Climate Audit

Barclay E MacDonald

Nicely written and very helpful to readers like me, who don’t have the information at our fingertips nor the time or inclination to wade through it.

Tom Fuller

I am reposting a comment here that I left on Angliss’ site, in case it somehow disappears:

Angliss, the argument from quantity you make in your previous post is so poorly constructed that it does not even rise to the level of wrong.

1. The model you choose to examine is incorrectly selected. The emails are the exact opposite of a bug that is being actively sought for removal. The emails related to ethical and professional misconduct are being actively concealed. Your argument that behaviour surrounding the emails can be compared to emails surrounding a software bug fails.

2. Your estimates of the quantity of emails is, as Mosher notes, not realistic (unless you include spam) and not relevant. The quantity of emails is not realistic. It does not correspond with levels of traffic reported by similar professionals in other contexts, and I would refer to the authors and ask if multiple addressees were counted as separate emails (which a server log analysis would mistakenly do). It is not relevant. People engaged in wrong doing live most of their lives and conduct most of their interactions exactly as do the innocent. Hence the ubiquitous phrase, ‘he was a quiet man. I never would have thought…’ Indeed, the more intelligent the ‘malefactor’ the harder the attempt to look and act like the innocent.

If you were attempting an honest investigation, you would have mapped the remits of the investigations done so far and charted their results, instead of saying that they had largely exonerated CRU and the individuals whose behaviour they looked at. Serious criticism of institutional and personal behaviour was noted. Serious allegations have yet to be investigated. You make no note of what is falling through the cracks.

Your argument that insufficient data is available for analysis is simply innumerate, and is essentially refuted every time a poll of 1,000 people is extrapolated to correctly predict an election. If you were to assume a total of 50 million emails involving the subjects of the controversy you would only need a sample of 666 emails to be able to make statistically significant statements at a 99% level of confidence with a confidence level of +/- 5%.

Every one of your arguments fails both logically and mathematically. It is obvious that you reached your conclusion and began to build a model that would provide you with data needed to support your conclusion.

That this level of Weird Science is serving to distract so many people in 2010 shows several truths: First, the political arguments of those who believe catastrophic global warming is the likeliest outcome of our emissions of CO2 are getting progressively weaker. Second, attempts to construct an electronic record rebutting criticism from Steve McIntyre and Mosher and myself have so far avoided a thorough examination of what we have written and the sources we have used. It is clear that you have not read much at McIntyre’s site, Mosher and my book, and it seems evident that you have not read all of the emails.

You found a cute little trick. You’re riding that pony for all the links and exposure you can get. But you are wrong on every count.

• Dave Andrews
  Posted Jun 13, 2010 at 3:38 PM | Permalink

  Not only does he fail logically and mathematically, he also fails on an historical basis by claiming that the emails’ context cannot be understood without interviewing the authors.
  On this basis we would not be able to understand anything about non recent history at all.

  • J Bowers
    Posted Jun 13, 2010 at 4:29 PM | Permalink

    I think you attribute too much certainty to historians. By your reasoning 99% of historians are actually now out of a job. Or, could it be that historians always strive to discover primary source evidence for a reason, but even then hold a question mark over their conclusions if the subject or witnesses are long dead and cannot be directly consulted?

  • Brian Angliss
    Posted Jun 13, 2010 at 4:38 PM | Permalink

    Fuller (#51): If you wish to prove that my prior analysis was wrong, simply stating that it is isn’t a great place to start. Prove it.

    While the emails aren’t bugs, they do represent an electronic record. And according to Aranda and Venolia, electronic records are “deeply unreliable,” as I pointed out in the other post. This is especially true of records that are likely filtered and/or analyzed using electronic methods instead of humans asking questions of other humans. The CRU emails very likely qualify, a point that Mosher claims you and he made in your book. The graphs are entirely clear – the electronic records almost always got the number and names of people involved wrong and almost always got the number and type of key events wrong, and in nearly every case, the errors were not discovered until the paper’s authors sat down with the people involved in the electronic record and talked to them. Let’s not forget that

    the paper found that the electronic records including the email records were missing or had incorrect data, failed to include events that were critical to solving the bug, didn’t describe structural issues and problems related to group dynamics and internal company politics, and had very little explanation of why things were done. For example, the authors found that the steps required to reproduce a bug, the list of corrective actions taken to try and fix the problem, and the root cause were often missing from the electronic records. The bugs often had lifespans that started in advance of the official record or ended either far before or well after the bug was actually declared “fixed.” The authors found that the officially responsible person (ie the bug’s “owner”) was not the person actually responsible for fixing the bug 34% of the time and were totally unrelated 11% of the time. Furthermore, in 7% of the bugs, all of the people listed in the electronic record had no relationship at all to the bug.

    At best, Tom, you and Mosher got half the story by talking to McIntyre et al. But given the findings of Aranda and Venolia, you’ve probably got much less than that. I can certainly appreciate that you don’t want to believe it, but the quantitative and qualitative findings of Aranda and Venolia can’t be dismissed by assertion – they have to be disproven, and thus far you’ve not even come close.

    The estimates of the quantity of Jones’ emails (which I didn’t use in my post, if you recall) are not realistic. But the quantity of emails for the electrical engineer (me in my day job, FWIW), the marketing professional, the english professor, the home manager, and the climate scientist (Tom Wigley) are all based on estimates from the people involved. If you have proof that those numbers don’t “correspond with levels of traffic reported by similar professionals in other contexts,” provide it please. I’d be interested in knowing if my own email totals were high or low compared to other EEs.

    You’re mathematical argument is the worst you’ve made yet. 666 emails would be necessary if they were randomly selected. You and Mosher make the point that they aren’t randomly selected – they were selected from the hundreds of thousands to millions of emails based on certain search criteria. Therefore, it’s not possible to make claims about the emails that apply to the entire population of emails. In public opinion survey terms, you’re using results from a self-selected survey to draw conclusions about the population at large when the best you can really do is draw conclusions about the respondents.

    Which is a great way to make my argument for me, BTW. Thanks.
    [Used your 4:54pm edit to update this earlier version. I hope I did that correctly, Brian!]

    • David S
      Posted Jun 13, 2010 at 5:02 PM | Permalink

      I have no idea where this numbers game comes from. If the emails provide compelling evidence of wrongdoing, and to me they do, then it doesn’t matter at all how they were compiled or how many others there are. This is not a sampling exercise, it is real life, with a lot riding on it. You don’t need to extrapolate from a smoking gun.

    • Brian Angliss
      Posted Jun 13, 2010 at 6:26 PM | Permalink

      No, David, but you do have to understand the reason that there was a gunshot in the first place. To use an admittedly extreme example, the difference between murder and self-defense depends on motive, location, and who is holding the gun. At this point there remains a serious question as to whether we have enough context to understand all the factors involved.

    • Tim
      Posted Jun 13, 2010 at 7:18 PM | Permalink

      1) The emails do not stand alone. SteveMc and others have been making claims for years which are corraborated by the emails.
      investigation by neutral body that seeks information from everyone. If there is a enough prima facia
      2) If there is a gun and dead body there is always an evidence then it goes to trial where all witnesses are cross examined by a lawyer seeking a conviction.

      None of 2) has occurred with the emails. The investigations that we have had are a joke because they amount deliberately avoided looking at any contrary evidence and simply accepted the accused explanations with no cross examination.

      IMHO, it is the unwillingness to investigate the issues properly which is most damning because it shows that the climate science establishment cannot be trusted to provide the public with a truth analyses of science or anything else.

    • RDCII
      Posted Jun 13, 2010 at 9:25 PM | Permalink

      Actually, Angliss, it turns out that the reason that all the emails aren’t available to provide a full context is that East Anglia has gone to extraordinary and, by some accounts Prima Facie illegal efforts to avoid satisfying FOI requests.

      If there are existing emails that could provide a less deceitful perspective on the available emails, then all East Anglia has to do to re-establish trust is…release them. Why don’t they do that?

      It’s not just that what the available emails say is damning, it is that if there are any historical exonerating emails…where are they? It seems like it would be in EVERYONE’s best interest to release them…if they existed.

      Perhaps you could try an FOI request to make all the emails available?

      Steve: most recently, the University refused my FOI request for the attachments to the controversial Wahl-Briffa emails>. The parties don’t wish to “clarify” these emails after all.

    • thefordprefect
      Posted Jun 13, 2010 at 11:25 PM | Permalink

      Steve: most recently, the University refused my FOI request for the attachments to the controversial Wahl-Briffa emails>. The parties don’t wish to “clarify” these emails after all.

      Steve the person requesting McIntyre and McKitrick emails got the same response. Why? Perhaps you would like to expose ALL your communications with McKitrick to public scrutiny. McKitrick has no more right to keep his emails private than does jones&co – both sides are university based.

      So lets keep things open – show your emails McIntyre!

    • bender
      Posted Jun 13, 2010 at 11:35 PM | Permalink

      Do you have any hint of any evidence that Steve has ever done anything as unscrupulous as what the hiders of the decline did? I bet not. And that’s the difference here. When your work smells fishy, you are obliged to come clean. Steve’s work does not smell fishy.

