Note: I wrote the following over three years ago, but never posted it. It seems particularly apposite to do so in light of the disclosure of the now infamous CRU document drop (or heist). The documents now available provide even more evidence of the McKitrick and McIntyre accusation of scientific misconduct, with individuals other than Mann being complicit in said misconduct.
Even absent any outright manipulation of results–which the documents (especially the comments in the code) also strongly suggest occurred–the failure to disclose data and methods is unpardonable. This piece includes an analysis of intellectual property rights issues as they relate to science generally, and climate science in particular.
I will probably write more later (holiday schedule permitting) on an aspect of the Wegman report that I didn’t discuss in the original post: the climate science “clique” (Wegman’s word). The emails suggest that clique is far too polite a word. “Cabal” or “mafia” seems closer to the mark. But, hopefully, more on that later.
Here’s what I wrote (in China, as a matter of fact) in August 06:
The NAS recently wrote a report which made it abundantly clear that the Mann “Hockey Stick” used to bolster the case for anthropogenic global warming was fundamentally flawed, but out of politeness–or cowardice–failed to pronounce the verdict in so many words. Indeed, the summary of the NAS report and subsequent press coverage it seemed determined to avoid pointing out that Emperor Mann had no clothes.
Hence, it was left to the Wegman Report to proclaim that Mann is indeed as naked as a jaybird. Wegman and his co-authors–very reputable statisticians all–make it abundantly clear that the statistical foundations of Mann’s work are extremely defective. Indeed, Wegman et al sustain all of the McIntyre-McKitrick criticisms of Mann’s work.
From this date hence, anyone who uses the Hockey Stick to bolster any argument that the current increase in global surface temperatures is unprecedented, and is attributable to anthropogenic influences, is displaying either willful ignorance or shocking intellectual dishonesty.
A couple of points.
First, I have read a good deal of the literature on global warming, and have always been quite unimpressed with much of the statistical work in this field. Wegman recommends that future work in this area include contributions from mainstream statisticians, which seems right.
Second, and perhaps more important, even the Wegman report (by design, it appears) tippy-toes around the issue of scientific conduct–or misconduct. At the very least, McIntyre and McKitrick raised serious questions about sloppiness by Mann and his co-authors, and indeed made a plausible affirmative case for scientific misconduct. Moreover, the consistent failure of Mann–and others–to disclose fully methods, data, and code is deplorable.
It is interesting to contrast the lack of transparency in climate research–at least the proxy record research that M-M focused on–with how scientific and statistical evidence is handled in litigation. I do a good deal of work as an expert witness. In such cases, if you make a mistake, it will be discovered. Every line of code, every piece of data, every method is discoverable. The opposing side’s experts will comb over every piece of information you disclose–and you must disclose everything. Opposing counsel will grill you for seven and one-half hours (in a federal case), asking you about every aspect of your methodology and results. You better check, double check, and check again. You can’t get away with claiming your code is proprietary (as Mann has done) or playing bait-and-switch with your data (as McIntyre and McKitrick claim that Mann also did.) It’s the Full Monty.
Although many academics express disdain for expert witness work because it is supposedly tainted by the fact that it is bought and paid for by interested litigants, the adversarial process makes it very difficult for seriously flawed or biased work to survive to influence the outcome of the case. This is not to say that experts will not disagree. They will. Different experts can examine data in an honest and intellectually defensible way and arrive at different conclusions. However, the kinds of stunts Mann has pulled would never fly in litigation. The opposing counsel or the judge would make short work of that.
The stakes in the cases I have worked on have been in the tens of millions, the hundreds of millions, and in one instance, the billions of dollars. Cases of such magnitude are not exceptional. Because of the high economic stakes, each side expends considerable resources to present its own affirmative economic and statistical analysis, and to rebut that of its opponent.
As high as these stakes are, those in the climate change debate are substantially higher. Billions is spent on climate research. Moreover, policies to address climate change likely have direct and indirect costs amounting to the trillions of dollars. Scientific work like paleoclimatology or climate modeling is routinely used to justify the needs for such policies. Given these stakes, “discovery” procedures like those employed in civil litigation seem more than justified. Specifically, full disclosure of data, methods, and computer code to other researchers is warranted. This will impose some costs on researchers, but knowing the burdens in advance they can take the appropriate measures to mitigate these costs. Moreover, it is essential to note that these procedures reduce costs as well–researchers seeking to replicate or evaluate others’ studies will incur a far smaller expense if methods, data, and code are made more transparent. Moreover, greater transparency will reduce the likelihood that scientific errors will go uncorrected. When research is policy relevant, as much climatological and some economic research is, such errors can be extremely costly. We have already paid a very high price to protect Michael Mann’s vanity, and the interests of those with a stake (intellectual, reputational, financial, or political) in the hypothesis of anthropogenic global warming. The point is to get the right answer, not the politically correct answer–and to get it in a timely fashion. Transparency is essential to achieve this objective.
An analogy with intellectual property law is apposite here. The key issue is whether disclosure will reduce the amount of original research by more than it will increase the “follow on” research. My intuition suggests the answer is no inasmuch as most of this research is not commercial in nature. Under these circumstances, protecting the original research through a form of “trade secrets” barrier is inappropriate. An analysis along the lines presented in the Landis-Posner book The Economic Structure of Intellectual Property Law is apposite here. Landis-Posner note the tension inherent in protecting intellectual property. On the one hand, protection generates a stream of rents that induce production of intellectual property; absent such protection, copy-cats could merely reproduce the property at virtually zero cost without incurring the expense and effort of creating it in the first place, thereby freeriding off the efforts of the inventor and undermining his incentive to innovate. On the other hand, increasing IP protections raises the costs that others incur to produce new works. That is, one can view existing work as an input to new, follow on work; high IP protections raise the cost of this input, leading to less work building on the old.
Inasmuch as basic science–including paleo-climatalogical research–is not funded by the sale of the intellectual property, but through grants and other mechanisms, the benefits of IP protection are small. One could argue that scientists are motivated by non-pecuniary factors, such as the prestige of being the first to publish a result. At most, this justifies the protection of IP to the point of publication. Once a result has been produced and published, the benefits of protecting the IP are minimal, while the costs are appreciable. Given the cumulative nature of scientific discovery and the centrality of the replication process, past research is an essential input to future research, and raising the cost of utilizing past research imposes a burden without a corresponding benefit. I am very hard pressed indeed to provide any economic justification for Mann and others to refuse to disclose their computer code, or to make it burdensome to access their data, or to impede replication of their work by publishing only obscure–and arguably misleading–descriptions of their methodology.