I emailed Scott Armstrong my questions about his climate change bet, and he graciously responded:
The bet is about the size of the forecast errors, not about direction.
There are better ways of designing a competition. I had to go for something simple. But I hope that researchers will discuss and apply these better approaches. There is a whole chapter on evaluation in the Principles of Forecasting book.
There are substantial benefits in having Al Gore take the challenge. It would be a cooperative venture in the sense of encouraging better science. So I have not given up. Surely he should be able to find five minutes over the next half year or so. If that fails, we have a back-up plan.
In a follow-up, I pressed him: “Will you bet with anyone other than Gore if he doesn’t change his mind?” His response:
We do have someone in mind, but there is some flexibility.
Armstrong has gotten my hopes up that a serious climate change bet is in the cards. Fingers crossed.
READER COMMENTS
aaron
Jul 10 2007 at 2:38pm
Basically, I think all that is needed is an agreed upon measure that can be calculated each month and season. Betting on 5+ year moving average, using data that is smoothed before input isn’t very exciting.
Comments are closed.