Bon Mot - In an earlier set of posts I pointed out that the actual temperature results of the past decade had to be "adjusted" to make it look as they were going up. That wasn't relevent to the results from models, but you have hit on a point I forgot in my first post on this article.

That is, model runs can also be "tuned" to replicate historical results, although it is difficult as you rightly suggest. They have to run the model multiple times, with different parameters - lots of adjustments can be made - until they find a result that is stable (doesn't crash or give obviously wrong results) and is something like the historial line for a few decades.

They they have the front to suggest that the model, because it followed the historical line will then say something useful about the future. Right! Look at the graphs of model results in the 2007 IPCC report. They don't "hindcast" for more than a few decades because they don't hindcast the little ice age.. its not there at all in the models. They are, in other words, nearly useless. the report gets around that obvious problem by simply not showing model results back beyond a few decades..

The following extract from President Václav Klaus at the Global Warming Policy Foundation’s Inaugural Annual Lecture.19 October 2010.

“To reduce the interpretation of the causality of all kinds of climate changes and of global warming to one variable, CO2, or to a small proportion of one variable – human-induced CO2 – is impossible to accept. Elementary rationality and my decades-long experience with econometric modeling and statistical testing of scientific hypotheses tell me that it is impossible to make strong conclusions based on mere correlation of two (or more) time series. In addition to this, it is relevant that in this case such a simple correlation does not exist.
The rise of global temperature started approximately 150 years ago but man-made CO2 emissions did not start to grow visibly before the 1940s. Temperature changes also repeatedly moved in the opposite direction than the CO2 emissions trend suggests.
Theory is crucial and in this case it is missing. Pure statistical analysis does not explain
or confirm anything. Two Chinese scientists, Guang Wu and Shaomin Yan, published a
study in which they used the random walk model to analyze the global temperature
fluctuations in the last 160 years. Their results – rather unpleasantly for the global warming alarmists – show that the random walk model perfectly fits the temperature changes.
Because “the random walk model has a perfect fit for the recorded temperature … there is no need to include various man-made factors such as CO2, and non-human factors, such as Sun” to improve the quality of the model fit, they say. It is an important result. Do other models give a better fit? I have not seen any.”