Perhaps like me, you watched the BBC Two Horizons program Monday night presented by Sir Paul Nurse, president of the Royal Society and Nobel Prize winning geneticist for his discovery of the genes of cell division. Perhaps like me you were particularly keen to see this program because it had been hyped that day as containing an interview with the bellicose Telegraph columnist James Delingpole where this fellow was purportedly made to look a bit stupid. I find it difficult to read Delingpole's writing. It's not that he's a particularly bad writer, it's just that he's so damn smug and full of himself (unlike me ;-). Check out his latest for a taste. That one is actually quite mild.
Delingpole has angered me with his vitriolic denunciations of peer reviewed science and stubborn denial of man-made climate change, and this has caused me to say some things about him from time to time that have been somewhat uncharitable. Perhaps this is because I always assumed he was a shill. Well I think I may owe Delingpole an apology. Having watched that Horizon program I am now convinced that he believes every word of it. And I now find that I feel a bit sorry for Delingpole, although he probably doesn't deserve it. Here's what happened. You can watch the video for yourself:
Poor James. He really believes there's some kind of mainstream science "warmist" conspiracy against the brave outliers who dare to challenge the consensus. He really believes that "climategate" is a real scandal. He fails to understand that it is a common practice in statistics to splice together two or more datasets where you know that the quality of data is patchy. In the case of "climategate", researchers found that indirect temperature measurements based on tree ring widths (the tree ring temperature proxy) is consistent with other proxy methods of recording temperature from before the start of the instrumental temperature record (around 1950) but begins to show a decline in temperature after that for reasons which are unclear. Actual temperature measurements however show the opposite. The researcher at the head of the climategate affair, Phil Jones, created a graph of the temperature record to include on the cover of a report for policy makers and journalists. For this graph he simply spliced together the tree ring proxy data up until 1950 with the recorded data after that using statistical techniques to bring them into agreement. What made this seem particularly dodgy was an email intercepted by a hacker in which Jones referred to this practice as a "Mike's Nature trick", referring to a paper published by his colleague
Poor James. I'm struck by one thought as I watch that video, and it's this. He could have recovered. When Nurse asked Delingpole the very straightforward question of whether he would be willing to trust a scientific consensus if he required treatment for cancer, he could have said "Gee, that's an interesting question. Let me think about that and why it's different." If he had an ounce of humility he could have saved himself. Instead, he became defensive and lost his focus. Eventually he would make such regrettable statements as this one: "It is not my job to sit down and read peer-reviewed papers because I simply haven’t got the time, I haven’t got the scientific expertise… I am an interpreter of interpretation." Oh dear. Would he trust a scientific consensus if his life depended on it? He evidently wasn't prepared for that question, which does seem a bit strange. I'll bet he's thought of at least a dozen answers to it now. In case he needs one more, here it is. In a parallel universe where James Delingpole is not the "penis" that Ben Goldacre describes him to be, he might have said the following:
Gee, that's an interesting question. Let me think about why it's different. (Thinks) Well, it seems to me that when evaluating a scientifically agreed treatment for a disease such as cancer, we have not only all the theory to peruse and the randomized and blinded trials, but also thousands if not millions of case studies where people have undergone the intervention. We have enough data to estimate a person's chances of recovery and know that on average they will do better. When discussing climate change, we really only have the one case study. Just the one earth. And it's a patient that has not undergone any intervention. The scientific consensus is therfore entirely theoretical and intangible. This makes it more difficult for the lay person such as myself to trust it.It seems to me that this would be a fair, if somewhat misguided, answer to the question. And that brings me to the final point, upon which Sir Paul ended the program saying "Scientists have got to get out there… if we do not do that it will be filled by others who don’t understand the science, and who may be driven by politics and ideology."
More on that here: http://www.leftfootforward.org/2011/01/james-delingpole-climate-conspiracy-theorist/
Re the Nature trick: if proxy tracks instrumental from 1850 to 1960 but then diverges for unknown reasons, how do we know that the proxy is valid for reconstructing temperatures in periods prior to 1850?
