Clayton : If you can convince someone to pay you to engage in useless speculation then more power to you. But my guess is that - outside of tax-funded grants - you’re not going to find very much demand for such “services.”
I don’t recall asking for this such grants. Also, I have a different way of viewing it, I view this line of discussion as useful. I’m quite familiar with both the computability issues we face in this realm, and the lack of constants, and the sheer volume of raw real world data that encounters each individual. I’m also familiar with the incompleteness issue of information, meaning that each individual has only an incredibly small subset of data with which to work.
Clayton : All these people are impeding you from getting what you want so, in the purely materialistic view of things, they’re all the same to you. Obstacles to your desired state.
This is the whole reason why Mises’s idea of human action is so profoundly unpopular. It is simply not allowed to reason about humans as frankly purposeful beings.
I’m not sure the distinction between frankly purposeful beings and purposeful beings.
It might make sense to break this discussion of the different categories of social action out as a separate topic.
So, social action is action which imputes the categories of human action (including the intentional stance) to the “target” of the action. I could not engage in social action without imputing the categories of human action to the target. Let’s not forget that social action still has the same categories as the more general category of action, including that it uses means to achieve an end. A social action then necessarily is designed to use a human being as means to an end. If I pick up your body and move you, I am not imputing intentionality to you. I am not engaged in social action, I am engaged in a non-social action. Moving your body by picking you up, does not involve imputing human categories of action to you. However, I would most likely have to knock you unconscious first, which in and of itself would be a social action. Because it imputed intentionality to you. I would have to imagine that you are able to act, and would resist my movement of you. Therefore, knocking you out would be considered an action to render you “unintentional”, just like a rock, or a dead animal, or a stream of water.
Social action imputes choice, knowledge, preference, understanding, valuation, etc. to the target of the action. So one could diagram a generic form of social action. My action is intended to alter your action. The only way that my action could alter your action is by changing your internal understanding of the future. THE ONLY WAY. If I don’t change something about how you perceive the future. Now it simply remains to be seen if we can find differences between actions I take in the way your perception is altered.
If I point a gun at you to compel an action, then there are 3 realities in play. One where nothing happened which presumably was the state of reality that you prefer above the other 2. The reality in which I shoot you, the least valued reality. And the reality in which I don’t shoot you and you do the thing I want. This third reality is necessarily more preferable than the one where I shoot you and necessarily less preferable to the one in which I leave you alone. So, we can demonstrate not only that you are worse off before the interaction, but that the I was the proximage cause the detrimental effect. So the hierarchy of realities is (get shot) < (give me 5$/don’t get shot) < (do nothing)
The second example I’d use as a categorically different, type of social action would be an exchange. I offer you something for money. In this case 3 realities are in play also. The first reality is the one in which no action by either of us is taken. The second reality is one in which you do the thing I want, which in this case is to give me 5$. That reality (as above) is less preferable to the one in which you take no action. But in this case I present a future in which I have 5$ and you have a pack of cigarettes from me. This reality is now one that you prefer to taking no action, and suddenly I’ve found a way to cause the action I wanted. So the hierarchy of realities is (give me 5$) < (do nothing) < (give me 5$/get pack of smokes).
Those are categorically different but in both cases they are social action, in that I’m using the target of my action as means to an end, get 5$. Obviously, in microcosm, the first case of the threat is a higher profit to me. But, of course action is not performed in a vacuum. There is history and memory and thus consequences in other social settings to both actions.
I think it’s useful to diagram this stuff in a more formal way. We can see things we might not see from word descriptions alone, just like we can see things in other disciplines by using math.
Which leads me to a couple of other objections you make :
David B : Think of it like using CAD tools to design cars,and then running the models through aerodynamic simulations or crash simulations.
Clayton : But we already know that’s impossible. Please see my post above regarding the difference between input-output functions and causal systems.