      Do you understand why the two situations are completely different?

    • bender
      Posted Jun 13, 2010 at 11:39 PM | Permalink

      Ford forgot to quote this part of what Steve wrote:
      “The parties don’t wish to “clarify” these emails after all.” Perhaps Ford doesn’t understand the irony here: people like Gavin et al. declaring that the emails only look bad when taken out of context. The truth is context makes it worse.

    • RDCII
      Posted Jun 14, 2010 at 1:52 AM | Permalink

      You seem to have glossed over what I said…that making the emails that would provide a “corrected” perspective on the released emails available would be good for EVERYONE. Do you disagree that, if such emails exist, it would be in East Anglia’s best interests to release those emails?

      Demanding emails from Steve based on the above observation is, in my opinion, a deliberate non-sequitor to the discussion in question. Do you have anything to contribute to the question of whether we should make our own judgements on emails with the context we have, since East Anglia is not providing any emails to give us more context?

    • Eric Barnes
      Posted Jun 14, 2010 at 10:18 PM | Permalink

      snip – policy

    • Steven Mosher
      Posted Jun 13, 2010 at 5:24 PM | Permalink

      Brian:

      “At best, Tom, you and Mosher got half the story by talking to McIntyre et al. But given the findings of Aranda and Venolia, you’ve probably got much less than that. I can certainly appreciate that you don’t want to believe it, but the quantitative and qualitative findings of Aranda and Venolia can’t be dismissed by assertion – they have to be disproven, and thus far you’ve not even come close.”

      During the period of Nov 17th to Nov 19th I talked to Steve Mcintyre on probably 4 or 5 separate occassions. I detail the content of these discussions in a couple places. Go find them.
      My discussions with Steve centered around VERIFYING that the mails in the package were mails he sent. We discussed his Nov 13th FOIA appeal rejection. We discussed the mail he received from UAE.

      From Nov 19th to jan 12th, the publication date of the book I can’t recall having a single conversation with him. My last email to or from him was Nov 19th, 2009. On jan 14th the publication date of the book, we emailed again. On or about Jan 13th we had a phone conversation discussing the post
      “The mosher timeline”

      So, I got virtually NONE of the story by talking to mcintyre.
      I did what I am trained to do. I read the mails. I read the posts. I constructed a narrative of who said what.

      I confine my conclusions to what the mails actually say.

      Now, you on the other hand, characterize steve’s work without having read even the smallest portion of the extant texts. Similarly with our work.

      No amount of additional mails will make those that exists go away.

      I like your bug metaphor. But here is the problem.

      Jones doesnt deny that he asked mann to delete mails.

      Are there any POSSIBLE mails that would make this request
      legit? Can you imagine a mail that would make this request legitamate? When jones agrees this was a mistake, can you at least take him at his word?

      Then can you ask the question that we asked.

      WHY? why did Jones ask mann to delete the mails?

      Well, the mails that lead up to the request give you a narrative.
      That time space is VERY short. So you are not talking about a sample of millions of mails. You are talking about mails sent
      or a very short period.. 2-3 days. The mails in question appear to be the complete record of all correspondence on this matter. There were only 2-3 people on these mails
      Palmer, Jones, Osborne, Briffa. However, if you would like to argue that there might be more mails in this short period, please feel free. And what precisely COULD those mails say?
      what missing evidence could make Jones actions laudable?

      lets say you send your employees 50 emails a day.
      One of those emails lands in my lap. That email says

      ” Mike, the legal department says I am being disposed on a patent case.and there is a discovery request. Can you please delete all emails I sent you about the ACME’s xyz circuit patent?”

      Do I really, need to see the other 49 emails to draw a conclusion about that request?

    • Green Sand
      Posted Jun 13, 2010 at 5:28 PM | Permalink

      Re: Brian Angliss (Jun 13 16:38),

      I have to agree with Brian:-

      “This is especially true of records that are likely filtered and/or analyzed using electronic methods instead of humans asking questions of other humans.”

      How about for once we have one set of “humans” i.e. any inquiring body asking questions of another set of “humans” i.e. McIntyre and McKitrick?

      You never know we might just get a resolution that all “sides” understand.

    • Steven Mosher
      Posted Jun 13, 2010 at 5:47 PM | Permalink

      Actually Brian, this is what I wrote to you:

      “With regard to the issue of deleting mails or the selection of mails. Its clear that whoever collected the mails did not read them all. I cover this in the book. It’s also highly likely that some form of automated collection process or automated selection process was used. very simply, the mails were not selected by a human. There are a number of mails that are clearly housekeeping mails. mails like out of office replies. Any human editor would delete these mails. They are pure noise. There are enough of these mails, mails which have nothing of import, to suggest that the mails were collected and filtered by a harvesting program. A program that looked for certain authors, and certain key words.
      In the first 24 hours of reading the mails this became clear to me. It was not simply a dump of all mails. If that was the case, there would have been a very low signal to noise. in 13 years one would expect more than 1 or 2 “out of office” replies. Neither was it a human selected corpus. No person trudging through these mails and sending them to others would include mails that are not relevant. Again, the out of office replies, the housekeeping mails, mails about this conference or that conference. these mails are caught up in the net because they mention certain key words, like YAMAL, SRES, GCM or because they mention key authors. Finally, the person who “selected” the mails or wrote the program that selected the mails, did not have a firm grasp on all of the issues.
      they read some of the mails, but did not understand all of the issues. For example, they highlighted this mail: 0939154709.txt That’s important because it shows that they didnt understand what they were reading. What they thought was a ‘bombshell” ,cutting off proxy data in 1960, was well known in the skeptic community. there are other things we can suggest about the person who did this, age, gender, education level, sense of humor. But in the end, whodunnit is a distraction from the main event. It fascinates people because they can’t answer the questions. Personally, I am fine with living in that uncertainty. We have to act under conditions of uncertainty. Whether that means acting to save the planet when the science is less than certain, or acting to save the process of science when we are uncertain about the motives of people who released the mails.
      Doubt should not stop us from taking reponsible action on climate change, and doubt should not stop us from calling for changes in how climate science is done. We want more transparency and more openness. The motives of the hacker dont change our principles. How could they? ”

      Now, when you interview me and take selections, did you honestly think that I would ignore the way you choose to edit things and select things? When YOUR argument is that we havent read enough, did you honestly think you could get away with criticizing Steve without reading his complete corpus.? Again, you’re not an idiot, but what were you thinking

    • WillR
      Posted Jun 13, 2010 at 6:01 PM | Permalink

      Re: Brian Angliss (Jun 13 16:38), Brian:

      I think you would have to prove that a discussion of software bugs and the accuracy of the email traffic relating to those bugs — etc. Was relevant to this particular issue.

      You have asserted that the accuracy of the electronic record of tracking software bugs and doing “break fixes” is somehow relevant to a discussion of paleo-climatology.

      Hopefully you can show how this is so. I don’t see it and I suspect that others don’t either.

      Since a court of law routinely accepts email as “evidence”, then it seems it might be ok here as well.

      To answer your questions before you ask, I have run large software projects and the follow on maintenance and I have run R&D projects. SO I am familiar with the ground you cover.

    • Brian Angliss
      Posted Jun 13, 2010 at 6:30 PM | Permalink

      I have established the relevence (and the paper’s authors have concurred with my explanation in the comments of the post) here. I recommend that you read the post in question and the Aranda and Venolia paper as well.

    • WillR
      Posted Jun 13, 2010 at 7:29 PM | Permalink

      Re: Brian Angliss (Jun 13 18:30),

      I don’t know if all the software authors had a PhD — or even a masters.

      Let’s look at this in a slightly different manner. The assertion is that people with a PhD who run research projects, write long technical reports and, write grant proposals and routinely make involved technical presentations are in the habit of writing emails wherein they say things they don’t mean and write things that are easily misunderstood. Furthermore they are in the habit of creating technical graphs that can be easily misconstrued because they routinely make a lot of mistakes. Furthermore they are in the habit of contradicting themselves because of these mistakes. (assumption).

      However, I read every email in the file and I did not see a lot of retractions and corrections in the email record — which would be consistent with your assertion. Perhaps you can point this out or explain the lack.

      Why would a gang like this be advanced millions of dollars in grant money? Good question?

      Why would people change social and economic policy in the values of trillions of dollars of public money? Reasonable question?

      When you assure me that the authors of the software study were studying people at the PhD level, that were in fact the R&D directors then it would be worth my time. Is it so? If they were studying junior programmers and customers then I could see that the study might be what you said — as little as you offered of the basis of the study.

      Hope that’s more clear.

      Now if they want to accept your defense, then I suspect they are about to lose any hope of further grant money.

    • RomanM
      Posted Jun 13, 2010 at 7:42 PM | Permalink

      Re: Brian Angliss (Jun 13 18:30),

      snip – politeness rules

      From the A & V paper:

      Our data come exclusively from Microsoft, and the extent to which our results are valid for other companies is not clear without replications. As is well known, Microsoft has tens of thousands of employees, millions of daily users, and many interconnected products. These are all forces that shape coordination dynamics.

      Not only are they talking about programmers who are dealing with technical programming problems, but the environment is not exactly Microsoft magnitude. Perhaps you can show me where the study is replicated to show that there is any relevance to a UEA size workplace whatsoever. This particular issue was in fact not addressed by Mr. Aranda’s comments on your blog.

      Your over-the-top post is an example of exactly the reason that a lot of well-educated intelligent people don’t buy the type of silliness offered by many proponents of CAGW.

    • willard
      Posted Jun 15, 2010 at 4:17 PM | Permalink

      snip – foodfight rules

    • RomanM
      Posted Jun 15, 2010 at 4:41 PM | Permalink

      snip

      Perhaps, you don’t understand that the paper referred to by Mr. Angliss had ZERO relevance to the subject under discussion. This had been pointed out to him clearly in several prior comments, but he refused to acknowledge this. I merely “played his game” by also indicating that the authors restricted any claim of application to a very specific situation which had no similarities in either content or environment to the CRU emails.
      snip

    • willard
      Posted Jun 15, 2010 at 5:35 PM | Permalink

      snip

      Perhaps Ambliss should amend his essay to forewarn his readers that it’s not scientific, or not scientific enough. Perhaps only irritation can explain why Ambliss’ work must be dismissed on the sole ground that it’s not scientific. Who knows?
      snip

    • willard
      Posted Jun 15, 2010 at 7:20 PM | Permalink

      snip

      The argument only shows that the extent of the results are valid is not clear without replication. From this lack of clarity, RomanM concludes that Ambliss’ work is nonsensical. This conclusion is a non sequitur and, independently, irate.

    • RomanM
      Posted Jun 15, 2010 at 8:05 PM | Permalink

      The argument only shows that the extent of the results are valid is not clear without replication.

      No, it says more specifically, that to use these results in a situation outside of where they were developed is unjustified.

      However, the “nonsense” of the use of the results had already been indicated by previous posters in the fact that there was no similarity whatsoever between the underlying situation dealt with in the paper (in particular, programmers working on technical problems in software development) regardless of the environment. This was simply something Angliss was unwilling to provide justification for. I merely pointed out that the authors themselves said that the environment was important and that there work did not extend outside of it.

      No non sequitur involved here. Somewhat irritated, maybe, irate, no.

    • willard
      Posted Jun 15, 2010 at 8:43 PM | Permalink

      Compare and contrast.

      From the article:

      > Our data come exclusively from Microsoft, and the extent to which our results are valid for other companies is not clear without replications.

      My version:

      > The argument only shows that the extent of the results are valid is not clear without replication.

      RomanM version:

      > No, it says more specifically, that to use these results in a situation outside of where they were developed is unjustified.

      And so from “not clear” we jump to “being unjustified”. The gap is less important than talking about “nonsense”, but there is still a gap. And so the non sequitur yet remains.

    • RomanM
      Posted Jun 15, 2010 at 9:17 PM | Permalink

      Frankly this word-parsing approach of yours is a waste of time.

      If the authors state that the extent for which some results are valid is not clear, then there is NO justification for applying those results under conditions which are radically different unless further evidence has been provided that they remain valid in the new environment. This is a general principle in most areas of science and not specific to this particular case.

    • MrPete
      Posted Jun 15, 2010 at 9:10 PM | Permalink

      Re: willard (Jun 15 17:35),
      The problem is larger and deeper than replication. The paper simply does not apply, as I describe elsewhere in this thread.

      Summarizing:
      * A&V metadata was human-entered and a significant portion of the erroneous data (e.g. severity, category, owner, etc). Email metadata is automatic and correct.
      * A&V message content is peripheral to the work done; bugs can and are resolved without change to the bug DB, fixing up the bug DB is an “optional” task. Email content was core communication of the team.
      * A&V data by its very nature is correctable through further interaction. Email malfeasance by its very nature truly exposes malfeasance.

      The name Climategate derives from Watergate, which is a good analogy for this discussion: when a recorded voice conversation reveals nefarious activity, further conversation will not “correct” or “fix” what was revealed. If it was nefarious when recorded, it remains thus.

    • willard
      Posted Jun 16, 2010 at 7:26 AM | Permalink

      MrPete,

      Angliss’ argument is not straightforward and the criticisms that you stated earlier, repeated on his site, and summarized here, need to be answered. I will let him answer for them, as I never wished to defend him here.

      The important point, as far as I am concerned, is that your argument need not be a scientific one. Neither is Angliss’. RomanM relies on it needing to be scientific for him to reach his conclusion. He just made that clear.

      That Angliss’ argument needs to be scientific to be justified, or even meaningful, is not so obvious as to cause irritation when one does not see it. If we are to apply this same kind of reasoning, we would have to say that the comparison between Watergate and Climategate is meaningless. But, of course, we can see some relationships.

      And so we get to the first reason why I replied: if this so obvious why Angliss is wrong, being over-the-top does not help to buy anything, without never really providing reason for or against any given position.
      
      Steve: I will get to this but there are so many things that I have to deal with that it’s hard to pick every spitball off the wall. It’s unfortunate that you can’t recognize how weak Angliss’ arguments with existing materials. He’s retracted one point already.

    • Brian Angliss
      Posted Jun 21, 2010 at 10:54 PM | Permalink

      RomanM – I have provided extensive explanation of precisely why the A&V paper applies to the CRU emails in comments at both of the original S&R posts. Perhaps if you’d care to point out there why it’s actually irrelevant, I’ll be able to address your concerns.

    • gens
      Posted Jun 16, 2010 at 8:59 AM | Permalink

      Willard,

      Based on your website, you seem to have a particular interest in communication and language. This makes your comments rather mystifying as you seem unaware of basic scientific language. RomanM is quite correct – the type of language used by A&V is quite boilerplate and the implications of non-reliance are clearly understood. And as RomanM says, this is a well accepted principle that crosses scientific disciplines. This is rather basic stuff.

    • Tom Fuller
      Posted Jun 13, 2010 at 7:38 PM | Permalink

      Angliss, this is why your argument doesn’t even rise to the level of wrong. The emails obviously were not a random sample. Since they were not, there is no need to consider the remaining body of emails, either quantitatively or quantitatively. However, because you chose to introduce the remaining career-level emails as a reason to criticize the small sample leaked for Climategate, it’s worth pointing out that that small number is sufficient to amount to a reliable dataset.

      It is your handwaving between differing and contradictory arguments that causes this. Your arguments fail.

      It would be easier to prove your analysis was wrong had you shown any sign of doing any analysis. I see no such sign at all.

    • MrPete
      Posted Jun 13, 2010 at 7:49 PM | Permalink

      Re: Brian Angliss (Jun 13 16:38),

      according to Aranda and Venolia, electronic records are “deeply unreliable,”…This is especially true of records that are likely filtered and/or analyzed using electronic methods instead of humans asking questions of other humans. The CRU emails very likely qualify, a point that Mosher claims you and he made in your book. The graphs are entirely clear – the electronic records almost always got the number and names of people involved wrong and almost always got the number and type of key events wrong, and in nearly every case, the errors were not discovered until the paper’s authors sat down with the people involved in the electronic record and talked to them…

      I have some deep and extensive experience with electronic records, communications, and bug reporting.

      Brian, you’re far from having even a reasonable case.

      1) A&V didn’t show that “electronic records are unreliable.” They showed that randomly selected bug tracking logs are unreliable records of the work done associated with fixing the bug.

      2) Climategate emails are demonstrably in a different space:

      a) Climategate emails were not randomly selected. They were clearly selected through a filter of some kind.

      b) Climategate emails were not a minimal document recording various tidbits of information at the periphery of the work. Climategate emails were an essential part of the core communication at the center of this team’s work.

      Thus, in at least these two essential ways, the Climategate emails do not fit the paradigm of this paper.

      Further, as others have alluded, it is not necessary for the emails to ennumerate every aspect of the Climategate story for them to accurately depict the events that they do describe.

      In a bug report, it is quite reasonable for an initial report to say “X, Y and Z are wrong” and then a subsequent communication via another channel to correct that report. And we would say the initial report was “incorrect.”

      However, in the case of Climategate emails, we don’t need to establish the idea that a communication was never later “corrected.” All we need to establish is the content of the communication, and at least a subset of the parties involved. And that, by definition, has taken place.

      You are on very thin ground to assert that the Climategate emails are incorrect as recorded.

      You could easily assert that other communication took place. But for the major implications of the emails, what other communication would negate these messages?

      Imagine for a moment that we were not talking about email, but rather recorded phone conversations. Suppose all we had were a few hours of phone calls recorded… damaging phone calls involving a few senior leaders.

      Now it is true that other phone calls may well have taken place. But the damage is done by noting what was said in the given calls.

      That’s what happened with Watergate.

      And that’s what happened with Climategate.

