Terms and Models

Everyone keeps models in their head. It’s the way we humans make sense of the world. For simplicity’s sake, lets model the model as a linear equation. The crudest models have one term: X caused Y, x = y. This is the stuff of the moonbat right and left. War = oil, abortion = murder. As you work your way along the bell curve towards the center of mass, the models get a little more complicated. People add weights to the terms: 0.9*abortion = murder (for those who don’t feel that rape victims need to keep the child). As you work your way further up the bell curve, more sophisticated thinkers begin to add terms: 0.9 early stage abortion + 1.0 partial birth abortion = murder*.

People can even hold different models for different situations. For example, in my personal universe, 0.8 abortion = murder. Rape victims and mothers of grossly deformed fetuses are exempt in my personal morality. However, I also have an equation for public morality borne of my libertarian beliefs, and if x does not equal y in your world, have at it. Don’t expect me to pay for it, but I’m not going to stand outside your clinic and call you a murderer, either. Human life is too complex to apply simple models of morality to other people, they are for you own guidance in making choices in this world.

As you get to the center of the human mental distribution, you find that most people have a mix of complex and simple models in their heads, but since most people can’t keep track of more than 3 things at once, most people have 3-term models as their most complicated case. More than one standard deviation past the mean, and you begin to get to the intellectuals, or partial-intellectuals such as me. Models begin to get really complex (My models are mostly partial differential equations, but then, I was trained to think that way.) Let’s stick with the linear equation analogy, although it gets a bit dicier as the models get more complicated. Now both variables and weights can interact. In the words of Murphy, commenting on the perversity of physical models: constants aren’t, variables won’t.

Lets take the disagreement that Lex and I have over the nature of Russian civilization. I do not think that we have fundamentally different terms in our model, it’s the weight we place on each term that differs. He places a lot of weight on the Byzantine term, I don’t. But we both recognize that the weight of the Byzantine term interacts with the weights of other historical events and trends that reinforce those Byzantine characteristics. Personally, I think that a civilizational influence has about 500 years to play out., and will die out unless reinforced by another historical influence. For example, the Yankee tendency to mind other people’s business outlined in “Albion’s Seed” would be of lesser importance in America today if we had not absorbed a lot of European socialists in the late 19th and early-to-mid 20th Century immigration waves. So, in my model, if A is the weight on the Yankee butinkski tendency, and B is the weight on Euroweenie socialism**, then A = A(t) + B(t+150)*0.3A(t0), where A(t) is a function of A that makes it decline over time, and A(t0) is the original impulse. Now we’ve introduced time dependence into the model (I told you my models were complex), and I’m not sure I’m going to make sense to anyone aside from the other voices in my head.

It is easy to talk to someone with a model that contains basically the same terms and weights as yours. Talking to someone with fewer terms than you usually means that you teach them, and talking to someone with more terms in his or her model means that you are the student. Assuming an ability to learn on the part of both parties (not always a good assumption, I know, believe me, I know, I was a TA), problems in communication usually arise from people fixing their weights in a radically different way. Lex and I have no problem getting at the root of the difference in our worldviews, and we have productive interactions talking about the origins of those differences. The difference in our weighting terms is small enough to have a conversation. When those weights diverge too much, however, communication becomes problematic. For example, when I talk about politics and economics, my fundamental assumption is that most people are shitheads. Let’s construct a simple model for my beliefs:

Economic and Political Disorder = 0.8 Shitheads – 0.6 Capitalist system of reality checks – 0.1 Government intervention to prevent tragedy of the commons

A good system can overcome some of the idiocy, and the market system, and the scientific method, are good ways to overcome the inherent capacity of humans to stuff their brains with manure. But the high Shithead weight and the low government weight are connected in my brain. Put someone in power and that person’s personal Shithead quotient goes up, so the government weight can never get very high in my model. Lets call the weights in those terms A. B. and C. In my model, C = 1-A-D, where D is a term to represent corruption. Socialists assume that only business people are shitheads, and that government solutions are superior to market solutions, so they think of a model such as this one:

0.8 Shitheads – 0.05 Capitalist system of reality checks – 0.9 Government intervention to prevent tragedy of the commons

Their C term is something like a normalized: 1/A-0.05D. I can’t talk to someone who assumes that human beings lose their tendency to become shitheads once those humans enter government service. Such a person needs a lot more life experience, and / or capacity to examine readily discernable evidence, before we can have a meaningful conversation.

