Back in February, Pat Michaels dropped something of a bombshell. It was an exposé of the Achilles heel of alarmist climate science.

He did it in a lecture, just posted to YouTube on June 11, at a joint meeting of the Independent Institute and the National Association of Scholars. What he showed was that practically all the climate models on which the Intergovernmental Panel on Climate Change and lots of national governments rely are wrong—hopelessly, egregiously, starkly wrong. And all (but one) in the direction of exaggerating CO2’s warming effect.

Why? Because modelers tune their models to reproduce the 20th-century temperature record. Yet the models supposedly model how global average (near-surface) temperature responds to rising atmospheric CO2 concentration. And the problem with that is that there’s no way the first half (roughly) of the 20th-century temperature record was driven by CO2.

Michaels demonstrates that by showing that, if the comparatively small amount of added CO2 in the first half of the century drove as much warming as occurred then, then the much greater amount of CO2 added in the second half should have added a whole lot more warming than actually occurred.

Having made that point, Michaels reminds us that good scientists, when observations contradict predictions based on their models, revise the models so as to come closer to the observations. But that’s exactly the opposite of what’s happening in the alarmist climate modeling community.

The “Coupled Model Intercomparison Project” 5th generation models (CMIP-5) simulated, on average 2.75 times as much warming as actually observed by radiosondes (temperature sensors carried aloft by weather balloons) from 1979–2018, as Michaels showed with this graph, courtesy of John Christy:

Now the CMIP-6 models are starting to come out. One would expect that refinements would bring their simulations closer to the observations. The opposite is the case. The new models—as many as had been reported as of late 2019—simulate not 2.75 but roughly 3.5 times the warming observed by radiosondes, as shown in this graph:

In short, far from improving the models, the modelers are making them worse than ever. After the hundreds of millions of dollars spent on them, one would hope for a different outcome.

Perhaps it’s driven not so much by honest scientific endeavor as by an agenda.