QUOTE (Wheel @ Aug 25 2008, 12:10 pm)

I think it unlikely that a thesis proven by reference to computer models and no experimental data would be accepted for publication by a journal of applied science. There are journals (International Journal of Theoretical Physics) for theoretical disciplines, however the climate isn't one of them.
This is exactly my point. Scientific models are sometimes published for the sake of advancing the researcher's reputation as well as to provide information on new techniques, as well as stake out an research area. It is appropriate in my mind that testing models against real data is done by teams with no vested interests in the results where possible.
QUOTE (Wheel @ Aug 25 2008, 12:10 pm)

There is a paucity of data, since the climate is ephemeral, and often very localised. For instance, in the past few years Arctic ice has been decreasing (until this year), while Antarctic ice has been increasing.
Climate, as the term is conventionally used, is not localised and refers to an aggregate of local conditions. Weather patterns in the Arctic vs weather patterns in the Antartic are local and by themselves don't say anything conclusive. There is a tonne of weather and other data available out there, not least from studies aimed at weather prediction. We have ice cores dating back thousands of years, not to mention human records, sea temperatures etc. The problem is not lack of data, it's trying to integrate it so the models match reality.
QUOTE (Wheel @ Aug 25 2008, 12:10 pm)

Predicting individual responses isn't comparable to forecasting complex chaotic systems. For the most part, experiments are carefully designed to reduce the number of variables or simplify systems, as you illustrated.
The same principles of reducing complexity to key elements, experiementation and comparison against result apply. All scientific principles that apply to testing relatively simple propositions also apply to large enterprises like climate modelling. Modelling chaotic systems goes on all the time, and there are many chaotic system we have some understanding of (turbulent fluid flow, epidemiology etc).
QUOTE (Wheel @ Aug 25 2008, 12:10 pm)

In engineering the test is simple - often, it's as simple as 'did it break'? The AGW thesis can't be tested like that, and I don't see how it's possible to disprove it without waiting for a few decades. Even so, there'd be no control.
If only! But as you implicitly admit here, we can disprove these models it if we wait. I'd argue we can do better using historical data. And consider what you mean by "control". By definition we have only one planet to test on, and expecting control cases is a bit unrealistic. Control cases are not an intrinsic requirement of a theory, but they're nice to have if available.
QUOTE (Wheel @ Aug 25 2008, 12:10 pm)

By the way, the IPCC's models didn't predict the last 10 years' temperature plateau, so you have to wonder about them.
Indeed. But there is enough reports of other data out there indicating global average temperatures are rising, but I've not enough time to gather some examples together unfortunately. The scientific consensus that on average the trend is upwards, but as far as I know it does not require that each year must be steadily hotter.
QUOTE (Wheel @ Aug 25 2008, 12:10 pm)

I think the people who contribute to the International Journal of Forecasting would disagree with you.
That journal is just a mash-up of long term modelling techniques in many disparate fields as far as I can tell, and is only about 20 years old. It is not a journal of a specific scientific discipline at all, which supports my contention there is no such thing as a forecasting scientist as a separate field.