Let me address this specifically. You know this is not true. Those simulations are useful, and they never perfectly or accruately model real-world conditions. They model with sufficient general accuracy to inform us of aggregious errors, or potential unforseen consequences that were not anticipated. We use weather prediction systems every day. We know that weather systems are dynamical and chaotic, and that the predictions are statistical models and are notoriously inaccurate beyond a very short period. Imperfect prediction has value, it just has less value than perfect prediction. The myth as you point out is perfect prediction. But that doesn’t remove the usefulness of any data that we get that informs us of pressures in a dynamical system. For example, in the weather field. Knowing that the temperature of the ocean water in the gulf of mexico reaches 89 degrees on average is very important information if we know something about how water temperature plays a role in hurricane formation. The simulations we run show us that the severity and strength of storms is much higher when the average temperature of the ocean is higher than when it’s lower.
We still don’t know where and when the hurricanes will develop and what paths they will take. We may even be wrong about the intensity, but we know that one condition is in place and that it contributes to higher intensity (and thus destructive) storms. Is that useful? Again go back and look at what you said about input/output functions and causal systems. Then consider this quote:
Hayek : But ultimately, of course, it goes back to the assumption of what the economists pleonastically call “given data,” this ridiculous concept that, if you assume the fiction that you know all the facts, the conclusion you derive from this assumption can apply directly to the world. My whole thinking on this started with my old friend Freddy Bennan joking about economists speaking about given data just to reassure themselves that what was given was really given. That led me, in part, to ask to whom were the data really given. To us, it was of course [given] to nobody. The economist assumes [the data] are given to him, but that’s a fiction.
This is a great point. But wait a minute think more deeply. One of the mistakes of historical anticipation of the future of weather prediction was that we would get better and eventually know the outcome. That we would somehow get a linear progression in the time period over which it gave us accurate predictions, as we got more and more precise in our modeling and in our data collection. But was we know now from chaos theory this is not correct. The more complex the system the more rapidly slight variations in initial conditions become massive large scale differences in outcome. We suffer from an information incompleteness issue. Again, as I said before the only system with sufficient computational power to capture all the data and produce the actual outcome is … reality!
But let’s dig even deeper. Dynamical systems have other features, and these are the ones that allow for the human brain, and for the type of “imprecise” calculation that humans do about the future that leads to success. Feedback signals regulate the conditions of the system. Postivie feedback loops reinforce a behavior, and therefore tend to destabilize the system. Negative feedback loops act as a degenerative force. If you have a stable system, there is a high amount of negative feedback in the system. Tracing feedback loops in a system as complex as human society isn’t sufficient to give specific and precise prediction of the timing and location of events in reality in the future. But we could use it to tell us the equivalent of what an extremely warm ocean temperature tells us.
Clayton : Modern computers are abysmally stupid. NB: I’m a computer engineer.
I program for a living. Computers aren’t actually that stupid. But we, as programmers, are. The substrate for computing and the software running in it that nature has produced in our brains, is phenomenally more advanced by comparison. So, we’re stupid in terms of producing computational devices and algorithms, nature is far more effective (in a non-intentional way.) I think I wasn’t clear. If you watched the video presentation I linked, Gerard Sussman’s presentation said that the human brain can do in 7 operations things we can’t even conceive of doing at all with computers. He used a pattern recognition problem, that’s incredibly simple for the human mind, but is damn near impossible today with modern programming techniques. The rest of his presentation focused on how he thinks one of the problems is that we get stuck with a programming paradigm, (logical, functional, object oriented, aspect-oriented, model-driven) and then drive it into the ground. We have hammers and try to convert problems into nails. He proposed that it might be that if we’re going to figure out how to create programs that can solve problems of these higher complexities we need to be experimenting with a wider variety of computational paradigms. And he presented a couple. He presented them based on some intuitive observations of how people actually solve problems in reality. The signal propogator one was an interesting one, drawn from his experiences teaching electrical engineering courses. (I know a little electronics as a consequence of my dad’s EE degree and my home environment, 4+ years of electronics in high school, and occupational training later in the military).
