Whenever you are trying to model some complex phenomenon, the fit of the model to the data can be improved by adding more adjustable parameters. A complex phenomenon will usually be dependent on a large number of factors, but the fact that the model fits the data better when you incorporate an additional factor may or may not mean that new factor is important: it might just mean that the additional parameter(s) you have incorporated are improving your fit. This is another thing the David Sangster told me: ‘With enough adjustable parameters, you can fit a camel.’
So there is a tension between the complete model, which contains all the factors that ought to be physically important – but might be meaningless because of all the guesstimated parameters you have put in to quantify these factors- and the simple model, which ignores things that might be physically important, but also avoids adjustable parameters. If you go too far in one direction, you get a model that can fit any possible data; too far the other, you get the well-known ‘assume a spherical horse’ punchline.
This also means that when you are modelling a complex phenomenon, you will tend to base your model on the processes that are best known, where you don’t have to pick numbers out of the air for your adjustable parameters, and you will ignore if you possibly can the role played by processes that are less understood, which would force you to bring in rubbery parameters.
Now to place myself beyond the pale. Some time ago I made the assertion:
‘Anthropogenic global warming is a fact, but we shouldn’t do anything about it.’
The second part of this statement is a considered opinion, based on facts and reasoned deductions from them. The first part of this statement, I have realised over the last few months, is based on an irrational mood.
That is: in the laboratory, and considering the atmospheres of the planets in toto, there is a perfectly splendid mechanism by which increasing the concentration of atmospheric carbon dioxide should increase temperatures. It is a really good mechanism, based on rock-solid physics. But is there any evidence that this mechanism is responsible for observed temperature change globally? Evidence, in the scientific sense, is where a model has predictive value: it does not just fit the data we have, but tells us what future data is going to look like. I did not examine this question before I made the statement above. Instead, I relied on the irrational mood that it seemed like wishful thinking that there was some sort of feedback mechanism providentially cancelling out this Greenhouse warming effect.
Let us consider these two famous graphs:
What do they tell us? They show us a correlation between carbon dioxide concentration and average global temperature. They also tells us, very clearly, that there are factors other than carbon dioxide which contribute to the world’s temperature.
We could also draw graphs that show some sort of a correlation between sunspot activity and global temperature, and earthshine and global temperature, and the number of pirates and global temperature. The last of these three graphs would be a joke circulated by the Church of the Flying Spaghetti Monster. The other two are graphs where it is easy to construct a testable mechanism for how the correlation might work. These mechanisms are not as solid or as well understood as the Greenhouse mechanism. They rely on more rubbery adjustable parameters. If we ignore them, do we have a spherical horse? If we include them, do we have a camel?
What is signal, and what is noise, in the Hadcrut3 temperature curve?
An idea that was in fashion when I was an undergraduate was the Gaia hypothesis of James Lovelock. You don’t hear much about it nowadays. You might remember that it was all about negative feedbacks keeping the global ecosystem in balance, life keeping things tickety-boo for life. I bring it up here as a hand-waving justification for a recent shift in my irrational mood: given that there is a grain of truth in Lovelock’s ideas, it now seems to me reasonably likely that there would be a negative feedback mechanism tending to minimise the effects of any carbon dioxide we add to the air.
I must now revise my assertion:
‘Anthropogenic global warming is a conjecture with limited predictive value, and we shouldn’t do anything about it.’
And I have to apologise for some of the slighting references to global warming denialists I have made previously.
And unfortunately I have nerfed one of the major motivations for establishing this blog, which was to use any perceived authority associated with my real name to push the line that we shouldn’t take any action to stop anthropogenic global warming. By denying AGW to be a fact, I have placed myself outside the pale of civilised discourse and disqualified myself from making any statements on the issue that will be taken seriously.
Son cosas de la vida…
