Expertise

by Henry on October 20, 2010

“Clive Crook on his blog today”:http://blogs.ft.com/crookblog/2010/10/the-intelligent-use-of-experts/

bq. idolizing experts and disdaining the supposedly ignorant masses is at least as dangerous. The intelligent use of experts is not straightforward. Technical expertise tends to be narrow, sometimes extremely narrow. Many policy-oriented experts are only too pleased to exceed their limits, pronouncing widely and authoritatively on matters they understand hardly any better than non-experts.

My immediate reaction while reading this was that even if the underlying claim is right, it still sounds a bit rich when it comes from someone whose paid job is every week to pronounce widely and authoritatively on matters where he does not possess any obvious expertise. And then I read the next sentence.

Economists and climate scientists spring instantly to mind.

Most readers of _Crooked Timber_ are probably unaware of the “little spot of bother”:http://climateprogress.org/2010/08/04/atlantic-editor-clive-crook-fabricates-another-quote-to-smear-michael-mann/ that Crook got into over climate science a couple of months ago. In a quite remarkably offensive post, entitled “Climategate and the Big Green Lie,” Crook “told us that”:http://www.theatlantic.com/politics/archive/2010/07/climategate-and-the-big-green-lie/59709/

bq. I also believe that the Climategate emails revealed, to an extent that surprised even me (and I am difficult to surprise), an ethos of suffocating groupthink and intellectual corruption. The scandal attracted enormous attention in the US, and support for a new energy policy has fallen. In sum, the scientists concerned brought their own discipline into disrepute, and set back the prospects for a better energy policy. I had hoped, not very confidently, that the various Climategate inquiries would be severe. This would have been a first step towards restoring confidence in the scientific consensus. But no, the reports make things worse. At best they are mealy-mouthed apologies; at worst they are patently incompetent and even wilfully wrong.

bq. … Further “vindication” of the Climategate emailers was to follow, of course, in Muir Russell’s equally probing investigation. To be fair, Russell manages to issue a criticism or two. He says the scientists were sometimes “misleading” — but without meaning to be (a plea which, in the case of the “trick to hide the decline”, is an insult to one’s intelligence).

When pushed on this, he issued a “purported clarification”:http://www.theatlantic.com/politics/archive/2010/08/more-on-climategate/60857/ which suggested equally clearly that Mann and his colleagues had been actively dishonest.

bq. Joe Romm tells me to “retract [my] libelous misinformation and apologize to Michael Mann”. … I think the only issue of substance in his complaint is the charge that I failed to notice that there were two Penn State investigations of Mann, not one, and that both had cleared the accused. Of course I was aware of the form of the inquiry, though I concede that the post was not as clear about the two phases as it should have been. …

bq. It seems to me, and I dare say to other open-minded readers, that the talk in the emails of a “trick to hide the decline” raised the reasonable suspicion that a trick had been used to hide the decline. “[T]o the contrary,” says the report. The “trick” has no connotation of trickery, but merely denotes a “statistical method”. Striving to keep a straight face, let us accept this. What about “hide”? Is it all right to employ a “statistical method to hide the decline”? Why was anybody trying to “hide the decline”? (One response might be: because the data which showed the decline were unreliable. Fair enough, but then this rather casts doubt on the whole series, doesn’t it, not just on the part that shows a decline?). I am not competent to discuss the science, and do not pretend to be. But here is what I see when I read the “trick” email and then the report. An explanation is required. Mann’s account strains credulity, yet is readily accepted.

I have no direct knowledge of the intervening discussions which led to Crook publishing rather substantially “amended”:http://www.theatlantic.com/politics/archive/2010/09/climategate-corrections-and-revisions-1/63070/ “versions”:http://www.theatlantic.com/politics/archive/2010/09/climategate-corrections-and-revisions-2/63071/ of his original posts some weeks later. But whatever their motivations, they represent a quite remarkable _volte face._ This bit in particular presents a striking contrast with the paragraphs quoted above.

