Who's confused?

[From Bill Powers (2003.02.16.1706 MST)]

Rick Marken (2003.02.16.1400) --

>Actually, Rick's model is running. I'm attaching it to this post. People
with Excel can >run it. It's not very well documented but it does produce
plots showing leakage, Q/Q' >and GNP varying over time. You can put a
leakage value in (in the purple cell under >"Enter Leakage value") and that
serves as the amplitude of a sinusoidal variation in >leakage over time.

How do you know it's running correctly, if you haven't changed the things
that were wrong with it? All models do _something_.

>The translation of Econ003 into Excel has created some questions that
perhaps Bill can >answer. Right now, my main question concerns some
statements in procedure >wconsumerlevel1:

Vw := Vw + (Gw-Uw-Dw*Vw)*dt

Vw := Vw + (Gw-Uw)*dt

These two integrations of Vw follow each other immediately. Is this right?
Seems like only one (the first one) is needed.

That's one of those corrections you make and then forget to take out the
incorrect line. You are right, only the first line is needed. The effect of
leaving in the second line is to double the effects of purchasing and using
goods -- as if twice as many were being purchased, and twice as many were
being used, per unit time. Doubling the purchasing doubles the gain of the
associated inventory control system, so has almost no effect (just makes
control tighter). Doubling the usage makes the worker work twice as much,
however, which is not a negligible effect. Good catch. I don't believe that
error is present in Econ004.

Wish I knew how to run those models in Excel.

Best,

Bill P.

[From Rick Marken (2003.02.16.1400)]

Bill Williams (UMKC 16 Feburary 2003 9:39 AM CST) --

At present, the basic core of your model appears to be running corectly. So far, however, you haven't used the model to develop a testable hypothesis. Rick's model isn't running, he's still has a basic problem with internal consistency. I've got a very simple model running. I like it because it is so simple I think it may actually be possible to explain to people what it is doing and how it is doing it. It's generating hypothesis four so far have been tentatively confirmed using Bureu Labor Stat figures ( 200 data points 1950 -200 ).

Actually, Rick's model is running. I'm attaching it to this post. People with Excel can run it. It's not very well documented but it does produce plots showing leakage, Q/Q' and GNP varying over time. You can put a leakage value in (in the purple cell under "Enter Leakage value") and that serves as the amplitude of a sinusoidal variation in leakage over time. Press the "Run" button to run the model. You will see the numbers change for a few iterations and then I turn the display off to do the calculations. Right now the leakage amplitude is set at .02 (average of 2% leakage, varying from 0 to 4%). Try entering a different leakage value (say .05) and run it again and see what happens to the GNP growth curve.

Although the program is not well documented at least you can see that it does indeed work. It would be easy to try fitting the model to actual GNP and Q/Q' data by finding the appropriate variation in leakage over time. I haven't done that because the model is fundamentally flawed for the reasons Bill gave. the main problem is the the composite producer/consumer controls PQ (a product) rather than Q and expenses (PQ) separately.

I am currently trying to put Bill Powers' model into spreadsheet form. I think his approach is better than mine because the variables being controlled seem more realistic (as does the model of the economic environment in which this control occurs).

The translation of Econ003 into Excel has created some questions that perhaps Bill can answer. Right now, my main question concerns some statements in procedure wconsumerlevel1:

Vw := Vw + (Gw-Uw-Dw*Vw)*dt

Vw := Vw + (Gw-Uw)*dt

These two integrations of Vw follow each other immediately. Is this right? Seems like only one (the first one) is needed.

Best regards

Rick

EconModel2003.xls (99 Bytes)

···

--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313

[From Rick Marken (2003.02.16.1545)]

"Williams, William D." wrote:

I wouldn't call a model which you acknowlde is "fundamentally flawed" a working model.

Why not? It works.

According to you I suppose anything that compiles or isn't kicked out by the interpreter is a running MOdel ???

No. In the case of my model, the model is designed to control P'Q' and P'Q'-PQ' in the face of disturbance (variations in the leakage variable, which subtracts from P'Q'). It does that (it controls successfully), so in that sense it works.

What does your working model do? Can you post the code and .exe file so we can see it in action?

Best

Rick

···

--
Richard S. Marken
MindReadings.com
marken@mindreadings.com
310 474-0313

[From Rick Marken (2003.02.17.0850)]

Bill Powers (2003.02.16.1706 MST)]

Rick Marken (2003.02.16.1400) --

>Actually, Rick's model is running.

How do you know it's running correctly, if you haven't changed the things
that were wrong with it? All models do _something_.

