“Beware the Tipping Point, my son!
That little wee thing, the difference it makes!
Beware its Connector Theory, and shun
Its luminous Balderdash!”
With appropriate apologies to the great logician whose words I abused, we consider the mysterious, misunderstood Tipping Point. Beware! for it is as rare as a Jabberwocky, everywhere lurking and ready to pounce, to devour and overcome. We can but sense its approach, its true shape is camouflaged. We only know that it was here after it has gone. Alas!
Or such is the popular imagination. The truth, as it frequently is, is mundane.
In physics, a tipping point is that point in a dynamic system where the system switches from one state of equilibrium to another. Take a cardboard box and flick it with your finger. If the box is light enough, it will wiggle a bit but it will return to its resting state. Now push it over, with one of the edges touching the floor acting as a hinge. Then let go. If you push just a little, the box will return to its original state (perhaps after bouncing around). But there will come a point which if you push past it, the box will fall over. That point is literally and physically the tipping point.
The equations of motion and the physical properties of the box govern the tipping point. And, of course, tipping points are present in many, but not all, physical systems. It’s also important to understand that the mere presence of a tipping point, its theoretical possibility, does not guarantee that this point will ever be reached. Finally, there is no “good” or “bad” about states of equilibrium. Those judgments require reference to non-physical criteria. Whether the box lies on this side or that is only interesting if you have money on the outcome or something breakable inside.
The climate is vastly more complex than a falling box. It is impossible—not just unlikely, but impossible—to model the climate precisely (actually, our expertise is such that we can only model an idealization of a box and not even a real box). We do not and cannot measure the climate well. The best models are gross simplifications where we abandon precision and instead seek to mimic statistical properties of the climate. Even that is a tricky, error-prone business.
A tipping point in the climate is that point at which the system changes from one state of equilibrium to another. It is not, it is certainly not, true that the state into which the system changes is inevitably worse than the previous state. If there are true tipping points in our real climate, passing through one does not in any way imply conditions must become worse. They might, they might not is the best we can say (conditions might even improve).
Secretary of Energy Steven Chu wants climate modelers to overcome their squeamishness and place equations for tipping points into climate models. He said, “To be sure, if you start to model the tipping points you put in much larger uncertainties, but there is a difference between uncertainty and inaccuracy.”
Chu is right that there is a difference between uncertainty and inaccuracy, but his reasoning is flawed. If you add, merely for the sake of adding, tipping points into model equations and those equations do not represent reality you introduce inaccuracy and increase uncertainty. And that is not the worst of it.
If you design a model that includes tipping points, and then run that model, you will likely find that the results evince tipping points. But it would be a mistake to issue the press release, “Tipping Points Found In Climate, Scientists Concerned.” You included the tipping points, it is therefore no shock you found what you included. You audience becomes more certain than it deserves to be.
Why, this is just like building into a model subroutines which provide for a positive feedback on temperature with carbon dioxide. When that model is run, it will show positive feedbacks on temperature with carbon dioxide. Is it worth a press release to say so?
Chu is worried that the “long tail of the damage tail is out there” and that climate models aren’t capturing these statistical properties. It is true that climate models do not (so far) make skillful predictions, thus they cannot adequately assess risk. Whether the risk is actually higher or lower is unknown. Including tipping points in climate models to artificially inflate risk, just to produce fatter “long tails”, is not good science. Modelers are right to be leery of tipping points.
Thanks to Marc Morano for the original link.