Having just read beso's defence of climate models, I can affirm he does not work for the University of East Anglia. Here is their personal opinions, never expected to be public but since yesterday now are:
/// Climate Models ///
<3111> Watson/UEA:
I'd agree probably 10 years away to go from weather forecasting to ~ annual
scale. But the "big climate picture" includes ocean feedbacks on all time
scales, carbon and other elemental cycles, etc. and it has to be several
decades before that is sorted out I would think. So I would guess that it will
not be models or theory, but observation that will provide the answer to the
question of how the climate will change in many decades time.
<5131> Shukla/IGES:
["Future of the IPCC", 2008] It is inconceivable that policymakers will be
willing to make billion-and trillion-dollar decisions for adaptation to the
projected regional climate change based on models that do not even describe and
simulate the processes that are the building blocks of climate variability.
<2423> Lanzante/NOAA:
While perhaps one could designate some subset of models as being poorer in a
lot of areas, there probably never will be a single universally superior model
or set of models. We should keep in mind that the climate system is complex, so
that it is difficult, if not impossible to define a metric that captures the
breath of physical processes relevant to even a narrow area of focus.
<1982> Santer:
there is no individual model that does well in all of the SST and water vapor
tests we've applied.
<0850> Barnett:
[IPCC AR5 models] clearly, some tuning or very good luck involved. I doubt the
modeling world will be able to get away with this much longer
<5066> Hegerl:
[IPCC AR5 models]
So using the 20th c for tuning is just doing what some people have long
suspected us of doing [...] and what the nonpublished diagram from NCAR showing
correlation between aerosol forcing and sensitivity also suggested.
<4443> Jones:
Basic problem is that all models are wrong - not got enough middle and low
level clouds.
<4085> Jones:
GKSS is just one model and it is a model, so there is no need for it to be
correct.
--------------------------------------
I think we've found a consensus there.