Good Enough for Rock and Roll
Most of you know that I work for about half the year as a technical production editor at a Big International Scientific Institution. I edit peer-reviewed academic research papers on the subject of transportation. Dull you say? Heck of dull. There are interesting topics but they always seem to go to the other production editors. They get the camels and the women in Chad who have to get their goods to market, or the railroads. I've never had railroads. Or shipping. Or trucking.
I get a lot of pavement and cement. And the Holy Grail. The Holy Grail is modeling: creating a mathematical/computer model or simulation that will reliably predict, for predicting how the traffic that uses a grid of streets will increase over time or use the grid over the course of a day. You can see that governments would be interested in the Holy Grail. It would be a huge savings; it would be almost like striking oil: empirical work is expensive, time consuming, and troublesome (counting cars, or sending out surveys to thousands of households).
The math for these things can be daunting, and usually the desired outcome is a a comparison of the results of the model or simulation with previously acquired empirical data. These volumes are the biggest, i.e., have the largest number of papers, and it is all very meta. I can go through 20-30 of these and these papers and am almost stupefied with boredom. A lot of the aren't making a new model out of whole cloth but just testing out a tweak, or finding a new way to test some part or aspect of the problem: small increments of improvement if any at all, and usually, the conclusion is that further research is needed. In short, it takes a very long time to put a model on the road.
I had two thoughts about this: The first one was that there's something weirdly compelling about any representative object: I mean, as soon as a small child carries out the intention of drawing a face, no matter how crudely, we recognize the face. We can pick out lifelike features in the most abstract imagery, if we are given the least encouragement, the least suggestion. Mathematical models don't have faces, but they are lifelike in another way, which is that once you know how to read them, you can see that they are life-like, even if they don't actually "work", that is, do what they are supposed to do. They do manage to do something that literally looks like what they are supposed to do. In effect then they are another imitation of life. And of course computer models and simulations are enormously appealing. If they weren't, there would be no computer games.
Games don't really appeal to me. I like really simple ones like various forms of solitaire, but role playing games bore me. They are like the opposite of novels, to me. I have this evil suspicion that there are people who would prefer to live in a world with a finite number of possible events, with each contingency goverened by a fixed set of rules. Things supposedly get more lifelike as these events and elements get more complicated; worse, they are interested in life as a subject only in so far as it is represented in a gamelike way. And this is the complete opposite of the way I'm interested in life.
It's not that models don't work; I'm not being some sort of virtual Luddite here. It's that people are susceptible to a bias when it comes to judging whether a model works. There's "work" as in "the little cars go zipping around the fictional streets, stopping at stop lights etc." that is, it's a convincing image of something happening; and there's "works" as in "can be a basis for predicting the behavior of variables under the full assault of contingency, as in real life." The first is easy, the second is not. The bias is a certain difficulty in making this distinction. It's like thinking a Thomas Kinkade painting is "real."
I started thinking about this before the financial horror show began, because models are used in the securities business, to predict the price behavior of things like derivatives, i.e., to take the gambling out of gambling. The thing about gambling is that when the game is rigged it still has to look as if it's not rigged. So a model that looks like it works is as good as a model that actually works, if it will move product. And that itself is worth a gamble.
So it seems to me that investment companies, the creators and sellers of these model-based investment products, didn't need to find out whether they actually worked -- because at a certain fundamental level it didn't matter. At this point places like Lehman Bros. had become like that one old school chum of Bertie Wooster's who can handicap anything and opens a book on which vicar will preach the longest sermon, or on the Young Mothers' Egg and Spoon Race.
You could gamble with these fictions, and you could market them. It helps to actually have something of value to sell when you're marketing, but it's not essential. And then you market your own astuteness and technical wizardry, and you're off. So it seemed to me, anyway, and I did wonder if some of this mayhem was the result of the sort of business-world impatience with how things get done, with the labor of making, with the finding out of adverse facts. I mean, this sort of magical thinking about the source of value in an organization can turn up anywhere: I saw it in my last corporate job, I saw it when I was teaching at a university, and I saw it in the dot-com boom (it was the defining characteristic of the New Economy; people would tell me, speaking down to an idiot, how companies didn't need to make a profit selling products any more; in the New Economy everybody was going to get rich selling stocks back and forth and somehow the Newness of the New Economy was the way money would somehow mysteriously keep flowing up from somewhere -- venture capitalists' bottomless pockets maybe -- they were the real magicians) and keep everyone in the game for ever.) The same thing happened with the housing bubble, of course. There is a movement afoot to blame the recession on minorities and the poor (It is what people do all over the world, of course. When there is a crisis, go beat up on some person who doesn't have anything to do with it, but who can't fight back. Of course when Americans do it it is special and godly and nice unlike when other people do it.) but the catastrophe is the result of something that got much less press than housing prices until it began to come apart, which was the credit bubble. There have been deflating housing bubbles before, and while they were no fun for the participants, there was nothing on this scale. It's that the mortgages were turned loose in the credit markets and borrowed and bet and hedged against and leveraged and turned into fairy money, and, let's suppose for a minute that the that this bubble was growing roughly over the same time period as the housing bubble: notice that it's a kajillion times bigger (that is an obscure economic term I occasionally employ to impress people), and that gives me some idea of the scope and speed of the growth of the credit bubble. From that I infer prodigious activity, the creating, buying, selling and trading of these debt products at almost inconceivable speed and in volumes that it's hard to wrap my mind around.
Reflecting on that atmosphere it seems to me unlikely that the creators of these things would give them the kind of critical scrutiny that traffic engineers give to their models of freeway capacity.
My life coach says I need to get over my love of being right, but I think it may take me a while.