    • geronimo
      Posted Jun 14, 2010 at 2:50 AM | Permalink

      Out of context is a weak argument in interpreting the emails, but claiming they have been changed by electronic transfer tests the level on credulity to destruction.

      How could these “changes” be made and what are they? Are we expected to believe that these changes somehow made perfectly innocent emails into evidence of refusal to provide data, to delete emails, to “hide the decline”, to get editors sacked, to keep papers out of consideration by the IPCC, to avoind complying with the FOIA etc. If there are changes then why have they all been in the direction of maligning the CRU? Why did the CRU put them into a special file? (I doubt it was to comply with the FOIA requests that were coming in that doesn’t make sense. So why?)

      Is that your hypothesis Brian? if not could you please explain exactly what errors or changes you are inferring.

      Thanks, by the way, for coming on these threads and engaging, if we did more of it it would probably bring the temperature of the debate down and we might get somewhere.

    • bluegrue
      Posted Jun 16, 2010 at 3:44 AM | Permalink

      Re: MrPete (Jun 13 19:49),

      I think, you misread A&V here.

      1) A&V didn’t show that “electronic records are unreliable.” They showed that randomly selected bug tracking logs are unreliable records of the work done associated with fixing the bug.

      A&V randomly selected the bug cases to investigate. That was the only random step in the study. Each of these selected bugs was treated as a seperate case. For each case A&V collected the complete bug tracking logs, e-mails associated with it, the complete electronic record. After that they contacted the people involved. Here’s the procedure in A&V’s own words.

      All of our cases followed the same methodology. First, we queried a product division’s bug database to find a case fulfilling our criteria. We obtained as much information as we could from its electronic records, including the events in its audit trail, all the bug record’s data fields, data on its owners and on everybody that had participated in any action related to the bug, and links to source code repositories. From that point, we traced backwards by contacting the people that had last touched or were referenced by the bug record. If they were not relevant to the history (a common case, due to bulk edits of bugs), we kept tracing back to find agents that were relevant for the bug.

      In each case A&V found the complete electronic log to lack crucial info. So your point 2a) does not apply.

      b) Climategate emails were not a minimal document recording various tidbits of information at the periphery of the work. Climategate emails were an essential part of the core communication at the center of this team’s work.

      So are phone calls, papers, letters, meetings at workshops and conferences, chats over a cup of coffee/tea in the cafeteria, seminars in the individual institutions, visits, … .

      I think A&V is applicable. The system at Microsoft is meant to catch the important aspects of a bug and fails. The filtered e-mails from CRU are even worse, as even the unfiltered set was never meant to be a complete record of the science and interactions of the scientists.

    • MrPete
      Posted Jun 16, 2010 at 11:18 AM | Permalink

      Re: bluegrue (Jun 16 03:44),
      You raise some interesting aspects.

      Random selection. It’s true that A&V filled out the data on their randomly selected cases. However, I suggest it is important that there’s a significant difference between randomly selected cases vs filtered cases, insofar as applicability of the study is concerned. I’m not enough of a statistician to flesh that part out in depth.

      As for the centrality of “phone calls, papers, letter, meetings” etc… you’ve ignored what I said. You’ve also ignored the significance of a bug tracking system.

      Yes, a bug tracking system “is meant to catch the important aspects of a bug.” However, even if the system fails to do that, the bugs get fixed. Even if nobody ever corrects the bug database, the bugs get fixed. Even if the bug database is 100% incorrect in the details, the bugs get fixed. I shouldn’t have to prove that: as A&V have shown (and is obvious to any QA manager), bug databases are notoriously bad at a detail level. It’s pretty awesome if they have a reasonably accurate general picture of the flow of bug reports and fixes and/or can give pointers to who to talk to if you need to learn more.

      A set of recorded conversations (whether voice, email, etc) are radically different, and particularly so for email:
      – The metadata is accurate, period. The parties involved in the conversation are known (could be others as well but certainly we know which email boxes were source and destination!) The date/time sequence is also known.
      – The content is accurate. Either it was directly created by the source (email) or is a direct recording of the conversation.

      Here’s a simple way to recognize the huge gulf between the two data sets: are they valid court evidence, “discoverable” as they say? I’m not a lawyer but have been involved in court discovery processes and in records-management.

      By law, email records are discoverable documents that must be managed according to legal requirements (in a US corporation, you can’t legally destroy email except according to specific policies–typically a minimum 5 year retention policy–especially after Sarbanes-Oxley!) The same is NOT true for a bug database. It’s “just” a database, known to be error-prone. There are no legal requirements to maintain old copies of non-document databases.

      If A&V is applicable, then email records, and voice recordings as well, should not be valid court evidence. They are valid court evidence. Therefore, you are incorrect.

      Case dismissed. 🙂

    • bluegrue
      Posted Jun 16, 2010 at 3:50 PM | Permalink

      Re: MrPete (Jun 16 11:18),
      you are still misunderstanding the comparison. It’s not “filtered cases” vs “randomly selected cases”. The CRU e-mails correspond to a single case in the A&V paper.

      A single case in A&V collects not just the bug db, as you erroneously claim (read the excerpt above), but all pertinent electronic traces, including all e-mails(!) of all people involved in the process of resolving the specific bug, not just of those people mentioned in the bug db. Even in this ideal situation, where you have a complete electronic trace, the image is imcomplete and inconclusive.

      If you look at the CRU e-mails, the situation corresponds to a single case in A&V, but the availablity of data is much worse. Just one of many channels of communication – the e-mails – is available, and the e-mails obviously have been heavily filtered, likely by looking for key persons and key words. You don’t know the filters and you can not even be sure, that even more e-mails were intentionally sorted out by hand in a second filtering step.

      Whether or not any of the data is “discoverable” or not is irrelevant to the question, whether or not the e-mails give you a good image of what has happened.

    • MrPete
      Posted Jun 16, 2010 at 6:46 PM | Permalink

      Re: bluegrue (Jun 16 15:50),

      You are misunderstanding both the data and what I’ve said. I make no suggestion that A&V didn’t collect all the info about what was involved in fixing the bugs. Of course they did (to the extent possible.) That has no bearing on what I’m saying. It also doesn’t matter whether we liken the Climategate emails to a single “bug” thread or a set of “bug” threads.

      One or a thousand, a bug database is an inherently erroneous record of what transpired in resolving bugs.

      Email documents are inherently accurate records of what was said in those emails, who said it, and when. It is irrelevant what other communications took place. And the “discoverability” of the documents is 100% relevant because of that.

      You are arguing about “whether or not the e-mails give you a good image of what has happened.”
      I (and others) are saying the emails give an accurate record of what was said in the emails. And that’s all we need to know.

      If President Nixon said “xyzzy” on a Watergate tape, and that was a damaging revelation, then other tapes, documents, conversations, etc wouldn’t change the damaging nature of the fact that he said “xyzzy.”

      If Jones wrote “xyzzy” in a Climategate email, and that was a damaging revelation, then other emails, documents, conversations, etc wouldn’t change the damaging nature of the fact that he said “xyzzy.”

      The fact that a bug database is often (usually!) an inaccurate and inconclusive record of what ultimately happens when fixing a bug is completely irrelevant.

      The fact that there may be other emails, other conversations, etc is also irrelevant.

      Were the revelations in the Climategate emails damaging? Either they were or they were not. Other conversations don’t matter when it comes to that question.

    • Brian Angliss
      Posted Jun 18, 2010 at 10:23 AM | Permalink

      This is what I wrote at S&R in response to this basic criticism.

      First, you are correct that A&V used a random selection of bugs from a large record. This means that their results are broadly applicable to bug records. That the CRU emails were selected isn’t to the question of whether the A&V paper’s results apply or not. Random selections mean you can apply conclusions to other related samples, while conclusions drawn from selections can only be applied to the selections.

      This is essentially equivalent to public opinion polling – you can draw conclusions on the entire country based on a small sample (assuming you can correct for self-selection biases among the population you’re polling) if the sample is randomly drawn. However, self-selected polls (the American Association for Public Opinion Research calls these “SLOP”) only apply to the people who respond to the poll and cannot be used to draw any conclusions about the population at large.

      So, the question is whether or not the results described in A&V apply more broadly to other electronic records, specifically emails. I’ll address that shortly.

      Second, you’re only partly correct on the metadata. Bug repositories often enter ID numbers and time/date tags automatically and most modern trackers I’m familiar with also automatically enter the “filed by” fields when a new bug is entered. This is equivalent to the from, date, and similar fields in an email. The other inputs into the bug tracker are manually entered, but so are the emails’ contents. The bulk of the useful information in both emails and bug trackers are human-entered and thus subject to human error.

