21 December 2007

When asteroids attack

Asteroid May Hit Mars in Next Month (AP).

Scientists tracking the asteroid, currently halfway between Earth and Mars, initially put the odds of impact at 1 in 350 but increased the chances this week. Scientists expect the odds to diminish again early next month after getting new observations of the asteroid's orbit, Chesley said.

If they expect the odds to diminish, why haven't they set them lower in the first place? I think this may be another issue of mean-median confusion -- say, there's a 1 in 3 chance that the odds will go up to 1 in 25 next month and a 2 in 3 chance it'll go away completely. But the statement seems kind of silly. Apparently efficient market theory doesn't apply to asteroids.


Anonymous said...

Efficient estimates would occur if it were a betting market. If the scientists had real money on it, they wouldn't wait until next month to get accurate. :-)

Aaron said...

Maybe it's because I'm a physics student, but this makes perfect sense to me! Let r be the asteroid's perigee distance—the closest it will get to the center of the Earth. Suppose that the asteroid will hit the Earth if r < 6400. Suppose also that our current measurements suggest that r = 7000 km, with a standard deviation of 100 km. It's easy to see, in this model, why knowing r more precisely would be expected to decrease the estimated probability of the asteroid hitting the Earth.

Anonymous said...

I suspect it's a case of the "It looks like it's heading that way because of X, but we can't be sure inless we have more data"