Thursday, January 18, 2007

Fun with numbers!

I just followed a link to this anti-global warming "argument" by some guy named James Lewis:

So in the best case, the smartest climatologist in the world will know 100 variables, each one to an accuracy of 99 percent. Want to know what the probability of our spiffiest math model would be, if that perfect world existed? Have you ever multiplied (99/100) by itself 100 times? According to the Google calculator, it equals a little more than 36.6 percent.

That sounds convincing...except, well, it's not:

Actually, James, I have multiplied (99/100) by itself 100 times and, oddly enough, it's 36.6% on any calculator, not just Google's. Well, OK, but what if the accuracy was known to, say, 99.3% instead of 99%? Then you get 49.5%. Bump that up to 99.7% and you get 74%! Wow! Numbers are like magic! See how easy it is to just plug in arbitrary numbers into an arbitrary equation and come out with an answer that satisfies your argument?

It's always interesting to see someone use a short column to throw around a bunch of numbers, usually in the form of "Multiply this times that, and then multiply that times this, and watch the numbers shrink! Ergo, [scientific theory here] is false!" You see this kind of thing a lot in Creationist/Intelligent Design literature.

So anyway, who is "James Lewis"? According to the blurb at the end of the article, he's an "academic scientist", and apparently "James Lewis" is a pseudonym. But there's a link to a personal blog of his, Dangerous Times, a sparsely-posted blog whose links all lead right back to American Thinker. What kind of "academic scientist" is "James Lewis", then? That's slightly important, since as PZ Myers pointed out the other day, the kind of scientist you are has some bearing on this stuff. You wouldn't go to a computer science Ph.D. for a precis of the most recent developments in particle physics, after all. This is why places like the Discovery Institute are always listing their rosters of "scientist" supporters but rarely actually indicating their areas of expertise.

And besides, I'm not terribly sure I'd overly rely on a guy willing to go on record as saying of George W. Bush:

By the measure of moral clarity and courage, George W. Bush is right up there with the best in American history.

Yup, there's a voice worth taking seriously on history and global warming.

1 comment:

Anonymous said...

More importantly, it demonstrates an abject lack of understanding the means by which uncertainties propagate in calculations. Generally speaking, the method for calculating this propagation involves summing over the product of each uncertainty with the deriivative of the calculated function wrt to variable in question.

The upshot of this is that when you have some calculated quantity that depends on 100 variables, you're taking 100 partial derivatives (one for each variable), and multiplying each one by it's uncertainty, and then adding them in quadrature (think pythagorean theorem). In the end, you find that your uncertainty is driven primarily by a few variables that have either large uncertainty, or large partial derivatives (meaning that the function depends very strongly on the variable in question)or large both.

So, the initial supposition, and the retort were equally silly, although it's obvious that the retort was meant to be silly...