The model is running correctly with respect to the goal I had when constructing
the model, which was to have the model behave in the way that TCP's equations
behave. So my model behaves correctly in the sense that it produces values of
Q/Q' and inflation, both as a function of leakage rate, that correspond to those
predicted by TCP's equations. It also behaves correctly because it successfully
controls for P'Q' and P'Q'-PQ'. I agree that the model probably doesn't work
correctly in the cosmic sense, one reason being that it doesn't control what
people are likely to actually be controlling. But I think the ultimate way to
test the correctness of a model is to see how well it accounts for data. My model
has not been tested against data yet but I imagine that if it were it would do
poorly (I would hope, given its flaws). And your model has not been tested
against data yet, either. So we don't yet know whether it is any more correct than
any other model. But Bill Williams says he has a working model that he has
successfully tested against data so I think it behooves us to take a serious look
at that model. So I would really appreciate it if Bill W. would post that model
(as source and as an .exe file, if possible). It would also be nice if he could
post a graph showing the fit of model behavior to data. I think that such a
demonstration of fit of model behavior to data would be _very_ nifty.

Best regards

Rick

···

---
Richard S. Marken, Ph.D.
Senior Behavioral Scientist
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

[From Bill Williams UMKC 16 Feburary 2003 9:39 AM CST]

Cc:

[From Bill Powers (2003.02.15.2038 MST

The expressions you cite are avery confused (mathematically) mixture that doesn't do what is intended.

How a rate varible is represented is represented is a matter of convention-- as long that is as consistency is maintained. I had the lesson about units in my first physics class, I understood it then, I haven't forgotten what it means now. What you said in the last post doesn't contribute anything to clarifying the issues. So, I think there is a very real question involved here about what is going on. Who is confused about what and why there is this confusion.

You've pestered me now for more than a decade and half about this business about the macro model. It's repeatedly begun and ended the same way. You always begin by seeking my help in regard to your interest in the economics question. My contribution has consistently been first to your dad, and now to you, that if you want to pursue this as "a serious endeavor" that rather than doing so as an exercise in issolated reflection that the exercise become a part of the on-going discussions in the intellectual communities concerned about econmic issues and economic theory. Efforts given to such exercises in issolation do not offer much hope that whatever results are generated will make any contribution to the development of economic thought. An example of this is what happened to your dad's efforsts. How many of his books were sold, and how many times has it been cited? There are no question costs involved, if you choose to pursue economic questions as a part of a some community of inquirers. Membership in the community involves something in the way of a pattern of behavior that maintains the community as basis for inquiry. For your dad this sort of suggestion was unthinkable. He'd worked it all out, now people could read what he'd written and profit by it. And, that was that. With your dad, it took about 20 minutes to see that he was headed, and very determined to go, at least in regard to his interest in econmics, in a direction that was doomed. Recently when you commented upon his well thumbed copy of Keynes' _Gen Th_ I was quite surprized. Given what he's said about Keynes I assumed that he'd never read a page of Keynes's stuff. ANd, to paraphrase Rick, (with some modifications ) "Just because somebody's reading a book doesn't mean they are reading the book." What ever your dad's preconceptions were they didn't permit him to read Keynes with understanding. Even with my very brief contact with your dad, I'm confident when he did attempted to read Keynes he did so with a great deal of determination. But, what he got out of Keynes was a product of the preconceptions with which he read Keynes rather than what he could have gotten out of the _Gen TH_. The result was what he said about Keynes, if by chance someone read what your dad had written immeadiately indicated the presence of massive misconceptions about contemporary economic theory. Other than Rick, and the Power's family, I don't think your dad persuaded anyone. Are there actually people out there who think that TCP's leakages thesis is an effective way of thinking about macro economics?

Now, to turn to your case. A while back in a posting you asked for an indication of how many people are following the Econ thread to the extent that they look at the code, compile the program and fiddle with the simulation. I don't recall that anyone answered that request. So, I assume that aside from myself and Rick we are largely talking to ourselves. THis isn't intended as a mean spirited jab. I'd rather more people were interested but my assessment, which anyone who wishes to can correct, is that the econ thread isn't of much interest. At least it isn't of such a character that many people are inclined to participate by commenting upon or contributing to it. Runkel and Nevin are the only exceptions, to You, Rick and myself.

At present, the basic core of your model appears to be running corectly. So far, however, you haven't used the model to develop a testable hypothesis. Rick's model isn't running, he's still has a basic problem with internal consistency. I've got a very simple model running. I like it because it is so simple I think it may actually be possible to explain to people what it is doing and how it is doing it. It's generating hypothesis four so far have been tentatively confirmed using Bureu Labor Stat figures ( 200 data points 1950 -200 ).

I'm surprized you've managed to get your model working, given the direction from which you've approached the problem. As I've said, while I haven't yet checked the model out in detail, I'm confident its working correctly. When I have time I plan to strip out the display code, and re-write the program so I can see more easily what is going on. Until that time I won't be entirely confident about the details. But, I don't that's important. Knowning that the program works, given time I could rewrite the model starting from stratch.
My question, however, which I've repeated from time to time, is where is this project going? Or, what seems to me to be a closely connected question, where is it now? Who at present, in your opinion, understands it? And, what sort of path do you have in mind to increase the number of people who understand it?

