Garbage In, Garbage Out

The Mail on Sunday has obtained the final draft of a report to be published later this month by the UN Intergovernmental Panel on Climate Change (IPCC), the ultimate watchdog whose massive, six-yearly ‘assessments’ are accepted by environmentalists, politicians and experts as the gospel of climate science.

They are cited worldwide to justify swingeing fossil fuel taxes and subsidies for ‘renewable’ energy. 
Yet the leaked report makes the extraordinary concession that the world has been warming at only just over half the rate claimed by the IPCC in its last assessment, published in 2007.

Back then, it said that the planet was warming at a rate of 0.2C every decade – a figure it claimed was in line with the forecasts made by computer climate models.

But the new report says the true figure since 1951 has been only 0.12C per decade – a rate far below even the lowest computer prediction. 
The 31-page ‘summary for policymakers’ is based on a more technical 2,000-page analysis which will be issued at the same time. It also surprisingly reveals: IPCC scientists accept their forecast computers may have exaggerated the effect of increased carbon emissions on world temperatures – and not taken enough notice of natural variability. 
They recognise the global warming ‘pause’ first reported by The Mail on Sunday last year is real – and concede that their computer models did not predict it. But they cannot explain why world average temperatures have not shown any statistically significant increase since 1997. 
They admit large parts of the world were as warm as they are now for decades at a time between 950 and 1250 AD – centuries before the Industrial Revolution, and when the population and CO2 levels were both much lower. 
The IPCC admits that while computer models forecast a decline in Antarctic sea ice, it has actually grown to a new record high. Again, the IPCC cannot say why. 
A forecast in the 2007 report that hurricanes would become more intense has simply been dropped, without mention.

This year has been one of the quietest hurricane seasons in history and the US is currently enjoying its longest-ever period – almost eight years – without a single hurricane of Category 3 or above making landfall.
 Lies, damned lies, and statistics, indeed.

Nothing in this article surprises me in the slightest. A computer model is only as good as its assumptions. Anyone who knows how to program models into statistical packages, or program his own models in languages like C++ or JAVA- I can do the first easily, and the second at a pinch- knows this basic truth. I work with financial models involving some very strong assumptions about the nature of financial markets. I know those assumptions very well, which is why I know the limitations and problems with the models that I work with. That is why I happen to be very, very good at what I do- because I know when my spreadsheet model is spitting out something accurate, and when it's spitting out complete nonsense.

Financial pricing models, like Black-Scholes-Merton for pricing vanilla options, or the Heath-Jarrow-Morton framework for interest rate derivatives, are generally well understood by practitioners. The literature is publicly available. The expertise is spread out across many, many organisations the world over. The code that you need to write to build your own Black-76 or SABR pricer is publicly available- I have a lot of R code designed to do exactly this sitting in my personal files, for instance.

The same assuredly cannot be said of climate models. Those models are proprietary, owned by large organisations that have a vested interest in public funding and thus have no desire to expose themselves to sunlight in order to show just how bad their predictions really are. The models are not transparent, it's not easy for a programmer like me to look at the code on a public website or repository and then try to back-test that code against my own data sets. And as I've pointed out before, many of the most commonly used models for climate forecasting are poorly programmed and built with an intrinsic assumption that temperatures increase in direct proportion with CO2 emissions.

If your model's predicted value does not match the actual value produced in the real world, your model is garbage. It's that simple. Unfortunately, the climate sophists of this world don't seem to particularly care about little things like "facts" or "statistics" or "real science".


  1. The nastiest prediction shows a change in temperature of about ~1.25 degrees in 40 years. Honestly, that doesn't sound that bad.

    One of the first things you learn in physics 101 is that fundamental equations like F=ma are models and don't accurately represent reality. There is no reason to trust models with multiple variables based on insufficient data; because logically, mapping long term climate change based off about 50 years of data is rather ambitious at best and a total failure at worst.

    I think Sheldon would agree that climate science - like geology - isn't even a real science.

    1. Bazinga.

      (I stopped watching "The Big Bang Theory" right around the time that Keoni Galt published a screed against it a while back, but I do tend to agree with Sheldon on quite a few things. Not least of which is his theory that women are "bat-crap crazy".)

      While I disagree with the specific idea that F = ma doesn't accurately represent reality- I would say that it is more precise to state that this equation works as a special case of a completely sound general theory- I do agree that models that try to predict the future without accurately backtesting against the past are simply not worthwhile.

      Climate data sets go back thousands of years, but very few, if any, models that I've heard of have successfully backtested against those data sets properly. Let me put it this way- as much as people hate bankers and traders, if our risk managers tried to deploy a model that did not successfully backtest within acceptable parameters against at least 3 years' worth of data, they would be promptly fired. Yet in climate science, politicians and "scientists" use shoddy models and shaky conclusions derived from them to justify trillion-dollar boondoggles, because science, and they all still have jobs- including Michael "Hockey Stick" Mann. That is flatly unacceptable.


Post a comment

Contact the Didact:

Popular Posts