Evaluation of the Hughes emails
David W. Schnare, Esq. Ph.D.
This is a lengthy article and covers several related topics. As previously reported, we have now received the Malcolm Hughes emails discussing issues related to the 1,000-year temperature reconstruction presented in the Mann, Bradley, Hughes 1998 & 1999 papers (MBH98 & MBH99). The collection contains 7,511 pages of emails and attachments and other records, of which 93% (6,999 pages) were withheld and subsequently ordered to be released. The 512 pages that were voluntarily released are in a file entitled:
can be downloaded from there and are paginated as “ABOR/MH/Non-Priv-001” through “ABOR/MH/Non-Priv-00512”.
The previously withheld documents are in a file entitled:
Malcolm Hughes Withheld Documents
can be downloaded from there and are paginated beginning with “ABOR/MH/Priv-000037” and ending with “ABORF/MH/Priv-007274”, but are not fully sequentially ordered.
The University was permitted to continue to withhold 275 pages. Those pages involve student records, ongoing research, and peer review evaluations. My comments below cite to the “Priv” pages and can be quickly found using the Adobe search function on the page numbers.
An Introductory Note
I do not know and have never spoken with Professor Malcolm Hughes. I once saw him in a courtroom but we were never introduced. I wish we had met. I think we would have found much in common and might have forged a friendship, being of the same generation and formal academic background. Based on a review of Dr. Hughes public records alone, I find he is a thoughtful man and genuinely honest as regarding his science. He is not the instigator of emotional arguments and in the 7,511 pages of his public records, I found not a single mean-spirited statement from him. Nor did I see him encouraging attempts to remove editors or even criticize them. Rather, he tends to be the adult in the room, entering the fray when Michael Mann attempts to push beyond the bounds of proper academic behavior (as established in the American Association of University Professors’ Ethical Guidelines), and otherwise leaving to Mann and others the braggadocio and crankiness that resulted in a handful of unhelpful criticisms and the resultant mea culpa’s necessary to maintain some level of working relationship amongst this small group of co-authoring academics.
Topics covered below include Dr. Hughes professionalism, how to protect confidentiality when needed, the failure of many to keep research logs, inconsistencies between the Hughes collection and the Mann collection recently made public, civility, the MBH versus the McIntyre & McKitrick papers, and a never-before seen (and failed) effort by Mann to explain away the divergence of tree ring data from actual temperatures. Because WUWT is best when presenting and discussing science, I begin with the divergence and borehole issues.
Divergence and the Need to Hide the Decline
ABOR/MH/Priv-005599 & ABOR/MH/Priv-005613 & ABOR/MH/Priv-005620
The “hide the decline” divergence issue is directly addressed here in a paper that was prepared for that purpose. It claims the MBH reconstruction does “predict” the 1990s temperatures. However, Hughes and Bradley were against submission of the note to Nature and it was never published.
Notably, this does not show the actual tree ring proxy data from after 1980 that Briffa and Osborne published that shows the divergence. The scientific question that remains 20 years after MBH98 is as to why the tree ring data diverges and what this means with regard to the pre-1900 reconstruction of MBH98 & MBH99.
ABOR/MH/Priv-005618 & 005624
Mann argues that, taking Bradley’s concerns into account, his reconstruction methods still work. Actually, he graphs this and it shows the adjustments do not work well. Of course, he doesn’t show uncertainty on the graph and provides no analysis of statistical association between recorded temperature and the proxy reconstruction.
The chart below is from email ABOR/MH/Priv-005618. Note, the dotted line (Mann’s tree-ring temperature reconstruction) is well below the actual temperature record after 1980 and does not show the clear divergence known to exist after 1992, something it seems he was unwilling to show for that reason.
Notably, Mann admits only the high elevation bristlecone pine chronologies do “a good job in calibrating/cross-validating against the instrumental record . . .” Hence, Bradley’s comment that the note and its failure of the 1990’s reconstruction to track the actual temperature “well enough” condemns an effort to send a note to Nature as not worthwhile.
Hughes agrees with Bradley and is against publishing. See ABOR/MH/Priv-005632. This is a rare instance when Hughes chooses to hide data that would have moved the discussion forward within the scientific community. It would have encouraged others to take a harder look at the utility of proxy data and spur new work on a high-quality proxy data set that comes out nearer to present. He stated:
That this new version of your post-1980 calculations should be so sensitive to the omission of a single record is very worrying indeed. It should also be noted that nothing much happens in the ‘new’ reconstruction until the last three years. I fear this would give a wonderful opportunity to those who would discredit the approach we used in MBH 1998 and 1999. They would almost certainly seize it to attack the use of the MBH99 reconstruction in the IPCC. On reflection, I think it would be much wiser for us to keep our powder dry, and if challenged in a creditable forum point out that we are working on assembling a dense and high quality proxy data set that comes out nearer to present.
Mann’s response indicates why he agrees to not publish, “While I actually think it does disprove the assertion they often make, I’m inclined now to agree that there are enough weak points that it might just open up other holes for them to attack us on.” At ABOR/MH/Priv-005633 This is different from “negative findings” being unpublishable. It is a clear desire to hide negative results that would have otherwise moved the scientific discussion forward.
The Borehole Debate
Borehole data impeaches much of the tree-ring data used in MBH and a paper that challenged MBH98 also opens the door to what appears to be selection bias in the borehole data MBH used. I don’t follow this work closely so I don’t know if this is new information to WUWT readers or is a rehash of an old discussion. Nevertheless, it shows the extent to which Mann goes to try to protect MBH98.
Esper and Mann disagreed on how and when to use borehole data. When a third party enters the discussion, providing data to support Esper over Mann, Mann’s reaction is to dun the Science Letters Editor. See ABOR/MH/Priv-005543. Yet, this is exactly how scientific discussions are supposed to be handled – through subsequent publication of confirming or impeaching observations. Mann claims the Pollack contribution has been “soundly discredited” in a Mann paper that Science rejected and was subsequently to be published in the “technical” literature. However, Science and J. Geophysical Research have two different audiences; Science has a responsibility to support a discussion of a paper it had already published and Science was giving Mann the opportunity to engage in that discussion by preparing a response letter to be published with the Pollack letter. Mann simply did not want the Pollack data to be seen and he succeeded. See ABOR/MH/Priv-005571
ABOR/MH/Priv-005542. Again, the Mann versions of the emails fail to include the related and relevant graphics.
Showing how different Mann, Esper and Pollack reconstructions are.
In another example (at ABOR/MH/Priv-000082), Tom Crowley argues “if we cannot make a case to our colleagues, why muddy the waters further”. It is one thing to recognize that you cannot demonstrate new knowledge. It is entirely another to begin from a presumption of knowledge and realize you do not have the observations necessary to convince others. The correct approach is to admit ignorance and highlight to your colleagues that this is an avenue worth of pursuit. Mann seems not to recognize this as how scientists should behave.
Bradley arguing that borehole data that does not support their approach should not be used. The manuscript should indicate that kind of decision. It calls into question all use of borehole data, one would think.
This is a good example of the (potential? Actual?) biases entered into their paper. They simply did not include anything that altered their prior planned outcome and did not explain why.
When Nature did not accept the attack on Pollack, they sent it to Science. There is harsh criticism of both the Nature editor and the Pollack paper, neither appropriate, and Hughes calls Mann out on that. See, ABOR/MH/Priv-005813. Bradley calls him out as well, wanting the following taken out: “but are based on what many in the paleoclimate research community feel are deep flaws in methodology and data quality.” Obviously, Mann does not “speak” for the entire community and Bradley believes Mann is not in step with the “community” in that regard.
Hughes demonstrate the position of a responsible scientist, writing “We should avoid all appearance (and reality) of pressuring any journal or organization.” This is a very strong indication of his integrity and commitment to good science and the proper means to engage the scientific community. But, Mann isn’t having any of it. See, ABOR/MH/Priv-005944. Bradly and Hughes cave on this. See, ABOR/MH/Priv-005950
A good example of Dr. Hughes’ performance of good scholarship is found at ABOR/MH/Priv-000309 thru ABOR/MH/Priv-000315 (read from the bottom up). Therein he explained to his colleagues that for the paper on which this team was working, he was shifting the tree-ring part of the dataset they would use, explaining why and opting for the dataset containing only those that meet chosen a priori standards, apparently rather than choosing a dataset that would produce a desired outcome. Normally, this effort would not be a cause for meritorious celebration, but amongst this coterie, it was an unusual success for good science and Dr. Hughes should be recognized for that result.
In another example of laudable scholarship, Dr. Hughes realized Mann had not properly archived the data used in the MBH98 paper. In November, 2003, five years after publication of MBH98, Hughes makes it clear that the team needs to do what it should have done in 1998, make available a “well documented and thoroughly checked file containing the data we used in MBH98.” See, ABOR/MH/Priv-001385. As Dr. Hughes explains, “I think it is important that we dispose of the data problems completely, and, if any errors crept in during the original collation of the dataset, we document them. I fully understand and agree with your approach of not being drawn into a guerilla campaign with these guys. They are not my main concern. I am concerned, however, about how our scientific colleagues view our work in the future, and I believe we need to do this for this reason. This affects all three of us [Mann, Bradley and Hughes].”
During the litigation, when accused of asking for these public records for the exclusive purpose of harassing these academics and seeking to cherry-pick emails intended to embarrass them, I responded that we had no idea what was in the emails, but that their release would allow the public to see how they engaged in their scientific pursuits. Further, the emails of an honest scientist would exonerate them from any accusations of misbehavior. I believe Dr. Hughes’ emails do just that, and I commend him to you as an example of a good scientist. I leave to others evaluation of Hughes’ correspondents.
As I explained to the Court, release of these emails would do more than open to public view the processes by which academics produce important scientific papers. It would also explore core behaviors on which there has been argument amongst Constitutional lawyers regarding academic freedom and the need for a scholar’s privilege. I briefly address these below, citing to emails for that purpose.
Ability to have confidential discussions
The University argued that emails should not be released as they constitute the means by which confidential discussions took place, and their release would chill such discussions. Our expert testimony dispelled that argument, but these emails do so as well. Notably, Dr. Hughes resorts to the telephone when he finds a need to address interpersonally sensitive topics (see, e.g., ABOR/MH/Priv-000202). In his decision in the legal matter that resulted in release of these emails, Judge Marner specifically noted that where the need for confidentiality in communications arose, the telephone was available, indicating that emails are neither a complete nor necessary replacement for voice-to-voice communications, and thus release of the emails would not harm the ability of academics to cooperate or necessarily chill the research effort.
Lack of a Research Log
Scientists are taught to keep research logs that chronical their work and allow for two things, duplication of the work and full recollection of what is done. In 2011, I asked the University of Virginia for documents from 15 of their faculty scientists that evidence the keeping of a research log associated with each of those scientists’ most recent peer-reviewed papers. Not one of them kept any discernable research log. As a result, no one could duplicate their work and they had no basis by which to make authoritative statements as to what they did. Mike Mann was one of those 15.
The Hughes documents shows the problem with this kind of misbehavior. Apparently the MBH99 paper stated that they used 28 chronologies for the western U.S. Keith Briffa and Jan Esper, who were preparing their own reconstruction, wanted to know which were used. Dr. Hughes was able to identify only 27 that were used (see ABOR/MH/Priv-000292). Mann, without a research log, had to rely on Scott Rutherford (then at Roger Williams University, but one of Mann’s Post-Doctoral research assistants) to reconstruct what they had done, thus independently confirming Dr. Hughes research notes or files. While Mann kept no log, Dr. Hughes appears to have done so, or at least the practical equivalent.
Inconsistencies between the Hughes and Mann email collections
In an effort to blunt our release of the Hughes emails, Michael Mann released the emails he claimed were withheld in the Virginia litigation. Unlike the Hughes collection, the Mann collection does not include attachments. As mentioned below, some of these attachments are highly valuable historical presentations of how they achieved their results. One example of note is the repeated efforts by Mann to impeach the borehole data that significantly undermines the MBH99 temperature reconstruction. See, ABOR/MH/Priv-005542. In another, at ABOR/MH/Priv-005618, Mann does his best to revive the legitimacy of his hockey stick reconstruction by attempting to unhide the decline – that is, to find proxy data that did not suffer from the divergence between post-1980 the proxy-based temperature reconstruction and measured temperature. Mann’s repeated efforts to salvage something out of his approach was eventually cashiered by both Professor Hughes and Professor Bradley. The Mann collection does not include all this material. The Hughes collection does.
The Mann collection is also missing interesting emails that are in the Hughes collection. At ABOR/MH/Priv-000425, and nearby, is a continuation of the discussion on which tree-ring datasets to use. This is an October 2002 discussion that appears to be the underpinning of a 2002 AGU Fall Meeting presentation entitled “Proxy-Based Reconstruction of Surface Temperature Variations in Past Centuries”, where the U.S. team and the U.K. team try to find common ground on northern hemisphere hot and cold periods. This is missing from the Mann collection, as is a discussion on how Mann and Crowley disagree regarding a recalibration of the tree ring data used in MBH98 and MBH99 that undermines the hockey stick. See, ABOR/MH/Priv-000523.
ABOR/MH/Priv-006407. This is an example of an improper redaction (“Henry’s student” was redacted.). The Mann email release does not contain any redactions and should have (e.g., phone numbers and private email addresses).
In late 2000, efforts to replicate the MBH98 & MBH99 temperature reconstructions had already begun to invalidate the hockey stick output. Of particular concern by Mann was the efforts by Tom Crowley, Phil Jones and Keith Briffa. Dr. Hughes describes the contretemps as rising from a “mental block accepting the basic idea of your [Mann’s] methods.” In contrast, Mann describes this as
“Phil and Keith (and sadly, Tom) simply just don’t seem know what the hell they’re doing here or, if they do, they are being intentionally deceptive.” See, ABOR/MH/Priv-000523 – ABOR/MH/Priv-000529.
Michael Mann has demonstrated a lack of civility in the past and that behavior is on full display in the Hughes collection. In comparison, Dr. Hughes does not engage in such behavior.
The Esper/Mann disagreement was not completely civil. Mann directly accuses a Science Journal editor of “bad judgment” for allowing publication of a paper with which Mann disagrees. The appropriate response is a letter in science pointing out weaknesses in the paper, which Mann and Trenberth attempt to get published (independently). Trenberth’s letter was not published. Mann’s was.
MBH versus MM
For those of you interested in the interactions among the MBH threesome as regards the critique of their work by McIntyre & McKitrick, you may wish to look at ABOR/MH/Priv-001717. Clearly, the MBH team was in need of a competent statistician and probably wished they had one to work with them in 1998. In addition, the emails contain the unpublished “REPLY TO ‘Global-scale temperature patterns and climate forcings over the past six centuries: A comment.’ By S. McIntyre and R. McKitrick”. Nature published the M&M Comment but chose not to publish the MBH reply. The M&M Comment is available at: http://www.uoguelph.ca/~rmckitri/research/fallupdate04/MM.resub.pdf
An interesting side note that I have not seen mentioned previously was the response of Nature to the M&M critique. Nature wanted a correction to MBH98. “Having carefully gone through your responses and after discussing the matter with my colleagues, we feel that we will need you to publish a correction to your initial paper (MBH98), stating the differences in the data sets used compared to those listed in the Supplementary Information of the original paper. The full data, as available on your ftp site, will be published as Supplementary Information to that correction.” See, ABOR/MH/Priv-001935.
For those without the time to read the 7,511 pages of Hughes’ material, below are some call-outs that I believe help reveal the nature of work done by this small team of academics during this seminal period of time.
An example of normal discussions between scientists. They were within the norms for civility, addressed specific scientific questions, challenged each other’s analysis, and recognized the strengths and weaknesses in the work. Clearly, each side was defensive, but no ad hominem.
Concern about keeping funding agencies happy is made clear, but with regard to credibility, not outcomes.
Interestingly, Hughes notes that reconstructions of 1000 years ago, “should be viewed as very preliminary” but this is not how the MBH99 is portrayed. Notably, Hughes also identifies what is needed to shore up the utility of these reconstructions.
Hughes contributions to this dispute were small, even tempered and professional.
Hughes rarely discloses his concern about defending his work, unlike Mann. He does, however, have such concerns about his work holding up. “It seems that you and Tom did a pretty good job, as, for readers without knowledge of what was going on, the effect on our work is probably no worse than neutral”. The previous emails in the chain indicate they had to defend their work against reviewers and beat back criticisms enough to get the paper accepted.
Showing how Mann picks peer reviewers.
Bradley showing problems with the proxies
Reference to talks Hughes had in Japan makes clear why this kind of record is valuable. He identifies individuals who have new ideas, ones that would significantly caution the utility of Hughes own work.
Example of a pay to publish behavior.
Finally, the Overpeck collection will become available in January. It is a much larger collection (~95,000 pages) and, in the main, addresses preparation of the IPCC Fourth report. That collection contains a large amount of formatting code which makes much of it unreadable. As soon as I have removed that coding and taken an initial look at the collection, I’ll make it available in the same manner as we did the Hughes collection.
via Watts Up With That?
December 10, 2018 at 12:38PM