Bishop Hill An early leaving present
Mar 15, 2015 Royal Society As Paul Nurse heads towards the exit door of the Royal Society later this year, Mike Kelly has sent him an early leaving present, a withering attack on the society's handling of the climate change issue.
...Human-sourced carbon dioxide is at best one of many factors in causing climate change, and humility in front of this complexity is the appropriate stance.
Yet the Society continues to produce a stream of reports which reveal little sign of this. The latest example is the pre-Christmas booklet A Short Guide To Climate Science. Last year also saw the joint publication with the US National Academy of Sciences (NAS) of Climate Change: Evidence And Causes, and a report called Resilience. Through these documents, the Society has lent its name to claims – such as trends towards increasing extreme weather and climate casualties – that simply do not match real-world facts.
Both the joint report with the NAS and the Short Guide answer 20 questions on temperatures, sea-level rises and ocean acidification. But a report today by the academic council of the Global Warming Policy Foundation, which includes several Society Fellows and other eminent scientists, states the Society has ‘left out’ parts of the science, so the answers to many of the questions ought to be different.
I have personal experience of this selectivity. Last year, at the request of the president, I produced a paper that urged the Society’s council to distance itself from the levels of certainty being expressed about future warming.
I said it ought at least to have a ‘plan B’ if the pause should last much longer, so calling the models into still more serious question. I got a polite brush-off.
Reader Comments (167)
Laurie Childs, Harry Passfield, et al: you are all either remarkably patient or incredibly stupid (I do suspect the former, but cannot discount the latter) in your correspondence with Onbylying. He has consistently shown himself to be an inveterate liar, as I have pointed put earlier in this thread, yet you continue to discuss points with him as if he would accept what you say with any honesty. What he fails to realise is that those who lie cannot see a truth when it is presented; he can only see any response as a lie, and seeks the motives behind it. Finding none, he will happily provide his own.
He is a troll, and is revelling in the unwarranted attention that you are giving him. Do us all a favour, and ignore him.
RR, one of the reasons that sceptics can debate and largely warmists can't is because we rehearse our arguments with people like OBA. Would we look this stuff up if we weren't trying to prove points? Next time we won't need to look it up.... probably. Sometimes we use our comments as a way to explore our own thoughts about the issues. Sometimes they're interesting.
Hi Rad Rod! You are absolutely correct about Bornbyaccident. And as I am not known for being remarkably patient.... :-)
However, I do get t'd off with commenters who come on here and flash their pseudo-credentials ("in my professional opinion", said Bornbyaccident) or seek to belittle and patronise the reputations of 12 (TWELVE, FFS!) eminent scientists in a drive-by shooting, that I like to see if there is any redeeming feature. There isn't. They're only here because this blog gives them the coverage that their own blogs can only dream of. They usually type solely with their left hand but If they had any integrity (they can't even spell that!) we would see them pop up on places like WUWT and CA, to name but two. ymmv.
Cheers.
Dang! Wrong again! I had expected you to meekly admit your folly and desist – but, no! You both provide excellent reasons as to why you should continue fattening up the goose for the slaughter. I shall check my rather long hate-list to see if you’re on it yet, and then see if I can get these wax dolls (fee-Bul, 9.95, job lot of fifty) to work. Let me know if you are getting twinges in your knees.
Oooh, yes – I, too, am a professional in my field. Am I also allowed to add “in my professional opinion” to my arguments to silence dissention by inspiring awe?
"A professional in your field", RR.....hmmmm. Well, I'm sure it's not the oldest profession. :-) I hope I escape your hate list....Cheers, Harry.
Not sure your comment deserves a response... and why not, anyway? If I have the looks, why should I not use them for personal gain? (Mind you, the first phrase of that sentence holds the key to that point...)
How are your knees, by the way? Please say you are getting unaccustomed twinges, else I will suspect these dolls may not be as good as they were sold. Perhaps special needles are needed...
