Sunday, September 21, 2003

The Causality Fallacy

We all love explanations. Why did she do that? How does that work? The trouble is, many systems don't lend themselves to simple explations. Our obsession with cause and effect leads to many dangerous behaviors, from seeking scapegoats, to over-simplifying complex problems that lack a single cause.

The NPR show On the Media got me thinking about this. Bob Garfield did a great piece on "single factor analysis" on September 12th. At the close of every trading day, reporters not only say, "The Dow closed down 50 points," they also give the reason: a Commerce Department report, poor earnings by a bellwether stock, profit taking. This is rubbish, of course: there is no single reason why three billion share trades on average moved one way or another. They do it, and we demand it, because There Has To Be A Reason.

One can multiply examples indefinitely: A post mortem has to show a cause of death. A crime has to have a limited number of perpetrators. A war was won or lost because of the actions of one man. Jack left Jill because she was unfaithful. The space shuttle Columbia crashed because of NASA culture.

This is not just because we live in a sound bite culture where everything has to be explained in two sentences. Stories since time immemorial have boiled reality down into a few characters and some pivotal events. This is because our learning brains have evolved to extract a few key factors from the environment, and associate them with events. This is evidently a supremely adaptive behavior, at least for the environment we faced in the course of evolution -- we're still around, aren't we?

However, Homo sapiens can now influence and build very complex systems that do not have simple explanations. Take chaotic systems; their exquisite dependence on input values means that one simply cannot explain why one outcome was reached rather than another. The global warming polemic is a perfect case where human nature forces the debate into a supposedly decidable fight over whether (a) global warming exists, and (b) whether industrial carbon dioxide is exacerbating it. The climate is a chaotic system; even the largest computers can't (and won't ever, it is said) predict the weather more than three days in advance.

If we are to think about the world in a way that mimics reality, then, we have to give up the habit of seeking cause and effect. Let's just ignore the argument that this means that science is irrelevant; it doesn't - cf. quantum mechanics and non-linear systems theory, to pick just two examples. What's really at stake is just as easy as obeying the command, "Add two plus two, but don't think about pink elephants."

Since it's in our nature to seek simple causes, we can't avoid seeking simple causes. We can, though, be aware that we have this weakness, and try to accomodate it. Tren Griffin taught me about psychological basis for investor misjudgement, as expounded by Charlie Munger in his on speech on "24 Standard Causes of Human Misjudgment". Munger calls it "patterened irrationality", and he gives examples of how one can gain wisdom by applying knowledge of our frailty. For example, he cites "the Warren Buffett rule for open-outcry auctions: don't go." If you know that a market has been set up to make you act irrationally, don't use it.

I'm not denying causality here. I'm claiming that the unaided human mind (unaided by cultural props like science and wisdom, that is) can only cope with a very simple form of causality: an effect has a single, simple, proximal cause. That's evidently false, and what one might call the Causality Fallacy.

So what? What should one do about the causality fallacy? Well, I don't know Warren Buffett, but even I know I'm no Warren Buffett. Still, here are some first thoughts:

  • Accept that many things simply cannot be explained in a satisfactory way. God moves in mysterious ways, it is said. They may be mysterious simply because our brains have evolved to see only simple motives, not ones that have the complexity of reality.

  • If you think something has a simple explanation, try to find another one.

  • If the answer is Either/Or, look for a third possibility.

No comments: