[Martin Taylor 2008.12.14.11.31]
[From Bill Powers (2008.12.13.1546 MST)]
Martin Taylor 2008.12.13.11.02 --
Living control systems steal energy from other sources and reject the final products to their environments, so they aren't really closed-loop systems either -- is that your argument?
Not at all. What I am saying is that you can't always ignore the energy source, at least not for control systems that have to act in a physical world, as opposed to the simulation world.
True. It would be fairly easy to add the energy requirement, if you want to do so, and make provision in the model for both using energy to control things and acquiring more energy when a local store of it is depleted.
You miss the key point, that even with a single simple control loop, the requirement for energy to drive the forces that counter the disturbance ensures that the system is non-linear unless the energy source is infinite. Why this matters is that when the simple control loop exists in an environment that contains other such loops whose actions affect either the environmental feedback function or the disturbance, non-linearity is almost sure to make the dynamics of the system become chaotic.
This, inevitably, would lead into a model of economics. One form of symbolic energy is money because money, for large numbers of consumers, represents physical work done or at least time spent away from other possible activities in order to manage the expenditure of physical energy to produce things. Note that this also covers barter. To make money, you have to take time off from doing the pleasurable or vital things that involve spending money.
I don't think you can properly substitute money for energy. The conflicts you mention are real, and to a large extent all our control activities are in support of what must be an intrinsic variable, a sufficient energy flow. It is also true that localised entropy changes are associated with energy flow. That association implies that if Bagno is right (as I believe he must be almost as strongly as I believe PCT must be correct, and for the same reasons), then the use of money as a medium for communication about structure means that it will also be associated with the deployment of energetic resources. In simpler language, if you use money to acquire structure in the form of something someone else built, you both reduce the energy you would have otherwise had to use to build the structure, and import the structure itself. The two go together qualitatively, but quantitatively they only correlate. You can sort-of substitute money for energy if money is actually an information transmission medium, as bagno argues it must be.
There's a difference between dealing with one control loop and many. If I eat all your food, you don't control very well for very long subsisting on my waste products.
That concept would make a nice basis for an economic model. It introduces the idea of interpersonal competition and conflict, and scarcity as well. This is not hard to bring into a model.
Yes, indeed. Have you actually read the paper by Bagno that I have on occasion recommended here, and that has been dismissed because it didn't include a PCT description of the behaviour of people?
I tried to read it, but was unable to get far with it. Information theory doesn't impress me as useful for making models, though I can't give you enough rational basis for that to make any difference in what you think about it.
Perhaps you would find my own take on Bagno, at the same URL, a bit easier to read. Having re-read it last night, I find gaping holes in my 20-year-old logic in some places -- notably the final conclusion -- and the language is rather naive, but nevertheless I think much of it still holds up moderately well.
You say: "Information theory doesn't impress me as useful for making models, though I can't give you enough rational basis for that to make any difference in what you think about it." I tend to agree with the first part of the comment. It's closely analogous to saying "Thermodynamics doesn't impress me as useful for making models", which I also think is true. What Information Theory and Thermodynamics both do is to constrain the possibilities for models, and to provide one of many possible ways of analysing the performance of a particular model, often in comparison to some idealized optimum model.
What Bagno does is not to model the economy, except in the most gross macro scale (the kind that Rick likes to use). Rather, he shows that no matter what the model, certain effects will happen. The thermodynamic equivalent would be to say that if you try to drive an engine by connecting it between a block of hot metal and a block of cold, the engine will run for a while, but will eventually run down. It doesn't matter what kind of engine it is (the model); it will run down and eventually stop unless something external continues to heat the hot block and to cool the cold one.
One of the key ideas is that of trust in the money supply, measured as uncertainty as to the future value of money measured against the structure a money unit can buy you.
I would say that trust and faith are exactly what you need when you're uncertain about the future but are required to act anyway. If you had a good model, on the other hand, you wouldn't need trust or faith because you would know that the model will provide the best basis for action you can find, so there's no need to go into conflict about the future. When you give up on modeling and succumb to uncertainty, you need trust just to keep trying.
If you have a chaotic system to model, then even if your model is exactly perfect in every detail, you still won't be free of uncertainty about the future unless your knowledge of the present is also infinitely precise. Your perfect model might well give wildly divergent results for the future for two very close approximations to the present conditions. Uncertainty as to what will happen is inherent in the perfect model.
That's not really the point, however. Reduction of uncertainty always means work, no matter what the domain. How much work for how much uncertainty reduction is the question that applies in every transaction.