bq. Well. It seems to me, and I dare say to other open-minded readers, that the talk in the emails of a “trick…to hide the decline” raised the reasonable suspicion that a trick had been used to hide the decline. “[T]o the contrary,” says the report. The “trick” has no connotation of trickery, but merely denotes a “statistical method”. Let us accept this. What about “hide”? Is it all right to employ a statistical method to hide the decline? Why was anybody trying to “hide the decline”?

bq. Other inquiries and statements by the scientists have explained that a proxy temperature series-used to estimate temperatures going back centuries, before instrumental data were available-shows an anomalous decline in recent decades. This is what the “statistical method” was intended to hide. The purpose was not to fool anyone, but to avoid confusion. All right, but to claim that the issue goes away once you understand this would be wrong. How far to trust a series that breaks down in recent decades (for reasons not yet understood) in estimating temperatures over centuries is a judgment call. Whether it was right to hide the decline for the purposes of the WMO graph, not merely by curtailing the suspect series and ending it early, but by splicing on actual temperatures, is another. Such judgments are, or ought to be, open to question. And how forthright to be in explaining what has been done is another issue worthy of debate. …

bq. I am not competent to discuss the science, and do not pretend to be. But here is what I see when I read the “trick” email and then the Penn State report. An explanation is required. The report offers only half an explanation: “trick” means “statistical method”. No contrary opinions are sought or heard. On this basis the report finds “no substance” in the criticism.

Suggestions that it is hard to ‘keep a straight face’ while entertaining the notion that climate scientists had meant the word ‘trick’ to refer to a statistical method and that Mann’s account “strains credulity” have been replaced with vaguely-worded platitudes about the “judgment calls” involved in dealing with data series that have gone funny in recent decades. The claim in the first post that it is an ‘insult to one’s intelligence’ to suggest that the Climategate emailers were not being deliberately misleading has disappeared entirely.

I don’t know exactly why Crook climbed down in this abject fashion. To use his own term, these posts are “mealy-mouthed apologies,” albeit “mealy-mouthed apologies” of the kind that clearly had some considerable difficulty making it past the craw. I imagine that Crook prides himself on what he fondly believes to be incisive and trenchant prose that cuts to the heart of the matter (or some such cliche). Being obliged, for whatever reason, to describe a topic that he has strong opinions on as “another issue worthy of debate” must have hurt.

But I’m wandering away from my main argument. It is obvious to anyone who spends any time at all with statistically minded scientists (and social scientists) that they commonly use the term ‘trick’ in just the way that the reports suggest it was used. When a statistician talks about e.g. a clever statistical trick, they are usually not suggesting that they want to lie with numbers, and you are likely to get very confused if you think that they are.

Hence, Crook’s imputation that the “trick” was surely employed toward dishonest ends, was not only potentially libelous, but a clear demonstration that he simply didn’t know what he’s talking about. He had no expertise on the topic, and under his own rules, should have shut up about it. If he had, it would have spared him some considerable embarrassment.

Summing up, I can certainly understand why Crook has an animus against climate scientists. I can even understand why, in some intellectually confused fashion, he links discussions of climate science to the question of what happens when people pronounce confidently on topics where they have no legitimate expertise, and suffer the painful consequences. But his memory seems to be malfunctioning. It wasn’t the climate scientists who got their arses handed back to them on a plate. It was Clive Crook. I trust he’ll be grateful for the reminder.

{ 60 comments }

1

David 10.20.10 at 10:19 pm

All this obsession with the “trick” emails and discussion comes ages after it was well and clearly put in context. Did he go to sleep for eight months, or something? What part of understand doesn’t he understand?

2

Antti Nannimus 10.20.10 at 10:20 pm

Hi Henry,

I can’t be bothered to follow all the links you’ve provided us in this controversy because I’m too busy enjoying the unusually warm and very dry autumn we’re having here in Minnesota this year.

Have a nice day!
Antti

3

bigcitylib 10.21.10 at 12:50 am

At the bottom, there is a question: did somebody “sexy up a graph”, as I’ve heard it put. And thinking on it, I’d say that the report in question did, in that it presented some results in the best light possible. But then the “messiness” behind the graph is discussed at length in the IPCC text, so there was no real deception involved.

