Science trumps politics and patronage: Al Gore is 'losing the bet' against scientific forecasting

Guest essay by Kesten C. Green

Global warming is a forecasting problem. The claim of the IPCC and sympathetic alarmists is that if we don’t stop emitting carbon dioxide, the Earth will be dangerously warmer in the future. How can they know that?

To put it another way, given the state of knowledge, are the IPCC using forecasting methods that are known to provide accurate forecasts?

In order to draw attention to this question, Professor Scott Armstrong in 2007 challenged former Vice President Al Gore to a bet on what would happen to global average temperatures over the next 10 years. Mr. Gore was getting a lot of media coverage at the time for his claims that temperatures on planet Earth were at a “tipping point” due to human emissions of CO₂. He nevertheless declined, on the grounds that he does not bet.

With Scott Armstrong, I had published a paper evaluating the procedures that the IPCC were relying on for their scenarios of future dangerous warming[1]. We found that the procedures violated 72 out of 89 relevant evidence-based forecasting principles.

If you are not sure on whether that is bad or not, think of how many violations of evidence-based procedures by ground crew or pilots that are typically associated with a major airline crash. One or two, perhaps? Bear in mind, too, that commercial air travel is a much simpler, and better understood, situation than global climate change.

How would you react if the pilot announced before take off that he was sure that your flight was special, and so he would not be following the usual procedures?

With the draconian nature of climate policy and regulatory responses to the global warming alarm, getting the forecasting right would seem to be a fundamental requirement for avoiding making the wrong decisions. What if, as some scientists believe, we are in for another period of long-term global cooling?

With these concerns in mind, Scott Armstrong went ahead with the bet, in order to highlight the importance of using evidence-based (scientific) methods for forecasting. We set up theclimatebet.com site to monitor progress on “The Global Warming Challenge”.

The Global Warming Challenge is a notional 10-year bet between Al Gore—represented by the IPCC’s 0.03ºC per annum “business as usual” increase in global average temperatures—and Scott Armstrong—represented by the Green, Armstrong, and Soon no-change forecast[2]. Specifically, Armstrong bets that temperatures will equal the 2007 annual average against the scenario that temperature will increases from that level at a rate of one-twelfth of the IPCC annual increment per month.

The arbiter of the bet is the University of Alabama and Huntsville’s lower troposphere global mean temperature anomaly series originated and maintained by John Christy and Roy Spencer from satellite data. The final day of reckoning will be early in January 2018, when the UAH observation for December 2017 is released.

Is it really possible that the simple no-change forecast of 21st Century temperatures is better than the IPCC projections from expensive and complex computer models?

Yes, it is. That conclusion is consistent with the evidence Scott Armstrong and I present in our recently published review of evidence on the effect of complexity on forecasting[3]. We found that using complex methods increases forecast errors relative to the forecasts from simple methods that decision makers could understand by 27% on average. We expect that the results of The Climate Bet will increase that average.

Even more importantly, the IPCC has no regard for the Golden Rule of Forecasting[4]. The Golden Rule is to be conservative when forecasting by staying close to cumulative knowledge about the situation and about forecasting methods.

The IPCC scenarios are derived entirely in contravention of that fundamental rule of forecasting. The dangerous manmade global warming scenarios are premised on the unscientific (unconservative) assumption that “this situation is different”. Forecasting research tells us that ignoring the Golden Rule typically leads to an increase in forecast error of around 45%.

So how is the Global Warming Challenge progressing?

An up-tick in temperature anomalies in June saw Mr. Gore and the warming scenario score the first win against the no-change forecast since January of 2013, nearly two-and-a-half years ago. The outlook for the dangerous warming scenario remains bleak, however. Over the 7.5 years of the Armstrong-Gore Bet so far—we have now passed the ¾ mark—the errors that have arisen from projecting temperature to increase at a rate of 3°C per century are more than 50% larger than the errors from the no-change forecast.

The chart presents the entire history of the bet, to date, and the table shows the latest three years of data from UAH, and the Armstrong and Gore forecast figures.

[1] Green, K. C. & Armstrong, J. S. (2007). Global warming: Forecasts by scientists versus scientific forecasts. Energy & Environment, 18(7+8), 997–1021. Available online from http://www.forecastingprinciples.com/files/WarmAudit31.pdf

[2] Green, K. C., Armstrong, J. S., & Soon, W. (2009). Validity of climate change forecasting for public policy decision making. International Journal of Forecasting, 25, 826–832. Available online from http://www.kestencgreen.com/gas-2009-validity.pdf

[3] Green, K. C. & Armstrong, J. S. (2015). Simple versus complex forecasting: The evidence. Journal of Business Research, 1678–1685. http://dx.doi.org/10.1016/j.jbusres.2015.03.026.

See also simple-forecasting.com.

[4] Armstrong, J. S., Green, K. C., & Graefe, A. (2015). Golden Rule of Forecasting: Be conservative. Journal of Business Research, 1717–1731. http://dx.doi.org/10.1016/j.jbusres.2015.03.031

See also goldenruleofforecasting.com.