ReplyDeleteThis is a good question and one I'm not sure I can answer it to anyone's satisfaction. We seem to have good agreement among several forms of temperature proxy going back centuries and with direct measurements back to 1880. There is divergence in more recent years and there are several theories as to why that might be. Some possible explanations here:
ReplyDeletehttp://www.skepticalscience.com/Tree-ring-proxies-divergence-problem.htm
In the physical world we can never be absolutely certain of anything. Rene Des Cartes showed it was impossible to prove that everything he sensed wasn't manipulated by some invisible demon. It is necessary to first make certain assumptions about the universe that we observe. After that, we can only go with the best theories available that allow us to make scientific progress.
This is why the divergence should have been shown to the policymakers and why what Jones did was wrong. By deleting the divergence and replacing it with instrumental data, he makes it look as though the proxies always track instrumental. He has hidden the uncertainty from the politicians.
ReplyDeleteIt seems a bit disingenuous to try and overstate the importance of the WMO document - the only case of splicing? - when they also wrote a paper in Nature (in 1998) about the divergence issue:
ReplyDeleteReduced sensitivity of recent tree-growth to temperature at high northern latitudes
If there was a conspiracy to "hide the decline" then they didn't do it very well.
Andy
ReplyDeleteA couple of things there. Firstly, the decline was hidden in AR3 and AR4 as well (although the authors were persuaded to add text about the deletion of the decline for AR4). It's not the splicing that's the problem so much as the hiding - you are right that WMO was the only time they spliced the instrumental data in, but it's the deletion of the decline which is the problem. (I think I'm right in saying that they padded the endpoint with the instrumental data in AR4 though.)
It is undisputed that the decline was discussed in Briffa's papers. The allegation is that Jones (in WMO), Briffa (in AR4, partially at least) and Mann (in AR3) hid uncertainty from policymakers. An attempt has been made to defend this by pointing, as you do, to discussion of the divergence problem in the primary literature. This is not a defence against the allegation made. The reader of the report should surely know about the divergence? Otherwise they are misled about the uncertainties.
There's another aspect too. If a particularly dedicated reader of these reports followed the citation to Briffa's original paper they would find a series with a decline, not a rising trend at the end. They would learn that there was a divergence problem and then they would scratch their heads and wonder why the series they were looking at in the report was heading up, while the cited paper said it was going down. There would be no explanation of the deletion of the divergence and its replacement with instrumental data. To my mind this is not supportable.
This is the point - the IPCC reports (and the WMO one) are not an accurate reflection of the primary literature.
Although Jones used the phrase "hide the decline" and the word "trick" in his email, I'm not convinced that this was done with the intention of causing a deception. That charge needs to be proven. For my part I am satisfied with the explanations given. What does constitute dishonesty on the other hand is the illegal accessing of unauthorized files on a server. The contents of those files were then manipulated to paint an incomplete and misleading picture.
ReplyDeleteI disagree: the IPCC reports are a good reflection of the primary literature.
ReplyDeleteThey assess the importance of problems (e.g. the divergence problem) and, in this case, present the best estimate of NH temperature over the last however-many years.
Sure, there are layers of complications but I don't think anyone reads these things thinking otherwise. Apart from Dellingpole, that is, who simply doesn’t have the time sit down and read peer-reviewed papers. ;)
Andy
ReplyDeleteThe divergence problem was not assessed at all in AR3 - it was deleted and not mentioned. How can that be a good reflection of the primary literature?
From the TAR:
ReplyDelete"There is evidence, for example, that high latitude tree-ring density variations have changed in their response to temperature in recent decades, associated with possible non-climatic factors (Briffa et al., 1998a)."
...so I think it's wrong to say it wasn't mentioned. It's not a deep analysis though, I'll give you that.
But it's also 10 years old and superseded by the AR4, which does a much better job.
But still deletes the decline! I'm sorry but deleting data that goes against your preferred hypothesis is not right.
ReplyDeleteWould you do it?
Bishop Hill seems to have a problem with deliberately excluding a portion of data which is known to be unreliable (because of direct measurement) as the reasons for this unreliability are not clear. What do we say to that?
ReplyDeleteFlay
ReplyDeleteYes, that's sort of it. The problem is more with the bit that's left in though. Do we know it's reliable? The answer is no we don't.
