[PlayadelSoul]

Wow, I found an IPCC member who is skeptical.
BBC NEWS | Science/Nature | No consensus on IPCC's level of ignorance

Slim models

To me, the elevation of climate models to the status of definitive tools for prediction has led to the temptation to be over-confident.

Here is how this can work.

Computer models are the basic tools which are used to estimate the future climate. Many scientists (ie the mere mortals) have been captivated by an IPCC image in which the actual global surface temperature curve for the 20th Century is overlaid on a band of model simulations of temperature for the same period.




Models 'key' to climate future
The observations seem to fit right in the middle of the model band, implying that models are formulated so capably and completely that they can reproduce the past very well.

Without knowing much about climate models, any group will be persuaded by this image to believe models are quite precise.

However, there is a fundamental flaw with this thinking.

You see, every modeller knew what the answer was ahead of time. (Those groans you just heard were the protestations of my colleagues in the modelling community - they know what's coming).

In my view, on the other hand, this persuasive image is not a scientific experiment at all. The agreement displayed is just as likely to do with clever software engineering as to the first principles of science.

The proper and objective experiment is to test model output against quantities not known ahead of time.