The most interesting revelation in climate gate was how little climate science’s media cheerleaders knew about the actual science, including folks like Pearce, Revkin, and Monbiot. As an enthusiastic amateur with some background in Phil and Hist of Science, the CRU emails struck me as entirely unsurprising. Making science is like making sausages, and tough shit for the delicate. If science reporters didn’t know this already, they haven’t been paying attention. (And deep down they probably do know this, because many scientific debates play out in a similar “mean girl” style, albeit before smaller audiences)

4

david 10.21.10 at 1:29 am

I’m so old i remember when good liberals used to explain how Clive Crook was one of the reasonable ones. Brad Delong I’m thinking of you.

He’s been awful from the start, and his arguments have always been the experts say x so x is right, or the experts say y so x is right, so x is right.

x equals what class privileges wants.

5

Jon H 10.21.10 at 2:10 am

I assume ‘trick’ is used in the same way one might say one has a ‘trick’ for doing a quick divide of an unsigned integer by 13, which consists of multiplying it by 1321528399.

It’s a trick, but it’s neither inaccurate nor dishonest.

6

Francis 10.21.10 at 5:15 am

david: neoliberals still seek “reasonable righties” with whom to have rational discussions about policy etc. Crook seemed fit the bill for a while. However, about the time that Megan McAddled joined the Atlantic, it no longer welcomed reasonable righties. Various forms of lunacy (try Sullivan’s recent posts about taxation, including such memorable lines as: “Why are so many on the left incapable of acknowledging that many people who are rich – but, of course by no means all of them – earned it the hard way?”) are now required.

Crook is apparently trying to keep his job; the results speak for themselves.

7

hix 10.21.10 at 6:59 am

This one is fun:

“Technical expertise tends to be narrow, sometimes extremely narrow. ”

“climate scientists spring instantly to mind.”

Clearly, climate scientists steped out of their narrow speciality by talking about climate change.

8

bad Jim 10.21.10 at 8:44 am

Economists and climate scientists spring instantly to mind.

Under the circumstances, we ought to excuse ourselves if the first thing that comes to our minds is the Dunning-Kruger effect: “I know I’m not wrong” or the epitome of cognitive dissonance: “I was right to be wrong.”

We’re not as young and naive as we used to be. Can’t we simply recognize narcissism when we run into it?

9

Robert 10.21.10 at 9:43 am

I studied enough economics on my own to know that orthodox economists, even when speaking about their own area of expertise, are completely unreliable and untrustworthy. On the other hand, I accept the academic consensus on climate change without having studied it much.

As I understand it, Tim Lambert’s conclusions are essentially mine. Yet his areas of study on his own time are the reverse of mine – lots on climate change, not so much on economics.

Are we justified in our beliefs? How? Perhaps one justification is the size and nature of the communities of dissenters in each field.

10

Brett Bellmore 10.21.10 at 11:14 am

As I read his argument in both cases, though phrased differently, it boils down to, “‘Trick’ might be statisticians’ shorthand for ‘technique’, but what is ‘hide’ shorthand for? Are we supposed to be ‘hiding’ things?” And, “If the tree proxy is known to be unreliable during modern times for unknown reasons, is the appropriate response to really to rely on it only during periods for which we lack any independent basis for knowing when it was wrong?”

These actually are reasonable points to raise. There are numerous reasonable points to raise, in fact. They don’t rise to the level of refutation, but, given the extent to which climatologists are relying on models with an embarrassing number of free parameters, trust IS essential. And it’s simply absurd to deny that we have now plenty of reasons not to extend much trust.

I know it irritates Warmers to be told they must do it all again, and this time in the cold light of day, with their critics watching. (Yes, listening to carping from people not fit to shine their shoes.) That they’re going to have to go out and LOOK AT those temperature stations. That their models need to be more physics based.

Always irritated me in high school to be told I had to show my work, too.

11

P O'Neill 10.21.10 at 11:46 am

Speaking of climate change and since we have at least one Italian speaker here, check out the bizarre situation that was revealed by an apparent climate change sceptic op-ed in La Stampa. The denouement is in the comments and the situation is explained more fully here.

12

Jerry Vinokurov 10.21.10 at 1:05 pm

Yes, Brett, you’ve really got those climate scientists pegged with their non-physics-based models. I can’t believe it took this long for some genius to come along and point out that climate science needs to be based on physics. They’ve been doing it wrong all this time!

