Previously, I have argued that climate models can duplicate history only because they are fudged.  I understand this phenomenon all too well, because I have been guilty of it many times.  I have built economic and market models for consulting clients that seem to make sense, yet did not backcast history very well, at least until I had inserted a few "factors" into them.

Climate modelers have sworn for years that they are not doing this.  But Steve McIntyre finds this in the IPCC 4th Assessment:

The strong emphasis placed on the realism of the simulated base state provided a rationale for introducing ‘flux adjustments’ or ‘flux corrections’ (Manabe and Stouffer, 1988; Sausen et al., 1988) in early simulations. These were essentially empirical corrections that could not be justified on physical principles, and that consisted of arbitrary additions of surface fluxes of heat and salinity in order to prevent the drift of the simulated climate away from a realistic state.

Boy, that is some real semantic goodness there.  We are not putting in fudge factors, we are putting in "empirical corrections that could not be justified on physical principles" that were "arbitrary additions" to the numbers.  LOL.

But the IPCC only finally admits this because they claim to have corrected it, at least in some of the models:

By the time of the TAR, however, the situation had evolved, and about half the coupled GCMs assessed in the TAR did not employ flux adjustments. That report noted that ‘some non-flux adjusted models are now able to maintain stable climatologies of comparable quality to flux-adjusted models’

Let’s just walk on past the obvious question of how they define "comparable quality" or why scientists are comfortable when multiple models using different methodologies, several of which are known to be wrong, come up with nearly the same exact answer.  Let’s instead be suspicious that the problem of fudging has not gone away, but likely has just had its name changed again, as climate scientists are likely tuning the models but with tools other than changes to flux values.  But climate models have hundreds of other variables that can be fudged, and, remembering this priceless quote…

"I remember my friend Johnny von Neumann used to say, ‘with four parameters I can fit an elephant and with five I can make him wiggle his trunk.’" A meeting with Enrico Fermi, Nature 427, 297; 2004.

We should be suspicious.  But we don’t just have to rely on our suspicions, because the IPCC TAR goes on to essentially confirm my fears:

(1.5.3) The design of the coupled model simulations is also strongly linked with the methods chosen for model initialisation. In flux adjusted models, the initial ocean state is necessarily the result of preliminary and typically thousand-year-long simulations to bring the ocean model into equilibrium. Non-flux-adjusted models often employ a simpler procedure based on ocean observations, such as those compiled by Levitus et al. (1994), although some spin-up phase is even then necessary. One argument brought forward is that non-adjusted models made use of ad hoc tuning of radiative parameters (i.e., an implicit flux adjustment).

Update:  In another post, McIntyre points to just one of the millions of variables in these models and shows how small changes in assumptions make huge differences in the model outcomes.  The following is taken directly from the IPCC 4th assessment:

The strong effect of cloud processes on climate model sensitivities to greenhouse gases was emphasized further through a now-classic set of General Circulation Model (GCM) experiments, carried out by Senior and Mitchell (1993). They produced global average surface temperature changes (due to doubled atmospheric CO2 concentration) ranging from 1.9°C to 5.4°C, simply by altering the way that cloud radiative properties were treated in the model. It is somewhat unsettling that the results of a complex climate model can be so drastically altered by substituting one reasonable cloud parameterization for another, thereby approximately replicating the overall intermodel range of sensitivities.