Bishop Hill: "The problem is more with the bit that's left in though. Do we know it's reliable? The answer is no we don't."
ReplyDeleteThis is the reason sceptics say the Hockey Stick is broken. McIntyre and McKitrick's work should have led to climate scientists accepting this simple point and moving on.
The fact that climate science refused to do this - with Dr. Kevin Trenberth still using the word 'deniers' of people like M&M at the American Meteorological Society this week, having refused to answer Steve McIntyre's polite questions about where he had gone wrong back in 2005 - has led to a collapse of confidence in everything else they are saying. This must be put right.
Flay: "The contents of those files were then manipulated to paint an incomplete and misleading picture"
ReplyDeleteJK: The 30 second solution is for Jones to order the release of ALL CRU emails.
End of "incomplete".
End of alleged "cherry picking."
End of "out of context."
Why doesn’t he - they are publically owned documents produced by public employees while on public time on publically owned email addresses and servers.
Thanks
JK
Here's what I have to say about that. A hacker broke into a server and stole some files (supposedly found by doing a grep). It seems a shame that he/she did not steal the entire email mailbox as this would have solved that problem. Perhaps he/she did after all. The picture painted was incomplete because only the "damning" evidence was presented. Even from the evidence presented we get reasonable explanations from those involved.
ReplyDeleteThe research is publicly funded, but the public do not have an automatic right to see all communications. The Freedom of Information Act has limits. Try asking for all police correspondence and see where you get. I don't see why Phil Jones should be expected to hand over more information in order to add context to some that was stolen. Most of us are satisfied. Additional information is unlikely to satisfy the loud little handful though. At worst there is scope for further misrepresentation.
Flay, you say
ReplyDelete"Bishop Hill seems to have a problem with deliberately excluding a portion of data which is known to be unreliable (because of direct measurement) as the reasons for this unreliability are not clear."
Let's just be clear that the issue is not one of data quality, what is unreliable is the relationship between tree rings (measured as well as they ever have been)and temperature.
Bishop Hill seems to agree with me. Lack of reliability of modern proxy data casts doubt on overall reliability as the reasons for the divergence are unclear. If the data continued to track direct measurements then one could regard non-climactic signal components as a constant.
ReplyDeleteI believe that all the (3000, IIRC) emails Briffa took home "for safekeeping" are subject to an FOI request, so we should get some more detail at some point. I must say though, that it is a bit unlikely that these will mitigate things for CRU because they would surely have released any that demonstrated that their accusers were mistaken (as indeed Briffa did on one occasion).
ReplyDeleteJust to be clear, I meant non-climatic not non-climactic. I doubt any of this is particularly climactic. ;-)
ReplyDeleteFlay @06:48
ReplyDeletethe plummeting public confidence in the police in general being nothing to do with their lack of forthrightness, transparency and accountability then? The latest agent provocateur bonking under-the-covers eco-bobbies trundling around in Volvo XC90s (good bed in the back with the seats down) revelations having nothing to do with climate change...? Given that some of the eco-activist groups receive direct government funding? Funny old world.
On a wider note, the uninformed or simply stupid shrill partisanship in the climate debate is out of hand, the politicised alarmism provoked folk to look a little closer at what was being claimed, and it is very clear that overstating the case without supporting information / observation / exposed analysis hasn't been a success - one doesn't require a PhD to grasp this simple thing.
You lambast Delingpole but it's truly well past time that those in the warm camp had a clearout of the SWP agitators and "care in the community" cases that set up shop on the left (warm)side of the "debate"
And actually, left=warm , right=denier how deeply dunb (and anti science / knowledge in principle)is that?
The media are key in all this, and the BBC's contribution does not cover it in glory or even engender any respect what-so-ever. The Horizon program hasn't changed that.
The idea that consensus over-rides observation ? hmmm...
Gordon the Fence Post Tortoise. Okay, whatever. Yawn. I'll tell you what. Conspiracy theories are thrilling, really. But you can't prove them so it's no good spouting them. Delingpole annoys me because he doesn't care about the science. He really doesn't. He cares about the conspiracy. I'm happy for someone to change my mind. In fact I'm rather liking a contrary theory from Henric Svensmark right now. I'll bet you've never heard of him. Moreover, I'll bet you really couldn't give a shit. Thanks for the comment. Have a nice day.