13

Maggie 10.21.10 at 1:21 pm

suffered through several increasingly difficult years of statistics as an undergrad research student – TRICKS are what help us parcel out the extraneous, the variability in come cases – they help us get to the TRUTH, not cover it up – this is just at the most basic level – tricks to those who are pro’s in the field get even closer to what is actually going on –

14

ajay 10.21.10 at 1:31 pm

Yes, listening to carping from people not fit to shine their shoes.

Brett, just quickly, without looking it up, what’s the dry-air adiabatic lapse rate? Roughly.

Bear in mind that, for atmospheric physics, this is not (cf. CP Snow) the equivalent of asking “have you read a play by Shakespeare?”, but more like asking “what was Shakespeare’s first name?”

Based on your ability to answer this question, consider why you feel qualified to opine on the appropriateness of current global climate models.

15

Michael Drake 10.21.10 at 1:47 pm

“It is obvious to anyone who spends any time at all with statistically minded scientists (and social scientists) …”

That example makes the usage seem more obscure than it is. One can find a “trick” to reliably backing up computer files, stringing a guitar; rebuilding a carburetor, replacing a flush valve flapper, or whatever. A “trick” in this sense is just a convenient way to get a particular job done. And practically everyone uses the word this way.

16

SamChevre 10.21.10 at 1:48 pm

I’ll happily admit that I don’t have the technical skill to decide whether the specifics of a climate model are appropriate.

But I can say two things, because they are common to all models, and require no knowledge of climate modeling specifically to know:

1) Your model result is no more reliable than your inputs.
2) Your free parameters are no more reliable than your calibration data.

17

Jerry Vinokurov 10.21.10 at 1:51 pm

Yes, listening to carping from people not fit to shine their shoes.

It’s not my shoes you need to worry about shining; it’s the shoes of actual climate scientists. You know, the ones that do the actual science.

18

Steve LaBonne 10.21.10 at 2:01 pm

I’ll happily admit that I don’t have the technical skill to decide whether the specifics of a climate model are appropriate.

If you had any sense you’d stop there.

19

SamChevre 10.21.10 at 2:38 pm

Steve,

Can you explain why?

Look: I’m an insurance modeler, with a very strong mathematical modeling background. I strongly doubt that anyone else on this board can do even basic life expectancy modeling.

If I told you that I have a model for causes of death, and I got the data to calibrate it by asking the guys at church what they expected to die of and consulting the obituaries in the Onion, you don’t have to know anything about the interaction between frailty and life expectancy to know that it is not going to give useful answers. You don’t NEED to know anything about life expectancy modeling to know that my calibration data is so poor as to make the model useless, whatever else it gets right.

20

Francis 10.21.10 at 2:43 pm

Brett and Sam: people redo the work all the damn time. Heard of the Clean Climate Code project? Or that Mann completely redid his hockey-stick paper and came out with essentially the same results?

21

Steve LaBonne 10.21.10 at 2:45 pm

Can you explain why?

Because people who start out by admitting that they don’t know what the fuck they’re talking about should shut the fuck up. Is this too complicated a concept for your simple mind to assimilate?

If I told you that I have a model for causes of death…

You know fuck-all about climate data and how they’re calibrated.

Deaf people are not usually hired as music critics. There’s a reason for that.

22

mds 10.21.10 at 2:47 pm

I strongly doubt that anyone else on this board can do even basic life expectancy modeling.

Really? That’s interesting, because you’re almost certainly not the only one here with a “very strong mathematical modeling background.” So you’re simultaneously asserting that (1) you have specific expertise in life expectancy modeling that other mathematical modelers couldn’t trivially replicate, and (2) your generic mathematical modeling background qualifies you to take potshots at those with specific expertise in climate modeling.

23

SamChevre 10.21.10 at 2:56 pm

(1) you have specific expertise in life expectancy modeling that other mathematical modelers couldn’t trivially replicate, and (2) your generic mathematical modeling background qualifies you to take potshots at those with specific expertise in climate modeling.

Yes.

