Bryan Caplan  

Bauman Climate Bet Bleg

PRINT
Reply to Cardinal Rodriguez Ma... About That High-Quality Insura...
The Economist confirms that global warming has paused over the last 15 years:
Between 1998 and 2013, the Earth's surface temperature rose at a rate of 0.04°C a decade, far slower than the 0.18°C increase in the 1990s. Meanwhile, emissions of carbon dioxide (which would be expected to push temperatures up) rose uninterruptedly. This pause in warming has raised doubts in the public mind about climate change.
The rest of the piece discusses explanations for the pause:
A convincing explanation of the pause therefore matters both to a proper understanding of the climate and to the credibility of climate science--and papers published over the past few weeks do their best to provide one. Indeed, they do almost too good a job. If all were correct, the pause would now be explained twice over.
I'm not qualified to evaluate any of this research.  As a matter of general epistemic policy, though, I put very little stock in after-the-fact explanations.  And that seems to be all the latest research is. 

Indeed, for a long time, most specialists just downplayed the facts:
As evidence piled up that temperatures were not rising much, some scientists dismissed it as a blip. The temperature, they pointed out, had fallen for much longer periods twice in the past century or so, in 1880-1910 and again in 1945-75 (see chart), even though the general trend was up. Variability is part of the climate system and a 15-year hiatus, they suggested, was not worth getting excited about.
Or disputed the facts:
An alternative way of looking at the pause's significance was to say that there had been a slowdown but not a big one. Most records, including one of the best known (kept by Britain's Meteorological Office), do not include measurements from the Arctic, which has been warming faster than anywhere else in the world. Using satellite data to fill in the missing Arctic numbers, Kevin Cowtan of the University of York, in Britain, and Robert Way of the University of Ottawa, in Canada, put the overall rate of global warming at 0.12°C a decade between 1998 and 2012--not far from the 1990s rate. A study by NASA puts the "Arctic effect" over the same period somewhat lower, at 0.07°C a decade, but that is still not negligible.
Or moved the goalposts:
It is also worth remembering that average warming is not the only measure of climate change. According to a study just published by Sonia Seneviratne of the Institute for Atmospheric and Climate Science, in Zurich, the number of hot days, the number of extremely hot days and the length of warm periods all increased during the pause (1998-2012).
To repeat, I'm not qualified to debate these experts.  But their reaction seems fishy to me, and I am more than qualified to bet against the consensus on the basis of that perceived fishiness.  Of course, since the experts are claiming knowledge, and I'm claiming ignorance, the odds should be in my favor.  Last week on Twitter, I publicly offered to bet at 2:1 odds that the global warming pause will continue for another 15 years. 

Yoram Bauman has nobly accepted my offer.  Indeed, he offers better terms than I requested: 3:1 odds, and I win if (according to according to ) the average global annual land + ocean temperature increase between 2014 and 2028 inclusive is less than or equal to +.05 C.

My main question: Do Bauman's terms raise any red flags?  In particular, does anyone have any problem with his suggested data source?

Otherwise, I propose only cosmetic changes:

1. Since we already have some data from 2014, the bet should run from 2015 to 2029.

2. Yoram proposed Wheat and Chessboard stakes, but I prefer to simply bet my nominal $333.33 against his nominal $1000.

3. If stops publishing data during this period, we call the bet off.  Alternately, Yoram can propose back-up data sources, and we only call the bet off if they all stop publishing data.

My main reservation is the past reluctance of global warming skeptics to bet.  But after a 15-year pause that was not widely predicted in advance (also known as "predicted"), I'm ready to take my chances.

P.S. My Bettor's Oath states: "When I lose a bet, I will admit defeat, pay promptly, and hold my tongue - never protesting that I was "really right."  If I have caveats or reservations, I will declare them when I make the bet - not after I lose it."  So let me state one caveat here and now: I expect to lose this bet.  I would not make it at even odds.  I am betting because I think climate experts know less than they think, not because I know future climate with any confidence.


Comments and Sharing





COMMENTS (23 to date)
Ben writes:

The intertubes being what they are, it seems quite likely that that particular URL will not stay the same for the next 15 years. So if you take the bet as presently defined, I don't expect either of you to win; I expect it to be called off. If you both really want to make this bet, then you ought to settle on some kind of backup plan for the data source.

I'm also not sure precisely what you mean by the "average global annual land + ocean temperature increase between 2015 and 2029 inclusive". Are you comparing the mean temperature in 2015 to the mean temperature in 2029, and ignoring the intervening years? If not (I hope not), then I suppose you are talking about fitting a regression line to the mean temperatures for all the years 2015 to 2029 inclusive, and comparing the fitted temperature for 2015 to the fitted temperature for 2029? So the mean temperature in 2029 could actually be lower than the mean temperature in 2015, and Bauman could still win – the fitted temps from the regression are all that matter, yes? That seems to me to be the sensible way to do it; otherwise the bet will mostly be determined by whether 2015 and 2029 in particular happen to be unusually hot or cold, which is not the point.