ReplyDeleteBH
ReplyDeleteWould you still be argueing that the data that don't show a climatic signal should be included if they were wrong in the positive direction? I doubt it.
It's clearly not a perfect situation. The correlation was good for 100 years and then some of the proxies diverge, probably because of anthropogenic influences. It's reasonable to assume that those influences didn't exist before 1850 so I think the method is ok.
But in short, I don't think there's a big issue here, it's only become important because of Phil Jones' "decline" quote.
Andy
ReplyDeleteWell I hope I would. In my book I think I was scrupulous in not defending Soon and Baliunas's use of proxies that were responding to precipitation even though it would have helped the sceptic "cause" to do so (I know there's a case that more heat leads to more rainfall, but it's a remove too far in my book).
The latest from the literature on the subject (D'Arrigo et al 2008) is that the evidence that the decline is limited to the 20th Century is "limited". It is not therefore reasonable to assume that deleting the divergence is justified. It is misleading. If this was, say a medical trial or a listing particulars for a big company, people would be going to jail.
The divergence problem became important when Mann deleted it from the graph in 3AR.
I'm surprised that you're so interested in presenting data that are known to be wrong! If it's well established that they're wrong and say that you're not showing them then that sounds reasonable. Maybe that hasn't been done everywhere but that doesn't mean that the underlying science is bad.
ReplyDeleteI'm not sure I agree with your interpretation of D'Arrigo et al. either, here's a longer quote:
"Although limited evidence suggests that the divergence may be anthropogenic in nature and restricted to the recent decades of the 20th century, more research is needed to confirm these observations."
That sounds quite similar to what I said before. Although I hope they're not suggesting that "unlimited" evidence is required!
They are not "wrong" per se. They diverge for an unknown reason. This may or may not have affected earlier periods. This is a fundamental uncertainty in the data and to present the data without its fundamental uncertainties in place is misleading.
ReplyDeleteAs I said before, if this were a medical trial or an IPO, people would have been jailed. Seriously. So here's a question. Why would it be OK for climatologists to remove uncertainties from view but not medical researchers? (Or are you saying that it doesn't remove the uncertainties from view?)
I'm honestly trying to understand your position here. I just cannot see how anyone can defend this. As Jones' colleague Paul Dennis noted on the Simon Singh thread, it's indefensible.
Remember that this was just a graph on the cover of a report. As I understand it, the body of the report showed the divergence. Is this not the case? I think several people made errors of judgement and then were not forthright for one reason or another. Perhaps that one graph made all the difference. Who knows?
ReplyDeleteI think you're being overdramatic with the comparison with a medical trial. It doesn't stand up to any scrutiny at all. But let's run with it...
ReplyDeleteThe only analogous situation I can think of is if you start a drug trial with one dose and then two thirds of the way through the trial you double the dose (or something similar). I assume that this doesn't happen but I know nothing about drug trials.
When you look at the results you see a good response in the first period but that changes after the dose changes. What do you conclude from this? Mostly that you're not very good at designing drug trials. But also that something was looking good in the first period.
If you publish these results I suppose you'd talk most about the first period or just do another trial to clarify the result, assuming anyone would give you ethical clearance after making a mess of the first trial.
Obviously, with studies of past climate there's no option of going back and controlling the forcing conditions so the best option is to try and understand the response change.
So we're stuck with a similar situation - good results for the first two thirds (1850-1960) and then divergence for the final third, which doesn't tell us much useful.
But we know, as I think we established from D'Arrigo et al. (2008), there is some evidence to suggest that the divergence is anthropogenic in nature and restricted to the recent decades of the 20th century.
So what's my view of this? Well, I don't think that the divergence problem was explained well enough in the IPCC TAR or in the WMO report. It seems that they had the choice between laboriously describing and justifying the method in documents aimed at non-scientists (which had already been done in the literature) or just showing the period that was known to be uncontaminated with non-climatic factors. They went with the latter.