I’m also asserting (3) that claiming specific expertise in life expectancy modeling does NOT exempt me from basic modeling criticism, like “your data source isn’t reliable.”

*Remeber, “Climategate” was about the input data, not the climate models.

24

Alex 10.21.10 at 2:59 pm

physics-based

Which of:

1. Heat flows from a hotter to a colder body.
2. Air with more CO2 absorbs more IR than with less.
3. Energy can neither be created or destroyed.

do you think is wrong?

25

Steve LaBonne 10.21.10 at 3:02 pm

*Remeber, “Climategate” was about the input data, not the climate models.

And as has been confirmed by repeated, well-publicized investigations- which you evidently haven’t bothered to find out about despite the fact that they’re even referred to in the original post, or else you feel free to just lie about- there was no “Climategate” and no significant problem that the supposed “Climategate” revealed with the input data. The whole thing was a right-wing wankfest from start to finish. Which reinforces the point that you don’t know WTF you’re talking about, but are just ignorantly spouting your ideology- just like Clive Crook.

26

ScentOfViolets 10.21.10 at 3:39 pm

Look: I’m an insurance modeler, with a very strong mathematical modeling background. I strongly doubt that anyone else on this board can do even basic life expectancy modeling.

Do tell. Among other things, I teach statistics and once did a bit of contract work on CDC-sponsored study to determine the effect of lifestyle habits on on health.

I imagine that I’m very far from being the only one here with that level of expertise in these sorts of things. So just what do you mean by that last statement? That you’re presuming that the rest of us don’t even know what, say, a Poisson distribution is, or a hypergeometric? That we don’t know anything about parametric statistics? What?

To use the words of another poster, what “specific expertise in life expectancy modeling that other mathematical modelers couldn’t trivially replicate” do you have? Don’t be shy.

27

SamChevre 10.21.10 at 3:50 pm

OK, somehow I’ve gotten this on a tangent. My point is that you don’t need to be a good life expectancy modeler to know that a model based on data from the Onion is a useless model.

Similarly, you don’t need to know the dry-air adiabatic lapse rate to know that if the data you are putting into your model aren’t reliable, the results won’t be reliable.

(I’m inclined to think that “Climategate” was a great hue-and-cry about trivial and unimportant errors, but the supposed errors weren’t the kind that requires familiarity with the specifics of climate modeling to comment on.)

28

Eggplant 10.21.10 at 4:10 pm

Thanks for sharing with us your modelling expertise, SamChevre. I’m sure the scientific community will be delighted to learn that unreliable data lead to unreliable results.

29

bunbury 10.21.10 at 4:13 pm

I think we are all aware of the GIGO concept but I’m not clear what that’s got to do with anything. Climate scientists do not as a matter of course use The Onion as a data source so what is the point? Clive Crook seems to think there is something that needs explaining as a result of an illegally obtained email containing the word “trick” but every examination of what was being talked about has found nothing of the kind.

As Alex and not enough other people point out, we are very confident that we are putting more carbon dioxide into the atmosphere and that the primary effect of this will be to warm the atmosphere. That’s not quite enough to establish that there is man made global warming but there has been a fairly heroic effort to measure whether or not there has been and the strong consensus is that the result of those efforts is a demonstration that there is indeed, as we would naturally expect, man made global warming. At the same time, despite the vast resources available fromoil companies and sceptical governments, no-one has found the underpants gnomes that have been turning down the sun or releasing secret reserves of liquid nitrogen or magic dust into the atmosphere to counteract the undeniable direct effect of extra carbon dioxide.

Not believing in global warming is a lot like believing in the Loch Ness monster. It is in some limited sense possible that there is such a thing but in general we don’t believe in monsters and many concerted efforts have failed to find one. That doesn’t stop eccentric millionaires continuing to look for it or enterprising tour operators playing on people’s nagging doubts.

PS SC, I am not a climate scientist but my cousin is but I can do useful things with mortality models. You will know that any such model has sever limitations but they get used in any case because some assumption needs to be made.

30

Jerry Vinokurov 10.21.10 at 4:43 pm

Why do people think that off-hand comments offered on internet message boards serve as a substitute for genuine scientific labor? I’m honestly curious to know why you, SamChevre, or anyone else for that matter, believes (or seems to believe) that a concept so obvious as “garbage in-garbage out” would have escaped actual climate scientists. I mean, seriously, do you actually think that decades of climatology are now invalid because no one thought to calibrate their thermometry or something?

