Fudge Factors

Some people are so interesting that just seeing their name is enough to make me go investigate. Freeman Dyson is one of those names for me — he’s a famous physicist who hasn’t limited his interest or thinking to his own field. When I think of Freeman Dyson, I think of a fine mind, well stuffed with knowledge, romping about the open fields of questions about our universe. Reading his thinking is an adventure.

So yesterday, when I noticed that TCS Daily had published an interview with him, I had to go read it. It’s a very unfocused interview, ranging all over the place. In the middle of it, Dyson said this:

Concerning the climate models, I know enough of the details to be sure that they are unreliable. They are full of fudge factors that are fitted to the existing climate, so the models more or less agree with the observed data. But there is no reason to believe that the same fudge factors would give the right behavior in a world with different chemistry, for example in a world with increased CO2 in the atmosphere.

As someone who has created computer models in the past, hearing that they are full of fudge factors is pretty much the same thing as saying they are full of another soft brown substance, one that doesn’t smell so good. What Dyson is describing is a process that every modeler is all too familiar with. It goes like this: first, you study the problem you’re trying to model until you believe you understand it.

Just as an example, suppose you’re trying to model what happens when you throw a ball. Your model tries to predict where the ball will hit the ground. So you study the physics of this motion, and build a model that takes into account the speed and direction of the throw, gravity, the air resistance, the wind, etc. Then you test the model — you put some values in for speed and direction of the throw, the wind speed, and so on. Your model calculates where the ball will land. Then you actually throw the ball to see where it actually lands.

And oops! The ball doesn’t land exactly where your model said it would. Instead, let’s say it lands 2 feet short, and 6 inches to the left of where your model predicted. What do you do now? There are two basic ways to “fix” your model. One way is to say to yourself “Ok, there must be something I don’t understand…", and go back to study the problem some more until you do understand why your model didn’t match reality. This sort of reaction brings deeper understanding of many kinds of physical processes, and models used in this way have contributed much to human understanding of many things. The other way to fix your model is to add “fudge factors”. In our example above, perhaps you multiply the distance predicted by your model by 0.976 (fudge factor one) and the leftward motion by 0.994 (fudge factor two). Those numbers aren’t based on any understanding of why the unfudged model is different than reality; they’re just the factors needed to line the unfudged model up with reality.

The problem with fudge factors is that they mask what’s really going on. To continue our example with the ball, if a scientist investigated the mismatch further, he might discover that the spin on the ball was also contributing to its trajectory — and when he corrects for the spin (by building it into his model), the model can predict the landing point to within an inch. Now this model can be used to confidently predict the landing point of the ball. The model with the fudge factors, though, will continue to perform poorly — because the spin on the ball can be different on every throw.

So Dyson has discovered that the climate models used to predict global warming are full of fudge factors, and he’s as suspicious of them as I am. Knowing what I do about Freeman Dyson, I put a high value on his judgment in such matters. Not only is he possessed of a fine, well-educated mind — he is also possessed of independence, both of his intrinsic spirit and of any connections to the global warming “movement” amongst climatologists. His observations about the fudge factors can be made without any particular expertise on climatology; all one needs to know is the math and the elements of modeling. Both things are elemental tools for any physicist, so Dyson is well-equipped there.

My skepticism on the global warming hysteria just got turned up another big notch…