In this particular case, I am arguing that the problem with the methodology is that nobody is checking it, and indeed, it is uncheckable.

Actually, that is exactly what we’ve been doing.

You are arguing that there are problems with the methodology used by climate modelers.
I am asking you what your evidence is.
You are replying that nobody knows and expecting me to take your criticism at face value.

You’ve made an allegation that you haven’t backed up.

Nobody knows the whole of the methodology. We do know bits.

I haven’t said that I’m absolutely certain it’s wrong. I’m saying that there’s no evidence to indicate that it’s right.

Suppose I were to say I have absolute proof that AGW is wrong. You would naturally ask what my evidence for this was. I would reply that nobody knew, but that since you haven’t shown any flaws in my proof we can go ahead and dismiss AGW without further ado. And what I say is true. You haven’t found any flaws in my proof, and neither has anybody else, because nobody has checked it. Granted, one person managed to figure out some bits of it and said he thought the results it produced were “meaningless” but how can we tell if he was right? Since nobody has checked, and that one guy didn’t have a complete understanding, it isn’t really the identification of an actual flaw in my proof, is it? Maybe my proof is right?

This is data associated with a published, peer-reviewed paper. The scientific method, as I was reminded above, relies on other scientists being able to replicate and check published science to find and remove any errors. But in this case, this is impossible because the methods were not public, were sufficiently non-trivial that Harry spent three nightmare years disentangling the code rather than start again from scratch, and where even Harry said he could not reproduce exactly what was done.

And a lot of people seem to think this is perfectly normal and nothing to worry about.

“And it’s a result that has held up repeatedly since.”

By “result” I assume you have subtly shifted topics to the hockeystick pattern again, rather than the “result” I was actually talking about which was MBH98/99 specifically.

I’ll ask again. Did the MBH98 reconstruction pass the R2 verification test? Had Mann calculated it at the time?

So nobody knows the methodology, but you’re absolutely certain that it’s wrong.

Suffice it to say that A.) I think you’re absolutely wrong on the first point and B.) You apparently have no evidence to claim that there was anything for peer review to catch. Let me know when you can answer the question “What is wrong with the methodology used to calculate the “flagship data product?”.

I didn’t say “the hockeystick pattern” I said the MBH98/99 Hockeystick, the famous one. It was published. It got past journal peer-review. It got past the IPCC review.

And it’s a result that has held up repeatedly since. Again, what is your complaint here? That they didn’t apply your preferred statistical analysis?

Well, I replied to that, but it seems to have disappeared. Not sure why. I’ll try again.

First paragraph – this isn’t a fragment of a conversation. All of these questions ought to have been answered by the enquiries.

Second paragraph – if you read the rest of the file, and bear in mind that we don’t have even the level of access that Harry does – you’ll see why it is impossible for outsiders to “look at the methodology”. Not even the CRU’s own research staff can determine the methodology.

As it happens, I do have a fair idea of what he’s talking about, and I’d tend to agree with him. But it’s really besides the point. To have Harry’s concerns left hanging unanswered like this is problem enough. The question is what have the scientific community done about this? What should they have done? How is extended peer-review supposed to operate in a case like this?

Third paragraph – I’m already aware of both of those, and already knew that the arguments in them had been shown to be incorrect years ago. But more to the point, neither of them answer this specific question.

Fourth paragraph – I didn’t say “the hockeystick pattern” I said the MBH98/99 Hockeystick, the famous one. It was published. It got past journal peer-review. It got past the IPCC review. It got past the paleo-climate chapter’s lead editor (for obvious reasons, given who that was). It got past the rest of the scientific community, the journalists, governments, activists for years subsequently.

It’s no use trying to distract attention away from MBH98 by offering me a set of alternative targets. Once we’ve settled what’s what with MBH, then we can move on to pointing out the obvious flaws in all the rest. But there’s not much point if we can’t even answer a straightforward scientific question about a specific reconstruction.

Did the MBH98 reconstruction pass the R2 verification test? Had Mann calculated it at the time?