      We can also look at the A&V paper to see if they cover the differences between a bug repository and emails discussing the contents of the bug repository, and the answer is clearly “yes.” Their “Level 2” analysis corresponds to an automated analysis of email records. In this case, they’re investigating whether an automated analysis of the emails produced a correct record as compared to . We can see this quantitatively by looking at Tables 2 and 3 (reproduced below) and seeing how many events and agents were identified when transitioning from a Level 2 to a Level 3 analysis (human sense-making). In every case, transitioning from looking at only the email record (and the complete email record, not a self-selected email record as in the case of CRU) to human-based analyses turned up additional events, and in every case but one, there were changes in who was involved in fixing the problems.

      link to image

      This same effect was observed qualitatively as well. Here’s some quotes from the paper that specifically mention errors in the email or general electronic records:

      People that took actions concerning the bug were often not mentioned in the record or in email communications (C2, C3, C7, C9, C10). [4.2.3]

      The extent of a participant’s contribution was easy to misjudge based on electronic traces: high frequency and intensity of interaction did not imply high level of contribution.[4.2.3]

      It is unrealistic to expect all events related to a bug to be found in its record or through its electronic traces. Naturally, most face-to-face events left no trace in any repository. But in some occasions, the key events in the story of a bug had left no electronic trace; the only way to discover them was through interviews with the participants. [4.2.5]

      Some of the concrete discrepancies we found might be acceptable for large-scale automated analyses of coordination. Others, such as most of the People issues in 4.2.3 and the missing links from bug records to source code change-sets, are far more serious. [6.2]

      The clearest finding in our study, the difference between the minable version and the true version of a bug’s history, should not be Microsoft-specific, as it depends not on corporate culture but on the amount and quality of the information that can be economically and efficiently captured electronically. [7]

      Third, you are correct that the content of the bug database is peripheral and serves as a storage location rather than a core communication link like email. However, for your overall point to hold, the email communications would also have to be peripheral rather than a core communication method. The tables above, however, quantitatively reject that assertion.

      Aranda pointed out above that my comparison to a Level 2 analysis is not quite right even though I think it more correctly captures the type of information included in the electronic record. Instead, the emails themselves represent the base level, or Level 1. He explains it a little better at his personal blog, where he says

      It seems as if the CRU case is more of an attempted level 3 analysis with very incomplete (and cherry-picked) level 1 data, whereas the East Anglia inquiries would be a level 4 analysis. There also seems to be some malice involved in the original analysis. So the link to our paper is not that straightforward.

      Not straightforward, but I think that I’ve explained the link reasonably well at this point. The fact that the original dataset is obviously incomplete (while A&V acquired a nearly complete dataset for each of their cases and still found the electronic records “deeply unreliable”) only strengthens my argument. As does the fact that the emails are clearly cherry-picked and not a representative sample (and thus self-selected). As to malice, that casts even more doubt on the emails, as we can’t know whether exculpatory emails were purged from the record or not.

      After all, if Jones sent an email that said “Please ignore that deletion email. Getting over the flu here and wasn’t thinking clearly.” and it wasn’t included in the record, that would radically change the nature of what we think we know about the emails. And the only way to know would be for the victims of the release to expose themselves even more.

    • Brian Angliss
      Posted Jun 18, 2010 at 12:50 PM | Permalink

      I wrote this in response to MrPete at S&R, and while it’s not a direct response to this exact comment, it does attempt to address his particular complaints.

      First, you are correct that A&V used a random selection of bugs from a large record. This means that their results are broadly applicable to bug records. That the CRU emails were selected isn’t to the question of whether the A&V paper’s results apply or not. Random selections mean you can apply conclusions to other related samples, while conclusions drawn from selections can only be applied to the selections.

      This is essentially equivalent to public opinion polling – you can draw conclusions on the entire country based on a small sample (assuming you can correct for self-selection biases among the population you’re polling) if the sample is randomly drawn. However, self-selected polls (the American Association for Public Opinion Research calls these “SLOP”) only apply to the people who respond to the poll and cannot be used to draw any conclusions about the population at large.

      So, the question is whether or not the results described in A&V apply more broadly to other electronic records, specifically emails. I’ll address that shortly.

      Second, you’re only partly correct on the metadata. Bug repositories often enter ID numbers and time/date tags automatically and most modern trackers I’m familiar with also automatically enter the “filed by” fields when a new bug is entered. This is equivalent to the from, date, and similar fields in an email. The other inputs into the bug tracker are manually entered, but so are the emails’ contents. The bulk of the useful information in both emails and bug trackers are human-entered and thus subject to human error.

      We can also look at the A&V paper to see if they cover the differences between a bug repository and emails discussing the contents of the bug repository, and the answer is clearly “yes.” Their “Level 2” analysis corresponds to an automated analysis of email records. In this case, they’re investigating whether an automated analysis of the emails produced a correct record as compared to . We can see this quantitatively by looking at Tables 2 and 3 (reproduced below) and seeing how many events and agents were identified when transitioning from a Level 2 to a Level 3 analysis (human sense-making). In every case, transitioning from looking at only the email record (and the complete email record, not a self-selected email record as in the case of CRU) to human-based analyses turned up additional events, and in every case but one, there were changes in who was involved in fixing the problems.

      link to image

      This same effect was observed qualitatively as well. Here’s some quotes from the paper that specifically mention errors in the email or general electronic records:

      People that took actions concerning the bug were often not mentioned in the record or in email communications (C2, C3, C7, C9, C10). [4.2.3]

      The extent of a participant’s contribution was easy to misjudge based on electronic traces: high frequency and intensity of interaction did not imply high level of contribution.[4.2.3]

      It is unrealistic to expect all events related to a bug to be found in its record or through its electronic traces. Naturally, most face-to-face events left no trace in any repository. But in some occasions, the key events in the story of a bug had left no electronic trace; the only way to discover them was through interviews with the participants. [4.2.5]

      Some of the concrete discrepancies we found might be acceptable for large-scale automated analyses of coordination. Others, such as most of the People issues in 4.2.3 and the missing links from bug records to source code change-sets, are far more serious. [6.2]

      The clearest finding in our study, the difference between the minable version and the true version of a bug’s history, should not be Microsoft-specific, as it depends not on corporate culture but on the amount and quality of the information that can be economically and efficiently captured electronically. [7]

      Third, you are correct that the content of the bug database is peripheral and serves as a storage location rather than a core communication link like email. However, for your overall point to hold, the email communications would also have to be peripheral rather than a core communication method. The tables above, however, quantitatively reject that assertion.

      Aranda pointed out above that my comparison to a Level 2 analysis is not quite right even though I think it more correctly captures the type of information included in the electronic record. Instead, the emails themselves represent the base level, or Level 1. He explains it a little better at his personal blog, where he says

      It seems as if the CRU case is more of an attempted level 3 analysis with very incomplete (and cherry-picked) level 1 data, whereas the East Anglia inquiries would be a level 4 analysis. There also seems to be some malice involved in the original analysis. So the link to our paper is not that straightforward.

      Not straightforward, but I think that I’ve explained the link reasonably well at this point. The fact that the original dataset is obviously incomplete (while A&V acquired a nearly complete dataset for each of their cases and still found the electronic records “deeply unreliable”) only strengthens my argument. As does the fact that the emails are clearly cherry-picked and not a representative sample (and thus self-selected). As to malice, that casts even more doubt on the emails, as we can’t know whether exculpatory emails were purged from the record or not.

      After all, if Jones sent an email that said “Please ignore that deletion email. Getting over the flu here and wasn’t thinking clearly.” and it wasn’t included in the record, that would radically change the nature of what we think we know about the emails. And the only way to know would be for the victims of the release to expose themselves even more.

      I apologize if I post this twice – I thought I’d posted it once, but it hasn’t shown up after an hour or so. Either the original got caught in a spam trap or I accidentally navigated away before hitting “post comment.”

    • Baa Humbug
      Posted Jun 15, 2010 at 5:20 AM | Permalink

      Re: Brian Angliss (Jun 13 16:38),

      Wow!! You’ve just made O J Simpson and his attorney proud. Well done.

    • Steven Mosher
      Posted Jun 16, 2010 at 3:29 PM | Permalink

      Brian,

      The approach of interviewing people is vital. Let’s look at the study you cite and see the important things that electronic records Lack

      4.2.6 Rationale. Probably the hardest questions to
      answer without human sense-making and participant
      collaboration were the “why” questions: In C4, why did
      a developer choose another as a required code
      reviewer, but a third as an optional reviewer? In C10,
      why was there no activity in a bug record for weeks
      after a few bouts of minute-by-minute updates and
      frantic emails? Why were the Status or Resolution
      fields in C2, C4, and C8 incorrect? Why in C5 did a
      triage group conclude that the bug would not be fixed?

      SO, did you ask Overpeck “why” Briffa would think to label this mail confidential? did you ask briffa Why? did anyone ask Jones why he asked mann to delete mails. In short, did you adopt the very approach of the study you cited? there are other issues as well, but lets start with that. I wont ask if you offered a 500 dollar gift certificate to reponsdents.

    • Brian Angliss
      Posted Jun 21, 2010 at 10:58 PM | Permalink

      Actually, I did ask Briffa why, but because he requested that I not quote him, I’m not at liberty to discuss what his answer was.

      I did take the approach that I claim is necessary as I contacted the principles in the emails I looked into as best I could and I asked you and McIntyre both followup questions that related to my writing.