Curious how the moment we come to a scientific topic which has political implications, all of a sudden someone’s barely-tangentially-related experience in a superficially similar field becomes a substitute for the work of real experts. What a fascinating coincidence.

31

Steve LaBonne 10.21.10 at 4:59 pm

Curious how the moment we come to a scientific topic which has political implications, all of a sudden someone’s barely-tangentially-related experience in a superficially similar field becomes a substitute for the work of real experts. What a fascinating coincidence.

The same kind that we biologists deal with all the time WRT evolution. The strategies employed by climate denialists are sickeningly familiar to us.

32

james 10.21.10 at 5:00 pm

SamChevre:

Are you asking why the tree ring data was used at all as opposed to removed completely?

33

bianca steele 10.21.10 at 5:08 pm

The annoying thing about the point brought up in the OP is that Clive Crook may well believe he has the expertise to form a reliable judgment[1] about both climate change in general and the e-mails in particular, on the basis of the skills taught in journalism schools and his semi-insider access to other insiders.

[1] Even weirder, he may think nobody, including himself, has the ability to form such a reliable judgment, but that he is better at it than anybody who doesn’t have his journalistic skills, his insider access, and his “detachment.”

34

Substance McGravitas 10.21.10 at 5:32 pm

on the basis of the skills taught in journalism schools

In journalism schools they teach you that words like “trick” can have only one meaning.

35

SamChevre 10.21.10 at 5:41 pm

RE: 31

I don’t have an opinion on the tree ring data, or any other particular data set. I was just annoyed by #13, which seemed to imply that you have to understand climate modeling to have any opinions on anything about it, including pointing out that GIGO applies to it just like it does everywhere else.

36

Steve LaBonne 10.21.10 at 5:46 pm

Implying that the actual experts have somehow missed something blindingly obvious that some random internet jackass has noticed (which is unlikely, to say the least) is also a familiar creationist tactic. Say, why aren’t there any crocoducks?

37

Jerry Vinokurov 10.21.10 at 5:48 pm

I don’t have an opinion on the tree ring data, or any other particular data set. I was just annoyed by #13, which seemed to imply that you have to understand climate modeling to have any opinions on anything about it, including pointing out that GIGO applies to it just like it does everywhere else.

You do, in fact, have to have an understanding of climate modeling to understand what counts as worthwhile data for a climate model. I almost have trouble believing that I just wrote the preceding sentence.

38

Eggplant 10.21.10 at 6:05 pm

Okay, I just heard back from the scientific community. It turns out the whole field of statistics is devoted to turning unreliable data into reliable results. Who knew.

39

luminous beauty 10.21.10 at 6:58 pm

SamChevre makes a common error conflating climate models, which are physical models based on physical principles; and statistical models, which are based on statistical regressions of time series data. The basis inputs of climate models can be all zeros and ones or a random set of numbers which are then iterated until the model acquires stability or a set of real world values representing a single iteration. AOGCMs are not inputed with regressed time series of the historical temperature record.

Sam the Horse is wronger than wrong.

40

Salient 10.21.10 at 7:47 pm

Implying that the actual experts have somehow missed something blindingly obvious that some random internet jackass has noticed (which is unlikely, to say the least) is also a familiar creationist tactic.

Not limited to creationists, either. :)

41

Steve LaBonne 10.21.10 at 7:50 pm

And here I was deliberately avoiding that one out of consideration for some of our hosts. ;)

42

Salient 10.21.10 at 8:00 pm

Eh, it’s not a comment on them, and we can balance it out. :)

43

LH 10.21.10 at 11:48 pm

When the “trick” controversy first surfaced, I did a search on one of my own books, a statistical software manual. Turns out I’d used the word “trick” 6 times, never in a context of deception but rather in sentences like,
“When solutions are needed, a simple trick called ‘centering’ often succeeds in reducing multicollinearity in polynomial or interaction-effect models. ”

If some journalists haven’t heard about this type of use, or don’t believe it when told, their education has skipped practical science.

