Betting on Global Warming

While I still have a possible bet in the pipeline, Brian Schmidt now has a real bet lined up for a total of \$6-\$9k. Personally I think that 0.15 oC/decade is more likely than not, but that less is a realistic possiblity given natural variablity. So while its good from an expected return view it could potentially be bad from a PR view. I wonder how important the latter is… probably not very, in 10 years time.

Brian has a spread of bets across 10, 15 and 20 years which is an interesting way of doing things. I’m still looking forward to making big money (a-la James :-) when a real market for these things developes and reselling good-odds bets for a sure fire profit becomes possible.

1. #1 copyworld
2007/04/25

I just came across your blog about Global Warming and wanted to drop you a note telling you how impressed I was with the information you have posted here. I also have a blog about Global Warming so I know I’m talking about when I say your site is top-notch!Global Warming Keep up the great work, you are providing a great resource on the Internet here!

2. #2 Joseph j7uy5
2007/04/25

The other question is, what will \$9,000 be worth in twenty years’ time?

3. #3 Brian Schmidt
2007/04/25

Legitimate point on the PR risk. Hopefully I’d only lose on the even-odds bet but win on the 2:1, so the PR effect would be limited. And I only lose on even-odds if the increase is less than 0.13C, as opposed to voiding, so the risk is lower.

4. #4 uBeR
2007/04/25

5. #5 Brian Schmidt
2007/04/25

Not inflation-adjusted. I’d guesstimate the 20 year bets will be worth half their present value, but YMMV.

6. #6 Benjamin Franz
2007/04/25

The first comment from “copyworld” is some kind of spam link.

7. #7 mugwump
2007/04/27

I too am looking forward to making big bucks out of the alarmists.

If +0.15 is 1:1 and +0.1 is 2:1, then assuming the second derivative of odds is reasonably small, a 0.0 rise in 10 years should be worth 8:1 and -0.05 16:1.

I’ll take the 0.05 fall in 10 years at 16:1. US\$10,000.

8. #8 Brian Schmidt
2007/04/30

Mugwump – you’re showing the danger of extrapolating. The 1:1 and 2:1 bets were rough guesses of what both my opponent and myself perceived as a relatively good bets – trying to draw mathematical equations from that extending to the outer edges of a graph line is ridiculous.

But, if you think the 16:1 fall is a good bet for your side, then by your logic (not mine) you should also think the 2:1 bet I negotiated is good. Contact me if you’re truly interested.

9. #9 mugwump
2007/04/30

Brian, I did say “assuming the second derivative of odds is reasonably small” (more accurate would have been “assuming the second derivative of log-odds is reasonably small”).

Your bet is interesting because it gives us two points on your log-odds curve, from which we can get a first-order estimate of your log-odds function, which is all I am doing (fitting a straight line to the log-odds).

Assuming your log-odds function is twice-differentiable (definitely not an unreasonable assumption), a straight-line fit is locally accurate. So the question comes down to what is “local”. Local here is measured in terms of the independent variable, temperature. I would have thought -0.05C/decade wasn’t so far from +0.15C/decade and +0.1C/decade as to make my estimate “ridiculous” (as you put it). Eg, if you do a Taylor-series expansion about +0.1C/decade the second-order term at -0.05C/decade is 0.0225 (= 0.15 squared). At 0.0C/decade it’s 0.01 (=0.1 squared). So the second derivative has to be pretty big to make the second-order terms significant.

I might take you up on your 2:1 bet after I model my own subjective probability function in more detail. I know enough already to definitely take you up on the implicit 16:1, 8:1 and perhaps even 4:1 bets.