In all likelihood, if these issues are resolved by you and Bauman, I would be interested in taking the same bet against you. If you're interested in that, let me know. :->

Steve writes:

*here

Glen writes:

If it were me, I would offer Bauman better odds, but with a different dividing line between winning and losing. You're essentially betting that the rate of change over the next 14 years will be the same as, or less than, the rate of change over the last 14 years. But the rate of change over those years was *substantially* less than the climate models predicted. If the global temp had gone up twice as fast as it did (0.08 deg C per decade instead of 0.04 deg C), that would still have been less than half of the predicted rise (0.18 deg C). So... I would offer something like 2:1 odds with a break-point of 0.07 deg C per decade, or about 0.10 deg C over 14 years.

Joshua writes:
after-the-fact explanations

Does this not define the majority of economics research?

Mike H writes:

Right now my best guess is rising CO2 levels are balancing falling solar output. If the predictions that the sun will continue to wane until mid century prove true, I think Bryan has a chance. Still, a rise of only 0.05C per decade seems almost indistinguishable from noise, so Bryan deserves at least 3:1 odds because of the overwhelming confidence of the climate modelers.

I imagine this is why Bryan expects to lose. We all know CO2 puts upward pressure on global temperatures, but it will be interesting if he loses with a value just over 0.05C. I can't imagine the climate scientists will be happy with that, even though they should. They should want to be wrong after all.

J writes:

I echo Glen's comments, I think 0.05C is a little too low given the predicted 0.18C, I think a lot of scientists even claim the time frame from 1998-2029 should average 0.18 a decade increase. This would suggest a much higher increase over the next fifteen years.

So putting Brian in the skeptical of consensus camp any rise of 0.09C per decade or less over the next fifteen should put Brian as the clear winner for the rate of warming being overstated.

I think 0.05C is a little low as even a 1/3 as predicted would cause Brian to lose. It'll be interesting to see how many people are willing to take Brian up given his low estimate.

ThomasH writes:

It looks fair to me (but ocean temperatures to what depth?), but a better bet would be one that covered both temperature and ocean acidification as both are summary indicators of climate change.

Handle writes:

Don't rely on a terrestrial data set, and instead use satellite observations, which will be free of various issues with surface-station-derived data. Measurements from space eliminate all those issues.

Good data sets are available here at the University of Alabama at Huntsville site:

http://www.nsstc.uah.edu/data/msu/t2lt/

See Dr. Roy Spencer's latest chart of this data for the global lower atmosphere here (and which looks like a 0.1 degree/decade increase trend, so you are undershooting by 66%):

http://www.drroyspencer.com/wp-content/uploads/UAH_LT_1979_thru_April_2014_v5.png

For what it's worth, an increase of 0.05 degrees C by 2029 would be a massive victory for Lukewarmers. Remember, the average alarmist forecast is for 2 degrees C by 2100, which would require something faster than 0.2 degrees per decade on average from now on. 0.05 C in 15 years is 1/6th that rate.

Chris Wegener writes:

I certainly agree that ignoring the light coming at us in the tunnel is a
great plan.

Handle writes:

@Chris Wegener:

Um, XKCD says we'll get 3.5 degrees of warming in the next 86 years, or 0.04 degrees per year, which is four times faster than trend, which means to catch up, rates will have to be much faster than even that in the coming decades. That's a very extreme forecast. I'd be happy to bet you lots of money that we won't see warming that extreme within this century.

Josh Sacks writes:

1) Are you betting the world or US temp? Not clear from description. World favors you as there is less volatility.

2) For the purposes of measuring changes from the present, you almost certainly want to us a satellite-based dataset: http://en.wikipedia.org/wiki/UAH_satellite_temperature_dataset
Historical datasets make various compromises in order to maintain comparability with the past. Given that you don't care about the past, you should use the most accurate, auditable, and reproducible dataset available.

3) I agree with Glen wrt. temperature and odds. The heart of the informed skeptic position is not that CO2 doesn't matter, but that the amount of warming is modest ~1.5C over next century. That's 0.15C/decade.
Predictions of high GW cost are based on 3-5C temperature changes, so GW activists should be willing to be on 0.3C/decade at reasonable odds.

John Csekitz writes:

While I may be wrong I think this a better bet as our skepticism is more of the impending disasters and doom used to sell the increased burden.

