A few comments on Mann’s interview on BBC4 in which Ross McKitrick and I were discussed.
Given the seeming efforts to backpedal on the role of the hockey stick in promoting Kyoto, it was interesting to hear the BBC reporter state that it was "hard to over-estimate how influential that study [MBH98] and that image [the hockey stick] has been".
Question 1: What do you make of the criticisms? Mann stated that the criticisms of the hockey stick graph were "spurious", "not real science", were "politically motivated" and were "ad hominem and somewhat vitriolic in nature". I have tried to avoid ad hominem comments and believe that I have done so for the most part; I would have thought that an evenhanded survey of Mann’s comments and our comments would not conclude that our comments were "ad hominem and somewhat vitriolic in nature".
Question 2: The questioner cited from the abstract of MM03, rather than MM05, and so the question was not a particularly good one. Mann said that every one of our claims had been "discredited". Given that, for example, von Storch, Zorita, Zweiers, Muller, the GRL referees, Mia Hubbert and others have endorsed our criticism of the MBH98 principal components methodology, it is certainly incorrect to state that every one of our claims has been discredited. Mann has argued at realclimate that the PC error doesn’t "matter" – a point which does not actually refute the problem (as opposed to arguing about the damages caused by the problem). We strongly disagree with the argument anyway, see comments at Errors Matter #1, Errors Matter #2, Errors Matter #3 and Was Preisendorfer’s Rule N used in MBH98?.
I have seen no response which even begins to deal with our criticisms of the incorrect benchmark for RE significance, the lack of statistical significance of the MBH98 reconstruction, the lack of robustness to the presence/absence of bristlecone pines – which are the focus of MM05, which reflects additional archival information made available in response to MM03. However, even for MM03, I recently took a look at how the specific criticisms of MM03- see Scorecard for MM03 – have borne up and I don’t see that they have been "discredited". Mann went on to say that we "got a hold of a corrupted datafile" even though the correct data was available at their site. Here Mann is re-hashing a dispute that has nothing to do with MM05 (and little to do with MM03). The "corrupted datafile" was located at Mann’s FTP site and was the dataset to which we were directed in response to our request for the FTP location of MBH98 data. When we noticed problems in this datafile, we asked Mann to specifically verify that this was the dataset used in MBH98. Rather than suggesting that this dataset was flawed, he said that he was too busy to respond to this or any other request. We were not asking for any hand-holding here, just for confirmation that this was the correct dataset. Given that this dataset was at Mann’s FTP site and we were directed to it, it seems disingenuous to say that "we got a hold of a corrupted datafile".
After publication of MM03, this dataset was deleted and another data directory suddenly materialized – see discussion here. Because we noticed problems with the dataset, we totally reconstructed the dataset and carried out fresh principal components calculations. In Scorecard, I acknowledge that 2 of 10 criticisms in MM03 may in fact apply to the now deleted dataset, but also point out that there is some evidence that Rutherford at least used the now deleted and "corrupted" file pcproxy. Mann then said that we "haven’t explained" why we "deleted proxy data". Our calculations in MM05 (and in MM03) are based on the complete proxy network. We point out that quite different results are obtained depending on whether the PC4 from a correct PC calculation is included or excluded in the regression calculations – this is a perfectly valid sensitivity analysis. We also point out that quite different results are obtained depending on whether the bristlecone pines – acknowledged to be a flawed proxy – are included or excluded in the calculations.
Mann has done calculations to show that you get high 15th century values if you don’t include U.S. tree ring data in the 15th century; this is simply because, a fortiori, the bristlecones are excluded. Given that he has claimed that their calculations are "robust" to the presence/absence of dendroclimatic indicators, I don’t see how he can have it both ways – how can it be robust to the presence/absence of dendroclimatic indicators and require the inclusion of the bristlecone pines to get a cold 15th century?
Question 3 and 4: The questioner asked about refusing to disclose workings and methods. Mann denied this. His answer here was rather at odds with his statement reported in the WSJ, where he said disclosing the code would be giving in to "intimidation", but the questioner did not notice this. Mann said that he had provided a "detailed description" of the algorithm. In July 2004, Mann provided an archive which (for the first time) was sufficient to determine which series were used in which calculation step.
There is source code for the tree ring principal components calculation at his FTP site (which was how we were able to detect the problem). However, there is no source code for the steps going from the proxy network to the NH temperature calculation. There is no digital version of the 15th century step through to the calibration and verification period. While we are able to substantially replicate the main features of his reconstruction, the description of the algorithm is insufficient to replicate MBH98 sufficiently to (for example) replicate an RE statistic of 0.51 for the 15th century step. Given that substantial inaccuracies in his description of methodology have already been identified (e.g. the inaccurate characterization of the PC method) and the broad reliance on MBH98, we see no reason why the source code should not be disclosed. In any event, it is a matter of fact that source code for steps after the PC calculations for tree ring networks have not been disclosed.
Question 5: The questioner asked about "being forced to retract" certain claims, which Mann denied. [I'll finish this later...]
John writes: the link to the interview doesn’t appear to work and I’m trying to get the website support for the "Today programme" to recognize that there’s a fault.