Rad,
With me it's definitely the latter. I'm just too stupid to know any better. I am a sceptic after all. I usually do manage to ignore the trolls, on the whole at least, but this one makes me laugh and I just can't resist it. I'll try to keep myself in check. I don't wanna end up on that list of yours :-)
How are my knees, RR? You got me there. Oh, for sure they're older than my teeth, and they're getting a pension (early retirement!!). As for the needles...you got me there too. I'm like you, I just KBO (hint: Churchill - but you knew that, I'm sure).
Onbyaccident Mar 17, 2015 at 9:36 AM
"What is your background here?"
Nowhere as extensive as yours - just their use in modelling a drum surface, dabbling with Hilbert–Schmidt's theorem and coping with Schrödinger's equation in quantum mechanics.
Never learned any stats doing that stuff. Nor it appears did you.
Dang! Looks like I am going to have to send these voodoo dolls back (do fee-Bul do returns, I wonder?) – I can’t even hit the chuckle muscles, let alone smaller parts, like knees. Maybe I should ditch the use of sewing needles, and use knitting needles instead.
I am ambivalent about feeding trolls. From my experience on this site we have never succeeded in getting one of them to acknowledge even the base facts. (Mar 16, 2015 at 2:17 PM | Unregistered Commente rMCourtney) What I find links them all is a distinct lack of proper scientific scepticism or objectivity and their flat refusal to join in a proper scientific debate with the well qualified contributors on this site. I wonder if they truly believe what they spout.? If not then their ethics must be called into question. So on balance after a basic exchange I think we should ignore them; Banging your head against a brick wall is a futile and painful practice.
@ LC
interesting that you first ask for the quibbles of the Mann approach before you ask for my thoughts on the weakness of the M&M analysis. Let me start off wwith the latter which is more telling and the more important. Some preamble first.
PCA is a generalisation of an eigenfunction technique which in simple terms allows for efficient filtering of data. It is especially useful for potentially correlated datasets. I say efficient as it allows one to define the significant contributors to the ways that a dataset can move. Efficient also in the sense that one can use a certain amount of the PCs to reconstruct answers within the dataset space. In that sense it can be used statistically to represent potential future movements of the process but that is not its use re Mann et al. It is used as I've mentioned before in many areas (e.g yield curve reconstructions in Finance, image reconstruction etc). The correlated datasets arise from either direct observations of temperature (and lets be clear here we discuss sea surface temperatures) or proxy sets which are reasonably thought to react to climatic changes. It is possible (and this you will need to trust me on as I don't want to go through steps involving covariance matrices and the derivation of their eigenfunctions) to derive a set of functions (the principal functions and their principal values) which can be used to determine the shape of of the the tempreature record from differently sourced proxy data sets (e.g. the problem being is how do we normalise the data in such a way that the tree ring dataset implied temp changes can be meaningfully added to the ice core samples wich can be added to the real observed data etc etc. In their initial 1998 paper Mann et al had a series of such datasets. One of the points of PCA is that you would not want to use all of the PCs to reconstruct the final answer as the particular PC is transmitted weakly. There are ways to do this. Statisticians use what they call selection rules. Not being a statistician I'd prefer to use a cut off based on what mathematicians call the spectrum of the data - by this I mean that you would look at the principal values and cut off beyond the level that they would be transmitted more weakly than the noise in the system. In this exercise though they amount to the same as (in my terms) the spectrum falls off relatively rapidly (i.e. you probably could not use more than say 5/6 of them in a 20+ dataset problem - memory a bit hazy but that being the order I remember). Being lazy I don't recall how many PCs Mann used - however M&M decided to use a) less PCs based on b) a PC set they composed themselves.