“Nature shares with Science and PNAS (Proceedings of the National Academy of Sciences) the distinction of being tied as the world's three most prestigious scientific journals, and an article is not published in these journals unless it has undergone extremely rigorous scientific peer-revue; so, climate-change deniers will have no professional credibility in attacking this study, as the Koch brothers and their friends can reasonably be expected to do, since they profit so much from what causes global warming - the burning of carbon-based fuels.
According to this study, the tropics, which are the near-equatorial region of this planet that's almost 100% impoverished, and that has thus contributed virtually nothing to global warming, will begin the period of permanent catastrophe starting in approximately 2020; but the (cooler) moderate-latitude countries, such as in North America and Europe, will begin this catastrophic period in or around 2047.”

And this “This past week, at the annual meeting of the American Association for the Advancement of Science (AAAS), experts warned that, "In 2020, the UN has projected that we will have 50 million environmental refugees," the AFP reports.”

bill writes:

The bet should be against the hottest year to date, not against 2015. 1998 is a good example why. Pretend you were making this same bet in 1997 and then look at the records from 1997 to 2013 to see why. I think (writing from memory) that of the 10 hottest years on record, 9 of them are in the 21st century. The only one from the 20th century is 1998. So the trend since 1998 looks flat because 1998 was an extreme spike. Check the trend to date from 1997 or 1999. Or just run a 5, 7 or 10 year moving average. If you look at those three moving averages, you may wish you didn't make that bet. Though for $333, who cares? And if you get to use 2015, you might get lucky and get another exceptionally hot spike like 1998.

blink writes:

Bryan, it is admirable that you and Yoram have agreed in principle to a bet. As you say, betting on the long-shot, you requested odds and expect to lose. As the terms of the bet are coarse, it raises the question: What do you actually believe?

Perhaps after ironing out the details, you (and Yoram, too) would answer: At what cutoff would you accept even odds? It would be interesting to know whether your beliefs differ on this point as well.

Jameson writes:

The fundamental issue here, on which I'm a bit disturbed by Caplan's attitude, is on whether it's "fishy" to be wrong (or have incomplete information) about short run outcomes in climate science. There is a huge difference between completely understanding climate dynamics and knowing that, for instance, CO2 emissions are hugely responsible for major changes. Expecting a model that predicts decade by decade is probably pure fantasy. How accurate do you think you could be predicting the next recession?

That said, Caplan did ask for odds, so I guess that makes sense.

Just one more thing, perhaps even more disturbing: 0.04 degrees per year is not a "pause." If temperatures continue increasing like that for a whole century, that is pretty huge.

Josiah writes:

I wish I had seen this before you finalized the bet, as I would've warned you that you were making a mistake (not that you would've listened, necessarily).

Fortunately $333.33 isn't a lot of money, so you'll be fine.

Michael Stack writes:

@BryanCaplan,

WHen you say that you 'expect to lose', are you saying that in the multiverse, you lose most of the time, but still win at least 33% of the time? Or are you saying that you expect to lose more than 66% of the time? If it is the latter, why wouldn't you have asked for the odds you expect?

Michael Stack writes:

@Jameson,

The increase of 0.04% C was per decade, not per year.

Josiah writes:

The increase of 0.04% C was per decade, not per year.

It's neither. The bet is whether average temperature during 2015-2029 is more than 0.05C warmer than average temperature during 2000-2014.

bastiches writes:

[Comment removed pending confirmation of email address. Email the webmaster@econlib.org to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Eliezer Yudkowsky writes:

I expect you to lose this one. It feels a lot like I thought the circumstances were just before I bet against the Higgs boson (90% of the possible range had been eliminated, and they were trying to predict a new physical field based on complicated arguments). Which I lost, because whaddaya know, the Higgs was there after all.

What odds would you set on a 0.20C rise this decade?

aaron writes:

Brian, if I were you, I'd stick to the satellite data. The GHE is strongest at the surface, but the ocean has a high specific heat. The atmosphere temps are more sensitive to greenhouse radiation and much of the additional heat at the ocean surface is evaporated back into the atmosphere and released high in the troposhere when the vapor condenses.

If cycliality is factor in warming, I'd expect the rate to be nearly half of the 70s to 2000 warming. There likely to be nearly 10 more years of haitus and resumption of the 1.2C/century trems. I'd only expect about a 1/3 of the 70ss-2000 warming trend for your time-frame.

aaron writes:

Oops. I thought you were taking the otherside, the satellite metric would favor Yoram on the technical point (atmosphere will warm more than surface w/ ocean surface).

The HADCru, GISS, and NOAA data do have analytical methodology that bias the trend up though, so satellite still might be preferred. Also, it is less likely to be affected by changes in methodoloy.

Nice to see you're taking the skeptic side.

I'd take the bet if I had the patience, however, I'd go for a higher temp or bigger payout. Probably .1C would bring you close to 50/50.

Comments for this entry have been closed
Return to top