Clayton : Please read this. You have an extremely naive view of the issues.
I’ll continue the discussion that Hayek was having :
Hayek : In fact, there’s no one who knows all the data or the whole process, and that’s what led me, in the thirties, to the idea that the whole problem was the utilization of information dispersed among thousands of people and not possessed by anyone. Once you see it that way, it’s clear that the concept of equilibrium helps you in no way to plan, because you could plan only if you knew all the facts known to all people; but since you can’t possibly know them, the whole thing is vain and a misconception partly inspired by this concept that there are definite data which are known to anyone.
High : Do you feel that mathematics has an important role to play in economic theory?
Hayek : Yes, but algebraic mathematics and not quantitative mathematics. Algebra and mathematics are a beautiful way of describing certain patterns, quite irrespective of magnitudes.
There’s one great mathematician who once said, “The essence of mathematics is the making of patterns,” but the mathematical economists usually understand so little mathematics that they believe strong mathematics must be quantitative and numerical. The moment you turn to accept this belief I think the thing becomes very misleading–misleading, at least, so far as it concerns general theory. I don’t deny that statistics are very useful in informing about the current state of affairs, but I don’t think statistical information has anything to contribute to the theoretical explanation of the process.
I mean, all these things I’ve stressed–the complexity of the phenomena in general, the unknown character of the data, and so on–really much more point out limits to our possible knowledge than our contributions that make specific predictions possible.
This is, incidentally, another reason why my views have become unpopular: a conception of scientific method became prevalent during this period which valued all scientific fields on the basis of the specific predictions to which they would lead. Now, somebody pointed out that the specific predictions which [economics] could make were very limited, and that at most you could achieve what I sometimes called patterned predictions, or predictions of the principle.
This seemed to the people who were used to the simplicity of physics or chemistry very disappointing and almost not science. The aim of science, in that view, was specific prediction, preferably mathematically testable, and somebody pointed out that when you applied this principle to complex phenomena, you couldn’t achieve this.
This seemed to people almost to deny that science was possible. Of course, my real aim was that the possible aims of science must be much more limited once we’ve passed from the science of simple phenomena to the science of complex phenomena. And there people bitterly resented that I would call physics a science of simple phenomena, which is partly a misunderstanding, because the theory of physics ends in terms of very simple equations. But that the active phenomena to which you have to apply it may be extremely complex is a different matter. The models of physical theory are very simple, indeed.
So far as the field of probability, that’s another part. But it is this intermediate field, which we have in the social sciences, where the elements which have to be taken into account are neither few enough that you can know them all, nor a sufficiently large number that you can substitute probabilities for the new information.
The intermediate phenomena field is a difficult one. That’s a field with which we have to deal both in biology and the social sciences. And they’re complex. They become, I believe, an absolute barrier to the specificity of predictions that we can arrive at. Until people learn themselves that they can’t achieve these ends, they will insist on trying. They will think that somebody who does not believe [this specificity can be achieved] is just old-fashioned and doesn’t understand modern science.
Please read through that dialog very carefully. Economics is like every other science, in that the only thing those sciences can deduce is very simple rules. The only difference is that they do have quantities at the root, and they can in fact model and predict some very simple phenomena. We call them experiments. Howver, they can only predict in very narrow and very specific conditions, and as soon as the predicted phenomena enters the world the system becomes impossibly complex. We actually have decent tools and techniques for dealing with dynamical systems. The problem is that the rest of science is still stuck imagining that they’ve predicted more than they have, not that economics is trying to do something uniquely impossible to economics. Weather prediction is the best example of the kinds of advancements we could make in economics. I’ll reiterate, science is like economics in that it is not able to predict the effects of dynamical systems, and for the same reason : insufficient data, and inaccurate rules.
