The main “evidence” for the CAGW hypothesis is that there exist climate simulations which (1) assume great sensitivity to CO2; and (2) are consistent with past temperatures. The problem with this is that in the universe of possible climate simulations, there must exist “false simulations,” i.e. simulations which are consistent with temperatures of the past 50 to 100 years, but do so by coincidence and are not accurate climate simulations. Such “false simulations” can be expected to diverge from reality and cannot be relied upon to accurately predict the future.
One can see that this is true by observing that there are many climate simulations in use out there with very different assumptions about climate sensitivity, and yet they all track past temperatures pretty well. Clearly at least some of these simulations are “false simulations.”
The fact is that it’s very difficult to simulate complex systems accurately. This is true because different elements of the system interact in many different ways, building up uncertainty at each step and making it increasingly difficult to accurately simulate the system over longer amounts of time. The classic example of this is predicting weather, which is very difficult to do more than a week or so in advance. So the default assumption should be that a simulation is unreliable.
Ok, so how do you know whether you have a false simulation or a good simulation? The answer is very simple: You test it. You have the simulation make predictions. If most of those predictions come true, then you can start having some confidence in the simulation.
Unfortunately, the simulations which have been used to predict CAGW have not been tested in this way. Instead they are tested by seeing how well they compare to past data. But this is silly. It’s very easy to make “predictions” with the benefit of hindsight.
So it turns out that the CAGW hypothesis is mainly based on simulations which are unreliable and untested. This is weak evidence at best.
Objections / FAQ
4.1 But the Simulations used to Predict CAGW are Based on Real Physical Principles
This is a bit like saying that Sunny Delight is “made with real juice,” or that the boardgame Risk is based on real geographical principles. In other words, all of the simulations necessarily contain simplifying assumptions which can lead to incorrect results. Even if the simulations are also based — in part — on undisputed physical laws.
The “physical principles” argument essentially implies that we can trust the simulations because their results follow more or less inexorably from basic first principles. But if this were true, then all of the simulations would give the same results, which they certainly do not. Again, the simulations also contain simplifying assumptions.
4.2 But Some Simulations Have Been Tested – Just Look at Hansen’s Simulation from the 80s
Sure, and Hansen’s simulation has been diverging from reality for some time now. Current temperatures are closest to Hansen’s scenario which assumed that CO2 levels would stabilize, which they have not. Of course it is possible that this result is not statistically significant. But in that case, it only shows that Hansen’s simulation has not been put to the test — if the results really are not significant.
4.3 But Skeptics Have Not Done any Better at Predicting Temperatures
That may very well be true, but this argument is attacking a bit of a strawman. I’m not claiming that I can accurately predict future temperatures — I’m claiming that Jim Hansen cannot. At least not as well as he thinks he can. Put another way, it’s not necessary for me to accurately predict the future to win this debate.
4.4 Perhaps Any One Simulation is Unreliable, But We Can Get Reliable Results by Averaging the Results of Many Simulations
This argument rests on the mistaken assumption that errors and mistakes will always cancel out. The assumption is incorrect because there may be “systematic error.” If most or all of the simulations miss the mark in a similar direction, then the average will be off. And there is no way to know that there does not exist such systematic error. The great Richard Feynman touched on this issue with the following parable:
Nobody was permitted to see the Emperor of China, and the question was, What is the length of the Emperor of China’s nose? To find out, you go all over the country asking people what they think the length of the Emperor of China’s nose is, and you average it. And that would be very “accurate” because you averaged so many people. But it’s no way to find anything out; when you have a very wide range of people who contribute without looking carefully at it, you don’t improve your knowledge of the situation by averaging.
The fact is that there are a lot of aspects of the climate which are still not understood very well by anyone. So averaging different simulations is a lot like the situation in Feynman’s parable.
4.5 Perhaps It’s Impossible to Simulate and Predict Weather over Periods of a Few Years, But It Is Possible over Long Time Periods Like 50 – 100 Years. It is Like the Difference Between Predicting the Percentage of Heads in 10 Coin Flips Versus Predicting the Percentage of Heads in 10,000 Coin Flips
It’s also possible that weather is chaotic over short AND long time periods. There is no a priori reason to believe that weather “evens out” over periods from 50 to 100 years. Indeed, during the Little Ice Age, temperatures were quite cold for well over 100 years. Could anyone have predicted the Little Ice Age beforehand? If the Little Ice Age was predictable, then what caused it? If you cannot answer that question, you must consider the possibility that weather is chaotic and unpredictable even over periods of 50 – 100 years.
At a minimum, nobody has shown me proof, or even compelling evidence, that climate is predictable on 50 – 100 year time periods. So it seems to me that if a butterfly in China can cause a storm in New York, it’s also possible that the same butterfly could cause a little ice age.