Almost nobody has very much information about the current state of influences that will affect economic variables, whether they have a good model or not, and for most people involved in everyday transactions the uncertainty has more to do with lack of knowledge of present conditions than poverty of model. Usually, it's not a bad bet that inflation will continue to be near what it has been in the last little while, and as variable as it has been in the last little while. Without investing a lot of effort (and expense) into getting better information about current conditions, it's hardly worth much to work on getting a better model. That's for academics, not for me wondering whether to shell out on a new car now rather than next year.
Fundamentally, money is a communication medium. It allows one person to tell another how valuable some structure is.
I doubt that. It tells another person what its value to you is, but it says nothing about the objective value of a structure. I suspect that this is what you meant. Is there any such thing as objective value?
No. Why would you think there might be, or that I might suggest such a thing? Money allows one person to tell another how valuable some structure is to them, in linguistic terms that mean something to both parties. That "something" means "value to me" for both parties. As noted in the next paragraph you quoted.
"Structure" can be embodied in an object or an idea, but it is something created by a living organism, controlling its perceptions. The value isn't the same to a producer as to a buyer, and if the value (translated into the perceived value of money units in both minds separately) is lower for the seller than for the buyer, a free transaction is possible.
I don't think the seller needs to know anything about the buyer's values or vice versa.
Of course not. And I explicitly said not. All the seller needs to know is that what the buyer is offering is more valuable to the seller than the thing being sold. The seller may guess that to the buyer the thing being sold is more valuable than the compensation being offered, but for the transaction to take place that's not required. For the buyer, the thing being sold must have more value than the compensation she is offering. Clearly, if there was an "objective value" for both the thing being sold and the compensation, a pair of values agreed by both parties, no transaction would ever take place.
You don't need to know the cause of a disturbance to counteract its effects. If I want to buy something from you and you ask too high a price, all I know is that I don't want to pay that much. I don't know what value you actually place on it; I find out by raising my bid a little until I exceed my limit, or until you surprise me by saying "OK." If I hit my limit I never do find out what value you placed on the item -- I get only a lower limit for that transaction on that day.
You describe part of the conflict that usually exists in any transaction, between the control systems in the buyer for quantity of "thing being sold" and for quantity of "compensation" (e.g. money). "Value" is defined by where the conflict is balanced.
Of course having gone through that bargaining process, I can tell you afterward approximately what value you placed on the item. That's no help in future transactions, though it might give me a statistical idea about what sort of opening bid to make for the same item in the near future. Even that idea's iffy -- if you sold me the next to last one, your price for the last one imnmediately after the first sale might be a lot higher.
To which I can only repeat "Of course". And if in the interim I won a lottery jackpot, the value of a dollar to me has declined dramatically, though the value of the item in question may not have changed at all, so I might charge a considerably higher price than last time. Or, my new bank balance may be higher than my reference level for it, making my value for an extra dollar go negative, so I might give you the item along with some money to take it. Values do change over time.
I might know that energy is conserved, and that in a given situation the sum of potential and kinetic energy is constant. But you have to tell me how the objects involved are going to move next; I can only predict that however they move, the sum will be constant.
Which tells you that if you create a model that produces a different result, that model won't work in the physical world. Build a model that says cars get 60 mpg by burning water into H2O2, and you have a model that doesn't work thermodynamically. And that's the kind of thing Bagno is telling us about the economic world. If (in a closed economy) you don't have either inflation or a continual (and therefore eventually unsustainable) increase in volunteer work, you will have an economy that runs down in the same way that the engine running between a hot and cold block of metal runs down. Since money is created by debt, either personal, corporate, or government debt must persist to support the required inflation. Nothing is said about how these requirements are met, and though the extension seems to follow naturally, Bagno doesn't deal with open economies that can import structure (as Germany did for a couple of decades during the Wirtschaftwunder period) while exporting entropy to their trading partners (the excessive inflation of the rest of the West during the same period).
Bagno isn't model building, in your sense. He's giving you constraints that tell you whether your model might behave as you hope if it was actually put into effect in the real world. A model that says inflation must be kept too low is one that, if put into effect, would lead to depression. A model that says inflation should be kept at some too-high level is one that would fail because it would lead to runaway inflation through positive feedback of the perception of uncertainty. And so forth. It's up to you to make your model. Having done so, you may be able to determine whether it is viable within the inviolable constraints.
Generalization is an operation that generally has no inverse. That's my generalization about generalization.
That is a Major conclusion with a Colonel of truth.
Martin