The Shotgun Blog
Wednesday, June 13, 2007
As a bonus feature to my article, The gospel of green, in the June 4 WS, here is an excerpt from the interview I did with David Orrell, a computational scientist and author of Apollo's Arrow: The Science of Prediction and the Future of Everything. I tracked him down to London, England, where he now works. I started our conversation by introducing the topic, green religion, and then had asked whether scientific predictions are being taken on faith these days;
DAVID ORRELL: That aspect of science as a whole has definitely taken on some of the characteristics of a religion. The whole business of prediction has typically been the preserve of religion since the Delphic Oracle and before. It’s typically been the established churches that have been in charge of telling us where we’re from and what’s going to happen in the future. Science has taken that role over from religion to explain both our origins and what the future holds.
The problem is that it has gotten to the point where people really want to predict the future and scientists want to be able to supply these predictions, so it gotten to the point where scientists don’t question their predictions enough. Scientists are famous for their skepticism, but when it comes to predictions their skepticism seems to be put on hold.
Religions have good aspects and bad aspects, so it’s not totally disparaging to say that science plays some roles of a religion. There are community aspects, and the whole caring about the meaning of life, etc. and all that’s great. I guess the problem can come when you have a priestly class of people who are interested in protecting their own version of the world. That’s sort of the sense I got a little bit of with some of these scientific models of the atmosphere. For example, the IPCC report that came out at the beginning of the year with the latest climate change predictions. If you actually listened to what the prediction was, it was basically saying that if carbon dioxide doubles from pre-industrial levels, the warming should be between 2 and 4.5 degrees and definitely not less than 1.5. So that’s a big spread. But on top of that, with other uncertainties, like exactly how much carbon dioxide is going to get released the prediction becomes something like 1.2 to 7 degrees. So it was incredibly vague. Most the media didn’t pick up on that. They all said that this is evidence, complete proof, the world is warming and then they spun off all these more detailed predictions which scientist had kind of thrown out. It was weird because there weren’t very many people who were saying, hang on a sec, this prediction is so vague it’s not really even useful. It doesn’t really tell us anything in detail. It’s a strange thing. It’s sort of like if you mention the Oracle in ancient Greece giving some really mysterious utterance and everyone goes, ‘Oh they’ve predicted the fall of the king!’
It’s one of those things where the prediction is kind of vague, but very authoritative at the same time. The weird thing is though that that prediction actually underestimates the error in the model. For example, the statement that warming could be between 1.5 and 4.5 degrees, if you actually look at the computer models… they did some experiments a couple of years ago at Oxford University where they randomly varied several of the variables that control the way clouds are modeled, and they ended up getting a much larger spread than that. And some models showed no cooling and other ones shot off to above 10 degrees. My background is more in engineering, and engineer modelers tend to have a built in conservative approach, mainly because they have something personal on the line. You design something and it fails, doesn’t work, then you are responsible for it. But in predictions of the future, no one really has anything personal at stake. There’s a big market for predictions, but it’s not something we go back and check very often. It’s the same kind of thing in economics as well. Again, there’s a huge market for predictions. There are loads of people who claim they can predict what’s going to happen on the markets, but if you go back and look at their track record, more often than not they’re no better than random.
In the priestly class I am thinking of people who want to protect positions of power and that would include funding. The whole climate change community is dependent in a way on … if they said our models don’t work and we should just pack up on go home, then that would be it, that wouldn’t work, right? I get the feeling they’ve lost some of the skepticism that scientists normally have, that they have sacrificed it more for job security. I don’t think scientists are going to like to hear that.
I support the environmental movement. But I think it has some contradictions. One of them is that a lot of environmentalists support this Gaia theory idea, that earth could be viewed as a living entity. So I thought it was a contradiction that on the one hand they say that the earth is this incredibly complex thing, similar in complexity to a living being, but on the other hand saying we can predict it and model it and say exactly what is going to happen to it in 10 or 20 years time. Those two things don’t really gel. One of the properties of living things is that they are extremely hard to predict. And I think there is a lot of truth in that idea, that the earth is kind of a complex, self-regulating system. If you imagine climate change as putting an extra layer of insulation on your house, then it is kind of predictable. But if you think of climate change as putting a blanket on a person or an animal who is cold and shivering, then you can imagine a computer model might not be the best thing for that because it depends on the response of the system. I kind of feel the environmentalists are throwing their lot in too strongly with the mechanistic, scientific models. This really doesn’t convince skeptics because skeptics can easily point to flaws and not be convinced.
TrackBack URL for this entry:
Listed below are links to weblogs that reference Archery:
I'd like to repeat that personal responsibility thing.
Last time I worked with a structural engineering model (aka finite element analysis) it came with a note that "computer models are not a substitute for experimentation". That is, "If you use our model to design something and it fails, it's your problem, not ours, buddy."
Remember, these are structural analysis models: vastly less complex than a climate model and with a much better track record. And yet even these guys disclaim responsibility for the results.
When liability is involved, people get a whole lot more humble about their ability to predict.
Posted by: pete e | 2007-06-14 2:23:58 AM
The comments to this entry are closed.