All this is a very long-winded way of introducing the next post I want to do on Chinese regionalism. I was putting it off until I read this post over at the Coalition of the Swilling, and it reminded me that I’ve been meaning to review two books here, “The China Dream” by Joe Studwell, and “The Coming Collapse of China” by Gordon Chang. I personally believe that the latent regionalism in Chinese society will reassert itself during the coming financial crisis. I just do not know how much weight to put on the “regionalism” term. Just as experience, better education, and vastly improved communications devices have impacted the peaks and valleys in the stock market cycle, so too will those factors affect the inflection of Chinese regionalism in the wake of the retreat of central authority that will surely accompany the looming financial crisis.

* I’m normalizing all of these weights to make the maximum weight equal 1. Think of them as percentages.

** And just to piss Ralf off, I will point out that it was German Romantic philosophical muddlers who gave the world both Communism and National Socialism. When I think of a statist Euroweenie, the caricature in my head always has a German accent. Of course, when I think of an independent-minded Western European scientist, they, too have a German accent. ;-)

11 thoughts on “Terms and Models”

  1. I think the same as John Jay even in matters such as morality where solid numbers do not exist.

    I find it interesting how people educated in measurement driven fields like science, engineering and business think in numbers whereas people educated in the liberal arts do not. For those in the humanities, words seem vitally alive somehow. Words express concrete reality and form the basis of all thought. The humanist seem to believe that, like the sorcerers of old, they can control anything in the world if they can just speak the right words in the right order.

    For the measurement driven, words are merely labels on boxes containing the jumble of information that comprises different concepts. Change the contents of the box and the function of the word changes. Fill a new box and slap a label on it and, bingo, new word.

    I think this divide is what makes postmodernism so funny to those in the measurement driven world. For the postmodernist, the boxes and their labels define and control the world. They create reality. For the measurement driven, the boxes and their labels are just temporary and disposable conveniences. Freighting them with great import is like worshiping one’s sock drawer.

  2. When comparing models, I think you also have to look at the relative sensitivity of each model to errors in estimating its variables. For example, centrally planned systems tend to be very sensitive to the kinds of people who are running them. If the people are OK you might end up with something like a kibbutz. But if the people are shitheads you might end up with something like the USSR. There’s a lot of variability in that system’s output. OTOH, a relatively open society with a small government and a high degree of decentralization and political competition might have a narrow range of outcomes depending on whether the people in charge of the government are OK or shitheads. From both an engineering and a political standpoint I prefer the latter type of system.

  3. I should add that one of the big problems with the first type of system is that it’s difficult to know whether someone is OK or a shithead without actually giving him power and seeing how he handles it — by which time, if he is a shithead, it may be too late to do anything about it. So any system whose success depends on selecting the right people is a nonstarter as far as I’m concerned. And of course there are many voters who think shitheads are OK, so as a practical matter selecting OK people would always be difficult even if we knew how to evaluate people better than we currently do.

  4. Thank you John. You just formulate in a concise terms what I have in mind for a while about the problem of communication. Your post just opened some clogged mental ducts for me.

  5. Jonathan – that real-world systems are often extremely sensitive to initial inputs is why my mental models are PDEs rather than linear equations, but I thought that using a PDE model would lose everyone who has not studied differntial equations.

  6. Jonathon, I’m going to disagree with you in a limited and specific sense, not with your overall comment.

    One of the ways to judge a system is by its product. If you construct a manufacturing plant which is totally up to date in every respect, but the appliance that comes out the door is faulty, then there is a systemic disfunction which must be located and corrected.

    It is the same with a political system, or a cultural system. If the end product of leadership selection is a grotesque parody of a rational, compassionate, thoughtful human being, and the laws and rulings produced are sinister caricatures of morally sound legislation, then it is not the particular nature of the people involved which creates the problem, but the very essence of the system itself.

