" ... In 2001 the IPCC noted, quietly and near the back of one of its reports, “In climate research and modelling, we should recognise that we are dealing with a coupled non-linear chaotic system, and therefore that the long-term prediction of future climate states is not possible.”
So the best they could do was run the models repeatedly and hope reality would lie somewhere in the mix of different outcomes.
Which still wasn’t very good since the IPCC also warned “Integrations of models over long time-spans are prone to error as small discrepancies from reality compound.”
OK, but don’t climate models seem to reproduce much of the 20th century well?
Yes, but only because scientists “tune” them to fit the past once they know the result, which is the opposite of making reliable forecasts.
Two decades later modelers surveyed about this awkward point didn’t even agree on whether retrofitting models to predict the past was proper scientific procedure.
The question of whether the twentieth-century warming should be considered a target of model development or an emergent property is polarizing the climate modeling community, with 35% of modelers stating that twentieth-century warming was rated very important to decisive, whereas 30% would not consider it at all during development.
... Wait, how can there be basically a 50/50 split among climate scientists over the question of how they build their models, when there’s a supposed 97% consensus on everything else?
Simple, the 97% myth is just that, a myth, and if you haven’t done so yet, have a look at our video on the subject which, we’re proud to note, is quickly closing in on a million views.
... climate scientists don’t even agree on how their models should be put together when they’re the ones doing it.
So you can hardly expect them to agree on how the real climate system is put together.
... tuning is often seen as an unavoidable but dirty part of climate modeling, more engineering than science, an act of tinkering that does not merit recording in the scientific literature.
... there’s a downside to tuning, and to pretending it doesn’t happen.
Although tuning is an efficient way to reduce the distance between model and selected observations, it can also risk masking fundamental problems and the need for model improvements."