Monday, September 6, 2010

Another month, another humilation for "scientific forecaster" Scott Armstrong


The original site Armstrong was using to track his bogus "bet," HubDub, has now vanished along with Armstrong's integrity and self-respect. The creators, who evidently have recognized in themselves a certain talent for make-believe, are now running a sports fantasy site. Fortunately Intrade is on the case; above you seeing Armstrong's chances of "winning" his own, self-designed, "Global Warming Challenge." It's now trading at 3.5%. No word from Armstrong, and still not updating the "bet." And while we're on the subject . . .

August UAH: 0.511C

. . . he's lost another month. That makes 7 out of 7 this year, meaning he loses 2010 as a whole (again, using his own nonsensical method of counting "winners" month by month of a ten-year prediction, and counting the winner of the most months as the winner of the year.) He's lost ten of the last eleven months. Here's a shocker -- he hasn't updated the status of the bet since March:

But here at the Idiot Tracker, we're happy to supply the assist to the eponymous Dr. Armstrong:

For April:

Armstrong: 0.263C
Gore (fictional): 0.348C
Actual (UAH): 0.50C

"Winner": GORE

For May:

Armstrong: 0.263C
Gore (fictional): 0.351C
Actual (UAH): 0.54C

"Winner": GORE

For June:

Armstrong: 0.263C
Gore (fictional): 0.354C
Actual (UAH): 0.44C

"Winner": GORE

For July:

Armstrong: 0.263C
Gore (fictional): 0.357C
Actual (UAH): 0.489C

"Winner": GORE

For August:

Armstrong: 0.263C
Gore (fictional): 0.36C
Actual (UAH): 0.511C

"Winner": GORE

You could set up this bet in a more honest, evenhanded way which might be semi-valid in terms of shedding light on what is happening with our climate. Needless to say, that's not the process Armstrong came up with, which makes it all the more amusing that he has managed to deal himself a losing hand from a stacked deck. The trouble with improving the process is that it make victory for Gore more or less assured. But let's look at what such a bet might be like.

For one thing, Armstrong has arbitrarily chosen the coolest of the four widely used global temperature data sets, Roy Spenser's UAH data:

NOAA NCDC: +0.163 C/decade

NASA GISS: +0.166 C/decade

UEA CRU: +0.158 C/decade

RSS LT: +0.164 C/decade

UAH LT: +0.140 C/decade

But even more basic, the notion that month-to-month temperature anomalies determine whether climate change is happening or not is silly. Obviously, if climate change is occurring, it's not going to change the fact that La Ninas lower temperatures, that solar activity goes up and down, or other random factors push the temperatures up or down. The sensible thing to do, and what scientists in fact do, is using moving averages. In other words, the last point in their graph is somewhere in the past, and they determine the average temperature using past and future readings. This smooths out the short-term fluctuations -- if each point on your graph represents a eleven-year average, chances are in that eleven years there have been a couple of La Ninas, and also a couple of El Ninos balancing them out. One really hot or really cold month no longer jerks the line up or down dramatically, which translates to the statistical equivalent of refuting global warming by pointing to snow on the ground (but what do you expect, really, when a guy with a degree in marketing proclaims himself a "scientific forecaster"?)

I mentioned eleven years; the sunspot cycle is eleven years*, so this is a commonly used period for averaging in climate studies. Why didn't Armstrong use an eleven-year moving average? Take a look at what such an average shows:

The herky-jerky of the annual anomalies is gone. In its place we see a smooth upward trend of rising temperature, with occasional brief pauses or accelerations. By the simple expedient of taking a moving average over time, the short-term fluctuations vanish, and the upward trend alone remains.

You can see how pointless a ten-year bet using this data would be. Temperatures have been going up steadily for 35 years; they aren't going to suddenly stop. It's hotter than it was ten years ago; ten years ago it was hotter than twenty years ago; twenty years ago it was hotter than ten years before that. Which is why, despite all the vocal deniers in the world, the ones proclaiming global warming is an obvious scam with no empirical evidence, mere "lying with graphs" as one denier put it, no one seems to be incline to put their money against a 0.15C rise in annual temperatures -- even at 30:1 odds.

1 comment: