Climate bets: Evans vs Schmidt

Dramatis personae

Coolist: David Evans, mathematician and engineer. 

His wikipedia page is here, and there's a fairly lengthy page at desmogblog. The strapline of one of his websites demonstrates his concerted Coolism:
Natural influences mostly explain recent global warming. The 2020s will be cooler than the 1980s. 

Warmist: Brian Schmidt, environmental lawyer. 

He used to blog here, but nowadays mainly blogs at Rabbett Run

The bet

This one, established in 2007, is a bit more complex than most as it has three time periods and two odds for each period. As Schmidt explained at the time:

We have three bet periods -10, 15, and 20 years - and two bets for each period - an even-odds bet and a 2:1 bet in David's favor. The even-odds bet centers around a temperature increase rate of 0.15C/decade with a 0.02 void margin on either side (bet voids if temps increase between .13 and .17C/decade). The 2:1 bet centers on 0.1C/decade with a .01 void margin. Even-odds bets are for $1,000 each, and the 2:1 bets increase over time, with me betting $1,000, $2,000 and $3,000, and David betting half that. My exposure is $9,000; his is $6,000.

That's the basic bet. We're using five-year averaged Nasa GISS data.

So...the first bet period (correct me if I'm wrong) is for the average of 2015-2019 compared with the average of 2005-09. The link in the original post to the NASA dataset is broken, but my guess is that they're using the Land-Ocean Temperature Index (LOTI), which is one of the main indices for global average surface temperature. The 2005-09 LOTI average temperature anomaly was +0.622 degC (compared to the 1951-80 mean). This means that for the evens bet:

- the central betline for 2015-19 is +0.772, i.e. 0.622 plus 0.15
- if 2015-19 averages less than +0.752, David Evans wins
- if 2015-19 averages more than +0.792, Brian Schmidt wins
- anything inbetween +0.752 to +0.792 is in the void zone, i.e. no winner 

How's it going so far?
Here are the NASA GISS LOTI values for the completed years of the first bet period:
2015: +0.872
2016: +0.997
2017: +0.903

This gives an average-to-date of +0.924. 2018 is almost complete and is currently at +0.82 (Jan-Oct average). If it stays at that level, the 2015-2018 running average-to-date will be +0.898.

In other words: Brian Schmidt is winning this at a canter. For him to not win the evens bet, 2019 LOTI would have to be below +0.370, and it hasn't been that low since 1996. It's not going to happen; there have been occasional drops of around 0.2 deg C from one year to the next in the past, but not half a degree.  

Here's a chart summarising the bet. 

David Evans is probably expecting to win parts 2 & 3, given his view that the 2020s will be cooler than the 1980s. That would be pretty amazing though; in NASA GISS LOTI the average temperature anomaly for 1980-89 was +0.25 degC. The graph above would have to have its Y-axis extended downwards a few notches...

Anyway, it would be great if either Brian Schmidt or David Evans were willing to offer their comments here. I'll get in touch to see if either of them are up for it.

An additional graph to go with my comment below:


  1. >"Brian Schmidt is winning this at a canter."
    Yes, but perhaps worth mentioning that super El Nino of 2015-2016 helped quite a bit.

    >"It's not going to happen"
    True and perhaps it is such a slam dunk it doesn't need to be mentioned that we look like heading into another weak El Nino making such a drop even less likely. (Also that Nov Dec usually have higher anomalies than year to date.)

  2. David Evans was good enough to send me his thoughts on the bet (thank you David). For some reason it didn't save or register when he tried to use the comment facility here on the blog, so he emailed me instead. Here is his text, verbatim:

    I made the bet in 2007 ( because the theory about carbon dioxide had no empirical evidence to back it, only a sensitivity calculation and computer models based on theory. The data from the upper troposphere contradicts the theory. So presumably something else was causing the warming, and sooner or later it would relent. Ok, I'll bet on that.