44

Steve LaBonne 10.22.10 at 2:46 am

If some journalists haven’t heard about this type of use, or don’t believe it when told, their education has skipped practical science.

That’s pretty much a given with that lot. Exceptions are quite rare.

45

Cryptic ned 10.22.10 at 3:28 am

I have to say that I’ve been in science about ten years and never heard the word “trick” used to mean “technique”, “algorithm”, “shortcut” or whatever else it might mean in the parlance of these statisticians and mathematicians. The use isn’t as widespread as some are suggesting.

46

Substance McGravitas 10.22.10 at 4:30 am

I hope it’s widespread enough that when you hear someone say “math tricks” you don’t say “That sounds mean and dishonest!”

47

Matt Austern 10.22.10 at 6:14 am

And I see that I used the word “trick” seven times in my book (a technical book about computer programming), with no implication of dishonesty or deception in any of them. Not counting uses of the word “tricky”. For example, “Your implementation might use special tricks to speed up string assignment, but then again, it might not.” And “Sometimes we can use a trick that exploits the C++ type inference mechanism.”

It’s not such an uncommon usage.

48

dsquared 10.22.10 at 7:23 am

1) Your model result is no more reliable than your inputs.

surely the whole point of statistics is that if you have enough of them and the errors are uncorrelated, your model can be more reliable than your inputs?

49

sg 10.22.10 at 8:32 am

google seems to turn up a large number of scholarly articles concerning statistical models that contain sentences like “…For each case, we apply a simple trick to reduce the number
of dimensions required, making the graphical presentation more manageable and…”

It’s as if, were I a journalist, a mere google search for “trick statistical model” would have been sufficient to show me that the phrase in the climategate emails was completely harmless.

Maybe google is in on the trick?

50

Nick Barnes 10.22.10 at 8:37 am

That’s Clear Climate Code, not Clean Climate Code, and it’s now part of the Climate Code Foundation.

Some comments seem to reflect a fantasy that climate scientists are unaware of data quality issues, or neglect them. In fact, many climate scientists spend most of their working lives addressing these questions. More or less every observational paper has at least a section on data quality, and there are papers, projects, and whole groups working on it, in various parts of climate science. Understanding the provenance and likely quality of your data has always been essential to the field. For an outsider’s introduction to the subject, specifically with reference to global climate modelling, read the recent and excellent “A Vast Machine” by Paul N. Edwards.

51

Steve LaBonne 10.22.10 at 12:39 pm

I have to say that I’ve been in science about ten years and never heard the word “trick” used to mean “technique”, “algorithm”, “shortcut” or whatever else it might mean in the parlance of these statisticians and mathematicians. The use isn’t as widespread as some are suggesting.

I have to say that I’ve been in science a lot longer than that and I’ve heard it a million times, about laboratory as well as mathematical techniques. Apparently you need to get out more.

52

Cryptic ned 10.22.10 at 12:51 pm

Thanks Steve, I am now convinced that climate change isn’t occurring, because you think it is and you are so smug.

53

Steve LaBonne 10.22.10 at 1:49 pm

Thanks Steve, I am now convinced that climate change isn’t occurring, because you think it is and you are so smug.

I am in awe of your supremely rational thought processes.

54

Jerry Vinokurov 10.22.10 at 2:47 pm

I have to say that I’ve been in science about ten years and never heard the word “trick” used to mean “technique”, “algorithm”, “shortcut” or whatever else it might mean in the parlance of these statisticians and mathematicians. The use isn’t as widespread as some are suggesting.

And I’ve been in science somewhat less and I hear that usage all the time. The idea that the word “trick” is even remotely controversial here is absurd.

55

hellblazer 10.22.10 at 8:14 pm

Enjoyed the post, enjoying most of the comments. Though I think at #39 the disdain should be aimed at goats rather than horses.

56

dhogaza 10.22.10 at 9:14 pm

SamChevre:

I don’t have an opinion on the tree ring data, or any other particular data set. I was just annoyed by #13, which seemed to imply that you have to understand climate modeling to have any opinions on anything about it, including pointing out that GIGO applies to it just like it does everywhere else.