I’d like to come back to my CAD/simulation example one more time. I think you missed something I said. Those models are less about describing what will happen, and more about identifying failure conditions the knowledge we do have implies that we might miss because of our inability to model the conditions we do know about in an effective way. It’s about understanding how the chaotic conditions of a real environment interact with the structure and whether or not the structure is sufficiently stable to tolerate these stresses. Our governmental institutions are NOT stable, the politices they put out destabilize the social environment. We can and do present this interpretation and understanding in word explanations, but I think we can do more. It’s not about predicting the exact reality a system will produce; it’s about demonstrating how real world phenomena can produce predictable failures traceable to specific structural features. We can capture the Peter Schiff example in a more formal way. We could probably simulate or simple models of these phenomena. We’ll always have the chaotic unpredictability issue. What we’re looking for is to introduce featues into the dynamical system that add stability and reduce or eliminate known risks; we’re not looking for perfectly predictable outcomes. If we are, we’re idiots. See Nassim Taleb’s books The Black Swan, and Fooled by Randomness as deep discussions of how we misapply the predictive value of the mathematical and scientific tools that we have.
But you asked me about this in your response, and I have to completely disagree with your assumption.
David B : It doesn’t mean that the simulations would accurately reflect specific events in reality.
Clayton : Then what the hell is the point? I have two programs on my computer that simulate the motions of the heavens (Stellarium and Celestia). I have no interest or use nor does anyone have a use for a program that simulates “something like” the motions of the heavens “but not actually” the motions of the heavens.
I believe you make the same mistakes scientists of other disciplines make, described by Hayek above. We can never get perfect prediction from our math or our theory. Even those programs you describe are incapable of accurately reflecting the state of the sky.. Oh wait, they can but only because the time scale on which change happens is so much longer than what you will perceive in your lifetime, that’s a problem of scale not of qualitative difference. Again, you assume somethings true that’s not, we don’t need perfect prediction to gain useful knowledge from simulations.
Weather simulations don’t produce the exact weather, they just give us approximations that statistically tend to converge on a certain set of conditions. They then use those convergences and convert them to verbal predictions of the weather conditions in the near future. That seems to have value to people, even though it’s not the equivalent of knowing where orion is in the sky at 1am tonight. (Summer I think it’s not visible at that time, but I’m not sure.)
Here are a couple of articles about how man uses simulations as part of a theoretical/simulation feedback loop that expands our knoweldge both about reality and about our models/theories.
http://www.lifeslittlemysteries.com/1993-string-theory-big-bang.html
These types of simulations don’t show our universe as it is, but they help us understand how the universe works, and how to model current conditions going forward. With these kinds of models (in any problem domain) we can understand the kind of effects specific changes might have, not with absolute certainty, but with more confidence, and less uncertainty. Anyway in which we reduce uncertainty would be a good thing. They look at the results generated by these simulations and what they get (when compared with reality) helps them to understand where they are right in their assumptions and data and where they are wrong. Not by comparing the exact conditions, but by comparing the structural differences.
I for one, would like to be able to show in a more real way, how certain types of policies must perform (not with absolute certainty about the specific timing of events) but by demonstrating that the future results are likely to converge in very specific ways: Housing Bubble, Debt bubble, Currency failure, etc.
You are going to argue that the information incompleteness, and the black box nature of valuing, are in fact so limiting that we can’t do this at all. I disagree, simply because the human mind somehow manages to construct successful action. I believe you’ve accepted a fallacy of the natural sciences, and that fallacy is eloquently described by Hayek above. Any ability to predict a micro results in experiemental conditions is misinterpreted as the ability in practice to predict with absolute certainty future states of reality. This cannot be done. The human brain is, however, an example of a computational device that is proficeint at solving problems with incomplete data, and we are experts in dealing with “black box” values of others.
Our brains intuitively handle complexity issues through pattern identification and recognition. We could be better, in our sciences and in our tools at assisting us in what we’re already good at. That’s my point, I’ll stick with it.
Above you said to me, “I think this is a naive view.”
Do you still think so?
But I think it captures some of my thinking, the question is how to self-regulate without crushing growth and innovation. My hope of course is to trend toward privately owned competing regulation of human behavior.