Scott Johnson at Ars Technica examines the question.
The article hits some of the key points, but in my opinion lies in a somewhat infertile place in the conversation; too long for a quick reading and neither deep enough to merit serious study nor engaging enough to hold the interest of someone reading out of general interest.
The in depth explanation genberal interest article that I’ve always intended to write, of what a climate model is, what its correct uses are, and how it can be misused, remains unwritten and remains on my endless to-write list.
I think the best bit of it, the sort of thing I would aim for in such an article, is
If you only tune in to public arguments about climate change or read about the latest study that uses climate models, it’s easy to lose sight of the truly extraordinary achievement those models represent. As Andrew Weaver told Ars, “What is so remarkable about these climate models is that it really shows how much we know about the physics and chemistry of the atmosphere, because they’re ultimately driven by one thing—that is, the Sun. So you start with these equations, and you start these equations with a world that has no moisture in the atmosphere that just has seeds on land but has no trees anywhere, that has an ocean that has a constant temperature and a constant amount of salt in it, and it has no sea ice, and all you do is turn it on. [Flick on] the Sun, and you see this model predict a system that looks so much like the real world. It predicts storm tracks where they should be, it predicts ocean circulation where it should be, it grows trees where it should, it grows a carbon cycle—it really is remarkable.”
Some other actual appreciations of the amazing accomplishment appear in the article, but this one captures it most clearly. Still, the fact that climate models are actually a triumph of science doesn’t really shine through, unfortunately.
And there is this key point:
One surprisingly common misconception about climate models is that they’re just exercises in curve-fitting. The global average temperature record is fed into the model, which matches that trend and spits out a simulation just like it. In this (mistaken) view, having a model that compares well with reality is a necessary outcome of the process. This doesn’t demonstrate that climate models can be trusted to usefully project future trends, but this line of thinking is mistaken for several reasons.
There’s obviously more to a climate model than a graph of global average temperature. Some parameterizations—those stand-ins for processes that occur at scales finer than a grid cell—are tuned to match observations. After all, they are attempts to describe a process in terms of its large-scale results. But successful parameterizations aren’t used as a gauge of how well the model is reproducing reality. “Obviously, since these factors are tuned for, they don’t count as a model success. However, the model evaluations span a much wider and deeper set of observations, and when you do historical or paleoclimate simulations, none of the data you are interested in has been tuned for,” Schmidt told Ars.
This reads as overzealously trimmed in editing to me. It’s a crucial point, and indeed one which is widely misunderstood. Read any naysayer essay on modeling and you’re almost inevitably going to see the models dismissed as curve-fitting. How and why this is completely wrong deserves the attention of anyone interested in the topic.
This video pitch is not aimed at making that point, but anyone watching it through will see that what it is talking about is not just curve-fitting!