    If you elect the general run of politician in a republican form of limited and power balanced government such as ours, the limits of the structure prevent many of the excesses found in a system of arbitrary power and unlimited state authority.

    Stalin prospered in the system of proletarian dictatorship because that system’s pattern of rewards and punishments fit his personality and moral character.

    The reason that the system collapsed under Gorbachev was that his timid attempts at reform, as minimal as they were, were antithetical to the very nature of a totalitarian state. The internal resonances that resulted brought down the house as the support beams cracked and buckled.

    If Gorby could have brought himself to be more Stalin, as Brezhnev seemingly could during his reign, the SU might have waxed as it did in the 70’s instead of waning as it actually did.

    The byzantine complexity of any major cultural system molds and rewards some personalities, and breaks and rejects others. Ghandi admitted that the tactics he used against the British would be very dangerous if employed against Nazis. MLK would have been arrested and disappeared into the gulag instead of leading rallies of hundreds of thousands in the capitol if he had tried his civil rights campaign in the SU, or China, or Cuba, so much in the news today.

    Lord Acton was right, with a vengeance.

    Totalitarian systems select those who match their internal ethics, and produce the only product they possibly can produce.

    It’s not the luck of the draw. It’s a sure thing.

  7. I look forward eagerly to the China post. Have you seen Will the Boat Sink the Water? The Life of China’s Peasants, which was reviewed here?

    It seems relevant to the issue of serious stresses which China faces.

    As Jim Bennett put it, in a talk at the Hudson Institute:

    People are mesmerized by the growth curves in China right now. They see pictures of the big skyscrapers in Pudong and, you know, they’re extremely impressed by this, but they’re not looking the fact that China is on the wrong side of a huge transition problem.

    If you look at these transition problems in small countries like Taiwan and South Korea, which have very similar social structures, this was a big crisis. It was a huge crisis in Japan, which is not so similar, but had some similar issues, and it led to a, you know, major world war.

    China’s got big problems. I hope they work through them peacefully, without an enormous amount of disruptions; but, you know, I’m not going to lay odds on that it’s anything like 80 percent chance of success there. I think they’ve got a 50-50 chance of getting through their democratic transition without major problems and disruption, which are going to be I think the big international crisis of the 2030 or 2040.

    Or sooner.

  8. Shannon, you continue to cement your place in my list of my favorite writers. Great comment.

    John, great post.

    The specific “abortion=murder” examples you gave, I think, demonstrate a problem with a lot of people’s models: they don’t define their terms carefully enough. Abortion IS always the taking of an innocent human life (which we generally call “murder”), but abortion is sometimes the correct choice to make. Both of the models you gave have some validity — one uses the term “murder” correctly, while the other recognizes that abortion is not always the wrong choice to make (how often “not always” is determines the parameter value.)

    I model morality as a value-based (rather than harm-based) system, wherein we try to protect and enhance those things we value. “Innocent human life” is way up there on the list, so in most cases we try to protect it, but there are a few things we value more in some circumstances. When both a mother and the child in her womb are innocent, and by some fluke of biology only one will live, I value the mother more than the child so I see an abortion as tragic but necessary — it’s totally against one thing I value, but the only way I can preserve the other. When the lives of innocent children in Lebanon are weight against continued rocket attacks against Israel and the continued existance of Hezbollah, I view military strikes as tragic but necessary for the same reason. But if we had the technology to make more precise strikes, or to save both the mother and child, those should be done because they’d better preserve / protect the things I value. (This is a practical application of some of the concepts from the courses I’ve taken on optimizing multi-variable functions with time-dependent constraints. Yeah, I’m a math nerd too.)

    I think the biggest difference between most people in the West is that we organize our values in slightly different orders. Or, to use the terminology above, we’ve assigned different parameter values in the function we’re optimizing. I think if average people realized that, we’d have fewer accusations of “your side is teh EVIL” and more people willing to work with one another to accomplish things. But, a lot of people have crappy one-variable models with a binary parameter (“abortion=bad”, “war=bad”, “kittens=good”).

Comments are closed.