at least that long to be able to reproduce your
- theoretical presentation from basic principles, particularly as you
- haven't said what they are.
-
- The kids are fine. Roxanne (the dog) has become fond of
- chasing rabbits. I want to alter that behavior. I think
- she is controlling for the hunt & chase. I don't think it
- is for food. I am trying to find a way to get her to
- substitute tennis balls & frisbees. Stay tuned.
-
- I'll bet you aren't going to do this by setting up reception/emission
- patterns that reinforce in some places and cancel in others. I'll bet
- you're going to do it pretty much the way PCT would describe it. That
- is, if we on this net could ever agree on how PCT would describe it.
- In the mean time, let us know what you do and how it works. That, at
- least, is DATA.
- ----------------------------------------------------------------------
- With small pups, start with two boxes which fit together to
- make a covered sleeping space just slightly larger than the
- pup. They will quickly be ready to wait for the yard or the
- paper.
-
- I said you were a genius.
- ----------------------------------------------------------------------
- -
- Best to all,
-
-
- Bill P.
-