    I didn't discover the relationship between solar output and earthly temperatures (in the frequency domain) until 2013. This relationship, which led to the notch-delay solar theory (, is that the Earth's surface temperature follows the Sun's total output but with a delay of one sunspot cycle. The Sun's output turned down significantly from 2004, and the sunspot cycle following that was about 13 years. So the surface temperature might be expected to turn down on average from about 2017. This would mean we placed the bet too early, but we didn't know that in 2007.

    The fluctuations due to the large La Nina in 2008 and the massive El Nino in 2016 have given Brian a leg up, which is unfortunate because El Ninos and La Ninas seem to have little to do with the underlying trend we are betting on. At this point in late 2018, with a year left to run, it is almost certain that Brian will win the first pair of bets, due to be settled in 2020.

    Taking a wider perspective to reduce the effect of temporary fluctuations, and using the more credible satellite temperature record (covers the whole planet, not dependent on thermometers in artificially warming locations, and past readings are not adjusted), the trend for 1979 - 2018 is about 0.13 °C per decade.

    I now hope to have a shot at winning the 2025 bet, and winning the 2030 bets.

    1. In addition to Hautbois' points before, I'll add a few more points rebutting what David Evans said above. This won't be an exhaustive list, since David made a large number of false claims:

      The UAH satellite-based analysis does adjust past data, and has a long history of bad adjustments which other people then had to correct (ex: 10.1126/science.1216273 , 10.1126/science.310.5750.972). In fact, UAH's most recent version was adjusted in a way that reduces post-1998 warming, which would include the time-period David Evans looks at. It's this adjusted UAH analysis that David cites (figure 7 of: 10.1007/s13143-017-0010-y).

      David cherry-picked a low outlier, since UAH shows less warming than virtually every other analysis (table 2.3 on page S18 of 10.1175/2019BAMSStateoftheClimate.1).

      UAH is a bulk troposphere analysis, and thus does not measure the same thing NASA GISS' surface analysis. A number of other surface analyses validate GISS trend. And there are satellite-based (skin) surface temperature analyses such as AIRS; they show a warming trend large enough to rebut David's prediction as well ( from 10.1088/1748-9326/aafd4e ; 10.3390/rs12020218).

  3. Hi folks!

    The above sounds pretty accurate to me, including that general consensus that I'm highly likely to win both of my bets with David for the 10 year time period. I'll add that I've had some good luck with 2015-2016 strong El Nino, but it is partly balanced by subsequent weak La Ninas, as well as a weak El Nino in 2007.

    In my 2007 blog post, I guessed that I might lose one of my 6 bets. The most likely chance for that to happen is in the 10-year time period, where signal-to-noise ratio is lowest. I'd say I'm in even better shape for the 15 and 20 year bets.

  4. Many thanks David and Brian for your comments - I appreciate both of you taking the time to engage.

    David - a couple of things in response if I may, from my warmist perspective!

    Firstly, you mention the underlying trend that you were betting on. I've added a graph above showing the 20 years prior to 2007 (when you made the bet), with the trendline extended out to 2029. If 2015-19 simply followed that trendline, Brian would most likely win anyway. He probably wouldn't have needed the 2016 El Nino, although it's obviously contributing to a more emphatic win. You would have needed a downwards break from the underlying trend which I think is what you were expecting, and continue to expect for the medium term.

    Secondly, you say "the more credible satellite temperature record" shows around 0.13 deg. C warming per decade. Are you ignoring the RSS lower troposphere dataset? This shows 0.198 deg. C warming per decade for 1979-2018, according to Kevin Cowtan's temperature trend explorer at A simple average of UAH and RSS lower troposphere datasets gives 0.164 deg. C warming per decade for 1979-2018. For comparison, the average of the 5 surface temperature datasets shown at that site for the same period gives 0.177 deg C warming per decade.

    And "...the theory about carbon dioxide had no empirical evidence to back it...", that's for another day and probably another forum.


Post a comment

Popular posts from this blog

Climate bet scorecard

Climate bets: Bashkirtsev & Mashnich vs Annan