However, I don't think that this was done to deceive; I think it was done to present what was at the time the best estimate of NH temperature for the last 1000 years. This was largely the conclusion that was drawn from the inquires into the contents of the CRU emails as well.
Andy
ReplyDeleteThe emails show us otherwise. In the weeks before the trick was prepared, Briffa talks about being under pressure to present a "nice tidy story of unprecedented warmth" (the quote is from memory but I think those are the exact words). It was done to avoid giving sceptics anything to get their teeth into.
I think I've shown pretty conclusively that the inquiries were not fair or honest (see my GWPF report). I don't think that anyone can reasonably suggest that they were.
Another thing though - I'm still not clear if you are recognising that the uncertainty was hidden the way they did it. What's your position here?
ReplyDeleteI don't think that the uncertainty was "hidden". I think it was decided that it wasn't the most important part of that work - the reconstruction that is perfectly justified using the uncontaminated data is.
ReplyDeleteThey could've just bound together all the papers that they cite, which would've covered all the uncertainty and why it doesn't undermine the method, but that would've defeated the point of the process.
Ask yourself this: if you were to find that you had invested in a company whose directors had decided that some uncertainty in their business plan was "not the most important" and, as a result, had decided not to tell you about it, would that be acceptable to you? If you then lost all your money because the uncertainty turned into reality, would you shrug your shoulders and say to yourself "Oh well, I guess they didn't think it was important at the time"?
ReplyDeleteI admire your fortitude in defending what happened, but really Andy, stop digging.
This quote about liberties taken with climate science in a movie is pertinent:
ReplyDelete"In a quick survey of the Fall [AGU] Meeting audience, Perkowitz estimated that about half were offended by the liberties [The Day After Tomorrow] took and half indicated that it is okay to stretch the truth in a movie to raise public awareness about an issue."
An inconvenient truth is emerging from ‘climate science’. Key physics in the climate models is probably wrong and has been so since the 1950s. So, the IPCC’s predictions of high-feedback CO2-AGW are probably baseless and it seems this was known by insiders by about 2003 yet AR4 went ahead with the incorrect science.
ReplyDeleteTo explain it you go back in History. Carl Sagan developed aerosol optical physics to explain Venusian clouds. It’s the relationship between ‘albedo’ and ‘optical depth’. It apparently fits real data and Sagan noticed it predicted that if you polluted clouds, the reduction of droplet size increased ‘albedo’ thus cooling the Earth.
This led him to conclude CO2-AGW was being hidden by man-made pollution shielding the Earth. His ex-students, Lacis and Hansen, introduced the physics to NASA. It’s why the modellers assumed high feedback CO2-AGW.
Data are shown in Figure 2.4 of AR4. Predicted median heating due to AGW is c. 1.6 W/m^2 but you only measure 0.4 W/m^2. ‘Cloud albedo effect’ cooling, 1.75 times as large as that, makes up most of the difference. But there’s a problem: by 2004, there was no experimental evidence of it. Apparently to keep it in AR4, NASA claimed 'enhanced surface reflection' from polluted clouds: there’s no such physics.
When I realized AR4 included fake science, I set out to identify where Sagan went wrong. Look at clouds about to rain, larger droplets, they get dark underneath so have higher albedo [it blinds glider pilots], the reverse of what the models predict. The answer is a second optical process, direct backscattering at the upper cloud surface.
Because this process is switched off by pollution, it’s another form of AGW [which better explains palaeo-climate data]. So, there's no absolute proof of any net CO2-AGW qhich for all we know could controlled by a process involving reduction of water vapour concentration.
This so-called settled science is falling apart.
Could you please tell me why my Twitter notification extension sent me a link to this arrogant smug alarmist asshole who attacks James Delingpole by the usual lies? I don't want to be getting this garbage. Whoever it was, I am going to find you and eliminate you from my list of contacts.
ReplyDeleteLuboš Motl, as you can see we had a fairly sensible discussion here in the comments until you came along. You will find that I am none of the things that you call me. In fact I agree that the science is inconclusive. James Delingpole doesn't give a shit about the science. All he cares about is "conspiracy". All the same, I thank you for your extremely constructive and illuminating comment. I will leave it up to show that I have a sense of humour.
ReplyDelete