Rained yesterday here in the city of cities. Must be because of climate change, right? Hey. The climate did change and it did rain. What more evidence do you want?
Bonus trivia question: name a period in which the climate on this island earth never changed.
That’s right, guppies. It’s a trick question. There is no such period! The climate has always changed; therefore, it is rational to suppose that it always will.
“Briggs, you fool. When people say ‘climate change’, they don’t mean the climate changed. They mean the nature of the climate has remained stationary up until some point, after which is changed—and even fools like you know this means a change in the statistical nature of the climate.”
So that if you say, as I’ve heard you say, that temperature is “normally distributed” and yesterday the high was 62F and today the high is 38F, you’d say the temperature didn’t change?
“Well, you know what I mean.”
Funny definition, that. Besides, even if the model—your normal distribution—remained the same from day to day, it’s still true that some thing or things caused the high to plunge (i.e. change), right?
“The climate didn’t change. The temperature did.”
That either makes no sense or it’s a circular argument. Your model—your normal distribution—doesn’t say diddly about what caused the observed change. Do you agree?
“It doesn’t have to! If the model doesn’t change, the climate didn’t change!”
You avoided the question while committing the Deadly Sin of Reification. You’re defining “climate” as your model, and not as the real-life observations. At least you’re in good company. Like that of Gerrit Hansen, Maximilian Auffhammer (cool name), and Andrew R. Solow who wrote a peer-reviewed paper of the same name as today’s post in the Journal of Climate, which makes the same mistakes you make (vol 27, 15 Nov 2014, pp 8297-8301).
These authors define a “stochastic” “point process”, which is to say, a probability model, which describes the uncertainty in event occurrences. Like, say, blog posts. Once a day here—a good and highly accurate model! Which is not meant as a joke. Why?
Their probability model, like any probability model, announces, conditional on specified premises, which usually include past observations, the probability some proposition is true. Thus, given our blog point process model and the history of posts at the venerable WMBriggs.com, you might say the probability “A post shows tomorrow, 11 December 2014″ is high. Tune in tomorrow to see how useful this model is.
Again, this isn’t a joke. I started this blog seven, eight years ago. Back then the climate was different, and so was my posting frequency. It was on the order of twice to thrice a week.
That means were I to fit a point process model to this history of posts, it would show a correlation with climate change. There would be parameters inside this model which would measure this association, parameters which I could use to quantify the correlation, I mean.
Now the real question: is the changing climate causing me to write now daily posts?
Of course not!
And it’s silly to suggest that it is, even though the parameters in our model show a “significant” correlation (with time or climate change). The two things—climate and my fevered imagination—have nothing to do with one another.
Just to be perfectly crystalline transparently forcefully clear, a parameter in a point process model, which might be pegged to time or some other external thing, cannot say why the observed series changes or doesn’t change.
Here’s our authors (pp. 8297-8298):
Given that an event has occurred after the climate has changed, was it or was it not caused by climate change? This question implies that, once climate has changed, the point process of events represents the superposition of a point process of events that would have occurred in the absence of climate change and a point process of events that would not have occurred in the absence of climate change and are, therefore, attributable to climate change. Moreover, these point processes must be independent; otherwise, the former would inherit a climate change effect through the latter.
They assume a model which has a rate indexed by a parameter, and after “climate change” this parameter increases, and the increase is “attributed” to climate change.
As you can see, this whole thing, start to finish, confuses the nature of causality. What does “once climate has changed” even mean? If we knew the climate changed at some point, and how this new “climate” (and the old) caused, say, hurricanes or lightning strikes, then we don’t need a probability model. We’d just say, “There will be this many hurricanes and lightning strikes”—and we could not be wrong.
We certainly don’t need a model to tell us if a hurricane struck. We can just look. And if we don’t know the precise causes of the hurricane, it’s silly to claim it was “caused” by climate change. Some thing or things caused the hurricane before the climate changed and some thing or things will cause hurricanes after the climate changes. The probability model just can’t say anything definitive about causes.
Over the 30-yr period 1950–79, there were a total of 39 intense North Atlantic hurricanes while over the following 33-yr period, 1980–2012, there were 53 such hurricanes. If we assume that the effect of climate change over the entire 63-yr period was to increase the rate of these hurricanes, then the ML estimate of the estimated probability that a hurricane in the later period is attributable to climate change is 0.19 and an approximate 0.95 confidence interval for this probability is (-0.17, 0.44).
Again, this makes no causal sense. Add emphasis: “If we assume that the effect of climate change….was to [cause an] increase” then the estimated probability that the hurricane is “attributable”, i.e. caused by, “climate change” ought to be 100% or nothing.
Scientists spend far too much time on these vaporous models when they’d be better off searching for causes and in understanding physics.