  • Steven Mosher
    Posted Jun 13, 2010 at 5:28 PM | Permalink

    ya dave, in my old feild of study we called this the “intentional fallacy”

• John Murphy
  Posted Jun 17, 2010 at 6:40 AM | Permalink

  Tom Fuller

  You are right. I tried to get (through FOI request to that dud Palmer at UEA) all the emails passing between Briffa, Osborn and the rest of the hockey team. I was refused. Don’t ask me the stated reasons. The reasons were concocted to justufy a predetermined decision made outside the FOI Act. I can read them a mile away because I’ve been dealing in FOI requests here in Australia for 20 years. It doesn’t take long to learn the signs.

  In light of Brian’s defence, I’m entitled to conclude, as I have, that the whole population of emails (a 100% sample) would merely reinforce the message conveyed byteh one’s we have.

  If Brian knows something we don’t (e.g. he has all the other emails) then he should release them. Otherwise, he should stop arguing through his hat.

ZT

So Brian isn’t interested in answering Steve’s question? What should one conclude from that?

Skip Smith

I’d like to urge a bit of civility here. Unlike many other critics and drive-by smearers, Brian Angliss is actually here engaging us in a discussion.

In my opinion the most fruitful way forward would be to focus on a topic that has a basis in data and fact — the various “tricks” used in temperature reconstructions. Brian has said he’s interested in looking at the numbers himself, so let’s make that happen. Let’s show him the “before” and “after” from various “tricks,” and provide him with the data and code to replicate everything.

dearieme

Look, dendros, tree rings have turned out to be lousy proxies for temperature. Nit-picking at the detail is pointless – the weight of evidence is now overwhelming. McIntyre has done you all a favour by drawing the world’s attention to the deficencies in “treemometer” data; you are now free to find other topics on which to spend your lives.

ianl8888

Angliss seems to have fooled most of the replies with simple straw man tactics

The core of the issue is that Jones felt impelled to “hide the decline” from the public gaze (correctly assuming that the public would not, cannot, check the published literature)

Angliss comes nowhere near addressing the issue and clearly has no intention of doing so … just ignore straw men

• thefordprerfect
  Posted Jun 13, 2010 at 11:33 PM | Permalink

  ianl8888
  Posted Jun 13, 2010 at 7:59 PM
  I repeat a post on the first thread:
  I am “public” but just following links brought up the decline well and truly discussed.

  McIntyre does not dispute a rise in temperatures. So the plots show 1800s to 2000s accurately. I.e. the temperature is rising. Is the plot wrong for the public who would not understand error bars etc.?

  Neil Fisher Posted Jun 12, 2010 at 7:51 PM
  If you read the document by Briffa referenced by the WMO leaflet the decline is well discussed:
  Fig 5 description:
  Note the recent disparity in density and measured temperatures (T) discussed in Briffa et al., 1998a, 1999b).

  text page 96

  A significant outcome of this ‘hemispheric-scale’ comparison between the summer temperatures and NHD1 (Fig. 5, also confirmed in the low-frequency density curve) is an apparent divergence in the post-1950 trends (Briffa et al., 1998b). High-frequency associations (not shown here) remain strong throughout the whole record, but average density levels have continuously fallen while temperatures in recent decades have risen…
  As yet, the reason is not known, but analyses of time-dependent regional comparisons suggest that it is associated with a tendency towards loss of ‘spring’ growth response (Briffa et al., 1 999b) and, at least for subarctic Siberia, it may be connected with changes in the timing of spring snowmelt (Vaganov et al., 1999).There are important implications in this observation not least the possibility of biased regression coefficients in attempts to reconstruct past low-frequency temperature change based on long density series calibrated against recent temperatures. These may overestimate past tem¬perature levels and underestimate the extent of apparent 20th century warming (but see later discussion and that in Briffa et al., 1998c).

  and section 5.3 seems a fair discussions on the problem of tree ring use:

  5.3. Changing Climate Sensitivity and ‘New’ Growth Influences
  In the foregoing discussion, we have alluded to the fact that tree-growth, as represented in various standardised tree-ring chronologies in various parts of the world, often seems anomalous in the 20th century as compared to earlier centuries. Though my no means always true, this is apparent often enough to support speculation that `unusual things are happening’ across large areas…
  UV radiation and a host of other environmental factors may all exert their influence on tree growth. To varying extents, such factors have always affected trees, but their recent influences, and especially the extent of their corn-bined influence, requires serious investigation, beyond that undertaken to date.

  Please remember the document refered to by McIntyre is a small pamphlet not an in depth peer reviewed document.
  From the pamphlet another regerence is:
  [http://www] .pages.unibe.ch/products/newsletters/NL99_1_scr.pdf

  In the immediate future, work will
  continue on important statistical issues
  related to the processing and interpretation
  of all of the various tree-ring collections.
  Potential anthropogenic influences
  on recent tree growth will become
  an increasingly important focus of the
  work. Increased tree productivity during
  the 19th and early 20th centuries and
  post-1950 declines in tree density trends
  have recently been identified in our
  data. The extent, detail and implications
  of these phenomena have yet to be further
  explored. Chronology confidence
  and the expression of climate forcing are
  most strongly expressed on short (annual
  to century) timescales. New

  This hidden decline is shouted from the rooftops in numerous articles and publications. How is it hidden?
  

  • geronimo
    Posted Jun 14, 2010 at 1:09 PM | Permalink

    TFP:”This hidden decline is shouted from the rooftops in numerous articles and publications. How is it hidden?”

    Shouldn’t you be addressing this question to Prof Jones? He is after all the person who wrote “hide the decline”.

    Also if the decline hasn’t been hidden can you point me to any publication from any of the hockey team that shows a downturn of temperature in the 29th century.

    As ever watch the pea, the real issue here isn’t that the hid the decline, it is that the decline makes the proxies unreliable for any part of the reconstruction.

  • ianl8888
    Posted Jun 14, 2010 at 5:30 PM | Permalink

    @tfp

    “How is it hidden?”

    By truncating the tree-ring proxy reconstruction in the early 1960’s and tacking on the instrument record ! ie. “hide the decline”

    Ho hum…

  • Geoff Sherrington
    Posted Jun 15, 2010 at 8:18 PM | Permalink

    Thefordprefect: The question of the decline, “How is it hidden”?

    The answer about the decline, “There are no statements of quantity or magnitude or coverage, and no mention that the text books will need rewriting.”

• Brian Angliss
  Posted Jun 18, 2010 at 10:18 AM | Permalink

  Would you care to point out any straw men I created and why they’re straw men? I haven’t done so intentionally and would like to see any unintentional ones so I can correct them, either by addressing the real point or by showing that the straw men are not actually straw men.

Tom Fuller

Angliss, here are my assumptions about how your project was born. Feel free to correct any misimpressions that I have. They are not really relevant to the results, which I will return to.

Assumptions

1. It appears to me that you wanted to rebut statements made by McIntyre, Mosher and myself. (I am not criticizing you if that is the case–although acknowledging it in your blog posts would have been helpful as well as refreshing.)

2. You (again, naturally, and this is also not a criticism) did not want to wade through 1,000 posts at CA, plunk down $16.99 for a book by people you don’t like just to pick through it for a couple of blog posts, and you didn’t want to read the commentary surrounding existing chronological versions of the emails and didn’t have time to plough through the raw collection. (Again, for a couple of blog posts I don’t have a problem with this.)

3. You encountered the study referenced above and immediately made a connection between their findings and what could be said about the Climategate emails.

It’s what happened next that is important. From reading your two blog posts on the issue and the comments made subsequently here and on your weblog, you refer only to a previously published study regarding software bugs and selected passages from the book Climategate, as well as correspondence with McIntyre and Overpeck. You have arranged them to form a narrative that fits your worldview and the only original work on your part was the creation of your estimates of email volumes for the period covered by the Climategate emails and your estimates of the percentage of the total provided by the Climategate emails.

What I don’t see any evidence of:

1. Examining the work you are criticizing.
2. Evaluating the suitability of the Microsoft study to the Climategate model, or looking for parallel studies or analysis.
3. Attempting to reason through why the behaviour of the Climategate scientists might be different from software programmers.
4. Understanding how quantitative and qualitative data sets should be described, analyzed and reported on.

The problem as it seems to me is that you are trying to paint a faux-scientific patina on what are speculative and kinda fun blog posts. Which is okay until you start slamming people who also use electronic publishing tools.

I don’t think you did your homework. I think it shows.

• Brian Angliss
  Posted Jun 14, 2010 at 5:26 PM | Permalink

  Assumption #1: incorrect. I’ve had experiences reading McIntyre’s work, and I’ve found that some of it is very good while some of it is not. That I found statements that I felt needed rebuttal is entirely coincidental – the first post on how much context is in the CRU emails was the entire point of my contacting Mosher and McIntyre. I didn’t even find you until about 85% of the way through writing the first piece, when I was fact-checking something Mosher said you and he wrote in the book – that’s when I came across the excerpts on your website. In the process of writing the first piece, however, I found a significant number of what I consider to be inaccurate, inconsistent, or misleading statements by Mosher and McIntyre. It was after I did so that I decided to run a companion piece to the original where I addressed the statements that I found. I have enough material that I haven’t used yet (they were more of a “grab bag” and didn’t really fit the main themes of the second piece, so I cut them out to save over 1000 words) to run at least one more piece.

