A while ago I discussed a Nature Climate Change article by Greg Hollin and Warren Pearce called Tension between scientific certainty and meaning complicates communication of IPCC reports. A number of us submitted a response, which appeared today together with a reply from Pearce & Hollin.
What’s maybe interesting about our response is that it includes writers, bloggers, physical scientists, and social scientists; not all of whom fit nicely into a single category. The main thing we were trying to point out was that Hollin & Pearce appear to have misunderstood what was presented in the IPCC press conference. Their claim seemed to be that the IPCC fell into a certainty trap because they were using the “hottest decade” as an indicator of our certainty about anthropogenic global warming (AGW), while dismissing the so-called “pause”, which was also a decadal indicator. In other words, they were using one decadal indicator to stress our certainty about AGW, while dismissing another.
However, these two indicators are not equivalent. The “hottest decade” refers to the 2000s being the hottest decade in a record going back to 1850. Also, on more than one occasion, they explicitly pointed out that each of the preceding three decades has been warmer, at the Earth’s surface, than any preceding decade since 1850. So, the “hottest decade” was referring to a long-term warming trend, not something short-term. The so-called “pause”, on the other hand, refers to the surface – over the last decade or so – warming more slowly than we might have expected. Although it does appear that the surface has indeed warmed more slowly in the last decade or so than was expected, such decadal variability is expected and does not somehow challenge the fundamentals of AGW.
Furthermore, the claim that the IPCC press conference dismissed the so-called “pause” is slightly odd given that there were numerous questions about the “pause” and numerous lengthy responses. That the responses were not “yes, the ‘pause’ is extremely important and really challenges our understanding of AGW”, or some variant of that, does not mean that it was dismissed. One of the responses even explicitly said
let’s be careful not to interpret that in terms of trend, but very interesting work is being done by the assessment to try to understand the elements of these variability and to do a bit of attribution, which is a tricky issue.
So, not only was our comment suggesting that the IPCC did not fall into a “certainty trap”, it was also pointing out that there was little evidence to suggest that the so-called “pause” was dismissed.
Warren Pearce and Greg Hollin have just written a blog post about their experience in writing their article and the response to our comment. What they seem to be suggesting is that part of the issue is that what they were doing was inductive, while the physical sciences is more commonly deductive. The idea being that in the physical sciences you typically use data to test a theory/hypothesis, rather than starting with the data and trying to see if it suggests anything interesting.
Personally, I think this is a rather simplistic interpretation of what happens in reality. In some circumstances scientists do collect data so as to test a theory/hypothesis. However, in many cases the data shows something interesting, and scientists then try to understand what the data is suggesting. The subtlety, though, is that you’re typically constrained by existing theories, or the known laws of physics. Often you don’t construct some kind of new basic theory so as to explain the data, you simply apply the known basic theories. Of course, you might construct a new model so as to explain the data/observations, but that’s not quite the same as developing some kind of brand new theory. In fact, as this article suggests, it’s often circular; it’s sometimes hard to say specifically if theory is driving observations, or if observations are driving theory.
I’m also not quite sure why it makes any difference whether one’s approach is inductive or deductive. What you present should be a reasonable representation of reality, whether you approached it inductively (“the data looks interesting, why is that?”) or deductively (“I have a theory/hypothesis, let me collect some data to test it”). For example, either the IPCC fell into a trap by using one indicator to stress the certainty of AGW while dismissing another essentially equivalent indicator, or they didn’t; either the IPCC dismissed the so-called “pause”, or they didn’t. It can’t really be both.
Of course, those of us who wrote the comment in response to the original Hollin & Pearce paper think that the IPCC neither fell into a certainty trap, nor dismissed the so-called “pause”. In some sense this seems like a pretty simple thing that we should be able to establish and agree on. It seems, however, that this is unlikely to happen anytime soon. I find this rather unfortunate.