I would be somewhat convinced by the skeptic argument if they could tweak a GCM to show natural factors can account for the recent warming.
What recent warming? Do you mean the last Grand Solar Maximum which warmed the Earth from the late 70's until about 2000? Or some other warming that has escaped those of use huddle over heaters & dressed in warm clothes (Even in Victoria, Australia - it is very cold right now.
Why? If it cannot be demonstrated through running the numbers on the physics then what is it based on? A hunch? Line fitting? Statistics?
It seems odd to accept an explaination that doesn't have a known physical basis while dismissing a certain explaination that does have a known physical basis.
I think that this is a reasonable way to proceed SoCold.
But for two issues:
* This would not be by 'tweaking' a GCM that is obviously incorrect. As that model has, buried within it, assumptions that are wrong. What would be needed would be a model that actually attempted to model the real world without these guiding design assumptions - and where when something was unknown - then it was not guessed. This makes things somewhat difficult as convection and clouds and their effects (more powerful than simple radiation) are unknowns. I would think with a different mind-set it would be relatively simple to replace small effects like variations in gas absorption of IR with basic parametrization. But this is not feasible with large unknowns. This honest way forward then is to say 'we cannot model this as NO-ONE UNDERSTANDS some of the major aspects of the chaotic workings of the climate'. Chaotic systems are extremely difficult to model when all aspects are known and understood. When several major aspects are at best guessed and some are just not known - any correlation of model output with reality will be pure chance.
This raises the second:
* Is 'running the numbers on the physics' really feasible? The climate is a chaotic system driven by chaotic sub-systems that react and feedback on each other dependent on their various states at the time. Some of these systems such as the PDO and ENSO and the associated trade winds that cause them and are themselves driven by them are extremely complex and poorly understood. It is not even possible for the turbulent fluid flow in a gas turbine engine to be modeled with accuracy where EVERY input is known and controlled. Yet climatologists think they can model the chaotic climate systems when they don't even understand the known inputs and have no knowledge of some other inputs? We are not talking about climate issues only known to the erudite - meteorologists and climatologists cannot even agree on the effect of CLOUDS. Now this might seem strange to people who have known about clouds since they were infants - but there is no agreement on their effects. As is seen quite readily from satellite imagery and for those North of 50oNorth seen almost all winter and most of the summer - clouds of various types from low stratus to frontal clouds and convective clouds are actually quite common meteorological phenomena - yet THEY ARE NOT UNDERSTOOD.
So my thoughts after considering these two issues are that your challenge indicates that you do not understand what you are asking. Science is not clever enough to 'model' 3 bodies in orbit around each other - yet the infinitely more complex climate can be modeled by 'tweaking' someone's failed GCM model?
This leaves your dismissive
"If it cannot be demonstrated through running the numbers on the physics then what is it based on? A hunch? Line fitting? Statistics?"
I think that the current GCM models that are all incapable of modeling the current climate changes correctly, have proven that 'you cannot demonstrate _ANYTHING_ through 'running the numbers on the physics'. You are looking for a La Place's Daemon.
So what is the alternative hypothesis based on - Well if a system is too complex to model or not fully understood - what can be done is to treat it as a black box or gray box, and assess its reaction to stimuli. This also requires some understanding of the length of time from stimulus to reaction, as well as any periodicity in the stimuli. It also requires a search for the stimuli that appear to have an affect on the system.
Record all the inputs into the 'gray box' and record the behavior of the gray box system. Then see if there is any association or correlation between the stimuli and any behavior changes in the system or its outputs. This is called SCIENCE.
It should be noted here that from the climate system point of view CO2 is CO2. Nature does not react differently to CO2 from an SUV to CO2 from mature trees' respiration.
Over a long period several thousands of years of paleoclimatology (but only perhaps as far as the current continents and mountains exist.) variance in CO2 (measured from proxies) has NOT correlated with global temperatures (measured from proxies). However, solar activity (measured from proxies) HAS correlated with global temperatures (measured from proxies). Not only that but using the periodicity of the solar changes - it has been possible to forecast the climate - with far more accuracy than GCM.
So now the cry is "AHHH but you cannot say _how_ changes in solar activity cause the changes in the climate - so the connection cannot be proven or accepted."
Well by the same argument, the current climatologists cannot agree on how CLOUDS affect the climate - so nor can the GCM models be accepted.
At least the Solar cycle approach does correlate with climate changes with accuracies over several thousand years and appears to forecast them with some accuracy. Whereas the GCM based on CO2 causes global warming have no correlation even in the last 8 years.
Finally The normal state of the Earth's climate appears to be Ice Age - with ice several kilometers thick on NH continents as far south as ~50o North. The intervening inter-glacial periods are short. Yet we find that climatologists are calling 'normal temperature' those temperatures that pertained in the 1970s to 1990s. A period that matches the 'satellite' era and also a period of high solar activity. This is self important assumption that 'if it is in my lifetime it is normal' is INCORRECT. Normal for the Earth is cold - VERY COLD and yet people are concerned about warming??
What would be needed would be a model that actually attempted to model the real world
GCMs are such models. There's no branch of alternative physics that GCMs use. GCMs are based on accepted physics and are attempts to actually model the real world.
I disagree modelling is a waste of time because the climate is too complex/chaotic. The reason I disagree with this is that all the models, started from different starting conditions show similar overall warming trends. Clearly on long timescales and with sufficient forcing climate models give a robust general result (ie one that isn't affected by small changes to start conditions or small differences in design or parameterization). I suspect this is simply because the co2 forcing is so high and it dominates producing a clear and unambiguous signal.
The only plausible argument against models is that current understanding of physics may be significantly wrong. So we have a choice to either accept the result from knowledge we do have or to embrace the possibility of knowledge we don't have.
As for proxy records, they compatible with the models. The proxy records are themselves used to constrain the physics which eventually goes into the models.
edit: and the solar record from the 70s to 00s cannot explain the warming in that time. If it could then a model would be able to show that.
I'm going to model my pressure cooker using established physics. Because I don't believe that water vapour has any effect I am going to ignore this and only model the CO2 in the cooker. For some reason it never seems to work as expected.
Models will only be viable if you model everything, else garbage in garbage out.
"Then you aren't modelling using established physics..."
Which is precisely why all the AGM models fail. Ignoring solar, convection, ocean currents plus applying almost arbitrary adjustment values just to give the correct results for todays temperature does not make a valid model. The IPCC's models are no better than my pressure cooker example.
You cannot selectively apply "established physics" as "established physics" tells you this will not work.