The New Math - Global Warming Edition

Anybody surprised that global warming nutters have fudged statistics when it comes to climate change?

No?

Well, check this out anyway:

The United Nations has just released, with much hoopla, a summary of its report on the current state of global-warming science. This is the fourth in a series of reports that the U.N. Intergovernmental Panel on Climate Change (IPCC) has issued every five years since 1996. From 1996 through today, these reports have asserted a steadily increasing level of certainty that human activities cause global warming. The current summary indicates that the IPCC is “90% confident” that we have caused global warming. The summary further implies that if we double the concentration of carbon dioxide (CO2) in the atmosphere, the IPCC is 90 percent onfident that we will cause further warming of 3° C /- 1.5° C.

But what do these statements of confidence really mean? They are not derived mathematically from the type of normal probability distributions that are used when, for example, determining the margin of error in a political poll (say, /- 5%). IPCC estimates of “confidence” are really what we would mean by this word in everyday conversation—a subjective statement of opinion. This is a very big deal, since bounding the uncertainty in climate predictions is central to deciding what, if anything, we should do about them.

When trying to predict something like mean global temperature (a meaningless number if ever one existed) based on factors which drive it, it is important to understand how much noise is in the model, particularly when one cannot feasibly measure the entire population (say, taking temperature readings across the entire planet, or spanning epochs).

I usually explain this with a light dimmer switch analogy.

Think of a situation where we want to predict how illuminated a room will be at a given moment. The position of the dimmer switch would be one factor, how open the blinds are another, the amount of sunlight hitting the windows outside a third. Other factors bear in mind, such as whether the room is lined with reflective mirrors or painted flat black, how many light fixtures, how luminous the bulbs are, etc. It really is a fairly complex problem with plenty of variables interacting simultaneously.

The only way to figure it out is to test it in various configurations----switch up all the way, high intensity bulbs, mirrored walls, blinds wide open, full sun exposure, etc.

We then statistically analyze our data to see which factors best indicate how illuminated the room is (guess we'd need a light meter or something to measure that too).

Some of these factors will prove to be significant, but only at a given level of confidence. Because of all the interactions and natural variation (what if dust drifts by the window just as we measure? Or a cloud?) add noise to our model, we cannot be 100% certain our result applies for a given variable at any given moment in time. We therefore apply a range called a confidence interval (or band) to the results. This shows up most often in the margin of error in opinion polls, but it applies widely. At a given level of confidence, the true value may be anywhere along this range. If at 95% confidence we say the room is illuminated to the tune of 27 candlepower +/- 3 candlepower (or insert your favorite unit here), we're saying the REAL value might be 24, might be 30, might be anywhere in between. And oh, by the way, there's also a 5% chance it'll be some number less than 24 or greater than 30. Anything's possible, right?

So given all this wonderful statistical stuff, how do climatologists crank out wonderfully precise estimates "to 90% confidence" of temperatures 100 years from now?

They make them up.

If a climatologist tells you it will rain next Tuesday, you can leave your umbrella home because his computer models can tell no such thing. Your Aunt Hilda's probably better predicting it based on her rheumatism, which at least has some basis in reality.

Think I'm exaggerating?

How much money do you think you'd make if you could predict with 90% confidence where a hurricane would land and how strong it would be?

How about if you could predict the low temperature next spring for citrus growers in Florida or California?

Or if you could simply tell us with 90% confidence whether it will be hot and dry next summer in the plains states or cool and wet?

All of these things are simpler problems than global warming, and yet I don't see anybody cleaning up in insurance, agriculture, or the futures markets as a result.

Hell, if we could predict such things with 60% confidence we'd beat Vegas odds every time.

So what gives?

Two things:

1. You can't predict the future based on the past. Things change. If you were hanging around Earth 66 million years ago, you'd swear dinosaurs would rule forever. Things change.

2. You can't pluck one variable out of a multivariate model and extrapolate the result with any reasonable degree of confidence. It is not as simple to say that man-made carbon emissions go up, global temperatures go up, no matter what the people cashing grant checks will tell you while their grad student stats whiz assistants smash their slide rules in frustration.

Don't believe the hype, people.