  Assumption #2: partially correct. I’ve read every email to which I’ve been provided specific reference, as well as a number of others. Probably about 10% of them to date, although I’m slowly working through the others (I have the complete archive on my computer, although I find eastangliaemails.com is more user-friendly). You’re correct, however, that I don’t have the time to read 5+ years of Climate Audit posts – I’m an electrical engineer who works in aerospace at a 40+ hour per week job. I admit to being tempted to acquire your book, however, although I’ll wait until after the Muir Russel review has posted their results – I expect that their review will overwhelm anything you’ve done, even if they agree 100% with everything you said.

  3: Incorrect. I encountered the paper well after I started this project, about 2 weeks into doing research (it took me about a month of research and interviews and another week of writing ) when I was pointed to it by someone whom I’d asked for research assistance with a note that it might be applicable to what I was working on. I’ve been interested in this ever since I realized that my own personal experience as an EE informed me that your approach was fundamentally flawed (shortly after the House of Commons report was released, when I first read Tim Osborn’s quote), and the paper provided a non-subjective way to illustrate that point.

  You’re correct that I arranged the quotes into a narrative, but that’s what journalism is – taking the available facts and creating a narrative out of them. However, I interviewed more than just Overpeck and McIntyre – I contacted Eugene Wahl for comment (he refused multiple requests), Keith Briffa (who responded but asked not to be quoted), Martin Vermeer, Steven Mosher, Gavin Schmidt, Tom Wigley, and a few others. The difference is that, while I had an idea where the paper was going, it was guided by my research rather than my personal opinions. In fact, Tom Wigley and Mosher’s responses sent my original idea careening while reading the paper sent my work off in a new, but related, direction.

  • steven Mosher
    Posted Jun 15, 2010 at 12:39 AM | Permalink

    I will suggest that you NOT use eastanglia.com. It is an incomplete record and you will miss some things. ( I think one Jones hyperbole in particular)

    Read all the mails. It took me a couple days the first time through.
    I think the guy who did the guardian book say he almost read all the mails.

    Read all of Climate audit. For two years bender chided me for not doing my homework. So I finally buckled down and read the whole damn thing.

    Read the early real climate posts 2004-2005.

    There is one mail in particular that is missing. Its in there buts its missing. that’s a riddle.

    I will give you a clue.

    Party A at CRU sent a mail to party B… not at CRU

    Party B responded to party A and his response has party A’s original mail attached.

    The original mail ( the thing in your sent folder) is not in the file.
    The mail from party B ( which would be in your “in box”) IS in the file.. and it has the original mail as part of the mail thread.

    This is a test of whether you read carefully and consider everything in front of you.

    Then you can ask.. what happened to the original mail? perhaps somebody deleted mail they sent but not mail they received…..

    ok thats too many hints..

    Then another puzzle.

    Read the last mail. What is in that attachment? The attachment is not in the files, BUT if you read all the mails and climateaudit carefully you should be able to ascertain what is in that attachment. If not, you need to read everything again.

ZT

I doubt that a software bug report has ever been written that instructs readers to delete their email on a given topic!

(But perhaps Brian’s professional experience is more sophisticated).

Ron Cram

Brian,
You admitted an error and struck out a significant portion of your blog posting but the facts are still not as clear to most readers as they could be. You struck out this quote from Steve:

There is no valid statistical procedure supporting the substitution of tree ring proxy data going the wrong with instrumental temperature data to create a false rhetorical impression of the coherence of the proxy data. (emphasis added, source)

And you struck out this portion you wrote:
If the scientists had actually substituted or replaced the tree ring proxy data with instrument data, then McIntyre and Fuller would have a valid claim of fraudulent behavior by Phil Jones et al.

I am certain you can write it better than I can.

geo

What are these “all the other proxy records” that cover the period before ~1500AD that the tree rings track well with?

• Jay
  Posted Jun 14, 2010 at 5:33 PM | Permalink

  They are the proxies that create the +0.5/-0.9 degree variance in the reliability of the various incarnations of the hockey stick models. If, and only if, ancient tree ring proxies are robust temperature proxies can they be used in the models to moderate or eliminate the mid-evil warming period.

  Angliss and his type are not merely haggling about which data sets were used in the 20th century and which were ignored, smoothed, homogenized or hidden behind other lines on a graph, they are fighting to preserve the uniqueness of late 20th century warming. Recall that the catastrophe community has longed blamed CO2 for recent warming, because their research and models found no other causes for the “unprecedented” warming in the past 25 years or so. If the med-evil warming period was warmer than the present, without the presence of elevated anthropogenic CO2 in the atmosphere, then their “CO2 as poison” theory dies on the vine.

  • Robert
    Posted Jun 14, 2010 at 5:59 PM | Permalink

    snip – oT

    • Craig Loehle
      Posted Jun 14, 2010 at 7:01 PM | Permalink

      Robert: the logic is this. It has been claimed that late 20th Century was unprecedented in temperature and that nothing could have caused the warming except CO2. If past periods were warmer, such as MWP, then one must be able to properly model the MWP to way WHY it was warm and go from there, but no one has been able to do that yet. What a warm MWP also says is that IT IS POSSIBLE to get a warm period without high CO2 because we know CO2 was not elevated 1000 yrs ago. If the MWP was warm, it is no longer ok to ignore things just because we can’t model them.

    • Tom Gray
      Posted Jun 14, 2010 at 9:54 PM | Permalink

      There is also the social science meme that M Tobis is currently describing on this blog in another thread. According to this meme, climate scientists have special insight into climate that is a result of their training and experience. From this insight it can be understood their intuition can be relied upon to discern results that are denied to other people whatever their training. Thus the strictures of statistics do not apply to them in their climate work since their intuition allows them to produce valid results without following strict statistical procedures.

      The absence of MWP and thus the strict dependence of global temperature on CO2 concentration was one of these intuitions. If the MWP did exist then what would be falsified is not just an hypothesis about global temperature but the entire meme of the special intuition of the consensus in climate science.

      This meme has been pretty widespread for a while. It has recently been expanded to a broader meme that puts the problem of AGW is a spcial place. Since AGW is potentially catastrophic and that the findings of consensus climate science are difficult for the mass of the population to understand then the issue of AGW should be removed from democratic control. The response to AGW should thus be reserved fro the special insight of the intellectual/scientific elite that can understand it. The existence of the MWP would again place this meme of elite accommodation in the mitigation of AGW ito jeopardy.

      The existence of the MWP is much more than a physical science issue.

    • Robert
      Posted Jun 14, 2010 at 10:21 PM | Permalink

      Craig Loehle,
      It is true that you cannot ignore the causes of the MWP and so on but my point I feel is fully logical. That a larger amplitude of variability in the Holocene is not “the final nail in the coffin” of AGW. Personally, I think that there are natural influences and anthropogenic. The idea that since 1850 CO2 has driven climate is ridiculous but I think it is equally ridiculous to discount any CO2 influence since the 1980s or 90s. Every legitimate climate scientist knows that the MWP had to be more than it was previously depicted, but whether it is warmer than right now is I believe up for serious debate. I’m inclined to think that it is unlikely that the MWP was warmer than current temperatures but perhaps on par is not hard to believe. We will learn soon enough if we have actually passed certain thresholds. If we see an ice free september sea ice extent in the next few years then I think I will have to concede that the Anthropogenic portion is higher than I previously anticipated.

    • Baa Humbug
      Posted Jun 15, 2010 at 5:36 AM | Permalink

      Re: Robert (Jun 14 22:21),

      If we see an ice free september sea ice extent in the next few years then I think I will have to concede that the Anthropogenic portion is higher than I previously anticipated.

      What has sea ice extent got to do with anthropogenic influences?

    • John Murphy
      Posted Jun 17, 2010 at 7:10 AM | Permalink

      Robert

      You are right. The MWP had to be warmer than today, at least in England, Germany and Greenland. When Bede the Venerable was living at the monastery in Jarrow in Northumbria they were growing grapes for wine. Can’t do that now – too cold. In southern Germany at about the same time, the vineyards were about 300 m higher upslope than you can grow graoes now. People lived in Greenland and grew grain – rye and oats to be sure but still grain. Cann’t do that now. they called labrador “Vinland” – not a name you’d use today.

      No one can accuse Bede and his contemporaries of being biased sceptics who twist the data. they just reported their lives and dies over 1,000 years ago,

      Until the proxies can explain those comtemporaneous records (and 100s of others to the same effect), they are not worth two bob, as we say here, and certainly not worth even that as the basis for decisons to spend a cazillion trillion dollars. Make no mistake, that’s why Mann chose the “last 1,000 years” for his “uprecedented” temperatures and “unprecedented” rate of temperature increase. that is teh basis

    • John Murphy
      Posted Jun 17, 2010 at 7:15 AM | Permalink

      Sorry, that comment was meant for Craig.