It is the latter bit that Mann calls the fraud in the M&M work and I'd agree. Remember Mann was held to be correct by an NSF inquiry into the dispute. I could only get close to the way they (M&M) reconstruct the data if I selectively take out certain datasets and recompute the PCs. These datasets coincided with the ones that had some of the the greater "hockey stickness" within the data themselves. Even removing these the HS shape was still there albeit not as strong as in the Mann reconstruction. Tamino also comments on this and notes that a particular dataset was omitted simply because the data had the first 4 years of a 600 year series not available (see my previous link). I ask you now my little builder friend. Does that seem reasonable?
Secondly - M&M not only then do not use the same PC set of Mann but when they reconstruct theirs they use only a limited sample. This could be justified if the dataset was especially poor (i.e. lots of noise etc) but again the analysis of the spectrum I indicated above would show this. They don't use then the full richness of the PC datasets to reconstruct their temporal temperature histories.
I'm not asking you to replicate Mann's data. I'm quite glad you did not take up my offer of having my old spreadsheets as they are all on my old PC. However once you have the data and some canned routines to build the covariance matrix and thence compute the eigen system (i.e. principal components and their associated pricipal values) it is relatively trivial to play about with omitting datasets, recomputing PCs etc etc ... I sense that M&M did this to try and get their desired history in order to debunk Mann. Hence my contempt I spoke of in an earlier message for that work.
Maybe this is of use to you. Maybe not. You see I don't think you (and people like you) would be swayed by scientific arguments. I did the above analysis as when it appeared was a totemic attack on a graph that to the denying community was anathema. One that were it to be correct and in the absence of a natural variation (be it in the sun, be it a volcanic eruption or lack thereof) would really be a significant proof of AGW. On this blog I've noted significantly that (when not getting the ad hom attacks) people here are more attracted to the pedantry whenever a "warmist" such as I or aTTP turn up rather than debate the science. Hence the MCourt argument around my use of word "reasonable" (when the answer is there in the info he sends me or in links I send but he clearly did not read) or (I think it was HP) who didn't like my use of the words "professional opinion" and like to stick to it. I don't care if you keep my use of the term "PCA analysis" gives me away (as what - someone who forgets that the "A" in "PCA" also stands for "Analysis" .... ho hum).
In summary then the Mann et al work stands. The HS stands. Who would have thought that all of this writing stems from someone just flipping me a list of people most of whom I do not consider as knowing more than I do on the subject of climate change (and I don't say that arrogantly).
Ah yes before I forget - the "quibble". I always wondered why Mann et al needed to use PCA analysis at all here. Yes useful for different proxy set combinations but the HS shape is there in each of them to varying degrees. If it were not then it would not have ended up in the final reconstructions. However does not invalidate the work - just a bit like using many hammers to knock in the same nail.
And here I thought the issue was about centering the data over the full period and the way not doing so over-weights those series that have extremes at the end in the PC .
Go back and read it all again would be my advice. And you really don't need those old spreadsheets to understand.
I suspect you can't see the wood for the Bristlecone.
"The hockey stick stands" No, it lies, you poor fool; you've been shafted.
============
'oba'
That's an awful lot of pseudoscientific blather to cover for the fact that low order eigenvectors of PC analyses of worthless noise can only result in more worthless noise. But you could always graft an unrelated instrumental record on the end if you want to pretend you have a signal. That sometimes works.
The only person on the world who *still* has faith in the hockey stick (and who also uses this style of disinformatory obfuscation) is the SkS inmate Rob Honeycutt.
Rumbled?
Ah - hadn't seen responses here. In order
@ HAS
Nope - the centering issue was an irrelevance and would have been obvious. Actually in this instance I would (and did) use the M&M "approach" as it is just more conventional but has little if any impact on final result. However would not change PCs.
@ SNTF
err - neither Mann nor M&M used the lower order eigenfunctions so I did not criticise. No point trying to point out something wrong in my analysis when it wasn't an issue in mine nor the published research. Rumbled? Non! Is that REALLY the best you can do?
To both of you. Try doing the reconstruction yourselves. At least HAS seems to have read some of the analysis (as managed to regurgitate the blown up centering business) but SNTF...no biscuit.
oba,
Even your name is a lie.