Again, you’re asking me to read Harry’s mind. Who is Harry? What is the wrong method he’s complaining about? Does it actually give the wrong answers? Did anyone answer Harry’s question satisfactorily? I don’t know, and apparently you don’t know either, because you’re only looking at fragments of a conversation.

Now if there is something fundamentally wrong with the “flagship gridded data product” it should be straightforward for some skeptic to take a look at that product, look at the methodology and identify the flaw Harry was talking about. You still haven’t bothered to show me that anyone has done this, so we’re arguing hypotheticals here.

As to your question about MBH98/99, the questions you’re asking have been addressed repeatedly, in particular here ( http://www.realclimate.org/index.php/archives/2004/12/myths-vs-fact-regarding-the-hockey-stick/ ) and most recently here ( http://www.realclimate.org/index.php/archives/2010/07/the-montford-delusion/ ) .

Notice that your comments tend to fall under MYTH#1 The “Hockey Stick” Reconstruction is based solely on two publications by climate scientist Michael Mann and colleagues (Mann et al, 1998;1999). MBH98/99 are not the only studies to have identified the “hockey stick” pattern. It has been reproduced and verified with additional data several times now.

Thank you. You have cheered me up, with that.

Jinchi,

I’ve asked all sorts of scientific questions. And there must have been a hundred comments that could be summed up as “just read the IPCC reports!”, “just read the peer-reviewed literature!”, or worst of all, those ones sending me off on a wild goose chase to some non-scientific anti-sceptic polemic website. As if I was some sort of idiot who didn’t know anything about the debate.

I asked one above. Harry said that he was seriously worried that “our flagship gridded data product” was calculated by a wrong method, rendering a significant part of the output “meaningless”. Was he wrong? I don’t necessarily expect anyone to weigh in with a technical exposition on interpolation algorithms, but I would expect somebody to be able to comment on how the extended peer-review process would be expected to handle this. (Sean made a good attempt.)

We mentioned above the “hide the decline” charts – this is another scientific question. If your temperature reconstruction method doesn’t work, is it scientifically and statistically valid to cut off the bits of your reconstruction that don’t fit the real temperature, and add in the real data?

We’ve also been skirting around the MBH98/99 question above. When you reconstruct temperature from indirectly from other records, it’s useful to keep back part of the real temperature data to test the accuracy of the reconstruction. If the reconstruction does not correlate with what actually happened, you probably have a case of spurious regression. The standard way of measuring this is by means of the squared Pearson correlation coefficient rho, sometimes called the r-squared or R2 test. Did MBH98/99 verification pass the R2 test? Did Mann know the result at the time? I’ve asked these questions before, and didn’t get an answer.

You see? We have lots of scientific questions. Do we have any scientific answers, though?

That’s OK with me. It just makes me sad.

What makes me sad is the number of climate denialists who can cite chapter and verse from a series of stolen emails, but never bother to frame anything as a scientific question. There must have been a hundred comments on Chris’s blog that could be summed up as ‘Just read the emails!’ as though some revelation would strike simply by repeating the mantra. And when anyone challenged their interpretation of the emails, they responded with indignant outrage.

I’m not going to try to read your mind as you try to read Harry’s mind. If you’ve got an actual point, why don’t you simply state it instead of sending us off on a wild goose chase to some climate skeptic website.

Oh, it’s already been done, to the extent that there’s anything much to do. You’re just ignoring it. Or dismissing it. Or accepting the slightest pretext to claim it’s been debunked.

Again. I still have no idea what you’re even complaining about. Now you’re telling me your point has been proven. Why don’t you start by pointing me to this cracker-jack analysis instead of sputtering about my lack of interest in wasting my time trolling through the CRU email dump.

You’ve seen the opinion polls. Is that something that you think you ought to care about?

Not as a scientist, I don’t.

Forgive me if I don’t take you seriously on that.

But if you want to look at it that way, these compilations of climate data are not exactly pure research, either. They’re just an application of statistics – of a form already well known in econometrics and other sciences.

That’s a good point, actually.