For the record: I like working on computers. I followed quite some computer trainings in the last decades and am the humble creator of several webapps in the last 8 years. I have no problem stating that computers are useful things, to say the least.
Also for the record: I like computer models! I use them regularly at work (okay, nothing to do with climate, but financial). Models are useful things. A lot of things can be modeled and afterwards what if-scenario’s can be calculated from that model. A lot of things could be learned from those kind of analyses. I am glad such tools are available.
That being said, mathematical models have their limitations. There are a few important things to consider. It needs as much data of the components as the process exists of. The more components are included, the better, but at least the major components should be there. As accurate as possible. Some will have a direct relationship (nice), other an indirect relationship (therefor the need to fill in with assumptions). The more gaps/assumptions, the less likely this result would actually be observed in real world (or the more likely the result will diverge from what happens in the real world).
The issue I have with climate models have is that their field of study is an intrinsic chaotic system. The more gaps in our knowledge of the system, the more gaps in the representation and/or the more assumptions are used in the model and the more uncertain the output will be. Mathematical models of chaotic systems are not evidence of anything. Their output depends on the assumptions that were put in by the creators in the first place. For example, if CO2 is assumed the major component that defines climate, then it should not be a big surprise that climate models “find” that the climate is very sensitive towards CO2. Yet the standstill of temperatures learns us that there are other components that are as important as CO2 is.
Although I knew the pitfalls of modeling of chaotic systems, I never considered this a point in climate models. I assumed scientists had enough knowledge of the climate to favor reliable output. I thought the climate models were weather models, but better. Weather models work quite well on short term, why shouldn’t climate science be so advanced that their models would work as well as the weather models? If you ever want a reason why the public need to be convinced that “the science is solid”, this is it.
Look it like this: we know our knowledge of the system is not complete. We know there are pieces missing, but we don’t know how many. Could be 1%, 10%, 50%, 90%,… We know we discover new pieces on a regular basis, so the total count of missing pieces could be high. Well, suppose such a system and, based on what we do know, we then claim to be able to forecast what this system will do almost 100 years in the future from now. Would we believe any of this? Would we take drastic measures based these calculations?
Let me recapitulate: climate is incredibly complex, many elements are not understood, many things are still being under study, there are a lot of discussion on what the major component is. Then a mathematical model is constructed on this and even less elements (clouds, cycles,…) are put into it. So how exactly would the output meaningful for predicting … wait for it … a century ahead?!?!