Falsifying models

In some scientific disciplines it is difficult to successfully advance a model, especially those based on observational science alone (astrophysics and earth sciences immediately comes to mind). Yet models can still be falsified straightforwardly based on those same empirical observations. The earth’s QBO is one I am working on, and according to Richard Lindzen is easily falsifiable :


Our QBO model should be easy to falsify as all that it requires is to demonstrate that the QBO cycles and lunar cycles are incommensurate — i.e. that they don’t line up according to Lindzen’s rules shown above.

The alignment of lunar and solar nodes with QBO cycles

Or is it not easily falsifiable? The cycles match so, again according to Lindzen, it is not so easy to dismiss.

Of course, the obvious problem with earth sciences models, and climate science models in particular, is that they can’t be falsified by the controlled experiments — since a lab as big as the solar system can’t be built. Thus models are only gradually accepted as further evidence is collected.

So the QBO is a prime example of a climate model that could be straightforwardly falsified based on the premises that Lindzen made. As it turns out, the behavior of the QBO, which Lindzen built his career around, is an EXACT match to a basic lunar+solar cyclic forcing. Again, easy enough to falsify.

Yet, earth sciences are glacially slow to pick up on new models, mainly because of this lack of experimental control. Murray Gell-Mann noted this during his celebrated tenure — “the more the evidence was there, the less they believed it”

Murray Gell-Mann 2009 interview : “Gell-Mann on Conventional Wisdom”

So with that, this is what happened yesterday:


That was me that Roundy is referring to. I was simply following through on a suggestion that Richard Lindzen made concerning identifying the role of atmospheric tides. After all these years, it’s still the case that the QBO cycle matches the lunar tidal cycle. Hard to falsify. I bet Roundy can’t falsify it either, other than through some non-testable hand-waving.

Roundy’s comments started from this thread, where they are questioning an alternative model.