    • David Stockwell
      Posted Jun 14, 2010 at 10:34 PM | Permalink

      I was thinking of other examples of meme’s or norms that we don’t all share. Eg: regarding atmospheric circulation there is a CS intuition that they will change in a permament way as a result of CO2 increase (the weakening of the Walker Circulation say). When I point out that there is no significant change in any of the NINO indices, the response is ‘well it hasn’t happened yet’. Still the lack of significant change didn’t stop Vecchi and Power et.al. trumpeting it as a finding.

      Another response by a climate scientist sums it up: “The field has gone beyond the blind application of statistics – David Jones.”

      Now I have extensive modelling experience, but I believe more than ever in Feynman’s exhortation to throw the model out if it disagrees with observations.

      I say this as an example in the hope that after they get through ‘storming’ on the other thread, they might get around to ‘norming’ or discussing those substantive assumptions where we differ.

      I think this ‘special insight’ thing is a kind of a rot that has set in because of excessive simulation and we need to get back to the real science falsification, but I am open to the debate over this norm.

    • Judith Curry
      Posted Jun 15, 2010 at 5:47 AM | Permalink

      David, I like the issue you raise, I am digging deep into this right now also, read a bunch of papers on the philosophy of similation. Hope to find a good thread somewhere to discuss this issue

    • Steven Mosher
      Posted Jun 16, 2010 at 3:16 PM | Permalink

      Count me in. I have some questions on how the attribution studies were set up.. methodological concerns. oh ya and like david, old modeler here. war games and such

    • Tony Hansen
      Posted Jun 15, 2010 at 4:43 PM | Permalink

      ‘…kind of a rot that has set in because of excessive simulation.’

      Might I suggest ‘excessive stimulation’ as the likely cause.

Geoff Sherrington

Hilton – which awards?

Without wanting to impede discussion, surely scientific points are finally paramount on this thread and paramount of these is the question of whether dendroclimatology has any continuing value for reconstructions.

After considering the obfuscatory social comments of apologists on this thread, it remains the case, known for some decades now, that if any trees can be used to recreate climate (especially temperature) then they are special trees and there is no way to decide in advance why a tree is special and which ones should be chosen in future. It follows that if there is divergence from the instrumental record now, past divergence cannot be ruled out. Therefore, all past dendroclimatic climate temperature estimates sould be expunged.

It would be beneficial to science if several reliable proxy climate methods could be shown to be reliable within useful error envelopes and I for one wish for them. They are the ones where effort should go.

I have several times criticised the structure of climate science for its near absence of review points and its near absence of decision points that say “Go” or “No-go”. It is not adequate that researchers can write grant aplications that continue the work. It is required that the go/no-go review process be invoked whenever there is a serious stumble, or at preset periods in a program of work.

The same should be done to proxy reconstructions based on all diverse methods. If they reach a no-go point, stop them.

The arteries of climate science proxies are clogged. It’s bypass time.

LearDog

I am waiting for Angliss’ “aha!” moment! It would be a MUCH more interesting article or post for him to write – “you know, I actually took time to investigate (with some help and coaching along the way), and do you know what? The Deniers are right!”

It would reflect well on his ethics, humility and intelligence.

But I’m not holding my breath….

• TGSG
  Posted Jun 15, 2010 at 9:58 PM | Permalink

  “But I’m not holding my breath….”

  Good choice that. 🙂

  And Jeff Id nailed the central theme up thread a ways.

E O'Connor

Tom Gray

Re M Tobis’ musings on climate scientist’s special insight/intuition that allows them to discern results denied to other people so the strictures of statistics do not apply to them in their climate work as their intuition allows them to produce valid results without following strict statistical procedures.

This reminds me of the first line of a song popularised by Muddy Waters –

“Got my mojo working, but it just won’t work on you”

• Craig Loehle
  Posted Jun 16, 2010 at 10:09 AM | Permalink

  If your mojo is strong but you still can’t predict anything and there are divergences laying around unexplained, then maybe mojo isn’t enough. Just sayin’

  • E O'Connor
    Posted Jun 17, 2010 at 8:58 AM | Permalink

    Indeed. Actually, M Tobis is on record as saying,
    “Proposing that matters of fact are decided by intuition or faith is a very very very bad idea.”

    What irked me was the pompous attitude in his posts, which I perceived as ‘Only we climate scientists have the knowledge and understanding, so trust us, we know what we are doing and the cheque is in the mail.’

    His visit here was done in a moment of weakness to grace the lion’s den with his presence.

John Murphy

Looking at Briffa’s data with the decline “unhided.” Has anyone ever considered that Briffa’s tree rings are recording something like the truth? That is, that the instrumental data has something wrong with it and is reporting too high?

John Murphy

Tom Fuller

You are right. I tried to get (through FOI request to that dud Palmer at UEA) all the emails passing between Briffa, Osborn and the rest of the hockey team. I was refused. Don’t ask me the stated reasons. The reasons were concocted to justufy a predetermined decision made outside the FOI Act. I can read them a mile away because I’ve been dealing in FOI requests here in Australia for 20 years. It doesn’t take long to learn the signs.

In light of Brian’s defence, I’m entitled to conclude, as I have, that the whole population of emails (a 100% sample) would merely reinforce the message conveyed byteh one’s we have.

If Brian knows something we don’t (e.g. he has all the other emails) then he should release them. Otherwise, he should stop arguing through his hat.

• steven Mosher
  Posted Jun 18, 2010 at 1:13 AM | Permalink

  Its this easy.

  Sit down and write the mail that would make Jones request to mann to delte mails more acceptable..

  Dear Mike,
  I know I told you to delete mails. That was wrong. I wrote in haste and we are under a great deal of pressure here. After a day’s reflection it seems to me that we should just trust the system. Holland has his request in and it was denied on the grounds of confidentiality. Palmer expects an appeal. Be that as it may. If we lose the appeal and have to turn the mails over, then I suppose Keith and Gene will come under some criticism for bending the rules a bit. Osborn, who likes to research these things, actually thinks that the IPCC rules don’t prevent the kind of exchange that Keith and Gene had. But you know the skeptics will have a field day. I suppose we will have to tough it out, put our faith in the science and press on.

  Best Phil.

thefordprefect

John Murphy Posted Jun 17, 2010 at 6:14 AM

I hope you are not serious with your ludicrous statement!

Trees do not grow at -40C
trees can grow slowly at -10C
trees can grow slowly at 0C
trees can grow moderately slowly at 5C
trees can grow moderately quickly at 10C
trees can grow well at 20C
trees can grow slowly at 50C
trees do not grow at 100C
Can you see any shape to the curve of tree growth vs temperature?

Not being a dendro type person I do not know the exact growth pattern but it must all follow a similar shape.

The shape and peak growth temperature will be modified by available food and water.
A rapid increase in temperature may overload the present root structure and growth may be slowed until more area/depth can be grown into. In a forest this would be difficult.

Warmer (not necessarily hotter) temperatures may actually cause the root uptake to mis-match the transpiration rate and growth could diminish.

It is painfully obvious that trees CAN be a +ve thermometer over some temperature range and a -ve thermometer over another range.

Is this an impossible scenario? —-NO
Is this the cause of the decline —-possibly
Is the cause being researched —–Briffa says it needs to be.

• Gord Richens
  Posted Jun 18, 2010 at 11:05 AM | Permalink

  Don’t forget the assertion by some climatologists that sites that were poor proxies for local temperatures could still be good proxies for global temperature reconstructions, thanks to their so-called “teleconnections.”

• MrPete
  Posted Jun 18, 2010 at 1:26 PM | Permalink

  Re: thefordprefect (Jun 18 04:19),

  It is painfully obvious that trees CAN be a +ve thermometer over some temperature range and a -ve thermometer over another range.

  Very good, TFP! Now, can you derive the next implication?

  If tree growth (plants in general actually) can either correlate or anti-corralate to temperature, how do you use them as a reliable temperature proxy for paleoclimate, ie. back when there were no thermometers around to act as a reference?

• steven mosher
  Posted Jun 18, 2010 at 3:40 PM | Permalink

  Re: thefordprefect (Jun 18 04:19),

  And in no case do you see 5 sigma spurts in growth from changes of 1-2C. The tree biologist explained this to us very clearly. You do see these excursions when the tree suffers from mechanical damage (strip bark) and it hapens to species that are krumholtz.

  So ford, when you see an excursion of growth that is beyond the bounds of changes driven by temperature, do you

  1. Use that core and average it with others ( hide the incline)
  2. Toss that tree core out.
  3. test with and without that core reporting the analysis
  4. truncate the proxie series because the tree OVER responds to temperature changes.

justbeau

Desmog is a Canadian web site for championing AGW, aimed at the general public. From a glance, the site looks to be an AGW PR outloet. If so, Angliss would seem a fan of the Team rather than a player. Should one expect anything sensible to be communicated from DeSmog?