The problem as I see it is that the only potential audience you've got for the forseeable future will want to know right up front before they spend any time attempting to understand what you've done is "How would you express this as a Keynesian model?" When a couple of weeks ago I was attempting to understand what you'd done ( While the code has extensive comments, to see what you've done requres looking past the parts of the program that create the display. And, there is no documentation, no flow charts, etc to explain what is going on. ) I decided it would be easier to write my own version, a Keynesian version, of a control theory macro model than it would be to understand your model. And, it only took half an hour or so. But, then it is a model intended from the start to be so simple, that as Geof Harcourt says,
"You can explain it to the children on sunday morning." It was only after I had my model going that I was able to understand what you'd done. When you talk about economic questions you do so in a way that, aside from whether you are actually making mistakes, certainly sounds as if you think about economic questions in a period-sequence format. The result at least in my pereception is that to undertand how Econ004 works requires looking beyond the way you describe it. And, I wasn't able to do this without a working model ( a Keynesian model ) from which to think about the process. Unfortunately
despite my best efforts, I can't explain the Keynesian system to you. For whatever reasons which I don't pretend to understand, when you attempt to think about the Keynesian system it doesn't make any sense _for you_.

When I felt I coud I've attempted to avoid becoming involved with you in such disussions because I've sensed that behind it lay some incrediablely powerful emotions in connection with your conception of political and economic affairs, and with your conflict with your dad. To enter into this realm with you seems to unavoidablely to involve encounters with monsters and the ghosts of a struggle between you and your dad. One of the lessons of a control theory approach to problems is avoid such conflicts if possible. I don't percieve your last post as being in anyway malicious. But neither do I understand what you've said as having anything to do with me, or what I was saying. The argument which I was expressing was that it is possible to re-write the Keynesian system as a simultaneous system, insert an agent ( based upon control theory into the system ) and escape from all the confusions which have characterized previous efforts to develop macro-economic theory.
If the control theory group had a larger group interested in, and equiped to work on the technicalities underlying the problem of economic theory, I think it might be possible to identify where the problem lies and get it sorted out. But, things being what they are, I don't see anyway we can resolve the economics issue between us. You'll have to find somebody else to talk to about economics. What I'm doing, apparently because it is expressed in terms of the Keynesian system, doesn't make any sense to you. I don't see that there is anything that I can do about this.

I understand a bit about mental blocks myself. When I finished the disertation I experience an enourmous frustration because there was nothing further I could do with the work that had gone into identifying the proper mode in which to express econmic theory. At the time it looked like a completely abstract formalistic academic exercise. It wasn't until I took a look at the apendix to BCP and saw that you'd used the proper equational format, that the work had any constructive use. Negitively the dissertation had served as source for criteria which rejected all the stuff I was reading in psychology as faulty. Maybe Weiner's got it right, but I couldn't follow the way he expressed stuff. The point is I could have inserted control theory agents into a Keynesian system long ago-- if the experience of running into a wall after the dissertation hadn't been so painful. So I wasn't inclined to revisit the painful topic of macro economic theory and instead worked at doing free standing applicatins like the Giffen, and other paradoxs applications. After having accumulated a dozen or so such applications, I was beginning to think enough was enough. It was only when I was trying to figure out what your model did that I began thinking "It ( a macro model ) doesn't have to be this complex." that I saw a way of doing a really simple, but still fundamentally correct, macro model. Seeing the Giffen model work will probably always be the most significant experience of my life. But, aside from that seeing the Veblen/Dueseenberry macro model comes close, probably because it makes use of the work on the dissertation in a direct way which the Giffen effect model doesn't. In doing so seeing the model run in the way I anticipated, wipes out was has been an enourmous error in which the dissertation looked to many people like utter nonsense. I never thought so, but it was, at the time, discouraging.

I've been asked to prepare the Veblen/Duessenberry model as a chapter in a book the economics department here plans to publish. As it looks now neither the model or the chapter will benefit from your criticism or the positive contribution you might make to it. As you know, once you get something running it isn't really rocket science. But, economics is sufficiently complex that it can be difficult to see how something works-- espeically if you approach the question with preconceptions that interfer with the argument.
And, that it seems to me is what's happened to you.

best

Bill Wlliams

···

Subject: Re: Savings and investment

[From Rick Marken (2003.02.16.1400)]

You say your model is running:

Although the program is not well documented at least you can see that it does indeed work. It would be easy to try fitting the model to actual GNP and Q/Q' data by finding the appropriate variation in leakage over time. I haven't done that because the model is fundamentally flawed for the reasons Bill gave. the main problem is the the composite producer/consumer controls PQ (a product) rather than Q and expenses (PQ) separately.

I wouldn't call a model which you acknowlde is "fundamentally flawed" a working model. According to you I suppose anything that compiles or isn't kicked out by the interpreter is a running MOdel ???

Bill Williams

···

-----Original Message-----
From: Richard Marken [mailto:marken@MINDREADINGS.COM]
Sent: Sun 2/16/2003 4:01 PM
To: CSGNET@listserv.uiuc.edu
Cc:
Subject: Re: Who's confused?