Well, gosh, you at least need to understand that the tree ring paleoclimate reconstructions (and other paleoclimate reconstructions) aren’t used by climate modelers to keep from making embarrassingly ignorant statements about how climategate shows that models are calibrated with bad data.

Embarrassingly ignorant statements like this:

m also asserting (3) that claiming specific expertise in life expectancy modeling does NOT exempt me from basic modeling criticism, like “your data source isn’t reliable.”

*Remeber, “Climategate” was about the input data, not the climate models.

Assuming “input data” refers to data input into climate models.

You’re simply wrong. Go read about it. Plenty of good stuff on the intertubes. You can download the documentation and source to NASA GISS’s Model E if you’d like.

At an even more basic level, as pointed out above by Luminous Beauty, you at least need to understand the difference between a statistical model (like your life expectancy stuff) and physical models (such as GCMs). Assuming you want to avoid making embarrassingly ignorant statements, that is. If your goal is to embarrass yourself, soldier on …

57

dhogaza 10.22.10 at 9:24 pm

Brett Ballimore:

As I read his argument in both cases, though phrased differently, it boils down to, “’Trick’ might be statisticians’ shorthand for ‘technique’, but what is ‘hide’ shorthand for? Are we supposed to be ‘hiding’ things?” And, “If the tree proxy is known to be unreliable during modern times for unknown reasons, is the appropriate response to really to rely on it only during periods for which we lack any independent basis for knowing when it was wrong?”

These actually are reasonable points to raise.

If it were true, perhaps so. They’re not relied on “only during periods for which we lack any independent basis for knowing when it was wrong”.

False for a couple of reasons:

1. Only some, not all, of the tree ring series suffer from the problem (which is well-known, and openly discussed among scientist, and in IPCC reports, etc). Of those that do, they only suffer the problem for a relatively short period of time covered by the instrumental record – they match well for about 2/3, and don’t for about 1/3.

2. They’re calibrated against non-tree ring proxy data for a lengthy stretch of time before the instrumental record, and match well.

There are numerous reasonable points to raise, in fact. They don’t rise to the level of refutation, but, given the extent to which climatologists are relying on models with an embarrassing number of free parameters, trust IS essential. And it’s simply absurd to deny that we have now plenty of reasons not to extend much trust.

I know it irritates Warmers to be told they must do it all again..

Why? Because people like you and SamChevre don’t know that the paleoclimate reconstructions you’re talking about aren’t used by modelers? Because you don’t understand the context of the divergence problem?

Of course it irritates climate modelers to be told “you have to do it all over again” by people who don’t have the first clue as to how climate models work and how this work is in no way dependent on the tree ring-based paleoclimate reconstructions. Not at all. Nada.

I love the down-is-up, black-is-white world of the denialist. “I’m wrong and don’t know the first thing about what I’m talking about, therefore YOU MUST DO YOUR WORK OVER AGAIN, while I’m looking over your shoulders making ignorant complaints about the job you’re doing”.

It’s liking having you lean over your car mechanic, telling him, “you *must* soak the air cleaner in olive oil, and fill the cylinders with water, I *insist*, because I don’t trust you”.

58

maidhc 10.23.10 at 6:38 am

There’s a viewpoint that unless Science can totally explain everything there is to know with 100% accuracy, then Science is totally discredited.

Science can’t explain why honeybees are dying? Then evolution is a hoax!

59

Daragh McDowell 10.23.10 at 3:15 pm

I stopped reading Crook’s blog after I noticed that there was a substantial groove starting to wear in my desk due to repeated head bangings…

60

Ken Fabos 10.24.10 at 9:37 pm

I’m not surprised that journalists and opinion writers are still repeating the same discredited misconceptions and misunderstandings – and imagine they will undermine their own credibility by admitting when they are wrong. They clearly prefer undermining it by insisting they aren’t. That they do so and complain about ‘unqualified’ bloggers pointing out that they are wrong just seems like pettiness. Is there any point pointing our that it’s not hard to find bloggers are much better qualified than they are? Thank you Internet. And thank you bloggers – sure, you aren’t all created equal but it’s not that hard to sift for quality as well as qualifications.

Comments on this entry are closed.